International RILEM Conference on Synergising Expertise towards Sustainability and Robustness of Cement-based Materials and Concrete Structures: SynerCrete’23 - Volume 1 (RILEM Bookseries, 43) 3031332105, 9783031332104
This book highlights the latest advances, innovations, and applications in cement-based materials (CBM) and concrete str
115 46 218MB
English Pages 1379 [1363] Year 2023
Preface
Organisation
Contents
RILEM Publications
Keynote
Substituting Natural Pozzolan with Artificial Derived from Industrial Perlite Waste for Mortar Production
1 Introduction
2 Experimental
2.1 Testing Raw Materials
2.2 Mortars
2.3 Grouts
3 Conclusions
References
Modelling of Cement-Based Materials
A Benchmarking of Slag Blended Cement Hydration Models
1 Introduction
2 Models
2.1 Brief Overview of Cement Composition
2.2 Ordinary Portland Cement Analytical Models
2.3 Analytical Models for Slag Blended Cement
2.4 Numerical Models for Slag Blended Cement
3 Comparison
3.1 Summary of Models’ Characteristics
3.2 Comparison on the Quantity of Consumed and Produced Phases
4 Conclusions
References
Refining Kinetic Models for SCM Reactivity in Blended Cements
1 Introduction
2 Data
3 Kinetic Model
4 Results and Discussion
4.1 Parameterisation of SCM Reaction Kinetics
4.2 Temperature Dependence
4.3 Alkali and Ca(OH)2 Supply
5 Conclusions and Perspectives
References
Atomistic Dissolution of β-C2S Cement Clinker Crystal Surface: Part 1 Molecular Dynamics (MD) Approach
1 Introduction
2 Computational Detail
2.1 Model Construction
3 Results and Discussion
3.1 Pre-hydration of ß-C2S
3.2 Dissolution of Calcium from the First Layer of (100) Surface of β-C2S: Case I & Case II, also Including Various Neighbour Scenarios
3.3 Dissolution of Silicate from the First Layer of (100) Surface of β-C2S: Case III with Various Neighbour Scenarios
3.4 Dissolution of Calcium from the Second Layer of (100) Surface of β-C2S: Case VI with Various Neighbour Scenarios
3.5 Upscaling the Dissolution Rate for (100) Surface of β-C2S
4 Conclusions
References
Einstein Explains Water Transport in C-S-H
1 Introduction
2 Transport at the Molecular Scale
2.1 From Displacement to Self-diffusion
2.2 From Self-diffusion to Permeability
3 Molecular Simulations Assessing Transport Properties of Water Confined in C-S-H
3.1 Self-diffusion from Equilibrium MD
3.2 Hydrodynamics Under Forced Flow from NEMD
4 Transport at the Gel Scale: A Mean-Field Approximation
5 Results
5.1 Transport Properties at the Molecular Scale at the Molecular Scale
5.2 Transport Properties at the Molecular Scale at the Gel Scale
6 Conclusions
References
Is Thermal Pressurization in C-S-H Relevant for Concrete Spalling?
1 Introduction
2 Models and Methods
2.1 C-S-H Structure and Force Field
2.2 Simulation Details
2.3 Confining Pressure Computation
3 Results and Discussion
3.1 Confining Pressure
3.2 Water Content
4 Conclusions
References
Multiscale Modeling of the Dielectric Response of C-S-H
1 Introduction
2 Molecular Models and Methods
2.1 Dielectric Response from Molecular Simulations
2.2 Simulation Details
3 Upscaling the Dielectric Permittivity
4 Results
4.1 Static Dielectric Response at the Molecular Scale
4.2 Frequency-Dependent Response at the Molecular Scale
4.3 Dielectric Response of C-S-H Gel
5 Conclusions
References
Quantum Mechanically Informed Kinetic Monte Carlo Models for Hydrogen Diffusion in BCC-Iron
1 Introduction
2 Hydrogen as an Interstitial in BCC-Iron
2.1 Jump Parameters
3 Conclusion
References
Numerical Modeling of Water Transfer in Geomaterials: Application to a Concrete Tunnel Subjected to Both Drying and Liquid Overpressure
1 Introduction
2 Water Transfer Model
2.1 Unsaturated Zone
2.2 Saturated Zone
3 Numerical Simulation of Radioactive Waste Storage Tunnel
3.1 Description of the Geometry and the Boundary Conditions
3.2 Results Analysis
4 Conclusions and Perspectives
References
Replicating the Failure Mechanism of a Real-World Event with the Lattice Discrete Particle Model
1 Introduction
2 Modelling Strategy
3 Results
4 Summary and Conclusions
References
Thermomechanical Investigations for the Design of Reinforced Concrete Facings
1 Introduction
2 Model
2.1 FE-Model
2.2 Studied Scenarios
2.3 Development of Mechanical Properties
3 Results
3.1 Temperature and Stress Development
3.2 Crack Width Calculation
4 Conclusions
References
Modelling of the CO2 Uptake by Recycled Concrete Aggregates
1 Introduction
2 Program
2.1 Model Equations
2.2 Application to the Modelling of Carbonation at Stockpile Scale
2.3 Application to the Modelling of Carbonation at Grain Scale
3 Results and Discussion
3.1 Atmospheric Carbonation of RCA at Stockpile Scale
3.2 Carbonation at Grain Scale
4 Conclusions
References
A Multiscale Multiphysics Platform to Investigate Cement Based Materials
1 Introduction
1.1 Industrial Context and Issues
1.2 A Software Toolbox Dedicated to Concrete
2 Brief Description of Vi(CA)2T Toolbox
2.1 Overview of Embedded Physical Models
2.2 Verification and Validation
3 Practical Applications
3.1 Creep Response for Containment Building Structural Analysis
3.2 Permittivity Measurements to Assess Water Content in Structure
4 Conclusion and Next Steps
References
Estimation of Protected Paste Volumes by Dirichlet Tessellation Associated with Point Processes of Air Voids
1 Introduction
2 Experimental Procedures
2.1 Materials and Mix Proportions of Concretes
2.2 Image Acquisition and the Automatic Linear Traverse Measurement of Air Voids
2.3 Conversion of Air Voids to a Planar Point Process and Dirichlet Tessellation of Cement Paste Matrix
2.4 Simulation of Point Patterns and Determination of the Characteristic Tile Area
2.5 Characteristic Distance Defined by the Nearest Neighbor Distance Distribution Function for a Point Process
3 Results and Discussion
3.1 Dirichlet Tessellation Associated with an Actual Point Pattern
3.2 Dirichlet Tessellation for Simulated Point Patterns with the Same Point Intensity
3.3 Relationship Between Characteristic Distances of the Nearest Neighbor Distance Distribution Function and the Tile Sizes of Dirichlet Tessellation
3.4 Relationship Between the Surface Density of Air Voids and the Tile Sizes for 95% Protection
4 Conclusions
References
MASKE: Particle-Based Chemo-Mechanical Simulations of Degradation Processes
1 Introduction
2 Methodology
2.1 The MASKE Simulator
2.2 The Calcium Hydroxide System in This Article
3 Results
3.1 Stress-Free Equilibrium and Dissolution/Growth of a CH Nanocrystal
3.2 Pressure-solution Creep
3.3 Crystallization Pressure
4 Conclusion
References
Temperature-Dependent Behavior of Mature Cement Paste: Creep Testing and Multiscale Modeling
1 Introduction
2 Multiscale Modelling of Cementitious Materials Using Methods from Continuum Micromechanics
2.1 Basics of Mean-Field Methods of Continuum Micromechanics
2.2 State-of-the-Art Multiscale Model for Strength and Creep Upscaling and Mechanical Properties of Hydrate Gel Needles
3 Temperature-Dependent Stiffness of the Hydrate Gel Needles
3.1 Macroscopic Experiments
3.2 Validation of the Temperature-Dependent Values of the Modulus of Elasticity of Table 1 by Means of Ultrasonic Pulse Velocity Tests
3.3 Multiscale Modelling
4 Conclusions
References
Development of an Experimental-Numerical Approach to Model Cement Paste Microstructure Using Quantitative Phase Assemblage from XRD and Thermodynamic Modeling
1 Introduction
2 Materials and Methods
3 Methodology
4 Results and Discussion
4.1 Experimental and GEMS Phase Assemblage
4.2 Calibrated µic Microstructure Model
4.3 µic Calibrated 3D Microstructure and Upscaling Perspectives
5 Conclusions
References
Experimental and Numerical Investigations on Concrete Abrasion of Hydraulic Structures
1 Introduction
2 Experimental Study
2.1 Sediment-Laden Flow Impact Method
2.2 Raw Materials and Mixture Design
2.3 3D Scanning Method and the Data Analysis
2.4 Experimental Results
3 Numerical Analysis
3.1 3D Meso-Scale Concrete Model
3.2 Material Property
3.3 Loads and Boundary Conditions
3.4 Simulation Analysis Results
4 Conclusions
References
Multi-physics Modelling of Concrete Shrinkage with the Lattice Discrete Particle Model Considering the Volume of Aggregates
1 Introduction
2 Multi-physics Framework
2.1 Mechanical Model
2.2 Hygro-Thermo-chemical (HTC) Model
2.3 Improved Shrinkage Model
2.4 One-Way Coupling
3 Simulations of Experimental Campaign
3.1 Experimental Description
3.2 Calibration of Model Parameters
4 Results and Discussion
5 Conclusions
References
Building Information Modelling
Enhanced Interoperability between Geotechnical and Structural Engineering for 3D Building Models
1 Introduction
2 Benefits and Limitations of the Subgrade Reaction Method
3 Engineering Approach to Exchange Settlement Troughs
3.1 3D Multi-storey Building
3.2 Exchanging Nodal Displacements
4 Conclusion
References
Industry 4.0 Enabled Modular Precast Concrete Components: A Case Study
1 Introduction
2 State of the Art
2.1 Precast Concrete Components and Modular Construction
2.2 Industry 4.0 and Digital Twin
2.3 The Asset Administration Shell
3 Concept
3.1 Serial Production of Precast Concrete Modules
3.2 The Digital Twin of Precast Concrete Modules
4 Application
4.1 Design
4.2 Digital Twin Representation of the Y-Shaped Precast Concrete Module
4.3 Production
4.4 Results
5 Conclusion
References
Integral BIM-Based Planning
1 Introduction
2 Integral Planning Process
3 Case Study
3.1 Project A
4 Analysis and Discussion
5 Conclusions
References
RecycleBIM Approach Towards Integrated Data Management for Circularity: Proof of Concept in a RC Building
1 Introduction
2 The RecycleBIM Project
3 Data: Information Needs and Database
4 The Web-Based Tool
5 Application to RC Building
6 Conclusions
References
Towards Standardization of Data for Structural Concrete: Product Data Templates
1 Introduction
2 PDTs: Standards and Existing Initiatives
3 Product Data Templates for Concrete Products
3.1 Methodology of EN ISO 23386 e 23387 for PDT Creation
3.2 Product Data Template Proposal
4 Conclusion
References
Non-standardised Testing Techniques
An Innovative Experiment for Air Pressure Measurements in Crack Models Representative of Real Cracks in Concrete
1 Introduction
2 Materials and Methods
2.1 Preliminary Remarks
2.2 Samples’ Realization
2.3 Experimental Device and Procedures
3 Results and Discussion
4 Conclusions
References
Monitoring of Capillary Pressure Evolution in Young Age Concrete Using High Capacity Tensiometers
1 Introduction
2 High Capacity Tensiometer Design for Monitoring Capillary Pressure in Early Age Concrete
2.1 Ceramic Filter
2.2 Pressure Transducer
2.3 Water Reservoir and Sheath
3 Saturation, Calibration, Measuring Range and Effect of Temperature
3.1 First Saturation
3.2 Calibration
3.3 Measuring Range
3.4 Effect of Temperature
4 Monitoring the Evolution of Capillary Pressure in Concrete Using HCT-Cs
5 Conclusion
References
A Transient Re-touching of Carbon Fiber to Cement Interface Under Single Fiber Pullout Testing with Direct Current Measurement
1 Introduction
2 Materials and Methods
2.1 Materials
2.2 Preparation of the Single Carbon Fiber Specimens
2.3 Linear Sweep Polarization Voltammetry (LSV)
2.4 Direct Current Measurement During the Whole Pullout Process
3 Results
3.1 Effect of Polarization
3.2 Electromechanical Single Carbon Fiber Pullout
4 Conclusions
References
Effect of Moisture on the Piezoresistive Properties of Aluminosilicate-Based Building Materials
1 Introduction
2 Experimental Part
2.1 Martials and Sample Preparation
2.2 Testing Procedures
3 Results and Discussion
4 Conclusions
References
Thermal Gradient in Large Concrete Test Bodies: A Macroscale Experimental Approach
1 Introduction
2 Experiments
2.1 Methodology
2.2 Preparation of the Formwork
2.3 Concrete Mixture
3 Results
3.1 Temperature Profiles
3.2 Thermal Gradient
4 Discussion
5 Conclusion
References
Developing a New Rapid, Relevant, and Reliable (R3) Method for Accelerated Measurement of Carbonation Progress at Gas Overpressure
1 Introduction
2 Development of a Device for Automated Accelerated Carbonation Testing
2.1 Description of the Setup
2.2 Automatic Control of Pressure and CO2 Concentration
2.3 Limitations of the Measurement Setup
3 Experimental Program
3.1 Materials and Pre-conditioning
3.2 Carbonation Progress at Constant Gas Overpressure
3.3 Carbonation Progress at Cyclic and Stepwise Increasing Gas Overpressure
3.4 Results and Discussion
4 Conclusions
References
Assessing Cement Matrix Permeability by Neutron Dark Field Imaging
1 Introduction
2 Preliminary Assessment of the Me-S-H Nanocomposite and its Effect
3 Neutron Dark Field Imaging
3.1 Sample Preparation
3.2 DFI Image Analysis
4 Discussion and Conclusions
References
Dam Concrete in Situ Creep Tests. Experimental Setup and Results from Six Large Concrete Dams
1 Introduction
2 Experimental Setup
2.1 General Aspects
2.2 Moulds
2.3 Location of the System
2.4 Strainmeters
2.5 Oil Piping
3 Comparison of the Results from Different Dams
4 Final Remarks
References
Preliminary Analysis of Non-destructive Test Methods to Evaluate the Self-healing Efficiency on the Construction Site
1 Introduction
2 Materials and Methods
2.1 Vascular Network
2.2 Concrete Composition and Specimen Preparation
2.3 Microscopy
2.4 Moisture Content
2.5 Air Permeability
2.6 Water Permeability with Karsten Tubes
2.7 Resistivity
2.8 Ultrasound
3 Results and Discussion of Laboratory Tests
3.1 Microscopy
3.2 Moisture Content
3.3 Air Permeability
3.4 Water Permeability with Karsten Tubes
3.5 Resistivity
3.6 Ultrasound
3.7 Summary
4 Application of Test Methods on Site
5 Conclusion
References
Open-Source EMM-ARM Implementation for Mortars Based on Single-Board Computer
1 Introduction
2 EMM-ARM Application to Mortars: Original Setup
3 Hardware of the New Design
3.1 Computing Device
3.2 Excitation of the Specimen
3.3 Response Acquisition of the Specimen
4 Software of the New Design
4.1 Multiprocessing on Raspberry Pi
4.2 Modal Parameters Identification Method
5 Preliminary Results and Discussions
5.1 Tests Conducted
5.2 Results
5.3 Discussions
6 Conclusions
References
Understanding the Degradation of Concrete Structures During the Nitrification Process for the Treatment of Wastewater: A Lab Biological Degradation Test
1 Introduction
2 Materials and Methods
2.1 Materials Description
2.2 Protocol for in Situ Exposure
2.3 Protocol for Lab Biological Test
3 Results
3.1 Monitoring of in Situ Exposure
3.2 Results for Lab Biological Test
4 Conclusions Remarks
References
A Cost-Effective Micro-controller Based System for EMM-ARM Tests in Cement Paste
1 Introduction
2 Original EMM-ARM Test Setup for Hardening Cement Pastes
3 Proposed System
3.1 Overview
3.2 Data-Acquisition System Hardware
3.3 Choice of Accelerometer
3.4 Source Code for Data-Acquisition
3.5 Post-processing Software
3.6 Bill of Materials and Cost Estimation
4 Validation
4.1 Methodology
4.2 Experimental Results
5 Conclusions
References
Analysis of Concrete Transient Thermal Deformation in the Context of Structures Submitted to Various Levels of Temperature and Mechanical Loading
1 Introduction
2 Materials and Methods
2.1 Materials
2.2 Experimental Program
3 Results and Discussion
3.1 Free Strains in Autogenous Conditions
3.2 Total Deformation and Basic Creep Strain
3.3 Coupled Effects of Temperature and Loading Rate
4 Conclusion
References
Semi-circular Bending Test to Evaluate the Post Cracking Behaviour of Fibre Reinforced Concretes
1 Introduction
2 Semi-Circular Bending Test (SCB)
2.1 Inverse Analysis
3 Methodology
3.1 Materials and Dosage Definition
3.2 Inverse Analysis Procedure
4 Results and Discussions
4.1 Experimental Results
4.2 Inverse Analysis Results
5 Conclusions
References
Development of Damage Monitoring Techniques During Fatigue Compression Test on Concrete Specimen
1 Introduction
2 Materials and Methods
2.1 Test Setup for Mechanical Testing
2.2 Fatigue Monitoring
2.3 First Non-destructive Method: Ground Penetrating Radar
2.4 Second Non-destructive Method: Modal Analysis
2.5 Third Non-destructive Method: Stiffness Damage Test, SDT
3 Results
3.1 Fatigue Monitoring
3.2 First Non-destructive Method: Ground Penetrating Radar
3.3 Second Non-destructive Method: Modal Analysis
3.4 Third Non-destructive Method: Stiffness Damage Test, SDT
4 Discussion
5 Conclusion
References
Comparison of Different Approaches for Quantification of Amorphous Phase in Hydrated Cement Paste by XRD
1 Introduction
2 Materials and Methods
2.1 Materials
2.2 Methods
3 Results
4 Conclusions
References
Innovative FWD Testing on Concrete Slabs
1 Introduction
2 Multi-directional FWD Tests
2.1 Test Protocol
2.2 Experimental Data from the New Slab A2-54440
2.3 Experimental Data from the Old Slab A2–54003
2.4 Asymmetries in the Structural Behavior of the Slabs
3 Structural Analysis of the New Slab
3.1 Identification of a Uniform Modulus of Subgrade Reaction
3.2 Extension and Consideration of a Uniform Eigendeflection, w0
4 Conclusions
Appendix A
References
Numerical Simulations for the Determination of Chloride Diffusivity in Reinforced Concrete Under Tensile Load
1 Introduction
2 Methodology
2.1 Accelerated Migration Test
2.2 Numerical Simulation
3 Results of the Sensitivity Studies
3.1 Electrical Field
3.2 Chloride Upstream Concentration
3.3 Distance Between 2 Cells
3.4 Influence of a Straight Crack
4 Conclusion
References
Valorisation and Recycling of Non-binder Components of Concrete
Deconstructable Concrete Structures Made of Recycled Aggregates from Construction & Demolition Waste: The Experience of the DeConStRAtion Project
1 Introduction
2 Design and Prototyping
2.1 Design of the Structural Components
2.2 Prototyping Activities
2.3 Raw Materials and Mixture Proportioning
2.4 Casting Procedures
2.5 Post-tensioning of RAC Modules with Shape Memory Alloy Bars
3 Full-Scale Flexural Tests
3.1 Description of Test Setup
3.2 Test Results
4 Conclusion
References
Experimental Investigation of the Influence of Hemp Particles on Hydration Kinetics of Multicomponent Mineral Binder
1 Introduction
2 Experimental Procedure
2.1 Materials
2.2 Experimental Methods
3 Results and Discussion
3.1 Hydration Kinetics Analysis of the Binder
3.2 Semi-quantitative Methods of Hydration Tracking
4 Conclusions
References
Development of Concrete Mixtures Based Entirely on Construction and Demolition Waste and Assessment of Parameters Influencing the Compressive Strength
1 Introduction
2 Experimental Program
2.1 Materials
2.2 Mixture Design
2.3 Curing and Testing
3 Results and Discussion
3.1 Compressive Strength
3.2 SEM/EDX Analysis
3.3 MIP Analysis
4 Conclusions
References
Utilisation of COVID-19 Waste PPE in the Applications of Structural Concrete
1 Introduction
2 Materials, Mix Design and Methodology
3 Results and Discussion
3.1 SEM-EDS Analysis
3.2 Compressive Strength Results
4 Conclusions and Recommendations for Future Work
References
Microbial Induced Calcium Carbonate Precipitation (MICP) Treatments for the Reduction of Water Absorption of Recycled Mixed Aggregates
1 Introduction
2 Material and Methods
2.1 Recycled Mixed Aggregate (RMA)
2.2 MICP Reagents
2.3 MICP Treatment Methods
2.4 Preparation of Mortar Samples
2.5 Testing Methods
3 Results and Discussion
3.1 Preliminary Tests
3.2 MICP Treatments
3.3 Mortar Properties
4 Conclusions
References
Use of Recycled Carbon Fibres in Textile Reinforced Concrete for the Construction Industry
1 Introduction
2 State of the Art
3 Materials and Methods
4 Results
4.1 Recycled Carbon Fibre Nonwoven Reinforcement
4.2 Recycled Carbon Fibre Yarn Reinforcement
4.3 Comparison of Results of Different Recycled Carbon Fibre Reinforcements
5 Conclusion and Outlook
References
Valorization of Sulphidic Mine Tailings as Artificial Aggregate: Implementation in Cement-Based Materials
1 Introduction
2 Materials and Methods
3 Results and Discussion
4 Conclusions
References
Influence of the Composition of Original Concrete on the Carbonated Recycled Concrete Aggregates Properties
1 Introduction
2 Materials and Methods
2.1 Materials
2.2 Methods
3 Results and Discussion
4 Conclusion and Perspectives
References
Evaluation of Eco-friendly Concrete Release Agents Based on Bio-Waxes
1 Introduction
2 Materials
2.1 Mould Release Agent (MRA)
2.2 Formwork Surfaces
2.3 Cement
3 Methods
3.1 Part 1 Study of Interfacial Properties Between MRA and Formwork Surface
3.2 Part 2 Study of Interaction Between Concrete and Formwork
4 Results
4.1 Part 1 Study of Interfacial Properties Between MRA and Formwork Surface
4.2 Part 2 Study of Interaction Between Concrete and Formwork
5 Conclusion
References
Durability Characterization of Concrete Using Seashell Co-products as Aggregate Replacement
1 Introduction
2 Materials
2.1 Raw Materials
2.2 Concrete Mix Design
3 Methods
3.1 Porosity Accessible to Water Measurement
3.2 Methods for Chloride Resistance Measurements
3.3 Method for Gas Permeability Measurements
4 Results and Discussion
4.1 Porosity Accessible to Water Results
4.2 Apparent Chloride Diffusion and Resistivity Results
4.3 Gas Permeability
5 Conclusion
References
Production Waste Fibres as a Sustainable Alternative for Strengthening Cementitious Composites
1 Introduction
2 Waste Fibre Streams
3 Characterization of Waste Fibres
3.1 Methods
3.2 Results and Discussions
3.3 Discussion
4 Conclusions
References
Effect of Elevated Temperatures on Concrete Made with Ash from Wood Biomass and Recycled Polymer Fibers from Waste Rubber
1 Introduction
2 Experimental Program
2.1 Materials and Concrete Mixes
2.2 Specimens – Curing, Dimensions and Heat Treatment
3 Results and Discussion
3.1 Compressive Strength
3.2 Modulus of Elasticity
3.3 Ultrasonic Pulse Velocity
3.4 Gas Permeability
4 Conclusion
References
Use of Marble Sludge Waste and Polypropylene Fibers in Developing Eco-friendly Strain Resilient Cementitious Composites
1 Introduction
2 Research Significance
3 Materials and Specimens
4 Experimental Results and Analysis
4.1 Compressive Behaviour
4.2 Tensile Behaviour
5 Conclusions
References
Test Methods to Evaluate Durability of Blended Cement Pastes Against Deleterious Ions (TC 298-EBD)
A Novel Uniaxial Penetration Approach to Investigate External Sulfate Attack on Blended Cement Pastes
1 Introduction
2 Materials and Methods
2.1 Materials
2.2 Sample Preparation
2.3 Macroscopic Observations
3 Results and Discussion
3.1 Appearances
3.2 Mass Variations and Expansion
3.3 Chemical Compositions in C-S-H Hydrates
3.4 Sulfate Ingress Maps
3.5 Insights on the Salt Crystallization Degradation
4 Conclusions and Perspectives
References
The Penetration of Chlorides Within Cement Pastes Under an Electric Field
1 Introduction
2 Materials and Methods
2.1 Cementitious Mixtures
2.2 Chloride Ion Penetration Resistance Test
2.3 Porosity and Water Permeability Tests
2.4 Stereomicroscope and SEM Observation
3 Results and Discussion
3.1 Chloride Penetration, Porosity and Water Permeability
3.2 Macroscopic Chloride Profiles
3.3 Microstructure
3.4 The Chloride Profiles
4 Conclusions
References
Chloride Binding in Slag Containing Composite Cements
1 Introduction
2 Materials and Methods
3 Results and Discussions
4 Concluding Remarks
References
XCr/√t as an Indicator of the Resistance Against Bulk Chloride Diffusion
1 Introduction
2 Materials and Methods
2.1 Investigated Systems
2.2 Specimen Preparation
2.3 Bulk Diffusion on Cement Pastes
2.4 Chloride Profiles by μXRF
3 Results and Discussion
3.1 Chloride Profiles and Reference Chloride Intensity
3.2 Determination of Dapp and Fitting of Chloride Profiles
3.3 √Dapp vs. XCr/√t for Different Reference Chloride Contents
4 Conclusion
References
Calibration of Tang’s Model for Concentration Dependence of Diffusion in Cementitious Materials
1 Introduction
2 Diffusion Coefficients are Defined by Mathematical Models
2.1 Pore Solution (and Surface) Electro-chemistry Effects
2.2 Multi-ionic Frictional Effects for Concentrated Electrolytes
2.3 Pore Structure Geometrical Effects
2.4 Updated Tang’s Model
3 Methodology
4 Results and Discussion
5 Conclusion
References
Self-sufficient Reactive Transport Modelling in Cement-Based Materials with Low-Carbon Footprint
1 Background
2 Modelling Framework
2.1 Transport Modelling
2.2 Pore Partitioning Model
2.3 Modelling Reactive Processes
2.4 Thermodynamic Modelling
3 Summary
References
Numerical Simulation of Chloride Ion Ingression in Mortar Incorporating the Effect of ITZ Using an Integrated COMSOL-IPHREEQC Framework
1 Introduction
2 Model Description
2.1 Thermodynamic Model
2.2 Homogenization Method
2.3 Transport Model
3 Results and Discussion
4 Conclusion
References
Resistance of Lime-Natural Pozzolan Mortars in Salt-Laden Environments
1 Introduction
2 Materials and Methods
3 Results and Discussion
3.1 Content of Anions
3.2 Mass Changes
3.3 Changes in Microstructure
4 Conclusions
References
Performance of Migrating Corrosion Inhibitors in Cracked Reinforced Concrete Exposed to Marine Environment
1 Introduction
2 Concrete Mix Design, Casting and Curing
2.1 Materials
2.2 Specimen Types
2.3 Curing Conditions
2.4 Exposure Conditions
3 Corrosion Evaluations
4 Results and Discussion
5 Conclusions
References
Slag or Reacted Binder, Which Dissolves First in Sulphuric Acid?
1 Introduction
2 Materials and Methods
2.1 Materials
3 Results and Discussion
4 Conclusions
References
Design and Performance
Assessing Early-age Dynamic Elastic Modulus in High-Performance Concrete
1 Introduction
2 Experimental Information
2.1 Materials and Mix Proportions
2.2 Test Methods
3 Results and Discussion
3.1 Relationship Between Ed and Es
3.2 Influence of Curing Temperature on Early-Age Ed and Es
3.3 Influence of Sustained Tensile Loading on Early-age Ed and Es
4 Conclusions
References
Development of Filling Grout Material for Boulder Ground
1 Introduction
2 Fundamental Study
2.1 Challenges in Implementing Filling Procedure of Grout Material into Boulder Ground
2.2 Evaluation Method of Tensile Strength
3 Test Overview
3.1 Mix Proportions of Grout and Test Flow
3.2 Method of Mixing Materials
3.3 Experimental Test
4 Test Results
4.1 Test Results of “Prototype”
4.2 Compressive Strength
4.3 Relationship Between Vane Shear Strength and Flow Value
4.4 Tensile Strength
4.5 Pressurized Bleeding Rate
4.6 Effect of Mixing Silica Fume and Bentonite
5 Summary
References
Design and Development of Multi-faceted Engineered Concrete
1 Introduction
2 Design Principle
3 Material
4 Experimental Program
5 Results and Discussion
6 Conclusion
7 Future Work
References
Assessment of Deviations in Concrete Properties Quantified Under Laboratory Conditions and from the Construction Site
1 Introduction
2 Investigation Program
2.1 Concrete Compositions
2.2 Specimen Design and Preparation
2.3 Performance Testing
2.4 Results and Discussion
3 Conclusion and Outlook
References
Performance of Powder Actuated Fasteners as Direct Fastenings in Steel Fibre Reinforced Concrete
1 Introduction
2 Fundamental Background Knowledge and Scope of this Study
2.1 Aim and Structure of the Study
3 Experimental Set-Up and Design
4 Results and Interpretation
4.1 Setting Assessment – Penetration Depth
4.2 Setting Assessment – Surface Damage
4.3 Pull-Out Load Assessment and Interpretation
5 Summary
References
Fiberglass Mesh Reinforced Rendering Mortar: Effect of Fiberglass Reinforcement
1 Introduction
2 Materials and Experimental Methods
2.1 Materials
2.2 Experimental Methods
3 Results and Discussions
4 Conclusion
References
Effect of Spatial Variability on the Failure Behaviour of a Reinforced Concrete Shear Wall
1 Introduction
2 Presentation of the Case Study
3 Parametric Study
4 Global Response of the Shear Wall
5 Cracking Patterns
6 Conclusion
References
In-situ Casting Method and Durability of Cementitious Materials at Deep Seafloor
1 Introduction
2 Field Trials of In-situ Mortar Casting
2.1 In-situ Casting with Conventional PC Mortar
2.2 In-situ Casting with a Unique Injection Mortar Kit
3 Deep-Sea Exposure of Hardened Cement Pastes
3.1 Mix Proportions of Paste Specimens
3.2 Experimental Results and Discussion
4 Conclusion
References
Axial Strength of Pile Head Embedded with Steel Column: Effect of Reinforcing Bar on Axial Strength
1 Introduction
2 Overview of Experiment
3 Failure Pattern and Load-Displacement Relationship
4 Strength Evaluation
5 Conclusion
References
Probabilistic Assessment of RC Piers Considering Vertical Seismic Excitation Based on Damage Indices
1 Introduction
2 Description of Model
2.1 Finite Element Modeling
3 Probabilistic Seismic Demand Models and Damage Limit States
3.1 Damage Limit States
4 Comparison of Results
4.1 Forces Demands
4.2 Fragility Analysis
5 Summary and Conclusion
References
New Conceptions and Constructive Methods for Pumped Storage Hydropower Plants
1 Introduction
2 Modeling Parameters for 3D Printed Process (Analytical Approach)
3 Case Study and the Choice of the Shape
4 Material
5 Numerical Modeling in the Fresh State
6 Numerical Modeling in the Hardened State
7 Concluding Remarks
References
Flexural and Shear Performance of Precast Prestressed Composite Beams
1 Introduction
2 Materials and Method
2.1 Research Program
2.2 Four-Point Bending Test – Flat Interface Beam
2.3 Three-Point Bending Test – Rib-shaped Beam
3 Experimental Results
3.1 Four-Point Bending Test – Flat Interface Beam
3.2 Three-Point Bending Test – Rib-shaped Beam
4 Numerical and Analytical Calculations
4.1 Parameters of Finite Element Model (FEM 3D)
4.2 FEM Results
4.3 Analytical Calculation - Model Code 2010
5 Summary and Conclusion
References
Structural Health Monitoring and Maintenance Management
A Novel Service Life Prediction for Reinforced Concrete Infrastructure Systems
1 Introduction
2 Background
2.1 Transportation Network Optimization Approach
2.2 System Reliability Approach
2.3 Predicted Performance Curve
3 The Subway System (SUBSYS) Performance Model
3.1 Station System Performance Model
3.2 Tunnel System Performance Model
4 The SUBSYS Model Application and Results
4.1 Stations Performance Curves
4.2 Tunnels Performance Curves
5 Conclusion
References
Monitoring of Reinforced Concrete Structures: Disposal of Low and Intermediate Level Radioactive Waste
1 Introduction
2 Methodology
3 Results
3.1 Temperature
3.2 Deformation
3.3 Oxygen Availability
3.4 Concrete Resistivity
3.5 Corrosion Potential
3.6 Corrosion Rate
4 Discussion
5 Conclusion
References
Development of a Low-Budget Monitoring System for Expansion Joints with Real-Time Data Analyses
1 Introduction
2 Development of a Low-Budget Monitoring System for Expansion Joints
3 Real Time Data Analyse Using Edge Impulse
4 Condition Monitoring of the Expansion Joint
5 Conclusion and Outlook
References
Correlations Between Localized Pitting Corrosion and Deflection in Reinforced Concrete Beams Subjected to Accelerated Corrosion
1 Introduction
2 Experimental Programme
2.1 Geometry and Reinforcement Layout
2.2 Test Setup and Instrumentation
2.3 Characterization of Rebar Corrosion
2.4 Calculation of Volume Loss Due to Pit Corrosion
3 Results and Discussion
4 Conclusions
References
Strain and Temperature Monitoring in Early-Age Concrete by Distributed Optical Fiber Sensing
1 Introduction
2 Full-Scale Field Tests
2.1 Description of the Specimens
2.2 Sensor Deployment and Monitoring
3 Theory and Method
3.1 Development of Concrete Properties
3.2 Determination of Stresses Based on Temperature and Strain Measurements
4 Results and Discussion
5 Conclusions
References
Investigation of the Impact of Concrete Surface Treatment Methods on the Interfacial Bond Strength
1 Introduction
2 Interface Bond Factors
2.1 Microcracks
2.2 Surface Roughness
2.3 Cleanliness of Substrate Surface
2.4 Free Water on Substrate Surface
2.5 Reduction of Interfacial Bond Strength
3 Experimental Method
4 Results and Discussion
5 Conclusions
References
Piezoresistive Self-compacting Concretes (PSSC) with Carbon Fibers (CF) and Nano-fibers (CNF) for Structural Health Monitoring
1 Introduction
1.1 Self Compacting and Sensing Concrete
1.2 Electrical Properties and Piezoresistivity
2 Materials
3 Methodology and Testing Setup
3.1 Electrical Resistivity Experimental Measurement
3.2 PZR Experimental Measurement
4 Results and Discussion
5 Conclusions
References
Structural Health Monitoring of Reinforced Concrete Beam-Column Joints Using Piezoelectric Transducers
1 Introduction
2 Materials and Test Setup
2.1 Test Setup and Instrumentation
2.2 Instrumented Monitoring
3 Results
3.1 Hysteretic Behavior
3.2 SHM Method Based on the PZT Transducers
3.3 Damage Quantification
4 Conclusions
References
Flexural Damage Evaluation in Fiber Reinforced Concrete Beams Using a PZT-Based Health Monitoring System
1 Introduction
2 Materials and Methods
2.1 Beam Manufacturing
2.2 Set up and Tested Specimens
2.3 Instrumented Monitoring
2.4 Data Analysis
3 Test Results, Voltage Response, and Data Analysis
4 Concluding Remarks
References
Calibration of Multi-physics Models on Weakly Instrumented Structures: Applications to Containment Buildings
1 Introduction
2 Methodology
2.1 Numerical Framework
2.2 Available Data
3 Results
3.1 Thermo-Hygric Modelling
3.2 Mechanical Modelling
3.3 Conclusion
References
Rehabilitation of Underground Garages—A Risk-Based Decision-Making Process
1 Introduction and Background
2 Methodology
2.1 Inspection Data
2.2 Rehabilitation Methods
2.3 Estimation of the Reliability Index
3 Results
4 Summary and Outlook
References
Early-age and Long-term Cracking Analysis in RC Structures (TC 287-CCS)
Early-Age to Long-Term Numerical Simulation of Concrete Members Tested in Adjustable Restraining Frames
1 Introduction
2 Experimental Campaign
2.1 General Description
2.2 Experimental Data
3 Finite Element Model
3.1 Geometry and Mesh
3.2 Material Models
4 Results
5 Conclusions
References
Effect of High Temperature at Early Age and Mineral Additives on Drying Shrinkage of Concrete with Blast Furnace Slag Cement
1 Introduction
2 Experimental Program
2.1 Specimen
2.2 Curing and Drying Conditions
3 Results and Discussion
3.1 Drying Shrinkage and Mass Loss of Mortar Without GGBFS Depending on Curing Temperature
3.2 Drying Shrinkage and Mass Loss Mortar with GGBFS Depending on Curing Temperature
4 Effect of Gypsum and Limestone Powder on Drying Shrinkage of Mortar with GGBFS Exposed to High Temperature Curing
4.1 Experimental Program
4.2 Result and Discussion
5 Conclusion
References
Effect of Member Geometry on the Modification Factor for the Degree of Restraint Before Cracking in Order to Account for the Effect of Cracking
1 Introduction
2 Model Description
3 Calculations
3.1 Scope of Calculations
3.2 Changes of the Modification Factor (Rσcr)
3.3 Changes of Cracks Width
4 Conclusions
References
An Analytical Approach for Calculating Crack Width of RC Members: Pure Shear Method
1 Introduction
2 Assumptions and Derivation of Pure Shear Method
3 Verification of Model Using RC Ties and Beams
3.1 Analysis of RC Ties
3.2 Analysis of RC Beams
4 Conclusions
References
Significance of Thermal Eigenstresses on the Risk of Cracking due to Concrete Hardening with Focus on Ground-Slab Types
1 Introduction
2 Eigenstresses and Typical Member Behavior
3 Nonlinear FE Calculations for Ground Slab-type Members
3.1 Calculation Model and Calculation Steps
3.2 Material Parameters
3.3 Parametric Study
3.4 Results
4 Discussion and Conclusion
References
Calculation of Steel Stresses in Cracked Reinforced Rectangular Concrete Elements Loaded in Bending
1 Introduction
2 Methods for Calculating s in Elements Loaded in Bending
2.1 Assuming a Triangular Compression Zone
2.2 Assuming Linear Interpolation Between Mext and MR
3 Development of the New Semi-analytical Method
4 Results and Validation
5 Conclusions and Discussion
References
Crack-Resistance of 25 Cements Determined by the Ring Shrinkage Test
1 Introduction
2 Materials and Methods
3 Results and Discussion
3.1 Isothermal Calorimetry
3.2 Ring Shrinkage Test
3.3 Discussion
4 Conclusion
References
Cracks Detection During Early-Age Concrete Hydration Using Distributed Fibre Optic Sensing: From Laboratory to Field Applications
1 Introduction
1.1 Distributed Fibre Optic Sensing DFOS
1.2 Early-Age Concrete Cracking
2 Design of DFOS Measurements
2.1 Selection of DFOS Technique
2.2 Design of DFOS Sensors
2.3 Installation
2.4 Thermal Compensation
2.5 Taking Measurements and Post-processing
3 Examples of Applications - Laboratory
3.1 Non-cracked Concrete
3.2 Reinforced Concrete Beams
4 Examples of Applications – In Situ
4.1 Prestressed Concrete Girders
4.2 Railway Bridge
5 Summary
References
Low Viscosity, High Temperature Stable Geopolymer for Crack Injection and Cavity Filling with Optional Increase of Volume and Preload
1 Introduction
2 Development of an Injection Material
2.1 Compressive and Bending Tensile Strength
2.2 Workability
3 Preload
3.1 Preload Via Injection
3.2 Preload Via Grouting
4 Foaming
5 Further Examinations
5.1 High-Temperature
5.2 Use of Recycled Materials
5.3 Crack Injection
6 Conclusion and Outlook
References
Numerical Simulation for Early-age Cracking Mitigation in Durable RC Deck Slab on Multiple Span Steel Box Girder Bridges Considering Thermal and Stepwise Construction Stresses
1 Introduction
2 Structural Details of Fudaigawa Bridge and Strategies for Early-Age Cracking Mitigation
2.1 Structural Details of Fudaigawa Bridge
2.2 Strategies for Early-Age Cracking Mitigation
3 Analysis of Stepwise Construction Stress in Fudaigawa Bridge
3.1 Stepwise Construction Plan for the Concrete Placement of the RC Deck Slab
3.2 Analysis of Generated Stepwise Construction Stresses
4 3D FEM Thermal Stress Simulation and Verifications
4.1 Three Leveled Systematic Thermal Stress FEM Simulation and Verifications
4.2 Results of Early-Age Thermal Stress FEM Simulation
4.3 Influential Factors Affecting Early-Age Thermal Stress
5 The Combined Effect of Thermal and Stepwise Construction Stresses
6 Conclusions
References
Modelling of Moisture Transport in Cracked Concrete by Using RBSM and TNM
1 Introduction
2 Numerical Model
2.1 Rigid-Body Spring Model (RBSM)
2.2 Time-Dependent Constitutive (TDC) Model
2.3 Truss-Network Model (TNM)
2.4 Moisture Transfer Model
2.5 Dynamic Microstructural Change Model
2.6 Liquid Water Uptake Through Cracks
2.7 Numerical Flow
3 Simulation of Anomalous Liquid Water Uptake in Cracked Concrete
3.1 Introduction of Reference Experiment
3.2 Details on Numerical Implementation and Materials Parameters
3.3 Numerical Results
4 Discussion
5 Conclusion
References
The Internal Curing Effect of Pre-saturated Lightweight Aggregate on Cementitious Material
1 Introduction
2 Materials and Test Methods
2.1 Materials and Mixture Proportions
2.2 Test Methods
3 Results and Discussion
3.1 Water Absorption of LWA
3.2 Mechanical Properties
3.3 Autogenous Shrinkage
4 Conclusions
References
Parametric Calculation Tool for Flexural Crack Width in Concrete Slabs Assuming Seismic Damage
1 Introduction
2 Analytical Crack Width Models
2.1 DIN EN 1992-1-1
2.2 DIN EN 1992-1-1-Revision
2.3 ACI 224R-01 R08 - Beam
2.4 ACI 224R-01 R08 - Slab
3 Model Structure
3.1 Subroutine Input
3.2 Subroutine Design of the Cross-section
3.3 Subroutine Calculation of the Crack Widths/Crack Spacing
4 Results
4.1 Verification of the Model
4.2 Parametrised Evaluation of the Crack Models
5 Conclusion
References
Thermal Energy Storage in Cementitious Composites (TC 299-TES)
Analysis of Methods Reducing Early Age Shrinkage of Ultra-light Foam Concrete with Phase Change Material
1 Introduction
2 Overview of Methods Reducing Shrinkage
3 Materials and Methods
4 Results
5 Conclusions
References
An Experimental Study on the Thermo-mechanical Properties of Cement Mortar with Textile Fibers for Building Applications
1 Introduction
2 Materials and Methods
2.1 Materials and Sample Preparation
2.2 Samples Characterization Methodology
3 Evaluation of Characterization Results
3.1 Bulk Density Testing
3.2 Mechanical Characterization
3.3 Thermal Characterization
4 Conclusions
References
Investigation of Combined Electronic and Ionic Thermoelectric Concrete
1 Introduction
2 Experiment
2.1 Sample Fabrication
2.2 MnO2 Powder
2.3 Material Characterization
2.4 Evaluation Setup
2.5 Results and Discussion
3 Conclusions
References
Experimental Characterization and Modelling of Geopolymers and Hybrid Materials for Solar Thermal Energy
1 Introduction
2 Methods
3 Finite Element Modelling
3.1 Basic Equations
4 Results
4.1 Mechanical Properties
4.2 Thermal Properties
4.3 Finite Element Simulations
5 Conclusions
References
Computational Design of Building Envelopes as Thermal Metamaterials
1 Introduction
2 Characterization of NRG-Foams
2.1 Heat Capacity and Enthalpy
2.2 Thermal Conductivity
3 Metamaterial Design for the Optimization of the Building Energy Performance
3.1 Definition of the Building
3.2 Definition of the Local Weather
4 Results
5 Conclusions
References
Smart-Earth Multifunctional Cement Composites for Sustainable Constructions: Thermal and Sensing Characterization
1 Introduction
2 Concept
3 State of Art
4 Smart Earth
4.1 Materials
4.2 Production of the Sample
4.3 Strain-Sensing Behaviour
5 Experimental Methodology
5.1 Setup and Devices
5.2 Tests’ Description
6 Results and Discussion
7 Conclusion
References
Innovative PCM-Enhanced Concrete Tiles for High Performance Buildings
1 Introduction
2 Literature Review
3 Development of an Innovative Tile
3.1 Concrete Mix
3.2 Mechanical Properties
3.3 Thermal Properties
4 Solatrium House
4.1 Project: The Need for Concrete Thermal Storage
4.2 Energy Saving
5 Results and Discussion of Energy Saving
6 Conclusions
References
Phase Change Materials Shape Stabilized in Biochar for Energy Efficiency and Structural Strength Enhancement in Buildings
1 Introduction
2 Materials and Methods
2.1 Materials and Sample Preparation
2.2 Experimental Methodology
3 Results from the Thermal Investigation
4 Results from the Mechanical Investigation
5 Conclusions
References
The Effect of Salt-Impregnation on Thermochemical Properties of a Metakaolin Geopolymer Composite for Thermal Energy Storage
1 Introduction
2 Materials and Methods
2.1 Materials and Sample Preparation
2.2 Methods
3 Results and Discussion
3.1 Phase Assemblage
3.2 Heat Storage Capacity
3.3 Water Sorption
4 Conclusions
References
Hygrothermal Measurement of Heavy Cob Materials
1 Introduction
2 Materials and Methods
2.1 Materials
2.2 Method
3 Results and Discussion
3.1 Thermal Properties
3.2 Hydric Properties
4 Conclusion
References
Novel Cement-Lime Composites with Phase Change Materials (PCM) and Biomass Ash for Energy Efficiency in Architectural Applications
1 Introduction
2 Materials and Paste Compositions
3 Experimental Program
3.1 Mineralogical Characterization Testing Procedure
3.2 Hardened Properties Testing Procedures
4 Experimental Results and Analysis
4.1 Hardened Properties
5 Conclusions
References
Cement Based Materials with PCM and Reduced Graphene Oxide for Thermal Insulation for Buildings
1 Introduction
2 Experimental Section
2.1 Materials
2.2 Paste and Specimen Preparation
2.3 Characterization Techniques
3 Results and Discussion
3.1 Morphological Characterization
3.2 Characterization of Thermal and Electrical Properties
4 Conclusions
References
Hygrothermal Behaviour and Durability of Bio-aggregate Based Building Materials (TC 275-HBD)
Rilem TC 275 HDB – International RRT on MBV Measurement of Vegetal Concrete
1 Introduction
2 Materials and Methods
2.1 Materials
2.2 Methods
3 Discussion and Results of the Round Robin Test
4 Conclusion
References
Rheological Behavior of 3D Printable Bio-Concretes Produced with Rice Husk
1 Introduction
2 Materials and Mixing Protocol
2.1 Materials Under Investigation
2.2 Mixing Procedure
3 Testing Procedure Methods
3.1 Scanning Electron Microscopy (SEM)
3.2 Isothermal Calorimetry
3.3 Rheological Tests
4 Results and Discussion
4.1 Rice Husk Properties
4.2 Isothermal Calorimetry
4.3 Rheological Tests Results
4.4 Printing Tests
5 Conclusions
References
Flax Fabric-Reinforcement Lime Composite as a Strengthening System for Masonry Materials: Study of Adhesion
1 Introduction
2 Experimental Procedure
2.1 Materials
2.2 Mechanical Tests
3 Results and Discussions
3.1 Manual and Mechanical Adhesion Test
3.2 Flexural Test
4 Conclusion
References
Characterisation of Hemp Shiv and its Effect on the Compressive Strength of Hemp Concrete
1 Introduction
2 Experimental Programme
2.1 Materials and Testing Method
2.2 Design of Experiments
2.3 Mix Procedure
3 Results and Discussions
3.1 Hemp Shiv Characterisation
3.2 Hemp Concrete Results
3.3 Isoresponse Charts of the Proposed Models of the Key Variables
4 Conclusion
References
Thermal Study of Hemp Concrete Behavior when Subjected to High Temperatures by X-ray Microtomography
1 Introduction
2 Materials and Methods
2.1 Materials and Mix Design
2.2 3D Thermal Tomography for Construction Materials
2.3 Digital Volume Correlation for 3D Tomographic Scans
2.4 Keyence Optical Microscope and Scanning Electron Microscope (SEM)
3 Results and Discussion
3.1 Representative Elementary Volume Characterization
3.2 Digital Volume Correlation
3.3 Keyence and Scanning Electron Microscope Imaging
4 Conclusion
References
From the Lab Scale to the Construction Site Scale: Properties of Hemp Thermal Insulation
1 Introduction
2 Materials and Methods
2.1 Tested Fibres
2.2 Microscope
2.3 Bulk and Specific Densities
2.4 Thermal Conductivity
2.5 Demonstrator
3 Results and Analysis
3.1 Fibrous Structure
3.2 Thermal Conductivity as a Function of the Density
3.3 Demonstrator: In-Situ Measurement
4 Conclusion
References
Recommend Papers
![Second RILEM International Conference on Concrete and Digital Fabrication: Digital Concrete 2020 [1st ed.]
9783030499150, 9783030499167](https://ebin.pub/img/200x200/second-rilem-international-conference-on-concrete-and-digital-fabrication-digital-concrete-2020-1st-ed-9783030499150-9783030499167.jpg)
File loading please wait...
Citation preview
RILEM Bookseries
Agnieszka Jędrzejewska · Fragkoulis Kanavaris · Miguel Azenha · Farid Benboudjema · Dirk Schlicke Editors
International RILEM Conference on Synergising Expertise towards Sustainability and Robustness of Cement-based Materials and Concrete Structures SynerCrete’23 - Volume 1
International RILEM Conference on Synergising Expertise towards Sustainability and Robustness of Cement-based Materials and Concrete Structures
RILEM Bookseries
Volume 43
RILEM, The International Union of Laboratories and Experts in Construction Materials, Systems and Structures, founded in 1947, is a non-governmental scientific association whose goal is to contribute to progress in the construction sciences, techniques and industries, essentially by means of the communication it fosters between research and practice. RILEM’s focus is on construction materials and their use in building and civil engineering structures, covering all phases of the building process from manufacture to use and recycling of materials. More information on RILEM and its previous publications can be found on www.RILEM.net. Indexed in SCOPUS, Google Scholar and SpringerLink.
Agnieszka J˛edrzejewska · Fragkoulis Kanavaris · Miguel Azenha · Farid Benboudjema · Dirk Schlicke Editors
International RILEM Conference on Synergising Expertise towards Sustainability and Robustness of Cement-based Materials and Concrete Structures SynerCrete’23 - Volume 1
Editors Agnieszka J˛edrzejewska Department of Structural Engineering Silesian University of Technology Gliwice, Poland
Fragkoulis Kanavaris Technical Specialist Services, Materials ARUP London, UK
Miguel Azenha ISISE University of Minho Guimaraes, Portugal
Farid Benboudjema Laboratoire de Mécanique Paris-Saclay ENS Paris-Saclay Gif-sur-Yvette, France
Dirk Schlicke Institute of Structural Concrete Graz University of Technology Graz, Austria
ISSN 2211-0844 ISSN 2211-0852 (electronic) RILEM Bookseries ISBN 978-3-031-33210-4 ISBN 978-3-031-33211-1 (eBook) https://doi.org/10.1007/978-3-031-33211-1 © RILEM 2023 No part of this work may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, microfilming, recording or otherwise, without written permission from the Publisher, with the exception of any material supplied specifically for the purpose of being entered and executed on a computer system, for exclusive use by the purchaser of the work. Permission for use must always be obtained from the owner of the copyright: RILEM. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors, and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, expressed or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. This Springer imprint is published by the registered company Springer Nature Switzerland AG The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Preface
Understanding and controlling climate change originating from human activities presents one of the greatest challenges of our time. The construction industry—and especially concrete, the most widely used building material—has a significant impact on this. The expected new construction of the future, but also the preservation of our existing building and infrastructure stock, therefore already invites us to use the diversity of concrete and concrete-like materials not only with regard to their performance, but also in particular with regard to their potential for sustainable construction. To fully comprehend the intricate behaviour of cement-based materials and concrete structures throughout their lifespan, researchers and practitioners must adopt multidisciplinary approaches. However, interdisciplinary work presents significant challenges as it requires merging distinct methods and approaches while facilitating effective communication among scientific fields with varying terminologies, perspectives, and collaboration practices. Following the legacy of COST Action TU1404 “Towards the next generation of standards for service life of cement-based materials and structures”, which initiated the SynerCrete conference series in 2018, the 2023 Edition continued its effort to assemble relevant stakeholders with the goal of SYNERgising their expertise and expediting the exchange of knowledge among international and interdisciplinary communities. By fostering new developments, this initiative aimed to produce stronger and more sustainable solutions for cement-based materials and conCRETE structures. SynerCrete’23 was held between 14 and 16 June 2023 on Milos in Greece—a mesmerising island in the Cyclades complex. Apart from its natural beauty, Milos is also known for its deposits of natural pozzolans, such as high-quality kaolinite and bentonite. The event was hosted at the Milos Conference Centre–George Eliopoulos located in Adamas, Milos Island’s main port. Situated on the Milos Gulf, the Milos Conference Centre is housed within a refurbished kaolin processing plant, demonstrating an exceptional example of industrial architecture from its time. The event provided a platform for participants to network, exchange knowledge and explore potential collaborations in the field of cement-based materials and structural concrete. The conference has focused on various fields of knowledge, with a non-exhaustive list of topics related to: • alternative concrete formulations for adaptation to climate change, including recent studies on supplementary cementitious materials, alkali-activated materials, nonPortland cement binders; • developments in approaches to modelling of cement-based materials at different scales in time and space; • novel / non-standardised testing techniques; • structural health monitoring and maintenance management; • digital approaches to structural concrete, including BIM, 3D printing and machine learning;
vi
Preface
• resource-responsible building, including the use of recyclable and circular materials and approaches in concrete and concrete structures; • innovative methods for design and construction of concrete structures. These proceedings of SynerCrete’23 feature 240 papers authored by the researchers at all career levels and other professionals—practising engineers and manufacturers—as well as by the invited speakers. A special mention should be made about a significant contribution from RILEM Technical Committees who organised their special topical sessions during the event: • TC 275-HBD: Hygrothermal behaviour and durability of bio-aggregate based building materials; • TC 281-CCC: Carbonation of concrete with supplementary cementitious materials; • TC 283-CAM: Chloride transport in alkali-activated materials; • TC 287-CCS: Early-age and long-term cracking analysis in RC structures; • TC 298-EBD: Test methods to evaluate durability of blended cement pastes against deleterious ions; • TC 299-TES: Thermal energy storage in cementitious composites; • TC 302-CNC: Carbon-based nanomaterials for multifunctional cementitious matrices; • TC MCP: Accelerated mineral carbonation for the production of construction materials. The aforementioned works have been showcased in two volumes of the present proceedings. Volume 1 encompasses the topics related to modelling and testing of cementbased materials as well as to the investigation of the structural behaviour. Volume 2 is devoted to the topics of materials production and reuse and durability of cement-based materials and concrete structures. Organised as a full RILEM event, SynerCrete’23 has been also scientifically cosponsored by the International Federation for Structural Concrete (fib), American Concrete Institute (ACI) and Japan Concrete Institute (JCI). Finally, the words of appreciation go to almost 100 members of the Scientific Committee, whose diligent work ensured the high quality of the contributions to the conference as every single contribution shown in this publication has been checked, commented and approved by a set of at least two independent reviews.
Preface
vii
We hope that the proceedings of SynerCrete’23 will be a valuable resource for researchers and practitioners alike and will contribute to the ongoing efforts towards sustainable and robust concrete materials and structures. Agnieszka J˛edrzejewska SynerCrete’23 Chair Fragkoulis Kanavaris SynerCrete’23 Chair Miguel Azenha Scientific Committee Chair Farid Benboudjema Scientific Committee Chair Dirk Schlicke Scientific Committee Chair
Organisation
Scientific Committee Miguel Azenha (Chair) Farid Benboudjema (Chair) Dirk Schlicke (Chair) Ali Abbas, UK Jean-Luc Adia, France Syed Yasir Alam, France Ouali Amiri, France Sofiane Amziane, France Carmen Andrade, Spain Shingo Asamoto, Japan Umberto Berardi, Canada Susan Bernal Lopez, UK Violeta Bokan Bosiljkov, Slovenia Ruben Paul Borg, Malta Alexandra Bourdot, France Matthieu Briffaut, France Antonio Caggiano, Italy Alejandro Caldentey, Spain Arnaud Castel, Australia Jean Philippe Charron, Canada Fanjie (Sam) Chen, Australia Özlem Cizer, Belgium Florence Collet, France Alexis Courtois, France Aveline Darquennes, France Nele De Belie, Belgium Abobakr Elwakeel, UK Eduardo Fairbairn, Brazil Rui Faria, Portugal Miguel Ferreira, Finland Etore Funchal de Faria, Brazil Ivan Gabrijel, Croatia Dariusz Gawin, Poland José Granja, Portugal Tulio Honorio, France Ole Jensen, Denmark Agnieszka J˛edrzejewska, Poland
ISISE, University of Minho, Portugal ENS Paris-Saclay, France Institute of Structural Concrete, TU Graz, Austria
x
Organisation
Sri Kalyana Rama Jyosyula, India Gintaris Kaklauskas, Lithuania Fragkoulis Kanavaris, UK Antonios Kanellopoulos, UK Anja Klausen, Norway Barbara Klemczak, Poland Eduardus Koenders, Germany Konstantin Kovler, Israel Markus Krüger, Austria Jacek Kwasny, UK Laurie Lacarriere, France Rodrigo Lameiras, Brazil Marco Liebscher, Germany Paulo B. Lourenço, Portugal Lino Maia, Portugal Ippei Maruyama, Japan Enrico Masoero, UK Thomas Matchei, Germany Shishir Mundra, Switzerland Sree Nanukuttan, UK Cécile Oliver-Leblond, France Małgorzata Paj˛ak, Poland Arnaud Perrot, France Nivin Phillip, India Bernhard Pichler, Austria Mário Pimentel, Portugal John Provis, UK Vlastimir Radonjanin, Serbia Ketan Ragalwar, USA Mezgeen Rasol, France Michael Raupach, Germany Kyle Riding, USA Emmanuel Roziere, France Łukasz Sadowski, Poland Małgorzata Safuta, Poland Jacqueline Saliba, France Florence Sanchez, USA Jorge Sánchez Dolado, Spain Javier Sánchez Montero, Spain David Santillan, Spain Suresh Seetharam, Belgium Alain Sellier, France José Sena-Cruz, Portugal Marijana Serdar, Croatia Carlos Serra, Portugal
Organisation
Ruben Snellings, Belgium Mohammed Sonebi, UK Carlos Sousa, Portugal Stéphanie Staquet, Belgium Prannoy Suraneni, USA Vít Šmilauer, Czech Republic Reignard Tan, Norway Luping Tang, Sweden Jean-Michel Torrenti, France Julio Torres Martín, Spain François Toutlemond, France Neven Ukrainczyk, Germany Jens Peder Ulfkjaer, Denmark Luca Valentini, Italy Gideon van Zijl, South Africa Roman Wan-Wender, Belgium William Wilson, Canada Mateusz Wyrzykowski, Switzerland Guang Ye, The Netherlands Ismael Yurtdas, France Kamyab Zandi, USA Hongzhi Zhang, China Xiangming Zhou, UK Mariusz Zych, Poland
Institutional support
xi
xii
Organisation
Gold sponsors
Contents
Keynote Substituting Natural Pozzolan with Artificial Derived from Industrial Perlite Waste for Mortar Production . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Maria Stefanidou
3
Modelling of Cement-Based Materials A Benchmarking of Slag Blended Cement Hydration Models . . . . . . . . . . . . . . . Jack Atallah, Harifidy Ranaivomanana, François Bignonnet, and Stéphanie Bonnet
23
Refining Kinetic Models for SCM Reactivity in Blended Cements . . . . . . . . . . . Ruben Snellings
34
Atomistic Dissolution of β-C2 S Cement Clinker Crystal Surface: Part 1 Molecular Dynamics (MD) Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . K. M. Salah Uddin, Mohammadreza Izadifar, Neven Ukrainczyk, Eduardus Koenders, and Bernhard Middendorf
43
Einstein Explains Water Transport in C-S-H . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Tulio Honorio
54
Is Thermal Pressurization in C-S-H Relevant for Concrete Spalling? . . . . . . . . . Fatima Masara, Tulio Honorio, and Farid Benboudjema
66
Multiscale Modeling of the Dielectric Response of C-S-H . . . . . . . . . . . . . . . . . . Sofiane Ait Hamadouche, Tulio Honorio, Thierry Bore, Farid Benboudjema, Franck Daout, and Eric Vourc’h
76
Quantum Mechanically Informed Kinetic Monte Carlo Models for Hydrogen Diffusion in BCC-Iron . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Gonzalo Álvarez, Alvaro Ridruejo, and Javier Sánchez
88
Numerical Modeling of Water Transfer in Geomaterials: Application to a Concrete Tunnel Subjected to Both Drying and Liquid Overpressure . . . . . Aya Rima, Laurie Lacarrière, Alain Sellier, and Minh-Ngoc Vu
96
xiv
Contents
Replicating the Failure Mechanism of a Real-World Event with the Lattice Discrete Particle Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . G. Lifshitz Sherzer and A. Mitelman
105
Thermomechanical Investigations for the Design of Reinforced Concrete Facings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Eva Maria Dorfmann, Dirk Schlicke, and Ngyuen V. Tue
116
Modelling of the CO2 Uptake by Recycled Concrete Aggregates . . . . . . . . . . . . Philippe Turcry, Bruno Huet, Jonathan Mai-Nhu, Pierre-Yves Mahieux, and Thomas Pernin
125
A Multiscale Multiphysics Platform to Investigate Cement Based Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Julien Sanahuja, François Soleilhet, and Jean-Luc Adia
136
Estimation of Protected Paste Volumes by Dirichlet Tessellation Associated with Point Processes of Air Voids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Kazuya Ohyama and Shin-ichi Igarashi
148
MASKE: Particle-Based Chemo-Mechanical Simulations of Degradation Processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Enrico Masoero
159
Temperature-Dependent Behavior of Mature Cement Paste: Creep Testing and Multiscale Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Eva Binder, Markus Königsberger, Rodrigo Díaz Flores, Herbert A. Mang, Christian Hellmich, and Bernhard L. A. Pichler Development of an Experimental-Numerical Approach to Model Cement Paste Microstructure Using Quantitative Phase Assemblage from XRD and Thermodynamic Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mohammed Krameche, William Wilson, and Arezki Tagnit-Hamou Experimental and Numerical Investigations on Concrete Abrasion of Hydraulic Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Qiong Liu and Min Wu Multi-physics Modelling of Concrete Shrinkage with the Lattice Discrete Particle Model Considering the Volume of Aggregates . . . . . . . . . . . . . Yilin Wang, Roman Wan-Wendner, Giovanni Di Luzio, Jan Vorel, and Jan Belis
171
181
191
203
Contents
xv
Building Information Modelling Enhanced Interoperability between Geotechnical and Structural Engineering for 3D Building Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Haris Felic, Dirk Schlicke, Andreas-Nizar Granitzer, and Franz Tschuchnigg Industry 4.0 Enabled Modular Precast Concrete Components: A Case Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Simon Kosse, Patrick Forman, Jan Stindt, Jannik Hoppe, Markus König, and Peter Mark Integral BIM-Based Planning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . José Alejandro Arellano Pericón and Dirk Schlicke RecycleBIM Approach Towards Integrated Data Management for Circularity: Proof of Concept in a RC Building . . . . . . . . . . . . . . . . . . . . . . . . Artur Kuzminykh, Manuel Parente, Vasco Vieira, José Granja, and Miguel Azenha Towards Standardization of Data for Structural Concrete: Product Data Templates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mohamad El Sibaii, Renan Rocha Ribeiro, Ricardo Dias, José Rui Pinto, José Granja, and Miguel Azenha
217
229
241
252
263
Non-standardised Testing Techniques An Innovative Experiment for Air Pressure Measurements in Crack Models Representative of Real Cracks in Concrete . . . . . . . . . . . . . . . . . . . . . . . . Jean-Louis Tailhan, Giuseppe Rastiello, Jean-Claude Renaud, and Claude Boulay
279
Monitoring of Capillary Pressure Evolution in Young Age Concrete Using High Capacity Tensiometers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Armin Jamali, Joao Mendes, Brabha Nagaratnam, and Michael Lim
289
A Transient Re-touching of Carbon Fiber to Cement Interface Under Single Fiber Pullout Testing with Direct Current Measurement . . . . . . . . . . . . . . Shaofeng Qin and Jishen Qiu
301
Effect of Moisture on the Piezoresistive Properties of Aluminosilicate-Based Building Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Pavel Rovnaník, Ivo Kusák, Pavel Schmid, and Libor Topoláˇr
313
xvi
Contents
Thermal Gradient in Large Concrete Test Bodies: A Macroscale Experimental Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Thierry Houndonougbo, Thierry Chaussadent, Loïc Divet, Joao Custodio, and Jean-François Seignol Developing a New Rapid, Relevant, and Reliable (R3 ) Method for Accelerated Measurement of Carbonation Progress at Gas Overpressure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Benedikt Grimm, Sebastian Münchmeyer, Thomas Kränkel, Christoph Gehlen, and Charlotte Thiel Assessing Cement Matrix Permeability by Neutron Dark Field Imaging . . . . . . Luca Valentini, Gregorio dal Sasso, Fabio Castiglioni, Matteo Busi, Giorgio Ferrari, Maria Chiara Dalconi, Markus Strobl, and Gilberto Artioli Dam Concrete in Situ Creep Tests. Experimental Setup and Results from Six Large Concrete Dams . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Carlos Serra, João Conde Silva, António Lopes Batista, and Nuno Monteiro Azevedo Preliminary Analysis of Non-destructive Test Methods to Evaluate the Self-healing Efficiency on the Construction Site . . . . . . . . . . . . . . . . . . . . . . . Tim Van Mullem, Gerlinde Lefever, Arthur Decuypere, Erik De Vleeschouwer, Yasmina Shields, Laurena De Brabandere, Didier Snoeck, Dimitrios G. Aggelis, and Nele De Belie
325
336
347
355
368
Open-Source EMM-ARM Implementation for Mortars Based on Single-Board Computer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Thomas Russo, Miguel Azenha, and José Granja
380
Understanding the Degradation of Concrete Structures During the Nitrification Process for the Treatment of Wastewater: A Lab Biological Degradation Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Yasmine Werghi, Tony Pons, Marielle Guéguen Minerbe, Marcos Oliviera, Sam Azimi, Vincent Rocher, and Thierry Chaussadent
392
A Cost-Effective Micro-controller Based System for EMM-ARM Tests in Cement Paste . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Renan Rocha Ribeiro, José Granja, Rodrigo Lameiras, and Miguel Azenha
399
Contents
xvii
Analysis of Concrete Transient Thermal Deformation in the Context of Structures Submitted to Various Levels of Temperature and Mechanical Loading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Robin Cartier, Hugo Cagnon, Thierry Vidal, Jean-Michel Torrenti, Alain Sellier, and Jérôme Verdier
413
Semi-circular Bending Test to Evaluate the Post Cracking Behaviour of Fibre Reinforced Concretes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Pedro Paulo Martins de Carvalho and Rodrigo de Melo Lameiras
425
Development of Damage Monitoring Techniques During Fatigue Compression Test on Concrete Specimen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Laura Kerner, Renaud-Pierre Martin, Mezgeen Rasol, Jean-Claude Renaud, and Léopold Denis Comparison of Different Approaches for Quantification of Amorphous Phase in Hydrated Cement Paste by XRD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Antonina Goncharov and Semion Zhutovsky Innovative FWD Testing on Concrete Slabs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Rodrigo Díaz Flores, Valentin Donev, Mehdi Aminbaghai, Luis Zelaya-Lainez, Ronald Blab, Martin Buchta, Lukas Eberhardsteiner, and Bernhard L. A. Pichler Numerical Simulations for the Determination of Chloride Diffusivity in Reinforced Concrete Under Tensile Load . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Amandine Asselin, Jean-Philippe Charron, Clélia Desmettre, Farid Benboudjema, and Cécile Oliver-Leblond
436
449
460
473
Valorisation and Recycling of Non-binder Components of Concrete Deconstructable Concrete Structures Made of Recycled Aggregates from Construction & Demolition Waste: The Experience of the DeConStRAtion Project . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Marco Pepe, Julien Michels, Giulio Zani, Marco Carlo Rampini, and Enzo Martinelli Experimental Investigation of the Influence of Hemp Particles on Hydration Kinetics of Multicomponent Mineral Binder . . . . . . . . . . . . . . . . . . Dmytro Kosiachevskyi, Kamilia Abahri, Anne Daubresse, Evelyne Prat, and Mohend Chaouche
487
500
xviii
Contents
Development of Concrete Mixtures Based Entirely on Construction and Demolition Waste and Assessment of Parameters Influencing the Compressive Strength . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Gurkan Yildirim, Emircan Ozcelikci, Musab Alhawat, and Ashraf Ashour Utilisation of COVID-19 Waste PPE in the Applications of Structural Concrete . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Shannon Kilmartin-Lynch, Rajeev Roychand, Mohammad Saberian, Jie Li, and Fangjie Chen
510
521
Microbial Induced Calcium Carbonate Precipitation (MICP) Treatments for the Reduction of Water Absorption of Recycled Mixed Aggregates . . . . . . . Brigitte Nagy, Johanna Zentner, and Andrea Kustermann
528
Use of Recycled Carbon Fibres in Textile Reinforced Concrete for the Construction Industry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Vanessa Overhage and Thomas Gries
540
Valorization of Sulphidic Mine Tailings as Artificial Aggregate: Implementation in Cement-Based Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Yury Villagran-Zaccardi, Liesbeth Horckmans, and Arne Peys
552
Influence of the Composition of Original Concrete on the Carbonated Recycled Concrete Aggregates Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sandrine Braymand, Sébastien Roux, Hugo Mercado Mendoza, and Florian Schlupp Evaluation of Eco-friendly Concrete Release Agents Based on Bio-Waxes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ojas Chaudhari, Giedrius Zirgulis, Isra Taha, and Dag Tryggö Durability Characterization of Concrete Using Seashell Co-products as Aggregate Replacement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Camille Martin--Cavaillé, Alexandra Bourdot, Nassim Sebaibi, and Rachid Bennacer Production Waste Fibres as a Sustainable Alternative for Strengthening Cementitious Composites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ana Bariˇcevi´c, Katarina Didulica, Branka Mrduljaš, and Antonija Oceli´c
560
570
581
593
Contents
Effect of Elevated Temperatures on Concrete Made with Ash from Wood Biomass and Recycled Polymer Fibers from Waste Rubber . . . . . . . . . . . . . . . . . Marija Jelˇci´c Rukavina, Ivan Gabrijel, Martina Kozlik, Vanja Žvorc, and Nina Štirmer Use of Marble Sludge Waste and Polypropylene Fibers in Developing Eco-friendly Strain Resilient Cementitious Composites . . . . . . . . . . . . . . . . . . . . Souzana Tastani, Paraskevi Christou, Christos Kostas, and Ioannis Ismail
xix
604
612
Test Methods to Evaluate Durability of Blended Cement Pastes Against Deleterious Ions (TC 298-EBD) A Novel Uniaxial Penetration Approach to Investigate External Sulfate Attack on Blended Cement Pastes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Qiao Wang, William Wilson, and Karen Scrivener
625
The Penetration of Chlorides Within Cement Pastes Under an Electric Field . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . F. Reichlin and C. Paglia
635
Chloride Binding in Slag Containing Composite Cements . . . . . . . . . . . . . . . . . . Arezou Babaahmadi and João Figueira
645
√ XCr / t as an Indicator of the Resistance Against Bulk Chloride Diffusion . . . . William Wilson, Fabien Georget, and Karen L. Scrivener
658
Calibration of Tang’s Model for Concentration Dependence of Diffusion in Cementitious Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Neven Ukrainczyk and Eddie Koenders
667
Self-sufficient Reactive Transport Modelling in Cement-Based Materials with Low-Carbon Footprint . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . O. Burkan Isgor and W. Jason Weiss
679
Numerical Simulation of Chloride Ion Ingression in Mortar Incorporating the Effect of ITZ Using an Integrated COMSOL-IPHREEQC Framework . . . . Siventhirarajah Krishnya, Yogarajah Elakneswaran, Yuya Yoda, and Ryoma Kitagaki Resistance of Lime-Natural Pozzolan Mortars in Salt-Laden Environments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Martin Vyšvaˇril, Patrik Bayer, and Karel Dvoˇrák
686
697
xx
Contents
Performance of Migrating Corrosion Inhibitors in Cracked Reinforced Concrete Exposed to Marine Environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Igor Lapiro, Guy Zur, Ela Ofer-Rozovsky, Rami Eid, and Konstantin Kovler Slag or Reacted Binder, Which Dissolves First in Sulphuric Acid? . . . . . . . . . . . Nana Wen, Arne Peys, Tobias Hertel, Vincent Hallet, and Yiannis Pontikes
707
715
Design and Performance Assessing Early-age Dynamic Elastic Modulus in High-Performance Concrete . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Arosha Dabarera, Liang Li, Vishvendra Singh Jamwal, Nisarg Satapara, Xifeng Liu, and Vinh Dao
725
Development of Filling Grout Material for Boulder Ground . . . . . . . . . . . . . . . . . Tomohiko Abe, Egy Crystal Soesilo, and Hiromi Fujiwara
737
Design and Development of Multi-faceted Engineered Concrete . . . . . . . . . . . . . Nabodyuti Das and Prakash Nanthagopalan
753
Assessment of Deviations in Concrete Properties Quantified Under Laboratory Conditions and from the Construction Site . . . . . . . . . . . . . . . . . . . . . David Ov, Juan Mauricio Lozano Valcarcel, Thomas Kränkel, Rolf Breitenbücher, and Christoph Gehlen
764
Performance of Powder Actuated Fasteners as Direct Fastenings in Steel Fibre Reinforced Concrete . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Panagiotis Spyridis, Alhussain Yousef, and Konrad Bergmeister
775
Fiberglass Mesh Reinforced Rendering Mortar: Effect of Fiberglass Reinforcement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Pascale Saba, Tulio Honorio, Xavier Brajer, and Farid Benboudjema
787
Effect of Spatial Variability on the Failure Behaviour of a Reinforced Concrete Shear Wall . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Wafaa Abdallah, Jacqueline Saliba, Sidi Mohammed Elachachi, Zoubir Mehdi Sbartaï, Marwan Sadek, and Fadi Hage Chehade In-situ Casting Method and Durability of Cementitious Materials at Deep Seafloor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Keisuke Takahashi, Tetsu Akitou, and Mari Kobayashi
799
811
Contents
xxi
Axial Strength of Pile Head Embedded with Steel Column: Effect of Reinforcing Bar on Axial Strength . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Kunie Ikeuchi, Tetsu Usami, and Yasuyoshi Miyauchi
818
Probabilistic Assessment of RC Piers Considering Vertical Seismic Excitation Based on Damage Indices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . S. Mahboubi, M. R. Shiravand, G. Shid, and M. Kioumarsi
828
New Conceptions and Constructive Methods for Pumped Storage Hydropower Plants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Eduardo M. R. Fairbairn, Larissa D. F. Santos, Oscar A. M. Reales, Marina B. Farias, Rodolfo G. M. Andrade, and Alfredo Q. Flores Flexural and Shear Performance of Precast Prestressed Composite Beams . . . . Jakub Zaj˛ac, Łukasz Drobiec, Julia Blazy, and Krzysztof Grzyb
840
851
Structural Health Monitoring and Maintenance Management A Novel Service Life Prediction for Reinforced Concrete Infrastructure Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Nabil M. Semaan Monitoring of Reinforced Concrete Structures: Disposal of Low and Intermediate Level Radioactive Waste . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Nuria Rebolledo, Julio E. Torres, Servando Chinchón, Javier Sánchez, Sylvia de Gregorio, Inmaculada López, and Manuel Ordóñez Development of a Low-Budget Monitoring System for Expansion Joints with Real-Time Data Analyses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Lukas Ambros, Natalie Binder, Christian Hölzl, and Markus Vill Correlations Between Localized Pitting Corrosion and Deflection in Reinforced Concrete Beams Subjected to Accelerated Corrosion . . . . . . . . . . David Dackman, Ignasi Fernandez, Carlos G. Berrocal, and Rasmus Rempling Strain and Temperature Monitoring in Early-Age Concrete by Distributed Optical Fiber Sensing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Carlos G. Berrocal, Ignasi Fernandez, Ingemar Löfgren, Erik Nordström, and Rasmus Rempling
867
879
891
902
913
xxii
Contents
Investigation of the Impact of Concrete Surface Treatment Methods on the Interfacial Bond Strength . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . August Jansson, Ignasi Fernandez, Carlos Gil Berrocal, and Rasmus Rempling Piezoresistive Self-compacting Concretes (PSSC) with Carbon Fibers (CF) and Nano-fibers (CNF) for Structural Health Monitoring . . . . . . . . . . . . . . Javier Puentes, Irene Palomar, and Gonzalo Barluenga Structural Health Monitoring of Reinforced Concrete Beam-Column Joints Using Piezoelectric Transducers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Maria Naoum, George Sapidis, Nikos Papadopoulos, Emmanouil Golias, and Constantin Chalioris Flexural Damage Evaluation in Fiber Reinforced Concrete Beams Using a PZT-Based Health Monitoring System . . . . . . . . . . . . . . . . . . . . . . . . . . . George Sapidis, Maria Naoum, Nikos Papadopoulos, and Maristella Voutetaki Calibration of Multi-physics Models on Weakly Instrumented Structures: Applications to Containment Buildings . . . . . . . . . . . . . . . . . . . . . . . . F. Soleilhet, J. Sanahuja, and J.-L. Adia Rehabilitation of Underground Garages—A Risk-Based Decision-Making Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Juan Mauricio Lozano-Valcarcel, Thomas Kränkel, Christoph Gehlen, and Angelika Schießl-Pecka
925
935
945
957
969
981
Early-age and Long-term Cracking Analysis in RC Structures (TC 287-CCS) Early-Age to Long-Term Numerical Simulation of Concrete Members Tested in Adjustable Restraining Frames . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Cláudio Ferreira, Dirk Schlicke, Carlos Sousa, and Miguel Azenha
995
Effect of High Temperature at Early Age and Mineral Additives on Drying Shrinkage of Concrete with Blast Furnace Slag Cement . . . . . . . . . . . 1008 Tatsuya Usui, Shingo Asamoto, and Shintaro Miyamoto Effect of Member Geometry on the Modification Factor for the Degree of Restraint Before Cracking in Order to Account for the Effect of Cracking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1017 Mariusz Zych
Contents
xxiii
An Analytical Approach for Calculating Crack Width of RC Members: Pure Shear Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1029 Karolis Sakalauskas and Gintaris Kaklauskas Significance of Thermal Eigenstresses on the Risk of Cracking due to Concrete Hardening with Focus on Ground-Slab Types . . . . . . . . . . . . . . . . . . 1038 Christina Krenn and Dirk Schlicke Calculation of Steel Stresses in Cracked Reinforced Rectangular Concrete Elements Loaded in Bending . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1047 I. Anton van der Esch, Rob Wolfs, and Simon Wijte Crack-Resistance of 25 Cements Determined by the Ring Shrinkage Test . . . . . 1059 V. Šmilauer, P. Reiterman, and B. Slánský Cracks Detection During Early-Age Concrete Hydration Using Distributed Fibre Optic Sensing: From Laboratory to Field Applications . . . . . . 1069 Rafał Sie´nko, Łukasz Bednarski, Tomasz Howiacki, and Kamil Badura Low Viscosity, High Temperature Stable Geopolymer for Crack Injection and Cavity Filling with Optional Increase of Volume and Preload . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1081 Hendrik Morgenstern and Michael Raupach Numerical Simulation for Early-age Cracking Mitigation in Durable RC Deck Slab on Multiple Span Steel Box Girder Bridges Considering Thermal and Stepwise Construction Stresses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1093 Thanh Ngoc Phan, Akira Hosoda, Yoichiro Tsujita, and Ayana Shirakawa Modelling of Moisture Transport in Cracked Concrete by Using RBSM and TNM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1106 Puttipong Srimook and Ippei Maruyama The Internal Curing Effect of Pre-saturated Lightweight Aggregate on Cementitious Material . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1118 Runqi Hao, Hannawi Kinda, and Darquennes Aveline Parametric Calculation Tool for Flexural Crack Width in Concrete Slabs Assuming Seismic Damage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1128 Alhussain Yousef and Panagiotis Spyridis
xxiv
Contents
Thermal Energy Storage in Cementitious Composites (TC 299-TES) Analysis of Methods Reducing Early Age Shrinkage of Ultra-light Foam Concrete with Phase Change Material . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1143 Barbara Klemczak, Jacek Gołaszewski, Grzegorz Cygan, Aneta Smolana, and Małgorzata Gołaszewska An Experimental Study on the Thermo-mechanical Properties of Cement Mortar with Textile Fibers for Building Applications . . . . . . . . . . . . . 1153 Rabeb Ayed, Emiliano Borri, Gemma Gasa, Salwa Bouadila, and Luisa F. Cabeza Investigation of Combined Electronic and Ionic Thermoelectric Concrete . . . . . 1163 Mostafa Yossef, Seyedabolfazl Mousavihashemi, Tanja Kallio, and Jari Puttonen Experimental Characterization and Modelling of Geopolymers and Hybrid Materials for Solar Thermal Energy . . . . . . . . . . . . . . . . . . . . . . . . . . . 1176 Irene Ramón-Álvarez, Sergio Sánchez-Delgado, Ignacio Peralta, Antonio Caggiano, and Manuel Torres-Carrasco Computational Design of Building Envelopes as Thermal Metamaterials . . . . . 1189 Víctor D. Fachinotti, Juan C. Álvarez Hostos, Ignacio Peralta, and Antonio Caggiano Smart-Earth Multifunctional Cement Composites for Sustainable Constructions: Thermal and Sensing Characterization . . . . . . . . . . . . . . . . . . . . . . 1199 Andrea Meoni, Claudia Fabiani, Antonella D’Alessandro, Anna Laura Pisello, and Filippo Ubertini Innovative PCM-Enhanced Concrete Tiles for High Performance Buildings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1209 Umberto Berardi Phase Change Materials Shape Stabilized in Biochar for Energy Efficiency and Structural Strength Enhancement in Buildings . . . . . . . . . . . . . . . 1222 Carolina Santini, Claudia Fabiani, Antonella D’Alessandro, and Anna Laura Pisello The Effect of Salt-Impregnation on Thermochemical Properties of a Metakaolin Geopolymer Composite for Thermal Energy Storage . . . . . . . . 1232 Lorena Skevi, Xinyuan Ke, Jonathon Elvins, and Yulong Ding Hygrothermal Measurement of Heavy Cob Materials . . . . . . . . . . . . . . . . . . . . . . 1243 Ouellet-Plamondon Claudiane and Kabore Aguerata
Contents
xxv
Novel Cement-Lime Composites with Phase Change Materials (PCM) and Biomass Ash for Energy Efficiency in Architectural Applications . . . . . . . . 1253 C. Guardia, A. Guerrero, and G. Barluenga Cement Based Materials with PCM and Reduced Graphene Oxide for Thermal Insulation for Buildings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1264 Edurne Erkizia, Christina Strunz, Jean-Luc Dauvergne, Guido Goracci, Ignacio Peralta, Ángel Serrano, Amaya Ortega, Beatriz Alonso, Francesca Zanoni, Michael Düngfelder, Jorge S. Dolado, Juan Jose Gaitero, Christoph Mankel, and Eduardus Koenders Hygrothermal Behaviour and Durability of Bio-aggregate Based Building Materials (TC 275-HBD) Rilem TC 275 HDB – International RRT on MBV Measurement of Vegetal Concrete . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1279 Florence Collet, Stijn Mertens, Paulina Faria, Sofiane Amziane, Thibaut Colinart, Camille Magniont, Sylvie Prétot, Romildo Dias Toledo Filho, and Méryl Lagouin Rheological Behavior of 3D Printable Bio-Concretes Produced with Rice Husk . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1288 Matheus P. Tinoco, Oscar A. M. Reales, and Romildo D. Toledo Filho Flax Fabric-Reinforcement Lime Composite as a Strengthening System for Masonry Materials: Study of Adhesion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1297 Ali Rakhsh Mahpour, Josep Claramunt, Mònica Ardanuy, and Joan Ramon Rosell Characterisation of Hemp Shiv and its Effect on the Compressive Strength of Hemp Concrete . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1307 Ahmed Abdalqader, Tahreer Fayyad, Mohammed Sonebi, and Su Taylor Thermal Study of Hemp Concrete Behavior when Subjected to High Temperatures by X-ray Microtomography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1318 Chady El Hachem, Joseph Moussa, and Kamilia Abahri From the Lab Scale to the Construction Site Scale: Properties of Hemp Thermal Insulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1330 Lily Deborde, Christophe Lanos, Florence Collet, Jules Delsalle, and Valentin Colson
RILEM Publications
The following list is presenting the global offer of RILEM Publications, sorted by series. Each publication is available in printed version and/or in online version.
RILEM Proceedings (PRO) PRO 1: Durability of High Performance Concrete (ISBN: 2-912143-03-9; e-ISBN: 2-351580-12-5; e-ISBN: 2351580125); Ed. H. Sommer PRO 2: Chloride Penetration into Concrete (ISBN: 2-912143-00-04; e-ISBN: 2912143454); Eds. L.-O. Nilsson and J.-P. Ollivier PRO 3: Evaluation and Strengthening of Existing Masonry Structures (ISBN: 2-91214302-0; e-ISBN: 2351580141); Eds. L. Binda and C. Modena PRO 4: Concrete: From Material to Structure (ISBN: 2-912143-04-7; e-ISBN: 2351580206); Eds. J.-P. Bournazel and Y. Malier PRO 5: The Role of Admixtures in High Performance Concrete (ISBN: 2-912143-05-5; e-ISBN: 2351580214); Eds. J. G. Cabrera and R. Rivera-Villarreal PRO 6: High Performance Fiber Reinforced Cement Composites - HPFRCC 3 (ISBN: 2-912143-06-3; e-ISBN: 2351580222); Eds. H. W. Reinhardt and A. E. Naaman PRO 7: 1st International RILEM Symposium on Self-Compacting Concrete (ISBN: 2-912143-09-8; e-ISBN: 2912143721); Eds. Å. Skarendahl and Ö. Petersson PRO 8: International RILEM Symposium on Timber Engineering (ISBN: 2-91214310-1; e-ISBN: 2351580230); Ed. L. Boström PRO 9: 2nd International RILEM Symposium on Adhesion between Polymers and Concrete ISAP ’99 (ISBN: 2-912143-11-X; e-ISBN: 2351580249); Eds. Y. Ohama and M. Puterman PRO 10: 3rd International RILEM Symposium on Durability of Building and Construction Sealants (ISBN: 2-912143-13-6; e-ISBN: 2351580257); Ed. A. T. Wolf PRO 11: 4th International RILEM Conference on Reflective Cracking in Pavements (ISBN: 2-912143-14-4; e-ISBN: 2351580265); Eds. A. O. Abd El Halim, D. A. Taylor and El H. H. Mohamed PRO 12: International RILEM Workshop on Historic Mortars: Characteristics and Tests (ISBN: 2-912143-15-2; e-ISBN: 2351580273); Eds. P. Bartos, C. Groot and J. J. Hughes PRO 13: 2nd International RILEM Symposium on Hydration and Setting (ISBN: 2-912143-16-0; e-ISBN: 2351580281); Ed. A. Nonat
xxviii
RILEM Publications
PRO 14: Integrated Life-Cycle Design of Materials and Structures - ILCDES 2000 (ISBN: 951-758-408-3; e-ISBN: 235158029X); (ISSN: 0356-9403); Ed. S. Sarja PRO 15: Fifth RILEM Symposium on Fibre-Reinforced Concretes (FRC) - BEFIB’2000 (ISBN: 2-912143-18-7; e-ISBN: 291214373X); Eds. P. Rossi and G. Chanvillard PRO 16: Life Prediction and Management of Concrete Structures (ISBN: 2-912143-195; e-ISBN: 2351580303); Ed. D. Naus PRO 17: Shrinkage of Concrete – Shrinkage 2000 (ISBN: 2-912143-20-9; e-ISBN: 2351580311); Eds. V. Baroghel-Bouny and P.-C. Aïtcin PRO 18: Measurement and Interpretation of the On-Site Corrosion Rate (ISBN: 2-912143-21-7; e-ISBN: 235158032X); Eds. C. Andrade, C. Alonso, J. Fullea, J. Polimon and J. Rodriguez PRO 19: Testing and Modelling the Chloride Ingress into Concrete (ISBN: 2-91214322-5; e-ISBN: 2351580338); Eds. C. Andrade and J. Kropp PRO 20: 1st International RILEM Workshop on Microbial Impacts on Building Materials (CD 02) (e-ISBN 978-2-35158-013-4); Ed. M. Ribas Silva PRO 21: International RILEM Symposium on Connections between Steel and Concrete (ISBN: 2-912143-25-X; e-ISBN: 2351580346); Ed. R. Eligehausen PRO 22: International RILEM Symposium on Joints in Timber Structures (ISBN: 2-912143-28-4; e-ISBN: 2351580354); Eds. S. Aicher and H.-W. Reinhardt PRO 23: International RILEM Conference on Early Age Cracking in Cementitious Systems (ISBN: 2-912143-29-2; e-ISBN: 2351580362); Eds. K. Kovler and A. Bentur PRO 24: 2nd International RILEM Workshop on Frost Resistance of Concrete (ISBN: 2-912143-30-6; e-ISBN: 2351580370); Eds. M. J. Setzer, R. Auberg and H.-J. Keck PRO 25: International RILEM Workshop on Frost Damage in Concrete (ISBN: 2-912143-31-4; e-ISBN: 2351580389); Eds. D. J. Janssen, M. J. Setzer and M. B. Snyder PRO 26: International RILEM Workshop on On-Site Control and Evaluation of Masonry Structures (ISBN: 2-912143-34-9; e-ISBN: 2351580141); Eds. L. Binda and R. C. de Vekey PRO 27: International RILEM Symposium on Building Joint Sealants (CD03; e-ISBN: 235158015X); Ed. A. T. Wolf PRO 28: 6th International RILEM Symposium on Performance Testing and Evaluation of Bituminous Materials - PTEBM’03 (ISBN: 2-912143-35-7; e-ISBN: 978-2-91214377-8); Ed. M. N. Partl PRO 29: 2nd International RILEM Workshop on Life Prediction and Ageing Management of Concrete Structures (ISBN: 2-912143-36-5; e-ISBN: 2912143780); Ed. D. J. Naus
RILEM Publications
xxix
PRO 30: 4th International RILEM Workshop on High Performance Fiber Reinforced Cement Composites - HPFRCC 4 (ISBN: 2-912143-37-3; e-ISBN: 2912143799); Eds. A. E. Naaman and H. W. Reinhardt PRO 31: International RILEM Workshop on Test and Design Methods for Steel Fibre Reinforced Concrete: Background and Experiences (ISBN: 2-912143-38-1; e-ISBN: 2351580168); Eds. B. Schnütgen and L. Vandewalle PRO 32: International Conference on Advances in Concrete and Structures 2 vol. (ISBN (set): 2-912143-41-1; e-ISBN: 2351580176); Eds. Ying-shu Yuan, Surendra P. Shah and Heng-lin Lü PRO 33: 3rd International Symposium on Self-Compacting Concrete (ISBN: 2-91214342-X; e-ISBN: 2912143713); Eds. Ó. Wallevik and I. Níelsson PRO 34: International RILEM Conference on Microbial Impact on Building Materials (ISBN: 2-912143-43-8; e-ISBN: 2351580184); Ed. M. Ribas Silva PRO 35: International RILEM TC 186-ISA on Internal Sulfate Attack and Delayed Ettringite Formation (ISBN: 2-912143-44-6; e-ISBN: 2912143802); Eds. K. Scrivener and J. Skalny PRO 36: International RILEM Symposium on Concrete Science and Engineering – A Tribute to Arnon Bentur (ISBN: 2-912143-46-2; e-ISBN: 2912143586); Eds. K. Kovler, J. Marchand, S. Mindess and J. Weiss PRO 37: 5th International RILEM Conference on Cracking in Pavements – Mitigation, Risk Assessment and Prevention (ISBN: 2-912143-47-0; e-ISBN: 2912143764); Eds. C. Petit, I. Al-Qadi and A. Millien PRO 38: 3rd International RILEM Workshop on Testing and Modelling the Chloride Ingress into Concrete (ISBN: 2-912143-48-9; e-ISBN: 2912143578); Eds. C. Andrade and J. Kropp PRO 39: 6th International RILEM Symposium on Fibre-Reinforced Concretes - BEFIB 2004 (ISBN: 2-912143-51-9; e-ISBN: 2912143748); Eds. M. Di Prisco, R. Felicetti and G. A. Plizzari PRO 40: International RILEM Conference on the Use of Recycled Materials in Buildings and Structures (ISBN: 2-912143-52-7; e-ISBN: 2912143756); Eds. E. Vázquez, Ch. F. Hendriks and G. M. T. Janssen PRO 41: RILEM International Symposium on Environment-Conscious Materials and Systems for Sustainable Development (ISBN: 2-912143-55-1; e-ISBN: 2912143640); Eds. N. Kashino and Y. Ohama PRO 42: SCC’2005 - China: 1st International Symposium on Design, Performance and Use of Self-Consolidating Concrete (ISBN: 2-912143-61-6; e-ISBN: 2912143624); Eds. Zhiwu Yu, Caijun Shi, Kamal Henri Khayat and Youjun Xie PRO 43: International RILEM Workshop on Bonded Concrete Overlays (e-ISBN: 2-912143-83-7); Eds. J. L. Granju and J. Silfwerbrand
xxx
RILEM Publications
PRO 44: 2nd International RILEM Workshop on Microbial Impacts on Building Materials (CD11) (e-ISBN: 2-912143-84-5); Ed. M. Ribas Silva PRO 45: 2nd International Symposium on Nanotechnology in Construction, Bilbao (ISBN: 2-912143-87-X; e-ISBN: 2912143888); Eds. Peter J. M. Bartos, Yolanda de Miguel and Antonio Porro PRO 46: ConcreteLife’06 - International RILEM-JCI Seminar on Concrete Durability and Service Life Planning: Curing, Crack Control, Performance in Harsh Environments (ISBN: 2-912143-89-6; e-ISBN: 291214390X); Ed. K. Kovler PRO 47: International RILEM Workshop on Performance Based Evaluation and Indicators for Concrete Durability (ISBN: 978-2-912143-95-2; e-ISBN: 9782912143969); Eds. V. Baroghel-Bouny, C. Andrade, R. Torrent and K. Scrivener PRO 48: 1st International RILEM Symposium on Advances in Concrete through Science and Engineering (e-ISBN: 2-912143-92-6); Eds. J. Weiss, K. Kovler, J. Marchand and S. Mindess PRO 49: International RILEM Workshop on High Performance Fiber Reinforced Cementitious Composites in Structural Applications (ISBN: 2-912143-93-4; e-ISBN: 2912143942); Eds. G. Fischer and V. C. Li PRO 50: 1st International RILEM Symposium on Textile Reinforced Concrete (ISBN: 2-912143-97-7; e-ISBN: 2351580087); Eds. Josef Hegger, Wolfgang Brameshuber and Norbert Will PRO 51: 2nd International Symposium on Advances in Concrete through Science and Engineering (ISBN: 2-35158-003-6; e-ISBN: 2-35158-002-8); Eds. J. Marchand, B. Bissonnette, R. Gagné, M. Jolin and F. Paradis PRO 52: Volume Changes of Hardening Concrete: Testing and Mitigation (ISBN: 2-35158-004-4; e-ISBN: 2-35158-005-2); Eds. O. M. Jensen, P. Lura and K. Kovler PRO 53: High Performance Fiber Reinforced Cement Composites - HPFRCC5 (ISBN: 978-2-35158-046-2; e-ISBN: 978-2-35158-089-9); Eds. H. W. Reinhardt and A. E. Naaman PRO 54: 5th International RILEM Symposium on Self-Compacting Concrete (ISBN: 978-2-35158-047-9; e-ISBN: 978-2-35158-088-2); Eds. G. De Schutter and V. Boel PRO 55: International RILEM Symposium Photocatalysis, Environment and Construction Materials (ISBN: 978-2-35158-056-1; e-ISBN: 978-2-35158-057-8); Eds. P. Baglioni and L. Cassar PRO 56: International RILEM Workshop on Integral Service Life Modelling of Concrete Structures (ISBN 978-2-35158-058-5; e-ISBN: 978-2-35158-090-5); Eds. R. M. Ferreira, J. Gulikers and C. Andrade PRO 57: RILEM Workshop on Performance of cement-based materials in aggressive aqueous environments (e-ISBN: 978-2-35158-059-2); Ed. N. De Belie
RILEM Publications
xxxi
PRO 58: International RILEM Symposium on Concrete Modelling - CONMOD’08 (ISBN: 978-2-35158-060-8; e-ISBN: 978-2-35158-076-9); Eds. E. Schlangen and G. De Schutter PRO 59: International RILEM Conference on On Site Assessment of Concrete, Masonry and Timber Structures - SACoMaTiS 2008 (ISBN set: 978-2-35158-061-5; e-ISBN: 978-2-35158-075-2); Eds. L. Binda, M. di Prisco and R. Felicetti PRO 60: Seventh RILEM International Symposium on Fibre Reinforced Concrete: Design and Applications - BEFIB 2008 (ISBN: 978-2-35158-064-6; e-ISBN: 978-235158-086-8); Ed. R. Gettu PRO 61: 1st International Conference on Microstructure Related Durability of Cementitious Composites 2 vol., (ISBN: 978-2-35158-065-3; e-ISBN: 978-2-35158-084-4); Eds. W. Sun, K. van Breugel, C. Miao, G. Ye and H. Chen PRO 62: NSF/ RILEM Workshop: In-situ Evaluation of Historic Wood and Masonry Structures (e-ISBN: 978-2-35158-068-4); Eds. B. Kasal, R. Anthony and M. Drdácký PRO 63: Concrete in Aggressive Aqueous Environments: Performance, Testing and Modelling, 2 vol., (ISBN: 978-2-35158-071-4; e-ISBN: 978-2-35158-082-0); Eds. M. G. Alexander and A. Bertron PRO 64: Long Term Performance of Cementitious Barriers and Re inforced Concrete in Nuclear Power Plants and Waste Management - NUCPERF 2009 (ISBN: 978-2-35158-072-1; e-ISBN: 978-2-35158-087-5); Eds. V. L’Hostis, R. Gens and C. Gallé PRO 65: Design Performance and Use of Self-consolidating Concrete - SCC’2009 (ISBN: 978-2-35158-073-8; e-ISBN: 978-2-35158-093-6); Eds. C. Shi, Z. Yu, K. H. Khayat and P. Yan PRO 66: 2nd International RILEM Workshop on Concrete Durability and Service Life Planning - ConcreteLife’09 (ISBN: 978-2-35158-074-5; ISBN: 978-2-35158-074-5); Ed. K. Kovler PRO 67: Repairs Mortars for Historic Masonry (e-ISBN: 978-2-35158-083-7); Ed. C. Groot PRO 68: Proceedings of the 3rd International RILEM Symposium on ‘Rheology of Cement Suspensions such as Fresh Concrete (ISBN 978-2-35158-091-2; e-ISBN: 978-2-35158-092-9); Eds. O. H. Wallevik, S. Kubens and S. Oesterheld PRO 69: 3rd International PhD Student Workshop on ‘Modelling the Durability of Reinforced Concrete (ISBN: 978-2-35158-095-0); Eds. R. M. Ferreira, J. Gulikers and C. Andrade PRO 70: 2nd International Conference on ‘Service Life Design for Infrastructure’ (ISBN set: 978-2-35158-096-7, e-ISBN: 978-2-35158-097-4); Eds. K. van Breugel, G. Ye and Y. Yuan
xxxii
RILEM Publications
PRO 71: Advances in Civil Engineering Materials - The 50-year Teaching Anniversary of Prof. Sun Wei’ (ISBN: 978-2-35158-098-1; e-ISBN: 978-2-35158-099-8); Eds. C. Miao, G. Ye and H. Chen PRO 72: First International Conference on ‘Advances in Chemically-Activated Materials – CAM’2010’ (2010), 264 pp, ISBN: 978-2-35158-101-8; e-ISBN: 978-2-35158115-5, Eds. Caijun Shi and Xiaodong Shen PRO 73: 2nd International Conference on ‘Waste Engineering and Management ICWEM 2010’ (2010), 894 pp, ISBN: 978-2-35158-102-5; e-ISBN: 978-2-35158-103-2, Eds. J. Zh. Xiao, Y. Zhang, M. S. Cheung and R. Chu PRO 74: International RILEM Conference on ‘Use of Superabsorbent Polymers and Other New Addditives in Concrete’ (2010) 374 pp., ISBN: 978-2-35158-104-9; e-ISBN: 978-2-35158-105-6; Eds. O. M. Jensen, M. T. Hasholt and S. Laustsen PRO 75: International Conference on ‘Material Science - 2nd ICTRC - Textile Reinforced Concrete - Theme 1’ (2010) 436 pp., ISBN: 978-2-35158-106-3; e-ISBN: 978-2-35158-107-0; Ed. W. Brameshuber PRO 76: International Conference on ‘Material Science - HetMat - Modelling of Heterogeneous Materials - Theme 2’ (2010) 255 pp., ISBN: 978-2-35158-108-7; e-ISBN: 978-2-35158-109-4; Ed. W. Brameshuber PRO 77: International Conference on ‘Material Science - AdIPoC - Additions Improving Properties of Concrete - Theme 3’ (2010) 459 pp., ISBN: 978-2-35158-110-0; e-ISBN: 978-2-35158-111-7; Ed. W. Brameshuber PRO 78: 2nd Historic Mortars Conference and RILEM TC 203-RHM Final Workshop – HMC2010 (2010) 1416 pp., e-ISBN: 978-2-35158-112-4; Eds. J. Válek, C. Groot and J. J. Hughes PRO 79: International RILEM Conference on Advances in Construction Materials Through Science and Engineering (2011) 213 pp., ISBN: 978-2-35158-116-2, e-ISBN: 978-2-35158-117-9; Eds. Christopher Leung and K. T. Wan PRO 80: 2nd International RILEM Conference on Concrete Spalling due to Fire Exposure (2011) 453 pp., ISBN: 978-2-35158-118-6, e-ISBN: 978-2-35158-119-3; Eds. E. A. B. Koenders and F. Dehn PRO 81: 2nd International RILEM Conference on Strain Hardening Cementitious Composites (SHCC2-Rio) (2011) 451 pp., ISBN: 978-2-35158-120-9, e-ISBN: 9782-35158-121-6; Eds. R. D. Toledo Filho, F. A. Silva, E. A. B. Koenders and E. M. R. Fairbairn PRO 82: 2nd International RILEM Conference on Progress of Recycling in the Built Environment (2011) 507 pp., e-ISBN: 978-2-35158-122-3; Eds. V. M. John, E. Vazquez, S. C. Angulo and C. Ulsen
RILEM Publications
xxxiii
PRO 83: 2nd International Conference on Microstructural-related Durability of Cementitious Composites (2012) 250 pp., ISBN: 978-2-35158-129-2; e-ISBN: 978-2-35158123-0; Eds. G. Ye, K. van Breugel, W. Sun and C. Miao PRO 84: CONSEC13 - Seventh International Conference on Concrete under Severe Conditions – Environment and Loading (2013) 1930 pp., ISBN: 978-2-35158-124-7; e-ISBN: 978-2-35158-134-6; Eds. Z. J. Li, W. Sun, C. W. Miao, K. Sakai, O. E. Gjorv and N. Banthia PRO 85: RILEM-JCI International Workshop on Crack Control of Mass Concrete and Related issues concerning Early-Age of Concrete Structures – ConCrack 3 – Control of Cracking in Concrete Structures 3 (2012) 237 pp., ISBN: 978-2-35158-125-4; e-ISBN: 978-2-35158-126-1; Eds. F. Toutlemonde and J.-M. Torrenti PRO 86: International Symposium on Life Cycle Assessment and Construction (2012) 414 pp., ISBN: 978-2-35158-127-8, e-ISBN: 978-2-35158-128-5; Eds. A. Ventura and C. de la Roche PRO 87: UHPFRC 2013 – RILEM-fib-AFGC International Symposium on Ultra-High Performance Fibre-Reinforced Concrete (2013), ISBN: 978-2-35158-130-8, e-ISBN: 978-2-35158-131-5; Eds. F. Toutlemonde PRO 88: 8th RILEM International Symposium on Fibre Reinforced Concrete (2012) 344 pp., ISBN: 978-2-35158-132-2, e-ISBN: 978-2-35158-133-9; Eds. Joaquim A. O. Barros PRO 89: RILEM International workshop on performance-based specification and control of concrete durability (2014) 678 pp, ISBN: 978-2-35158-135-3, e-ISBN: 978-2-35158-136-0; Eds. D. Bjegovi´c, H. Beushausen and M. Serdar PRO 90: 7th RILEM International Conference on Self-Compacting Concrete and of the 1st RILEM International Conference on Rheology and Processing of Construction Materials (2013) 396 pp, ISBN: 978-2-35158-137-7, e-ISBN: 978-2-35158-138-4; Eds. Nicolas Roussel and Hela Bessaies-Bey PRO 91: CONMOD 2014 - RILEM International Symposium on Concrete Modelling (2014), ISBN: 978-2-35158-139-1; e-ISBN: 978-2-35158-140-7; Eds. Kefei Li, Peiyu Yan and Rongwei Yang PRO 92: CAM 2014 - 2nd International Conference on advances in chemically-activated materials (2014) 392 pp., ISBN: 978-2-35158-141-4; e-ISBN: 978-2-35158-142-1; Eds. Caijun Shi and Xiadong Shen PRO 93: SCC 2014 - 3rd International Symposium on Design, Performance and Use of Self-Consolidating Concrete (2014) 438 pp., ISBN: 978-2-35158-143-8; e-ISBN: 978-2-35158-144-5; Eds. Caijun Shi, Zhihua Ou and Kamal H. Khayat PRO 94 (online version): HPFRCC-7 - 7th RILEM conference on High performance fiber reinforced cement composites (2015), e-ISBN: 978-2-35158-146-9; Eds. H. W. Reinhardt, G. J. Parra-Montesinos and H. Garrecht
xxxiv
RILEM Publications
PRO 95: International RILEM Conference on Application of superabsorbent polymers and other new admixtures in concrete construction (2014), ISBN: 978-2-35158-147-6; e-ISBN: 978-2-35158-148-3; Eds. Viktor Mechtcherine and Christof Schroefl PRO 96 (online version): XIII DBMC: XIII International Conference on Durability of Building Materials and Components (2015), e-ISBN: 978-2-35158-149-0; Eds. M. Quattrone and V. M. John PRO 97: SHCC3 – 3rd International RILEM Conference on Strain Hardening Cementitious Composites (2014), ISBN: 978-2-35158-150-6; e-ISBN: 978-2-35158-151-3; Eds. E. Schlangen, M. G. Sierra Beltran, M. Lukovic and G. Ye PRO 98: FERRO-11 – 11th International Symposium on Ferrocement and 3rd ICTRC - International Conference on Textile Reinforced Concrete (2015), ISBN: 978-2-35158152-0; e-ISBN: 978-2-35158-153-7; Ed. W. Brameshuber PRO 99 (online version): ICBBM 2015 - 1st International Conference on Bio-Based Building Materials (2015), e-ISBN: 978-2-35158-154-4; Eds. S. Amziane and M. Sonebi PRO 100: SCC16 - RILEM Self-Consolidating Concrete Conference (2016), ISBN: 978-2-35158-156-8; e-ISBN: 978-2-35158-157-5; Ed. Kamal H. Kayat PRO 101 (online version): III Progress of Recycling in the Built Environment (2015), e-ISBN: 978-2-35158-158-2; Eds. I. Martins, C. Ulsen and S. C. Angulo PRO 102 (online version): RILEM Conference on Microorganisms-Cementitious Materials Interactions (2016), e-ISBN: 978-2-35158-160-5; Eds. Alexandra Bertron, Henk Jonkers and Virginie Wiktor PRO 103 (online version): ACESC’16 - Advances in Civil Engineering and Sustainable Construction (2016), e-ISBN: 978-2-35158-161-2; Eds. T. Ch. Madhavi, G. Prabhakar, Santhosh Ram and P. M. Rameshwaran PRO 104 (online version): SSCS’2015 - Numerical Modeling - Strategies for Sustainable Concrete Structures (2015), e-ISBN: 978-2-35158-162-9 PRO 105: 1st International Conference on UHPC Materials and Structures (2016), ISBN: 978-2-35158-164-3, e-ISBN: 978-2-35158-165-0 PRO 106: AFGC-ACI-fib-RILEM International Conference on UltraHigh-Performance Fibre-Reinforced Concrete – UHPFRC 2017 (2017), ISBN: 9782-35158-166-7, e-ISBN: 978-2-35158-167-4; Eds. François Toutlemonde and Jacques Resplendino PRO 107 (online version): XIV DBMC – 14th International Conference on Durability of Building Materials and Components (2017), e-ISBN: 978-2-35158-159-9; Eds. Geert De Schutter, Nele De Belie, Arnold Janssens and Nathan Van Den Bossche PRO 108: MSSCE 2016 -Innovation of Teaching in Materials and Structures (2016), ISBN: 978-2-35158-178-0, e-ISBN: 978-2-35158-179-7; Ed. Per Goltermann
RILEM Publications
xxxv
PRO 109 (2 volumes): MSSCE 2016 - Service Life of Cement-Based Materials and Structures (2016), ISBN Vol. 1: 978-2-35158-170-4, Vol. 2: 978-2-35158-171-4, Set Vol. 1&2: 978-2-35158-172-8, e-ISBN : 978-2-35158-173-5; Eds. Miguel Azenha, Ivan Gabrijel, Dirk Schlicke, Terje Kanstad and Ole Mejlhede Jensen PRO 110: MSSCE 2016 - Historical Masonry (2016), ISBN: 978-2-35158-178-0, eISBN: 978-2-35158-179-7; Eds. Inge Rörig-Dalgaard and Ioannis Ioannou PRO 111: MSSCE 2016 - Electrochemistry in Civil Engineering (2016), ISBN: 978-235158-176-6, e-ISBN: 978-2-35158-177-3; Ed. Lisbeth M. Ottosen PRO 112: MSSCE 2016 - Moisture in Materials and Structures (2016), ISBN: 978-235158-178-0, e-ISBN: 978-2-35158-179-7; Eds. Kurt Kielsgaard Hansen, Carsten Rode and Lars-Olof Nilsson PRO 113: MSSCE 2016 - Concrete with Supplementary Cementitious Materials (2016), ISBN: 978-2-35158-178-0, e-ISBN: 978-2-35158-179-7; Eds. Ole Mejlhede Jensen, Konstantin Kovler and Nele De Belie PRO 114: MSSCE 2016 - Frost Action in Concrete (2016), ISBN: 978-2-35158-182-7, e-ISBN: 978-2-35158-183-4; Eds. Marianne Tange Hasholt, Katja Fridh and R. Doug Hooton PRO 115: MSSCE 2016 - Fresh Concrete (2016), ISBN: 978-2-35158-184-1, e-ISBN: 978-2-35158-185-8; Eds. Lars N. Thrane, Claus Pade, Oldrich Svec and Nicolas Roussel PRO 116: BEFIB 2016 – 9th RILEM International Symposium on Fiber Reinforced Concrete (2016), ISBN: 978-2-35158-187-2, e-ISBN: 978-2-35158-186-5; Eds. N. Banthia, M. di Prisco and S. Soleimani-Dashtaki PRO 117: 3rd International RILEM Conference on Microstructure Related Durability of Cementitious Composites (2016), ISBN: 978-2-35158-188-9, e-ISBN: 978-2-35158189-6; Eds. Changwen Miao, Wei Sun, Jiaping Liu, Huisu Chen, Guang Ye and Klaas van Breugel PRO 118 (4 volumes): International Conference on Advances in Construction Materials and Systems (2017), ISBN Set: 978-2-35158-190-2, Vol. 1: 978-2-35158-193-3, Vol. 2: 978-2-35158-194-0, Vol. 3: ISBN:978-2-35158-195-7, Vol. 4: ISBN:978-2-35158-1964, e-ISBN: 978-2-35158-191-9; Eds. Manu Santhanam, Ravindra Gettu, Radhakrishna G. Pillai and Sunitha K. Nayar PRO 119 (online version): ICBBM 2017 - Second International RILEM Conference on Bio-based Building Materials, (2017), e-ISBN: 978-2-35158-192-6; Eds. Sofiane Amziane PRO 120 (2 volumes): EAC-02 - 2nd International RILEM/COST Conference on Early Age Cracking and Serviceability in Cement-based Materials and Structures, (2017), Vol. 1: 978-2-35158-199-5, Vol. 2: 978-2-35158-200-8, Set: 978-2-35158-197-1, e-ISBN: 978-2-35158-198-8; Eds. Stéphanie Staquet and Dimitrios Aggelis
xxxvi
RILEM Publications
PRO 121 (2 volumes): SynerCrete18: Interdisciplinary Approaches for Cement-based Materials and Structural Concrete: Synergizing Expertise and Bridging Scales of Space and Time, (2018), Set: 978-2-35158-202-2, Vol.1: 978-2-35158-211-4, Vol. 2: 978-235158-212-1, e-ISBN: 978-2-35158-203-9; Eds. Miguel Azenha, Dirk Schlicke, Farid Benboudjema and Agnieszka Knoppik PRO 122: SCC’2018 China - Fourth International Symposium on Design, Performance and Use of Self-Consolidating Concrete, (2018), ISBN: 978-2-35158-204-6, e-ISBN: 978-2-35158-205-3; Eds. C. Shi, Z. Zhang and K. H. Khayat PRO 123: Final Conference of RILEM TC 253-MCI: Microorganisms-Cementitious Materials Interactions (2018), Set: 978-2-35158-207-7, Vol.1: 978-2-35158-209-1, Vol.2: 978-2-35158-210-7, e-ISBN: 978-2-35158-206-0; Ed. Alexandra Bertron PRO 124 (online version): Fourth International Conference Progress of Recycling in the Built Environment (2018), e-ISBN: 978-2-35158-208-4; Eds. Isabel M. Martins, Carina Ulsen and Yury Villagran PRO 125 (online version): SLD4 - 4th International Conference on Service Life Design for Infrastructures (2018), e-ISBN: 978-2-35158-213-8; Eds. Guang Ye, Yong Yuan, Claudia Romero Rodriguez, Hongzhi Zhang and Branko Savija PRO 126: Workshop on Concrete Modelling and Material Behaviour in honor of Professor Klaas van Breugel (2018), ISBN: 978-2-35158-214-5, e-ISBN: 978-2-35158-215-2; Ed. Guang Ye PRO 127 (online version): CONMOD2018 - Symposium on Concrete Modelling (2018), e-ISBN: 978-2-35158-216-9; Eds. Erik Schlangen, Geert de Schutter, Branko Savija, Hongzhi Zhang and Claudia Romero Rodriguez PRO 128: SMSS2019 - International Conference on Sustainable Materials, Systems and Structures (2019), ISBN: 978-2-35158-217-6, e-ISBN: 978-2-35158-218-3 PRO 129: 2nd International Conference on UHPC Materials and Structures (UHPC2018-China), ISBN: 978-2-35158-219-0, e-ISBN: 978-2-35158-220-6; PRO 130: 5th Historic Mortars Conference (2019), ISBN: 978-2-35158-221-3, e-ISBN: 978-2-35158-222-0; Eds. José Ignacio Álvarez, José María Fernández, Íñigo Navarro, Adrián Durán and Rafael Sirera PRO 131 (online version): 3rd International Conference on Bio-Based Building Materials (ICBBM2019), e-ISBN: 978-2-35158-229-9; Eds. Mohammed Sonebi, Sofiane Amziane and Jonathan Page PRO 132: IRWRMC’18 - International RILEM Workshop on Rheological Measurements of Cement-based Materials (2018), ISBN: 978-2-35158-230-5, e-ISBN: 978-2-35158-231-2; Eds. Chafika Djelal and Yannick Vanhove PRO 133 (online version): CO2STO2019 - International Workshop CO2 Storage in Concrete (2019), e-ISBN: 978-2-35158-232-9; Eds. Assia Djerbi, Othman OmikrineMetalssi and Teddy Fen-Chong
RILEM Publications
xxxvii
PRO 134: 3rd ACF/HNU International Conference on UHPC Materials and Structures UHPC’2020, ISBN: 978-2-35158-233-6, e-ISBN: 978-2-35158-234-3; Eds. Caijun Shi and Jiaping Liu PRO 135: Fourth International Conference on Chemically Activated Materials (CAM2021), ISBN: 978-2-35158-235-0, e-ISBN: 978-2-35158-236-7; Eds. Caijun Shi and Xiang Hu
RILEM Reports (REP) Report 19: Considerations for Use in Managing the Aging of Nuclear Power Plant Concrete Structures (ISBN: 2-912143-07-1); Ed. D. J. Naus Report 20: Engineering and Transport Properties of the Interfacial Transition Zone in Cementitious Composites (ISBN: 2-912143-08-X); Eds. M. G. Alexander, G. Arliguie, G. Ballivy, A. Bentur and J. Marchand Report 21: Durability of Building Sealants (ISBN: 2-912143-12-8); Ed. A. T. Wolf Report 22: Sustainable Raw Materials - Construction and Demolition Waste (ISBN: 2-912143-17-9); Eds. C. F. Hendriks and H. S. Pietersen Report 23: Self-Compacting Concrete state-of-the-art report (ISBN: 2-912143-23-3); Eds. Å. Skarendahl and Ö. Petersson Report 24: Workability and Rheology of Fresh Concrete: Compendium of Tests (ISBN: 2-912143-32-2); Eds. P. J. M. Bartos, M. Sonebi and A. K. Tamimi Report 25: Early Age Cracking in Cementitious Systems (ISBN: 2-912143-33-0); Ed. A. Bentur Report 26: Towards Sustainable Roofing (Joint Committee CIB/RILEM) (CD 07) (eISBN 978-2-912143-65-5); Eds. Thomas W. Hutchinson and Keith Roberts Report 27: Condition Assessment of Roofs (Joint Committee CIB/RILEM) (CD 08) (e-ISBN 978-2-912143-66-2); Ed. CIB W 83/RILEM TC166-RMS Report 28: Final report of RILEM TC 167-COM ‘Characterisation of Old Mortars with Respect to Their Repair (ISBN: 978-2-912143-56-3); Eds. C. Groot, G. Ashall and J. Hughes Report 29: Pavement Performance Prediction and Evaluation (PPPE): Interlaboratory Tests (e-ISBN: 2-912143-68-3); Eds. M. Partl and H. Piber Report 30: Final Report of RILEM TC 198-URM ‘Use of Recycled Materials’ (ISBN: 2-912143-82-9; e-ISBN: 2-912143-69-1); Eds. Ch. F. Hendriks, G. M. T. Janssen and E. Vázquez Report 31: Final Report of RILEM TC 185-ATC ‘Advanced testing of cement-based materials during setting and hardening’ (ISBN: 2-912143-81-0; e-ISBN: 2-912143-705); Eds. H. W. Reinhardt and C. U. Grosse
xxxviii
RILEM Publications
Report 32: Probabilistic Assessment of Existing Structures. A JCSS publication (ISBN 2-912143-24-1); Ed. D. Diamantidis Report 33: State-of-the-Art Report of RILEM Technical Committee TC 184-IFE ‘Industrial Floors’ (ISBN 2-35158-006-0); Ed. P. Seidler Report 34: Report of RILEM Technical Committee TC 147-FMB ‘Fracture mechanics applications to anchorage and bond’ Tension of Reinforced Concrete Prisms – Round Robin Analysis and Tests on Bond (e-ISBN 2-912143-91-8); Eds. L. Elfgren and K. Noghabai Report 35: Final Report of RILEM Technical Committee TC 188-CSC ‘Casting of Self Compacting Concrete’ (ISBN 2-35158-001-X; e-ISBN: 2-912143-98-5); Eds. Å. Skarendahl and P. Billberg Report 36: State-of-the-Art Report of RILEM Technical Committee TC 201-TRC ‘Textile Reinforced Concrete’ (ISBN 2-912143-99-3); Ed. W. Brameshuber Report 37: State-of-the-Art Report of RILEM Technical Committee TC 192-ECM ‘Environment-conscious construction materials and systems’ (ISBN: 978-2-35158-0530); Eds. N. Kashino, D. Van Gemert and K. Imamoto Report 38: State-of-the-Art Report of RILEM Technical Committee TC 205-DSC ‘Durability of Self-Compacting Concrete’ (ISBN: 978-2-35158-048-6); Eds. G. De Schutter and K. Audenaert Report 39: Final Report of RILEM Technical Committee TC 187-SOC ‘Experimental determination of the stress-crack opening curve for concrete in tension’ (ISBN 978-235158-049-3); Ed. J. Planas Report 40: State-of-the-Art Report of RILEM Technical Committee TC 189-NEC ‘NonDestructive Evaluation of the Penetrability and Thickness of the Concrete Cover’ (ISBN 978-2-35158-054-7); Eds. R. Torrent and L. Fernández Luco Report 41: State-of-the-Art Report of RILEM Technical Committee TC 196-ICC ‘Internal Curing of Concrete’ (ISBN 978-2-35158-009-7); Eds. K. Kovler and O. M. Jensen Report 42: ‘Acoustic Emission and Related Non-destructive Evaluation Techniques for Crack Detection and Damage Evaluation in Concrete’ - Final Report of RILEM Technical Committee 212-ACD (e-ISBN: 978-2-35158-100-1); Ed. M. Ohtsu Report 45: Repair Mortars for Historic Masonry - State-of-the-Art Report of RILEM Technical Committee TC 203-RHM (e-ISBN: 978-2-35158-163-6); Eds. Paul Maurenbrecher and Caspar Groot Report 46: Surface delamination of concrete industrial floors and other durability related aspects guide - Report of RILEM Technical Committee TC 268-SIF (e-ISBN: 978-235158-201-5); Ed. Valerie Pollet
Keynote
Substituting Natural Pozzolan with Artificial Derived from Industrial Perlite Waste for Mortar Production Maria Stefanidou(B) School of Civil Engineering, Aristotle University of Thessaloniki, Thessaloniki, Greece [email protected]
Abstract. Pozzolans were used selectively in combination with lime to produce mortars used in structures where watertightness was required such as aqueducts, baths and pipes many centuries ago. The appreciated properties of mortars containing pozzolans were the increased durability and strength and the dense structure in relation to pure lime mortars. Ancient engineers managed to produce composites which proved their durability and their study provides significant information to modern materials technology. Today, natural pozzolans are used as binders for the production of mortars in order to produce repair materials compatible to old authentic mortars and also in cement industry. Thus, the reserves of natural pozzolans are not sufficient for the explosive development of the construction sector. Additional sources of pozzolanic materials were sought, such as industrial by-products used as supplementary materials. In the case of their use, the purpose is not only the safe and massive disposal of large quantities of materials that would otherwise burden the environment, but also the utilization of their special characteristics to produce different types of composites. Recent studies have documented that waste perlite, may present induced pozzolanic reactivity and could be used as an alternative pozzolanic material. Additionally, the substitution of natural pozzolan with waste perlite in lime-based mortars and grouts, has proved to be beneficial for their overall performance. Keywords: Natural Pozzolan · Perlite By-Products · Sustainability · Mortars · Industrial Waste
1 Introduction Lime produced using traditional practice (low temperature burning at different maturing times) and slow setting was used usually with local origin aggregates of homogeneous distribution for the mortar production. The prehistoric use of pozzolanic materials added in lime mortars lies in the use of reactive clays that were selected by empirical criteria [1]. The use of natural pozzolan in lime mortars has been identified in Hellenistic mortars of specific applications, such as the substratum of floor mosaics in Pella and Vergina [2, 3], which were also manufactured according to the need of resistance to humidity. The use of pumice in ancient Linda (4th century BC) and Delos (2nd century BC) was © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 A. J˛edrzejewska et al. (Eds.): SynerCrete 2023, RILEM Bookseries 43, pp. 3–19, 2023. https://doi.org/10.1007/978-3-031-33211-1_1
4
M. Stefanidou
widespread in structures requiring waterproof as tanks and water pipes [4]. Efstathiades [4] mentions the use of pozzolan in Kamiros Rhodes as early as the 5th BC century. The art of “reinforcing” lime mortars with natural pozzolanic materials was certainly appreciated by the ancient Greeks, who created hydraulic lime-mortars by utilizing the volcanic ash that was deposited on the Island of Santorini and the sounding islands by the eruption of Thera in 1500BC. Additionally, pozzolanic materials in the form of crushed brick and tile were deliberately added in large quantity in mortars of Minoan Crete of c 1000 BC, ancient Greece [5] (Table 1). Table 1. Properties of old mortars Period/Monuments
Comp. Strength MPa
Porosity %
Specific gravity t/m3
Binder type L-lime P-pozzolan B-brick dust C-clay
Roman/Galerious Palace 4th A.D.
3.0–4.5
19–25
1.60–1.73
L+P
Early Byzantine/Hagia Sophia Thessaloniki 6th A.D.
2.0–6.0
20–25
1.55–1.75
L+P+B
Late Byzantine/Hagios Panteleimonas 14th A.D.
1.0–1.5
16.5–22
1.61–1.71
L+C
Ottoman/Alkazar 16th A.D.
1.0–1.5
20–25
1.54–1.56
L
Romans were the first that recorded the hydraulic properties of pozzolans and widely used them in construction [6]. Romans used them as pulvis puteolanus and referred to today as volcanic ash or pumice pozzolan. Lime–pozzolan technology was subsequently widely disseminated across the known-world under Roman occupation. Following the collapse of the Roman Empire in the fifth-century AD, knowledge of lime technology in Western Europe was gradually lost until its renaissance in 1755–1759 when English civil engineer John Smeaton, conducted extensive research into the hydraulic properties of lime–pozzolan binders. The practice, of the deliberate inclusion of pozzolanic materials, continued in Europe as demonstrated by the analysis of samples from the Byzantine Empire composing “kourasania”, mortars with extremely high-water tightness, containing brick dust and brick fragments in their composition, a type of artificial pozzolan [7], Venetian renders [8].
Substituting Natural Pozzolan
5
Using pozzolan in a mortar in relation to a neat lime mortar result in [9]: – The reduction of water required to obtain adequate workability. This is possibly due to the reduction of electrostatic phenomena that cause the grains of pozzolan under which the water molecules remain bound on the granules of the particles. Also, as rounded and smooth surface of the granules is the phenomena of friction are reduced and thus the mixture becomes fluid. – Reducing the tendency for shrinkage during curing. The reduction of water required is a key reason for the reduction of shrinkage phenomena. Also, stable compounds that begin to form early in mixed systems are dissuasive for the shrinkage phenomena. – Reduction of the porosity. The action of the fine material as a filler and the mode of crystallization and the formation of compounds in the mixtures pozzolanic assist to reduce porosity. – Increase the strength. The formation of stable crystalline compounds and reduce porosity are the main reasons for increasing the strength in pozzolanic mixtures. The modification in the matrix by the addition of pozzolan in pure lime binder, are due to reactions between the active ingredients of pozzolan (mainly silicon and aluminum) and portlandite. Thus, apart from the portlandite crystals provided in the structure of the lime mortars, the mortars containing pozzolana coexist C-S crystals and C-Al composition in fibrous form. – Increase the durability. Structural changes withing the binding tissue as well as stronger cohesion of the binder with the aggregates, results in higher durability. This also depends on several factors such as the composition of the pozzolan, the fineness, the amount of water involved in the reaction, the treated conditions of the fresh mixtures and the environment in which the material serves. Natural pozzolans had found place in durable concrete construction of the 20th century. Natural pozzolan is the only Supplementary Cementitious Material (SCM) proven to enhance, fortify and protect concrete long-term. Portland cements, which almost entirely replaced lime-based binders due to their superior strength and speed of set [10]. Only relatively recently has damage to the fabric of historic buildings ‘repaired’ with cement-based building materials highlighted the incompatibility of impermeable, rigid, cement-based materials with their historic lime-based counterparts [11]. Furthermore, it is recognized that the high strengths attained by Portland-cement based materials are unnecessary in many applications. Instead, it is considered appropriate in modern practice to mitigate the environmental impact of concretes by specifying Portland-composite cements, in which a substantial proportion of the Portland cement is substituted by secondary materials, despite this in some cases resulting in reduced strengths [12]. Durability of reinforced concrete structures under aggressive exposure and extreme climatic conditions has become a matter of great concern to the construction industry worldwide. Supplementary cementitious materials (SCMs), such as silica fume, fly ash, granulated blast furnace slag, metakaolin, natural pozzolan, sugarcane bagasse, waste foundry sand, etc., are commonly used to partially replace the ordinary Portland cement to enhance the durability of concrete [13]. Additional advantages gained by adding a pozzolan to the concrete mix design include reduced heat of hydration damage, increased long-term compressive strength, and a reduction in the massive carbon footprint from the production of Portland cement—natural pozzolans are not simply added, but are used to replace
6
M. Stefanidou
some of the cement. The pozzolanic reaction is a secondary, reaction to the primary reaction of Portland cement and water. The benefits of the pozzolanic reaction are many and can be summarized to the reduction of heat of hydration damage and escalate strength: NPs reduce the heat of hydration anywhere from 10–40% during the first 100 h, depending on the ultimate mix design, thus lowering the threat of thermal cracking and allowing for a cooler set. This slow pozzolanic hydration process can continue for months and even years, bringing the long-term strength of the natural pozzolan-based concrete well beyond that of ordinary Portland cement concrete. An indirect gain by using NPs in concrete is the protection against chemical attacks, mitigate alkali silica reaction (ASR) during concrete hardening, natural pozzolans readily react with calcium hydroxide as it becomes available, trapping any present alkali inside the densified cement paste, which alleviates capillary action and virtually eliminates ASR and efflorescence [14]. Driven by the need to dispose of the ash from the stacks of coal-fired power plants, the residual ash had similar composition to natural pozzolans forming what was called an artificial pozzolan. The cement industry soon transitioned from natural pozzolans to coal combustion residuals (fly ash) as their pozzolans of choice due to cost and availability. Artificial pozzolans, being a waste material, were easily given away or sold at a low cost [15, 16]. Today, several plants producing fly ash have been either shuttered or are scheduled to be shuttered in the coming years. There is a growing shortage of artificial pozzolan, and industry is looking for replacement pozzolans [17]. As a result of global action plans on climate change and related recommendations to stop the burning of coal for power generation, the availability of industrial by-products is expected to decrease in the next few years. Thus, the search for alternatives to industrial-by products, which are often used in concretes to increase their resistance, has led to more interest in the use of natural pozzolans, demonstrated to have a similar effect on the resistance of cementitious materials [18]. From the perspective of the principles of sustainable development, the use of pozzolanic materials in construction is a one-way street that cannot be ignored, due to the technical advantages and especially the durability that pozzolanic materials give to mortars and concretes, as well as the reduction of economic and environmental costs of modern projects. NPs are magmatic in origin and widely available worldwide, although like most naturally occurring compounds, all pozzolans are not the same. There are two types of NPs–Raw and Calcined. Most raw NPs are derived from magmatic sources such as pumice, pumicite, tuff, volcanic ash, and perlite. Calcined NPs are derived from various types of clays and shales. The most known definition (ACI 116R) is that pozzolans are siliceous or siliceous and aluminous materials which, in themselves, possess little or no cementitious value but when they are finely ground, in the presence of water, they react chemically with calcium hydroxide (CH) (at ordinary temperature) to form compounds possessing cementitious properties. The performance of NPs depends on the mineralogy and the fineness [9]. One simple experiment of grading a natural pozzolan indicate the different properties gained and the properties of mortars recorded (Table 2).
Substituting Natural Pozzolan
7
Table 2. Sieving pozzolan and recording the properties [9]. Pozzolan gradation (μm)
Reactive silica %
Specific gravity t/m3
Comp. Strength MPa - 28d
Porosity %
100–75
38.73
1.840
0.609
19.38
75–63
38.95
2.147
2.688
15.64
63–45
39.17
2.342
4.150
15.08
4.1 MPa) at both ages tested. Long-term the activity is higher than that of natural pozzolan (Table 4). In an effort to test the heat of hydration in the two compositions tested the use of TAM Air (by TA instruments) was used for the compositions shown in Table 5.
12
M. Stefanidou Table 4. Testing pozzolanicity according to ASTM C593-95 Materials
Compr. Strength (MPa)
Compr. Strength (MPa)
7 days
28 days
NP
8.80
9.10
D1S
7.76
10.09
Table 5. Compositions to control the heat of hydration. Lime
NP
DIS
W/B
NP
1
1
-
0.65
D1S
1
-
1
0.70
Fig. 5. Reveals that the accelerating period, which is the period after the first release of heat from the moment the slurry encountered water, changed when D1S was added. Specifically, the peak shifted from 9 to 15 h, with the addition of D1S. Also, the heat flux levels moved to lower levels than the reference composition. Heat flux levels after 4 days were ≈25 μW/g for LP, and ≈2.5 μW/g for D1S. According to Fig. 6, the cumulative heat flow of hydration per gram mass was initially higher for D1S indicating when the initial exothermic reaction of perlite was stronger than the reference composition. However, the heat flux decreased over the following days keeping the normalized cumulative energy constant, with NP continuing the reaction and surpassing the cumulative energy after 3 and a half days of reaction.
Substituting Natural Pozzolan
13
Fig. 6. Normalized cumulative heat flow of hydration per gram versus time for compositions NP and D1S.
2.2 Mortars After recording the properties of the by-product and found the chemical, physical and microstructure properties, mortar mixtures were manufactured, according to Table 6, using as reference mortar containing lime and natural pozzolan (LP) and a mixture where NP was totally substituted by D1S. The Binder/Aggregate (B/A) ratio was maintained in all mixtures at 1/2, whereas aggregates were natural of siliceous origin and gradation 0–4 mm. The Water/Binder (W/B) ratio was adjusted for achieving workability 15 ± 1 cm, in accordance to EN1015-3 [25]. EN1015-11[26] was followed for manufacturing and curing mortars, preparing prismatic specimens (dimensions 4 × 4 × 16 cm). Additionally, plates of 25 × 25 × 2 cm were made to record thermal conductivity for each composition. Table 6. Constituents and proportions of the mortar mixtures. Raw materials
Parts per weight LP
D1S
Hydrated lime powder
1
1
Natural pozzolan
1
-
D1S
-
1
Siliceous sand (0-4mm)
4
4
W/B ratio
0.71
0.57
Workability (cm) (EN1015-3:1999)
15.4
15.0
14
M. Stefanidou
Regarding the water demand of the mixtures (Table 6), it was asserted that the reference mortar (LP) showed high W/B ratio (0.71) while in D1S a significant reduction was observed. When it comes to physical properties recorded (Table 7) and from the evaluation of the results it was concluded that the replacement of NP by D1S, led to the same level of porosity, and water absorption while lighter mortars were produced. Table 7. Physical properties of the mortar compositions Compositions
Porosity (%)
Absorption (%)
Ap. Spec. gravity
28d
90d
28d
90d
28d
90d
LP
28.44
32.85
16.75
19.79
1.70
1.66
D1S
28.73
33.19
17.11
20.22
1.68
1.64
Mechanical properties were influenced by the presence of perlite by-products, while their strength development was remarkable (Fig. 7). Both flexure and compressive strength were increased by time but in the case of compression strength the LP composition recorded 46% increase from 28 to 365 days while the D1S composition for the same period recorded 160% strength increase. Recording the thermal conductivity properties of the samples at different ages, it seems that λ is reduced by time and the samples with D1S present lower values (Table 8). This can be due to the lighter nature of the samples with D1S. Generally, total replacement of NP by perlite by-products, for manufacturing restoration mortars for historic building, seems to be a feasible and environmental beneficial approach. Regarding the latter, the produced materials could be characterized by a low environmental footprint due to the limitation of the mining of NP needed for their manufacture, as well as the exploitation of perlite by-products that would be otherwise deposited outdoor. The positive influence of perlite by-products on lime-based mortars, could be synopsized as following: – Reduction of the W/B ratio around 15-30%. – Keeping porosity, water absorption at the same level while producing a lighter material. – The low strength development rate – Reduction of the λ value From the above-mentioned remarks, it may be asserted that the type and proportion of perlite by-products are of paramount importance, influencing directly the physical and mechanical properties of the modified mortars. The total substitution had also positive results, showing that D1S could be used as an alternative pozzolanic material in limebased mortars.
Substituting Natural Pozzolan
15
Flexural strength (MPa)
4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0
LP 28d
90d
180d
D1S 365d
LP 28d
90d
180d
D1S 365d
Compressive strength (MPa)
14 12 10 8 6 4 2 0
Fig. 7. Mechanical properties of LP and D1S compositions in time.
Table 8. Thermal conductivity at different ages recorded by thermal plate technique (HFM-100 Heat Flow Meter by Thermtest)
LP
AGE
Thickness
λ [W/(m*K)]
days
mm
10 °C
20 °C
90
28.0
0.9306
0.9692
27.0
0.8928
0.9381
26.0
0.6567
0.6687
25.0
0.6228
0.6411
D1S LP D1S
365
16
M. Stefanidou
2.3 Grouts In case where perlite by-product was added in grout mixtures, two compositions were manufactured and tested (Table 9). Fresh (Table 10) and harden state properties were tested in the produced grouts (Fig. 8). Table 9. Proportion by weight of the grout components Compos Air lime Natural pozzolan D1S W/B LP
1
1
-
0.90
D1S
1
-
1
1.05
Table 10. Results of the fresh properties of grouts Comp
Fluidity (sec)
Penetrability (sec) ASTM (1h)
EN (1h)
Volume stability (%)
ASTM
EN
setting
bleeding
LP
12.23
8.90
12.80
9.55
3.90
0.8
0.4
D1S
12.06
8.72
12.55
10.12
1.97
3.0
2.0
From the replacement of NP by D1S in grout mixtures the following derive: Replacing NP with D1S caused 10% increase of water demand to achieve the same fresh state properties. Maintaining the fluidity of the compositions over time, which is an important property determining their effectiveness, was achieved. The penetration time of the mixtures was improved using perlite products by 25–60%. The mechanical characteristics of the compositions increase significantly over time. • D1S is a fine and reactive material that has the potential to be used as a binder with high pozzolanic properties • D1S product can be characterized as a mortar, with a strong pozzolanic effect. According to the relevant regulation (ASTM C618) it can be characterized as a natural pozzolan (type N) and used without further processing in the production of lime-based mortars. This is based both on the fineness of the grinding of the material and on its high percentage of aluminosilicate components, a significant part of which remains active and can react with lime through aluminosilicate reaction • D1S can be used as a pozzolanic material in mortars and grouts for the restoration of historic masonry.
Substituting Natural Pozzolan
17
1.0 0.9
Flexural strength (MPa)
0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0
LP 28d
90d
180d
LP 28d
90d
180d
D1S 365d
4.0
Compressive strength (MPa)
3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0
D1S 365d
Fig. 8. Mechanical properties of grouts in time.
3 Conclusions As established by the present work, pozzolanic materials present a diachronic use. Their value starting from the early, empirical construction has been scientifically documented and has been institutionalized with regulations regarding their use. At the same time, the importance of pozzolans is great for the new development of building technology, which is based on sustainable development. In this context, limiting the extraction of natural pozzolan using industrial byproducts is an important research topic. Materials that are now an environmental burden are tested and utilized. The specific by-products have such characteristics (size, composition) that they can participate in the production cycle of materials with important properties, making them suitable for use.
18
M. Stefanidou
Acknowledgements. The study was implemented during the Project ‘PerliMat – Exploitation of industrial products from the perlite process for the development of increased added value, ready mixed restoration materials’, Co-financed by the European Regional Development Fund of the European Union and Greek national funds through the Operational Program Competitiveness, Entrepreneurship and Innovation, under the call RESEARCH – CREATE – INNOVATE (project code: T2EDK-01105).
References 1. Davey, N.: A History of Building Materials. Phoenix House, London, UK (1961) 2. Papayianni, I., Pachta, V., Stefanidou, M.: Analysis of mortars from the floor substratum of the Aiges Palace and proposals of restoration materials. In: Proceeedings of the 3rd Hellenic Conference of Restoration, ETEPAM (2012) 3. Papayianni, I.: A diachronic principle in construction. In: Swamy, R.N. (ed.) The Use of Mixed Type Binders: Durability Aspects. J.G. Cabrera Symposium on Durability of Concrete Materials, pp. 115–130 (1998) 4. Efstathiades, E.: Worldwide pioneering the ancient Greek construction materials technology and its applications in civil engineering work. Technical Times (2004). [in Greek] 5. Theodoridou, M., Ioannou, I., Philokyprou, M.: New evidence of early use of artificial pozzolanic material in mortars. J. Archaeol. Sci. 40(8), 3263–3269 (2013) 6. Pollio, V., Morgan, M.H.: Vitruvius: The Ten Books on Architecture. Dover Publications, New York, USA (1960) 7. Binda, L., et al.: Experimental study on the mechanical role of thick mortar joints in reproduced byzantine masonry. In: Proceedings of the International RILEM Workshop on Historic Mortars: Characteristics and Tests (PRO 12), pp. 227–247. RILEM Publications SARL (2009) 8. Charola, E., Henriques, F.: Hydraulicity in lime mortars revisited. In: Proceedings of the International RILEM Workshop on Historic Mortars: Characteristics and Tests (PRO 12), pp. 95–104. RILEM Publications SARL (2009) 9. Stefanidou, M.: Natural pozzolans as sustainable building materials. Env. Earth Sci. 5, 758 (2016) 10. Vicat, L.J.: A treatise on calcareous mortars and cements, artificial and natural. Donhead, Shaftesbury (1837, reprinted1997) 11. Delgado Rodrigues, J., Grossi, A.: Indicators and ratings for the compatibility assessment of conservation actions. J. Cult. Herit. 8(1), 32–43 (2007) 12. Grist, E.R., Paine, K.A., Heath, A., Norman, J., Pinder, H.: Compressive strength development of binary and ternary lime–pozzolan mortars. Mater. Des. 52, 514–523 (2013) 13. Djamila, B., Othmane, B., Said, K., ElHadj, K.: Combined effect of mineral admixture and curing temperature on mechanical behavior and porosity of SCC. Adv. Concr. Constr. 6(1), 69–85 (2018) 14. Fengyan, W., Xianghui, L., Yinong, L., Zhongzi, X.: Effect of pozzolanic reaction products on alkali-silica reaction. J. Wuhan Univ. Technol.-Mater. Sci. Edn. 21, 168–171 (2006) 15. Kasaniya, M., Thomas, M.D., Moffatt, E.G.: Efficiency of natural pozzolans, ground glasses and coal bottom ashes in mitigating sulfate attack and alkali-silica reaction. Cem. Concr. Res. 149, 106551 (2021) 16. Rodriguez, R.E., UribeAfif, R.: Importance of using natural pozzolan on concrete durability. Cem. Concr. Res. 32(12), 1851–1858 (2002) 17. Pekmezci, B.Y., Akyuz, S.: Optimum usage of a natural pozzolan for the maximum compressive strength of concrete. Cem. Concr. Res. 34(12), 2175–2179 (2004)
Substituting Natural Pozzolan
19
18. AlChaar, G.K., Alkadi, M., Asteris, P.G.: Natural pozzolan as a partial substitute for cement in concrete. The Open Constr. Build. Technol. J. 7(1), 33–42 (2013) 19. Rashad, A.M.: A synopsis about perlite as building material - a best practice guide for Civil Engineer. Constr. Build. Mater. 121, 338–353 (2016) 20. Fodil, D., Mohamed, M.: Compressive strength and corrosion evaluation of concretes containing Pozzolana and perlite immersed in aggressive environments. Constr. Build. Mater. 179, 25–34 (2018) 21. Stefanidou, M., Pachta, V.: Influence of perlite and aerogel addition on the performance of cement-based mortars at elevated temperatures. In: Proceedings of SBE19 – Sustainability in the built environment for climate change mitigation, Thessaloniki (2019) 22. Ró˙zycka, A., Pichór, W.: Effect of perlite waste addition on the properties of autoclaved aerated concrete. Constr. Build. Mater. 120, 65–71 (2016) 23. Pachta, V., Papadopoulos, F., Stefanidou, M.: Development and testing of grouts based on perlite by-products and lime. Constr. Build. Mater. 207, 338–344 (2019) 24. ASTM C 593-95: Standard specification for fly ash and other pozzolans for use with lime 25. EN 1015-3: Determination of consistence of fresh mortar (by flow table) 26. EN 1015-11: Determination of flexural and compressive strength of hardened mortar
Modelling of Cement-Based Materials
A Benchmarking of Slag Blended Cement Hydration Models Jack Atallah(B) , Harifidy Ranaivomanana, François Bignonnet, and Stéphanie Bonnet Nantes Université, Ecole Centrale Nantes, CNRS, GeM, UMR 6183, 44600 Saint Nazaire, France [email protected]
Abstract. Several slag-blended cement hydration models widely used in the cement research field are benchmarked in this work, with a focus on their hypothesis, assets and drawbacks. The effect of slag on the hydrating mixture and the hydration products composition and quantities is considered. Different cement hydration models are presented in this study: analytical models such as Chen & Brouwers’ and Kolani’s models and generic numerical hydration models as CEMGEMS and VCCTL. Powers’ and Tennis & Jennings’ models are also used for comparison for compositions without slag-substitution. Computed results obtained with these models such as the degree of hydration and the hydration products composition and quantities are compared with experimental results found in the literature. Keywords: Hydration · Cement · Slag · Degree of hydration · Multi-physics modelling
1 Introduction Concrete represents 90% of the construction market [1]. It is considered as the most consumed material in the world [2, 3], succeeding water. In the context of sustainable development, the optimization and reduction of the use of cementitious materials is therefore becoming a core issue for the reduction of pollution in the construction sector. One of the solutions that can insure this reduction is the partial substitution of clinker by mineral additions. These additions affect the hydration reactions and are generally industrial by-products such as fly ash, blast furnace slag and silica fume. While the hydration processes of ordinary Portland cements is widely studied in literature, the slagblended cement was given less attention. The influence of slag can be identified according to several aspects, especially the hydration reactions and the degree of hydration of the material. In a study involving slag-blended cement hydration modeling, one must select a model with the best suited characteristics. The model characteristics may be of a chemophysical nature (such as the chemical composition of the output phases) or a technical nature (the model execution time for example). The model choosing process should naturally be a first step before finally selecting a desired hydration model and basing © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 A. J˛edrzejewska et al. (Eds.): SynerCrete 2023, RILEM Bookseries 43, pp. 23–33, 2023. https://doi.org/10.1007/978-3-031-33211-1_2
24
J. Atallah et al.
further modeling hypothesis and exploration. Doing a benchmarking of existing models contributes to making a well-informed choice of a hydration model considering its purpose (research or engineer application) while being well informed on its limits. Several multi-component hydration models based on microstructure modeling have been proposed previously (CEMHYD3D [4], CEMGEMS [5, 6]…). These models describe best the microstructure evolution during hardening but require a long computing time, which is not favorable in applications that are time sensitive. Here lies the importance of analytical models, that provides reliable results in short periods of time such as Chen and Brouwers’ [7–10] and Kolani’s [11] models. Additionally, models [12, 13] that do not take slag substitution into consideration are also included for comparison purposes on ordinary Portland cement. The different models investigated hereafter are: – – – – – –
Powers’ model [12], Tennis and Jennings’s model [13], CEMGEMS’ model [5], VCCTL’s model [14], Chen & Brouwers’ models [8–10], Kolani’s model [11].
Note that several other slag blended cement hydration models have been proposed by Stephant [15], Merzouki [16], Königsberger [17], Kinomura [18], Elakneswaran [19], Thomas [20], Phreeqc [21]. These more recent models are promising but have not been treated in this benchmark due to time considerations. The paper is organized as follows: the models are presented in Sect. 2 and compared to experimental data in Sect. 3. A conclusion is drawn in Sect. 4.
2 Models The benchmarked models are detailed below, in order to assess their reliability along with other criteria such as computation time and physical input parameters (consideration of temperature, curing conditions, oxides composition for the reagents). Ultimately, phases quantities will be compared, especially the portlandite formation that can be found experimentally with a good reliability. Powers’ model [12] is applied for the pure clinker hydration only. It is used as reference and a foundation for further models. Tennis & Jennings’ model [13] improves on Powers model, by taking into consideration time and further phases details. But it is also non-applicable to slag blended cement. Chen & Brouwers’ model [8–10] represents an approximation of physical behavior based on comprehensive chemical formulas. This analytical model is relatively fast. It does not offer a proper kinetic modeling; therefore, it can be coupled with a kinetic model such as offered in Meinhard [21] or Mounanga [22]. Kolani’s model [11] is an iterative analytical model, with relatively fast results, giving more details such as kinetics and heat flux. CEMGEMS [5] is a fast-numerical model, but does not provide time depending results, and has technical limitations such as the imposition of reactive oxides which may not be present in the original mixture. VCCTL [14] is a numerical model based on the CEMHYD3D system widely used in literature but has a long computation time.
A Benchmarking of Slag Blended Cement Hydration Models
25
2.1 Brief Overview of Cement Composition Cement is a hydraulic binder that, when mixed with water, forms a paste that hardens and provides mechanical resistance and stability to the structure. The characteristics of the slag and cement composition used by Kolani [11] are shown in Table 1. Table 1. Clinker and slag oxide composition for the Cement paste used in Kolani [11]. Name
Cement notation
Chemical formula
Clinker composition
Slag composition
Calcium oxide, C or lime
CaO
63.13
42.30
Silicon dioxide, S or silica
SiO2
20.68
35.90
Aluminum oxide, or alumina
A
Al2O3
4.40
11.20
Iron oxide, or rust
F
Fe2O3
2.34
0.30
Magnesium oxide, or periclase
M
MgO
2.10
8.00
Sulfur trioxide
S¯
SO3
3.27
0.20
The hydration of cement starts instantly when mixed with water. The reaction progresses upon contact of the binder with water and occurs with heat release. Hydration reactions are affected by many factors such as: the phase composition of the cement and the presence of different ions in the crystalline phase, the particle size distribution, the water to binder ratio, the curing temperature, the presence of chemical admixtures [22]. Main hydration products and their chemical formulas are presented in Table 2. 2.2 Ordinary Portland Cement Analytical Models Powers’ Model. Powers’ model [12] is a simple model involving three different phases solely, with no further detailing: the hydration products, the anhydrous clinker and capillary pores. Different models found in the literature are based on Powers’ model but considering different cement additives (such as slag [15], fly ash [23] …). Because of the simplicity of its original model, it cannot be taken into consideration when developing a slag blended cement model, it is only studied here to be taken as a reference when talking about hydration of pure clinker, with no slag added.
26
J. Atallah et al. Table 2. Different hydration products and their chemical formulas.
Abbreviation
Chemical formula
Name
CH
Ca(OH)2 or CaO · H2 O
Calcium hydroxide
C-S-H
x CaO·SiO2 ·y H2 O
Calcium silicate hydrate
C-S-A-H
x CaO·SiO2 ·y Al2 O3 ·z H2 O
Calcium aluminate hydrate
AFt
C6 (A,Fe)S3 H32
Calcium trisulfoaluminate hydrate/ettringite
AFm
C4 (A,Fe)SH12
Calcium monosulfoaluminate
Cc
CaCO3
Calcium carbonate
HT
M5 AH13
Hydrotalcite
HG
C6 AFS2 H8
Hydrogarnet
AH
C4 AH13
Tetracalcium
Mc
Ca4 Al2 (CO3 )(OH)12 .5H2 O
Calcium Monocarboaluminate
The clinker, water and hydration products quantity evolve during the hydration process depending on the initial water to cement ratio (w/c) as well as the curing conditions. The hydration advancement degree is described using the mass ratio of hydrated clinker to initial anhydrous clinker (α). The volume fractions of the anhydrous clinker, hydration products and capillary pores are calculated using the following formulas: fa =
1−α kh α ρa w/c + (1 − kh )α ; fh = ; fcp = 1 + ρa w/c 1 + ρa w/c 1 + ρa w/c
(1)
W here: fa : volume fraction of anhydrous clinker [m3 /m3 ], fh : volume fraction of hydration products [m3 /m3 ], fcp : : volume fraction of capillary pores [m3 /m3 ], ρa : density of anhydrous clinker (to be taken equal to 3.13) [kg.m−3 ], kh : volume of produced hydrates (to be taken equal to 2.13) [m3 ]. Tennis and Jennings’ Model. Tennis and Jennings’s model [13] is a hydration model based on chemical reactions more detailed than Power’s model. It distinguishes the anhydrous clinker phases (Alite, Belite, Aluminate and Ferrite), gypsum and the various hydration products phases (calcium hydroxide, hydrogarnet, CH, AFm, Aft…). It also studies the kinetics of the reaction based on the Avrami’s equations for each phase. These equations describe the kinetics of the hydration of the different cement phases. The model is based on the stoichiometric reactions for the hydration of the principal phases of clinker: C3S, C3A, C2S and C4AF. The deficiency in this model lies in the hypothesis that all the ettringite will react to form monosulfate, even though experimentally, it is proven that some ettringite is found at the end of the hydration process [24]. Another problematic point is the Bogue [25]
A Benchmarking of Slag Blended Cement Hydration Models
27
calculation, converting the oxides present in the clinker to the clinker phases, which is only an approximation and doesn’t provide exact values [26, 27]. Further, according to the authors themselves, Avrami equations are too simplistic to describe finely the hydration kinetics of cement paste. This model predicts the volume fractions of various compounds. It also quantifies the porosity, partitioning it into capillary porosity and C-S-H porosity. It further distinguishes different C-S-H types (high-density C-S-H (HD C-S-H) and low-density C-S-H (LD CS-H)). It also calculates the chemical shrinkage and can be used to compute physical properties of the cement. Unfortunately, this model is only appropriate to the slag cement hydration computation. This model will serve as a reference for further models’ validation in the limiting case of no substitution by slag.
2.3 Analytical Models for Slag Blended Cement Chen & Brouwers’ Model. Chen & Brouwers [8–10] proposed three different models for each of the hydration of the pure clinker, alkali-activated slag and slag blended cement. These models are based purely on the chemical reactions and stochiometric calculations, based on different hypothesis for each model. Starting with the initial oxide compositions and their proportions in the mixture, the hydration of the two main constituents (clinker and slag) is investigated separately, and then established considering the interaction of the hydration products. A mixture of the products of both Portland cement’s hydration and slag’s hydration is to form. The amount of CH formed by the Portland cement hydration is partially consumed by the slag hydration (Pietersen and Bijen, 1994 [28]; Regourd, 1980 [29]; Richardson and Groves, 1992 [30]). The most questionable product is the C-A-S-H whose composition varies depending on its calcium to silicon C/S ratio and aluminum to silicon ratio A/S. Chen [7] proposed three different models for slag-blended cement based on stoichiometry calculations, mainly to determine the C/S and the A/S ratios in the C-A-S-H, which impact the stoichiometry of the hydration reactions and the molar volume and weights of the C-AS-H. It is well established that the C/S ratio is lower in slag blended cement pastes than that in ordinary Portland cement pastes, indicating that the slag reaction is consuming additional calcium. This additional demand is provided by the CH formed by the hydration of the Portland cement. The curing time has been found to have minor effect on the stoichiometry of C-A-S-H, contrarily to the slag proportions that modify strongly that stoichiometry. In fact, when increasing the slag over cement ratio, the C/S ratio in C-A-S-H decreases while the A/S ratio increases [7]. Three models are examined, depending on the different degrees of aluminum substitutions for silicone in C-S-H. These models are used to quantify the hydration products and to determine the composition of C-S-H based on stochiometric calculations. The three models have the following hypotheses: 1) Model 1: No CH enters the C-S-H formed by the slag hydration.
28
J. Atallah et al.
2) Model 2: CH enters C-S-H from the slag hydration to sustain a C/S ratio 1.8, a common value for ordinary Portland cement paste. 3) Model 3: Part of the CH enters C-S-H from the slag hydration. Once the three models have been developed, a comparison to a series of experimental results has been carried out. Only the third model is retained in the present work. The results given by this model do not explicitly depend on time, but on the degrees of hydration of clinker and slag. Kolani’s Model. Kolani [11] proposed an incremental model based on the identification of the degrees of hydration, heat exchange, and the hydric state. A set of partial differential equations is established to govern the evolution of the corresponding variables with time. This multiphasic model takes into consideration the difference in the kinetics of hydration of the slag and cement constituents, combined to a stochiometric chemical model. The interactions between the phases integrates a statistical approach to predict the stoichiometry of the C-S-H formed. The kinetics model proposed by Buffo-Lacarriere [30] is adapted to take into consideration the slag blended cement hydration’ specifications. The main equations governing this model can be resumed in these equations: The rate of hydration of the anhydrous phases i in {clinker, slag} is: α˙ i = Ai · gi · πi · hi · si
(2)
where α˙ i , Ai , gi , πi , hi , si are respectively: the time derivative of the hydration degree of phase i, a global fitting parameter, the dissolution activation, the water availability, the thermic activation and the chemical activation detailed further in [12]. The water and heat conservation equations: −→ W ˙ = −div −Dw0 exp(pW )− grad W + Qth,i · fi · α˙ i (3) W −−→ W · fi · α˙ i ρc · T˙ = −div −λgrad T + Qth,i
(4)
W , Q T , λ are respectively: the water content of the cement, where W , T , Dw0 , p, fi , Qthi thi the cement temperature, fitting parameter of the Mensi law (1998) [31], another fitting parameter of the Mensi law, anhydrous compound i dosage, water consumed by compound i, total hydration heat for compound i and the cement thermic conductivity.
2.4 Numerical Models for Slag Blended Cement CEMGEMS. CEMGEMS [5, 6] is a web platform that provides an easy user-interface thermodynamic models for cementitious material hydration in the cement chemistry and engineering domain. This web app is developed and maintained in Switzerland at CONGINEER Ltd, thanks to funding from the international Nanocem consortium. The CEMGEMS web app uses the developed UpToDate GEMS3K code to compute chemical speciation by Gibbs energy minimization (GEM) according to the standard thermodynamic data from the Cemdata18 (Empa) and PSI/Nagra chemical thermodynamic databases.
A Benchmarking of Slag Blended Cement Hydration Models
29
CEMGEMS is used to compute the hydration process, the interstitial pore solution, the heat exchange and much more using a simple, intuitive user interface with changeable input parameters offering the results under plot views and data tables formats. VCCTL. The Virtual cement and concrete testing laboratory (VCCTL) [4] is a virtual laboratory software where cement paste and concrete materials are computed in a virtual environment. It has been researched and developed for over 30 years [32–35]. The goal of this computational modeling of cement and concrete hydration is to provide an accurate and time-saving virtual fast prototyping tool that enables for the investigation of the characteristics of various hypothetical combination designs combining regularly used concrete components [36]. A solution like this might drastically minimize the amount of time and effort required to generate and qualify concrete mixes for industrial purposes. It operates on any random mixture of cement powders, fillers, aggregates or any other cementitious material under a wide range of curing conditions. Transport properties, alongside with thermal and mechanical properties are determined over the course of the hydration. It has a graphical user interface allowing the creation, hydration and estimation of the 3D cement material microstructure. It is based on the CEMHYD3D system, widely used in literature and for many cement types: slag blended cement [8], limestone blended cement [37], and has many applications such as porosity depercolation repercolation [38], microstructural studies [39].
3 Comparison 3.1 Summary of Models’ Characteristics A comparison between the models, regarding the input and output parameters are summarized in Table 3 below. As shown, each model presents different assets and limitations regarding the input and output parameters as well as the computational time. Table 3. A summary of the assets and limitations of some slag-blended cement hydration models.
30
J. Atallah et al.
3.2 Comparison on the Quantity of Consumed and Produced Phases To compare the model predictions of the quantity of consumed and produced phases, a simulation by Chen & Brouwers [8–10], Kolani [12], CEMGEMS [5] and VCCTL [9] models was done for slag-blended cements with clinker and slag oxide composition presented in Table 1. Four types of cement presented by [12] are treated with slag over cement ratio equal to 0%, 30%, 50% and 70%. These cements were treated in isothermal conditions and constant temperature equal to 20 °C. A slag substitution rate of 50% was taken, with a water over binder ratio equal to 0.5, a clinker hydration degree of 0.98 and a slag hydration rate of 0.34. These values of hydration degrees correspond to the maximum hydration degrees simulated using VCCTL at 365 days. The results of these calculations are presented in Fig. 1 below. The volume fractions’ comparison of the phases regarding the porosity, the C-S-H, the C-A-S-H shows relatively close results, at the exception of hydrates formation of which VCCTL shows a large underestimation. 100%
shrinkage Water
Volume fraction (%)
80%
C-S-H C-S-A-H Hydrates
60%
CH Slag
40%
C3S C2S
20%
C3A C4AF
0% VCCTL
Chen & Brouwers
Kolani
CEMGEMS
gypsum
Fig. 1. Phases comparison using VCCTL [14], Chen & Brouwers [8–10], Kolani [11] and CEMGEMS [5] models for a slag substitution rate of 50%, a water over binder ratio equal to 0.5, a clinker hydration degree of 0.98 and a slag hydration rate of 0.34. The key “hydrates” denotes all hydration products not already listed in the other.
Another type of validation is the comparison of the volume fractions provided by experimental campaigns carried out by Kolani [11]. For these types of comparison, experimental initial composition and conditions are taken into consideration, and the hydration rate of slag and clinker is pre-specified. Results are favorable, according to experimental data provided by Kolani [11], at the exception of VCCTL which overestimates the portlandite content. The validation of others hydration products presents several issues. First, the different models do not consider exactly the same hydration products. For example, calcium monosulfoaluminate is considered as a product of slag-blended cement hydration by Kolani [11] but not by Chen & Brouwers [8–10]). Second, experimental techniques such as XRD or TGA can be interpreted differently by various authors leading to different estimation of the nature and quantities of hydration products [37] (Fig. 2).
A Benchmarking of Slag Blended Cement Hydration Models 30% of slag substuon
0.15
0.1
0.05
0.15
0.1
0.05
0
0 0
0.2
0.4 0.6 clinker hydraon degree
0.8
0
1
50% of slag substuon
0.2
0.15
0.1
0.05
0
0.4 0.6 clinker hydraon degree
0.8
1
70% of slag substuon
0.2 Volume fracon of Portlandite (%m)
0.2 Volume fracon of Portlandite (%m)
0% of slag substuon
0.2 Volume fracon of Portlandite (%m)
Volume fracon of Portlandite (%m)
0.2
31
0.15
0.1
0.05
0 0
0.2
0.4 0.6 clinker hydraon degree
0.8
1
0
0.2
0.4 0.6 clinker hydraon degree
0.8
1
Fig. 2. Variation of the portlandite volume fraction depending on the clinker hydration degree as proposed by different models, validated by experimental results from Kolani [9]. Mixtures containing 0%, 30%, 50% and 70% slag substitution ratio respectively.
4 Conclusions A description was made of different types of models: analytical models such as Chen & Brouwers’ model or Kolani’s model, and numerical models such as CEMGEMS or VCCTL. – While each type of model has its own assets and drawbacks, analytical ones seem to provide reliable results (relatively close values of porosity, portlandite, C-S-H, C-A-S-H and other hydration products) in short computational time. – Numerical models require more computational time with outputs comparable to analytical model ones. – All these outputs models fit well to experimental data for the portlandite content, with a slight over estimation by VCCTL. Acknowledgements. This work was funded by the French National Research Agency (ANR) under grant DEMCOM ANR-20-CE22–0008-01.
32
J. Atallah et al.
References 1. Ollivier, J.P., Vichot, A.: La durabilité des bétons. Association technique de l’industrie des liants hydrauliques (2008) 2. Cement Technology roadmap 2009 - Carbon emissions reductions up to 2050. World Business Council for Sustainable Developement and International Energy Agency (2009) 3. Chen, C., Habert, G., Bouzidi, Y., Jullien, A., Ventura, A.: LCA allocation procedure used as an incitative method for waste recycling: an application to mineral additions in concrete. Resour. Conserv. Recycl. 54(12), 1231–1240 (2010) 4. Bentz, D.: CEMHYD3D: A Three-Dimensional Cement Hydration and Microstructure Development Modeling Package: Version 3.0. (2005) 5. Kulik, D.A., Winnefeld, F., Kulik, A., Miron, G.D., Lothenbach, B.: CemGEMS – an easy-touse web application for thermodynamic modeling of cementitious materials. RILEM Tech. Lett. 6, 36–52 (2021) 6. CemGEMS-2021–140-Supplementaryfile-1339–2–10–20210609.pdf 7. Chen, W.: Hydration of slag cement: theory, modeling and application. PhD Thesis, University of Twente, The Netherlands (2006) 8. Chen, W., Brouwers, H.J.H., Shui, Z.H.: Three-dimensional computer modeling of slag cement hydration. J. Mater. Sci. 42(23), 9595–9610 (2007) 9. Chen, W., Brouwers, H.J.H.: The hydration of slag, Part 1: Reaction models for alkali-activated slag. J. Mater. Sci. 42(2), 428–443 (2007) 10. Chen, W., Brouwers, H.J.H.: The hydration of slag, Part 2: Reaction models for blended cement. J. Mater. Sci. 42(2), 444–464 (2007) 11. Kolani, B., Buffo-Lacarrière, L., Sellier, A., Escadeillas, G., Boutillon, L., Linger, L.: Hydration of slag-blended cements. Cem. Concr. Compos. 34(9), 1009–1018 (2012) 12. Powers, T.C.: Fundamental aspects of shrinkage of concrete. Rev. Matér. 544, 79–85 (1961) 13. Tennis, P.D., Jennings, H.M.: A model for two types of calcium silicate hydrate in the microstructure of Portland cement pastes. Cem. Concr. Res. 30(6), 855–863 (2000) 14. Bullard, J.W.: VCCTL Software. NIST, Dec. 06, 2010. https://www.nist.gov/services-resour ces/software/vcctl-software. Accessed 09 Jun 2022 15. Stephant, S.: Etude de l’influence de l’hydratation des laitiers sur les propriétés de transfert gazeux dans les matériaux cimentaires. PhD Thesis, Dijon, France (2015) 16. Merzouki, T., Bouasker, M., Houda Khalifa, N.E., Mounanga, P.: Contribution to the modeling of hydration and chemical shrinkage of slag-blended cement at early age. Constr. Build. Mater. 44, 368–380 (2013) 17. Königsberger, M., Carette, J.: Validated hydration model for slag-blended cement based on calorimetry measurements. Cem. Concr. Res. 128, 105950 (2020) 18. Kinomura, K., Ishida, T.: Enhanced hydration model of fly ash in blended cement and application of extensive modeling for continuous hydration to pozzolanic micro-pore structures. Cem. Concr. Compos. 114, 103733 (2020) 19. Elakneswaran, Y., Owaki, E., Miyahara, S., Ogino, M., Maruya, T., Nawa, T.: Hydration study of slag-blended cement based on thermodynamic considerations. Constr. Build. Mater. 124, 615–625 (2016) 20. Thomas, J.J., et al.: Modeling and simulation of cement hydration kinetics and microstructure development. Cem. Concr. Res. 41(12), 1257–1278 (2011) 21. User’s Guide to PHREEQC – a Computer Program for Speciation, Reaction-Path, Advective-Transport, and Inverse Geochemical Calculations. https://wwwbrr.cr.usgs.gov/pro jects/GWC_coupled/phreeqc.v1/html/phqc_2.htm. Accessed 29 Nov 2022) 22. Shi, H., Zhao, Y., Li, W.: Effects of temperature on the hydration characteristics of free lime. Cem. Concr. Res. 32, 789–793 (2002)
A Benchmarking of Slag Blended Cement Hydration Models
33
23. Taylor, H.F.W., Mohan, K., Moir, G.K.: Analytical study of pure and extended Portland cement pastes. Part I: Pure Portland cement pastes. J. Am. Ceram. Soc. 68(12), 680–685 (1985) 24. Liu, L., Sun, W., Ye, G., Chen, H., Qian, Z.: Estimation of the ionic diffusivity of virtual cement paste by random walk algorithm. Constr. Build. Mater. 28(1), 405–413 (2012) 25. Bogue, T.L., Bogue, R.L.: Extinguish burnout in critical care nursing. Crit. Care Nurs. Clin. North Am. 32(3), 451–463 (2020) 26. Taylor, H.F.W.: Modification of the Bogue calculation. Adv. Cem. Res. 2(6), 73–77 (1989) 27. Taylor, H.F.W.: Cement Chemistry, 2nd edn. Thomas Telford, London (1997) 28. Pietersen, H.S., Bijen, J.M.: Fly ash and slag reactivity in cements: Tem evidence and application of thermodynamic modelling. In: Goumans, J.J.J.M., van der Sloot, H.A., Aalbers, Th.G. (eds.) Studies in Environmental Science, vol. 60, pp. 949–960. Elsevier (1994) 29. Regourd, M., Thomassin, J.H., Baillif, P., Touray, J.C.: Study of the early hydration of Ca3SiO5 by X-ray photoelectron spectrometry. Cem. Concr. Res. 10(2), 223–230 (1980) 30. Richardson, I.G., Groves, G.W.: Microstructure and microanalysis of hardened cement pastes involving ground granulated blast-furnace slag. J. Mater. Sci. 27(22), 6204–6212 (1992) 31. Mensi, R., Acker, P., Attolou, A.: Séchage du béton: analyse et modélisation. Mater. Struct. 21(1), 3–12 (1988) 32. Garboczi, E.J., Berryman, J.G.: Elastic moduli of a material containing composite inclusions: effective medium theory and finite element computations. Mech. Mater. 33(8), 455–470 (2001) 33. Bentz, D.P., Garboczi, E.J.: A digitized simulation model for microstructural development. Ceram. Trans. 16, 211–226 (1991) 34. Garboczi, E.J., Bentz, D.P.: Digital simulation of the aggregate—cement paste interfacial zone in concrete. J. Mater. Res. 6(1), 196–201 (1991) 35. Jennings, H.M., Johnson, S.K.: Simulation of microstructure development during the hydration of a cement compound. J. Am. Ceram. Soc. 69(11), 790–795 (1986) 36. Watts, B.E., Tao, C., Ferraro, C.C., Masters, F.J.: Proficiency analysis of VCCTL results for heat of hydration and mortar cube strength. Constr. Build. Mater. 161, 606–617 (2018) 37. Bentz, D.P.: Modeling the influence of limestone filler on cement hydration using CEMHYD3D. Cem. Concr. Compos. 28(2), 124–129 (2006) 38. Bentz, D.P.: Capillary porosity depercolation/repercolation in hydrating cement pastes via low-temperature calorimetry measurements and CEMHYD3D modeling. J. Am. Ceram. Soc. 89(8), 2606–2611 (2006) 39. Bentz, D.P.: Quantitative comparison of real and CEMHYD3D model microstructure using correlation functions. Cem. Concr. Res. 36(2), 259–263 (2006)
Refining Kinetic Models for SCM Reactivity in Blended Cements Ruben Snellings(B) KU Leuven, 3001 Leuven, Belgium [email protected]
Abstract. High-level replacement of clinker by supplementary cementitious materials (SCMs) offers a well-tried solution to reducing CO2 emissions of cement production. To sustain further clinker reductions, new sources and combinations of SCMs will need to step in and their impact on cement hydration will need to be understood as, for instance, in comprehensive hydration models. A key input for cement hydration models, whether based on thermodynamic, microstructural, or simple mass balance principles, are time-dependent data on reaction degrees of all cement constituents. These kinetic data are usually introduced either directly from experiments or from empirical models fitted to experimental data. While reliable reaction degree data are relatively straightforward to obtain for the well-known, crystalline clinker phases in cement, advanced analysis techniques are required to estimate reaction kinetics of amorphous phases, present in most SCMs. Generic kinetic models for SCM reaction are therefore of interest to explore and predict the impact of SCM reactions on the cement hydrate assemblage, and eventually, performance. This contribution proposes a new approach to refine kinetic models for SCM reaction by derivation of model parameters from reactivity screening tests, such as the R3 test, reported in the published literature and case studies. It is discussed how this approach can be used to model reactivity of SCMs and how it can be used and extended to investigate and parameterize kinetic interactions in hydration of SCMs. Keywords: Supplementary Cementitious Materials · Reactivity · Kinetic Models · R3 test · Cement hydration
1 Introduction The production of blended or composite cements offers a well-known pathway to lowering CO2 emissions [1]. To this end, about 900 Mt of supplementary cementitious materials (SCMs) are used in cement products as a replacement of Portland clinker [2]. Average clinker factors are currently situated around 0.76, yet have not significantly decreased over the last decade [1]. This lack of progress is reflecting slow adoption of low carbon cements in practice, but also indicates depletion or unavailability of well-known, performant SCMs, such as ground granulated blast furnace slag, in growth regions. Further increasing levels of clinker replacement will require optimising blended cement to make efficient use of all cement constituents, clinker and SCMs. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 A. J˛edrzejewska et al. (Eds.): SynerCrete 2023, RILEM Bookseries 43, pp. 34–42, 2023. https://doi.org/10.1007/978-3-031-33211-1_3
Refining Kinetic Models for SCM Reactivity in Blended Cements
35
In very low clinker cements (i.e. ≤50 wt.% clinker), the most reactive clinker phases (alite, aluminate) attain (near) full reaction in a matter of days, subsequently SCMs need to step in to contribute to the performance development and make up for the loss of clinker by dilution [3]. Strikingly, a literature survey indicated that SCMs fall short in terms of reactivity in very low clinker cements [4]. The higher the SCM content, the lower their reaction rate and degree of reaction. The underlying causes of the limitations on SCM reactivity are only partially understood and captured by models. The contribution of the SCM mainly depends on its intrinsic reactivity and how it is modulated by external factors such as temperature, supply of Ca(OH)2 , pore solution alkalinity or the availability of water or precipitation space [4]. The SCM reactivity is a key material property to be understood, predicted and enhanced to further increase clinker substitution. SCM reactivity encompasses both reaction rate and its integral, the reaction degree, at a given time. Adequate descriptions or models of SCM reactivity should include both aspects and interaction terms that account for external influencing factors. In modelling the hydration process of blended cements, it is crucial to accurately describe the reactivity of the SCM [5, 6]. The release of chemical species by the SCM markedly affects the hydration product assemblage, the composition of the hydrates and the porosity and pore size distribution of the hardened cement [7, 8]. Today, most hydration models incorporate either experimentally determined SCM reactivity data [3, 8, 9], or extrapolated reactivities from calibration experiments [10, 11]. Experimental determination of the degree of reaction of SCMs is tedious, in general, and limited by experimental accuracy [12]. Electron microscopy image analysis, X-ray diffraction, solid state NMR, selective dissolution treatments, mass balance etc. may all provide data and insight on SCM reactivity, yet each direct measurement technique has practical limitations related to treatment of signal overlap or estimation of correction parameters that may produce bias or imprecision [13]. Similarly, model estimation has shown limited success in predicting reactivity of SCMs that are significantly different from the original parameterised system [11, 14]. As a result the kinetic models usually require significant modification and tedious reparameterization for each new SCM type. This contribution aims to provide a pragmatic inroad to rationalisation and parameterisation of SCM reactivity. Heat release data obtained by the so-called R3 test for measurement of SCM reactivity are used to characterise the reactivity response as a function of time. A simple kinetic expression is proposed to describe the SCM heat release response. The potential use of the expression as a means to describe and understand SCM reactivity is illustrated by the analysis of available literature data for different SCMs and for different R3 model mix formulations.
2 Data The data used in this paper were published earlier by Avet et al. [15] and part of the underlying dataset treated by Li et al. [16]. For material characterisation data the reader is referred to the respective publications. The study by Avet et al. [15] introduced the concept of the R3 test for screening the reactivity of calcined clays. In the R3 test the SCM test material is made to react with
36
R. Snellings
a reagent mixture designed to trigger pozzolanic and hydraulic reactions. The reagent mixture intends to approximate the reaction conditions present in a hydrating blended cement (pH, available reactants). However, the reagent mixture does not include Portland cement or clinker to avoid measurement interferences and reagent variability. Instead, Ca(OH)2 , CaCO3 , K2 SO4 and KOH are the solid reagents that are mixed with water and the SCM at the onset of the test. The proposed mix proportioning is reproduced in Table 1. After mixing the samples are cured at 40 °C to expedite the SCM reaction and as such shorten the duration of the test. In the development of the test procedure a selection of mix formulation parameters were varied, including the portlandite to calcined clay mass ratio, the alkalinity of the mix (K2 O to calcined clay mass ratio), the sulfation level (SO3 to calcined clay mass ratio), and the water to solids mass ratio. Additionally, two curing temperatures were compared, i.e. 20 and 40 °C. In the final mix design Ca(OH)2 and water are provided in large excess to enable measurement of the unconstrained intrinsic reactivity of the SCM. Alkali, sulfate and calcium carbonate are mainly added to approximate the alkalinity and available anions in a typical hydrating blended cements. The R3 test adopts the cumulative heat release or the amount of bound water recorded for the test mixtures at a given time (3 or 7 days) as measures of SCM reactivity. In this study the heat release data are used as these provide a continuous measurement of reactivity over the duration of the test. The results used pertain to mixes comprising calcined clay 1, a commercial metakaolin and calcined clay 4, a calcined impure kaolinitic clay. Li et al. [16] presented a comparative study of SCM reactivity tests, one of which was the R3 reactivity test. 10 SCMs and one inert material were subjected to reactivity testing according to the selected test procedures by the participating laboratories. The SCMs comprised 2 calcined clays (CC1 and CC2), 1 natural pozzolan (Po), 2 slags (S1 and S8), 2 calcareous fly ashes (CFA1 and CFA2) and 3 siliceous fly ashes (SFA_E, SFA_I and SFA_R). To avoid interlaboratory variance, the R3 heat release test results contributed by one of the participating laboratories was retained and analysed in this study. This dataset was deemed representative and was situated within the established confidence intervals. Table 1. R3 test mix proportioning (mass ratios) [15] SCM
Ca(OH)2
CaCO3
K2 SO4
KOH
H2 O
1
3
0.5
0.118
0.025
5.4
3 Kinetic Model As in the R3 test all reactants except the SCM are in excess, the kinetic rate law of the SCM reaction in the R3 test can be treated as pseudo-first order. Therefore the rate law is expressed as in Eq. (1): −d [A]/dt = k[A]
(1)
Refining Kinetic Models for SCM Reactivity in Blended Cements
37
where A represents the SCM concentration, t the time and k the rate constant. The integrated rate law then becomes as given in Eq. 2: [A] = [A]0 e−kt or, if expressed as degree of reaction, α, in Eq. 3: α(t) = 1 − e−kt
(2)
(3)
Expressing the degree of reaction in terms of cumulative heat release, H, gives Eq. 4: H (t) = HR,ult 1 − e−kt (4) where H R,ult corresponds to the ultimate heat released by the full reaction of the SCM, which under ideal conditions identifies with the reaction enthalpy. Equation (4) is the central equation used in the data regression of the R3 heat release results. The reaction rate constants and the ultimate heat release were calculated for the SCMs studied by Li et al. Minor modifications to the R3 test can be made to derive additional parameters of interest. For instance, R3 test results measured for 20 and 40 °C curing can be used to study the temperature dependence of the reaction rate constant k. Here, a conventional Arrhenius equation was adopted to calculate the apparent activation energy Ea according to Eq. 5: k(T ) = Ae(−Ea/RT )
(5)
where T corresponds to the absolute temperature in K, R is the gas constant and A is a pre-exponential factor. Similarly, the kinetic model can be used to analyse the sensitivity of SCM reactivity to mix formulation parameters. The effect of alkalinity and sulfation levels was studied by fitting of Eq. 4 to the data of Avet et al. and derivation of the kinetic parameters. While Ca(OH)2 is supplied in excess in the R3 test, in real blended cements the availability of Ca(OH)2 is controlled by the hydration of clinker phases. Without availability of Ca(OH)2 , the SCM reaction slows down significantly. To model this strong deceleration of the reaction rate upon depletion of Ca(OH)2 in the mix, a Ca(OH)2 rate reduction term was introduced. An asymmetric logistic function (Gompertz function) was selected as it enables to capture the rapid deceleration of the reaction. The Gompertz growth model is often used to study the time evolution of growth processes that are limited by the availability of a reactant, however its chemical meaning is as yet unclear [17]. The Ca(OH)2 depletion term is introduced in Eq. 6: (6) H (t) = HR,ult (1 − exp(−kt)) 1 − exp −be−cX where b and c are fitting parameters, and X is the residual amount of Ca(OH)2 reactant in the paste at a given time. Parameters b and c were constrained to be the same for all Ca(OH)2 to SCM ratios considered. All data were normalized to the SCM mass, e.g. J/(g of SCM), before application and fitting of the kinetic model. The fitting routine used a generalized reduced gradient algorithm to minimize the residual sum of squares between the model and experimental data.
38
R. Snellings
4 Results and Discussion 4.1 Parameterisation of SCM Reaction Kinetics The experimental R3 test heat release results for the SCMs studied by Li et al. are fitted the kinetic model (Eq. 4). Figure 1a and 1b compare the experimental and modelled curves. The obtained reactivity parameters, i.e. the ultimate heat release and the rate constant are listed in Table 2. As can be observed in Fig. 1, the kinetic model enables to closely match the experimental heat release data.
Fig. 1. R3 test heat release curves of a range of conventional SCMs – experimental results are shown as solid lines, curves fitted by the kinetic model as dashed lines; a) data and models for the studied calcined clays, natural pozzolan and slags, b) data and models for calcareous and siliceous fly ashes. Experimental data from [16].
The fitted parameters reflect the material characteristics and general reactivity response of the different SCM types. Both calcined clays are highly reactive (high reaction rates), the difference in rate constant between CC1 and CC2 may be related to the calcination process (flash vs. soak calcination). CC1 as pure metakaolin also shows a higher ultimate heat release than CC2 matching the difference in initial kaolinite grade. The other SCM types invariably present lower rate constants, the slags being more reactive than the tested natural pozzolan and fly ashes. The calcareous fly ashes react more rapidly than the siliceous fly ashes, CFA1 showing a higher rate constant than CFA2 while having similar ultimate heat release. The siliceous fly ashes show low reaction rates, but quite different ultimate heat release values. Finally, the natural pozzolan (Po) shows the lowest ultimate heat release of the SCMs included in the study. Considered together, ultimate heat release and the rate constant provide a means to capture the reactivity measured by the R3 heat release test. The ultimate heat release provides insight in the nature and content of the reactive phase of the SCMs, while the rate constant constrains the instantaneous reaction of the SCM. 4.2 Temperature Dependence Variation of the R3 test curing temperature enables to directly estimate the temperature dependence of the SCM reaction. Figure 2 shows the experimental and modelled heat release curves for R3 tests of CC1 (data from [15]). The modelled curves match the
Refining Kinetic Models for SCM Reactivity in Blended Cements
39
Table 2. Ultimate heat release (H R,ult ) and rate constant (k) reactivity parameters fitted to the experimental R3 test results reported by [16]. Reactivity parameter
CC1
CC2 Po
H R,ult
[J/g SCM]
1044 496
k
[1/day] 2.28
1.15
S1
S8
CFA1 CFA2 SFA_E SFA_I SFA_R
558
417
419
318
562
278
0.81 0.70 0.51
0.31
0.16
0.04
0.18
137 504 0.6
Fig. 2. R3 test heat release curves of CC1 for curing at 20 and 40 °C. The inset shows the Arrhenius plot for derivation of the temperature dependency. Original data from [15].
experimental data well. As shown in Fig. 2, the Arrhenius equation (Eq. 5) can be readily used to derive an (apparent) activation energy of 81 kJ/mol for CC1, close to the value reported in the literature for silica fume [18]. Here the approach is illustrated using only two temperatures, preferably more data points are available for confirmation and error estimation. 4.3 Alkali and Ca(OH)2 Supply The kinetic model can also be used to study the effect of mix formulation on the reaction kinetics of SCMs. Here we discuss the effect of alkali level and Ca(OH)2 to SCM ratio. Obviously, other compositional parameters such as sulfation level or water to binder ratio could be investigated as well. Figure 3 illustrates the changes in ultimate heat release and rate constant for varying R3 model mix alkali concentration. The results for CC1 [15] indicate a reduction in the ultimate heat release with increasing alkali content, while in contrary the reaction rate constant increases significantly with increasing alkali content. These trends correlate
40
R. Snellings
with the short term accelerating effect of alkali on cement hydration at the expense of later age reductions in degree of hydration and performance [19].
Fig. 3. Calculated ultimate heat release (H R,ult ) and rate constant (k) parameters of calcined clay CC1 [15] for varying alkali levels (K2 O/SCM mass ratio) of the R3 test formulation. The Ca(OH)2 to SCM ratio was 1 to 1.
As part of the development of the R3 formulation, Avet et al. [15] experimented with different Ca(OH)2 to calcined clay ratios. A clear reaction limiting effect was observed when reducing the supply of Ca(OH)2 to ratios below 2 to 1. Verification by XRD confirmed that depletion of the Ca(OH)2 occurred in the pastes [15]. Figure 4 clearly illustrates that at early times initial reaction rates are similar, yet strongly drop for low Ca(OH)2 to SCM ratios when nearing Ca(OH)2 depletion. Based on the hypothesis that Ca(OH)2 depletion is the cause of the reaction deceleration, a Ca(OH)2 rate reduction term was added in Eq. 6. This term (introduced as a Gompertz function) strongly slows down the reaction rate when the limiting substrate, in this case Ca(OH)2 , is nearing depletion. Figure 4 also shows the fitted model curves for the 2 to 1 down to the 1 to 3 Ca(OH)2 to SCM mass ratios. The 3 to 1 data were taken as mother curve, all other curves were obtained by fitting the b and c constants of Eq. 6 and the Ca(OH)2 consumption rate (g/g SCM) to the data. The degree of reaction of CC1 was taken as the ratio of the instantaneous heat release to the ultimate heat release as determined from the CH:SCM 3:1 data. The specific Ca(OH)2 consumption of CC1 was fitted to be 2.28 g/g of CC1 as an average across all formulations. Slight overestimation of the heat release at low CH:SCM and underestimation of heat release at high CH:SCM may be explained by slight changes in the reaction stoichiometry, for instance the C/S ratio of C-A-S-H. Nevertheless, the close correspondence between the model and experimental data sustains the hypothesis. In a next step the Ca(OH)2 depletion term could be implemented in low clinker blended cements with limited supply of Ca(OH)2 .
Refining Kinetic Models for SCM Reactivity in Blended Cements
41
Fig. 4. R3 test heat release results for different Ca(OH)2 (CH) to SCM (CC1) ratios. Experimental results are shown as solid lines, modelling results as dashed lines. Experimental data from [15].
5 Conclusions and Perspectives This study proposes a straightforward kinetic model for derivation of SCM reactivity parameters from experimental R3 heat release test data. The kinetic model is based on a pseudo-first order rate law, and constitutes a rate and a yield term. The model was applied to literature R3 test results, in general close correspondence between the experimental data and the fitted models were obtained. The fitted rate constants and ultimate hydration heat (yield) terms could be readily used to quantitatively compare and qualify SCMs. The R3 test aims at measuring the intrinsic reactivity of SCMs in blended cement systems by providing excess Ca(OH)2 and water reagents. In real blended cements, the supply of both Ca(OH)2 and water may become rate limiting, therefore requiring rate reduction terms to be applied to modulate the SCM reactivity. As a first step, a Ca(OH)2 depletion reduction term was proposed and applied to model the reaction kinetics in R3 systems with lowered Ca(OH)2 availability. The close match between the experimental results and the model offers support to the model extension approach by rate reduction terms. In a next step the model curves could be used as input for kinetic modelling of blended cement hydration. Furthermore, it is of interest to explore if the model could be generalised by explicitly accounting for SCM fineness.
References 1. Environment, U.N., Scrivener, K.L., John, V.M., Gartner, E.M.: Eco-efficient cements: potential economically viable solutions for a low-CO2 cement-based materials industry. Cem. Concr. Res. 114, 2–26 (2018)
42
R. Snellings
2. Juenger, M.C.G., Snellings, R., Bernal, S.A.: Supplementary cementitious materials: new sources, characterization, and performance insights. Cem. Concr. Res. 122, 257–273 (2019) 3. Snellings, R., et al.: Hydration kinetics of ternary slag-limestone cements: impact of water to binder ratio and curing temperature. Cem. Concr. Res. 151, 106647 (2022) 4. Skibsted, J., Snellings, R.: Reactivity of supplementary cementitious materials (SCMs) in cement blends. Cem. Concr. Res. 124, 105799 (2019) 5. Ben Haha, M., De Weerdt, K., Lothenbach, B.: Quantification of the degree of reaction of fly ash. Cem. Concr. Res. 40(11), 1620–1629 (2010). https://doi.org/10.1016/j.cemconres.2010. 07.004 6. Glosser, D., Suraneni, P., Isgor, O.B., Weiss, W.J.: Estimating reaction kinetics of cementitious pastes containing fly ash. Cem. Concr. Compos. 112, 103655 (2020) 7. Gruskovnjak, A., et al.: Hydration mechanisms of super sulphated slag cement. Cem. Concr. Res. 38, 983–992 (2008). https://doi.org/10.1016/j.cemconres.2008.03.004 8. Deschner, F., et al.: Hydration of Portland cement with high replacement by siliceous fly ash. Cem. Concr. Res. 42, 1389–1400 (2012). https://doi.org/10.1016/j.cemconres.2012.06.009 9. Avet, F., Li, X., Scrivener, K.: Determination of the amount of reacted metakaolin in calcined clay blends. Cem. Concr. Res. 106, 40–48 (2018). https://doi.org/10.1016/j.cemconres.2018. 01.009 10. Kolani, B., Buffo-Lacarrière, L., Sellier, A., Escadeillas, G., Boutillon, L., Linger, L.: Hydration of slag-blended cements. Cem. Concr. Compos. 34, 1009–1018 (2012). https://doi.org/ 10.1016/j.cemconcomp.2012.05.007 11. Wang, T., Ishida, T.: Multiphase pozzolanic reaction model of low-calcium fly ash in cement systems. Cem. Concr. Res. 122, 274–287 (2019). https://doi.org/10.1016/j.cemconres.2019. 04.015 12. Scrivener, K.L., et al.: TC 238-SCM: hydration and microstructure of concrete with SCMs. Mater. Struct. 48(4), 835–862 (2015). https://doi.org/10.1617/s11527-015-0527-4 13. Durdzi´nski, P.T., et al.: Outcomes of the RILEM round robin on degree of reaction of slag and fly ash in blended cements. Mater. Struct. 50(2), 1–15 (2017). https://doi.org/10.1617/ s11527-017-1002-1 14. Glosser, D., Suraneni, P., Isgor, O.B., Weiss, W.J.: Using glass content to determine the reactivity of fly ash for thermodynamic calculations. Cem. Concr. Compos. 115, 103849 (2021) 15. Avet, F., Snellings, R., Alujas Diaz, A., Ben Haha, M., Scrivener, K.: Development of a new rapid, relevant and reliable (R3) test method to evaluate the pozzolanic reactivity of calcined kaolinitic clays. Cem. Concr. Res. 85, 1–11 (2016). https://doi.org/10.1016/j.cemconres.2016. 02.015 16. Li, X., et al.: Reactivity tests for supplementary cementitious materials: RILEM TC 267-TRM phase 1. Mater. Struct. 51(6), 1–14 (2018). https://doi.org/10.1617/s11527-018-1269-x 17. Anguelov, R., Borisov, M., Iliev, A., Kyurkchiev, N., Markov, S.: On the chemical meaning of some growth models possessing Gompertzian-type property. Math. Meth. Appl. Sci. 41(18), 8365–8376 (2018). https://doi.org/10.1002/mma.4539 18. Jensen, O.M.: The pozzolanic reaction of silica fume. MRS Online Proc. Libr. 1488, 45–54 (2012) 19. Mota, B., Matschei, T., Scrivener, K.: The influence of sodium salts and gypsum on alite hydration. Cem. Concr. Res. 75, 53–65 (2015). https://doi.org/10.1016/j.cemconres.2015. 04.015
Atomistic Dissolution of β-C2 S Cement Clinker Crystal Surface: Part 1 Molecular Dynamics (MD) Approach K. M. Salah Uddin1(B) , Mohammadreza Izadifar2 , Neven Ukrainczyk2 Eduardus Koenders2 , and Bernhard Middendorf1
,
1 University of Kassel, 34125 Kassel, Germany
[email protected] 2 Technical University of Darmstadt, 64287 Darmstadt, Germany
Abstract. The major concern of the modern cement industry is to minimize the CO2 footprint. Thus cement based on belite, an impure clinker mineral (CaO)2 SiO2 (C2 S in cement chemistry notation), which forms at lower temperatures, is a promising solution to develop eco-efficient and sustainable cementbased materials, used in enormous quantities. However, the slow reactivity of belite limits its application. To understand the reasons behind, dissolution mechanisms and kinetic rates at the atomistic scale provide fundamental insights. This work aims to understand the dissolution behaviour of different surfaces of β-C2 S providing missing input data and an upscaling modeling approach to connect the atomistic scale to sub-micro scale. First, a combined ReaxFF and metadynamicsbased molecular dynamic approach are applied to compute the atomistic forward reaction rates (RD) of calcium (Ca) and silicate species of (100) surface of β-C2 S considering the influence of crystal surfaces and crystal defects. To minimize the enormous number of atomistic events possibilities, a generalized approach is proposed, based on the systematic removal of the nearest neighbour’s crystal sites. This enables tabulation of data on the forward reaction rates of the most important atomistic scenarios, needed as input parameters to implement the Kinetic Monte Carlo (KMC) computational upscaling approach, presented in Part-2 paper of this proceedings. Keywords: cement hydration · belite · dissolution · ß-C2 S · free energy surfaces · facet properties · molecular dynamics simulation · ReaxFF · metadynamics
1 Introduction Concrete is a widely used construction material on earth and Ordinary Portland Cement (OPC) has been considered one of the main constituents of the concrete known as a binder, however, the CO2 emission during the production of cement is the main concern due to its environmental impact [1–3]. Conforming the future climate regulation, the use of cement clinker content needs to be reduced to develop environment-friendly concretes. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 A. J˛edrzejewska et al. (Eds.): SynerCrete 2023, RILEM Bookseries 43, pp. 43–53, 2023. https://doi.org/10.1007/978-3-031-33211-1_4
44
K. M. Salah Uddin et al.
Therefore, the key challenge is to produce cement clinkers with a lower CO2 emission while keeping similar or even improved hydration performance. Belite (ß-C2 S) based cement could be an alternative way to develop eco-concrete, due to its lower energy consumption (220 kJ/ton lower) and lower CO2 emission (10–15% lower) during the production compared to alite (C3 S) [4, 5]. However, the lower reactivity of belite is the key issue that needs to be solved. Hydration of cement is a complex process, especially the dissolution of different surfaces ß-C2 S at the atomistic scale are not fully elucidated yet. An advanced multiscale modelling approach will help to understand the mechanistic step of dissolution from atomistic to nano-scale. Molecular Dynamic (MD) simulation is a very effective method to calculate reaction mechanisms at an atomistic scale [6]. The current contribution aims to understand the dissolution reaction at the ß-C2 S-water interface at atomistic scale and develop the necessary parameters to connect the micro-scale level. Many methods have been developed for atomistic scale calculation over the years. Among them, DFT (Density Functional Theory) is considered to be the most accurate, however, it is not applicable to the larger system due to its incredibly high computation time [7]. In contrast, classical force field theory is unable to calculate the reaction, therefore, ReaxFF (Reactive Force Field theory) is developed to connect force field theory and quantum mechanics. ReaxFF has already shown great potential for computing the reaction mechanism including different fields successfully [8–13]. Nevertheless, it becomes computationally expensive during transition state (TS) calculation. Hence, metadynamics (metaD) is combined with ReaxFF to minimize the computational cost by enforcing the reaction using adding biased potential. This combined approach is already implemented successfully in portlandite dissolution [14]. In this work (Part-1), a multistep modelling approach has been taken to get depth information on the dissolution and reactivity of different surfaces of ß-C2 S at room temperature (298K). At first, the hydration of (100) surface of ß-C2 S was allowed for 600 picoseconds. Later, the pre-hydrated surface was used as an input to study the dissolution mechanism of calcium by using ReaxFF coupled with metaD [15]. Besides, the orientation of calcium and silicate are different depending on the number of neighbours. Therefore, the influence of crystal site neighbours was also calculated. Finally, all the calculated microscopic rate constant using transition state theory (TST) will be provided as an input for upscaling (in Part-2 contribution presented within this conference proceedings) using a Kinetic Monte Carlo (KMC) approach and calculating the overall rate of the dissolution [16].
2 Computational Detail The ReaxFF parameter (Si-O-H and Ca-O-H) used in this work is already been explained successfully the dissolution mechanism of portlandite and cement clinkers (C3 S) [17, 18]. MeatD is a powerful sampling method that enforces the reaction and calculates the transition state (TS) consuming lower computation time by adding the biased potential on a selected number of degrees of freedom, i.e., collective variable (CV) [19, 20]. The entire simulations were carried out by using ReaxFF in the LAAMPS (Largescale Atomic/Molecular Massively Parallel Simulator) platform [21], and, the additional PLUMED package was used as an extension of LAMMPS for metaD [22–24].
Atomistic Dissolution of β-C2 S Cement Clinker Crystal Surface
45
2.1 Model Construction In this work a β-C2 S crystal unit was used with the lattice parameter of a = 5.51 Å, b = 6.76, c = 9.33 Å, α = γ = 90° and β = 94.17° [25, 26]. The fresh cleaved orthogonal simulation cell with (100) surface of β-C2 S (21.90 × 27.04 × 18.56) × 10–30 m3 composed of 896 atoms was constructed by Avogadro and virtual nano lab (VNL) [27] and Avogadro [27, 28]. The geometry was optimized using with Hessian-free truncated Newton algorithm (HFTN), where the cutoff tolerances for energy and force are 4.18 × 10–4 and 4.18 × 10–8 kJ mol−1 respectively [29]. Afterward, a 1.18 × 10–26 m3 periodic cell filled with water was added on the top of the (100) surface of β-C2 S using packmole [30]. The density of water in the simulation cell was 1000 kg m−3 with a random distribution. At first, the constructed simulation cell was equilibrated for 150 picoseconds with 0.5 femtoseconds time steps using a Nose − Hoover thermostat (NVT) to reach the 298 K and 1 atm inside the simulation box (NVT ensemble). Later on, it simulation cell was hydrated for 600 picoseconds using Nose − Hoover barostat (NPT). The last geometry of the pre-hydrated (100) surface of ß-C2 S (after 600 picoseconds) was selected to calculate the dissolution mechanism using the combined approach of MetaD and ReaxFF. A periodic boundary condition was applied during the entire simulation. After analysing the hydrated surface, two different types of Ca were selected, the first Ca (Case. I) located between silicates (Ca-SiO4 row) and second, (Case. II) in between two Ca (Ca row). Each central Ca is surrounded by 4Ca neighbourneighbours located around the 5 Å radius of the centre. Hence, to calculate the influence of Ca neighbourneighbours different scenarios were considered. A calcium Ca-744 from (100) facet located in between two silicates (Case I, S0 ) is selected to calculate the dissolution mechanism from crystal surface to solution using the ReaxFF-metaD scheme. The selected CV was the distance between the COM (centre of mass) of the crystal and the specified calcium atom and biased potential was added as a Gaussian with frequency 40. Moreover, each Gaussian hill is added every 0.02 picoseconds (hill height of 6.28 kJ/mol and a full width at half-maximum of 0.2 × 10–10 m). The simulation was run for 500 picoseconds (till converged) with NPT ensemble at temperature 298 K and the free energy landscape of dissolution was calculated. In order to calculate the influence of neighbourneighbours, a similar approach was applied in absence of 1,2,3,4 Ca neighbourneighbour (Green) those pre-deleted in different combinations total of 8 scenarios for each case were calculated (Fig. 1. II). Moreover, the probability of Ca2+ ions dissolution is comparatively higher than the silicate dissolution due to their strong electrostatic interaction with neighbourneighbouring Ca. Therefore, for the silicate dissolution, at first, all surrounding Ca atoms were removed to free the silicate row, and the similar modelling approach is applied to calculate the dissolution of the central silicate (marked yellow cross) before and after removing 1 and 2 neighbouring silicates located on both sides in the same row (Fig. 1. III). Besides, investigating the possibility of etch pit formation, the dissolution behaviour (FES) of selected Ca (marked yellow cross) from the second layer was calculated in the presence and absence of Ca, and the silicate of the upper layer (Fig. 1. IV). Finally, the calculated rate of individual dissolution of the most important events by transition state theory (TST) for the (100) surface of ß-C2 S [31], The full set of individual reaction rates is transferred as a basic input of upscaling KMC approach
46
K. M. Salah Uddin et al.
Case I
Case II
Case III
Case IV
Fig. 1. Dissolution scenarios of (I) central Ca (Ca-SiO4 row) (II) central Ca in the Ca layer from (100) surface of ß-C2 S in absence of 1,2,3,4 Ca neighbours (green) removed at different combinations (scenarios) for both cases (a-h). (III) Dissolution scenarios of central silicate in the presence (a) and the absence (b, c) of two nearest silicate neighbours located on both sides of the center in a row one by one manner. (IV) Dissolution behaviour of a central Ca located in between two silicates in the second layer in (a) the presence of all Ca and silicates of the first layer (except one Ca located above the selected Ca), (b) after removing all Ca within 5 Å distance from the selected Ca and (c) one silicate from the first layer.
Atomistic Dissolution of β-C2 S Cement Clinker Crystal Surface
47
to calculate the overall dissolution rate of (100) surface of ß-C2 S as well as investigate morphological changes during dissolution (submitted separately as part two contribution of this proceedings).
3 Results and Discussion 3.1 Pre-hydration of ß-C2 S Although in slower rates than C3 S, β-C2 S is responsible for the early strength development of concrete by providing more C-S-H (Calcium-Silicate-Hydrate) hydration products and less portlandite [15]. The dissolution behaviour of (100) surface of β-C2 S could lead to an in-depth overview of its reactivity toward hydration. Therefore, the reaction was carried out for 600 picoseconds for calculating the interaction between the surfaces of β-C2 S and water bulk. According to the observation, the β-C2 S has shown less reactivity in comparison to C3 S because of its compact crystal structures and the absence of free oxygen, since the free oxygen of C3 S is an influential factor that leads to a strong interaction with water bulk [6, 15] (Fig. 2).
Fig. 2. Representative snapshot for the (100) surface of β-C2 S after pre-hydration for 600 picoseconds at 298 K.
3.2 Dissolution of Calcium from the First Layer of (100) Surface of β-C2 S: Case I & Case II, also Including Various Neighbour Scenarios The selected central Ca atom is located in between two silicates in case I, and it is surrounded by the 4 Ca neighbours radially distributed within 5 Å. The dissolution of central Ca in presence of all 4 neighbours (Fig. 3a) is required to overcome 164.30 kJ/mol, however, removing the 1 Ca neighbour placed in Ca-row reduces the activation energy slightly by 154.00 kJ/mol. Besides the G value of -13.80 kJ/mol indicate the exergonic process. Besides, removing one Ca from the Ca-silicate row (Fig. 3c) influences significantly on the dissolution behaviour of central Ca by reducing the activation barrier to 44 kJ/mol. The influence was very close to scenario 3 (activation barrier of 52 kJ/mol) when 2Ca was missing from the Ca-row. (Fig. 3d). In addition, the dissolution was barrier less when 2Ca is missing from the Ca-SiO4 row and remain barrierless with the further reduction (in absence of 3 or 4 neighbours).
48
K. M. Salah Uddin et al.
Case I
(a)
(b)
(c)
(d)
(e)
(f)
(g)
(h)
(i)
Case II
Fig. 3. The representative free energy landscape for the Ca dissolution from (100) surface of β-C2 S. Case 1:the dissolution scenarios of central Ca in presence of 4 Ca neighbours (a) and the absence of the different numbers of 1,2,3 (b-f). Case II: dissolution of central Ca located in Ca row (Fig. 1. Case II) in the presence of all 4 Ca neighbours (g) and absence of 1,2 Ca neighbours (h-i) at 298K.
In case 2, the selected central Ca was placed in the Ca-row. The complete dissolution for the first scenario (in presence of 4 Ca neighbours, S0 ) required overcoming a significantly lower activation barrier of 44.60 kJ/mol compared to a similar scenario in case I, Since it does not have similar electrostatic interaction with the oxygen of the silicate the dissolution process was more thermodynamically favorable (G = −173.60 kJ/mol) in comparison to the Ca in case I. In addition, the barrierless dissolution of central Ca in absence of 2, 3, and 4 neighbouring Ca indicated that the probability of Ca dissolution from the Ca-row is much higher compared to the Ca-silicate row.
Atomistic Dissolution of β-C2 S Cement Clinker Crystal Surface
49
3.3 Dissolution of Silicate from the First Layer of (100) Surface of β-C2 S: Case III with Various Neighbour Scenarios Usually, the dissolution of Ca is faster compared to the silicates due to the strong electrostatic interaction of 4 oxygen with the neighbouring Ca. Consequently, the dissolution of Ca is much faster than the silicate. Therefore, to calculate the silicate dissolution all the surrounding Ca should be removed first. Then, the geometry was allowed for prehydration to create the ideal scenario for silicate dissolution, where the three oxygen of each silicate are protonated, and remain oxygen interacted with the Ca of the second layer. The silicate dissolution is not favourable (G = +31.00) in presence of both silicate neighbours. Nevertheless, the activation energy decreases from 89.00 kJ/mol to 67.70 kJ/mol, and it becomes thermodynamically favorable (G = −144.60) after removing one silicate neighbour. Besides, the further removal of the second neighbour, the dissolution was barrier-less (Fig. 4).
(a)
(b)
(c)
Fig. 4. The representative dissolution profile of the central silicate (marked in Fig. 1. III, S0 ) (a-c), from (100) surface of β-C2 S before and after removal of one and two silicate neighbours at 298 K.
3.4 Dissolution of Calcium from the Second Layer of (100) Surface of β-C2 S: Case VI with Various Neighbour Scenarios The dissolution of Ca from the second layer was not favourable due to their high activation barrier (355.70 kJ/mol) in perfect condition. However, β-C2 S dissolution is influenced by the crystal defect and the reaction barrier reduces to 161.90 kJ/mol after removing the Ca from the first layer. Besides, the silicate from the first layer still creates a steric hindrance which leads to the dissolution process endergonic (G = +89.70 kJ/mol). Nevertheless, removing the silicate located above the dissolution pathways of the Ca not only reduces the activation energy (44.30 kJ/mol) but also the dissolution become favourable (G = −62.70 kJ/mol) (Fig. 5, Table 1). 3.5 Upscaling the Dissolution Rate for (100) Surface of β-C2 S To upscale the atomistic simulations towards much larger timescales sub-micrometer system-sizes, kinetic Monte Carlo (KMC) simulations are applied. KMC requires inputs
50
K. M. Salah Uddin et al.
(a)
(b)
(c)
Fig. 5. The representative dissolution profile of the Ca from the second layer (marked in Fig. 1. IV, S0 ) (a-c), from (100) surface of β-C2 S in the different scenarios: before and after removal of Ca and silicate from the first layer at 298 K.
Table 1. Thermodynamic properties of different cases with possible scenarios of Calcium and silicate dissolution from the (100) surface of β-C2 S at 298K. NB represents the neighbour. Cases
Possible Scenarios of dissolution of central Ca/silicate
I
S0 : In presence of 4 NB 164.30
−78.00
S1 : In absence of 1 NB 154.80
−13.80
S2 : In absence of 1 NB
44.00
−182.00
S3 : In absence of 2 NB
52.00
−130.16
S4 : In absence of 2 NB
0.00
----
S5: In absence of 3 NB
0.00
----
S0 : In presence of 2 NB
44.60
−173.60
S1 :In absence of 1 NB
42.40
−94.60
S2 :In the absence of 2 NB
0.00
II
III
IV
Free Energy of Activation (G* ) kJ/mol
Free Energy Change (G) kJ/mol
----
S0 : In presence of 2 NB
89.00
+31.00
S1 :In absence of 1 NB
67.70
−144.60
S2 :In the absence of 2 NB
0.00
S0 : In presence of upper layer Ca
355.70
---+247.70
S1 : In absence of upper 161.90 layer Ca
+89.70
S2 : In absence of upper layer Ca and one silicate
−62.70
44.30
Atomistic Dissolution of β-C2 S Cement Clinker Crystal Surface
51
of the individual atomistic dissolution rates. This can be calculated using transition state theory (TST) from the activation barrier of all scenarios computed from MD simulation (ReaxFF and MetaD), thus ready to be used as an input for KMC (kinetic Monte Carlo) approach. Finally, KMC can predict the overall mesoscopic dissolution rate of (100) surface of ß-C2 S and morphological change at far-from-equilibrium, as presented in Part 2 paper of this proceedings.
4 Conclusions The primary objective of this research is to elucidate the dissolution mechanism ß-C2 S using a multiscale modeling approach. The paper explains the dissolution behaviour of (100) surfaces of ß-C2 S considering 4 different crystal site cases, most representative for the dissolution events, including effects of the first site neighbour configurations, called scenarios. The results of the MD simulations can be summarized as follows. • The Ca in the Ca-row was found more reactive compared to the Ca located in the Ca-SiO4 row. • The increase in rate of Ca and silicate dissolution by decreasing the number of neighbours was systematically quantified, thus providing parameters required for upscaling by KMC approach. • The silicate dissolution was unfavourable due to the high activation barrier, however, removing the neighbours increases the dissolution rate of the central silicate. • the dissolution of Ca from the second layer (case IV), was not possible due to hindrance effect of the (undissolved) first layer. However, already partial dissolution of atoms from the first layer creates a channel that enables the dissolution of Ca below, which may explain the initiation of etch pit formation during the cement dissolution. The calculated atomistic dissolution rates provide most critical scenarios, which enables an input for KMC upscaling. Acknowledgments. The authors thank the German Research Foundation (DFG) for funding the project with the number 455605608 titled: ‘Elucidation of initial cement dissolution mechanism by the gap-bridging multiscale modeling approach (CEM-bridge)’.
References 1. Lea, F.M.: Lea’s Chemistry of Cement and Concrete, 4th edn. Elsevier ButterworthHeinemann, Oxford (2004). 0750662565 2. Taylor, H.F.W.: Cement Chemistry. Thomas Telford Publishing, London (1997) 3. Barcelo, L., Kline, J., Walenta, G., Gartner, E.: Cement and carbon emissions. Mater. Struct. 47(6), 1055–1065 (2013). https://doi.org/10.1617/s11527-013-0114-5 4. Scrivener, K.L., John, V.M., Gartner, E.M.: Eco-efficient cements: Potential economically viable solutions for a low-CO2 cement-based materials industry. Cem. Concr. Res. 114, 2–26 (2018). https://doi.org/10.1016/j.cemconres.2018.03.015
52
K. M. Salah Uddin et al.
5. Sui, T., Fan, L., Wen, Z., Wang, J.: Properties of belite-rich portland cement and concrete in China. JCEA 9, 384–392 (2015). https://doi.org/10.17265/1934-7359/2015.04.002 6. Salah Uddin, K., Middendorf, B.: Reactivity of different crystalline surfaces of C3S during early hydration by the atomistic approach. Materials 12(9), 1514 (2019). https://doi.org/10. 3390/ma12091514 7. Izadifar, M., Dolado, J.S., Thissen, P., Ayuela, A.: Interactions between reduced graphene oxide with monomers of (calcium) silicate hydrates: a first-principles study. Nanomaterials 11, 2248 (2021). https://doi.org/10.3390/nano11092248 8. Chenoweth, K., van Duin, A.C.T., Goddard, W.A.: ReaxFF reactive force field for molecular dynamics simulations of hydrocarbon oxidation. J. Phys. Chem. A 112, 1040–1053 (2008). https://doi.org/10.1021/jp709896w 9. van Duin, A.C.T., Strachan, A., Stewman, S., Zhang, Q., Xu, X., Goddard, W.A.: ReaxFF SiO reactive force field for silicon and silicon oxide systems. J. Phys. Chem. A 107, 3803–3811 (2003). https://doi.org/10.1021/jp0276303 10. Manzano, H., Pellenq, R.J.M., Ulm, F.-J., Buehler, M.J., van Duin, A.C.T.: Hydration of calcium oxide surface predicted by reactive force field molecular dynamics. Langmuir 28(9), 4187–4197 (2012). https://doi.org/10.1021/la204338m 11. Fogarty, J.C., Aktulga, H.M., Grama, A.Y., van Duin, A.C.T., Pandit, S.A.: A reactive molecular dynamics simulation of the silica-water interface. The J. Chem. Phys. 132(17), 174704 (2010). https://doi.org/10.1063/1.3407433 12. Cheung, S., Deng, W.-Q., van Duin, A.C.T., Goddard, W.A.: ReaxFF(MgH) reactive force field for magnesium hydride systems. J. Phys. Chem. A 109(5), 851–859 (2005). https://doi. org/10.1021/jp0460184 13. Senftle, T.P., et al.: The ReaxFF reactive force-field: development, applications and future directions. npj Comput. Mater. 2, 9396 (2016). https://doi.org/10.1038/npjcompumats.201 5.11 14. Salah Uddin, K., Izadifar, M., Ukrainczyk, N., Koenders, E., Middendorf, B.: Dissolution of Portlandite in pure water: Part 1 molecular dynamics (MD) approach. Materials 15(4), 1404 (2022). https://doi.org/10.3390/ma15041404 15. Salah Uddin, K.M. Elucidation of Chemical Reaction Pathways in Cementitious Materials. https://doi.org/10.17170/kobra-202009211835 16. Izadifar, M., Ukrainczyk, N., Salah Uddin, K.M., Middendorf, B., Koenders, E.: Dissolution of Portlandite in Pure Water: Part 2 Atomistic Kinetic Monte Carlo (KMC) Approach. Materials (Basel) 15, 1442 (2022). https://doi.org/10.3390/ma15041442 17. Aktulga, H.M., Fogarty, J.C., Pandit, S.A., Grama, A.Y.: Parallel reactive molecular dynamics: numerical methods and algorithmic techniques. Parallel Comput. 38, 245–259 (2012). https:// doi.org/10.1016/j.parco.2011.08.005 18. Manzano, H., Pellenq, R.J.M., Ulm, F.-J., Buehler, M.J., van Duin, A.C.T.: Hydration of calcium oxide surface predicted by reactive force field molecular dynamics. Langmuir 28, 4187–4197 (2012). https://doi.org/10.1021/la204338m 19. Barducci, A., Bonomi, M., Parrinello, M.: Metadynamics. WIREs Comput. Mol. Sci. 1, 826–843 (2011). https://doi.org/10.1002/wcms.31 20. Laio, A., Gervasio, F.L.: Metadynamics: a method to simulate rare events and reconstruct the free energy in biophysics, chemistry and material science. J. Phys. Chem. A 71, 126601 (2008). https://doi.org/10.1088/0034-4885/71/12/126601 21. Plimpton, S.: Fast parallel algorithms for short-range molecular dynamics. J. Comput. Phys. 117, 1–19 (1995). https://doi.org/10.1006/jcph.1995.1039 22. Bonomi, M., et al.: PLUMED: A portable plugin for free-energy calculations with molecular dynamics. Comput. Phys. Commun. 180, 1961–1972 (2009). https://doi.org/10.1016/j.cpc. 2009.05.011
Atomistic Dissolution of β-C2 S Cement Clinker Crystal Surface
53
23. Tribello, G.A., Bonomi, M., Branduardi, D., Camilloni, C., Bussi, G.: PLUMED 2: new feathers for an old bird. Comput. Phys. Commun. 185, 604–613 (2014). https://doi.org/10. 1016/j.cpc.2013.09.018 24. Kylasa, S.B., Aktulga, H.M., Grama, A.Y.: PuReMD-GPU: a reactive molecular dynamics simulation package for GPUs. J. Comput. Phys. 272, 343–359 (2014). https://doi.org/10.1016/ j.jcp.2014.04.035 25. Chi, L., et al.: Hydration activity, crystal structural, and electronic properties studies of Badoped dicalcium silicate. Nanotechnol. Rev. 9, 1027–1033 (2020). https://doi.org/10.1515/ ntrev-2020-0082 26. Midgley, C.M.: The crystal structure of β dicalcium silicate. Acta Cryst 5, 307–312 (1952). https://doi.org/10.1107/S0365110X52000964 27. Schneider, J., et al.: ATK-ForceField: a new generation molecular dynamics software package. Model. Simul. Mater. Sci. Eng. 25, 85007 (2017). https://doi.org/10.1088/1361-651X/aa8ff0 28. Hanwell, M.D., Curtis, D.E., Lonie, D.C., Vandermeersch, T., Zurek, E., Hutchison, G.R.: Avogadro: an advanced semantic chemical editor, visualization, and analysis platform. J. Cheminform. 4, 17 (2012). https://doi.org/10.1186/1758-2946-4-17 29. Tuckerman, M.E., Alejandre, J., López-Rendón, R., Jochim, A.L., Martyna, G.J.: A Liouvilleoperator derived measure-preserving integrator for molecular dynamics simulations in the isothermal–isobaric ensemble. J. Phys. A: Math. Gen. 39, 5629–5651 (2006). https://doi.org/ 10.1088/0305-4470/39/19/S18 30. Martínez, L., Andrade, R., Birgin, E.G., Martínez, J.M.: PACKMOL: a package for building initial configurations for molecular dynamics simulations. J. Comput. Chem. 30, 2157–2164 (2009). https://doi.org/10.1002/jcc.21224 31. Laidler, K.J., King, M.C.: The development of transition-state theory. J. Phys. Chem. 87(15), 2657–2664 (1983)
Einstein Explains Water Transport in C-S-H Tulio Honorio(B) Université Paris-Saclay, CentraleSupélec, ENS Paris-Saclay, CNRS, LMPS - Laboratoire de Mécanique Paris-Saclay, 91190 Gif-Sur-Yvette, France [email protected]
Abstract. Water transport is critical for the durability and confinement performance of cement-based materials. C-S-H, the primary phase in hydrated cementbased materials, is nanoporous, contributing therefore to water transport. To understand water transport in C-S-H, it is necessary to deploy well-suited techniques and theoretical framework to deal with the nanoscale. Einstein linked the dynamics of a diffusing particulate system described by the mean-squared displacement (MSD) to the (self)-diffusion. The self-diffusion is, in turn, related to viscosity via the Stoke-Einstein relation. Using (equilibrium) molecular simulations, the selfdiffusion of water in C-S-H is computed via the Einstein MSD equation. Using non-equilibrium molecular simulations of pressure-gradient-driven flows in C-SH, it is shown that the Stokes-Einstein equation captures viscosity changes with the pore size expected in nanoporous materials. These results suggest that a fundamental link between diffusion and permeability can be established for C-S-H. Further, mean-field homogenization is deployed to get the effective diffusion and permeability of C-S-H at the gel scale. The results obtained are in good agreement with the available experimental data. In particular, it is finally possible to explain why the water permeability of the C-S-H gel (7 × 10−23 m2 ), as early calculated by Powers based on experiments, is so low. The strategy deployed here can be extended in future work to understand ion transport and unsaturated transport in C-S-H. Keywords: Diffusion · Permeability · Confined water · Bound water · Molecular Simulations
1 Introduction Water transport phenomena are critical for cement-based materials’ durability and performance specifications. The durability problems of concrete generally involve water and ions transport. The containment capacity of concrete is a crucial performance specification in the design and asset management of concrete infrastructures for energy production (dams, cement oil wells, nuclear power plants, and waste disposal structures) and sanitation (pipelines, sewerage, and sewage treatment). C-S-H, the major phase in hydrated cement-based materials, exhibits a bimodal pore size distribution [1, 2]: the smaller pores are interlayer pores (~ 1 nm), and the larger pores are gel pores (~ 3–10 nm). These two pore size classes correspond to the © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 A. J˛edrzejewska et al. (Eds.): SynerCrete 2023, RILEM Bookseries 43, pp. 54–65, 2023. https://doi.org/10.1007/978-3-031-33211-1_5
Einstein Explains Water Transport in C-S-H
55
micropore and mesopores classification of the IUPAC [3]. Water confined in micro and mesopores exhibit particular behavior differing from bulk water (e.g., [4]). To understand the behavior of confined water and how water in micro and mesopores contributes to transport in C-S-H, it is necessary to deploy well-suited techniques and theoretical frameworks to deal with the nanoscale. This work presents the theoretical framework explaining the dynamics in porous materials from the nanoscale. The results of Einstein on diffusion [5] are critical in setting up the fundamentals in this framework. Then, molecular simulations are preformed based on the theory presented. Next, mean-field homogenization is deployed to estimate gel transport properties using as input the information from the molecular scale. Comparisons with experimental data are made at the gel scale for both diffusivity and permeability.
2 Transport at the Molecular Scale 2.1 From Displacement to Self-diffusion In a series of seminal papers starting in 1905 (compiled later in a single document [5]), Einstein indicated that the “irregular thermal movements of the molecules of the liquid” are the origin of diffusive processes (including Brownian movement). He derived the diffusion equation, in an equivalence with Fick’s laws, by accounting for the flux of particles and their typical mean free path [6]. The associated solution in his formulation allows establishing a linear link between the time evolution of the mean-squared displacement (MSD) |r(t)|2 and the self-diffusion coefficient Ds [5]. One can rigorously write this relation as (e.g. [7]): Ds =
|r(t)|2 1 lim 2ds t→∞ t
(1)
where ds is the dimensionality, and t is the time. The self-diffusion coefficient is characteristic of the diffusion of an individual diffusing fluid particle [8]. This diffusion coefficient is associated with microscopic diffusion at equilibrium as described by: J = Ds ∇c
(2)
where J is the molecular flux and ∇c is the concentration gradient. The self-diffusion coefficient can be easily obtained from (equilibrium) molecular dynamics simulations. Experimental techniques such as Pulsed Field Gradient Nuclear Magnetic Resonance (PFG-NMR) and Quasi-Elastic Neutron Scattering (QENS) enables quantifying Ds [8]. The self-diffusion coefficient can be also be written in the form of a Green-Kubo relation (e.g. [7]): Ds =
1 ∞ N ∫ vi (0).vi (t)dt i 3N 0
(3)
as a function of the velocity auto-correlation function v(0). v(t) of the N particles in the system. It can be demonstrated that Eqs. (1) and (3) are equivalent [7].
56
T. Honorio
A collective diffusion coefficient D0 quantifying the dynamics of the fluid phase accounting for the particle correlations can be computed with [8]: D0 =
1 ∞ N N ∫ vi (0).vj (t)dt j i 3N 0
(4)
Ds and D0 differ from the velocity cross-correlation terms vi (0).vj (t) (with i = j) quantifying how the velocity of a particle is correlated to that of another particle. According to the linear response theory, D0 relates the flux and the chemical potential gradient ∇μp (e.g. [8, 9]): J = D0
c ∇μp kb T
(5)
with kb being the Boltzmann’s constant, and T being the temperature. In this framework, the fundamental force at the origins of diffusion is the gradient of the chemical potential of particles; thus, the diffusional processes originated from the flow of particles minimizing the free energy [9]. Finally, one can define the macroscopic diffusion coefficient Dt linking, at equilibrium, the flow J to the macroscopic concentration gradient ∇c : J = Dt ∇c
(6)
with Dt being the diffusion coefficient related to the transport in a porous solid subjected to thermodynamic gradients [8]. This is the diffusion coefficient obtained from macroscopic measurement at equilibrium [10]. Writing the chemical potential as μp = μ0 + kb T ln f , where f is the fugacity, one obtain the relation [10]: ∂ ln f Dt = D0 (7) ∂ ln c T f where ∂∂ ln ln c T is known as the Darker factor, which can be computed from the fluctuations of fluid particle number [11]. The three diffusion coefficients should be approximately equivalent DS ≈ D0 ≈ Dt under the assumptions of (i) negligible velocity cross-correlation terms vi (0).vj (t) with respect to auto-correlation terms; and (ii) linear adsorption isotherm so that the Darker factor is approximately 1 [8]. 2.2 From Self-diffusion to Permeability The Einstein-Stokes relation provides a fundamental link between self-diffusion and dynamic viscosity μ (e.g. [12]): Ds =
kb T 6π re μ
(8)
where re is the equivalent radius of the diffusion particle. The validity of this expression need to be confirmed for the case of confined water in C-S-H.
Einstein Explains Water Transport in C-S-H
57
Once the viscosity is quantified, the permeability if given using Darcy’s law (e.g. [13]): q=
κ ∇p = K∇i η
(9)
where q is flux due to a pressure gradient ∇p (or hydraulic gradient ∇i = ∇p/ρg), with ρ being the fluid density and g the gravity acceleration), and κ is the intrinsic permeability (dimensions of length [L2 ]) and K is the hydraulic conductivity (dimension of length per time [L.T−1 ]). Water self-diffusion [14–16], as well as other physical properties [17], are recognized to be pore size dependent in C-S-H: the smaller the pore the more solid-like the behaviour. Viscosity is expected to follow the trend [13]. Molecular simulation arises as a powerful tool to quantify this pore size dependency, as shown in the sequel.
3 Molecular Simulations Assessing Transport Properties of Water Confined in C-S-H Using Kunhi Mohamed et al. [18] C-S-H atomistic model (with structural formula (CaO)1.67 (SiO2 ). nH2 O, for a water content n), systems with one slit pore ranging from 1.18 to 10.63 nm (i.e., c-length of simulation boxes varying from 2.55 to 12.0 nm) are constructed. ClayFF [19] with SPC/E [20] water model are used to describe interatomic interactions. All simulations are performed with LAMMPS [21]. Tail correction are adopted for the Lennard-Jones interactions. Ewald summation is used to cope with long range electrostatics. Grand Canonical Monte Carlo (GCMC) simulation are used to fill the pores with water so that the water content corresponds to the osmotic equilibrium under liquid saturated conditions. Details of simulation and structural validation are provided in a previous work [17]. Equilibrium Molecular Dynamics (MD) and Non-equilibrium Molecular dynamics (NEMD) are performed to quantify the self-diffusion and the hydrodynamics under forced flow, respectively, as detailed in the following. In both cases, due to the slit pore geometry of C-S-H at the molecular scale, the response is expected to be a function of the effective pore size thickness H and transverse isotropic, with the self-diffusion and permeability tensor being written, respectively, in the form: ⎛
⎞ 0 D|| (H ) 0 Ds (H ) = ⎝ 0 D|| (H ) 0 ⎠ 0 0 D⊥ (H ) ⎛ ⎞ 0 κ|| (H ) 0 κ(H ) = ⎝ 0 κ|| (H ) 0 ⎠ 0
0
(10)
(11)
κ⊥ (H )
where the subscripts || and ⊥ stand, respectively, for the components parallel and perpendicular to the pore plane.
58
T. Honorio
3.1 Self-diffusion from Equilibrium MD The self-diffusion coefficient was computed using Einstein equation (Eq. 1). The C-SH atomic configurations equilibrated at various temperatures are first equilibrated in a 0.1 ns NVT run with Nosé-Hoover thermostat with damping of 100 timesteps. Next, the systems are equilibrated during 0.1 ns in the microcanonical ensemble (NVE). In the production runs, the MSDs are obtained from NVE simulations during a 1 ns run. This simulation length is sufficient to capture Fickean regime in bulk water. Twelve independent trajectories were considered to compute the mean values and standard deviations. Self-diffusion is known to exhibit periodic boundary conditions finite size dependence [22, 23]. The simulation results were corrected for these artefacts [16]. 3.2 Hydrodynamics Under Forced Flow from NEMD NEMD simulations are performed to simulate Poiseuille flow (i.e. flow due to a pressure gradient ∇p in a given direction) in C-S-H nanochannels. An additional constant force F = ∇pSs L/N (computed from ∇p in the considered direction, number of fluid particles N , and surface area and length of pore, Ss and L„ respectively) is applied to each fluid molecule. The value of F = 0.75 × 10−3 kcal/(mol.Å) is adopted as in [24]; this force corresponds to a ∇p ranging from 2.75 to 5 MPa/nm (from smaller to the larger pores). The pressure gradient adopted in NEMD simulations of Poiseuille flows is generally much larger than the values observed in experiments so that the velocity profiles obtained from the simulation is significantly larger than the noise originating from the irregular motion of molecules. The Reynolds number for the largest pore (for which higher velocities are expected) is on the order of the unit, which means that laminar flow conditions are ensured in all cases considered. The temperature of the interlayer fluid is kept via velocities rescaling, and the thermostating is applied only the in-plane direction orthogonal to ∇p as in [24]. A timestep of 1 fs timestep is used; velocity and density profiles are calculated per chunks slicing the thickness of the slit pore and averaged during 0.2 ns blocks. A 1 ns run was sufficient to reach steady-state flows for the system studied. Production runs are performed for 2 ns.
4 Transport at the Gel Scale: A Mean-Field Approximation Analytical micromechanics schemes are deployed to get the effective transport properties of C-S-H at the gel scale. A single formulation can be obtained for the various conductivity-like properties [25], including diffusion and permeability (or hydraulic conductivity). The self-consistent (or Bruggeman) scheme is adopted to approximate the effective properties. With this scheme the effective diffusion of a macroscopically isotropic heterogeneous material composed of N microscopically transverse isotropic equiaxed inclusions is given by: Dgel =
N i=1
fi Di ASC i
(12)
Einstein Explains Water Transport in C-S-H
59
with fi and ASC i being the volume fraction and concentration tensor associated with selfconsistent scheme for inclusion i. ASC i can be computed following usual derivation for conductivity-like properties (e.g. [26]). Similarly, for the effective permeability κ gel =
N i=1
fi κi ASC i
(13)
with the same concentration tensor ASC i . C-S-H gel is assumed to be composed of two phases: microporous solid particles (subscripts,s ) and gel porosity (subscripts,g ). Both phases can exhibit transverse isotropic behaviour. For a two-phase material and under the assumption of macroscopically isotropic behavior (which implicates that all orientations of the anisotropic equiaxed inclusions are equiprobable), the expression above simplifies into the following the implicit equation:
D
hom
=
7(1−fg )D||,s 5(1−fg )D⊥ 7f D||,s + 2Dhom − D g+2D hom D||,s +2Dhom ||,g 7(1−fg ) 5(1−fg ) 7fg + D +2Dhom − D +2εhom D||,s +2Dhom ⊥,s ||,g
5fg D⊥,g D⊥,g +2Dhom 5fg − D +2D hom ⊥,g
−
(14)
where f g is the gel porosity. Explicit analytical roots of this equation exists but are too lengthy to be reported here. Similar expression can be written for permeability but of course as a function of κ|| (H ) and κ⊥ = 0. Differently from ref. [16], D⊥ here is also assumed to be zero. A correction factor fH = H /(H + 4.11) is applied to solid microporous C-S-H particle behavior to account for the volume effectively occupied by the channel, as done for the permeability in ref. [13]. Only the real, non-negative solution of Eq. (14) has physical meaning for diffusion and permeability. The pore size associated with microporous solid particles should be the equilibrium basal spacing d eq = 13.7 Å (i.e., the most prevalent pore size in C-S-H layers stacks [17, 27]). Typical gel pore sizes ranges from 2 to 10 nm diameter [2].
5 Results 5.1 Transport Properties at the Molecular Scale at the Molecular Scale Self-Diffusion. The log-log plots of the time evolution of the MSD for water confined in C-S-H is shown in Fig. 1 according to various pore sizes. Fickean diffusion (i.e. MSD being linear with time) is onset for timescales exceeding 0.1 ps for systems with interlayer distance above 2.6 nm. Smaller pore sizes exhibits sub-diffusive behaviour, which is associated with glassy dynamics in the literature [14]. The MSD of the smaller roughly converges to a target value roughly corresponding to a dynamics of particles in solids. Self-diffusion coefficients are computed for systems exhibiting Fickean diffusion by applying Einstein equation (Eq. (1)) only to the domain in which the MSD is linear with time. For the parallel component, the description of diffusion must account for the collision with the pore walls (see ref. [16] for an application of a better-suited theoretical
60
T. Honorio
description than Einstein’s equation in this case). The self-diffusion coefficients are depicted in Fig. 2 as a function of the interlayer distance for various temperatures. The diffusion in the pore plane direction retrieves bulk-like behaviour for interlayer distances above approximately 6 nm. The diffusion perpendicular to the pore wall is significantly reduced as expected due to the presence of pore walls.
Fig. 1. Mean-square displacement of confined water oxygens in C-S-H for various interlayer distances d : components in the (A) parallel and (B) perpendicular directions to the pore plane.
Fig. 2. Self-diffusion coefficient of water confined in C-S-H for various interlayer distances d : components in the (A) parallel and (B) perpendicular directions to the pore plane. In (A), the horizontal dashed line refers to bulk water self-diffusion for each temperature. Full lines are the fits obtained with Eq. 15.
The following expression fits the pore size and temperature dependence of parallel component of the self-diffusion in C-S-H at the molecular scale: Ea (d ) 1 1 d − 20 exp − − (15) D (d , T ) = Dbulk (T ) 1 − exp − 16.7 R T 300 where the Dbulk (T ) is the bulk self-diffusion at temperature T ; Ea (d ) is the activation energy (which is also pore size dependent [16]) and R is the gas constant. The fits are
Einstein Explains Water Transport in C-S-H
61
also shown in Fig. 2. This expression shows that an Arrhenius type thermal activation properly capture the temperature dependence of diffusion at the molecular scale. The parallel component of water self-diffusion in interlayer pores (i.e., with interlayer distance corresponding to the equilibrium basal spacing d eq = 13.7 Å) is 0.00096Dbulk , which compares well with molecular simulations reporting self-diffusion in the range 0.0006Dbulk to 0.0018Dbulk [15]. The volumetric diffusion Dv = (2D + D⊥ )/3 for interlayer pores is 0.0006Dbulk which is close the value of 0.0004Dbulk obtained from inverse analysis using micromechanics [28]. Permeability. The velocity profiles obtained from NEMD simulation are show in Fig. 3 for various effective pore sizes H and temperatures. The viscosity of confined water is evaluated from fitting the velocity profiles of NEMD simulations. Direct simulations using the auto-correlation of the out-of-diagonal components of pressure tensor (e.g. [29]) are also performed for comparison. Figure 3(B) shows that both strategies yields similar viscosities in reasonable agreement with Stokes-Einstein relation. A fundamental link between diffusion and permeability in C-S-H is, therefore, established for C-S-H using this relation.
Fig. 3. NEMD results for various pore sizes and temperature: (A) velocity profiles at 300 K for various pores sizes; (B) comparison of viscosity of confined water obtained from fitting NEMD, direct simulations suing auto-correlation pressure tensor and Stokes-Einstein equation at 300 K; (C) velocity profiles at various temperature for H = 82 Å and corresponding (D) viscosities.
The NEMD results in C-S-H clearly show the presence of a negative slip length (noted b in Fig. 3(A)): water close to the pore surface are unaffected by the pressure gradient showing therefore zero velocity. This negative slip length is approximated 7.5 Å, which roughly corresponds to two layers of adsorbed water molecules. Negative slip lengths
62
T. Honorio
are expected for forced flow in nanochannels made from hydrophilic materials as is the case of C-S-H.
5.2 Transport Properties at the Molecular Scale at the Gel Scale Figure 4 shows the effective diffusion and permeability at the gel scale as a function of the gel porosity fg as obtained from the micromechanical model presented in Sect. 4 using as input the results from molecular simulations from the last section. The packing density ηg = 1 − fg of low and high density C-S-H is generally assumes to be ηg = 0.64 for low density (LD) and ηg = 0.74 for high density (HD) C-S-H [30]. The effective diffusivity computed for LD and HD C-S-H are 0.012Dbulk and 0.0034Dbulk , respectively. These values compare well with experimental data from the literature: 0.02–0.05Dbulk from QENS experiments [31]; 0.017Dbulk from Incoherent Elastic Neutron Scattering experiments [32]; and 0.03Dbulk from PFG-NMR experiments [33]. The effective permeability computed for LD and HD C-S-H are 10–15 × 10−23 m2 and 8.5–10 × 10−23 m2 , respectively. These values are in good agreement with the available experimental data. In particular, it is finally possible to explain why the water permeability of the C-S-H gel (7 × 10−23 m2 ), as early calculated by Powers based on experiments [34], is so low. For the permeability, the results obtained using input with negative slip and no-slip (i.e. stick) boundary conditions are compared. The case of isotropic behavior with slip boundary (κ|| = κ⊥ ) is also shown for comparison. The case with slip boundary and 2D behavior (κ|| = 0) is more consistent with the experimental evidence. Finally, Fig. 5 shows the temperature evolution of the effective diffusion and permeability of LD and HD gel under undrained conditions. The data can be used in multiscale modelling of transport phenomena in cement systems.
A
B
Fig. 4. Effective (A) diffusion and (B) permeability of C-S-H gel for various effective gel pore sizes Hg .
Einstein Explains Water Transport in C-S-H
A
63
B
Fig. 5. Effective (A) diffusion and (B) permeability of C-S-H gel as a function of the temperature adopting a effective gel pore size Hg of 4 nm.
6 Conclusions In this work, the fundamental aspect associated with diffusion and permeability were presented together with an application to the case of C-S-H at the molecular scale. The results were then used in mean-field homogenization to get the effective diffusion and permeability of C-S-H gel. The following conclusions can be drawn: – Diffusion coefficients. The fundamental description of diffusion provided by Einstein enables quantifying the effect of confinement on the reduction of self-diffusion coefficients. The differences of self-diffusion, corrected/collective diffusion and macroscopic diffusion are yet to be properly quantified. The bottom-up strategy outlines here using molecular simulation and mean-field homogenization can be helpful in this task. – Pore-size dependence of transport properties. Self-diffusion and viscosity are pore size dependent: the smaller the pore, the more solid-like the behaviour of confined water. Stoke-Einstein equation relates viscosity and self-diffusion and is consistent with both MD and NEMD simulations, which constitutes a coherent set of data on the pore size dependency of these properties. This result establishes a fundamental link between diffusion and permeability in C-S-H. The low permeability of C-S-H, first quantified by Powers [34], can be explained with remarkable precision by this pore size dependence of both diffusion and viscosity, the reduction in the dimensionality of the flow, and the negative slip boundary conditions [13]. – Temperature dependence of transport properties. The multiscale modelling approach proposed were used to quantify the temperature dependence of diffusion and permeability under undrained conditions. Arrhenius thermal activation properly capture the temperature dependence of diffusion and viscosity at the molecular scale. Due to the validity of Stoke-Einstein equation, the temperature dependence of permeability can also be established at the molecular scale. The strategy can be extended in future work to deal also with drained conditions.
64
T. Honorio
References 1. Muller, A.C.A., Scrivener, K.L., Gajewicz, A.M., McDonald, P.J.: Densification of C-S–H measured by 1H NMR relaxometry. J. Phys. Chem. C 117, 403–412 (2012) 2. Abdolhosseini Qomi, M.J., Brochard, L., Honorio, T., et al.: Advances in atomistic modeling and understanding of drying shrinkage in cementitious materials. Cem. Concr. Res. 148, 106536 (2021) 3. IUPAC: Compendium of Chemical Terminology (the “Gold Book”). Compiled by McNaught, A.D., Wilkinson, A. XML on-line corrected version: http://goldbook.iupac.org (2006-) created by Nic, M., Jirat, J., Kosata, B.; updates compiled by Jenkins, A. 2nd ed. Blackwell Scientific Publications (1997) 4. Brovchenko, I., Oleinikova, A.: Interfacial and Confined Water. Elsevier (2008) 5. Einstein, A.: Investigations on the Theory of the Brownian Movement. Courier Corporation (1956) 6. Metzler, R., Jeon, J.-H., Cherstvy, A.G., Barkai, E.: Anomalous diffusion models and their properties: non-stationarity, non-ergodicity, and ageing at the centenary of single particle tracking. Phys. Chem. Chem. Phys. 16, 24128–24164 (2014) 7. Boon, J.P., Yip, S.: Molecular Hydrodynamics. Courier Corporation (1991) 8. Coasne, B.: Multiscale adsorption and transport in hierarchical porous materials. New. J. Chem 40, 4078–4094 (2016) 9. Ichikawa, T.: Theory of ionic diffusion in water-saturated porous solid with surface charge. J. Adv. Concr. Technol. 20, 430–443 (2022) 10. Maginn, E.J., Bell, A.T., Theodorou, D.N.: Transport diffusivity of methane in silicalite from equilibrium and nonequilibrium simulations. J. Phys. Chem. 97, 4173–4181 (1993) 11. Jobic, H., Theodorou, D.N.: Quasi-elastic neutron scattering and molecular dynamics simulation as complementary techniques for studying diffusion in zeolites. Microporous Mesoporous Mater. 102, 21–50 (2007) 12. Batchelor, C.K., Batchelor, G.K.: An Introduction to Fluid Dynamics. Cambridge University Press (2000) 13. Honorio, T.: Permeability of C-S-H. Preprint. Available at (2022) 14. Youssef, M., Pellenq, R.J.-M., Yildiz, B.: Glassy nature of water in an ultraconfining disordered material: the case of calcium−silicate−hydrate. J. Am. Chem. Soc. 133, 2499–2510 (2011) 15. Qomi, M.J.A., Bauchy, M., Ulm, F.-J., Pellenq, R.J.-M.: Anomalous composition-dependent dynamics of nanoconfined water in the interlayer of disordered calcium-silicates. J. Chem. Phys. 140, 054515 (2014) 16. Honorio, T., Carasek, H., Cascudo, O.: Water self-diffusion in C-S-H: effect of confinement and temperature studied by molecular dynamics. Cem. Concr. Res. 155, 106775 (2022) 17. Honorio, T., Masara, F., Benboudjema, F.: Heat capacity, isothermal compressibility, isosteric heat of adsorption and thermal expansion of water confined in C-S-H. Cement 6, 100015 (2021) 18. Kunhi Mohamed, A.: Atomistic simulations of the structure of calcium silicate hydrates: interlayer positions, water content and a general structural brick model. In: Infoscience. http:// infoscience.epfl.ch/record/256963. Accessed 3 Feb 2021 19. Cygan, R.T., Liang, J.-J., Kalinichev, A.G.: Molecular models of hydroxide, oxyhydroxide, and clay phases and the development of a general force field. J. Phys. Chem. B 108, 1255–1266 (2004) 20. Berendsen, H.J.C., Grigera, J.R., Straatsma, T.P.: The missing term in effective pair potentials. J. Phys. Chem. 91, 6269–6271 (1987)
Einstein Explains Water Transport in C-S-H
65
21. Plimpton, S.: Fast parallel algorithms for short-range molecular dynamics. J. Comput. Phys. 117, 1–19 (1995) 22. Simonnin, P., Noetinger, B., Nieto-Draghi, C., et al.: Diffusion under confinement: hydrodynamic finite-size effects in simulation. J. Chem. Theory Comput. 13, 2881–2889 (2017) 23. Yeh, I.-C., Hummer, G.: Diffusion and electrophoretic mobility of single-stranded RNA from molecular dynamics simulations. Biophys. J. 86, 681–689 (2004) 24. Botan, A., Rotenberg, B., Marry, V., et al.: Hydrodynamics in clay nanopores. J. Phys. Chem. C 115, 16109–16115 (2011) 25. Torquato, S.: Random Heterogeneous Materials: Microstructure and Macroscopic Properties. Springer Science & Business Media (2002) 26. Hatta, H., Taya, M.: Effective thermal conductivity of a misoriented short fiber composite. J. Appl. Phys. 58, 2478–2486 (1985) 27. Honorio, T.: Monte Carlo molecular modeling of temperature and pressure effects on the interactions between crystalline calcium silicate hydrate layers. Langmuir 35, 3907–3916 (2019) 28. Stora, E., Bary, B., He, Q.-C.: On estimating the effective diffusive properties of hardened cement pastes. Transp. Porous Media. 73, 279–295 (2008) 29. Allen, M.P., Tildesley, D.J.: Computer Simulation of Liquids. Oxford University Press, New York (1989) 30. Constantinides, G., Ulm, F.-J.: The nanogranular nature of C-S–H. J. Mech. Phys. Solids 55, 64–90 (2007) 31. Li, H., Fratini, E., Chiang, W.-S., et al.: Dynamic behavior of hydration water in calciumsilicate-hydrate gel: a quasielastic neutron scattering spectroscopy investigation. Phys. Rev. E 86, 061505 (2012) 32. Fratini, E., Faraone, A., Ridi, F., et al.: Hydration water dynamics in tricalcium silicate pastes by time-resolved incoherent elastic neutron scattering. J. Phys. Chem. C 117, 7358–7364 (2013) 33. Korb, J.-P., Monteilhet, L., McDonald, P.J., Mitchell, J.: Microstructure and texture of hydrated cement-based materials: a proton field cycling relaxometry approach. Cem. Concr. Res. 37, 295–302 (2007) 34. Powers, T.C.: Structure and physical properties of hardened Portland cement paste. J. Am. Ceram. Soc. 41, 1–6 (1958)
Is Thermal Pressurization in C-S-H Relevant for Concrete Spalling? Fatima Masara(B)
, Tulio Honorio , and Farid Benboudjema
Université Paris-Saclay, CentraleSupélec, ENS Paris-Saclay, CNRS, LMPS - Laboratoire de Mécanique Paris-Saclay, 91190 Gif-Sur-Yvette, France [email protected]
Abstract. Upon heating, thermal pressurization of the fluid in a porous media may occur according to the poromechanical boundary conditions. This pressure build-up may exceed material strength leading to explosive spalling. Thermal pressurization is a mechanism evoked to explain fire spalling of concrete and is also relevant to applications such as well cement lining in petroleum engineering and nuclear waste disposal structures in which the material is subject to significant temperature changes. The contribution of water in C-S-H gel and interlayer pores to thermal pressurization is yet poorly understood. In this work, we study the evolution of the pressure in C-S-H for different pore sizes (micro-and mesopores) as a function of temperature in undrained and drained conditions using molecular simulations. For the drained case, we consider two different poro-mechanical conditions, the first one at 100% relative humidity (liquid saturated system), and the second one at a constant vapor pressure equal to 0.1 MPa. By analyzing the confining pressure, we show how confinement affects the pressure buildup in the three different poromechanical conditions. We also study water desorption in C-S-H interlayer pores and how confinement affects the liquid-gas water phase transition. We finally compare our model with other available data in the literature. Our collected data shows the importance of nanoscale processes to predict and understand thermal pressurization in cement-based materials. Keywords: Thermal pressurization · Confining pressure · C-S-H · Molecular simulations · Temperature
1 Introduction The pore fluid pressure may increase dramatically when subjected to high temperatures. “Thermal pressurization” is the term used to describe the increase in pore pressure when the temperature increases. Thermal pressurization in concrete causes the material to spall, which reduces the structure’s strength and service life. In other applications where the material is also subject to significant temperature variations, such as well cement lining in petroleum engineering and nuclear waste disposal structures, thermal pressurization is crucial [1–3]. To date, we still do not fully understand how water in interlayer pores and C-S-H gel contributes to thermal pressurization. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 A. J˛edrzejewska et al. (Eds.): SynerCrete 2023, RILEM Bookseries 43, pp. 66–75, 2023. https://doi.org/10.1007/978-3-031-33211-1_6
Is Thermal Pressurization in C-S-H Relevant
67
In this work, we aim to study thermal pressurization in C-S-H at the molecular scale. For this purpose, we perform molecular simulations for four C-S-H pore sizes from 13.7 up to 46.3 Å under three different heating conditions (one undrained, drained at constant partial pressure or constant relative humidity). We show that the evolution of the confining pressure in C-S-H interlayer micropores depends significantly on the pore size and the poromechanical condition. We also investigate water desorption and liquid-gas water phase change in the different pore sizes tested and compare the results with the bulk water ones. The outputs of this paper can give useful insights into thermal pressurization in cement-based materials and can also be used as an input for multiscale modeling.
2 Models and Methods 2.1 C-S-H Structure and Force Field As in the original C-S-H model adopted in this work (Kunhi Mohamed et al. [4]), two solid C-S-H layers with two slit pores separating the solid layers are built up. The solid layers are composed of stacks of calcium silicate, while the pores contain water molecules, calcium counterions, and hydroxides. The first pore has a constant interlayer spacing (13.7 Å interlayer distance as in Kunhi Mohamed et al. model [4]). The second pore has a variable interlayer spacing (four pore sizes are selected with interlayer distances from 13.7 up to 46.3 Å). The interlayer spacing d corresponds to the distance separating the two C-S-H layers along the z-axis. The molecular formula of CS-H is Ca1.67 SiO3.7 . nH2 O, with n standing for the variable number of water molecules according to the drained poromechanical conditions (Pr = 0.1 MPa or RH = 100% conditions) and the pore size. The ClayFF force field [5] and the SPC/E water model [6] combined identify the type of interactions between the different species present in the system. It should be noted that the same original structure (Kunhi Mohammed et al. [4]) and force field (ClayFF and SPC/E [5, 6]) were employed before in [7, 8]. 2.2 Simulation Details Two drained and one undrained heating conditions are simulated. For the drained case, we simulate a system either at a constant partial vapour pressure Pr or constant relative humidity RH. For one drained heating condition, we impose a constant partial vapor pressure Pr of 0.1 MPa (equivalent to 1 atm). The RH in this case varies according to the saturation vapor pressure P0 (RH = PP0r ). The saturation vapor pressure P0 (also known as the coexistence pressure) is calculated from the Clausius-Clapeyron equation with [9]: Hvap 1 1 (1) − P0 (T ) = Pi exp R Ti T where Pi and T i are the initial pressure and temperature, H vap is the enthalpy of vaporization, and R is the gas constant.
68
F. Masara et al.
In this work, we take a temperature T i of 300 K. The pressure Pi at 300 K is equal to 0.001 MPa. The enthalpy of vaporization H vap is equal to 44.2 kJ/mol (SPC/E) [9]. Simulations are performed for temperatures between 270 and 640 K; the latter value is close to the critical temperatures of the SPC/E water model (T c = 623–640 K [9–12]) and the experimental values by [13, 14] (T c = 639.5 K in [13] and T c = 647.1 K in [14]. For the other drained heating condition, we simulate a system at 100% RH (liquid saturated state), for which the vapor pressure Pr and the coexistence pressure P0 are equal. For the third case, we perform the simulations in undrained conditions. Unlike drained conditions, water desorption is prohibited in undrained conditions. In undrained conditions, (NVT) molecular dynamics simulations are performed, where N, V, and T refer to a constant number of atoms, volume, and temperature, respectively. For the two drained heating conditions, hybrid grand canonical Monte Carlo-molecular dynamics (GCMC-MD) simulations are carried out. The GCMC simulation phase is run in the μ VT ensemble for water molecules only (μ is the chemical potential of water and is obtained from the RH (μ − μ0 (T ) ≈ kB T ln(RH), where μ0 is the chemical potential of water in a saturated state, and kB is the Boltzmann constant). The MD simulation phase is run in the NVT ensemble for calcium counterions and hydroxides (the solid C-S-H layers are not time integrated during the MD phase to decrease the computation time). Such a hybrid system allows only water to move between the C-S-H pore and an external infinite reservoir filled with water until equilibrium is reached. The movement of the other particles between the two systems, the C-S-H pore and the external water reservoir is forbidden, which is necessary to keep the C-S-H pore electrically neutral since the reservoir is filled with water molecules only. Simulations are run with a 0.1 fs timestep under three dimensional-periodic boundary conditions. A temperature T i of 300 K is taken as an initial temperature, increased and decreased to 640 and 270 K respectively. For long-range electrostatic interactions, we use the Ewald summation method. For Lennard-Jones interactions, we apply tail corrections. To run the simulations, we work with Lammps [15]. To visualize the data, we use VMD (visual molecular dynamics). 2.3 Confining Pressure Computation The confining pressure here, is computed with the virial equation in the z-axis direction using [7, 8, 16]: Nmi vi2 + P= V
N i
ri .fi
(2)
V
Where mi and vi are the mass and the velocity of atom i, N is the number of atoms counting the periodic replicas, ri and fi are the position and the force acting on atom i.
Is Thermal Pressurization in C-S-H Relevant
69
3 Results and Discussion 3.1 Confining Pressure The confining pressure in C-S-H provides good details about the behavior of the phase upon temperature change (cohesion/repulsion, liquid-solid and liquid-gas water phase transitions) [17]. Figure 1 shows the confining pressure in the previously defined C-S-H pores and bulk water in the three different heating conditions. It should be noted that the confining pressure in C-S-H refers to the total pressure exerted by all the species present, which includes the solid particles and the electrolytes (water, calcium counterions and hydroxides). Drained (Pr = 0.1 MPa)
confining pressure [MPa]
Undrained 600
600
a
300
b
300
0
0 bulk water d=46.3 Å d=26.3 Å d=18.3 Å d=13.7 Å
-300 -600 -900 270
370
470 T [K]
570
-300 -600 -900 670
270
370
470 T [K]
570
670
confining pressure [MPa]
Drained (RH = 100%) 600
c
300 0 -300 -600 -900 270
370
470 T [K]
570
670
Fig. 1. Confining pressure of the different C-S-H pores and bulk water in the three poromechanical conditions.
In undrained conditions (Fig. 1a), the confining pressure increases significantly with temperature in the C-S-H pores and bulk water, which is expected since the water content remains constant (no desorption). The pressures shifting from negative to positive values as the temperature increases implies a disjoining behavior. The evolution of the confining pressure in undrained conditions doesn’t show a strong correlation with the pore size.
70
F. Masara et al.
The confining pressure in the 13.7 Å pore shows slightly larger values than in the other pores but possesses quite similar values to that of bulk water. The thermal pressurization coefficient, defined as the ratio of the confining pressure to the temperature, is computed for the C-S-H pores and bulk water and compared with other data in the literature. In the 46.3, 18.3, and 13.7 Å pores, the thermal pressurization coefficient possesses similar values (on average equal to 1.61 MPa/K). In the 26.3 Å pore, the thermal pressurization coefficient is lower (equal to 1.19 MPa/K). For bulk water, the thermal pressurization coefficient is higher than that in the C-S-H pores (equal to 1.78 MPa/K). The values of the thermal pressurization coefficient obtained here are on the order of magnitude of the experimental value of Ghabezloo (0.62 MPa/K [18]) and the simulation value of Bonnaud et al. (1.81 MPa/K [17]). In Pr = 0.1 MPa drained case (Fig. 1b), we observe a strong pore-size dependency of the confining pressure upon temperature change. In pores > 18.3 Å, the evolution of the confining pressure with the temperature shows three different trends with two notable transition points. Above ≈ 290 K, the confining pressure decreases with temperature to reach ≈ 0 MPa at 430 K in the 46.3 and 26.3 Å pores. Beyond 430 K, the confining pressure fluctuates around 0 MPa. The previous temperature is close to the temperature for which we observed a drop in the water content (Fig. 2a), that is, the cavitation temperature of confined water in these pores. The decrease in the confining pressure between 290–430 K in the 46.3 and 26.3 Å pores is caused by water desorption. In this temperature range, and before the cavitation temperature is reached, water is a liquid and expands when the temperature increases. The expansion of liquid water in drained conditions causes desorption of water molecules out of the pore and therefore, a decrease in the confining pressure. Below ≈ 290 K, the pressure decreases and reaches ≈ 400 MPa in the 46.3 Å pore and ≈ 600 MPa in the 26.3 Å pore at 270 K. In smaller pores (18.3 and 13.7 Å pores), the behavior of the confining pressure is quite different. Above a transition temperature (310 K for the 18.3 Å pore and 340 K for the 13.7 Å pore), the confining pressure decreases. The decrease in the confining pressure is also attributed to water desorption. Unlike larger pores, we observe here negative pressure values beyond ≈ 450 K. For the 13.7 Å pore, the confining pressure decreases more steeply compared to the 18.3 Å pore (more negative pressures are observed in the 13.7 Å pore). Another difference with pores > 18.3 Å is that here, we did not observe any transition in the confining pressure related to cavitation, instead, we only observed a shift from positive to negative pressure values when the temperature increases. Below the transition temperature, the pressure decreases slightly and ≈ 400 MPa in the 18.3 Å pore and ≈ 150 MPa in the 13.7 Å pore. As the pore size increases from 13.7 to 46.3 Å, the maximum confining pressure observed, shifts toward lower temperature values. We also note that this maximum confining pressure possesses higher magnitudes in larger pores. Since the decrease in the pressure beyond the maximum confining pressure is attributed to liquid water desorption in all pores, and since C-S-H is overall strongly hydrophilic, which makes water desorption more difficult in smaller pores, therefore the shift in the maximum confining pressure toward higher temperature values when the pore size decreases is reasonable. We also observed a similar tendency in the water sorption isotherms (Fig. 2b). For bulk water, we observe a similar trend as in the 46.3 Å pore with negative pressures reflecting a state of cohesion above ≈ 390 K. In contrast
Is Thermal Pressurization in C-S-H Relevant
71
to undrained conditions, we observed here large positive pressures at low temperatures and considerably lower pressures at intermediate and high temperatures. Given these differences, we assume that C-S-H will be susceptible to spalling when subjected to high temperatures in undrained conditions and when subjected to low temperatures in Pr = 0.1 MPa drained conditions. In RH = 100% drained case (Fig. 1c), the behavior of the confining pressure with the temperature also shows a clear pore size dependency as in the Pr = 0.1 MPa drained case. In the 46.3, 26.3 and 18.3 Å pores, the confining pressure increases with the temperature. Interestingly, the behavior of the confining pressure in these pores in the RH = 100% drained case is similar to the undrained behavior with slightly less steep slopes. This shows that C-S-H for these particular pore sizes will be susceptible to spalling due to heating as in the undrained case. In the 13.7 Å pore, the confining pressure shows a distinct behavior. Between 270 and 330 K, the confining pressure increases. From 330 to 480 K, the confining pressure remains approximately constant. Above 480 K, the confining pressure decreases and becomes negative beyond 580 K. The difference observed in the evolution of the confining pressure in the 13.7 Å compared to the other pore sizes might be due to effects of the pore surfaces that become more significant when the size of the pore decreases and therefore affects the values of the confining pressure. For bulk water, the confining pressure increases linearly with the temperature above 330 K. At 445 K, a shift from negative to positive pressures takes place, which indicates an expansion behavior. Below 330 K, the pressure increases slightly to reach approximately 0 MPa at 270 K. 3.2 Water Content Figure 2 shows the variation of the number of water molecules with temperature for the different C-S-H pore sizes and bulk water in the two drained heating conditions. In Pr = 0.1 MPa drained case (Fig. 2a), two different trends are observed according to the pore size. For pore sizes ≥ 26.3 Å, the evolution of the number of water molecules with the temperature is discontinuous (reflected as a sudden decay in the amount of water at a given temperature corresponding to the pore size). This sudden decay in the amount of water is due to cavitation, a phenomenon that occurs in pores of infinite dimensions (as in this work) and can occur in ink bottle pores due to a liquid-gas phase transition [19, 20]. In the 46.3 Å pores, cavitation occurs at 405 K. In the 26.3 Å pore, cavitation occurs at a higher temperature (425 K). The increase in the cavitation temperature when the pore size decreases means that desorption is more difficult in smaller pores, an effect mainly due to the overall hydrophilic nature of the two pore surfaces. For bulk water, cavitation occurs at 425 K, a higher temperature than in the 46.3 and 106.3 pore. In the smallest pores (18.3 and 13.7 Å pores), the evolution of the number of water molecules with the temperature is continuous (no cavitation). Such a transition from a discontinuous to a continuous behavior when the pore size decreases was also observed in previous works on C-S-H [8] and silica nanopores [21]. It should be noted that since we keep the solid layers rigid (they are not time integrated in the MD phase) and since we use nonreactive force field, the deformation of the solid layers upon to heating and any surface alteration due to temperature increase are not captured. C-S-H is known to change at high temperature with firstly OH recondensating into molecular water and, upon further
72
F. Masara et al.
temperature increase, amorphization and reconversion into unhydrated calcium silicate would finally take place. For the temperatures we consider here, some of these alteration would occur. For C-S-H with Ca/Si close to one of the C-S-H model adopted here, less than 20% of the OH ions would recondensate for temperatures reaching the critical temperature of water [22]. In order to assess the specific contribution of confining water in thermal pressurization, we impose a constant surface state in our simulations (i.e. the C-S-H surface is kept the same even at high temperatures and only interlayer water is desorbed, as shown in Fig. 3)). Future work can deal with the quantification of the influence of surface alteration with temperature.
Fig. 2. Confining pressure of the different C-S-H pores and bulk water in the three heating conditions.
In RH = 100% drained case (Fig. 2b), the evolution of the number of water molecules with the temperature is continuous in all the C-S-H pore sizes and bulk water. This indicates that there is no cavitation or a liquid-gas water phase change. Desorption of water here is possible due to the expansion of liquid water with the increasing temperature. In the 13.7 Å pore, water starts to desorb at a higher temperature than in the other pores, an aspect, due to again, the overall hydrophilic nature of the two pore surfaces. In Fig. 3, we show snapshots of the 46.3 Å pore at various temperatures during desorption in the two drained conditions. The atomic configurations of the 46.3 Å pore shows a cavity at 410 K in the Pr = 0.1 MPa drained case.
Is Thermal Pressurization in C-S-H Relevant
Drained (Pr = 0.1 MPa)
73
Drained (RH = 100%)
Bulk space surface adsorbed water
d=46.3 Å d=46.3 Å
Fig. 3. Atomic configurations of the 46.3 Å pore at various temperatures during desorption in the two drained conditions. Ca is intralayer calcium ions and Caw is interlayer calcium counterions.
4 Conclusions In this work, we perform molecular simulations to predict thermal pressurization in C-SH interlayer micropores and gel pores under three different poromechanical conditions. In undrained conditions, we observed a marked linear increase in the pressure upon heating in all the C-S-H pores and bulk water, because the water content remains constant. The slopes of the curves are similar in 13.7, 18.3, and 46.3 Å pores but possess larger values than in the 26.3 Å pore, as demonstrated from the thermal expansion coefficient. The thermal expansion coefficients obtained in this study are on the same order of magnitude as other experimental and simulation data from the literature. In drained RH = 100% conditions, we also observed an increase in the pressure upon heating for pore sizes greater than 13.7 Å. Confined water in C-S-H (interlayer or gel pores) contributes, therefore to thermal pressurization. In drained Pr = 0.1 MPa
74
F. Masara et al.
conditions, we observed a significant decrease in the pressure upon heating because of water desorption. The effect of poromechanical conditions is therefore not trivial: thermal pressurization is observed when the fluid is kept in liquid state but thermal negative pressurization can be observed upon vaporization (drained Pr = 0.1 MPa). In conclusion, we assume that C-S-H will spall when exposed to high temperatures due to thermal pressurization in undrained and RH = 100% drained conditions and when exposed to low temperatures in Pr = 0.1 MPa drained conditions due to the high pressures observed at low temperatures. These data can be helpful for multiscale modeling and for the prediction of thermal pressurization in cement-based materials. Acknowledgments. We thankfully appreciate the financial support of the French National Research Agency (ANR) via the project THEDESCO (ANR-19-CE22-0004-01).
References 1. Bu, Y., Chang, Z., Du, J., Liu, D.: Experimental study on the thermal expansion property and mechanical performance of oil well cement with carbonaceous admixtures. RSC Adv. 7(46), 29240–29254 (2017) 2. Poyet, S., Charles, S.: Temperature dependence of the sorption isotherms of cement-based materials: heat of sorption and Clausius-Clapeyron formula. Cem. Concr. Res. 39(11), 1060– 1067 (2009) 3. Bažant, Z.P., Kaplan, M.F.: Concrete at High Temperatures: Material Properties and Mathematical Models. Longman, Harlow, Essex (1996) 4. Kunhi Mohamed, A., Parker, S.C., Bowen, P., Galmarini, S.: An atomistic building block description of C-S-H - towards a realistic C-S-H model. Cem. Concr. Res. 107, 221–235 (2018) 5. Cygan, R.T., Liang, J.-J., Kalinichev, A.G.: Molecular models of hydroxide, oxyhydroxide, and clay phases and the development of a general force field. J. Phys. Chem. B 108(4), 1255–1266 (2004) 6. Berendsen, H.J.C., Grigera, J.R.: The missing term in effective pair potentials. J. Phys. Chem. 91(24), 6269–6271 (1987) 7. Honorio, T., Masara, F., Benboudjema, F.: Heat capacity, isothermal compressibility, isosteric heat of adsorption and thermal expansion of water confined in C-S-H. Cem. Concr. Res. 6, 100015 (2021) 8. Masara, F., Honorio, T., Benboudjema, F.: Sorption in C-S-H at the molecular level: disjoining pressures, effective interactions, hysteresis, and cavitation. Cem. Concr. Res. 164, 107047 (2023) 9. Fugel, M., Weiss, V.C.: A corresponding-states analysis of the liquid-vapor equilibrium properties of common water models. J. Chem. Phys. 146(6), 064505 (2017) 10. Alejandre, J., Tildesley, D.J., Chapela, G.A.: Molecular dynamics simulation of the rthobaric densities and surface tension of water. J. Chem. Phys. 102(11), 4574–4583 (1995) 11. Guissani, Y., Guillot, B.: A computer simulation study of the liquid–vapor coexistence curve of water. J. Chem. Phys. 98(10), 8221–8235 (1993) 12. Boulougouris, G.C., Economou, I.G., Theodorou, D.N.: Engineering a molecular model for water phase equilibrium over a wide temperature range. J. Phys. Chem. B 102(6), 1029–1035 (1998) 13. Perry, R.H., Green, D.W.: Perry’s Chemical Engineers’ Handbook. McGraw-Hill, New York, USA (1984)
Is Thermal Pressurization in C-S-H Relevant
75
14. Lemmon, E.W., Bell, I.H., Huber, M.L., McLinden, M.O.: NIST Standard Reference Database 23: Reference Fluid Thermodynamic and Transport Properties-REFPROP, Version 10.0, National Institute of Standards and Technology, Standard Reference Data Program, Gaithersburg (2018) 15. Plimpton, S.: Fast parallel algorithms for short-range molecular dynamics. J. Comput. Phys. 117(1), 1–19 (1995) 16. Abdolhosseini Qomi, M.J., Brochard, L., Honorio, T., Maruyama, I., Vandamme, M.: Advances in atomistic modeling and understanding of drying shrinkage in cementitious materials. Cem. Concr. Res. 148, 106536 (2021) 17. Bonnaud, P.A., et al.: Temperature dependence of nanoconfined water properties: application to cementitious materials. J. Phys. Chem. C 120(21), 11465–11480 (2016) 18. Ghabezloo, S.: Micromechanics analysis of thermal expansion and thermal pressurization of a hardened cement paste. Cem. Concr. Res. 41(5), 520–532 (2011) 19. Sarkisov, L., Monson, P.A.: Modeling of adsorption and desorption in pores of simple geometry using molecular dynamics. Langmuir 17(24), 7600–7604 (2001) 20. Vishnyakov, A., Neimark, A.V.: Monte Carlo simulation test of pore blocking effects. Langmuir 19(8), 3240–3247 (2003) 21. Bonnaud, P.A., Coasne, B., Pellenq, R.J.-M.: Molecular simulation of water confined in nanoporous silica. J. Phys.: Condens. Matter 22(28), 284110 (2010) 22. Zhang, Y., Zhou, Q., Ju, J.W., Bauchy, M.: New insights into the mechanism governing the elasticity of calcium silicate hydrate gels exposed to high temperature: a molecular dynamics study. Cem. Concr. Res. 141, 106333 (2021)
Multiscale Modeling of the Dielectric Response of C-S-H Sofiane Ait Hamadouche1,2(B) , Tulio Honorio1 , Thierry Bore3 Farid Benboudjema1 , Franck Daout2 , and Eric Vourc’h2
,
1 Université Paris-Saclay, CentraleSupélec, ENS Paris-Saclay, CNRS, LMPS – Laboratoire de
Mécanique Paris-Saclay, 91190 Gif-sur-Yvette, France [email protected] 2 SATIE, UMR CNRS 8029, ENS Paris-Saclay, Université Paris-Saclay, Gif-sur-Yvette, France 3 School of Civil Engineering, The University of Queensland, St Lucia, QLD 4072, Australia
Abstract. The interpretation of dielectric measurements in cement-based materials, as well as multiscale modeling strategies, requires the knowledge of the intrinsic permittivity of their various constituent phases. Calcium silicate hydrates (C-SH) is the major hydrated phase in concrete (when clinker is the main cement compound), but to date, its frequency-dependent complex dielectric response remains unknown. Direct experimental measurements of C-S-H intrinsic dielectric behavior are challenging due to the scales to be probed and the difficulties in isolating this component in cement systems. Molecular simulations arise as a helpful tool to provide reliable estimates of properties bottom-up. This study adopts a multiscale approach to estimate the complex dielectric response of C-S-H over a frequency range of [0; 100 GHz]. We perform molecular dynamics simulations to compute the frequency-dependent dielectric response of water in C-S-H using theoretical framework of Statistical Physics, which enables us to associate the microscopic decay in water polarization correlations to the dielectric response. Several configurations are considered by varying the interlayer distance, covering the range of pore sizes associated with interlayer pores and gel pores in C-S-H. The dielectric response is anisotropic and pore size dependent, as expected in layered materials. The results at the molecular scale are then used as inputs in a homogenization model to estimate the dielectric permittivity of C-S-H gel, which we compare with estimations obtained from inverse analysis based on Micromechanics. Our results are a valuable input for multiscale modeling of non-destructive testing and evaluation in cement-based materials. Keywords: Electromagnetic properties · Molecular Dynamics · Anisotropy · Permittivity · Homogenization
1 Introduction Electromagnetic measurement methods have been widely used to assess the hydric properties of concrete, properties which have an influence on its mechanical behaviour and durability [1]. Since water molecules are dipolar, they induce a higher dielectric permittivity than gases and solids, making these methods sensitive to water content. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 A. J˛edrzejewska et al. (Eds.): SynerCrete 2023, RILEM Bookseries 43, pp. 76–87, 2023. https://doi.org/10.1007/978-3-031-33211-1_7
Multiscale Modeling of the Dielectric Response of C-S-H
77
From cement paste to concrete scale, several experimental studies have investigated the dielectric permittivity of cement-based materials [2, 3]. As a function of composition, multiscale strategies have been used to model the complex frequency-dependent permittivity of cement-based materials [3, 4]. These approaches require, however, fundamental property data on constituent intrinsic behaviour that is lacking to date. The main contributor to the dielectric response of cement-based materials is the ionic aqueous solution present in the pores, with the water-water interaction contribution being predominant compared to the ion-ion and ion-water interactions [28]. There exist microporous phases in cement system (C-S-H, AF-phases) and their contribution to the dielectric response remains poorly understood. Numerous studies on the molecular scale show that water confined in nanometer sized pores behaves differently from capillary water [5]. Quantifying the contribution of these microporous phases will contribute to enhance the precision of multiscale modelling of the dielectric response of cement-based materials. In ordinary cement-based materials, calcium silicate hydrate (C-S-H) is the predominant hydrated phase. However, its frequency-dependent permittivity remains unknown, since the scales to be probed and the difficulties in isolating this component in cement systems make direct experimental measurements challenging. Molecular dynamics (MD) simulations are a reliable tool to estimate the dielectric permittivity of C-S-H. Masoumi et al. [6] used MD simulations to compute the static dielectric constant of water confined in C-S-H, and observed in-plane static dielectric constant ε|| ≈ exp (–hp /13.22), for a pore thickness hp in Å for C-S-H with Ca/Si of 1.7 and 1.1. The authors also observed a pore-size dependence of the static response. Performing a frequency-dependent dielectric response computation on C-S-H can assist unravelling the origins of the dielectric properties of the material. When the dielectric response of C-S-H is quantified with regard to pore size, dielectric methods can be used to investigate water content in the interlayer and gel pores. In this work, we adopt an interdisciplinary framework combining MD simulations, Micromechanics and dielectric relaxation modeling to get information on the behaviour of the main Civil Engineering material. MD simulations are performed to determine the dielectric response of confined water in C-S-H at the molecular scale, which is used as input in a homogenization model to estimate the frequency-dependent permittivity of C-S-H gel. The latter is then fitted using the relaxation model of Cole-Cole.
2 Molecular Models and Methods 2.1 Dielectric Response from Molecular Simulations Molecular dynamics simulations allow us to compute the time evolution of the total → (t), which is the sum of the dipole moment − dipole moment M μi (t) of each particle i with a charge qi . For a system with N particles: (t) = M
N i
− → μi (t) =
N i
→ qi − ri (t)
→ where − ri (t) is the position of the center of the particle i.
(1)
78
S. A. Hamadouche et al.
The dielectric response of water results from microscopic relaxation processes described by the auto-correlation function of water polarization φ ij (t) (for i, j = x, y, z) [7]: Mi (0)Mj (t) (2) φij (t) = Vcav kB T ε0 where k B is the Boltzmann constant, T is the temperature (in Kelvin), ε0 is the vacuum permittivity, and V cav is the volume of the pore occupied by confined water. Here, this volume is computed using the sum of the Voronoï volume of the oxygen atoms in water molecules. The frequency-dependence is computed via the Fourier transform of the components of the susceptibility tensor [8]. The susceptibility of water can be written as follows [7]: ∞ e−2π ift φij (t)dt (3) χij (t) = φij (0) − 2π i 0
where f is the frequency. The dielectric permittivity tensor ε can be computed directly from the susceptibility tensor. In a slit pore with an infinite slab of fluid confined between two walls, the dielectric tensor can be written as a function of the parallel ε|| and perpendicular ε⊥ components [8]: ⎛ ⎞ 0 ε|| (f ) 0 ε(f ) = ⎝ 0 ε|| (f ) 0 ⎠ (4) 0 0 ε⊥ (f ) in a Cartesian frame so that the parallel component is aligned with the slit pore plane. The frequency-dependent components of the dielectric permittivity tensor are obtained as [8]: ε|| (f ) = 1 + χ|| (f )
(5)
ε⊥ (f ) = 1 − χ⊥ (f )
(6)
In the case of C-S-H, the in-plane response can be approximated as ε|| (f) ≈ εxx (f) ≈ εyy (f). Due to translational invariance, ε⊥ is expected to be negligible. Also, due to the slit geometry, the out-of-diagonal terms are expected to be εxy (f) = εyx (f) ≈ 0. In this study we will focus only on the in-plane response. 2.2 Simulation Details The atomistic C-S-H model we use is based on Kunhi et al. [9] model with molecular formula Ca1.67 SiO3.7 . nH2 O. The Ca/Si of 1.67 is a common value observed in C-S-H [10, 11]. To describe the interatomic interactions, we use ClayFF [12] and SPC/E water [13]. An analysis of the structural features of the C-S-H model combined with this force field is provided in ref. [14] together with stability analysis under sorption. To study the influence of confinement, we consider various interlayer pore sizes ranging from 11.8 to 106.3 Å (Fig. 1 (A)), which spans the sizes associated with interlayer
Multiscale Modeling of the Dielectric Response of C-S-H
79
and gel pores in C-S-H as in [15]. Grand Canonical Monte Carlo (GCMC) simulations were performed to fill the system with the water content corresponding to ambient conditions and (liquid) saturated conditions (T = 300~K and RH = 100%) (see ref. [14] for details) Note that saturated nanometer pore sizes is a relevant condition to a broader range of RH since dehydrating such small pore would require a very low RH. Two layers are simulated, therefore two pore spaces are present in the simulated system. One pore is kept at the equilibrium basal spacing of 13.7 Å, while the other is changed (see Fig. 1(A)). The following analysis deals with water present in only this second pore.
Fig. 1. Multi-scale strategy adopted in this work.
We use LAMMPS [16] software for MD simulations. Only the water and ions in the interlayer space are time integrated during MD to gain computational efficiency and reduce fluctuations in the computation of φ || (t). For the integration of φ || (t) in Eq. (3), the upper integration limit must be chosen so that the function decay is integrated correctly by avoiding the integration of noise present in the signal from the simulation [17]. Since MD simulations are based on finite differences algorithms, very long runs are prone to accumulating numerical errors, resulting in noise present in the spectra of φ || (t). To avoid the integration of that noise and to reduce the computational time needed to compute Fourier transforms in large lists, we use a polynomial exponentially decaying aiming to capture potential oscillations in the φ || (t) profiles: Poly
φ||
(t) = εs exp[
−t ](b0 + b1 t + b2 t 2 + b3 t 3 ) τPoly
(7)
where εs is the static dielectric permittivity, τ poly the relaxation time, t the time, and b0 , b1 , b2 , b3 are fitting coefficients.
80
S. A. Hamadouche et al.
3 Upscaling the Dielectric Permittivity We adopt a microstructure representation of C-S-H gel, assuming that it is composed of two phases: microporous solid particles and gel porosity. The former is depicted as a transverse isotropic inclusion composed of C-S-H layers stacked according to the equilibrium basal spacing d eq = 13.7 Å, which should be the most prevalent configuration in stacked systems [14]. According to [18], confined water shows a transverse isotropic behaviour up to 100 nm. The authors studied the dielectric response of water molecules confined between hydrophobic surfaces, under the influence of external electric fields. Since the nature of the surfaces between which water molecules are confined have little influence on the dielectric response [19], we can assume that ε|| (f) = ε⊥ (f) in the gel porosity of our hydrophilic system. Various conventions are used in the literature to classify the different types of porosity in C-S-H according to their sizes. Jennings [20] proposed two characteristic pore sizes for C-S-H gel porosity from small-angle scattering experimental data: small gel pores [1–3 nm] and larger gel pores [3–12 nm]. In this study, to upscale C-S-H gel properties, we adopt three characteristic pore diameters in the average range of pore sizes: 2, 4 and 10 nm. Thus, we use as input the information obtained from molecular simulations regarding the interlayer distances of 1.37 nm (= d eq ) for the solid particles and 2, 4, and 10 nm for the gel pore (Fig. 1(B)). These inputs are used in the following homogenization model: ε
hom
=
7(1−fg )ε||,s ε||,s +2εhom 7(1−fg ) ε||,s +2εhom
+ +
5(1−fg )ε⊥ ε⊥,s +2εhom 5(1−fg ) ε⊥,s +2εhom
− −
7fg ε||,s ε||,g +2εhom 7fg ε||,g +2εhom
− −
5fg ε⊥,g ε⊥,g +2εhom 5fg ε⊥,g +2εhom
(8)
where f g represents the gel porosity, ε|| = εx,p = εy,p, εz,p = ε⊥,p the subscripts, s and,g stand, respectively, for the solid C-S-H particle and gel porosity. More detail about the homogenization model used here are available in [15].
4 Results 4.1 Static Dielectric Response at the Molecular Scale Figure 2 shows the in-plane static response ε|| of water confined in C-S-H as a function of the interlayer distance, along with the numerical and experimental values of bulk water for comparison. As reported by Masoumi et al. [6], the dielectric constant of water confined in C-S-H depends on the pore size and exhibits bulk-like behavior when the interlayer spacing exceeds 6 nm, while it decreases with the confinement of water. The decrease of the dielectric constant is a result of the forces that hinder polarization of water molecules near solid surfaces. To obtain the numerical value of bulk water, we computed the dielectric constant of (rigid) bulk SPC/E water also under ambient conditions (T = 300 K). We obtained ε = 71.2 ± 2.11, which is in agreement with other studies using simulations of SPC/E model [21, 22], and is close to the experimental value ε = 78.02 ± 0.20 (interpolated value for
Multiscale Modeling of the Dielectric Response of C-S-H
81
T = 300 K from the experimental study of [23]). We fit the in-plane response with the following expression:
h−X (9) ε|| = εbulk exp − Lcharac where εbulk is the dielectric constant of bulk water computed using MD simulations (= 71.2), h the interlayer distance, X = 7.1 Å is a shift in the interlayer distance, and L charac = 20 Å is the characteristic distance associated with the pore size dependence of the dielectric response. When h >> L charac , the static dielectric response of C-S-H approaches that of bulk water. Using a similar equation to fit their results, Masoumi et al. [6] obtained L charac = 13.22 Å. The authors computed the static response with a rigid SPC water model, which is known to yield a lower dielectric constant when compared to rigid SPC/E water model. For the same temperature, the authors obtained a static response ε|| = 64 for bulk water.
Fig. 2. Static response of water molecules confined in C-S-H as a function of the interlayer distance.
4.2 Frequency-Dependent Response at the Molecular Scale Based on the integration of the correlation function in Eq. (3), the frequency-dependent permittivity of confined water in C-S-H is illustrated in Fig. 3. To cope with the noise in MD results and gain in precision and efficiency in the computation of integrals in Eq. (3), we fit φ || (t) using Eq. (7). Figure 3 shows both the real and imaginary parts of ε|| (f), as well as the isotropic response of bulk water based on experimental and numerical (using SPC/E model) values reported by Buchner et al.[24] and Rinne et al.[25] respectively. The bumps that can be seen on the dielectric response may not necessarily have a physical meaning. The real part have negative values around 1 GHz for d = 13.7 Å, whilst the dielectric permittivity of common materials is generally non-negative [26]. As suggested
82
S. A. Hamadouche et al.
by [27], the relaxation frequencies f r of bound water are lower than that of bulk water, since the resonance frequency decreases as we increase the interlayer distance in the imaginary part.
___ ---
Bulk water
Exp SPC/E
Confined water h = 106.3 Å h = 76.3 Å h = 46.3 Å h = 26.3 Å h = 21.3 Å h = 16.3 Å h = 16.3 Å
Fig. 3. Frequency-dependent dielectric response of water molecules confined in C-S-H for different interlayer distances. The response of bulk water corresponds to experimental values according to [24], and simulations values using SPC/E water according to [25]. The dielectric responses regarding each interlayer distance are shifted 50 units upwards for readability.
The results obtained so far can now be used to get an expression of the pore size dependence of the complex frequency-dependent dielectric response in C-S-H. Cole-Cole model [28], which is one of the most used relaxation models, shows a good performance in fitting ε|| (f). With this model, the frequency dependence in the response reads: ε∗ = ε∞ +
εcc − ε∞ 1 + (i2π f τcc )1−αcc
(10)
where εcc and ε∞ are respectively the static and infinite frequency dielectric permittivity, τ cc the relaxation time and α cc the exponent parameter which has a value between -1 and 1 (with α = 0 corresponding to a Debye model). For εcc , we fixed the values of the static response obtained in Sect. 4.1 using Eq. (9) and ε∞ is taken equal to 1, since the force fields used in MD simulations do not include atomic polarization [29]. The parameters τ cc and α cc were determined by fitting (using least squares methods) ε|| (f) with the Cole-Cole model. The Cole-Cole model does not fit the negative values, nor the bumps observed in the ε|| (f) dielectric spectra. The values obtained for τ cc and α cc
Multiscale Modeling of the Dielectric Response of C-S-H
were fitted with expressions depending on the interlayer distance:
h − B0 + τ∞ τcc = B exp − Lcharac
83
(11)
αcc = −B1 (1 − exp[−B2 (h − B3 )])2 + B4
(12)
where h is the interlayer distance, L charac = 2.90 Å is the characteristic distance, τ ∞ = 14 ps is the value towards which the data converge, and B, B0 , B1 , B2 , B3 , B4 are other fitting coefficients whose values are reported in Table 1. Table 1. Coefficients of the fitting equations of τcc and αcc at the molecular scale. Coefficients
B [ps]
B0 [Å]
B1 [-]
B2 [1/ Å]
B3 [Å]
B4 [-]
Values
12655
3.330
0.589
0.307
16.046
0.673
4.3 Dielectric Response of C-S-H Gel We estimate the static and frequency-dependent permittivity of C-S-H gel using Eq. (8) in accordance with the representation of the C-S-H gel structure detailed in Sect. 3. The microporous solid particles as well as the gel porosity are represented as transverse isotropic inclusions. The results were obtained for different packing densities η = 1– f g . Table 1 summarizes the values of the static response for η = 0.64 and η = 0.74, corresponding respectively to low density (LD) and high density (HD) C-S-H, the two main packing densities of C-S-H [30]. Our values are in agreement with Guihard et al.[1] reporting a static dielectric permittivity of C-S-H in the range 2–20. In this reference, these values are obtained through inverse analysis of experimental data in the frequency range [200 MHz; 1GHz], assuming that the dielectric response of C-S-H is a real value that remains constant over the studied frequency range. Our results however show that some frequency dependency in C-S-H would arise, even at this relatively low frequency range, due to the contribution of interlayer pores. Table 2. Static permittivity of LD and HD C-S-H. dg = 2 nm
dg = 4 nm
dg = 10 nm
εgel (0) (LD)
8.75
10.40
11.08
εgel (0) (HD)
8.30
9.36
9.77
Figure 4 depicts the real and imaginary parts of εgel (f) for both LD (Fig. 4A) and HD (Fig. 4B) C-S-H for the three characteristic gel pore sizes considered here (2, 4, and 10 nm). We observe that the dielectric response at the gel scale also depends on the gel
84
S. A. Hamadouche et al.
pore size. Nevertheless, it does not have a major influence on the resonance frequency, which seems to be approximately the same whatever pore size considered. The dielectric response of C-S-H gel is fitted using a Cole-Cole model. As at the molecular scale, we fix the values of the static response (Table 2) for εcc . The two remaining Cole-Cole parameters (τ cc and α cc ) are summarized in Table 3.
A
B
Fig. 4. Frequency-dependent dielectric response of: (A) LD C-S-H, and (B) HD C-S-H.
Table 3. Cole-Cole parameters of LD and HD C-S-H. dg = 2 nm
dg = 4 nm
dg = 10 nm
τcc [ps] (LD)
281
245
251
αcc [-] (LD)
0.024
0.048
0.051
τcc [ps] (HD)
312
293
299
αcc [-] (HD)
0.018
0.016
0.016
5 Conclusions In this work, we used an interdisciplinary and multiscale approach to compute the dielectric response of water confined in C-S-H. At the molecular scale, molecular dynamics simulation were performed. The results were then used as input in a homogenization
Multiscale Modeling of the Dielectric Response of C-S-H
85
model in order to get the response at the gel scale. We estimate the intrinsic frequencydependent permittivity of C-S-H gel for the first time, providing important input for multiscale modelling of dielectric permittivity of cement-based materials. The main conclusions are: • Dielectric response of confined water in C-S-H at the molecular scale: Water confined in C-S-H shows a decreasing dielectric response with confinement, retrieving the dielectric response of bulk water for interlayer distances above roughly 6 nm. The Cole-Cole model is used to fit the frequency-dependent dielectric permittivity of confined water in C-S-H in order to capture its pore size dependence. • Dielectric response of C-S-H gel: We quantify the complex dielectric response of LD and HD C-S-H, which is input for multiscale approach estimating the dielectric response of cement-based materials. The values of the static response of LD and HD C-S-H are in the range of 2–20 reported in experimental studies [1]. In our analysis, we considered various gel pore sizes. The pore size considered for the gel porosity has an influence on the dielectric response, with an increase of the static response when the pore size increase, while the resonance frequency does not change much. Our results provide, therefore, an estimation of the frequency-dependent dielectric permittivity of LD and HD C-S-H. In future micromechanical modelling of the dielectric properties of cement-based materials, C-S-H must be considered as a phase showing a complex frequency-dependent behaviour. The next steps will focus on the effect of the relative humidity on the dielectric response of C-S-H.
References 1. Guihard, V., Patapy, C., Sanahuja, J., Balayssac, J.-P., Taillade, F., Steck, B.: Effective medium theories in electromagnetism for the prediction of water content in cement pastes. Int. J. Eng. Sci. 150, 103273 (2020). https://doi.org/10.1016/j.ijengsci.2020.103273 2. Guihard, V., Taillade, F., Balayssac, J.-P., Steck, B., Sanahuja, J., Deby, F.: Permittivity measurement of cementitious materials with an open-ended coaxial probe. Constr. Build. Mater. 230, 116946 (2020). https://doi.org/10.1016/j.conbuildmat.2019.116946 3. Camp, P.R., Bilotta, S.: Dielectric properties of portland cement paste as a function of time since mixing. J. Appl. Phys. 66(12), 6007–6013 (1989). https://doi.org/10.1063/1.343577 4. Guihard, V., Taillade, F., Balayssac, J.P., Steck, B., Sanahuja, J., Deby, F.: Prediction of cement-based materials’ water content with the use of electromagnetic homogenisation schemes. In: 2018 Progress in Electromagnetics Research Symposium (PIERS-Toyama), Aug. 2018, pp. 164–168. https://doi.org/10.23919/PIERS.2018.8597698 5. Abdolhosseini Qomi, M.J., Brochard, L., Honorio, T., Maruyama, I., Vandamme, M.: Advances in atomistic modeling and understanding of drying shrinkage in cementitious materials. Cem. Concr. Res. 148, 106536 (2021). https://doi.org/10.1016/j.cemconres.2021. 106536 6. Masoumi, S., Zare, S., Valipour, H., AbdolhosseiniQomi, M.J.: Effective interactions between calcium-silicate-hydrate nanolayers. J. Phys. Chem. C (2019). https://doi.org/10.1021/acs. jpcc.8b08146 7. Rinne, K.F., Gekle, S., Netz, R.R.: Ion-specific solvation water dynamics: single water versus collective water effects. J. Phys. Chem. A 118(50) (2014). https://doi.org/10.1021/jp5066874
86
S. A. Hamadouche et al.
8. Froltsov, V.A., Klapp, S.H.L.: Dielectric response of polar liquids in narrow slit pores. J. Chem. Phys. 126(11), 114703 (2007). https://doi.org/10.1063/1.2566913 9. Kunhi Mohamed, A., Parker, S.C., Bowen, P., Galmarini, S.: An atomistic building block description of C-S-H – Towards a realistic C-S-H model. Cem. Concr. Res. 107, 221–235 (2018) 10. Richardson, I.G.: Tobermorite/jennite- and tobermorite/calcium hydroxide-based models for the structure of C-S-H: applicability to hardened pastes of tricalcium silicate, β-dicalcium silicate, Portland cement, and blends of Portland cement with blast-furnace slag, metakaolin, or silica fume. Cem. Concr. Res. 34(9), 1733–1777 (2004) 11. Allen, A.J., Thomas, J.J., Jennings, H.M.: Composition and density of nanoscale calcium– silicate–hydrate in cement. Nat. Mater. 6(4), 311–316 (2007) 12. Cygan, R.T., Liang, J.-J., Kalinichev, A.G.: Molecular models of hydroxide, oxyhydroxide, and clay phases and the development of a general force field. J. Phys. Chem. B 108(4), 1255–1266 (2004) 13. Berendsen, H.J.C., Grigera, J.R., Straatsma, T.P.: The missing term in effective pair potentials. J. Phys. Chem. 91(24), 6269–6271 (1987) 14. T. Honorio, F. Masara, and F. Benboudjema, “Heat capacity, isothermal compressibility, isosteric heat of adsorption and thermal expansion of water confined in C-S-H,” Cement, p. 100015, Oct. 2021 15. Honorio, T., Carasek, H., Cascudo, O.: Water self-diffusion in C-S-H: effect of confinement and temperature studied by molecular dynamics. Cem. Concr. Res. 155, 106775 (2022). https://doi.org/10.1016/j.cemconres.2022.106775 16. Plimpton, S.: Fast parallel algorithms for short-range molecular dynamics. J. Comput. Phys. 117(1), 1–19 (1995). https://doi.org/10.1006/jcph.1995.1039 17. Ryckaert, J.-P., Ciccotti, G., Berendsen, H.J.C.: Numerical integration of the Cartesian equations of motion of a system with constraints: molecular dynamics of n-alkanes. J. Comput. Phys. 23(3), 327–341 (1977). https://doi.org/10.1016/0021-9991(77)90098-5 18. De Luca, S., Kannam, S.K., Todd, B.D., Frascoli, F., Hansen, J.S., Daivis, P.J.: Effects of confinement on the dielectric response of water extends up to mesoscale dimensions. Langmuir 32(19), 4765–4773 (2016). https://doi.org/10.1021/acs.langmuir.6b00791 19. Zhang, C., Gygi, F., Galli, G.: Strongly anisotropic dielectric relaxation of water at the nanoscale. J. Phys. Chem. Lett. 4(15), 2477–2481 (2013). https://doi.org/10.1021/jz401108n 20. Jennings, H.M.: Refinements to colloid model of C-S-H in cement: CM-II. Cem. Concr. Res. 38(3), 275–289 (2008). https://doi.org/10.1016/j.cemconres.2007.10.006 21. Rami Reddy, M., Berkowitz, M.: The dielectric constant of SPC/E water. Chem. Phys. Lett. 155(2), 173–176 (1989). https://doi.org/10.1016/0009-2614(89)85344-8 22. Fuentes-Azcatl, R., Mendoza, N., Alejandre, J.: Improved SPC force field of water based on the dielectric constant: SPC/ε. Phys. Stat. Mech. Appl. 420, 116–123 (2015). https://doi.org/ 10.1016/j.physa.2014.10.072 23. Gregory, A.P., Clarke, R.N.: Traceable measurements of the static permittivity of dielectric reference liquids over the temperature range 5–50 °C. Meas. Sci. Technol. 16(7), 1506 (2005). https://doi.org/10.1088/0957-0233/16/7/013 24. Buchner, R., Hefter, G.T., May, P.M.: Dielectric relaxation of aqueous NaCl solutions. J. Phys. Chem. A 103(1), 1–9 (1999). https://doi.org/10.1021/jp982977k 25. Rinne, K.F., Gekle, S., Netz, R.R.: Dissecting ion-specific dielectric spectra of sodium-halide solutions into solvation water and ionic contributions. J. Chem. Phys. 141(21), 214502 (2014). https://doi.org/10.1063/1.4901927 26. Fan, G., Sun, K., Hou, Q., Wang, Z., Liu, Y., Fan, R.: Epsilon-negative media from the viewpoint of materials science. EPJ Appl. Metamaterials 8, 11 (2021). https://doi.org/10. 1051/epjam/2021005
Multiscale Modeling of the Dielectric Response of C-S-H
87
27. Dobson, M.C., Ulaby, F.T., Hallikainen, M.T., El-rayes, M.A.: Microwave dielectric behavior of wet soil-part II: dielectric mixing models. In: IEEE Trans. Geosci. Remote Sens. 23(1), 35–46 (1985). https://doi.org/10.1109/TGRS.1985.289498 28. Cole, K.S., Cole, R.H.: Dispersion and Absorption in Dielectrics I. Alternating Current Characteristics. J. Chem. Phys. 9(4), 341–351 (1941). https://doi.org/10.1063/1.1750906 29. Honorio, T., Bore, T., Benboudjema, F., Vourc’h, E., Ferhat, M.: Dielectric properties of the pore solution in cement-based materials. J. Mol. Liq. 302, 112548 (2020) 30. Constantinides, G., Ulm, F.-J.: The effect of two types of C-S-H on the elasticity of cementbased materials: results from nanoindentation and micromechanical modeling. Cem. Concr. Res. 34(1), 67–80 (2004). https://doi.org/10.1016/S0008-8846(03)00230-8
Quantum Mechanically Informed Kinetic Monte Carlo Models for Hydrogen Diffusion in BCC-Iron Gonzalo Álvarez1
, Alvaro Ridruejo1(B)
, and Javier Sánchez2
1 Department of Materials Science. E.T.S.I. Caminos, Universidad Politécnica de Madrid,
28040 Madrid, Spain [email protected] 2 Instituto de Ciencias de la Construcción “Eduardo Torroja” (IETcc-CSIC), Madrid, Spain
Abstract. Hydrogen embrittlement is one of the main causes of catastrophic failure of structural components made of carbon steel. Among the underlying physical mechanisms associated with this complex phenomenon, hydrogen diffusion as an atomic interstitial in the body - centered cubic (BCC) lattice of pure iron plays a fundamental role. The main goal of this study is to characterize the fundamental events that controls the diffusion of hydrogen: the atomic jumps among stable or metastable lattice points (tetrahedral and octahedral sites). To this end, the best technique available is density functional theory (DFT), which is able to determine from first principles the atomic configuration and the energy landscape associated to the presence of hydrogen in the lattice. In this work, the strategies employed so far to obtain the jump parameters are reviewed, and a recently developed technique (Linear synchronous transient and Quadratic synchronous transient method) has been applied in order to improve the accuracy of previous results. Keywords: Hydrogen diffusion · BCC-Iron · DFT calculations · Ab-initio characterization · Hydrogen embrittlement
1 Introduction Hydrogen embrittlement (HE) is a process by which the toughness of a material decreases sharply in the presence of hydrogen. Due to its economic and social impact, hydrogen embrittlement has been studied since its formulation at the end of the 19th century [1, 2], resulting in a thorough comprehension of its macroscopic effects. However, a general consensus on the microscopic processes involved in the development of hydrogen embrittlement has not been achieved [3–7]. Several different currently coexist in the literature describing hydrogen embrittlement, all of them supported by several empirical observations. The following models should be mentioned: - The HIPT (Hydrogen Induced Phase Transformation) model is the only model with general consensus on the effective microscopic mechanisms governing embrittlement on its covered conditions. HIPT model explains embrittlement process through the formation of different phases, with drastically different properties in the presence of hydrogen © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 A. J˛edrzejewska et al. (Eds.): SynerCrete 2023, RILEM Bookseries 43, pp. 88–95, 2023. https://doi.org/10.1007/978-3-031-33211-1_8
Quantum Mechanically Informed Kinetic Monte Carlo Models
89
[8, 9]. Due to its nature, this model only covers those materials susceptible to phase transformation in the presence of hydrogen. - The HEDE (Hydrogen Enhanced DEcohesion) model was the first model proposed to describe hydrogen embrittlement by Pfeil on 1926 [10]. HEDE model is based on the exclusive presence of the HID (Hydrogen Induced Decohesion) mechanism, by which interstitial hydrogen weakens metallic bonding resulting in a significant decrease of cohesive energy of the material [11, 12]. - The HELP (Hydrogen Enhanced Localized Plasticity) model was first suggested by Beachem in 1972 [13]. HELP model explains embrittlement in the absence of decohesion mechanisms, due to the promotion of very local plasticity mechanisms, such as LEHDM (Local Hydrogen Enhanced Dislocation Motion) [14], AIDE (Adsorption Induced Dislocation Emission) [4], HESIV (Hydrogen Enhanced Strain-Induced Vacancy formation) [15] and DEFACTANT (DEFect ACTing AgeNT) [16] mechanisms. HELP model attributes the reduction of the fracture toughness of the material to a reduction of the volume involved in the fracture process instead of the lack of plastic deformation. - The HELP + HEDE model was first experimentally observed by Wang in 2009 [17]. This model assumes that both decohesion and local plasticity mechanisms are together responsible for the embrittlement phenomenon. This model is based on the negative selffeedback loop present in several embrittlement mechanisms due to the local distortion of the hydrogen environment, and the possibility of positive cross-feedback between mechanisms [18]. - The HELP mediated HEDE model was initially proposed by Novak et al. in 2010 [19]. This model suggests that initially a local plasticity mechanism is triggered due to the initial local hydrogen, and a posterior decohesion mechanism initialized due to the change in the local environment due to the plastic mechanism [20]. The main difficulty on the understanding of hydrogen embrittlement arises from its multiscale nature, spanning dissimilar length and time scales. Experimental characterization of this phenomenon is unable to cover the smaller scales due to a combination of the following factors: The low susceptibility of hydrogen for several techniques, the high mobility of hydrogen atom even at cryogenic temperatures, a non-negligible difference of behaviour between bulk and surface states and the significant impact of any sample preparation technique on hydrogen distribution. The numerical approach to hydrogen embrittlement therefore has to cover the wide range of scales influencing the phenomenon and its multi-physical coupled nature. This combination limits the use of standard simulation techniques to the characterization of individual elements involved in the degradation process. Over the last decades, special attention has been given to kinetic Monte Carlo (kMC) models to study the atomic scale of hydrogen embrittlement, due to their capability to deal with events with very different characteristic times [21], with a highlighted interest on quantum mechanically informed kinetic Monte Carlo models (Q-kMC). All kMC models are heavily parametrized and depend on the event rates prescribed, which may be physically based or not. Q-kMC models are characterized by a quantum mechanical computation of the different rates governing the system under study. These rates usually arise from ab-initio techniques such as density functional theory (DFT) calculations, and
90
G. Álvarez et al.
their reliability depends on the quality of the calculations [22, 23]. In this particular case, the simulation of hydrogen diffusion in iron with a Q-kMC model depends only (and critically) on a correct determination of the parameters controlling the discrete motion steps of hydrogen atoms. The purpose of this paper is to characterize the chemical potential landscape of interstitial hydrogen on the body centered cubic lattice of iron to properly characterize the physical parameters affecting diffusion, in both a defect free and vacancy containing lattice. These physical parameters are required to calculate a micro-structure dependent diffusion coefficient by the use of kMC models. The local diffusion coefficient can be further employed by homogenization technique to asses hydrogen response for steel components.
2 Hydrogen as an Interstitial in BCC-Iron Diffusion of hydrogen in the BCC-Iron is an interstitial phenomenon in which hydrogen atoms “jump” in between interstitial sites. The diffusion of hydrogen in BCC iron is a thermally activated process with negligible tunnelling contribution at room temperature. DFT calculations have shown a that only tetrahedral (T) sites are stable sites for hydrogen under equilibrium conditions. The transitions between T sites may occur either through the first-degree saddle (S) point (T-S-T) or through the octahedral (O) site (T-O-T), which is a second-degree saddle point on the BCC-cell (Fig. 1).
Fig. 1. Schematic of BCC-Fe cell with marked high symmetry sites. Octahedral sites (O) (red), tetrahedral sites (T) (Pink) and first degree saddle point (S) (blue). a) All interstitial sites in a cell. b) Representation of high symmetry points on a face
2.1 Jump Parameters As hydrogen diffusion is a thermally activated process, it can be described by an Arrhenius equation, in which both the attempt frequency ( 0 ) and the internal energy barrier (EB ) are temperature (T) dependent Eq. (1). = o (T )e
−E B (T ) kB T
(1)
Quantum Mechanically Informed Kinetic Monte Carlo Models
91
where kB stands for Boltzmann constant. The characterization of the diffusion phenomena is therefore reduced to the identification of the different possible transitions and the characterization of both their attempt frequencies and their energy barriers. Attempt Frequency: The evaluation of the attempt frequency for a first-degree saddle point can be computed using harmonic transition state theory from the vibrational frequency (ν) in the starting configuration (A) and the first-degree saddle point (S) using Eqs. (2) and (3). [24] 3N +3 o (T ) =
i=1 υA,i 3N +2 i=1 υS,i
3N +3 j=1
f (2π υA,j /kB T )
j=1
f (2π υS,j /kB T )
3N +2
f (x) = sinh(x)/x
(2) (3)
where is the reduced Plank constant. In the cases where the vibrational energy at the saddle point (2πνS ) is much greater than the thermal energy (kB T), Eqs. (2) and (3) can be simplified to Eq. (4) [24]. ⎤ ⎡ 3N +3 3N +3−D − 21 j=1 2π υA,j + 21 j=1 2π υS,j kB T ⎦ (4) exp⎣− 0 (T ) = 2π kB T Here, D stands for the degree of the saddle point (1 in the case of O sites and 2 for S-sites in BCC iron). A phonon analysis reported by de Andres et al. [25] shows that the vibrational energy on the O-site is approximately ten times higher than the thermal energy at room temperature. Energy Barriers: The characterization of the energy barrier can be performed by the energy difference between the initial configuration and the saddle point configuration. It is worth mentioning the relevance of the difference in zero-point energies between the initial and saddle point configuration (EZPE ). The energy barrier can therefore be expressed as (5). E B = ES − EA − EZPE
(5)
The evaluation of the energy of a specific configuration is a well understood method in DFT calculations [26–30], however the localization of all the relevant configurations may not be straightforward. Determination of Relevant Spatial Configurations: When considering the different configurations, they should be differentiated between (local) energy minimum configurations and non-minimum energy configurations, of which saddle point configurations are the most meaningful ones.
92
G. Álvarez et al.
The examination of minimum energy configurations can be performed by the combined use of a predictor configuration, a configuration which slightly deviates from the target configuration (local minimum) and a standardized geometry optimization method such as Plane-Wave Self-Consistent Field (PWscf) calculations in Quantum ESPRESSO [31] or the Broyden-Fletcher-Goldfarb-Shanno (BFGS) [32] method in CASTEP [33]. The finding of non-minimum energy configurations, however, is not as straightforward, as the previous methods are based on the evaluation of the forces exerted on the different particles involved and the displacement of said particles according to the forces until the energy minimum is reached. Three different methods will be discussed: -Nudge Elastic Band (NEB) [34]: This method was the first method proposed to find saddle point configurations. It is based on the sampling of a proposed minimum energy path between two minimum energy configurations with images of the system (states of the system in intermediate configurations) joined together with a spring or elastic band. The force acting on an image (Fi )is given the projection of the spring force (Fi S ) to the local tangent (to the trajectory and the true force (∇E(Ri )) perpendicular to the local tangent, Eq. (6). Fi = FSi ||| − ∇E(Ri )|⊥
(6)
This method allows for the approximation to a saddle point, but a rigorous convergence is not guaranteed. -Climbing image nudge elastic band (CI-NEB) [35]: An improvement over the classical NEB by modifying the force at the maximum energy image after a few iterations to the true force due to the potential energy with the tangent component to the trajectory inverted Eq. (7). Fimax = −∇E(Rimax ) + 2∇E(Rimax )|||
(7)
This modification ensures the localization of a saddle point along a trajectory joining the initial and final configurations. -Linear synchronous transient and Quadratic synchronous transient (LST/QST) method [36]. This method is based on the evaluation of computationally cheap intermediate configurations obtained by a linear interpolation between initial and final configurations, and a posterior refinement of the maximum energy structure by conjugate gradient, followed by an iterative process of evaluation of a local quadratic synchronous transient path, the evaluation of the maximum energy structure along the path and a posterior conjugate gradient refinement until convergence. This method allows for the search of saddle points deviating from the initial proposed energy path due to the iterative refinement of the whole structure, resulting in the possible obtention of lower energy saddle points for transitions over complex energy landscapes. The LST/QST has been used to calculate saddle-point configurations on the BCCFe + Hydrogen system by the evaluation of the energy barrier for hydrogen diffusion around a vacancy in a 3x3x3 BCC iron supercell with the presence of both a vacancy and an interstitial hydrogen (Fe53 H). The computations have been performed using the GGA approximation of the PBE form [37] under CASTEP simulation package [33] using nonlinear core corrected ultra-soft pseudopotentials (uspcc) [38]. Table 1 provides the energy barrier (EB ) for the transition of a hydrogen atom from interstitial site to an adjacent one before ZPE correction.
Quantum Mechanically Informed Kinetic Monte Carlo Models
93
Table 1. Hydrogen transitions and their corresponding energy barriers for the Fe53 H system (Before ZPE corrections). O stands for the octahedral site adjacent to the vacancy, while TX, corresponds to the Xth closest Tetrahedral site X’ and X” superindex order configurations with equivalent site distance by the biggest component along the cell axis. Transition
Energy barrier (meV)
O-O
240
O-T2
614
T2-O
negligible
T2-T2
76
T2-T3
131
T3-T2
73
T3-T3
65
T3-T4’
34
T3-T4”
119
T4’-T3
81
T4”-T3
39
T4’-T4’
91
3 Conclusion The different techniques used to calculate the physical parameters governing a QI-kMC model for hydrogen diffusion have been covered and exemplified over the BCC-Iron + Hydrogen system. The energy barrier for a hydrogen transition between neighbouring hydrogen positions in configurations around a vacancy have been provided showing that the octahedral site adjacent to the vacancy is the most stable configuration, while providing a description of the barriers to be overcomed by a hydrogen atom to move around the vacancy.
References 1. Johnson, W.H., Thomson, W.: II. On some remarkable changes produced in iron and steel by the action of hydrogen and acids. Proc. R. Soc. Lond. 23, 168–179 (1875). Dec 2. Reynolds, O.: On the effect of acid on the interior of iron wire. J. Franklin Inst. 99, 70–72 (1875). Jan 3. Thomas, R.L.S., Li, D., Gangloff, R.P., Scully, J.R.: Trap-governed hydrogen diffusivity and uptake capacity in ultrahigh-strength AERMET 100 steel. Metall. and Mater. Trans. A. 33, 1991–2004 (2002). Jul 4. Lynch, S.: Discussion of some recent literature on hydrogen-embrittlement mechanisms: Addressing common misunderstandings. Corros. Rev. 37, 377–395 (2019). Oct 5. Sofronis, P., Liang, Y., Aravas, N.: Hydrogen induced shear localization of the plastic flow in metals and alloys. Eur. J. Mech. A. Solids 20, 857–872 (2001). Nov
94
G. Álvarez et al.
6. Song, Curtin, W.A.: Atomic mechanism and prediction of hydrogen embrittlement in iron. Nature Materials 12, 145–151 (Feb 2013) 7. Martínez-Pañeda, E., Golahmar, A., Niordson, C.F.: A phase field formulation for hydrogen assisted cracking. Comput. Methods Appl. Mech. Eng. 342, 742–761 (2018). Dec 8. Motta, A.T., Chen, L.Q.: Hydride formation in zirconium alloys. Jom 64(12), 1403–1408 (2012) 9. Billone, M.C., Burtseva, T.A., Einziger, R.E.: Ductile-to-brittle transition temperature for high-burnup cladding alloys exposed to simulated drying-storage conditions. J. Nucl. Mater. 433(1–3), 431–448 (2013) 10. Pfeil: The effect of occluded hydrogen on the tensile strength of iron. In: Proceedings of the Royal Society of London. Series A, Containing Papers of a Mathematical and Physical Character, vol. 112, pp. 182–195 (Aug 1926) 11. Oriani, R.A., Josephic, P.H.: Hydrogen-enhanced nucleation of microcavities in aisi 1045 steel. Scripta Metallurgica 13, 469–471 (Jun 1979) 12. Gangloff, R.P.: Critical issues in hydrogen assisted cracking of structural alloys. In: Environment-Induced Cracking of Materials, pp. 141–165. Elsevier (Jan 2008) 13. Beachem, C.D.: A new model for hydrogen-assisted cracking (hydrogen “embrittlement”). Metallurgical Transactions 3(2), 441–455 (1972). https://doi.org/10.1007/BF02642048 14. Birnbaum, H.K., Sofronis, P.: Hydrogen-enhanced localized plasticity-a mechanism for hydrogen-related fracture. Mater. Sci. Eng., A 176, 191–202 (1994). Mar 15. Ogosi, E., Asim, U.B., Siddiq, A., Kartal, M.E.: Hydrogen effect on plastic deformation and fracture in austenitic stainless steel (Jun 2020) 16. Kirchheim, R.: On the solute-defect interaction in the framework of a defactant concept. Int. J. Mater. Res. 100, 483–487 (2009). Apr 17. Wang, R.: Effects of hydrogen on the fracture toughness of a X70 pipeline steel. Corros. Sci. 51, 2803–2810 (2009). Dec 18. Djukic, M.B., Sijacki Zeravcic, V., Bakic, G.M., Sedmak, A., Rajicic, B.: Hydrogen damage of steels: A case study and hydrogen embrittlement model. Engineering Failure Analysis 58, 485–498 (Dec 2015) 19. Novak, P., Yuan, R., Somerday, B.P., Sofronis, P., Ritchie, R.O.: A statistical, physical-based, micro-mechanical model of hydrogen-induced intergranular fracture in steel. J. Mech. Phys. Solids 58, 206–226 (2010). Feb 20. Nagumo, M.: Fundamentals of hydrogen embrittlement. Springer Singapore (2016) 21. Hammersley. J., Handscomb, D.C.: Monte carlo methods. Flecher & Son Ltd Norwick (1964) 22. Ramasubramaniam, A., Itakura, M., Ortiz, M., Carter, E.A.: Effect of atomic scale plasticity on hydrogen diffusion in iron: Quantum mechanically informed and on-the-fly kinetic Monte Carlo simulations. J. Mater. Res. 23(10), 2757–2773 (2008). https://doi.org/10.1557/JMR. 2008.0340 23. Du, Y.A., Rogal, J., Drautz, R.: Diffusion of hydrogen within idealized grains of bcc Fe: A kinetic Monte Carlo study. Physical Review B - Condensed Matter and Materials Physics 86, 174110 (2012). Nov 24. Kehr, K.W.: Theory of the diffusion of hydrogen in metals. In: Alefeld, G., Völkl, J. (eds) Hydrogen in Metals I. Topics in Applied Physics, vol 28. Springer, Berlin, Heidelberg (1978). https://doi.org/10.1007/3540087052_47 25. de Andres, P.L., Sanchez, J., Ridruejo, A.: Hydrogen in α-iron: role of phonons in the diffusion of interstitials at high temperature. Sci Rep 9, 12127 (2019). https://doi.org/10.1038/s41598019-48490-w 26. Nityananda, R., Hohenberg, P., Kohn, W.: Inhomogeneous electron gas. Resonance 22(8), 809–811 (2017). https://doi.org/10.1007/s12045-017-0529-3 27. Kohn, W., Sham, L.J.: Self-consistent equations including exchange and correlation effects. Phys. Rev. 140, A1133–A1138 (1965)
Quantum Mechanically Informed Kinetic Monte Carlo Models
95
28. Monkhorst, H.J., Pack, J.D.: Special points for Brillouin-zone integrations. Phys. Rev. B 13, 5188–5192 (1976) 29. Francis, G.P., Payne, M.C.: Finite basis set corrections to total energy pseudopotential calculations. J. Phys. Condes. Matter 2, 4395–4404 (1990) 30. Payne, M.C., Teter, M.P., Allan, D.C., Arias, T., Joannopoulos, J.D.: Iterative minimization techniques for ab initio total-energy calculations – molecular-dynamics and conjgate gradients. Rev. Mod. Phys. 64, 1045–1097 (1992) 31. Giannozzi, P. et al.: J. Phys.: Condens. Matter 29, 465901 (2017) 32. Byrd, R., Nocedal, J., Yuan, Y.: Global convergence of a class of quasi-newton methods on convex problems. SIAM J. Numer. Anal. 24(5), 1171–1190 (1987) 33. Clark, S.J., et al.: First principles methods using CASTEP. Zeitschrift fur Kristallographie 220, 567–570 (2005). May 34. Mills, G., Jonsson, H., Schenter, G.K.: Surf. Sci. 324, 305 (1995) 35. Chem, J.: Phys. 113, 9901 (2000). https://doi.org/10.1063/1.1329672 36. Govind, N., Petersen, M., Fitzgerald, G., King-Smith, D., Andzelm, J.: A generalized synchronous transit method for transition state location. Comput. Mater. Sci. 28, 250–258 (2003) 37. Ernzerhof, M., Perdew, J.P.: Generalized gradient approximation to the angle and systemaveraged exchange hole. J. Chem. Phys. 109, 3313 (1998). Nov 38. Louie, S.G., Froyen, S., Cohen, M.L.: Nonlinear ionic pseudopotentials in spin-densityfunctional calculations. Phys. Rev. B 26, 1738–1742 (1982). Aug
Numerical Modeling of Water Transfer in Geomaterials: Application to a Concrete Tunnel Subjected to Both Drying and Liquid Overpressure Aya Rima1(B) , Laurie Lacarrière1 , Alain Sellier1 , and Minh-Ngoc Vu2 1 LMDC (Laboratoire Matériaux Et Durabilité Des Constructions), Université de Toulouse,
UPS, INSA, 135, Avenue de Rangueil, 31 077 Toulouse Cedex 04, France [email protected] 2 Andra, 1-7 Rue Jean Monnet, 92298 Chatenay Malabry, France
Abstract. Concrete structures can go through cycles of drying and saturation during their lifetime. In special cases, such as tunnels, dams, and piles, this phenomenon may occur frequently. Drying and saturation cycles should be considered because they adversely affect the short-term and long-term behavior of the structure. In this paper, a water transfer model is proposed to predict the moisture migration and water content field within the studied porous material. The switching between saturated and unsaturated states is done continuously, without resorting to a classical three state variables model, but using liquid pressure as a single state variable in the governing water transport equation. First, the hydric transfer model is presented. The underground structure subjected to drying and resaturation is then simulated in order to demonstrate the capacity and performance of the hydric model to consider simultaneously negative and positive liquid water pressure, which was not previously the case, for instance, in the classical Richards formulation. This example demonstrates the potential of the model to describe water transfer by smoothly transitioning from a drying condition to a saturated condition. Keywords: Water transfer · Saturation · Drying · Porous material · Finite element
1 Introduction It is important to understand the behavior of structures throughout their lifetime. In porous materials such as concrete, several phenomena occur that affect the physical properties and performance of construction materials. One of these phenomena is the movement of moisture within the material due to its high porosity. This can lead to deterioration of concrete properties and cause major problems. So, it is important to understand the short- and long-term water transport. In this type of study, the coexistence of saturated at positive pressure and unsaturated conditions must be possible. Indeed, in © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 A. J˛edrzejewska et al. (Eds.): SynerCrete 2023, RILEM Bookseries 43, pp. 96–104, 2023. https://doi.org/10.1007/978-3-031-33211-1_9
Numerical Modeling of Water Transfer in Geomaterials
97
some special cases, such as piles, dams, and tunnels, the state of concrete can change between saturated and partially saturated conditions, which must be taken into account when calculating water transport in concrete structures. There are several studies in the literature addressing this type of problem. Philip and De Vries [1] and Luikov [2] proposed a theoretical model for water transport in porous materials. Bazant and Thanguthai [3] were inspired by this work and proposed a mathematical model that can describe the drying process of cementitious materials. Nonlinear diffusion equations can also be used to study water migration under gradient of moisture. In these models, water content or relative humidity (RH) can be maintained as the state variable, and the other variables are given as a function of it. These methods have attracted the attention of researchers due to their implementation simplicity and ease of use [4–7]. The liquid and vapor flows are combined in a single state variable (degree of saturation, water content, or relative humidity), leading to a single mass conservation equation to solve. This equation takes into account the evolution of the permeability of water and gas through a single equivalent diffusion coefficient. The main advantage was then to use a thermal analogy to solve any drying problem, without modifying the software. However, this model has some limitations. The main problem is that the positive liquid pressure cannot be considered and that the gas pressure must be assumed to be in equilibrium with boundary conditions. For this purpose, more complex models have been proposed considering that the movement of water can occur in three forms (liquid, vapor, and gas) [8–10]. These models take into account several phenomena that occur when modeling the movement of water. The main drawback of this last type of model is that three state variables lead to three coupled mass conservation equations to solve. Despite this major drawback, some numerical codes use this method to compute water motion. But, major modifications have to be done to codes in order to implement these equations. Since our goal is to provide a formulation compatible with a thermal analogy, but without the two limitations mentioned above, this paper aims to propose a hydric transfer model that uses only one state variable. Unlike previous single-variable formulations, the model is able to switch between positive water pressure and drying. Section 2 presents the relevant water transfer equations. In Sect. 3, an underground structure application is performed to demonstrate the performance of the model. Finally, Sect. 4 presents conclusions and perspectives.
2 Water Transfer Model In porous material, such as concrete, water transfer is controlled by liquid water, water vapor, and the dry air phase. Three mass balance equations are required to properly describe water transport in the material. The macroscopic approach is expressed by the following conservation equations for three state variables: ⎧ ∂ml − → ⎨ ∂t = −div( wl ) + S(v)→(l) + S(s)→(l) − → ∂mv (1) ∂t = −div(wv ) − S(v)→(l) ⎩ − → ∂ma ∂t = −div(wa )
98
A. Rima et al.
where ml , mv , and ma are the mass quantities for liquid, vapor, and gas, respectively. wl , wv , and wa are the mass flows of each constituent. S(v)→(l) is the mass quantity exchanged by evaporation or condensation and S(s)→(l) is the mass quantity exchanged by the phase change from solid to liquid. In this work, the mass of dry air is assumed to be negligible compared to the mass of vapor phase (of course this first assumption will define the application domain of the model). Consequently, the third equation of Eq. 1 can be neglected. Based on existing studies, simplifications can be adopted that allow the first two mass conservation equations to be reduced to a single one, as suggested in the pioneer works of Richards [11]. The obtained single mass conservation equation is as follows: ∂(ml + mv ) → → = −div(− wl + − wv ) + S(s)→(l) (2) ∂t The quantity of mass for the liquid and vapor phases is given as a function of the density ρ, the degree of saturation S l , and the porosity of the material φ (Eqs. (3) and (4)). ml = ρl .Sl .φ
(3)
mv = ρv .(1 − Sl ).φ
(4)
As a temperature rise is possible, the partial pressure of the dry air is assumed to be negligible compared to the vapor pressure. With this assumption, the gas pressure can be assumed to be equal only to the vapor pressure pv . The mass flux for each phase is assumed to be the product of the density and the transfer velocity. It is given by: kl −−→ − → wl = −ρl . krl grad (pl ) ηl kg Γ −−→ − → wv = −ρv . krg (1 + )grad (pv ) ηg pv
(5) (6)
with ρ i (i = l, v) the density of each phase, k i the permeability, k ri the relative permeability, η i the dynamic viscosity, and Γ the Klinkerberg coefficient. Substituting Eqs. (3), (4), (5), and (6) into Eq. (2), we notice that three state variables are required (pl , pv , and S l ). However, our main goal is to reduce the number of state variables to one in order to simplify the computation, keeping in mind that positive pressure must be able to coexist with drying. Since liquid pressure pl is present in both cases, it is chosen as the only state variable of the proposed water transport model. The other variables are assumed to be internal variables expressed as a function of liquid pressure pl . The water balance equation is then given by Eq. (7), where C l and K l are two coefficients to represent capacity and equivalent permeability, respectively. ∂pl −−→ = div(Kl grad (pl )) (7) ∂t The reason why this formulation has not been used so far is certainly related to the fact that the expression of pv as a function of pl is not explicit. This point will be discussed below. Cl
Numerical Modeling of Water Transfer in Geomaterials
99
2.1 Unsaturated Zone In the unsaturated state, vapor pressure pv is given as a function of liquid pressure pl via Kelvin law, in which vapor pressure and liquid pressure are in equilibrium (Eq. (8)). pc = pv − pl = −ρl
RT pv ln( ) Mv pvs
(8)
where pc is the capillary pressure, R is the gas constant, M v is the molar mass of water, and pvs is the saturation vapor pressure. Since this equation cannot be reversed analytically, another relationship between pv and pl is used (Eq. (9)). It is derived from the Kelvin law but without neglecting the molar volume of liquid. pv = pvs exp((
Mv )(pl − pvs )) ρl RT
(9)
Since the mass water must be computed, it is also necessary to use the water retention curve in porous media. For instance, the degree of saturation can be given by the analytical expression derived from [10] (Eq. (10)). Sl = [1 + (
pc nvgn −mvgn ) ] Mshr
(10)
where nvgn and mvgn are the Van Genuchten exponents and M shr is a parameter that accounts for the temperature effect. For the unsaturated state, the capacity term C l of Eq. (7) is deduced and becomes: m
n
pc nvgn −mvgn −1 pc nvgn −1 vgn v Mv Cl = (ρl − pRT ) vgn ] ( Mshr ) Mshr φ[1 + ( Mshr ) φ Mv 2 + ρl ( RT ) (1 − Sl )pv
(11)
The diffusion term Kl is also deduced and given by the following equation: Kl = ρl
kl 1 pv Mv 2 kg Γ ) krl + ( krg (1 + ) ηl ρl RT ηg pv
(12)
2.2 Saturated Zone In a saturated medium, water transport occurs exclusively through the liquid phase. Therefore, only liquid pressure pl is considered in the governing equation. In this case, the liquid pressure is given as a function of the mass of the liquid entering the material (Eq. (13)). pl = kw (
ml − φ) ρl
(13)
where k w is the bulk modulus of liquid water and φ is the porosity (assumed to be constant if the model is not coupled with the mechanical behavior of the matrix).
100
A. Rima et al.
Replacing the liquid mass in Eq. (2) by the expression of Eq. (13), we obtain the expression for the capacity C l and the equivalent permeability term K l (Eqs. (14) and (15)). Cl =
ρl kw
Kl = ρl
kl ηl
(14) (15)
The main advantage of these proposed expressions (Eqs. (11), (12), (14), and (15)) is that continuity is provided when switching between positive and negative liquid pressure zones. More details about the hydric model can be found in [12]. The proposed hydric model can then be used in a finite element code thanks to a thermal analogy, as expected. In our case, the software is Cast3M [13].
3 Numerical Simulation of Radioactive Waste Storage Tunnel 3.1 Description of the Geometry and the Boundary Conditions After presenting the water transfer model, a prediction of the hydric behavior of a nuclear waste disposal structure is carried out. The tunnel studied is the one proposed by the French Agency for Nuclear Waste Management (ANDRA). The tunnel consists of two concrete layers and is located at a depth of 500 m below the ground surface. The concrete is surrounded by «Callovo Oxfordian» soil (COX). The objective of this application is to determine the liquid pressure evolution in the underground structure, including concrete and soil, during its lifetime. The concrete layer in contact with the soil is 20 cm thick and plays the role of a compressible layer that absorbs the convergence of the soil over time. A second concrete layer forms the main part of the tunnel and has a thickness of 50 cm. Its inner radius is 4.35 m. After the excavation, two damaged zones (EDZ) with elliptical shape are observed around the concrete (see Fig. 1a). The geometry and dimensions of the underground structure are also illustrated in Fig. 1a. In general, several stages must be distinguished when simulating water transfer in a tunnel. In this study, the chronology is idealized with two main stages. The first includes the construction stage and the 100 year aeration. The second stage is the concrete and soil resaturation phase. These two stages are detailed below. The imposed hydric boundary conditions are presented in Fig. 1b. The soil liquid pressure boundary condition is placed far enough away from the tunnel so that it does not influence the kinetics of water transfer. This means that it is chosen in such a way that the disturbance of the initial liquid pressure field in the soil never reaches the imposed vertical boundary conditions. Currently, the hydric model is only implemented in the 3-D version of Cast3M. For this reason, a small thickness in the longitudinal direction of the tunnel is considered and a three-dimensional calculation is performed. Since the area around the concrete structure is damaged, different permeability parameters are assumed for the EDZ1, EDZ2, and soil. The parameters used for the water transfer model for the concrete and each phase of the soil are shown in Table 1.
Numerical Modeling of Water Transfer in Geomaterials
101
3.2 Results Analysis Stage 1: Placement of concrete - 100 years (drying) Since the tunnel is not closed at this stage, an aeration phenomenon occurs. In the excavated area there is a relative humidity (RH) of 50%. This RH corresponds to a liquid pressure of -95 MPa at a constant temperature of 20 °C. After excavation, soil disturbance occurs, causing a negligible value of liquid pressure pl in the damaged zone (EDZ1 and EDZ2). Initially, it is assumed that liquid pressure pl in the concrete is zero since it is negligible compared to the value prevailing in the excavated area. The water present in the pores of the concrete and soil is attracted toward the excavation zone due to the pressure difference between the ventilated area and the structure. The pore water evaporates, leading to a desaturation of the tunnel and the surrounding soil and a decrease in liquid pressure (drying).
(a)
(b)
Fig. 1. (a) Geometry and dimensions of the nuclear waste storage tunnel (b) Boundary conditions of the underground structure (dimensions in m).
The evolution of the liquid pressure as a function of time is shown in Fig. 2. After one year, the desaturation of the concrete and the two EDZs is significant. Then, this
102
A. Rima et al. Table 1. Hydric parameters provided by ANDRA
Parameter
Concrete
EDZ1
Ø (-)
0.135
5.6
4.5
4.2
kl (m2 )
9.59e-18
9.59e-17
9.59e-19
9.59e-21
kg (m2 )
1.5e-17
9.59e-14
9.59e-16
9.59e-20
Mshr0 (MPa)
19.6
EDZ2
17.7
17.7
Soil
17.7
mvgn (-)
0.33
0.34
0.34
0.34
nvgn (-)
1.5
1.52
1.52
1.52
desaturation progresses slowly to reach the undamaged zone of the soil. As time progresses, the liquid pressure gradient in EDZ1 tends to zero due to its high permeability. This gradient increases in EDZ2 as well as in the undamaged zone. With time, the desaturation continues. At 100 years, the liquid pressure in the concrete is almost equal to the imposed liquid pressure in the vent zone. 0
1 year
50 years
100 years
5.00E+06
pl (Pa)
-1.50E+07 -3.50E+07 -5.50E+07 -7.50E+07 -9.50E+07 0
5
10
15
20
x (m) Fig. 2. Evolution of the liquid pressure pl along line AB (stage 1).
Stage 2: 100 Years - Saturation After 100 years, ventilation ceases and resaturation of the underground structure begins. The previous boundary condition of 50% RH is transformed into a state of null flow. As a result, the water pressure in the concrete and soil increases. A quantitative study is performed to determine the time required to reach a new equilibriu. Figure 3 shows that the time required for the pressure to become positive is about 7143 years. 18000 years are required to retrieve the natural underground water pressure of 5 MPa along the horizontal line AB. The structure is then fully resaturated.
Numerical Modeling of Water Transfer in Geomaterials 100 years
200 years
7143 years
103
>18000 years
5.00E+06
pl (Pa)
-1.50E+07 -3.50E+07 -5.50E+07 -7.50E+07 -9.50E+07 0
5
10
15
20
x (m) Fig. 3. Evolution of the liquid pressure pl along line AB (stage 2).
4 Conclusions and Perspectives In this paper, a water transfer model presenting an analogy with the heat transfer equation is proposed for porous materials subjected to negative and positive water pressures. In this model, only the liquid water pressure is chosen as the state variable and all other variables and coefficients are specified in terms of this variable. The liquid pressure is used because, unlike the vapor pressure, the degree of saturation, and the water content, it exists in both cases: in the positive pressure zones and in the unsaturated zone. This has the advantage of providing continuity when switching between the overpressure and drying boundary conditions. The vapor pressure and the degree of saturation are expressed as a function of the liquid pressure using thermodynamic laws. The main underlying assumption is that the dry air pressure can be neglected with respect to the vapor pressure, which is reasonable as long as the porous medium is not too dry. One method to verify this assumption could be to compare it with experimental results, which is a first perspective of this work. The prediction of the hydric behavior of a nuclear waste storage tunnel shows the capacity of the proposed model to treat simultaneously positive and negative boundary conditions. The drying of the tunnel followed by its saturation are modeled. The time required to obtain only a positive pressure can be determined, as well as the time to reach the hydric equilibrium of the underground structure after resaturation. In the present work, we neglect the influence of temperature on the drying and resaturation kinetics. Thus, another perspective is to perform other simulations to consider this aspect of the hydric behavior. High temperature periods should accelerate the drying and resaturation process. Next, the proposed hydric model could be coupled to a mechanical formulation to study the effects of the deformation of the porous material on the porosity and permeability.
104
A. Rima et al.
References 1. Philip, J.R., De Vries, D.A.: Moisture movement in porous materials under temperature gradients. EOS Trans. Am. Geophys. Union 38(2), 222–232 (1957) 2. Luikov, A.V.: Systems of differential equations of heat and mass transfer in capillary-porous bodies. Int. J. Heat Mass Transf. 18(1), 1–14 (1975) 3. Bažant, Z.P., Thonguthai, W.: Pore pressure and drying of concrete at high temperature. J .Eng. Mech. Div. 104(5), 1059–1079 (1978) 4. Bažant, Z.P., Najjar, L.J.: Drying of concrete as a nonlinear diffusion problem. Cem. Concr. Res. 1(5), 461–473 (1971) 5. Mensi, R., Acker, P., Attolou, A.: Séchage du béton: analyse et modélisation. Mater. Struct. 21(1), 3–12 (1988) 6. Xi, Y., Bažant, Z.P., Molina, L., Jennings, H.M.: Moisture diffusion in cementitious materials moisture capacity and diffusivity. Adv. Cem. Based Mater. 1(6), 258–266 (1994) 7. Witasse, R., Georgin, J.F., Reynouard, J.M.: Nuclear cooling tower submitted to shrinkage; behaviour under weight and wind. Nucl. Eng. Des. 217(3), 247–257 (2002) 8. Baroghel-Bouny, V., Mainguy, M., Lassabatere, T., Coussy, O.: Characterization and identification of equilibrium and transfer moisture properties for ordinary and high-performance cementitious materials. Cem. Concr. Res. 29(8), 1225–1238 (1999) 9. Gawin, D., Majorana, C.E., Schrefler, B.A.: Numerical analysis of hygro-thermal behaviour and damage of concrete at high temperature. Mechanics of Cohesive-frictional Materials: An International Journal on Experiments, Modelling and Computation of Materials and Structures 4(1), 37–74 (1999) 10. Thiery, M., Baroghel-Bouny, V., Bourneton, N., Villain, G., Stéfani, C.: Modélisation du séchage des bétons: analyse des différents modes de transfert hydrique. Revue européenne de génie civil 11(5), 541–577 (2007) 11. Richards, L.A.: Capillary conduction of liquids through porous mediums. Physics 1(5), 318– 333 (1931) 12. Chhun, P.: Modélisation du comportement thermo-hydro-chemo-mécanique des enceintes de confinement nucléaire en béton armé-précontraint (Doctoral dissertation, Toulouse 3) (2017) 13. CEA: Cast3m. Commissariat à l’Energie Atomique (2016)
Replicating the Failure Mechanism of a Real-World Event with the Lattice Discrete Particle Model G. Lifshitz Sherzer(B)
and A. Mitelman
Department of Civil Engineering, Ariel University, Ramat Hagolan 65, Ariel, Israel [email protected]
Abstract. On Feb. 4th, 2022, a 9-m-high pre-fabricated concrete segment of the Israeli separation wall collapsed near the town of El Ram, north of Jerusalem. The failure occurred due to excavated soil wrongfully deposited against the back of the wall. The wall segments were not designed to carry lateral earth loading. As a result, they collapsed due to flexural failure. We back-analysed the failure event based on technical data and observations to obtain the wall loading. This analysis suggests that the internal moment at the time of failure was considerably greater than the capacity of the wall according to accepted standards. We then conduct numerical analysis using the Lattice Discrete Particle Model (LDPM), capable of capturing fracturing processes and complex failure mechanisms to replicate the actual collapse mechanism. This analysis accounts for the post-peak behaviour of the reinforced concrete up to the point of the wall’s collapse. Identifying the failure in advance was crucial and prevented catastrophic outcomes by allowing time to react. However, the reinforced concrete residual state is far from being fully understood. While engineers cannot rely on structural elements’ residual state during design, a better understanding of this state is crucial for proper response and mitigation measures following the onset of failure. Keywords: Multi scale · Wall collapse · LDPM · Residual state and failure mechanisms
1 Introduction Cantilever retaining walls are highly ubiquitous structures. However, there are relatively few published studies on the forensic geotechnical investigation of failures of these structures [1]. On Feb. 4th, 2022, a number of pre-fabricated concrete segments of the Israeli separation wall collapsed near El Ram, north of Jerusalem. The cause of failure in the current case study is trivial: a contractor wrongfully deposited excavated soil and boulders against the back of the 9 m high cantilever wall. The wall segments, that were not designed to carry earth loading, collapsed. The event of the collapse was caught on camera, revealing that the wall segments were tilting forward previous to their failure The authors G. Lifshitz Sherzer & A. Mitelman & contributed equally to this work. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 A. J˛edrzejewska et al. (Eds.): SynerCrete 2023, RILEM Bookseries 43, pp. 105–115, 2023. https://doi.org/10.1007/978-3-031-33211-1_10
106
G. L. Sherzer and A. Mitelman
(see Fig. 1a showing the wall during failure and Fig. 1b at final collapse). This indicates that the mechanism that triggered the collapse was a flexural tension crack that developed near the base of the wall.
Fig. 1. a) the wall tilting during the collapse. b) final wall collapse.
The greatest challenge in geotechnical engineering is the uncertainty that stems from heterogeneity of geological materials. However, in the current case, most factors that govern the earth pressures against the retaining wall can be assessed quite accurately. The earth fill was made of cohesionless granular land fill excavated and deposited to the space between the wall and rock slope. The fill contained sand, cobbles, and boulders. The soil can therefore be regarded as loose soil, in compliance with the assumptions of Coulomb theory. Additionally, complex phenomena such as swelling, consolidation, additional pressures due to compaction, do not apply. The rainfall in Jerusalem during the two months prior to the event of collapse can be regarded as insignificant, and was approximately 4 mm. Hence, dry conditions apply, and groundwater would have a negligible effect on wall internal forces. Given these conditions, it is argued that the current case study is very similar to a controlled full-scale experiment, and provides itself as an important opportunity for a back-analysis study. In this paper, we attempt to numerically back-analyse the cantilever wall failure event with the objective of successfully predicting the residual state. For this objective, we use the Lattice Discrete Particle Model (LDPM), a numerical method capable of capturing fracturing processes and complex failure mechanisms.
Replicating the Failure Mechanism of a Real-World Event
107
2 Modelling Strategy For assessing concrete behaviour, building codes and structural analysis tools at the current state of practice rely mainly upon empirical formulae and elastoplastic material models. However, the failure process of concrete is governed by fracturing. Therefore, numerical methods that integrate principles from fracture mechanics allow deepening the understanding of concrete structural failure [2]. The Lattice Discrete Particle Model (LDPM) is a numerical method that simulates the behavior of granular materials and particulate systems [3]. It introduces specific geometrical discretization of the space around aggregate particles. The geometrical generation in LDPM involves generating particles with a specified size distribution based on the Fuller approach [4], which are then placed randomly from the largest to the smallest into a computational grid, following a try-and-reject algorithm to avoid particle and boundary overlapping. Once the particles have been placed, an initial lattice connection between every four particles with Delaunay triangulation is built. Thereupon a VoronoiLike tessellation is generated [5]. The boundaries between all the Voronoi-Like cells, namely “facets,” create a fully connected network. These facets surfaces define the weak points of the structure, which are susceptible to failure. The original LDPM assumes that the aggregate is rigid and cannot fail. However, recent developments in the model have introduced new features such as aggregate failure through discretization approach [6] and ITZ deformability using a Multiphysics in LDPM (M-LDPM) [7]. The LDPM is governed by three failure equations; 1. Fracturing and cohesive behaviour, as follow; σbt (ε, ω) = σco (ω)exp[−Ho (ω)(< εmax − ε0 (ω) > /σc0 (ω))]
(1)
where σc0 denotes the yielding compressive stress, ω represents the degree of failure between shear and tension, Hco denotes the hardening Young’s modulus, brackets are used in the Macaulay sense, i.e., = max(x,0), and ε0 and εmax are, respectively, the strain at the yielding point and the maximum strain during the loading history. 2. Pore collapse and material compaction, as follow; σbc (εD , εV ) = {σco + (< εV − εc0 > Hc (rDV )} for εV≤ εc1 otherwise σbc (εD , εV ) = σc1 (rDV )exp[(< εV − εc1 > Hc (rDV )/σc 1(rDV ))]
(2)
where Hc denotes the initial hardening modulus and ε1 denotes the volumetric strain at the begining of the hardening σc1 (rDV ) = σc0 + (ε1 − εc0 ) Hc(rDV ) 3. Frictional behaviour, as follow, σbs (σN ) = σs + (μo + μ∞ )σNo − μ∞ σN − (μo − μ∞ )σNo exp(σN / σNo )
(3)
where σ s denotes cohesive stress, μo denotes the initial internal friction coefficient, μ∞ is the internal asymptotic friction coefficient, and σ No denotes the transitional stress.
108
G. L. Sherzer and A. Mitelman
Over the last two decades, the LDPM method has successfully been used for analysing the behaviour of plain concrete and Rebar Reinforced Concrete (RRC) [5, 8, 9]. For the rebar-facet interaction in LDPM modelling, a 1D beam of finite elements is incorporated into the original 3D structure. Figure 2 shows a typical representation of rebar (in red) and the surrounding LDPM mesh, where the blue triangle demonstrates the interaction. The concrete marked by c (blue dot) and the rebar denoted by r (red dot), together with both axial and radial springs representing the constraints, are depicted in Fig. 2. The penalty functions are used to determine the constraints connecting nodes c and r.
Fig. 2. Rebar Concrete interaction in LDPM.
The approximate displacement of a point within the LDPM mesh can be calculated using linear finite-element projections, as follows: Uc (t) =
4 i
Ni (xc )uci (t)
(4)
where ui c (t) denotes the displacement time histories of the four vertices of the tetrahedron that contains the rebar point, xc = xr are the nodal coordinates of the physical point in the LDPM mesh that corresponds to the rebar node, and Ni (xc) are the linear Lagrange shape functions. The design plans and technical specifications of the wall elements were obtained, and the model geometry and strength properties were built accordingly. The geometrical structure was generated in LDPM. In order to save computational time, the height of the wall was selected as 3m. This height was selected as it is centre of gravity of the ground lateral pressure. A width of 0.5 m is selected as representative slice from the wall. The wall was fixed at the bottom (See Fig. 3a) for the geometrical structure). Rebars were added to the geometrical structure according to the actual design of the wall. The steel area relevant for calculating the capacity for flexural failure at the wall base is 8.97 cm2 for per meter. The assigned steel material parameters are: elasticity modulus E = 200000 MPa, a density of 7800 kg/m3 , Poisson’s ratio of 0.3, and yielding strength of 500 MPa. The actual steel reinforcement in the concrete design is made of welded mesh, with 16 10mm diameter bars per element in the primary bending direction. The inner structure, including the size distribution of the aggregates located in the concrete block, is presented in Fig. 3b-3d.
Replicating the Failure Mechanism of a Real-World Event
109
Fig. 3. A) numerical set-up; ground pressure reacting on the wall and the Center Gravity (C.G) force that sets the height for the model and its geometrical structure b) inner structure with the surrounding concrete, c) inner structure including steel reinforcement (rebars) made of welded mesh (red) and aggregates distribution located in the block (blue), d) inner structure with the fixed support at the bottom (gray).
The load was imposed as a constant velocity of 0.05 m/s and applied to a node list in a 3 m height located at the top of the wall representing the C.G load of the ground pressure. The applied velocity is considered quasi-static loading. The events of the wall’s failure were recorded on film, showing evidence that the earth filling before the moment of failure was sloping above the top of the wall. Thus, for modeling purposes, we made a conservative assumption that the onset of flexural failure occurs when the earth filling is at the top level of the wall and that the additional filling is added during the residual state of the wall after the tensile cracking. Therefore, we calibrated the material to fit the macroscopic response to fail when the load reaches the C.G load of the earth filling 9m high. The C.G load was calculated as depicted in Eq. (5): Fmax,C.G = 1/2Ka γ H2 B
(5)
110
G. L. Sherzer and A. Mitelman
where K a denotes the effective lateral ground pressure coefficient, γ denotes the ground density of 2t/m3 , H denotes the wall’s height, and B denotes the out-of-plane width of the wall. The coefficient K a is calculated using Jaky’s equation for granular soils, was determined as 0.23: Ka = 1 − sinF
(6)
where F is the soil friction angle, assumed equal to 38º. This assumption is based on the angle of repose of the soil after the collapse, as measured from images, see Fig. 4. By placing the values in Eq. (5), the Fmax, C.G results in a value of 91398N.
Fig. 4. Soil angle repose after collapse.
Identification of LDPM Parameters The LDPM material parameters selected to fit this maximum load are as follows: the aggregate-to-cement ratio a/c = 3.3 and the water-to-cement ratio was based on the actual mixture w/c = 0.45. The cement content is 500 = kg/m3. The aggregate size distribution (which controls the mesh fineness) was assumed to have a characteristic coarse size within a range of a minimum aggregate size of do = 18 mm and a maximum of da = 20 mm in order to simulate a larger scale and save computational time. The Fuller coefficient nf = 0.5 was assumed as the default parameter since no available data was found for the specific sieve curve of the mix design. The identified model parameters are as follows: normal elastic modulus (stiffness for the normal facet behaviour) that governs the LDPM response in the elastic regime Eo = E/(1-2ν) = 40285 MPa corresponding to experiment model elasticity of E = 28200 MPa and to POisson’s ratio ν = 0.15, shear– normal coupling parameter α = 0.35, tensile strength σt = 6.5 MPa, tensile fracture energy Gt = 35 N/m, shear-to-tensile strength ratio σs /σt = 3 where σs is the facet strength for pure shear, softening exponent that governs the interaction between shear and tensile behaviour during softening at the facet level nt = 0.2, initial yielding compressive stress before the strain hardening with high-confinement condition σco = 140 MPa, initial hardening modulus ratio Hco /E0 = 0.45, Transitional strain ratio kco = 2.72, deviatoric strain threshold ratio kc1 = 1, deviatoric damage parameter kc2 = 5, initial friction coefficient μo = 0.4, asymptotic friction coefficient μ∞ = 0, the normal stress limit between μo and μ∞ is σNo = 600 MPa, densification ratio Ed /Eo . More information
Replicating the Failure Mechanism of a Real-World Event
111
and a deep understanding of the parameters can be found in [5, 8, 10]. See Fig. 5 for the global response of the calibrated parameters fitting the load calculated from Eq. (5).
Fig. 5. Numerical global response fitted to the C.G of the ground load
3 Results The steel reinforcement area is relatively low and does not significantly contribute to the capacity of the base cross-section. Therefore, once the wall is loaded beyond the tensile capacity of the concrete, a tension crack is initiated. The maximum capacity force that can react on the steel bars obtained from the simulations results are shown in Fig. 6. The total load of 91,398 N is distributed between the concrete wall and steel reinforcement. The steel reaches its elastic limit at 39,000 N (42.6% of the entire load), shown in Fig. 6. From this point onwards, the concrete carries the remainder of the loading. Hence, we examined the effect of crack propagation at different loading stages, starting from the stage where the steel reaches its capacity up to the residual stages. The evolution of tensile cracking, represented as the crack opening, is shown in Fig. 7. A pure tension mechanism is observed from the simulation, as cracking initiates at the bottom of the segment at the stretched fiber and propagates along the out-of-plane width (perpendicular to the stretching direction), as shown in Fig. 7. When the crack opens significantly, brittle failure occurs at the end of the residual stage.
112
G. L. Sherzer and A. Mitelman
Fig. 6. The maximum load acting on the steel reinforced bars
Nevertheless, an in-depth examination of the Material’s response is still required, as numerous factors can impact the flexural tensile strength, the formation of cracks and the degree of crack opening in the concrete segment. Such factors that affect the material’s response are related to the heterogeneity of the material: aggregate size, defects (voids), mineralogy composition, admixture types, moisture content, compaction, and curing conditions. Since the steel bars fail before the onset failure of the wall, the concrete is left to fully carry the external lateral loads, thereby increasing the effect of these factors. It can be assumed that these factors triggered the concrete to carry larger forces than the capacity of the wall according to accepted standards. Given that the LDPM approach is capable of replicating the global response based on the concrete components allows us to examine the peak load and the non-linear domain. To further examine the effect of concrete components, we repeated the simulation similarly to the previous one, but with a change in the minimum aggregate size to do = 15mm. This change in the minimum aggregate controls the degree of heterogeneity and irregularity of the internal structure, which is reflected in the mesh structure. Figure 8 shows the force displacement for both simulations. It is observed that an increase in the irregularity of the internal structure by decreasing the minimum aggregate size results in a decrease in the load capacity. In addition, it affects the cracks and the non-linear domain, increasing the Material’s brittleness (steeper slope). The aggregates play an essential role in the crack-bridging effect. The LDPM is based on the cohesive crack model [11], where the fracture process zone (FPZ) is simulated by a “Fictitious” crack that is able to transmit forces owing to the bridging effect affected by the heterogeneity of the material. The early softening domain has a steep slope followed by a tail. Decreasing the minimum aggregate size decreases the tail, and as a result, shortens the duration of the residual state (see Fig. 8).
Replicating the Failure Mechanism of a Real-World Event
113
Fig. 7. The concrete crack opening at a variety of loading stages after the rebars reaches its elastic limit, after which the remaining loads are transferred to the concrete a) a view of the concrete crack opening at the stage when the rebars reach maximum forces acting on the rebars, b) a view of concrete crack opening at the stage when the wall reaches its peak load and the axial forces reacting on the rebars, and c) a view of the concrete crack opening at the residual stages of the wall and the axial forces reacting on the rebars.
The simulation results showed an out-of-plane motion of the collapsed wall similar to the actual event. See Fig. 9 for actual and numerical simulation. Field reports testified that the apparent residual state lasted approximately two days up to the collapse of the wall. Consequently, the residual state controlled by the bridging effect allowed for identifying and preventing catastrophic outcomes.
114
G. L. Sherzer and A. Mitelman
Fig. 8. The global response of force-displacement for the simulation with a minimum aggregate size of 18mm (blue) and the simulation with a minimum aggregate size of 15 mm (orange).
Fig. 9. Out-of-plane motion of the actual wall, the simulated wall with the same resolution, and zoom-in view for better visualization of the details.
Replicating the Failure Mechanism of a Real-World Event
115
4 Summary and Conclusions In this paper we back-analyse cantilever wall failure due to earth loading and tensile cracking at the base. The geometric and material properties of the cantilever wall elements were obtained and the LDPM method was used for simulation. The main findings are as the following: 1. The internal moment at the point of failure is greater than the capacity of the wall according to accepted standards. 2. The failure mechanism is such that the steel capacity limit is reached, and the loading is gradually transferred to the concrete that begins to crack. 3. Simulation results show out-of-plane motion of the wall, in agreement to the actual event of failure. 4. The simulation was repeated in order to investigate the impact of reducing the minimal aggregate size and increasing the heterogeneity of the concrete matrix. Results show that this leads to a significant reduction in wall bending capacity.
References 1. Rao, V.V.S., Sivakumar Babu, G.L. (eds.): Forensic Geotechnical Engineering. DGE, Springer, New Delhi (2016). https://doi.org/10.1007/978-81-322-2377-1 2. Elmo, D., Mitelman, A.: Modeling concrete fracturing using a hybrid finite-discrete element method. Comput. Concr. 27, 297–304 (2021) 3. Cusatis, G., Pelessone, D., Mencarelli, A.: Lattice Discrete Particle Model (LDPM) for failure behavior of concrete. I: Theory. Cem. Concr. Compos 33 (2011). https://doi.org/10.1016/j. cemconcomp.2011.02.011 4. Fuller, W., Thomson, S.: The Laws of Proportioning Concrete. Trans. Am. Soc. Civ. Eng. 59, 67–143 (1907). https://doi.org/10.1061/taceat.0001979 5. Sherzer, G.L., Alghalandis, Y.F., Peterson, K., Shah, S.: Comparative study of scale effect in concrete fracturing via Lattice Discrete Particle and Finite Discrete Element Models. Eng. Fail. Anal. 135, 106062 (2022). https://doi.org/10.1016/j.engfailanal.2022.106062 6. Sherzer, G.L., Alghalandis, Y.F., Peterson, K.: Introducing fracturing through aggregates in LDPM. Eng. Fract. Mech. 261, 108228 (2022) 7. Li, W., Zhou, X., Carey, J.W., Frash, L.P., Cusatis, G.: Multiphysics lattice discrete particle modeling (M-LDPM) for the simulation of shale fracture permeability. Rock Mech. Rock Eng. 51, 3963–3981 (2018) 8. Sherzer, G., Gao, P., Schlangen, E., Ye, G., Gal, E.: Upscaling cement paste microstructure to obtain the fracture, shear, and elastic concrete mechanical LDPM Parameters. Materials (Basel). 10, 242 (2017). https://doi.org/10.3390/ma10030242 9. Alnaggar, M., Pelessone, D., Cusatis, G.: Lattice discrete particle modeling of reinforced concrete flexural behaviour. J. Struct. Eng 145 (2019). https://doi.org/10.1061/(ASCE)ST. 1943-541X.0002230 10. Lifshitz, S.G., Schlangen, E., Ye, G., Gal, A.E.: Evaluating compressive mechanical LDPM parameters based on an upscaled multiscale approach. Constr. Build. Mater. 251, 1–18 (2020). https://doi.org/10.1016/j.conbuildmat.2020.118912 11. Planas, J., Guinea, G.V., Elices, M.: Generalized size effect equation for quasibrittle materials. Fatigue Fract. Eng. Mater. Struct. 20, 671–687 (1997). https://doi.org/10.1111/J.1460-2695. 1997.TB00300.X
Thermomechanical Investigations for the Design of Reinforced Concrete Facings Eva Maria Dorfmann(B) , Dirk Schlicke, and Ngyuen V. Tue Institute of Structural Concrete, Graz University of Technology, Graz, Austria [email protected]
Abstract. The need for rehabilitation of existing structures is currently increasing and will continue to do so in the future. Reinforced concrete facings are considered an effective rehabilitation measure for damaged surfaces of massive concrete structures. The installation of the reinforcement for concrete facings, necessary for crack width control in the structure, is however time-consuming and expensive. The required reinforcement depends mainly on the development of the hydration heat and the resulting restrained deformations at an early age. This contribution presents a numerical model of the concrete facing and the existing structure for the simulation of the stress development due to the hardening of the concrete and the resulting crack formation. The results of the simulation should provide the basis for a future reinforcement optimization. For this purpose, the effect of different cements as well as the influence of the interaction with the existing structure are examined in this contribution. Keywords: Concrete facing · Early age · Numerical modelling · Crack width
1 Introduction For the repair of massive concrete structures, the application of a concrete facing represents an effective rehabilitation measure. This has now established itself as a standard method for the rehabilitation and reinforcement of sluice structures in German Waterways. In addition to its use in rehabilitation, reinforced concrete facings are also used to strengthen existing structures and ensure the impermeability of various structures. However, reinforcement installations for concrete facings according to [1] and [2] with two layers of reinforcement (Fig. 1) are often time-consuming and therefore expensive. The design of concrete facings and the necessary reinforcement ratio is closely linked to the prediction of cracking due to imposed deformation at an early age. Early age cracking in concrete facings is strongly dependent on both the material properties of the concrete facing and the interaction with the existing structure. The material properties and herewith imposed deformations of concrete facings at early age are predominantly characterized by hydration heat, shrinkage and simultaneously occurring strength evolution and creep. Additional imposed deformations occur with ambient temperature changes. Especially once the concrete temperature starts decreasing, the restraint of the existing structure leads to the development of tensile © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 A. J˛edrzejewska et al. (Eds.): SynerCrete 2023, RILEM Bookseries 43, pp. 116–124, 2023. https://doi.org/10.1007/978-3-031-33211-1_11
Thermomechanical Investigations
117
stresses in the concrete facing. If the tensile strength is reached, cracks might occur (Fig. 1). Main influencing parameters in this context are the concrete composition, the bonding behaviour between old and new concrete as well as the dimensions of the existing and the new structure.
Fig. 1. System overview as top view: (left) existing structure with reinforced concrete facing, (right) and cracking in facing due to the restrained deformation by the existing structure.
This contribution particularly concentrates on a detailed thermomechanical modelling of the concrete facing, the prediction of the stress development in the structure and the resulting risk of cracking, taking in consideration the evolution of the thermal and mechanical properties as well as the viscoelastic effects in hardening concrete. Furthermore, the effect of different concrete compositions and the influence of the bond between the concrete facing and the existing structure are investigated.
2 Model 2.1 FE-Model The following numerical investigations were conducted with a finite element model of a representative section of a sluice structure. The modelled structure consists of the already existing wall (old concrete) and the concrete facing with an explicit representation of the bond joint (transition zone between the existing structure and the concrete facing) and the reinforcement (Fig. 2). The basic model is a 2.5 m thick wall with a block length between the movement joints of 15 m, to which a 40 cm thick concrete facing is applied. The computational model was created using symmetry with the appropriate boundary conditions, and therefore only half of the block length was modelled. Furthermore, due to the large building expansions, the heat flow in the vertical direction is neglected and only a horizontal section is considered. The thickness of this section corresponds to the reinforcement spacing selected for the investigation. The study was carried out in two steps: First the temperature and stress development in the hardening phase in the concrete facing were evaluated. In a second step, cracking of the concrete was simulated, since the stresses exceed the tensile strength, and a crack width calculation was carried out. The material properties of the concrete facing are given in Table 1. The properties were modelled as time-dependent (as described in Sect. 2.3). The Young’s modulus of
118
E. M. Dorfmann et al.
Fig. 2. FE-model of the structure.
the old concrete was chosen to be 29,000 N/mm2 , while the stiffness of the bond joint, starting from the Young’s modulus of the old concrete, was varied for the following investigation. The bond between old and new concrete was considered only by modelling the bond joint. The anchors, which only affect the stresses in the concrete facing when a very low Young’s modulus is assumed, were not considered in the FE- model for the present study. The reinforcement (E s = 200,000 N/mm2 ) was modelled as a bar element, connected to the nodes of the concrete elements using contact elements according to the bond law in [4]. The minimum reinforcement was determined according to [2] for a maximum crack width of w = 0.25 mm and equals as = 27.43 cm2 /m. The distribution of the reinforcement follows the recommendation in [3] with 2/3 of the reinforcement on the outer part of the concrete facing and 1/3 on the back. Finally, the modelling of the crack formation process is accomplished with the implementation of discrete crack elements. Therefore, the nodes at the edge of the crack are connected with springs. Up to the tensile strength of the concrete, a linear material behaviour is assigned to the springs of the crack elements. After crack formation the stress-crack opening is described according to [5], taking into consideration the fracture energy. The crack width is then evaluated considering the relative displacement of the two edge nodes of the crack element at the surface of the concrete facing. Table 1. Material properties for the investigated concrete Concrete
C 30/37
fck [N/mm2 ]
30
fctm
[N/mm2 ]
2.9
Ecm
[N/mm2 ]
33,000
Since the focus of this study is set on the stress development and crack formation at an early age, the simulation is carried out for a period until equalization of concrete temperature with ambient conditions is reached (300 h = 12.5 days), whereby for an optimization of the calculation duration the size of the time-steps is successively increased.
Thermomechanical Investigations
119
The fresh concrete temperature is assumed to be 25 °C. Influences from temperature variations of the surrounding environment are neglected at this point of the study and a constant air temperature of 20 °C is assumed. In addition, the removal of the formwork is set to 3 days. 2.2 Studied Scenarios Different parameters were varied in the following simulation. On the one hand, the influence of two different concrete mixtures on the stress development and on the crack formation in the concrete facing was investigated. Therefore, the study was conducted for two different types of concrete mixtures: a concrete mixture with CEM I and a concrete mixture with CEM II. On the other hand, the interaction with the already existing structure was examined. This interaction is studied by varying the Young’s modulus of the transition zone between old and now concrete. The final goal is an optimization of the reinforcement in the future, therefore also the influence of the reinforcement layer in the back on the crack width was part of this study. The data for the study is summarized in Table 2. Table 2. Summary of the studied scenarios. Mixture
Distribution of the reinforcement
Stiffness of the transition zone [N/mm2 ]
CEM I
- Amin in 2 layers
- 100
- Without the reinforcement layer at the back
- 1000
- Amin in 2 layers
- 100
- Without the reinforcement layer at the back
- 1000
- 29,000 CEM III
- 29,000
2.3 Development of Mechanical Properties The modelling of the development of the mechanical properties is essential for the evaluation of the stress development and the resulting cracking in the concrete facing. The hardening process and the development of the mechanical properties are modelled based on the equivalent concrete age. The equivalent concrete age captures the dependence of the strength development on the temperature in the component. Due to the heat exchange with the environment and the heat flow into the existing, old concrete structure, the concrete facing shows different equivalent ages over its thickness. For the simulation of the hydration-related temperature development the heat release function according to Jonasson [6] was used (Eq. (1)). a t b· ln 1+ τeff
Q(t)=Qmax · e
k
(1)
120
E. M. Dorfmann et al.
The parameters for the hydration-heat development for both concrete mixtures are summarized in Table 3. The respective adiabatic temperature development for the mixtures with CEMI and CEMIII is shown in Fig. 3.
Fig. 3. Development of hydration heat.
The development of compressive and tensile strength of the concrete facing was modelled according to WESCHE [7]. Here, the development of the strength is described based on the time coefficient fβ (Eq. (2). w −β (2) fβ = exp −a · · teff WES − 28−βWES z The parameters listed in Table 3 were used for the investigated mixtures with CEMI and CEMIII. Table 3. Parameters for the model. Parameter JONASSON
Parameter WESCHE
a
b
τk
αmax
a · wz
βWES
CEM I
−3,3
−7,0
2,8
0,98
3,5
0,55
CEM III
−1,05
−0,015
1000
0,98
2
0,55
Figure 4 shows the relative development of tensile strength and Young’s modulus for both examined concrete mixtures. After 28 days both concretes have equal properties. Thus, the two concretes differ in the development of the mechanical properties and the magnitude of the hydration heat. Besides the thermal deformations, shrinkage leads to additional restrained deformations and therefore, to the development of additional stresses. In the present study, only the effect of basic shrinkage (without the effect of drying) was investigated. Due to its later development, drying shrinkage is not considered for the stress development at an early age. The basic shrinkage was modelled according to the normative specifications in [6] with εc,b,∞ = 0,05‰. The shrinkage development over time is simplified by linear coupling to the hydration process of the concrete.
Thermomechanical Investigations
121
Fig. 4. Development of the tensile strength and the Young’s modulus.
For the determination of the stress development significant attention has also to be paid to the creep behaviour of concrete. This was modelled based on the creep function in [2], with a modified tensile creep behaviour according to [9] and [10]. Based on the linear-elastic stress calculation the resulting creep deformation was evaluated for every time step and imposed in the calculation. A detailed explanation of the implemented algorithm for creep can be found in [10].
3 Results 3.1 Temperature and Stress Development The temperature development shown below (Fig. 5) is evaluated on the surface and in the middle of the concrete facing, at the contact area between facing and old concrete as well as in the old concrete. In general, the results agree with the expected temperature development shown in [1]. The maximum temperature is reached in the middle of the facing layer and is approx. 46 °C for the mixture with CEM I, while it is only approx. 37 °C for the mixture with CEM III. Considering the adiabatic temperature development, a large heat dissipation into the surrounding environment and into the old concrete can be stated herewith. The temperature gradient in the facing layer with the mixture CEM I is also more pronounced than with mixture CEM III.
Fig. 5. Temperature development in the concrete facing and the existing structure.
122
E. M. Dorfmann et al.
Figure 6 shows the stress development in the concrete facing due to the temperature development at the symmetry axis in longitudinal direction (x-axis) in the middle of the facing. The stresses shown in the figure correspond to the maximum stresses occurring in the structure. These stresses in the middle of the concrete facing are compared to the characteristic value of the tensile strength fct0,05k , with fct0,05k = 0.7·fctm .
Fig. 6. Elastic and viscoelastic stress development for different concrete mixtures.
The initial increase in temperature leads to the development of compressive stresses at the beginning of the hardening phase. With the beginning of cooling the compressive stresses are increasingly reduced until tensile stresses start to develop. These resulting tensile stresses in the facing exceed the tensile strength of the concrete and crack formation has to be expected for both concrete mixtures. The influence of creep on the stress development is also shown in Fig. 6. Due to creep the initial compressive stresses are slightly smaller, whereas the tensile stresses are smaller compared to the elastic stress development. However, no remarkable difference can be observed in terms of the time of crack development. 3.2 Crack Width Calculation Based on the model for the simulation of the stress development, crack elements were subsequently implemented in the concrete facing. The first crack element was implemented at the symmetry axis, corresponding to the section with maximum stresses, and then the number of cracks was increased manually until no more cracks would occur. The number of occurring cracks in the model for both concrete mixtures are: 5 cracks for a mixture with CEMI and 4 cracks for a mixture with CEMIII. The crack width is evaluated on the surface of the concrete facing by considering the relative displacement of the crack edge nodes. Figure 7 shows a summary of the crack width calculations with two reinforcement configurations (i) two layers acc. to [3] and (ii) only one layer at the surface. Following statements can be made: – When using a CEM III the crack width caused by thermal deformations can be reduced. However, the crack width does not differ significantly, when shrinkage is considered. – The better the bond behaviour between concrete facing and existing structure (regarded by different Young’s moduli of the bond joint), the smaller the resulting crack widths on the surface of the facing layer.
Thermomechanical Investigations
123
Fig. 7. Crack width for the studies scenarios.
– As to be expected, shrinkage leads to an increase of the crack width. – The reinforcement layer at the back of the concrete facing has generally only a small influence on the surface crack width. In all cases the surface crack width is larger when the second reinforcement layer is not considered, however, the influence on the surface crack width is greater, when the stiffness of the bond joint is the lowest. – As the Young’s modulus of the bond joint decreases, the influence of the reinforcement layer in the back on the crack width becomes larger. – The minimum reinforcement was evaluated for a maximum crack width of w = 0.25 mm. According to the simulation, this maximum crack width is exceeded only in one case (when a low stiffness of the bond joint and no reinforcement layer at the back are considered).
4 Conclusions Restrained thermal stresses in concrete facings lead most likely to the formation of cracks. The aim of the concrete facing design is therefore not to prevent cracking overall but to limit the crack width with an efficient reinforcement installation. The studied scenarios in this contribution indicate the possibility for an optimization of the reinforcement in terms of crack width limitation. In fact, the use of a concrete mixture with CEM III has a significant effect when it comes to thermal crack formation due to its low hydration heat development. However, in combination with shrinkage almost similar crack widths occur in CEMIII compared with CEMI. The usage of CEMII is therefore only beneficial in combination with shrinkage reducing agents. Furthermore, the importance of the bond behaviour with the existing structure for the limitation of the crack width was shown in the study. In summary the first results of the investigation show the potential for an optimization of the reinforcement. Further detailed investigations should lead to the development of an efficient design of concrete facings in the future. Acknowledgements. The work presented in this paper is part of the investigation conducted within the research project “Untersuchung zur erforderlichen Bewehrung in Vorsatzschalen” supported by the Federal Waterways Engineering and Research Institute (BAW), Germany.
124
E. M. Dorfmann et al.
References 1. ZTV-W LB 219: Zusätzliche Technische Vertragsbedingungen – Wasserbau (ZTV-W) für die Instandsetzung der Betonbauteile von Wasserbauwerken (Leistungsbereich 219), Bundesministerium für Verkehr und digitale Infrastruktur Abteilung Wasserstraßen, Schifffahrt, Ausgabe (2017) 2. DIN EN 1992-1-1 – Eurocode 2: Bemessung und Konstruktion von Stahlbeton- und Spannbetontragwerken – Teil 1-1: Allgemeine Bemessungsregeln und Regeln für den Hochbau. Beuth Verlag GmbH, Berlin (2013) 3. FuE-Abschlussbericht: Bewehrungsoptimierung an Betonvorsatzschalen (2009) 4. König, G., Tue, N.V.: Grundlagen und Bemessungshilfen für die Rissbreiten-beschränkung im Stahlbeton und Spannbeton sowie Kommentare, Hintergrundinformationen und Anwendungsbeispiele zu den Regelungen nach DIN 1045, EC2 und Model Code 90. DAfStb Heft 466. Deutscher Ausschuss für Stahlbeton, Berlin (1996) 5. Model Code: fib Model Code for Concrete Structures 2010. International Federation for Structural Concrete (2013) 6. Jonasson, J.-E.: Slipform construction – calculations for assessing protection against early freezing. Swedish Cement and Concrete Institute, Stockholm (1984) 7. Wesche, K.: Baustoffe für tragende Bauteile, Band 2: Beton und Mauerwerk. Vieweg+Teubner Verlag (1993) 8. Röhling, S.: Zwangsspannungen infolge Hydratationswärme. Verlag Bau + Technik, Düsseldorf (2009) 9. Schlicke, D.: Minimum reinforcement for restrained concrete. PhD Thesis, Graz University of Technology, Austria (2014) 10. Heinrich, P.J.: Effiziente Erfassung viskoelastischer Eigenschaften bei der Spannungsermittlung von gezwängten Betonbauteilen. PhD Thesis, Graz University of Technology, Austria (2018) 11. Turner, K.: Ganzheitliche Betrachtung zur Ermittlung der Mindestbewehrung für fugenlose Wasserbauwerke. PhD Thesis, Graz University of Technology, Austria (2017)
Modelling of the CO2 Uptake by Recycled Concrete Aggregates Philippe Turcry1(B) , Bruno Huet2 , Jonathan Mai-Nhu3 , Pierre-Yves Mahieux1 , and Thomas Pernin3 1 LaSIE, UMR 7356, CNRS, La Rochelle Université, La Rochelle, France
[email protected]
2 Holcim Innovation Center, Saint Quentin Fallavier, France 3 CERIB, Epernon, France
Abstract. The CO2 emissions due to the production of cement is a partly reversible phenomenon known as carbonation. The CO2 uptake occurs at each step of the life cycle of a structure, from the cradle to the grave as soon as concrete surfaces are in contact with atmospheric CO2 . After concrete crushing, this carbon sink is amplified by the increase of material surface. However, the study of the carbonation of recycled aggregates is difficult to carry out in the laboratory or in situ. Physico-chemical modelling is therefore a powerful tool which should make it easier to explore the CO2 uptake. We propose here results from numerical simulations of carbonation at both the grain and the stockpile scales obtained during the French Project “Fastcarb”. Our purpose is to evaluate the efficiency of a process on recycled aggregates, and the influence of parameters assumed to control the CO2 uptake rate, such as ambient CO2 concentration, grain size and water content. Experimental data, gas diffusion tests and CO2 binding capacity and rate tests, give order of magnitude of time and length scales of interest through dimensionless numbers and are input to the full numerical model. The used model is formulated in terms of mass conservation equations for both water and CO2 . Both dimensionless numbers and numerical results confirm that an accelerated process using industrial gases with high concentration (pCO2 ~20%) should be preferred to an atmospheric carbonation (pCO2 ~0.04%) to take full advantage of the CO2 uptake by recycled aggregates on short time scale. Keywords: Recycled Concrete Aggregate · Carbonation · CO2 Uptake
1 Introduction Among strategies to decarbonize the concrete construction sector, one of them is based on the sequestration of CO2 due to carbonation during the life and end-of-life of structures [1]. Calcium oxides of the cementitious matrix can mineralize to a certain extent the gaseous CO2 released during the cement manufacturing process the limestone decarbonation. This uptake depends on many parameters, such as environmental conditions, concrete and cement compositions, structure lifetime. If the uptake during the use of the © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 A. J˛edrzejewska et al. (Eds.): SynerCrete 2023, RILEM Bookseries 43, pp. 125–135, 2023. https://doi.org/10.1007/978-3-031-33211-1_12
126
P. Turcry et al.
structure is not negligible (around 10 to 20% of the emissions from limestone decarbonation [2]), this is probably not the most effective way in binding CO2 quickly because reinforced concrete structures are designed to limit the rate of carbonation and the related risk of corrosion. At the end-of-life, concrete is crushed to produce recycled concrete aggregates (RCA) that are usually used for road application [3] or even in recycled aggregates concrete. Crushing increases the specific area of the material what should favorize carbonation. Thus, CO2 uptake of RCA is an emerging solution put forward for at least 10 years, e.g., [4]. Before re-use, RCA are usually stored during several weeks in piles several meters high (Fig. 1). Evaluation of CO2 uptake by RCA are often made by considering simple geometries (e.g., spherical shape) and uniform CO2 concentration through the stockpile, e.g., [5, 6]. The latter assumption was questioned by Thiery et al. [7]. From experiments and modelling, the authors show that gas diffusion through the stockpile must be considered for a correct assessment of the carbonation of such a reactive granular medium. However, as discussed by the authors, the main limit of their work is that the study was carried out in the case of beds of monodispersed cement paste particles.
Fig. 1. View of a stockpile of recycled aggregates after crushing (La Rochelle, France).
The study of RCA carbonation, at stockpile scale or even at grain scale, is challenging since the material is somewhat heterogeneous (large size range, mix of different waste particles, various proportions of attached mortar, etc.) and characterization methods of the CO2 uptake are not yet standardized. Therefore, modelling is undoubtedly a useful tool. In the following study, we present results from numerical simulations of carbonation at both stockpile and grain scale obtained during the French Project “Fastcarb” [8]. Our purpose is double: evaluation of the atmospheric carbonation rate of a RCA stockpile and evaluation of the influence of parameters on an accelerated process used to sequester CO2 . For the latter objective, we consider that the used process insures an optimal contact between gas and grains.
Modelling of the CO2 Uptake by Recycled Concrete Aggregates
127
2 Program 2.1 Model Equations The mass balances of the main reactants, i.e., gaseous CO2 , liquid water, carbonatable elements and CaCO3 , are considered: ∂ (ϕ(1 − S + SKH )[CO2 ]) = div DCO2 grad[CO2 ] − rC ∂t
(1)
∂ (ϕS) = div(DS gradS) + rH ∂t
(2)
∂nCaCO3 = rC ∂t
(3)
with: ϕ porosity (-), S water saturation degree (-), KH Henry’s constant (-), [CO2 ] CO2 concentration in the gas phase (mol/m3 ), DCO2 diffusion coefficient of gaseous CO2 (m2 /s), rC sink term modelling the consumption of CO2 by the material (mol/m3 /s), DS moisture diffusion coefficient (m2 /s), rH source term modeling the production of water due to carbonation of portlandite, nCaCO3 content of CaCO3 produced by carbonation equal to the content of bound CO2 (mol/m3 ). The diffusion coefficient was written according to the formula of Millington [9]: DCO2 = Dair ϕa (1 − S)b
(4)
with: Dair the diffusivity of CO2 in the air (m2 /s), a and b empirical exponents. The moisture transfer property DS (not detailed here) was calculated as a function of the water permeability and the water vapor desorption isotherm. The equations system was solved using the finite elements method (open-source code Fenics). 2.2 Application to the Modelling of Carbonation at Stockpile Scale The previous model was applied to simulate the atmospheric carbonation of a RCA stockpile. Our approach was different than that used by Thiery et al. [7]. While the latter modelled a double scale medium (intergranular gas diffusion, intragranular gas diffusion and CO2 binding by cement hydrates), we considered a reactive transfer with gas diffusion between grains and an empirical sink term. rC was built from experimental results obtained in a previous study [10]. In the latter, the consumption of CO2 by a bed of 0/20 mm RCA in room conditions (20 °C and 65% RH) was determined over 4 months using an experimental method similar to the one described in [11]. From this time-evolution, we deduced the carbonation rate of RCA as a function of the carbonation degree assuming that the amount of CO2 bound after 4 months was the maximal CO2 binding capacity of the tested RCA (Fig. 2). The sink term has the following expression: rC = ϕ(1 − S)[CO2 ]f(αRCA )
(5)
128
P. Turcry et al.
where: f(αRCA ) is the function describing the influence of the carbonation degree αRCA , ratio of the amount of bound CO2 and the maximal CO2 binding capacity. f(αRCA ) models the rate of binding due to carbonation of all the grains, from fines to coarse aggregates. Of course, it was obtained for specific defined moisture conditions (65% RH). We considered that, on first approach, these conditions are optimal for carbonation.
Carbona on rate (gCO2/kgRCA/s)
2E-08
1.5E-08
1E-08
5E-09
0 0
0.2
0.4 0.6 0.8 RCA carbona on degree (-)
1
Fig. 2. Atmospheric carbonation rate of RCA (in g of CO2 per kg of RCA per s) versus carbonation degree of RCA (dots are experimental data from [10]).
Gas diffusion tests were carried out on RCA beds of 20 cm with a packing density of 0.65 (value close to the packing density of a stockpile) and different water saturation degrees [10]. From these results, the exponents of the Millington formula were obtained (a = 1.5 and b = 1.6). Figure 3 gives a comparison between experimental values and calculated ones of the diffusion coefficient as a function of the water saturation degree (granular medium).
CO2 diffusion coefficient (m 2/s)
1.0E-05 1.0E-06 1.0E-07 1.0E-08 1.0E-09 Granular medium (exp)
1.0E-10
Granular medium (model : a=1.6 and b=1.5) Grain (model: a=3 and b=2.2)
1.0E-11 0
0.2
0.4 0.6 0.8 Water satura on degree(-)
1
Fig. 3. CO2 diffusion coefficients for inter or intragranular diffusion: values from diffusion tests on RCA bed (exp), calculated values from Millington model (granular medium of 35% porosity and concrete grain of 15% porosity). a and b are the exponents of Millington model.
Modelling of the CO2 Uptake by Recycled Concrete Aggregates
129
In the present case, ϕ is the porosity of the granular medium, the source term rH was not taken into account, desorption isotherm and water permeability (used to calculate DS ) are from literature [3]. Boundary conditions are the following: [CO2 ] = 0.04%, temperature = 20 °C and relative humidity = 65%. In the following, we consider a stockpile of 2000 m3 with an external surface of 600 m2 (values assessed from a real recycling platform). The computations were made for a 1D geometry. The length scale L over which diffusion and chemical reaction compete because their time scale is the same can be assessed with Eq. 6. De,CO2 [CO2 ] (6) L= rC
2.3 Application to the Modelling of Carbonation at Grain Scale To model carbonation at grain scale, the sink term was written as a function of the carbonation of portlandite and CSH assumed to be the main carbonatable elements: 1 rC = rCH + rCSH n
(7)
rCH, CSH = ϕSKH [CO2 ]kCH,CSH
(8)
with:
kCSH is a linear function of the carbonation degree of CSH. kCH is also a function modelling the influence of carbonation degree on portlandite carbonation but with a more complex mathematical form (“shrinking-core model” as proposed by [12]). In Eq. (6), n is equal to 3.4 (i.e., moles of CaO per mole of CSH). The exponents of Millington model were taken from [13] (a = 3 and b = 2.2). Note that in the present case the decrease of porosity due to carbonation of hydrates was also modelled. Expressions and values for water permeability and desorption isotherm come from [14]. The model was used to compare atmospheric carbonation, i.e., at an ambient CO2 concentration of 0.04%, and accelerated carbonation, i.e., at a concentration of 15%. Moreover, a parametric study was carried out to evaluate the influence of material parameters on the efficiency of an accelerated process at 20 °C and 15% CO2 (Table 1). Before carbonation, we consider moisture equilibrium between the grain and the ambiance (the ambient RH was calculated knowing the water vapor desorption isotherm). Note that the concentration of 15% was chosen because it is about equal to the concentration of cement plants gases [8]. The computations were done for spherical geometry.
130
P. Turcry et al. Table 1. Ranges of entry data used for the parametric study. Parameters
Values
Initial porosity (%)
15, 25
CaO content (mol/m3 )
1860, 3720, 5580
Water saturation degree (-)
0.3, 0.6, 0.9
Grain diameter (mm)
1, 4, 10, 40
3 Results and Discussion 3.1 Atmospheric Carbonation of RCA at Stockpile Scale We calculated the carbonation of a RCA stockpile for a duration of 4 months, what is a typical duration between two crushing campaigns to our knowledge. Figure 4 shows depth-profiles of bound CO2 content in the stockpile. While CO2 binding capacity of RCA is significant (here around 20 g par kg of RCA), CO2 uptake remains low at the stockpile scale: carbonation depth is lower than 4 cm over 4 months. Compared to the dimensions of the order of several meters of a stockpile, this carbonated thickness is very low. By integrating the profile, the total mass of CO2 bound at 4 months can be calculated from the external surface. Around 400 kg are so obtained, what represents less than 1% of the maximal mass of CO2 that could be sequestered if all the pile was carbonated. The carbonation depth (XC ) could have been assessed using the classical “square-root of time model” proposed by Papadakis et al. [15]: 2DCO2 [CO2 ]0 t XC = (9) nCO2 with nCO2 the CO2 binding capacity of RCA (mol/m3 ) and [CO2 ]0 the atmospheric CO2 concentration. By applying the previous equation for t = 4 months, DCO2 = 1.10–6 m2 /s and nCO2 corresponding to 20 g par kg of RCA, XC is found equal to about 4 cm, what is close to the layer thickness being carbonated according to the numerical simulations. Note however that the computed profiles in shown in Fig. 4 are not sharp. This means that the chemical reactions rate (sink term) cannot be considered instantaneous compared to the gas diffusion rate in the granular medium. The main assumption of the square-root of time model is thus not satisfied. Equation (8) gives only a rough assessment. The calculation of competitive length scale Eq. (6) using measured data yields a value of ~1 cm consistent with numerical profiles.
Modelling of the CO2 Uptake by Recycled Concrete Aggregates
131
25
Bound CO2 (g/kg)
20 15 7d 30d 120d
10 5 0 0
0.02
0.04
0.06
0.08
Depth (m)
Fig. 4. Bound CO2 profiles in stockpile of RCA (ambient RH = 65%).
In practice, the boundaries of the pile could move during the period between two crushing campaigns on the recycling platform. To favour the CO2 uptake, the carbonated layer should be removed regularly to renew the carbonation process in RCA layers more in depth. Moreover, we did not consider any variation of water content due to rain. Anyway, our calculation results indicate that the atmospheric carbonation of RCA is not the most efficient method to sequester CO2 quickly. An accelerated process with high CO2 concentration in moving bed reactor promoting the contact between CO2 and grains should be preferable. 3.2 Carbonation at Grain Scale Atmospheric Carbonation vs. Accelerated Carbonation. Figure 5 highlights the huge difference in binding rate between atmospheric and accelerated carbonation in the example of a 4 mm diameter grain. This difference is the result of the CO2 concentration used in the sink terms rCH and rCSH . 80 70 Bound CO2 (g/kg)
60 50 40
Accelerated
30
Atmospheric
20 10 0 0
30
60
90 120 Time (days)
150
180
Fig. 5. Mass of CO2 bound by a grain of 4 mm in diameter in natural carbonation at 0.04% CO2 or accelerated at 15% CO2 (porosity = 15%, S = 0.6, CaO content = 3720 mol/m3 ).
132
P. Turcry et al.
Simulations were also made to evaluate the influence of a natural carbonation prior to the accelerated carbonation. As shown in Fig. 6, the storage of a grain in atmospheric conditions during 30 or 90 days has minor effect on the CO2 uptake due to an accelerated process of 7 days. This example indicates that the initial state should have little influence on the amount of CO2 fixed. 30
Bound CO2 (g/kg)
25 20 30 days 15
90 days
10 5 0 0
15
30
45
60
75
90
105
Time (days)
Fig. 6. Mass of CO2 bound by a grain of 4 mm in diameter after 30 or 90 days of atmospheric carbonation followed by a 7-day long accelerated carbonation (initial properties: porosity = 20%, S = 0.6, CaO content = 3720 mol/m3 ).
Parametric Study. Figure 7 gives all the results from the numerical campaign, i.e., masses of bound CO2 after 1 day of accelerated carbonation as a function of the studied parameters. As expect, the CaO content has a major effect on the CO2 uptake, as does the porosity. It is interesting to note that diameter and water saturation degree have coupled effects. For the lowest saturation (S = 0.3), no effect of the size on the mass of bound CO2 can be observed, while for the highest one (S = 0.9) the mass decreases sharply with the diameter. The coupling between water saturation degree and diameter can be explained by the competition between gas diffusion and chemical reactions. At high water content, carbonation rate is mainly controlled by gas diffusion since the diffusivity decreases strongly with S (Fig. 3). On the contrary, at low water content, carbonation rate is controlled by chemical kinetics. Figure 8 highlights this competition by showing the profiles of bound CO2 content inside a 10 mm grain. In the case of high gas diffusivity (S = 0.3), the grain carbonates almost uniformly. At high saturation degree (S = 0.9), the profiles of CO2 content are sharper: a “carbonation front” progresses from outside to the core.
Modelling of the CO2 Uptake by Recycled Concrete Aggregates 18
18 Porosity=25% S=0.3
18
Porosity=25% S=0.6
16 14
14
12
12
12
10
10
10
8
8
8
6
6
6
4
4
4
2
2
2
0
0
0
10
20 30 Diameter (mm)
18
10
20 30 Diameter (mm)
14
0
40
Porosity=15% S=0.6
16
14
12
12
12
10
10
10
8
8
8
6
6
6
4
4
4
2
2
2
0
0
10
20 30 Diameter (mm)
40
20 30 Diameter (mm)
40
Porosity=15% S=0.9
16
14
0
10
18
18
Porosity=15% S=0.3
16 Bound CO 2 (g/kg)
0 0
40
Porosité=25% S=0.9
16
14
2
Bound CO 2 (g/kg)
16
133
0 0
10
20 30 Diameter (mm)
40
0
10
20 30 Diameter (mm)
40
Fig. 7. Mass of bound CO2 per kg of RCA after 1-day accelerated carbonation as a function of grain diameter and CaO content (orange: 1860, blue: 3720, grey: 5580 mol/m3 ). 1000
1000
3 days
1 day
800
CO2 content (mol/m3)
CO2 content (mol/m3)
3 days 0.2 day 600 400
1 day 0.2 day
800 600 400 200
200
0
0 0
1
2 3 Radius (mm)
4
5
0
1
2 3 Radius (mm)
4
5
Fig. 8. Profiles of bound CO2 content in a 10 mm grain during accelerated carbonation for two water saturation degree: S = 0.3 (left) and S = 0.9 (right).
From experiments on RCA (diameter higher than 1 mm), Sereng et al. [16] have shown the existence of an optimum of water content for which CO2 uptake due to accelerated carbonation is maximized. This optimum is approximately equal to 80% of the water absorption of the aggregates determined according to NF 1097-6 standard [17]. This optimum is classically explained by the competition between gas diffusion and chemical reactions, as recalled previously. Our simulations show however that the optimum should depend on the aggregates size (Fig. 9). When the grain diameter decreases, the water content maximizing CO2 uptake increases, since carbonation becomes more
134
P. Turcry et al.
and more controlled by chemical rate rather than by transfer of CO2 . Actually, the optimum uptake is obtained when the competiting length scale L, Eq. (6), matches the size of aggregates. 12
Bound CO 2 (g/kg)
10 Diameter 1 mm 4 mm 10 mm 40 mm
8 6 4 2 0 0.3 0.6 0.9 Water satura on degree (-)
Fig. 9. Mass of bound CO2 per kg of RCA after 1-day accelerated carbonation for different grain diameters and water saturation degrees. Dotted lines indicate main tendencies (yellow line for diameter = 1 mm and blue one for diameter ≥4 mm).
4 Conclusions Carbonation of crushed concrete can be an efficient solution to sequester CO2 (in the scope of CCU – CO2 , Capture and Utilization). However, to take advantage of this “natural phenomenon”, management of recycled concrete aggregate (RCA) should be improved. Our simulations show that, at the stockpile scale, diffusion of atmospheric CO2 limits the carbonation extent to a very thin layer (a few cm) of the pile carbonated over a 4-month storage period. An accelerated process should be preferred to quickly carbonate RCA after crushing. At the grain scale, simulations confirm that carbonation at high CO2 concentration is much faster than atmospheric concentration. Our parametric study reveals that the effect of RCA diameter and water saturation degree are coupled. At low water content (lower than 0.6), the size between 1 and 40 mm has little influence on the carbonation rate. Carbonation is mainly controlled by chemical reaction and not diffusion: carbonation of the grain is almost homogeneous. At high water content, the size becomes a major parameter in carbonation: CO2 is bound mainly on the “skin” of the grain. The other material parameters, i.e., porosity and content of carbonatable elements, have expected effects since the CO2 uptake rate is an increasing function of these parameters. In engineering practice, these results indicate that water content of RCA must be managed only in the case of coarse aggregates. Our study presents some limits. In the modelling of atmospheric carbonation of a stockpile, we did not consider phenomena such as wetting/drying due to climatic conditions or possible CO2 advection through the granular medium due to wind. In the
Modelling of the CO2 Uptake by Recycled Concrete Aggregates
135
modelling of accelerated carbonation at grain scale, we modelled grains as sphere. More realistic geometries should be considered in further works. Moreover, all calculation results should be compared to experimental ones. This requires the development of test method to correctly characterize CO2 uptake by RCA. Acknowledgements. The authors would like to acknowledge all the partners of the National Project Fastcarb.
References 1. Miller, S.A., Habert, G., Myers, R.J., Harvey, J.T.: Achieving net zero greenhouse gas emissions in the cement industry via value chain mitigation strategies. OneEarth 4, 1398–1411 (2021) 2. AFNOR: CEN/TR 17310 Carbonation and CO2 uptake in concrete, standard (2019) 3. Hou, Y., Mahieux, P-Y., Lux, J., Turcry, Ph., Aït-Mokhtar, A.: Carbonation rate of compacted recycled aggregates for sub-base layers of pavement. Constr. Build. Mater. 312 (2021) 4. Engelsen, C., Mehus, J., Pade, C.: Carbon Dioxide Uptake in Demolished and Crushed, Concrete. Technical Report Rep, Norwegian Building Research Institute, Oslo (2005) 5. Pade, C., Guimaraes, M.: The CO2 uptake of concrete in a 100-year perspective. Cem. Concr. Res. 37, 1348–1356 (2007) 6. Guo, R., et al.: Global CO2 uptake by cement from 1930 to 2019. Earth Syst. Sci. Data 13, 1791–1805 (2021) 7. Thiery, M., Dangla, P., Belin, P., Habert, G., Roussel, N.: Carbonation kinetics of a bed of recycled concrete aggregates: a laboratory study on model materials. Cem. Concr. Res. 46, 50–65 (2013) 8. Torrenti, J. M., et al.: The FastCarb project: Taking advantage of the accelerated carbonation of recycled concrete aggregates. Case Stud. Constr. Mater. 17 (2022) 9. Millington, R.J.: Gas diffusion in porous media. Science 130, 100–102 (1959) 10. Hou, Y., Mahieux, P-Y., Turcry, Ph., Lux, J., Aït-Mokhtar, A., Aurélia Nicolaï, A.: Plateforme de recyclage de déchets inertes du BTP : un puits de carbone gris? Acad. J. Civil Eng. 40 (2022) 11. Boumaaza, M., Huet, B., Turcry, Ph., Aït-Mokhtar, A.: The CO2 -binding capacity of synthetic anhydrous and hydrates: validation of a test method based on the instantaneous reaction rate. Cement Concr. Res. 135 (2020) 12. De Larrard, T., Bary, B., Adam, E., Kloss, F.: Influence of aggregate shapes on drying and carbonation phenomena in 3D concrete numerical samples. Comp. Mat. Sci. 72, 1–14 (2013) 13. Boumaaza, M., Huet, B., Pham, G., Turcry, Ph., Aït-Mokhtar, A., Gehlen, C.: A new test method to determine the gaseous oxygen diffusion coefficient of cement pastes as a function of hydration duration, microstructure and relative humidity. Mater. Struc. 51 (2018) 14. Gendron, F.: Carbonatation des matériaux cimentaires - étude de la diffusion du CO2 . PhD Thesis, La Rochelle Université (2019) 15. Papadakis, V.G., Vayenas, C.G., Fardis, M.N.: A reaction engineering approach to the problem of concrete carbonation. AIChE J. 35, 1639–1650 (1989) 16. Sereng, M., Djerbi, A., Omikrine Metalssi, O., Dangla, P., Torrenti, J. M.: Improvement of recycled aggregates properties by means of CO2 uptake. Appl. Sci. 11 (2021) 17. Afnor. NF 1097–6: Tests for Mechanical and Physical Properties of Aggregates-Part 6: Determination of Particle Density and Water Absorption (2014)
A Multiscale Multiphysics Platform to Investigate Cement Based Materials Julien Sanahuja(B) , François Soleilhet, and Jean-Luc Adia EDF Lab Les Renardières, Ecuelles, France [email protected]
Abstract. As the owner of an extensive fleet of power plants, EDF is committed to the safe and sustainable long-term operation of massive civil engineering facilities. This involves inspection and auscultation of the current state of the structure, and structural computations to anticipate durability and to test mitigation strategies. Both require relevant material properties for the concretes at stake, which often cannot be measured directly, for either practical or economical reasons. The design of next-generation plants involves concrete mixes not yet existing, but whose behavior is required by structural engineers, to ensure that each facility correctly fulfills its function. Investigating the properties and behavior of cement based materials is thus of paramount importance. In addition to classical lab experiments, EDF is currently developing a “virtual lab” dedicated to concretes. It is based on a multidisciplinary platform, built taking advantage of recent advances in cement based materials modeling. A multiscale and multiphysics approach is adopted: materials are morphologically described down to the scale where the physical processes can be uncoupled, and where elementary constituents whose properties are mix independent can be identified. Then, upscaling techniques are used to estimate properties and behaviors at the engineering scale. The platform, named Vi(CA)2 T, includes toolboxes to upscale mechanical behavior (stiffness, creep) and transport properties (dielectric permittivity). It adheres to an open design: new physics are straightforward to add. After discussing the platform design and the challenges raised by such an approach, examples of applications at EDF are presented. Keywords: Concrete · Micromechanics · Virtual Lab · Creep · Permittivity
1 Introduction 1.1 Industrial Context and Issues To manage the long term operation of power plants civil engineering facilities, EDF has adopted an integrated approach [4]. It is based on three pillars, structural analysis, inspection, and material knowledge, complementing each other (Fig. 1): • without structural analysis, anticipating the future behavior of structures or evaluating the impact of mitigation and reparation strategies is not possible; © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 A. J˛edrzejewska et al. (Eds.): SynerCrete 2023, RILEM Bookseries 43, pp. 136–147, 2023. https://doi.org/10.1007/978-3-031-33211-1_13
A Multiscale Multiphysics Platform
137
• without inspection, the current state of structures cannot be diagnosed; • without material knowledge and modeling, structure computations cannot use relevant material properties and behaviors, and raw data from indirect inspection techniques cannot be properly interpreted (ex: when water content is indirectly measured through dielectric permittivity).
Fig. 1. Overview of approach for long term operation of civil engineering structures.
The physical properties and mechanical constitutive behaviors of the concrete(s) making up the structure at stake are then required. In this respect, a comprehensive experimental characterization of every concrete of every civil engineering structure would be out of scope: • as made up from local raw materials, each concrete is almost unique: the number of different mix designs in the French fleet has been estimated as around 1000; • re-casting samples according to a mix designed several decades ago is not always feasible as the specifically used cement and aggregates may not be available anymore; • drilling core samples from the structure is not always conceivable, and can only reveal the current state of the material, not its whole history. Thus, modeling approaches represent an attractive alternative to estimate, from the knowledge of both the initial mix design and the aging conditions in the structure, the properties and behaviors of concrete. To be adaptable to a rather wide range of concrete mixes, these models clearly have to be as much physics-based as possible. Concrete being a highly variable, multi-scale material, governed by multi-physics processes, multi-scale models represent an appealing option to derive physics-based behaviors. Indeed, these approaches, based on homogenization or micromechanics, are able to bridge the scale where the physical processes occur to the scale of interest for the engineer. In other words, it is possible to go down to the scale where intricate behaviors and couplings observed macroscopically can be physically investigated and possibly un-coupled. Another interesting feature of multiscale modeling is the possibility
138
J. Sanahuja et al.
to break down concrete into individual material phases whose elementary properties can be expected to be unique from one concrete to another. 1.2 A Software Toolbox Dedicated to Concrete Recent progresses regarding physics-based modeling of cementitious materials make the approach promising. However, knowledge is scattered worldwide, and models from the literature, often relying on rather involved techniques, may not be directly usable by engineers. Thus, structural engineers had, for a long time, to mainly rely on empirical models whose predictions and adaptability from one mix to another can be questionable. To overcome this situation, the development of a software tool dedicated to concrete has been initiated at EDF R&D [10]: Vi(CA)2 T, for Virtual Cement and Concrete Ageing Analysis Toolbox. It is based on the concept of a “virtual lab” or “virtual material”. From the mix design details, it is able to estimate physical properties (such as the evolution of hydration degree, cumulated heat, or capillary porosity) and mechanical behaviors (such as the evolution of Young’s modulus, or basic creep). This paper proposes an overview of Vi(CA)2 T features, of the underlying models, of the verification and validation approach, and then a specific focus on some practical applications of the toolbox. For more details on the toolbox, see [14].
2 Brief Description of Vi(CA)2 T Toolbox Vi(CA)2 T is a software toolbox dedicated to concrete, able to estimate properties and behaviors from the mix design details. To do so, it embeds two types of models: • Homogenization models: taking advantage of upscaling techniques, from the amount and elementary behavior of phases, and from morphological models, the effective properties and behaviors are estimated (such as stiffness, basic creep compliance or dielectric permittivity). • Cement paste hydration models: combining chemistry and kinetics, from the initial mix design information, the amounts of each phase (anhydrous phases, hydrates, water and, by extension, porosity) and the hydration physical properties (temperature, released heat) are estimated as a function of time. After the pioneering works [10] on version 1, limitations were identified: accessibility and usage by engineers was difficult, and the capacity to incorporate new models was limited. Thus, the tool has been revamped to version 2 [9], to extend usage and applicability. Hydration modeling has been improved, considering additions such as silica fume or ground granulated blast furnace slag, and more realistic heat transfer conditions. Creep modeling has been extended to estimate the complete basic creep function. Moreover, quality assurance aspects have been improved, adding unit, verification and non-regression tests. Version 2 is implemented using Python 2.7, for historical reasons. The toolbox is currently being ported to Python 3.9. This version, 3, is more emphasized on generic usage, as upscaling models can be used way beyond cement based materials. It is thus designed as a generic package gathering physical models, so the user can build upon it
A Multiscale Multiphysics Platform
139
to implement his own applications. The main features of successive versions 1, 2 and 3 are highlighted on Table 1. Table 1. Main features of Vi(CA)2 T versions (version 3: current features as of january 2023). Version
1
2
3 (under development)
Implementation
Scilab
Python 2.7
Python 3.9
Main usage mode
Scilab toolbox to be used in scripts
standalone CLI/GUI application
Python package to be used in scripts
Hydration model
clinker
clinker + silica fume + blast furnace slag, heat conservation
as version 2
Morphological models
basic homogenization schemes
multiscale representation based on spheroids + morphologically representative patterns
as version 2
Linear properties models
elasticity
elasticity
elasticity + eigenstrains/stress, permittivity
Basic creep models
asymptotic creep and uniaxial creep creep rate compliance
Quality assurance
n.a
132 unit, verification, 195 unit, verification non regression tests tests
Benchmarking
n.a
hydration, elasticity, creep (COST action TU1404 [5, 8, 16])
creep (ACES European project): work in progress
Applications presented in this paper
n.a
n.a
creep of Vercors concrete (Sect. 3.1), permittivity of paste (Sect. 3.2)
response to any stress/strain history
2.1 Overview of Embedded Physical Models The toolbox includes several modules implementing physical models, which can be connected depending on the intended application. The global organization of these modules
140
J. Sanahuja et al.
is represented on Fig. 2, while the physical models are independently described in more details in the following subsections. Hydration Modeling. The purpose of hydration models is to predict the time evolutions of: • quantities of the elementary phases of cement paste (as aggregates are assumed to be chemically inert): anhydrous, hydrates, water, and, by extension, capillary porosity; • heat generated by hydration reactions, and/or temperature evolution; from, as input data, mix design information and thermal conditions. Time response constraints currently prevent the use in Vi(CA)2 T of hydration models providing a full 3D description of the evolving cement paste microstructure. Hydration is thus modeled through simplified kinetics approaches for clinker components (C3S, C2S, C3A, C4AF + gypsum), silica fume and blast furnace slag. Stoichiometry is enforced through commonly used simplified hydration reactions of these reactants. The heat conservation equation is also written. This modeling yields an initial value problem involving a system of first-order ordinary differential equations, which is numerically integrated.
Fig. 2. Schematic flowchart of Vi(CA)2 T.
A Multiscale Multiphysics Platform
141
Microstructure Modelling. The homogenization models estimating effective properties and behaviors use as inputs: the volume fraction and elementary behavior of each phase, and a morphological model describing the geometrical arrangement of these phases. The evolution of microstructure and volume fractions due to the physicochemical processes at stake, such as cement hydration or degradation mechanisms, should also be described. Full field homogenization, relying on an explicitly described morphology, cannot be used in Vi(CA)2 T currently, as these techniques are too computationally intensive. Mean field homogenization is used so the morphological model has to be simplified. Basically, a multi-scale approach is followed, where at each scale, microstructure is described from simplified features such as ellipsoids and morphologically representative patterns. Thus, at each scale, a so-called homogenization scheme can be built from the morphological representation. This scheme is translated, depending on the investigated physics (elasticity, dielectric permittivity), into a semi-analytical procedure to estimate the effective properties. Regarding concrete, a rather natural scale separation can be introduced between aggregates (including sand) and the cement paste phases: concrete can be considered as aggregates inclusions in a cement paste matrix. Refined models would introduce more scale separations in the aggregates sieve curve, for example between coarse aggregates and sand; or would directly use the continuous size distribution. To simplify, aggregates can be modeled as spheres. If required, an interfacial transition zone can be introduced around inclusions, using morphologically representative patterns. Morphology of cement paste is much less straightforward to model, and is still a matter of debates in the scientific community. While the (possibly remaining) anhydrous grains can be represented as spherical inclusions, hydrated phases are much more challenging to describe, especially the C-S-H gel. Observing microstructural details of the latter (at scales starting from a few nanometers) is indeed technically challenging. At the cement paste scale, several morphological models can be found in the literature. Based on some microscope observations and a trial and error process comparing the evolution of effective stiffness with respect to experimental data, a simplified morphology has been designed for cement paste [12]. The main idea is to build a morphological model as simple as possible but as detailed as necessary to produce relevant effective stiffness evolutions. The model has been straightforwardly extended to concrete, see Fig. 3. While this simplified representation considers anhydrous or hydrates as a whole (albeit with a low/high density split), the currently implemented morphology details the diverse anhydrous and hydrated phases. Knowledge on Micro-Mechanisms. Estimating mechanical properties or behaviors beyond elasticity often requires, as a starting point, informations on the main micromechanisms at the origin of the effective behavior in question. For example, as far as basic creep is concerned, consensus has not been reached yet on the main nano- or micro-mechanisms. Many hypotheses can be found in the literature [1], among which transfers between capillary and adsorbed water, water transfers to newly created microcracks, C-S-H sheets viscous sliding, etc. This debate about the relevant creep mechanisms is beyond the scope of this paper, so a pragmatic assumption has to be made. The sheet sliding mechanism, in C-S-H bricks, is considered here.
142
J. Sanahuja et al.
Fig. 3. Multiscale morphological model of concrete (2D sketch of the actual 3D model, made up of spheres and spheroids).
Thus, transforming the elastic multiscale model into a basic creep one only involves modification of the elementary behavior of C-S-H bricks for the strain mechanisms activating sheets sliding [11]. A Maxwell model is considered as a first step, so that the corresponding characteristic time is the only extra parameter with respect to the elastic model. Upscaling Approaches. While upscaling elasticity is straightforward (at least once the simplified enough morphological model has been designed, and the elementary phases stiffness and volume fractions are known), upscaling other mechanical properties such as strength or basic creep represents an ambitious challenge as far as cementitious materials are concerned. Regarding strength, the proper description of damage, microcracking and cracks coalescence seems to be a genuine challenge for current mean field homogenization approaches. Regarding basic creep, a classical simplification is to assume microstructure does not evolve after the stress loading step. The correspondence principle can then be taken advantage of: the Laplace-Carson transform changes non aging linear viscoelastic behaviors into elastic ones. The transport properties represent another difficulty, as the contrast between phase properties can be high (if not infinite). This significantly raises the effective properties sensitivity on morphology. The latter can only be represented in an approximate way by mean-field homogenization. Nevertheless, a transport properties upscaling module is implemented, and used to homogenize dielectric permittivity. In this case, the maximum ratio between phase properties is about 80, which is reasonable. 2.2 Verification and Validation Regarding verification, three different types of tests are implemented: unit tests which check the basic features, verification tests which check the consistency of model results with respect to theoretical or alternative approaches, and non regression tests which check that results from a complete study are identical to previous ones. EDF contributed to the European COST action TU1404 “Towards the next generation of standards for service life of cement-based materials and structures”. This
A Multiscale Multiphysics Platform
143
project offered the opportunity to participate to Europe-wide benchmarks on prediction of cementitious materials properties and behaviors [5, 8, 16]: the reader is referred to these cited papers for more details. Benchmark participations are always insightful for comparisons to experimental results but also to evaluate the scattering of results from various modeling strategies. Hydration and Elasticity. The first two benchmarks are about hydration and early age stiffness of cement pastes at w/c = 0.3 and w/c = 0.4. Vi(CA)2 T tends to overestimate hydration rate at early age, and to overestimate the Young’s modulus. It is important to note here that regarding these two benchmarks, Vi(CA)2 T has been used as is, with default parameters, without any fit on provided experimental data. Only cement datasheets and pastes mix designs have been used. This is “prediction” in its original, blind, sense. However, comparisons to experimental data and to results from partners motivate questions on the relevance of hydration model parameters. These parameters have historically been extracted from the literature, while many evolutions have since been implemented in the models. Reevaluating these parameters, through identification on experimental data, would now be useful. Non Aging Basic Creep. The third benchmark is about non aging basic creep. As hydration data is provided, the aim is to separate effects in the validation process: here only the micro-mechanical creep models are tested. This benchmark consists in two parts: • bridging length scale: predicting early age short-term creep of mortar and concretes from early age short-term creep results on pastes (obtained from the experimental procedure described in [7]); • bridging time scale: predicting creep of mature paste (loaded at 30 years) from the same early age short-term creep results on pastes. For this third benchmark, the creep behavior of the interface between C-S-H sheets, identified long ago [11, 13], has not been reused. Indeed, the morphological model has since evolved significantly (notably detailing anhydrous and hydrated phases in cement paste). Rather, C-S-H bricks creep compliance has been identified on results provided on pastes: short creep tests (3 min long loading) repeated every hour at early age (between 1 and 7 days) on w/c = 0.42, 0.45, 0.50 pastes. Default values are still used for other parameters. The benchmark results were first unveiled at the previous, Synercrete’18, conference, and then published [8]. Vi(CA)2 T reasonably predicts early age creep of the mortar, but tends to underestimate creep compliance of the concrete. Note that it is also the case of the two other participating models. The creep compliance of the mature w/c = 0.5 cement paste is very reasonably predicted.
3 Practical Applications This section provides an overview of practical applications of the Vi(CA)2 T version 3 toolbox at EDF R&D, to provide inputs for structural analysis and interpretation of results from non destructive techniques.
144
J. Sanahuja et al.
3.1 Creep Response for Containment Building Structural Analysis Structural computations obviously require materials constitutive behaviors. In some cases, tests have been performed on samples made up of the specific concrete. When the structure is heavily instrumented, parameters of the concrete constitutive behavior could be numerically identified comparing sensors data to FEM model results. However, in most cases in the field, neither experimental results nor abundant sensor data are available. In these situations, Vi(CA)2 T outputs can be taken advantage of, as “virtual experimental results”, to contribute to definition of materials constitutive behaviors. This method is detailed on Vercors mockup in companion paper [15]. The uniaxial basic creep compliance of Vercors concrete loaded at t 0 = 90 d is simulated using Vi(CA)2 T: the estimate correctly compares with experimental results from two different labs (Fig. 4). Note that no parameter has been fit on Vercors compliance results: the model developed for the COST action TU1404 creep benchmark (last part of Sect. 2.2) has been reused as is, with parameters identified on cement paste early age data provided by this benchmark. Here, the only input data for the micromechanical model is Vercors concrete mix design.
Fig. 4. Estimation of uniaxial basic creep compliance of Vercors concrete at t 0 = 90 d, compared to measurements on samples at EDF TEGG and MMC labs (2 samples per lab).
3.2 Permittivity Measurements to Assess Water Content in Structure Water content is a major parameter of concrete delayed behavior and durability. However, its non destructive measurement is indirect as it involves a physical property whose value depends on water content. Dielectric permittivity is often considered. It thus remains to evaluate the relation between water content and dielectric permittivity, which depends
A Multiscale Multiphysics Platform
145
on the concrete at stake. Empirical laws could be used but they are often designed for soils and are not specific. Experimental measurements are possible but they are heavily time consuming as a concrete sample has to be dried step by step, with permittivity measurements in-between, and concrete drying is a very slow process. A third option has been investigated recently [6]: building a model based on upscaling of permittivity. Such a model requires to bridge the scale where liquid and gas can be identified as distinct phases in pore domain, to the scale of concrete. The morphological model for a first application to cement paste is sketched on Fig. 5. The permittivities of elementary phases have been identified on either compacted powders or specifically designed mixes. Results from the model compare reasonably well to experimental data (Fig. 6). Note that once the morphological model is designed, simulating one curve takes a few seconds, while obtaining the experimental data for one paste took several months as many cycles “drying equilibrating saturation field - measuring permittivity” are required.
Fig. 5. Morphological model designed to estimate permittivity of cement paste from its water content [6] (2D sketch of the actual 3D model, made up of spheres and spheroids).
Fig. 6. Estimation of permittivity of cement pastes as function of saturation ratio, compared to measurements on samples, for different w/c ratios [6].
146
J. Sanahuja et al.
4 Conclusion and Next Steps Vi(CA)2 T is a toolbox dedicated to the investigation of properties and behaviors of cementitious materials through multiscale modeling. Its version 3 is currently under development, using Python 3.9, emphasizing a modular approach. Modules to upscale elasticity and non aging basic creep are already available. A transport module (with applications to dielectric permittivity) is also implemented. Practical applications encompass both structural computations and interpretation of non destructive testing measurements. Beyond the mentioned examples, other applications have been performed on mechanical degradation due to mechanisms involving local expansions, at the aggregate [3] and concrete [2] scales. Modules to investigate mechanical strength, aging basic creep, and loss of mechanical properties due to various degradation mechanisms are upcoming. Regarding validation, Vi(CA)2 T results are currently being compared to the ACES creep database [17].
References 1. Bažant, Z.: Prediction of concrete creep and shrinkage: past, present and future. Nucl. Eng. Des. 203(1), 27–38 (2001) 2. Chen, F., Sanahuja, J., Bary, B., Le Pape, Y.: Effects of internal swelling on residual elasticity of a quasi-brittle material through a composite sphere model. Int. J. Mech. Sci. 229, 107390 (2022) 3. Cheniour, A., et al.: FFT-based model for irradiated aggregate microstructures in concrete. Mater. Struct. 55, 214 (2022) 4. Galenne, E., Michel-Ponnelle, S., Salin, J., Moreau, G., Sanahuja, J.: The aim of computational methods for managing concrete structures in nuclear power plants. In: Euro-C, Computational Modelling of Concrete Structures, St. Anton am Arlberg, Austria (2014) 5. Guang, Y., et al.: GP2a - benchmark numerical simulation in micro-level RRT+. In: Synercrete18 - Interdisciplinary Approaches for Cement-Based Materials and Structural Concrete Synergizing Expertise and Bridging Scales of Space and Time, Funchal, Portugal, October 2018 6. Guihard, V., Patapy, C., Sanahuja, J., Balayssac, J.-P., Taillade, F., Steck, B.: Effective medium theories in electromagnetism for the prediction of water content in cement pastes. Int. J. Eng. Sci. 150, 103273 (2020) 7. Irfan-ul-Hassan, M., Pichler, B., Reihsner, R., Hellmich, C.: Elastic and creep properties of young cement paste, as determined from hourly repeated minute-long quasistatic tests. Cem. Concr. Res. 82, 36–49 (2016) 8. Königsberger, M., Honório, T., Sanahuja, J., Delsaute, B., Pichler, B.: Homogenization of nonaging basic creep of cementitious materials: a multiscale modeling benchmark. Constr. Build. Mater. 290, 123144 (2021) 9. Le Pape, Y., Sanahuja, J., Tran, N., Reviron, N., Charpin, L., Petit, L.: Vi(CA)2 T - virtual cement and concrete ageing analysis toolbox. In: 7th International Conference on Multiscale Materials Modeling, Berkeley, California, USA, October 2014 10. Le Pape, Y., Toulemonde, C., Masson, R., El Gharib, J.: Benhur and Vi(CA)2 T, two toolboxes to model concrete as an heterogeneous material combining analytical and numerical approaches. In: Concrete Modelling Conference, Delft, The Netherlands (2008) 11. Sanahuja, J., Dormieux, L.: Creep of a C-S-H gel: micromechanical approach. Int. J. Multiscale Comput. Eng. 8(4), 357–368 (2010)
A Multiscale Multiphysics Platform
147
12. Sanahuja, J., Dormieux, L., Chanvillard, G.: Modelling elasticity of a hydrating cement paste. Cem. Concr. Res. 37(10), 1427–1439 (2007) 13. Sanahuja, J., Dormieux, L., Le Pape, Y., Toulemonde, C.: Modélisation micro-macro du fluage propre du béton. Dans 19e Congrès Français de Mécanique, Marseille, France (2009) 14. Sanahuja, J., Tran, N.-C., Charpin, L., Petit, L.: Vi(CA)2 T v2: can a concrete material properties simulation code be both physics-based and engineer-friendly? In: Technological Innovations in Nuclear Civil Engineering 2016, Paris, France, September 2016 15. Soleilhet, F., Sanahuja, J., Adia, J.-L.: Calibration of multi-physics models on weakly instrumented structures: applications to containment buildings. In: SynerCrete23 - The International RILEM Conference on Synergising Expertise Towards Sustainability and Robustness of Cement-Based Materials and Concrete Structures, Milos Island, Greece, June 2023 16. Wyrzykowski, M., et al.: Numerical benchmark campaign of COST action TU1404 microstructural modelling. RILEM Techn. Lett. 2, 99–107 (2017) 17. Šmilauer, V., Havlásek, P., Dohnalová, L., Wan-Wendner, R., Bažant, Z.: Revamp of creep and shrinkage NU database. In: Biot-Bažant Conference, Northwestern University, Evanston, IL, USA, June 2021
Estimation of Protected Paste Volumes by Dirichlet Tessellation Associated with Point Processes of Air Voids Kazuya Ohyama and Shin-ichi Igarashi(B) Kanazawa University, Kanazawa 920-1192, Japan [email protected]
Abstract. Hardened cement paste is susceptible to frost damage. The use of an air entrainment agent makes concrete frost durable. This is explained by the presence of the protected paste volume in the vicinity of entrained air voids. In this study, a model using Dirichlet tessellation is proposed to estimate the protected paste volume associated with a random system of real air voids. The characteristic distance for the protection is defined by the ratio of the cumulative polygon tile areas to the entire cement paste. A critical distance is statistically determined by simulating random point patterns. To protect 95% of the paste area, the proposed model suggested that the average distance of about 0.22 mm should be protected at the maximum in a local region where air voids were sparsely present due to random fluctuation. This estimation was almost equivalent to the spacing factor recommended for the moderate exposure condition. This fact explains why the spacing factor has been useful as a safe characteristic distance parameter in determining air contents. The distance equivalent to the spacing factor reflects the most disadvantageous condition that can occur locally in the actual distribution of air voids against frost attack. Keywords: Air Void · Point Process · Dirichlet Tessellation · Simulation · Confidence Interval
1 Introduction Air entrainment is the most common and effective method to provide concrete with frost resistance. The frost resistance provided by air entrainment depends on not only the total air content but also the air void system quality [1]. The quality means the state of the random spatial arrangement of polydisperse air voids. The frost protection by those random air voids is explained by the concept of protected paste volume, which was originally proposed by Powers [2]. Each air void protects a local region of cement paste ranging from its surface to some distance against frost attack. If any location in the cement paste matrix is within protected regions around air voids nearby, the concrete is considered to have adequate freeze-thaw durability. In other words, to make the concrete durable against freeze-thaw cycles, the whole cement paste matrix should be covered by the patches of protected regions formed around air voids. The quality of the air void system greatly affects the coverage ratio by the protection patches. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 A. J˛edrzejewska et al. (Eds.): SynerCrete 2023, RILEM Bookseries 43, pp. 148–158, 2023. https://doi.org/10.1007/978-3-031-33211-1_14
Estimation of Protected Paste Volumes
149
If the size of the protected region is known in concrete, it may be possible to entrain air to meet the protection requirements. However, the protection range is determined by not only the geometrical arrangement of air voids but also material properties such as the local permeability of water in cement pastes, i.e., mix proportions and microstructure of concrete [2]. This is because the protection is closely related to the distance that water needs to travel to reach air voids. Hasholt [3] has proposed a new insight for this mechanism from connectivity between air and capillary pores, in which the probability of the connection is hypothesized to depend on the total surface area of air voids. In addition to those geometrical characteristics, several environmental conditions such as a freezing rate may also affect the size of the protected region. Thus, it would be difficult to specify the protection by air voids as a unique material property. However, at least, it is well recognized that the air void system in which spacing between air voids is shorter than a certain distance leads to freeze-thaw durable concrete. The distances between air voids or the volume fractions of cement paste within a certain distance from the air void are one of the decisive parameters in evaluating the freeze-thaw durability of airentrained concrete. Those parameters have been routinely evaluated by the conventional test method such as ASTM C456 and if a more sophisticated evaluation is required, some rigorous mathematical models have been proposed. The theoretical backgrounds and characteristics of those models are reviewed in detail by Snyder et al. [4]. The authors have recently proposed a simple model in which air voids observed in a 2D plane are treated as a planar point process [5]. In the model, the characteristic distance of air void spacing or the distance from a given location to the nearest air void is statistically estimated using the nearest neighbor distance distribution function for planar point processes. The characteristic distances obtained by the model are consistent with the spacing factors and have a good correlation with the spacing factor. Further, the procedure to evaluate the characteristic distance from an observed point pattern is quite easy as such it can be used as a routine test method for practitioners. It should be noted that the model can be also regarded as a method to estimate paste-air void proximity, which has been of great interest. As for the protection area in a point pattern, a certain polygonal area of cement paste can be allocated with each air void by applying a mathematical tessellation algorithm to the whole cement paste around air voids [6]. The various sizes of polygonal tiles associated with each point also reflect the nearest distance to an air void from a given location in cement paste. Once the tile tessellation is obtained, it is quite natural to assume that the nearest air void provides frost protection within the tile area. The objective of this study is to propose a model in which the Dirichlet tile area assigned to each air void is regarded as the region where the air is responsible for protection. The variations in the tile sizes due to different spatial arrangements of air voids are represented by the simulations generating random point patterns in the cement paste phase around aggregate particles. A critical distance for paste-air proximity is defined by the upper limit of the confidence interval of cumulated tile areas. The distances and volume ratios of the protected region inferred by the model are compared with the conventional spacing factor and the volume fraction of protected paste volumes estimated by the surface density of air voids. The significance of the conventional spacing factor is discussed in terms of the intrinsic random nature of air voids, which is estimated by the Dirichlet tessellation model.
150
K. Ohyama and S. Igarashi
2 Experimental Procedures 2.1 Materials and Mix Proportions of Concretes Ordinary Portland cement was used. Coarse aggregate was crushed gravel (density 2.66 g/cm3 , absorption 0.52%, maximum size 20 mm). Fine aggregate was river sand (density 2.58 g/cm3 , absorption 2.22%). A commercial AE agent (a modified rosin oxide) was also used. The design air content was 4.5%. Concrete specimens (100 mm × 100 mm × 400 mm) were produced. They were demolded at 24 h after casting and cured in water (20 ± 2 ◦ C). The mix proportion of the concrete is shown in Table 1. Table 1. Mix proportion of concrete W/C (%)
s/a (%)
Slump (cm)
Design air content (%)
Unit content (kg/m3 ) Water
Cement
Sand
Gravel
AE agent (%wt/C)
Measured air content (%)
50
45
10 ± 2
4.5
165
330
797
1003
0.007
5.0
2.2 Image Acquisition and the Automatic Linear Traverse Measurement of Air Voids After 21d curing, plate specimens (100 mm × 100 mm × 30 mm) were cut out. Their surfaces were polished to be able to see the edges of fine air voids following the procedure of ASTM C457. The central part of 60 mm × 60 mm in the polished section was used for the subsequent image acquisition. The whole observation area was segmented into 11 × 11 = 121 quadrats, and an image of each quadrat was saved into a computer using a digital camera attached to the microscope. The size of the quadrat is 6.14 mm × 6.14 mm of 2048 × 2048 pixels. The image size and resolution are 4M and 3 μm/pixel, respectively. Using the automatic system of image analysis, air voids in each quadrat were automatically identified by differences in shadows between direct and lateral lighting. The chord length of air voids and the spacing factor were also automatically calculated using software equipped with the equipment. Following the traverse length specified in the ASTM C457, the pitch of traverse lines was properly determined so that 4 traverse lines were used for each quadrat. The total traverse length for the 121 quadrats was 2745 mm. The spacing factor L and other records obtained by the automatic linear traverse measurement are given in Table 2. 2.3 Conversion of Air Voids to a Planar Point Process and Dirichlet Tessellation of Cement Paste Matrix The area fraction of the air voids in each quadrat (i.e., observation window W) was simply obtained by image analysis. Centroids xi (i = 1, . . . , n) of each air void were
Estimation of Protected Paste Volumes
151
Table 2. Records of the linear traverse measurement Air content in hardened concrete A (%)
Paste fraction p
Specific surface area α (/mm)
Spacing factor L (μm)
3.6
0.269
31.3
178
also determined by a general procedure of image analysis for particles. Then, all the air voids were converted into a set of spatial points whose positions agreed with the centroids of the original real air voids. This set of points was regarded as a spatial point process X = {xi : i = 1, · · · j, · · · n} ⊂ W . The Dirichlet tessellation associated with the point process X is computed using an R package [5]. A Dirichlet tile that includes a point xi ∈ X is a polygonal region, and any location within the tile is closer to the point xi than any other points xj=i ∈ X . This means each tile can be regarded as an area in which the internal point representing the air void should give a minimum level of protection from freeze-thaw attack. The tile region C(xi |X ) allocated to a point xi ∈ X is defined by Eq. (1): C(xi |X ) = u ∈ W : u − xi = minu − xj (1) j
where, u ⊂ W \ X is a given point within an observation field, i.e., a quadrat image. A schematic drawing of a random point pattern and its Dirichlet tessellation is shown in Figs. 1(a) and (b). The area of each tile was computed, and a histogram of the frequency of the tile areas was arranged (Fig. 1(c)). Based on the general distribution of data shown in the histogram, the distribution was approximated by the Gamma distribution (Fig. 1(c)). 2.4 Simulation of Point Patterns and Determination of the Characteristic Tile Area Point intensity λA of air voids per unit area of cement paste was determined by the total number (≈17500) of air voids and the area fraction of cement paste in 121 quadrats. Then a local point intensity λi in a quadrat was assumed to vary following the Poisson distribution with the parameter λA . The local point intensity λi was generated following the Poisson distribution. Using the generated intensity λi , a point pattern with the intensity λi was simulated for the cement paste area. The Dirichlet tessellation associated with the simulated point pattern was executed at every simulation and tile areas excluding aggregate particles were calculated. Then the frequency of the tile areas without aggregate was arranged to obtain a histogram. The histogram was approximated with Gamma distribution, and its cumulative probability density function was calculated (Fig. 1(d)). This procedure was repeated 199 times. The simulation for λi (i = 1, · · · 199} and the real point pattern, are considered as a test of a significance level of 0.05. Based on the cumulative probability density function for each simulation and the real distribution, the maximum tile area for covering 95% of the entire cement paste with protective patches
152
K. Ohyama and S. Igarashi
(b)
(a)
(c)
Frequency
Cumulative Probability
(d)
Tile Area
Tile Area
Fig. 1. Process to evaluate the Dirichlet tile areas: (a) upper left; random points (b) uppr right; Dirichlet tessellation (c) lower left; histogram and probability density function (d) lower right; cumulative probability and 95th percentile
around each point was determined (Fig. 1(d)). The tile areas also exhibit a certain variation. Thus, the tile areas corresponding to the upper limit of the confidential interval of 95% were presented as a histogram. This distribution is regarded as a variation of the furthest distances from a given location to the nearest air void [7]. 2.5 Characteristic Distance Defined by the Nearest Neighbor Distance Distribution Function for a Point Process For a point xi ∈ X , its shortest distance to the other point xj ∈ X is written as di = d (xi , X \ xi ). Then, for any r ≥ 0, the nearest neighbor distance distribution function G(r) is defined by the following probability equation: G(r) = Pr{di = d (xi , X \ xi ) ≤ r|xi ∈ X }
(2)
This function is a cumulative probability function of the distance r (Fig. 2). It converges to unity at long distances. The distance that corresponds to the probability of 0.5 is defined as a median distance R50 . The distance di = d (xi , X \ xi ) is the distance between two points included in the point process, X . A characteristic distance L is defined using the median distance R50 . R50 is modified to define L [4].
L = R50 −
DA 2
(3)
where DA is the mean diameter of air voids and is easily obtained via simple image analysis. The nearest neighbor distance distribution function G(r) is alternatively regarded as
Estimation of Protected Paste Volumes
153
the spherical contact distribution function when the point process is completely spatially random, i.e. the homogeneous Poisson point process. Therefore, the value L may also be interpreted as the distance from an arbitrary point u ⊂ W \ X , to the edge of the nearest air void. In other words, the characteristic distance L is interpreted as a parameter of cement paste-air void proximity also. 1.0 0.9 0.8 0.7
G(r)
0.6 0.5 0.4
Void
Void
0.3 0.2 0.1 0.0
R50
Distance r
Fig. 2. Characteristic distance L defined by the nearest neighbor distance distribution function G(r).
3 Results and Discussion 3.1 Dirichlet Tessellation Associated with an Actual Point Pattern Taking a nominal paste content and dispersion of aggregate particles into account, the quadrat including air voids shown in Fig. 3(a) was chosen as a representative region among 121 images. The Dirichlet tessellation of the point pattern is shown in Fig. 3(b). It seems air voids are clustered in a local region where aggregate particles are also clustered. The simulation generating random point patterns as air voids was also executed in this paste region. Since air voids are assumed randomly dispersed, an arbitrary selection of images does not lose generality as far as an entirely spatially random process is simulated.
(a)
(b)
1mm
Fig. 3. Image of air voids chosen as a representative region (a) and the Dirichlet tessellation associated with the point process of air voids (b).
154
K. Ohyama and S. Igarashi
Fig. 4. Histograms for the Dirichlet tiles plotted against tile areas (a) and against the equivalent radius of circles with the same area of each polygonal tile (b).
Figure 4(a) shows the histogram for the Dirichlet tiles associated with the real point pattern (Fig. 3(b)). The entire distribution is asymmetrical and has a long tail toward the larger tile area since the minimum size of the tile area is limited to zero whereas the maximum size is not fixed. Such a character of the entire distribution can be approximated with the Gamma distribution. Furthermore, a distribution shown in the histogram suggests that there exist some areas where there are few air voids due to fluctuation of random dispersion. Otherwise, the presence of large tiles also suggests that the air voids within the tiles are so large that other voids cannot get closer within their radius. Figure 4(b) also shows the histogram for the same tile areas as Fig. 4(a), but the class is the radius of circles equivalent to the areas of the original polygonal tiles. The distribution of the radius seems more symmetric than the direct estimation of polygonal areas in Fig. 4(a). The equivalent radius to the mean tile area is about 151 μm. The average diameter of air voids was 68 μm for about 17500 air voids tallied in 121 quadrats. If it is assumed that the tile of the average size has an air void with the mean diameter, the paste thickness that needs to be protected is 117 μm. This distance is smaller than the measured spacing factor of 178 μm and much smaller than the spacing factor of 200 μm which is usually recommended for concrete exposed to moderate climate conditions. Thus, if the air in this concrete has a protected region of 200 μm in thickness, in other words, the microstructure of the cement paste allows water to travel this distance during
Estimation of Protected Paste Volumes
155
the freezing process, the tiles with the average size could be sufficiently protected by air entrainment in the concrete. 3.2 Dirichlet Tessellation for Simulated Point Patterns with the Same Point Intensity The distribution of air voids is random although their point process has a short exclusive distance from each other due to the sizes of air voids themselves. The tile size to cover 95% of the paste area changes in various random patterns of air voids. Figure 5 shows the histogram of the maximum tile sizes to attain the coverage ratio of 95% of the whole paste area in the simulated point patterns.
Fig. 5. Maximum tile sizes to attain the area fraction of 95% to the entire cement paste matrix.
The distribution in the histogram is almost symmetrical and has less variance since those tiles are a set of specific sizes to reach the upper limit of the confidence interval. This is in contrast to the histograms shown in Figs. 4 (a) and (b) which are a series of samples from the smallest to the largest tiles in the real pattern. The average radius of the 95% maximum tiles of the simulation is about 250 μm, which is about 100 μm greater than the average tile size. Also, assuming here that an air void of the average diameter is present in the tile, the furthest distance from the surface of the air void to the edge of the tile is 216 μm (Fig. 6). Alternatively, assuming a monodisperse sphere determined by the average chord length in the linear traverse measurement and the spacing factor for regular cubic lattice arrangement, the distance from the sphere center to the farthest point is about 274 μm, which is approximately equivalent to the maximum tile size of 250 μm. It has been known that most of the cement paste matrix in conventional concrete is within the range distance of the spacing factor from the surfaces of air voids [7, 8]. The spacing factor estimates a greater distance than the real average distance to the furthest locations from each air void. The tile that corresponds to the upper limit of 95% confidence interval is the largest among the tiles occupying 95% of the protected region. Thus, the size of the protected region determined by the Dirichlet tessellation is consistent with the fact that the distance corresponding to the spacing factor covers most of the paste region. In other words, the distance equal to the spacing factor of
156
K. Ohyama and S. Igarashi
this concrete represents the real situation with less protection resulting from the sparse number of air voids due to the fluctuation of random distribution. This means that the distance comparable to the spacing factor is a safe distance for the freeze-thaw durability design which is realized in a real air void system. radius for covering 95% of cement paste radius of average tile
average size circle of air voids
151 212 250
68 mono-sized sphere assumed for calculation of the spacing factor
192
distance of spacing factor 178 (µm)
Fig. 6. Relationship between the tile sizes in the proposed model and the air void systems assumed for the conventional spacing factor.
3.3 Relationship Between Characteristic Distances of the Nearest Neighbor Distance Distribution Function and the Tile Sizes of Dirichlet Tessellation When a point process is completely spatially random, as mentioned before, the nearest neighbor distance distribution is interpreted also as the spherical contact distribution [5]. The latter has a similar perspective of paste-air proximity. Furthermore, it is also reflected in the Dirichlet tile size distribution. Figure 7 is the nearest neighbor distance distribution function of actual air void patterns. The median distance R50 and the characteristic distance L are 220 μm and 186 μm, respectively. The latter is as expected, comparable with the thickness to be protected in the average tile containing the mean air void (Fig. 4). The authors have pointed out that the characteristic distance L can be used as a parameter to quantify the spatial distribution of real air voids [5]. It is found from the similarity of the two distances that the characteristic distance L of an air void pattern may be also interpreted as the average thickness which the average air void should protect from the frost attack. To evaluate distances or probability of paste-air proximity rigorously, it is needed to consider the particle size distribution of air voids with various sizes. In particular, the nearest surface distance is different from the distance to the nearest sphere center in a polydisperse system like air voids. However, as shown in Fig. 7, even the point process ignoring the size of polydisperse air voids could give significant distances that are consistent with the concept of the protected paste volume. This fact suggests that the point intensity reflects the air content [5] and that the distances between points implicitly represent the size distribution of air voids and the exclusive distance between each other.
Estimation of Protected Paste Volumes
157
Fig. 7. Nearest neighbor distance distribution function for the real air void pattern.
3.4 Relationship Between the Surface Density of Air Voids and the Tile Sizes for 95% Protection The estimation of the maximum tile size as an upper limit of the 95% confidence interval is in principle the same as the evaluation of the volume fraction that spreads from the surfaces of air voids. The tessellation assumes various tile areas as minimum responsibility of protection which each air void should have to ensure frost resistance. It ignores the superposition of the protected regions. In other words, the model assumes the thickness where frost protection is required is not constant. Meanwhile, if the constant shell thickness of the protection is assumed, the protected paste volume can be simply estimated by the products of the surface density and the assumed thickness. The surface density, SV is calculated by the fundamental stereology equation; SV = 2 PL , where PL is the number of interface points between air and cement paste per unit length. This number of those interface points is automatically obtained by the conventional linear traverse measurement, i.e., twice the number of chords on the air voids. Then, if the thickness of the protection is assumed δ, the total volume of the protected paste is a product of SV δ. Assuming this volume fraction of 95% to the entire paste volume, the required thickness is estimated at 225 μm. Almost the same thickness as 225 μm is obtained, i.e., 227 μm if the equation of the relation between the specific surface α and the average chord length l is used. These values derived from the fundamental stereology are approximately equivalent to a tile size of 250 μm. Thus, the tile size estimated by the upper limit of the confidence interval is also consistent with the evaluation based on classical stereology rules and the assumption of constant shell thickness as protected paste volume. It is confirmed that the protection thickness comparable with the spacing factor results in the coverage of most of the paste region. Furthermore, because there is consistency between the proposed model in which air void sizes are ignored and the conventional estimation in which air voids are considered as monodisperse spheres, features of spatial point patterns seem to reflect the presence of various sizes of air voids in the point patterns.
158
K. Ohyama and S. Igarashi
4 Conclusions A model using Dirichlet tessellation associated with the point pattern of air voids was proposed to estimate protected paste volume against frost attack. In the model, each tile area is simply regarded as the region where the air void in the tile should provide the minimum protection. The polygon tile areas around air voids were accumulated in ascending order and specified a tile area that attained the upper limit of the confidence interval to cover 95% of the entire paste area. The major results obtained in this study are summarized below. • The area distribution of Dirichlet tile sizes associated with a real point pattern can be approximated by the Gamma distribution. • The distance comparable to the spacing factor is sufficiently larger than the furthest distance of the average tile area associated with an air void. • The tile area to attain the coverage of 95% of the entire cement paste with the protected region is almost comparable with the distance of the spacing factor. • The characteristic distance defined by the nearest neighbor distance distribution function for the air voids corresponds to the furthest distance of the average Dirichlet tile area. • The protection distance estimated by the proposed model does not contradict both the conventional spacing factor and the volume of the protected region estimated by the fundamental stereology equations. • Features of distances in the point patterns for air voids seem to reflect the fact that they have their sizes. Acknowledgments. The authors would like to thank Prof. Rolf Turner, University of Auckland, New Zealand, for his advice on R programming. This work was supported by JSPS KAKENHI Grant Number JP21K04211.
References 1. Tunstall, L.E., Ley, M.T., Scherer, G.W.: Air entraining admixtures: mechanisms, evaluations, and interactions. Cem. Concr. Res. 150, 106557 (2021) 2. Powers, T.C.: The air requirement of frost-resistant concrete. Proc. Highw. Res. Board 29, 184–211 (1950) 3. Hasholt, M.T.: Air void structure and frost resistance: a challenge to powers’ spacing factor. Mater. Struct. 47(5), 911–923 (2013). https://doi.org/10.1617/s11527-013-0102-9 4. Snyder, K., Natesaiyer, K., Hover, K.: The stereological and statistical properties of entrained air voids in concrete: a mathematical basis for air void system characterization. In: Mindess, S., Skalny, J. (eds.) Materials Science of Concrete VI, pp. 129–214 (2001) 5. Murotani, T., Igarashi, S., Koto, H.: Distribution analysis and modeling of air voids in concrete as spatial point processes. Cem. Concr. Res. 115, 124–132 (2019) 6. Baddeley, A., Rubak, E., Turner, R.: Spatial Point Patterns, Methodology and Applications with R. CRC Press, Boca Raton (2016) 7. Mayercsik, N.P., Felice, R., Ley, M.T., Kurtis, K.E.: A probabilistic technique for entrained air void analysis in hardened concrete. Cem. Concr. Res. 59, 16–23 (2014) 8. Son, Y., et al.: A 3D petrographic analysis for concrete freeze-thaw protection. Cem. Concr. Res. 128, 105952 (2020)
MASKE: Particle-Based Chemo-Mechanical Simulations of Degradation Processes Enrico Masoero(B) School of Engineering, Cardiff University, Cardiff, UK [email protected]
Abstract. The long-term performance of new cementitious materials is uncertain. To help predict durability and manage uncertainty, the models must capture the fundamental mechanisms driving degradation in new binders. Such mechanisms evolve slowly and are strongly chemo-mechanically coupled, all of which challenges the existing simulations. This article presents MASKE: a simulator of microstructural evolution based on interacting particles, which represent multiphase solid domains in an implicit ionic solution. Chemical reactions, sampled through Kinetic Monte Carlo to reach long time scales, determine precipitation/dissolution processes whose rates depend on solution chemical potentials and on particle interaction energy. Published results from MASKE have already addressed aggregation-driven precipitation of C-S-H nanoparticles, stress-driven dissolution of C3 S crystals at screw dislocations, and carbonation of calcium hydroxide in a C-S-H matrix. Here new results are presented, focussing on a nanocrystal of calcium hydroxide and discussing: (i) its chemical equilibrium and dissolution/growth kinetics in stress-free conditions, and (ii) the emergence of pressure-solution and crystallization pressure when the crystal is compressed between platens. Similar chemo-mechanical processes contribute to important degradation modes of concrete, such as creep, sulphate attack, and alkali-silica reaction. Keywords: Simulation · Degradation · Dissolution and growth · Pressure solution · Crystallization pressure
1 Introduction The current environmental targets are putting the concrete industry under pressure to reduce its carbon emissions [1]. Part of the solution is to transition to less CO2 intensive cements, such as LC3 and other alkali activated binders [2]. However, the long-term performance of these new binders is largely unknown, and there is documented risk that accelerated aging tests may not trigger representative mechanisms of slower degradation during service [3–5]. Modelling and simulation can support the transition, by providing a fundamental understanding of the degradation mechanisms in new materials, thus informing better extrapolations of short-term results to long time scales. Concrete degradation is underpinned by chemical reactions at the nanoscale and deformation at the microscale. At the nano-to-micro scale, free energy changes from © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 A. J˛edrzejewska et al. (Eds.): SynerCrete 2023, RILEM Bookseries 43, pp. 159–170, 2023. https://doi.org/10.1007/978-3-031-33211-1_15
160
E. Masoero
chemical reactions and mechanical stress are comparably large, hence strongly coupled chemo-mechanical processes emerge. Two examples are: (i) stress-driven dissolution, causing non-linear dissolution vs. supersaturation curves in various minerals [6] and contributing to creep deformations [7] (along with its dual process of dissolution-driven stress relaxation which also contributes to creep [8, 9]); (ii) crystallization pressure [10], underpinning degradation processes such as freeze-thaw, alkali-silica reactions, and delayed ettringite formation. Chemo-mechanical coupling is captured by ab-initio and atomistic simulations, but these are limited to length and time scales that are too small to address the degradation of concrete. At the nano-to-micro mesoscale, particlebased simulations have coarse-grained the results of atomistic simulations into potentials of mean force, used for studying mechanical processes directly at the mesoscale [11]. However, chemical evolution in particle-based simulations has been either neglected or oversimplified, which is a limitation for the degradation processes in concrete. This article adopts MASKE: a particle-based simulator of chemo-mechanical evolution at the nano-to-micro mesoscale, in complex mineral systems within a co-evolving ionic solution. A precursor of MASKE was used in 2017 to simulate precipitation of coarse-grained C-S-H nanoparticles [12], showing how both mechanical aggregation and solution saturation contribute to the early hydration rate curve of Portland cement. In its current version, except slightly different rate equations, MASKE was used in 2020 to simulate the dissolution of a C3 S crystal, where each particle represented one C3 S molecule [13]. Focussing on a high-stress intersection between a screw dislocation and the mineral surface, the simulations predicted the experimentally observed sigmoidal dependence of the dissolution rate on the solution supersaturation. Finally, MASKE was recently used to simulate carbonation, i.e. precipitation of CaCO3 particles in a system of dissolving calcium hydroxide and C-S-H, both modelled at the molecular scale (1 particle = 1 molecule) and in contact with an aqueous solution rich in CO3 -2 [14]. This article starts with an overview of the MASKE simulator in its current version, presenting its Kinetic Monte Carlo approach that allows sampling long timescales, and the rate equations for dissolution and precipitation events. The explanation is targeted to simulations with only one mineral species, Ca(OH)2 (called CH), where each particle represents one molecule. The system is parametrized for CH, including potentials of mechanical interaction between particles and thermodynamic and kinetic parameters that control the dissolution and precipitation reactions. A spheroidal nanocrystal is then built from an orderly agglomeration of CH particles. The crystal is first studied in stressfree conditions, predicting the mechanisms and rates of dissolution and growth at various levels of solution saturation, determined by the concentrations of Ca2+ and OH− ions in solution. Then two steel platens are added to compress the crystal at pressures between 10 and 90 MPa. With this added mechanical stress, the simulations capture the buildup of creep deformations from pressure-induced dissolution, as well as the swelling of the system when a high solution saturation drives strong crystallization.
MASKE
161
2 Methodology 2.1 The MASKE Simulator MASKE is a C++ code to simulate the kinetic evolution of microstructures featuring multiple solid phases in a fluid medium. The fluid is described by the average concentrations of chemical species in it, e.g. water and ions. The solids are modelled as spherical particles that interact mechanically through effective potentials: Fig. 1. One particle may be as small as a single molecule (as in this article) or it may coarse-grain thousands or millions of molecules. An implicit additional volume of solution Vsol can also be attached to the simulation box, to allow for solution buffering when needed. The evolution of the system is driven by continuous and discrete events, which MASKE places onto a common timescale. Continuous events are integrated over time with user-assigned time step; examples may be diffusion equations or equations of motion. This article will not consider continuous events but only discrete events, for which there is a time lag t between two events. Examples are the precipitation or dissolution of a solid particle, which also alter the solution by consuming or releasing ions. t is not assigned a priori and it typically changes during a simulation, depending on the evolving composition of the solution and mechanical interactions between particles.
Fig. 1. Left: a typical simulation snapshot (using OVITO [15]) with steel platen particles in grey, a crystal of real Ca(OH)2 particles for dissolution in red, and trial Ca(OH)2 particles for precipitation in blue. Right: harmonic interaction potential between pairs of Ca(OH)2 particles.
MASKE uses off-lattice rejection-free Kinetic Monte Carlo (KMC) to compute the t until the next discrete event. First a list of possible discrete events is created, e.g. dissolution of N existing particles and precipitation of a new particle at M possible sites. Each possible event is given a rate Ri and the KMC algorithm computes t as: 1 1 (1) · ln t = N +M u i=1 Ri u is a random number uniformly distributed between 0 and 1. An event is selected with probability proportional to its Ri and carried out. The rates Ri depend on the chemical reactions driving the events, for which the user must provide stoichiometries and thermodynamic and kinetic parameters. The rates also depend on the underlying
162
E. Masoero
mechanisms, e.g. classical rather than nonclassical nucleation, growth and dissolution. In this article, since each particle will represent only one solid molecule, Transition State Theory (TST) will be the starting point to express the dissolution and precipitation rates. The precipitation rate of a unimolecular particle in MASKE is: ΔGp‡ kB T −α/3 α/3−1 exp − = kp Qr,p V · Vt (2) rp = Qr,p V · Vt ‡ ‡ k T B hc γ kp is the rate constant of the precipitation reaction; MASKE explodes it into its parts to capture the effect of temperature T on the rates. This brings in the Boltzmann kB and Planck h constants and the activity coefficient γ‡ , concentration c‡ , and activation energy G‡p of the activated complex for precipitation in standard state. G‡p can be obtained from experiments at different T, but its value depends on the arbitrary choice of c‡ , hence the two parameters must be provided consistently. Qr,p is the activity product of the nr molecules in solution, with concentrations ci , that react to form the solid: n r γi ci (3) Qr,p = i=1
The activity coefficients γi are computed using a version of the Debye-Hückel theory that accounts also for uncharged molecules and concentrated electrolytes: √ √ −z 2i A I log10 (γi ) = (4) √ + bi I 1 + Bai I zi is the charge of the ith molecule, A and B are solvent-specific constants (0.51 and 3.29 nm−1 for water), ai is the hydrated radius of the molecule in solution, bi is a molecule-specific constant, and I is the ionic strength of the fluid with all its ns species: ns I= ci zi2 (5) i=1
To sample precipitation events, MASKE creates a lattice of trial particles with a user provided geometry, e.g. the cubic lattice in Fig. 1. V in Eq. (2) is the volume of one lattice cell. The trial particles are assigned the same interaction potentials as the corresponding species of real particles, and then the total interactions energy of the system is minimized while keeping all real particles fixed in their positions. This moves each trial particle to the most favourable precipitation site close to it (note the off-lattice nature of this operation). All the trial particles within the same attractive basin of a given set of real particles will ultimately move to the same local energy minimum, i.e. the same site. This implies that, for an infinitely fine sampling lattice with V → 0, the sum in Eq. (1) eventually attributes a cumulative volume weight Vt to the generic precipitation site, with Vt being the volume of the attractive interaction basin surrounding the site. α/3 Therefore, the overall rp in Eq. (1) associated to the site will eventually scale as Vt , ‡ ‡ where α is the spatial dimension of c . α = 3 if c is per unit volume and α = 2 if c‡ is per unit surface; the latter is typical for dissolution and precipitation processes. All in all, the units of rp in Eq. (1) are correctly number of events per unit time.
MASKE
The dissolution rate of a unimolecular particle in MASKE is:
ΔGd‡ α/3 kB T k Qr,d Vt rd = hc‡ γ‡ exp − kB T exp − UdkB−U T
α/3 k Qr,d Vt = kd exp − UdkB−U T
163
(6)
G‡d is the standard state activation energy of the dissolution reaction, kd its reaction constant, and Qr,d the activity product of the molecules reacting to dissolve the solid. Vt is the volume of the attraction basin surrounding the particle that is being sampled for dissolution. The second exponential term in Eq. (6) is new and it features the dissolutioninduced change in (position-dependent) excess free energy of a particle Ud , minus the same energy change for a stress-free particle occupying a kink position in a crystal lattice Uk . The rationale for this term is that adding or removing a stress-free particle in kink position should not alter the free energy of the solid. Particles that are less connected than a kink or under mechanical stress will be in a higher energy state than a kink, hence their dissolution would entail a Ud − Uk < 0 and thus a greater rate than better-connected, less stressed particles. The above rate equations may be used directly to sample precipitation and dissolution, but this may lead to inefficient simulations, with many precipitation events at energetically unfavourable locations, as rp in Eq. (2) is position-independent, immediately followed by fast dissolution. A way to avoid such fluctuations is to employ net rates: ΔGd‡ Qp,d Ud − Uk kB T α/3 net V exp − Q exp − − rd = rd − rd − = r,d kB T kB T Keq,d t hc‡ γ‡ (7) rpnet
= rp − rp−
ΔGp‡ = exp − kB T hc‡ γ‡ kB T
Ud − Uk Qp,p α/3−1 Qr,p − exp − V · Vt kB T Keq,p
(8) rd- in Eq. (7) is the backward reaction for dissolution, whose expression is analogous to rp in Eq. (2). Therefore rd- brings in the activity product of the products of dissolution, Qp,d . The equilibrium constant Keq,d of the dissolution reaction emerges because G‡p = G‡d + RT ln(Keq,d ). Analogous considerations apply to Eq. (8), where rp- is the backward precipitation reaction, Qp,p is the activity products of the products of precipitation, and Keq,p is the equilibrium constant of the precipitation reaction. Equations (7) and (8) apply to simple, one-step, unimolecular reactions, but richer rate expressions could be used instead to capture more complex reaction mechanism e.g. coarse-grained rates in [12]. 2.2 The Calcium Hydroxide System in This Article This article will apply MASKE to simulate the dissolution and precipitation of a nanocrystal of calcium hydroxide, Ca(OH)2 (or CH in cement notation), with externally applied forces, in a aqueous solution of Ca2+ and OH− ions. CH was chosen as an
164
E. Masoero
example to demonstrate the capabilities of MASKE, without targeting processes that are technologically impactful for this mineral specifically. Indeed, CH was chosen because it is simple and its kinetic and thermodynamic parameters are established in the literature. The nanocrystal is described as a cluster of individual spherical particles arranged in a face centred cubic (FCC) lattice: see Fig. 1. Each particle represents one CH molecule with volume vm,CH = 0.05567 nm3 , which gives a particle diameter DCH = 0.4737 nm. Pairs i,j of particles at a distance rij interact through a harmonic potential (see Fig. 1):
1 r − r0,CH − ε0,CH if rij ≤ rc,CH k (9) Uij,CH (rij ) = 2 ij,CH ij 0 if rij > rc,CH r0,CH = 0.741/3 DCH = 0.4285 nm is the equilibrium distance, with correction to remove the artificial porosity of 0.26 introduced by the FCC arrangement (more details in [13]). Kij,CH = ECH πD2CH / 4 r0,CH is the interaction stiffness, with ECH = 48 GPa the Young modulus of CH [16]. The bond energy at equilibrium is set to ε0,CH = γCH πD2 /6 (see rationale in [13]), where γCH = 68.4 mJ/nm2 is the CH-water interfacial energy [17]. rc,CH the interaction cut-off, set to when Uij returns to 0 at rij > r0,CH , which leads to rc,CH = 0.4571 nm and therefore to a bond failure strain (rc,CH /r0,CH – 1) = 7.2%. Particle dissolution and precipitation follow the reaction: Ca(OH)2 Ca2+ + 2OH−
(10)
with Keq,d = 6.30866 ·10–6 [18] (and therefore Keq,p = K−1 eq,d = 1.585125 ·105 ). The net rates follow Eqs. (7) and (8), with γ‡ = 1 as typical in reactions involving [19], G‡d = 74 kJ/mol for a c‡ = 1 μ mol/m2 (so α = 2) as per rate constant in [20], hence G‡p = G‡d + RT ln(Keq,d ) = 44.6 kJ/mol. The activity coefficients of the Ca2+ and OH− follow Eq. (4) with aCa = 0.486 nm, bCa = 0.15, aOH = 1.065 nm and bOH = 0.064 [18]. In all cases, the reactant of the dissolution reaction as well as the product of precipitation is just solid CH, hence Qr,d = Qp,p = 1. Ukink = γCH πD2 as per rationale in [13]. Vt is approximated as to vm,CH rc,CH /r0,CH . Precipitation is sampled using a cubic lattice of trial particles (Fig. 1) with linear cell size of r0,CH /2, leading to V = 0.00984 nm3 . A spheroidal nanocrystal is first obtained starting from a cubic FCC crystal of linear size 10 r0,CH and placing it in an undersaturated solution with cCa = 15 mmol/L: see Fig. 2a. A large buffer solution volume Vsol = 1024 nm3 is added, so any dissolution has negligible impact on the concentrations of ions in solution. Only Ca and OH ions are assumed to populate the solution (which is an oversimplification for future improvements), so in all cases in this article cOH = 2 cCa . Therefore, this initial dissolution phase was conducted at saturation index β = Qr,p /Keq,d = 0.786, as per Eqs. (3)-(5), where β < 1 indicates an undersaturated solution in which the crystal should dissolve. The spheroidal crystal is then tested for stress-free dissolution and precipitation by running a set of MASKE simulations, each with a different level of cCa (and thus cOH ) to impose β values between 0 and 10 (cCa between 0 and 39.1 mmol/L), always with Vsol = 1024 nm3 to fix the concentrations. Chemical equilibrium is expected at β = 1. The next simulation campaign starts with two added steel (S) platens, one above and one below the CH crystal, discretized using spherical steel particles on an FCC lattice:
MASKE
165
z y x (a)
Simulated dissolution time (seconds)
(b)
Fig. 2. (a) Creation of a spheroidal Ca(OH)2 nanocrystal by dissolution of an initially cubic one at saturation index β = 0.786. (b) Addition of steel platens to apply stress along Z.
see Fig. 2b. The steel particles interact with each other through a potential analogous to Eq. (9), with parameters: DS = 0.6 nm (arbitrary), r0,S = 0.741/3 DS = 0.5427 nm, kij,S = ES πD2 S /(4 r0,S ) with ES = 200 GPa, ε0,S = γS πD2 /6 with γS = 200 mJ/nm2 (arbitrarily set just to create stronger bonds than in CH), rc,S = 0.5614. The platens are not allowed to dissolve or precipitate: they will just apply mechanical stress to the CH crystal. To this end, a purely repulsive Hookean potential is set between CH and steel particles: Uij,CH-S = ½ kij,CH-S (rij – r0,CH-S ) for rij ≤ r0,CH-S , and Uij,CH-S = 0 for rij > r0,CH-S , with kij,CH-S = 2 [2 kij,CH 2 kij,S /(2 kij,CH + 2 kij,S )] and r0,CH-S = ½ (r0,CH + r0,S ). All the particles in the platen below the CH crystal are fixed in position, whereas the upper platen is fixed on the horizontal XY plane and free to move vertically along Z. A vertical downward force fZ is applied to each of the 1,058 particles of the upper platen, whose horizontal area is (16 r0,S )2 , hence the vertical pressure between platens is σ z,plat = 11.48 fz , in MPa if fz is in pN (10–12 N). The platens are initially brought into contact with the crystal by applying a small σ z,plat = 10 MPa and running 106 steps of energy minimization using the quickmin algorithm in LAMMPS; no dissolution nor precipitation are allowed at this point. Then, three separate simulations are run, by initially setting σ z,plat = 10, 45 and 90 MPa respectively and re-equilibrating the system mechanically via 106 more quickmin steps. Then a solution with β = 1 is brought into the picture (cCa = 16.43 mmol/L), within a cubic simulation box of linear size 20 nm and a relatively small added solution volume Vsol = 104 nm3 . The crystal per se would not dissolve at β = 1, but now the applied stress is expected to induce dissolution-driven creep, with the upper platen progressively moving downward as a result. Lastly, the initial system with σ z,plat = 10 MPa is immersed in a solution with β = 100 (cCa = 92.1 mmol/L), fixed through a Vsol = 1024 nm3 ; this time, the chemical drive for precipitation is expected to push up the upper platen to some extent, as an effect of crystallization pressure. In all the examples, per-particle virial stresses are computed, using each particle’s volume as the averaging domain [9].
3 Results 3.1 Stress-Free Equilibrium and Dissolution/Growth of a CH Nanocrystal Figure 3 shows the dissolution and growth of an initially spheroidal, stress-free CH nanocrystal when the surrounding aqueous solution of Ca2+ and OH− ions is kept at various levels of constant saturation index β. As expected, at any β < 1 the crystal fully
166
E. Masoero
dissolves; the morphology in Fig. 3c for β = 0 may actually represent an intermediate step during dissolution at any other β < 1. When β = 1 some particles precipitate but on a very long timescale, hence with very low rate. Such precipitation occurs because the rich initial morphology in Fig. 3a features possible precipitation sites which are more favourable than kink sites, hence their rp net in Eq. (8) benefits from the contribution of Ud . However, when these sites are exhausted, precipitation stops and the morphology in Fig. 3b, which is just slightly more orderly than the initial one in Fig. 3a, is obtained. When β > 1 the crystal grows, but not indefinitely because that entails a high energy barrier to precipitate adatom particles on top of flat crystal surfaces; this could be achieved through a very large β, crystallography defects, or allowing for fluctuations by using the straight rates in Eqs. (2) and (6) instead of the net rates here (see [12] for more discussion on this). Within the scope of this article, Fig. 3e shows that a moderately high β = 4 (with cCa = 27.7 mmol/L) leads to the same small rearrangements as at β = 1, although on a much smaller timescale. At β = 10, instead, Fig. 3f shows much more precipitation and the ability to create new terrace ledges (although not new adatoms), which is sufficient to sustain significant crystallization.
Fig. 3. (a) Initial Ca(OH)2 nanocrystal. (b, c, e, f) Dissolution/precipitation at various constant saturations β, and corresponding configurations. (d) Rate vs. β curve from all the simulations.
Figure 3d gathers the rates per unit area of the crystal at all the explored β. The rates are obtained from the initial gradients of Figs. 3b, c, e, f and assuming the crystal to be a perfect sphere of known volume N·vm,CH , where N is the number of particles in a simulation at any given time. Figure 3d shows that the rate at β = 1 is negligibly small, hence chemical equilibrium is well captured. At β > 1, the inability to sustain much precipitation underlies the flattening region between β = 1 and β = 6, after which the ability to precipitate new terrace ledges and significantly grow the crystal underpins the increasing gradient of the rate vs. β curve. The significant nonlinearity of the rate curve in Fig. 3d, despite using rates derived from linear TST, stems from MASKE’s ability to account for detailed morphology effects. Another strength of the KMC simulations,
MASKE
167
evidenced in Fig. 3, is their ability to predict long timescales and realistically low rates, despite the nano length scale of the system. 3.2 Pressure-solution Creep Figure 4 shows that a compressive stress applied to the nanocrystal, σ z,plat , induces dissolution and thus a progressive reduction of the distance between platens zt , i.e. creep from pressure solution. In all the simulations in Fig. 4, the solution is initially set to equilibrium with the solid, at β = 1. The figure also shows colour plots of the vertical stress per particle, σ z , which is correctly greater than σ z,plat at the points of concentrated contact between crystal and platens. When a small stress is applied, in Fig. 4a, small reorganizations of the crystal surface take place with dissolution of particles near the platens immediately followed by precipitation at a small number of favourable sites. As a result, the top platen moves a bit downward, but the underlying dissolution is very slow and the solution remains near chemical equilibrium at β = 1. When more stress is applied, in Fig. 4b, the timescale of dissolution gets smaller, so creep goes faster. The solution reaches a slightly higher β, between 1 and 2, which is enough to sustain a larger downward displacement of the upper platen but no major crystallization. When an even higher stress is applied, in Fig. 4c, the timescale reduces further and β stabilizes at around 6, which Fig. 3d highlighted as a critical threshold to enable more precipitation and crystallization by forming new terrace ledges. This mechanism leads to visible crystallization in Fig. 4c and to much more dissolution and vertical displacement. 3.3 Crystallization Pressure Figure 5 shows that setting the crystal to a high fixed β = 100 leads to widespread precipitation and crystallization, although still without the ability to create adatoms and therefore grow the crystal indefinitely. The precipitation process involves regions near the crystal-platen interfaces and is able to create sufficient vertical pressure to push upward the top platen, thus capturing qualitatively the effect of crystallization pressure. However, the current version of MASKE is likely to significantly underestimate the effect of crystallization pressure, in that the adoption of net rates and of energy minimization (instead of molecular dynamic) between discrete events of precipitation produce an averaging out of fluctuations that may add pathways for particles to squeeze in at the crystal-platen interfaces and thus further push the top platen upward.
168
E. Masoero
Fig. 4. Dissolution creep induced by a stress applied to the platens: (a) σ z,plat = 10 MPa, (b) σ z,plat = 45 MPa, (c) σ z,plat = 90 MPa.
Fig. 5. Expansion from crystallization under small compression σ z,plat = 10 MPa and β = 100.
4 Conclusion The MASKE simulator is able to describe fundamental processes where chemical reactions, such as dissolution and precipitation, are coupled with mechanical stress. The simulations also predict how the morphology of reacting minerals changes during the processes. In previously published works, MASKE was already used to simulate singlephase and multi-phase processes in concrete, also including some coupling between dissolution and stress along screw dislocations.
MASKE
169
This article has broadened the range of applications of MASKE, presenting a first set of systematic and quantitative results on creep from stress-driven dissolution, and first results on the effect of crystallization pressure. The simulations have focussed on a Ca(OH)2 nanocrystal made of agglomerated particles representing one molecule each, all within a aqueous solution featuring only Ca2+ and OH− ions. A number of extensions and improvements are possible now, such as studying other cement minerals (e.g. creep from solution-pressure is pronounced for gypsum), employing coarser particles each representing millions of molecules to reach larger length and time scales, and allowing for ion speciation in solution (an interface between MASKE and PHREEQC [21] is under implementation). Further extensions of MASKE are being considered too, e.g. to include ion diffusion in solution and the presence of particles representing bacteria whose metabolism also alters the solution chemistry, in particular to model self-healing concrete [22] and other bio-inspired systems for example in ground engineering [23]. Despite these possible improvements and ongoing developments, the results in this article already led to the following findings: • The Kinetic Monte Carlo algorithm in MASKE was able to predict realistic timescales (seconds to hours) and rates, even for the nanocrystal studied here; • Under stress-free conditions, the relationship between dissolution/growth rates and solution saturation index β is highly nonlinear, despite the rates are obtained from linear Transition State Theory. The nonlinearity stems from morphological details that MASKE captures. β = 6 emerges as a threshold level of saturation to overcome for significant crystallization to take place. • The simulations successfully predicted creep from stress-driven dissolution, confirming that both the extent and the rate of creep can be increased by increasing the intensity of the applied stress, here from 10 to 90 MPa. The simulations of creep also confirmed that significant crystallization is triggered only when the applied stress is sufficiently large to pushes the saturation of the solution up to β = 6 threshold. • When tested under a small stress of 10 MPa and high solution saturation β = 100, the crystal displayed some expansion which captures qualitatively the effect of crystallization pressure. However, MASKE would require additional features to describe the phenomenon of crystallization pressure in detail. All in all, this work has shown how MASKE can simulate in detail some of the complex, fundamental processes that underpin the degradation of concrete and of other mineral systems. There is scope for similar simulations to support the development of more reliable constitutive laws for future degradation models of both traditional and new cementitious materials.
References 1. International Energy Agency: Global status report: towards a zero-emission, efficient, and resilient buildings and construction sector (2017) 2. Monteiro, P.J., Miller, S.A., Horvath, A.: Towards sustainable concrete. Nat. Mater. 16(7), 698–699 (2017) 3. Editorial, N.: Concrete needs to lose its colossal carbon footprint. Nature 597, 593–594 (2021)
170
E. Masoero
4. Bernal, S.A., Provis, J.L., Brice, D.G., Kilcullen, A., Duxson, P., van Deventer, J.S.: Accelerated carbonation testing of alkali-activated binders significantly underestimates service life: the role of pore solution chemistry. Cem. Concr. Res. 42(10), 1317–1326 (2012) 5. Pouhet, R., Cyr, M.: Carbonation in the pore solution of metakaolin-based geopolymer. Cem. Concr. Res. 88, 227–235 (2016) 6. Lasaga, A.C.: Kinetic Theory in the Earth Sciences, vol. 811. Princeton University Press, Princeton (1998) 7. Pignatelli, I., Kumar, A., Alizadeh, R., Le Pape, Y., Bauchy, M., Sant, G.: A dissolutionprecipitation mechanism is at the origin of concrete creep in moist environments. J. Chem. Phys. 145(5), 054701 (2016) 8. Bažant, Z.P., Hauggaard, A.B., Baweja, S., Ulm, F.J.: Microprestress-solidification theory for concrete creep. I: aging and drying effects. J. Eng. Mech. 123(11), 1188–1194 (1997) 9. Masoero, E., Di Luzio, G.: Nanoparticle simulations of logarithmic creep and microprestress relaxation in concrete and other disordered solids. Cem. Concr. Res. 137, 106181 (2020) 10. Scherer, G.W.: Stress from crystallization of salt. Cem. Concr. Res. 34(9), 1613–1624 (2004) 11. Ioannidou, K., Labbez, C., Masoero, E.: A review of coarse grained and mesoscale simulations of C–S–H. Cem. Concr. Res. 159, 106857 (2022) 12. Shvab, I., Brochard, L., Manzano, H., Masoero, E.: Precipitation mechanisms of mesoporous nanoparticle aggregates: off-lattice, coarse-grained, kinetic simulations. Cryst. Growth Des. 17(3), 1316–1327 (2017) 13. Coopamootoo, K., Masoero, E.: Simulations of crystal dissolution using interacting particles: prediction of stress evolution and rates at defects and application to tricalcium silicate. J. Phys. Chem. C 124(36), 19603–19615 (2020) 14. Alex, A., Masoero, E.: Autogenous healing in cement: a kinetic Monte Carlo simulation of CaCO3 precipitation. In: Computational Modelling of Concrete and Concrete Structures, pp. 102–106. CRC Press (2022) 15. Stukowski, A.: Visualization and analysis of atomistic simulation data with OVITO–the open visualization tool. Modell. Simul. Mater. Sci. Eng. 18(1), 015012 (2009) 16. Wittmann, F.H.: Estimation of the modulus of elasticity of calcium hydroxide. Cem. Concr. Res. 16(6), 971–972 (1986) 17. Estrela, C., Estrela, C.R.D.A., Guimarães, L.F., Silva, R.S., Pécora, J.D.: Surface tension of calcium hydroxide associated with different substances. J. Appl. Oral Sci. 13, 152–156 (2005) 18. Lothenbach, B., et al.: CEMDATA18: A chemical thermodynamic database for hydrated Portland cements and alkali-activated materials. Cem. Concr. Res. 115, 472–506 (2019) 19. Domínguez, A., Jimenez, R., López-Cornejo, P., Pérez, P., Sánchez, F.: On the calculation of transition state activity coefficient and solvent effects on chemical reactions. Collect. Czech. Chem. Commun. 63(12), 1969–1976 (1998) 20. Bullard, J.W.: A three-dimensional microstructural model of reactions and transport in aqueous mineral systems. Modell. Simul. Mater. Sci. Eng. 15(7), 711 (2007) 21. Parkhurst, D.L., Appelo, C.A.J.: Description of input and examples for PHREEQC version 3—a computer program for speciation, batch-reaction, one-dimensional transport, and inverse geochemical calculations. US Geol. Surv. Techn. Methods 6(A43), 497 (2013) 22. Bagga, M., et al.: Advancements in bacteria based self-healing concrete and the promise of modelling. Constr. Build. Mater. 358, 129412 (2022) 23. Assadi-Langroudi, A., et al.: Recent advances in nature-inspired solutions for ground engineering (NiSE). Int. J. Geosynth. Ground Eng. 8(1), 1–36 (2022)
Temperature-Dependent Behavior of Mature Cement Paste: Creep Testing and Multiscale Modeling Eva Binder1,2,4 , Markus Königsberger1,3 , Rodrigo Díaz Flores1 , Herbert A. Mang1,4 , Christian Hellmich1 , and Bernhard L. A. Pichler1(B) 1 Institute for Mechanics of Materials and Structures, TU Wien (Vienna University of
Technology), Karlsplatz 13, 1040 Vienna, Austria [email protected] 2 Linnaeus University, Universitetsplatsen 1, 35195 Växjö, Sweden 3 ULB (Université Libre de Bruxelles), Avenue Franklin Roosevelt 50, 1050 Brussels, Belgium 4 College of Civil Engineering, Tongji University, 1239 Siping Road, Shanghai, China
Abstract. The elastic compliance of cementitious materials increases slightly, and their time-dependent “creep” compliance increases significantly with increasing temperature. The present contribution provides quantitative insight into this topic. Thereby, the focus rests on mature cement paste made from Ordinary Portland cement and distilled water. Macroscopic creep experiments were performed, in order to quantify both elastic and creep moduli in the range of temperatures from 20 °C to 45 °C. The experimental results regarding temperaturedependent values of the modulus of elasticity are validated herein using original results from ultrasonics testing. Finally, a multiscale model was used to establish a link to well-known stiffness constants of unhydrated cement clinker, as well as to temperature-dependent elastic and creep stiffness properties of nanoscopic hydrate-gel needles. Keywords: Creep testing · Multiscale modelling · Creep modulus · Elastic modulus · Thermal activation · Activation energy
1 Introduction Macroscopic creep of cementitious materials results from the viscoelastic behaviour of microscopic calcium-silicate-hydrates (C-S-H), the main hydration products resulting from the chemical reaction between ordinary Portland cement and water. It is challenging to establish models that allow for upscaling the creep properties of C-S-H to the macroscopic viscoelastic behaviour of concrete, because several transitions between different scales of observations are necessary. This is explained as follows. Concrete is a matrix-inclusion composite. It consists of aggregates (with a characteristic size from a few millimetres to a few centimetres) embedded in a continuous mortar matrix. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 A. J˛edrzejewska et al. (Eds.): SynerCrete 2023, RILEM Bookseries 43, pp. 171–180, 2023. https://doi.org/10.1007/978-3-031-33211-1_16
172
E. Binder et al.
Mortar is a matrix-inclusion composite. It consists of sand grains (with a characteristic size from a several dozens of micrometres to a few millimetres) embedded in a continuous cement paste matrix. Cement paste is a matrix-inclusion composite. It consists of unhydrated (= unreacted, and therefore, “residual”) cement clinker grains (with a characteristic size from a few single microns to a several dozens of micrometres) embedded in a continuous hydrate foam matrix [1]. The hydrate foam is a microheterogeneous material with a highly-disordered arrangement of its constituents. It consists of isotropically oriented hydrate gel needles (with a characteristic diameter from several single nanometres to several dozens of nanometres and a characteristic length from a few dozens of nanometres to a few single micrometres) in direct interaction with capillary pores of similar size [1]. The hydrate gel is a microheterogeneous material itself. It contains, e.g. solid C-S-H and gel pores. For the purpose of the present study, however, it is not necessary to resolve the microstructure of the hydrate gel. When it comes to multiscale modelling of mechanical properties of cementitious materials, carried out under the ambition to assign material constants (rather than material parameters), to the microstructural constituents (aggregates, sand, cement clinker, hydration products), their microstructures must be resolved at least down to the hydrate gel needles and the capillary pores. As for concrete, this requires four scale transitions: (i) from the scale of the hydrate gel needles and the capillary pores to the scale of the hydrate foam, (ii) from the scale of the hydrate foam and the residual clinker grains to the scale of cement paste, (iii) from the scale of cement paste and sand grains to the scale of mortar, and (iv) from the scale of mortar and aggregates to the scale of concrete. Because of these four scale transitions, methods of continuum micromechanics are particularly well suited for multiscale modelling.
2 Multiscale Modelling of Cementitious Materials Using Methods from Continuum Micromechanics 2.1 Basics of Mean-Field Methods of Continuum Micromechanics Continuum micromechanics models account for six key features of microheterogeneous materials, rather than for every detail of their microstructures [2]: (i) the hierarchical organization of the material as described for concrete in the Introduction, (ii) material constants assigned to the material constituents (= “material phases”), (iii) phase volume fractions quantifying how much volume of the microstructure is occupied by the individual material phases; note that the volume fractions of cement clinker, water, and hydration products change at early ages, because of chemical reactions; (iv) characteristic ellipsoidal shapes assigned to the material phases, ranging from plates, via oblate spheroids, to spheres, and via prolate spheroids, to cylinders, (v) the characteristic spatial orientation of non-spherical material phases, as well as (vi) the specific mode of interaction between the material phases which are present at the same scale of observation. As for (iv), the characteristic interaction between the matrix material and the inclusions of a matrix-inclusion composite is accounted for by the Mori-Tanaka scheme, while
Temperature-Dependent Behavior of Mature Cement Paste
173
the characteristic interaction between the constituents which are part of a (matrix-free) highly disordered microstructure is accounted for by the self-consistent scheme. Homogenization schemes of continuum micromechanics exploit analytical solutions of so-called Eshelby problems [3]. An Eshelby problem consist on an infinite three-dimensional matrix hosting one ellipsoidal inclusion. The famous Eshelby solution expresses the spatially uniform strain inside the ellipsoidal inclusion as a function of the remote strain imposed, as loading, on the infinite matrix surrounding the ellipsoidal inclusion. This analytical expression linking the remote strain to the inclusion strain involves the difference between the stiffness of the infinite matrix and the stiffness of the ellipsoidal inclusion, as well as a Hill tensor accounting for the ellipsoidal shape and the spatial orientation of the ellipsoidal inclusion. One auxiliary Eshelby problem is introduced for every material phase of the composite of interest [2]. The properties of the auxiliary inclusion are linked to the properties of the corresponding material phase, the stiffness of the auxiliary infinite matrix is linked to a characteristic stiffness of the actual composite of interest, and the auxiliary remote strain of the infinite matrix is linked to the strain experienced by the composite of interest, as explained next. The elastic stiffness, the ellipsoidal shape, and the orientation of every auxiliary inclusion are set equal to the properties of the corresponding material phase [2]. Given that hydrate gel needles are isotropically oriented in space, they are represented by infinitely many Eshelby problems [4]. The stiffness of the auxiliary infinite matrix (surrounding an inclusion representing one specific material phase) is set equal to a characteristic stiffness of the composite which contains the specific material phase [2]: (i) If the composite of interest is a matrixinclusion composite, the stiffness of the auxiliary infinite matrix of the Eshelby problem is set equal to the stiffness of the matrix of the composite. This leads to so-called MoriTanaka schemes. (ii) If the composite of interest is a composite with a highly disordered arrangement of the material phases, the stiffness of the auxiliary infinite matrix of the Eshelby problem is set equal to the homogenized stiffness of the composite. This leads to so-called self-consistent schemes. The auxiliary remote strain of the infinite matrix (surrounding an inclusion representing one specific material phase) is linked to the strain experienced by the composite which contains the specific material phase [2]. In this context, the spatially uniform strain state inside the auxiliary inclusion is used as an estimate of the volume-averaged strain of the corresponding material phase of the composite of interest. Inserting the estimated volume-averaged phase strains into the strain average rule allows for establishing a transfer tensor linking (i) the strain experienced by the composite of interest to (ii) the auxiliary remote strain of the corresponding Eshelby problems. 2.2 State-of-the-Art Multiscale Model for Strength and Creep Upscaling and Mechanical Properties of Hydrate Gel Needles The state-of-the-art continuum-micromechanics-representation of cement paste, allowing for upscaling of both the strength and creep, was introduced in [1]. The breakthrough at that time was the introduction of the hydrate foam as an essential intermediate scale (rather than introducing clinker grains, hydration products, and capillary pores at the
174
E. Binder et al.
same scale of observation), see Fig. 1. The two-dimensional sketches show qualitative properties of three-dimensional representative volume elements.
Fig. 1. Micromechanical representation of cement paste; after [5].
Because hydrate gel needles are very small, it is impossible to cut specimens out of them and to perform direct material testing on such specimens. As a remedy, the stiffness, strength, and creep constants of hydrate gel needles were identified by means of downscaling properties measured on cement pastes. The isotropic elastic stiffness constants of the hydrate gel needles were identified as follows. Poisson’s ratio of the hydrate gel was set equal to 0.24 [6]. The modulus of elasticity was identified to be equal to 29.2 GPa by means of downscaling “final stiffness” properties measured on very well hardened cement pastes [1]. These two values, allows for computing other stiffness constants of the isotropic hydrate gel, e.g. the bulk modulus amounts to 18.7 GPa, and the shear modulus to 11.8 GPa. The strength properties of the hydrate gel needles were identified as follows. In the interest of keeping the number of introduced material constants as small as possible, the first strength criterion introduced at the microscopic scale of the hydrate gel needles was a pressure-insensitive shear failure criterion [4]. The corresponding von Mises type deviatoric strength value was identified to be equal to 69.9 MPa, by means of downscaling the results of an early-age strength testing campaign carried out at TU Wien [7]. The corresponding model-predicted uniaxial compressive strength of the hydrate gel needles amounts to 123.5 MPa [8]. Very remarkably, the same uniaxial compressive strength is predicted by a Mohr-Coulomb-type pressure-sensitive shear failure criterion with a cohesion amounting to 50 MPa and an angle of internal friction amounting to 12° [8]. These two Mohr-Coulomb properties were identified based on the re-analysis of nanoindentation experiments into low-density C-S-H [9]. The cohesion of 50 MPa and the internal friction angle of 12° were finally translated into Drucker-Prager strength quantities in [10]. This publication also includes the extension of the multiscale model from the scale of cement paste to those of mortars and concretes. This extension explicitly accounts for stress concentration in interfacial transition zones surrounding the sand grains and the aggregates [11, 12]. The creep properties of the hydrate gel needles were identified as follows. In the interest of keeping the number of introduced material constants as small as possible, the first creep compliance tensor introduced at the microscopic scale of the hydrate gel needles was isochoric [13]. This describes that the hydrate gel shows viscoelastic behaviour, when subjected to any type of shear loading, while the material exhibits a purely elastic and therefore time-independent behaviour when subjected to isotropic (=
Temperature-Dependent Behavior of Mature Cement Paste
175
“hydrostatic”) loading. The newly introduced tensorial creep compliance tensor function contains two material constants: a creep shear modulus and a power-law exponent. The 20.9 GPa and the power-law exponent was identified to be equal to 0.25 by means of downscaling the results of several hundreds of three-minute creep tests of cement pastes with initial water-to-cement mass ratios of 0.42, 0.45, and 0.50, respectively, performed at 20 °C during the first week after production, see [14] for the creep tests and [13] for the identification of the material properties. The identified creep properties were used to predict the creep behaviour of cement paste with an initial water-to-cement mass ratio of 0.50, subjected at an age of 30 years to a 28-day lasting creep test, see [15] for the creep tests and [13] for model validation. The multiscale creep model was extended from the scale of cement paste to those of mortars and concretes in [16]. This extension explicitly accounts for a water migration phenomenon. During the production of mortars and concretes, made from oven-dried sand and aggregates, part of the mixing water is taken up by the open porosity of the sand grains and aggregates. Therefore, the actual water-to-cement mass ratio of the cement paste (serving as the binder between the sand grains and the aggregates) is smaller than the ratio of the masses of the used water and cement. The water, which is initially moving into the open porosity of the sand grains and the aggregates is later sucked back into the cement paste matrix. This is driven by the underpressure of water induced by self-desiccation.
3 Temperature-Dependent Stiffness of the Hydrate Gel Needles The reported isotropic viscoelastic stiffness constants, the bulk modulus of 18.7 GPa, the shear modulus of 11.8 GPa, the creep shear modulus of 20.9 GPa, and the powerlaw exponent of 0.25 are referred to as “constants”, because they are independent of the initial water-to-cement mass ratio and independent of the degree of hydration. Still, these values were identified for (almost) water saturated materials conditioned to 20 °C. The stiffness constants are expected to change with decreasing water saturation level, and with changing temperature. While the former dependency provides motivation for future research, the latter dependency is in the focus of the present contribution. The present section reports about the temperature-dependence of the viscoelastic stiffness properties of the hydrate gel [5]. As an original contribution, the experimental results regarding temperature-dependent macroscopic values of the modulus of elasticity reported in [5], see Table 1 in Subsect. 3.1, are validated herein using newly presented results from ultrasonics pulse transmission testing, see Table 2 in Subsect. 3.2. 3.1 Macroscopic Experiments The test protocol of [14] was used in [5]. Five cylinders of cement paste were produced with an initial water-to-cement mass ratio of 0.42. The diameters of the samples amounted to 7 cm. The lengths were equal to 30 cm. After demolding, the specimens were cured for three months, submersed in lime-saturated water conditioned to 20 °C. Then, they were taken out of the water, protected against drying, and installed (oneby-one) in a temperature chamber which is part of a uniaxial electromechanical testing machine. The whole setup was conditioned to either 20 °C or 30 °C or 45 °C. More
176
E. Binder et al.
than enough time was waited for the system to reach a stationary state, with isothermal conditions inside the temperature chamber. This way, all five specimens were subjected, one after the other, to three-minute creep tests at the three different temperatures. Thus, a total of 15 creep tests were performed [5]. In between two tests carried out on the same specimen, more than enough time was waited for the specimens to show complete creep recovery. Therefore, every test can be evaluated independently. In addition, the loading was limited with some 16% of the strength of the specimens reached at the time of testing. Therefore, creep testing was carried out in the linear creep regime, and it is very unlikely that the specimens were damaged [5]. During creep testing, the deformations of the specimens were measured between two aluminium rings clamped to the specimens, see Fig. 2. Strain histories were quantified by dividing the average change of length, recorded by five displacement sensors (LVTDs) evenly positioned around the specimen, by the measurement length of 16.4 cm. Stress histories were quantified by dividing the force readings of the testing machine by the cross-sectional area of the specimens.
Fig. 2. Setup for creep testing; after [5].
Test evaluation is carried out in the framework of the linear theory of viscoelasticity, using Boltzmann’s superposition principle. The bilinear stress histories (loading ramp and load plateau) entered the evaluation of the tests as input. A macroscopic uniaxial power-law creep function was used. It contains three material properties: the modulus of elasticity (referred to in the following as the “elastic modulus”), a creep modulus, and a macroscopic power-law exponent. The last of these three quantities is known to amount to 0.25. The elastic modulus and the creep modulus were identified such as to re-produce the measured strain evolutions in a best possible fashion. Thus, the identification results are 15 pairs of elastic and creep moduli, referring to the five specimens and to the three different testing temperatures [5]. The elastic moduli E were found to decrease slightly with increasing temperature. The creep moduli Ec were found to decrease significantly with increasing temperature, see Table 1 for average values taken from [5].
Temperature-Dependent Behavior of Mature Cement Paste
177
Table 1. Stiffness values obtained from evaluation of three-minutes quasi-static creep tests by means of the linear theory of viscoelasticity and Boltzmann’s superposition principle [5]. temperature
20 ◦ C E [GPa] 21.3
30 ◦ C Ec [GPa] 41.6
E [GPa] 21.0
45 ◦ C Ec [GPa] 34.5
E [GPa] 20.5
Ec [GPa] 22.9
3.2 Validation of the Temperature-Dependent Values of the Modulus of Elasticity of Table 1 by Means of Ultrasonic Pulse Velocity Tests In order to validate the temperature-dependent values of the modulus of elasticity listed in Table 1, ultrasonic pulse velocity measurements were carried out on two specimens which were previously subjected to three-minutes quasi-static testing, using a central frequency of 250 kHz. Employing the theory of elastic wave propagation through isotropic media, measured values of longitudinal and shear wave velocities were translated into values of the modulus of elasticity and Poisson’s ratio, see Table 2. The ultrasonics-derived values of the modulus of elasticity values (Table 2) agree, both in absolute and relative terms, well with the quasi-statically determined values of the modulus of elasticity (Table 1). This underlines the credibility of the data listed in Table 1, and it provides the motivation to continue with the analysis of temperature-dependent stiffness values of mature cement paste by means of multiscale modelling. Table 2. Stiffness values obtained from evaluation of ultrasonic pulse transmission tests by means of the theory of elastic wave propagation through isotropic media. temperature
20 ◦ C E [GPa] 21.2
30 ◦ C υ [–] 0.298
E [GPa] 20.9
45 ◦ C υ [–] 0.298
E [GPa] 20.3
υ [–] 0.298
3.3 Multiscale Modelling The multiscale model of [13] was extended in [5], in order to account for temperaturedependent creep properties. The model subdivides cement paste into three constituents: cement clinker, hydrate gel needles, and water-filled capillary porosity. Cement clinker does not creep. It is a linear elastic material, and its isotropic stiffness properties (bulk modulus = 116.7 GPa, shear modulus = 53.8 GPa) were treated as being temperature independent. Water is a liquid. It does not exhibit a solid stiffness. Assuming the capillary porosity to be well connected, drained conditions are assumed. This implies that no pore pressure builds up during mechanical loading of cement paste. Thus, the macroscopically observed changes of the elastic modulus and of the creep modulus, must be triggered by changes of the viscoelastic stiffness properties of the hydrate gel. They are quantitatively known at 20 °C, see the first paragraph of Sect. 3.
178
E. Binder et al.
The degree of hydration of the samples was identified such that the multiscale model reproduced the average macroscopic elastic modulus determined experimentally at 20 °C [5]. This delivered a degree of hydration of 80%. Inserting this value, together with the initial water-to-cement mass ratio of 0.42, into Powers’ hydration model allowed for quantifying the volume fractions of the three constituents. 91.4% of the volume of the tested cement paste consisted of hydrate foam, the remaining 8.6% were occupied by residual cement clinker. 81.2% of the volume of the hydrate foam consisted of hydrate gel needles, the remaining 18.8% were occupied by capillary pores. Results from molecular studies were used to model establish a predictive model accounting from the temperature-dependent properties of the hydrate gel. Poisson’s ratio of cementitious hydrates is known to be virtually temperature-independent [17]. The decrease of the elastic modulus of the hydrate gel with increasing temperature was introduced as a linear function. Its value at 20 °C was known from previous studies. Its slope was set equal to that found in molecular simulations of C-S-H [18]. Using the extended multiscale model for predicting the dependency of the macroscopic modulus of elasticity on temperature delivered a very good agreement with results from the experimental study described before [5]. As for the temperature dependence of creep, the power-law exponent of the hydrate gel was set equal to 0.25, independent of temperature. As for the creep shear modulus of the hydrate gel, an Arrhenius-type temperature-activation law was used [5]. The activation energy was set equal to that of the viscosity of bulk water. This was motivated as follows. Nanoscopic water migration processes trigger macroscopic creep of cementitious materials. Water at these small scales is confined, i.e. it is structures rather than disordered. The viscosity of structured water is different from the viscosity of bulk (disordered) water. However, the activation energy describing the temperature dependence of the viscosity of bulk and structured water was shown by means of molecular simulations to be the same [19]. Using this property as input, the predictive multiscale model was complete. The creep shear modulus of the hydrates at 20 °C was known from previous studies. Its change with temperature was modelled using the described Arrhenius law with the activation energy amounting to 17.57 kJ/mol. Using the extended multiscale model for predicting the dependency of the macroscopic creep modulus on temperature delivered a very good agreement with tests results of Table 1, see [5].
4 Conclusions The microscopic hydrate gel is an important part of the microstructure of cementitious materials. In the past 10 years, several studies provided access to its stiffness and strength properties. The most recent step in a longer development concerned the influence of the temperature on the viscoelastic behaviour of the hydrate gel [5]. Experiments and multiscale modelling concerned practically relevant temperature range from 20 °C to 45 °C. Poisson’s ratio of the hydrate gel is virtually constant. • Its modulus of elasticity decreases linearly with increasing temperature. This decrease amounts to some 1.57% provided that the temperature is increased by 10 °C. • The power-law creep exponent of the hydrate gel is virtually constant.
Temperature-Dependent Behavior of Mature Cement Paste
179
• Its creep shear modulus decreases non-linearly with increasing temperature, as described by an Arrhenius law. The activation energy is equal to 17.57 kJ/mol. The original contributions of the present paper, see Subsect. 3.2 and Table 2, contribute to the scientific debate regarding similarities and differences of moduli of elasticity, Young’s moduli, loading moduli, unloading-moduli, (quasi-)static moduli, and dynamic moduli. In this context, the following conclusions are drawn: • The evaluation of quasi-static tests within the framework of linear viscoelasticity, using Boltzmann’s superposition principle, yielded temperature-dependent elastic stiffness moduli of mature cement paste, which agree well with results obtained from the evaluation of ultrasonic pulse velocity testing based on the theory of elastic wave propagation though isotropic media, compare Tables 1 and 2. It is concluded that the E-values listed in these Tables are independent of the loading rate. • Given that the E-values listed in Tables 1 and 2 also refer to reversible deformation processes, it is concluded that the E-values comply with the fundamental thermodynamics-based definition of elastic stiffness moduli. This provided the motivation to address them as “elastic moduli” or “moduli of elasticity”, respectively.
References 1. Pichler, B., Hellmich, C.: Upscaling quasi-brittle strength of cement paste and mortar: a multi-scale engineering mechanics model. Cem. Concr. Res. 41(5), 467–476 (2011) 2. Zaoui, A.: Continuum micromechanics: survey. J. Eng. Mech. 128(8), 808–816 (2002) 3. Eshelby, J.D.: The determination of the elastic field of an ellipsoidal inclusion, and related problems. Proc. R. Soc. Lond. Ser. A Math. Phys. Sci. 241(1226), 376–396 (1957) 4. Pichler, B., Hellmich, C., Eberhardsteiner, J.: Spherical and acicular representation of hydrates in a micromechanical model for cement paste: prediction of early-age elasticity and strength. Acta Mech. 203(3), 137–162 (2009) 5. Binder, E., Königsberger, M., Flores, R.D., Mang, H.A., Hellmich, C., Pichler, B.L.: Thermally activated viscoelasticity of cement paste: minute-long creep tests and micromechanical link to molecular properties. Cem. Concr. Res. 163, 107014 (2023) 6. Constantinides, G., Ulm, F.J.: The effect of two types of CSH on the elasticity of cementbased materials: results from nanoindentation and micromechanical modeling. Cem. Concr. Res. 34(1), 67–80 (2004) 7. Pichler, B., et al.: Effect of gel–space ratio and microstructure on strength of hydrating cementitious materials: an engineering micromechanics approach. Cem. Concr. Res. 45, 55–68 (2013) 8. Pichler, B., et al.: The Counteracting effects of capillary porosity and of unhydrated clinker grains on the macroscopic strength of hydrating cement paste–a multiscale model. In: Mechanics and Physics of Creep, Shrinkage, and Durability of Concrete: A Tribute to Zdeˇnk P. Bažant, pp. 40–47 (2013) 9. Sarris, E., Constantinides, G.: Finite element modeling of nanoindentation on C–S–H: effect of pile-up and contact friction. Cem. Concr. Compos. 36, 78–84 (2013) 10. Königsberger, M., Hlobil, M., Delsaute, B., Staquet, S., Hellmich, C., Pichler, B.: Hydrate failure in ITZ governs concrete strength: a micro-to-macro validated engineering mechanics model. Cem. Concr. Res. 103, 77–94 (2018)
180
E. Binder et al.
11. Königsberger, M., Pichler, B., Hellmich, C.: Micromechanics of ITZ–aggregate interaction in concrete Part I: stress concentration. J. Am. Ceram. Soc. 97(2), 535–542 (2014) 12. Königsberger, M., Pichler, B., Hellmich, C.: Micromechanics of ITZ-aggregate interaction in concrete Part II: strength upscaling. J. Am. Ceram. Soc. 97(2), 543–551 (2014) 13. Königsberger, M., Irfan-ul-Hassan, M., Pichler, B., Hellmich, C.: Downscaling based identification of nonaging power-law creep of cement hydrates. J. Eng. Mech. 142(12), 04016106 (2016) 14. Irfan-ul-Hassan, M., Pichler, B., Reihsner, R., Hellmich, C.: Elastic and creep properties of young cement paste, as determined from hourly repeated minute-long quasi-static tests. Cem. Concr. Res. 82, 36–49 (2016) 15. Tamtsia, B.T., Beaudoin, J.J.: Basic creep of hardened cement paste a re-examination of the role of water. Cem. Concr. Res. 30(9), 1465–1475 (2000) 16. Irfan-ul-Hassan, M., Königsberger, M., Reihsner, R., Hellmich, C., Pichler, B.: How wateraggregate interactions affect concrete creep: multiscale analysis. J. Nanomech. Micromech. 7(4), 04017019 (2017) 17. Wang, X.F., et al.: Computational study of the nanoscale mechanical properties of CSH composites under different temperatures. Comput. Mater. Sci. 146, 42–53 (2018) 18. Xin, H., Lin, W., Fu, J., Li, W., Wang, Z.: Temperature effects on tensile and compressive mechanical behaviors of CSH structure via atomic simulation. J. Nanomater. (2017) 19. Zaragoza, A., et al.: Molecular dynamics study of nanoconfined TIP4P/2005 water: how confinement and temperature affect diffusion and viscosity. Phys. Chem. Chem. Phys. 21(25), 13653–13667 (2019)
Development of an Experimental-Numerical Approach to Model Cement Paste Microstructure Using Quantitative Phase Assemblage from XRD and Thermodynamic Modeling Mohammed Krameche(B)
, William Wilson , and Arezki Tagnit-Hamou
Université de Sherbrooke, Sherbrooke, QC J1K 2R1, Canada [email protected]
Abstract. Multiscale modeling is a powerful tool to understand and predict the overall mechanical and durability performance of concrete. However, considering properties at the microscale is challenging because of the highly heterogenous character of the cement paste microstructure, which includes porosity and several intermixed phases. The current work suggests a new experimental-modeling approach to model the 3D microstructure of cement paste, destined to multiscale modeling applications. Quantitative phase assemblage from X-ray diffraction (QXRD) conduced at different hydration ages combined with Gibbs Energy Minimization (GEMS) thermodynamic modeling serve to define microstructure phases to be modeled by μic model. Phase map from Electron Dispersive Energy Spectroscopy (EDS) mapping clustering is used to distribute microstructure phases into particles, hydrate layers or porosity. The μic microstructure model is then calibrated following an iterative process and using quantitative phase assemblages from thermodynamic modeling and QXRD analyses. The approach is applied to an Ordinary Portland Cement (OPC) system, for which the μic model is validated at the long-term hydration age. The results show that the calibrated μic model agrees with thermodynamics modeling results. This allows a more realistic representation of the cement paste microstructure. Therefore, a more accurate prediction of the mechanical behavior of concrete when the model is used as input for multiscale modeling of concrete. Keywords: Cement · Cement microstructure · μic · Multiscale modeling · Ordinary Portland cement (OPC) · Thermodynamics modeling · GEMS · X-ray diffraction
1 Introduction Scientific research is strongly engaged in the upscaling from microstructure scale properties to concrete properties. Several methodologies and approaches are being proposed to predict concrete long-term properties from lower levels features. Some approaches focus © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 A. J˛edrzejewska et al. (Eds.): SynerCrete 2023, RILEM Bookseries 43, pp. 181–190, 2023. https://doi.org/10.1007/978-3-031-33211-1_17
182
M. Krameche et al.
on multiscale modeling based on the atomic scale, namely by building up from C-S-H atomistic scale simulations to create the 3D microstructure [1, 2]. In other approaches, the microstructure of the cement paste is the starting level of multiscale modeling [3]. In these approaches, the micro-mechanical properties of microstructure phases should be quantified experimentally [3]. The properties are coupled to the 3D cement paste microstructure that should be modeled. Several models can be used to model the microstructure through hydration simulations, among them HYMOSTRUC3D [4] CEMHYD3D [5] and μic [6]. In the literature, few works have considered the calibration of microstructure models based on quantitative experimental analyses. In these works, the calibration is only limited to the major phases of the microstructure and ignores some anhydrous phases and hydrates. The purpose of the study is to improve the representativeness of cement paste microstructure modelling of OPC (ordinary Portland Cement) for upscaling applications. We addressed this by defining an experimental-numerical approach, in which all microstructure phases disclosed and quantified by quantitative X-ray diffraction (QXRD) and GEMS (Gibbs Energy Minimization) modeling are considered and quantitatively calibrated in the μic microstructure model.
2 Materials and Methods The study was conducted using an OPC system with w/b (water to binder) ratio of 0.4. An OPC type GU (general use) commercially available in Canada was selected. The chemical and mineralogical compositions are detailed in Table 1 and Table 2 respectively. The PSD (Particle Size Distribution) of cement powder was measured using LDS (Laser Diffraction Spectrometry). The characteristic sizes d90, d50 and d10 were found to be 52.1 μm, 17.3 μm and 2.5 μm respectively. Table 1. Chemical composition of cement by XRF. Oxides
CaO
SiO2
Al2 O3
Fe2 O3
SO3
K2 O
Na2 O3
MgO
(LOI)
Wt %
61.85
20.7
4.48
2.54
3.65
0.18
0.18
2.1
2.75
XRD-Rietveld on fresh slices and external standard were used to determine the phase assemblage evolution in time. Cement paste was prepared by mixing OPC and water for 2 min using a high shear mixer, then casted in sealed tubes. 24 h later, the samples were placed in sealed containers with minimal water contents to achieve water-curing with minimal leaching at a temperature of 20 °C. At the end of the curing period, the samples were cut into 2 mm thick slices and analyzed at 1, 3, 5, 7, 14, 30, and 120 days. A PANalytical X’pert Pro MRD diffractometer was used to acquire the XRD patterns at 50 mA and 40 kV, using Soller slits of 0.04 rad with incident divergence and anti-scatter slits of 0.5° on the incident beam. Scans were collected in the 5–70 2θ range with a step size of 0.0267°, within an acquisition time of 15 min. For thermodynamic modeling, the GEMS (Gibbs Free Energy Minimization Software) with the Cemdata18 database was used [7, 8]. The modeling used as input the
Development of an Experimental-Numerical Approach
183
Table 2. Mineralogical composition of cement by XRD. Phase
C3 S
C2 S
C3 A
C4 AF
Gypsum
Anhydrite
Bassanite
CaCO3
CaO
Wt %
56.9
20.0
6.4
7.0
3.4
0.2
0.6
4.4
0.6
mineralogical composition of cement shown in Table 2 and the evolution of hydration degrees of clinker phases (alite, belite, aluminate, ferrite and calcite) from QXRD phase assemblage. The 3D microstructure of the OPC paste was modeled with the μic model. This choice was motivated by the vector modeling approach offering unlimited resolution [6], as well as the compatibility and flexibility when coupling the microstructure model to micro-mechanical models for upscaling applications.
3 Methodology The list of microstructure phases to consider in the μic microstructure model was first defined according to XRD results performed at 7 hydration ages. For more accuracy, the list was updated latter to include hydrates expected to form in the GEMS modeling. At this level, phase densities were selected from Cemdata18 databases [8]. It is worth mentioning that density values do not change during hydration simulations in μic. Phase distribution in the microstructure model was chosen to simulate the phase distribution observed with phase maps, obtained by clustering chemical elemental maps from quantitative EDS mapping analyses [9, 10]. Cement particles were modeled as spheres containing the homogenized amount of alite, belite, aluminate, ferrite, calcium sulfates, and minor phases. While calcite was modeled as independent spheres containing only the relevant phase. As for hydrates, the cement particles are modeled to be covered by a layer of C-S-H followed by a second layer containing a mixture of C-S-H and AFt phases. While the calcite particles are covered by a deposit of C-S-H to simulate the filler effect. The porosity in the microstructure is filled by other spheres, representing the phases portlandite, MC (mono-carbo-aluminate), HC (hemi-carbo-aluminate), HT (hydrotalcite), HG (hydrogarnet), or a mixture of C-S-H and AFt. The simulation volume was set to 100 × 100 × 100 μm as a comprise between cement paste representativeness and calculation time. In this micro-volume, cement particles were randomly distributed according to the corrected PSD input. The latter was calculated from the experimentally measured PSD, by considering the probability of the modeled cement volume to contain particles of each size range. The resulting corrected PSD can be characterized by d90 = 22.9 μm, d50 = 8.7 μm and d10 = 1.5 μm. Hydration reactions of alite and belite phases were adopted from literature [11, 12]. While hydration reactions of aluminate, ferrite and calcite were developed by considering reactants and hydrates contents in the long term as provided by GEMS. Both reactions were controlled in μic by nucleation and growth kinetics followed by diffusion kinetics. In μic model, nucleation and growth kinetics are modeled using modified Avrami equation, while diffusion stage is modeled by an equation linking the reaction rate to the thickness of hydrates layers and their diffusivity coefficients [6, 11]. These equations
184
M. Krameche et al.
include the coefficients k (growth rate constant) and kdiff (diffusivities multiplicator) that represent the kinetics parameters to be calibrated. More details can be found in the research done by Shashank.B [6, 11]. Kinetics parameters were calibrated through an iterative process to adjust μic phase assemblage between 1 to 30 days to the experimental phase assemblage determined by QXRD for anhydrous materials and the GEMS phase assemblages for hydration products. QXRD and GEMS phase assemblages at 120 days were used for μic model validation.
4 Results and Discussion 4.1 Experimental and GEMS Phase Assemblage The experimental phase assemblage as determined by QXRD (Fig. 1) shows the diminution in time of the clinker phases volumes (alite, belite, aluminates, ferrite and calcite) and the volumes increase of the crystalline hydrates and amorphous content. The latter represents mainly C-S-H mixed with other possible amorphous phases. It was observed that MC and HC phases are precipitated in the sample at an early age. It is also noted that all hydration products remained stable during hydration as their volumes did not decrease over time. Namely ettringite which was stable in time. This is due to the calcite content of 4.4% by mass in the cement composition. The precipitation of HC and MC is promoted by the increasing amount of carbon in the pore solution, leading to decreasing aluminum concentration. Thus, more sulfate ions are available to form and stabilize ettringite [13, 14]. Thermodynamic calculations have been performed by assuming the hydration degrees of clinker phases provided by QXRD at different hydration ages. Kaolinite, gibbsite, thaumasite and quartz were suppressed in the equilibrium calculations due to their slow hydration kinetics in laboratory conditions. As shown in the thermodynamic phase assemblage in Fig. 2 HG(hydrogarnet) was predicted to form due to iron originally present in the ferrite phase. GEMS phase assemblage showed the presence of minor amounts of brucite phase at early age, because of periclase (MgO) hydration. With the hydration progress, the quantity of brucite was reduced to zero, giving the necessary Mg amount for the formation of the HT (OH-hydrotalcite) phase. Lower amounts of C-S-H were predicted to form compared to the amorphous content in QXRD phase assemblage. This can be explained mainly by the contribution of possible amorphous phases (e.g., MC, HT, …) in the amorphous content from QXRD. Moreover, a considerable gap can be noticed between the C-S-H composition in the GEMS model (atomic ratio: (Ca/(Al + Si) ≈ 1.30) and that quantified by QEDS in the sample microstructure (atomic ratio: (Ca/(Al + Si) ≈ 1.75). 4.2 Calibrated µic Microstructure Model The calibration process described in the methodology was performed on the kinetics of hydration reactions. Subsequently, the model was validated in the long term by comparing μic-predicted phase assemblage to QXRD and GEMS results at 120 days. As shown in Fig. 3 (a, b, c, d and e), the evolution of the clinker phases consumption profiles predicted by μic are in good agreements with the corresponding GEMS profiles
Development of an Experimental-Numerical Approach
185
Fig. 1. Phase assemblage for the OPC paste measured by QXRD at 1, 3, 5, 7, 14, 30 and 120 days.
Fig. 2. Phase assemblage for the OPC paste calculated by GEMS.
(imposed by QXRD), which means that kinetic parameters were properly calibrated. However, although the hydrate quantity profiles predicted by μic followed the same trends as the GEMS results, the amount of C-S-H at 30 days was overestimated by a factor of 1.3. This could be explained by the hypothesis and limits of each model. In terms of μic modeling, simplified chemical equations governing the formation of C-S-H were used from literature. These equations may not be stoichiometrically balanced due to the variations in the products formed, the composition and density of C-S-H. The latter was modeled in μic as a phase having a constant density which is contradictory
186
M. Krameche et al.
to findings on C-S-H composition, densification and ions adsorption [12, 15]. C-S-H density is controlled by the water content in the molecule, which is strongly affected by the curing relative humidity and the saturation state of the sample. In a fully saturated cement paste with w/b ratio equal to 0.4, C-S-H has four molecules of water, and the density is near the values 1900–2100 kg/m3 [12]. Values higher than 2500 kg/m3 can be reached under dried conditions [12]. Moreover, cement particles were modeled as spheres, which underestimates the contact surface between particles. This can lead to overestimation of the free surface free area as calculated by μic, which should affect proportionally the hydrates nucleation and growth. To address this problem and match μic quantities of C-S-H to GEMS phase assemblages, a corrective coefficient was adopted. This coefficient is added to the volumetric alite hydration reaction Eq. (1) in μic as follows: 1.0 C3 S + 1.31 H2 O → α(1.0 C.S.H inner + 0.57 C.S.H Outer ) + 0.59 CH
(1)
With a coefficient α of 0.65, the resulting C-S-H quantities were found to be in good agreement with C-S-H values provided by GEMS, as shown in Fig. 3h. The μic model was validated by predicting the phase assemblage of the same system at 120 days. A comparison between the 2 predicted μic phase assemblage and that obtained by GEMS is presented in Fig. 4. A good agreement can be observed between the two assemblages, with total solids volumes of 58.1 cm3 and 56.6 cm3 predicted by μic and GEMS respectively. 4.3 µic Calibrated 3D Microstructure and Upscaling Perspectives The μic modelling allows voxelization of the modeled micro volume when exporting phase locations. The voxel size was set to 250 nm as a compromise between calculation time and resolution. Figure 5 shows the heterogenous character of cement paste microstructure and the complex distribution of microstructure phases and porosity in the micro volume. The 3D microstructure at 120 days shows a phase distribution similar to the phase map results found in literature [6, 16], as described in the methodology. To prepare the future upscaling applications, micromechanical properties of microstructure phases were collected from NI-QEDS results [17], then homogenized for μic mixed phases (e.g., C-S-H and ettringite) according to the volume fraction of each sub-phase. Chemo-mechanical coupling is performed by allocating the resulting micromechanical properties values to each voxel hosting the concerned microstructure phase as shown in Fig. 6a for indentation modulus (M) and Fig. 6b for hardness modulus (HD). The figures illustrate the very high contrast in micromechanical properties between microstructure phases. In future works, the mechanical elastic and inelastic behavior of cement paste microstructure will be predicted using a new variant of the LDPM (Lattice Discrete Particle Model) coupled to the calibrated μic microstructure model. This may lead to a deeper understanding of the relation between the microstructure properties and the cement paste and concrete mechanical behavior at the macroscopic scale.
Development of an Experimental-Numerical Approach
187
Fig. 3. Comparison of phase volumes over time between the calibrated μic model and GEMS model.
188
M. Krameche et al.
Fig. 4. Predicted phase assemblage at 120 days: calibrated μic vs GEMS.
Fig. 5. μic calibrated 3D microstructure (a) before hydration, (b) at 120 days.
Fig. 6. Micromechanical propertied distributed in μic microstructure voxels (a) indentation modulus M, (b) hardness modulus HD
Development of an Experimental-Numerical Approach
189
5 Conclusions In this paper, the microstructure of OPC paste was modeled using the μic model to include all the microstructure phases revealed by QXRD and GEMS phase assemblages. Their quantities were matched to GEMS phase assemblage through the calibration of the hydration kinetics parameters and adoption of appropriate hydration reaction equations. The microstructure model was validated in the long-term hydration age and will serve as input for multi-scale modeling applications. The following conclusions can be noted: • QXRD & GEMS at different hydration ages provided the evolution of phase assemblage for an advanced calibration of μic microstructure model with main and minor phases (e.g., HC, MC, HT). • Calibration of kinetic parameters and “adequate volumetric equations” are sufficient to match μic phase assemblage to the experimental evolution of all anhydrous phases and the GEMS evolution of most hydrates. • The amount of C-S-H predicted by the calibrated μic was overestimated by 30% compared to the GEMS result. This was addressed by adopting a correction coefficient from the GEMS phase assemblage to adjust the amount of C-S-H provided by C3 S hydration in the μic model. • The calibrated μic model was validated by predicting the phase assemblage of the OPC system at 120 days with a total solid phase volume 2.6% higher than the value predicted by GEMS. • The proposed approach provides inputs for multiscale micromechanical modeling of OPC paste through calibrated μic model.
References 1. Hantal, G., et al.: Atomic-scale modelling of elastic and failure properties of clays. Mol. Phys. 112(9–10), 1294–1305 (2014). https://doi.org/10.1080/00268976.2014.897393 2. Abdolhosseini Qomi, M.J., et al.: Combinatorial molecular optimization of cement hydrates. Nat. Commun. 5 (2014). https://doi.org/10.1038/ncomms5960 3. Zhang, H., Šavija, B., Schlangen, E.: Combined experimental and numerical study on microcube indentation splitting test of cement paste. Eng. Fract. Mech. 199, 773–786 (2018). https:// doi.org/10.1016/j.engfracmech.2018.04.018 4. Evaluating compressive mechanical LDPM parameters based on an upscaled multiscale approach | Elsevier Enhanced Reader. Accessed 28 Nov 2022 5. Saeed, H.A., Tagnit-Hamou, A., Ebead, U.A., Neale, K.W.: Stoichiometric study of activated glass powder hydration. Adv. Cem. Res. (2015). https://doi.org/10.1680/adcr.10.00067 6. Bishnoi, S., Scrivener, K.L.: Studying nucleation and growth kinetics of alite hydration using μic. Cem. Concr. Res. 39(10), 849–860 (2009). https://doi.org/10.1016/j.cemconres.2009. 07.004 7. Kulik, D.A., et al.: GEM-Selektor geochemical modeling package: revised algorithm and GEMS3K numerical kernel for coupled simulation codes. Comput. Geosci. (2012). https:// doi.org/10.1007/s10596-012-9310-6 8. Lothenbach, B., et al.: Cemdata18: a chemical thermodynamic database for hydrated Portland cements and alkali-activated materials. Cem. Concr. Res. 115, 472–506 (2019). https://doi. org/10.1016/j.cemconres.2018.04.018
190
M. Krameche et al.
9. Münch, B., Martin, L.H.J., Leemann, A.: Segmentation of elemental EDS maps by means of multiple clustering combined with phase identification. J. Microsc. 260(3), 411–426 (2015). https://doi.org/10.1111/jmi.12309 10. edxia: Microstructure characterisation from quantified SEM-EDS hypermaps | Elsevier Enhanced Reader. Accessed 30 Nov 2022 11. Bishnoi, S.: Vector modelling of hydrating cement microstructure and kinetics, p. 190 (2008) 12. Taylor, H.F.W.: Cement Chemistry, 2nd edn. T. Telford, London (1997) 13. Martin, L.H.J., Winnefeld, F., Müller, C.J., Lothenbach, B.: Contribution of limestone to the hydration of calcium sulfoaluminate cement. Cem. Concr. Compos. 62, 204–211 (2015). https://doi.org/10.1016/j.cemconcomp.2015.07.005 14. Lothenbach, B., Le Saout, G., Gallucci, E., Scrivener, K.: Influence of limestone on the hydration of Portland cements. Cem. Concr. Res. 38(6), 848–860 (2008). https://doi.org/10. 1016/j.cemconres.2008.01.002 15. Investigation of C-A-S-H composition, morphology and density in Limestone Calcined Clay Cement (LC3) | Elsevier Enhanced Reader. Accessed 01 Dec 2022 16. Georget, F., Wilson, W., Scrivener, K.L.: Simple automation of SEM-EDS spectral maps analysis with Python and the edxia framework. J. Microsc. 286(2), 185–190 (2022). https:// doi.org/10.1111/jmi.13099 17. Wilson, W., Sorelli, L., Tagnit-Hamou, A.: Unveiling micro-chemo-mechanical properties of C–(A)–S–H and other phases in blended-cement pastes. Cem. Concr. Res. 107, 317–336 (2018). https://doi.org/10.1016/j.cemconres.2018.02.010
Experimental and Numerical Investigations on Concrete Abrasion of Hydraulic Structures Qiong Liu
and Min Wu(B)
Department of Civil and Architectural Engineering, Aarhus University, 8000 Aarhus C, Denmark [email protected]
Abstract. Long-term abrasion has been a significant durability problem for hydraulic concrete structures. This work proposes combined experimental and numerical approaches in investigating concrete abrasion behaviors. A laboratory setup has been built to simulate the abrasion process of concrete subject to sediment particles transported by water similar to natural conditions. Threedimensional (3D) scanning analyses of the abraded concrete samples are then conducted to investigate the abrasion surface morphology and quantify the abrasion depth. Based on the obtained experimental results, numerical calculation studies are performed, aiming to develop a model that can account for important mesoscale factors for concrete abrasion and predict the abrasion damage behaviors of concrete structures. Mesoscale concrete is modeled as a heterogeneous 3-phase material composed of coarse aggregate particles, cement mortar matrix, and interfacial transitional zones (ITZs). Numerical calculations based on the generated mesoscale model are carried out to further study the effects of the interfacial bond and aggregate distribution on the abrasion resistance of concrete. The preliminary results show that mesoscale properties of concrete are very relevant in understanding the abrasion mechanisms and behaviors. Keywords: Concrete Abrasion · Hydraulic Structures · Durability · 3D Scanning · Mesoscale Modeling
1 Introduction Concrete abrasion damage of hydraulic structures is a cumulative process from the abrasive effects of sediments, including sands, gravels, rocks, and other debris, contained in the water flow impacting the concrete surface [1]. Therefore, the variation of the abrasion rates with exposure time may be critical for the prediction of the service life of the concrete structures [2, 3]. Besides, concrete abrasion damage involves the process of energy conservation in which part of the kinetic energy of the sediment-laden flow is converted into the fracture energy exerted on the concrete [4, 5]. The formation of cracks and the occurrence of abrasion damage are highly related to the fracture energy. It can be concluded that the kinetic energy embodied in the sediment-laden flow is the driving force leading to damages caused by concrete abrasion. Consequently, hydraulic parameters, e.g. velocity and sediment content of the concerned flow, are important factors when it comes to concrete abrasion damage [6, 7]. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 A. J˛edrzejewska et al. (Eds.): SynerCrete 2023, RILEM Bookseries 43, pp. 191–202, 2023. https://doi.org/10.1007/978-3-031-33211-1_18
192
Q. Liu and M. Wu
Due to the complexity of hydraulic conditions and the heterogeneous nature of concrete composites, it is somehow difficult to predict the concrete abrasion rates accurately merely through pure theoretical methods. As a result, the study on concrete abrasion behaviors requires substantial experimental research. The widely used test procedures include underwater test methods [8, 9] and water-borne impact methods [10, 11]. Almost all of the work reported in the literature use parameters such as the abrasion mass/volume loss or abrasion depth to quantify the abrasion rates [2, 12–14]. Large variations and inconsistencies are observed in the obtained results due to the difference in the testing procedures and the data analysis methodologies adopted. Apart from experimental methods, numeric approaches, e.g. two-dimensional (2D) mesoscale models, have been developed for concrete abrasion simulation, e.g. in [1]. However, there are significant constraints, e.g. it seems impossible to consider the effects of heterogeneous and random properties of concrete on abrasion modeling analyses in the reported 2D model. Moreover, circular aggregate particles are considered and distributed only in the near-surface layer. In this context, more advanced numeric approaches need to be sought. In this work, the effects of exposure time of sediment-laden flow acting on the concrete surface together with the coupling effects of the flow velocity and the sediment/sand content on concrete abrasion damage are studied, using a state-of-the-art home-built setup. Besides, three-dimensional (3D) mesoscale concrete models composed of different material phases are developed and some preliminary results based on the numerical simulations are reported.
2 Experimental Study 2.1 Sediment-Laden Flow Impact Method The sediment-laden flow impact test, which aims to simulate the process of abrasion damage of concrete surfaces exposed to sediment particles transported by water in natural conditions, is shown in Fig. 1. Detailed information regarding the test set-up can refer to the previously published research in [15]. The main pump used in the test setup was to provide power for circulating sediment-laden flow and the small water pump was used to continuously agitate the sands in the container and promote the mixing of the sand in the water. The continuous impact loads induced by the generated sediment-laden flow abraded the top layers of the concrete specimen. Before and after the 8-h abrasion test, the concrete specimen was weighed to determine the abrasion material loss. To better quantify the surface abrasion damage of the concrete, a 3D scanning method was used, as detailed in Sect. 2.3. 2.2 Raw Materials and Mixture Design In this study, two concrete mixtures with water–to–cement (w/c) ratios of 0.65 and 0.50 were investigated to assess the abrasion damage of normal-strength concrete and highstrength concrete. The detailed mixture design is presented in Table 1. The measured 28-day compressive strengths of the two concrete mixtures are 31 MPa and 58 MPa, respectively.
Experimental and Numerical Investigations
193
Fig. 1. Schematic diagram of the home-built test apparatus [15].
Table 1. Concrete mixtures used in this study (kg/m3 ). w/c
Cement
Water
River sand
Gravel
Compressive strength (MPa)
4–8 (mm)
8–16 (mm)
0.65
230
150
600
320
880
31
0.50
330
165
700
263
788
58
2.3 3D Scanning Method and the Data Analysis After the sediment-laden flow impact test, a digital copy of the concrete surface morphology can be constructed using the SOL 3D scanner (from Global Scanning). The XYZ coordinates of scattered points on the concrete surfaces can be obtained from the 3D scanner so that the abrasion surface information can be analyzed further through, e.g. MATLAB software. In this study, a MATLAB script is developed to visualize the abrasion depth based on all the measured data points. As an example for illustration, Fig. 2(a) shows a photo taken for the abrasion damage of a concrete specimen subjected to an 8-h test; Fig. 2(b) shows the reconstructed digital copy of the concrete surface obtained from the 3D scanner; and Fig. 2(c) presents the analyzed abrasion depths of the concrete specimen. Undoubtedly, compared with the original photo of the abraded surface of the concrete (see Fig. 2(a)), the analyzed results based on the 3D scanning (see Fig. 2(c)) are more distinct and readable. The method increases the accuracy of the measured abrasion depth. Normally, direct measurement of the abrasion depth is challenging since it is rather shallow in many cases (e.g. on the order of several millimeters). Instead of measuring a few spots, this method provides a good overview of the whole abraded area. In addition, it not only measures the abrasion depth (including the distribution) but also the abrasion areas. Such information is of significant value to analyze and understand the abrasion phenomena.
194
Q. Liu and M. Wu
Fig. 2. Concrete specimen surface after 8-h long abrasion: (a) Real concrete surface after the test; (b) Abrasion surface morphology from the 3D scanner; (c) Abrasion depth profile.
2.4 Experimental Results Concrete abrasion damage is a cumulative process of material loss, which indicates that the exposure time of the flow acting on the surface, is an important influence parameter for concrete abrasion damage [16]. Figure 3 shows the abrasion damage morphology of concrete specimens with a high strength of 58 MPa subjected to 8-h, 16-h, and 24-h tests. As expected, the abrasion depth increases as the exposure time extends. When the abrasion time is only 8 h, the maximum abrasion depth of the concrete specimen is 4 mm. After the 16-h abrasion test, the maximum abrasion depth increases up to 6 mm. When the concrete slab is subjected to the 24-h test, the maximum abrasion depth is also nearly 6 mm but with much larger abrasion damage areas at large depth (>5 mm). It can also be noted that in the first 8-h test, the abrasion rate (characterized by the depth variation gradient) is the largest since the top layer of concrete is mainly composed of mortar that can be abraded away easily at the beginning of the test. However, as the abrasion damage propagates deeper and the aggregates are subjected to direct impinging loads, abrasion rates decrease which can be explained by the higher hardness of aggregates as compared with that of mortar [17, 18]. As mentioned earlier, the flow velocity and sediment content can greatly affect concrete abrasion rates. In the sediment-laden flow method, the flow velocity cannot be
Experimental and Numerical Investigations
195
Fig. 3. Effects of the exposure time on concrete abrasion: (a) 8-h test; (b) 16-h test; (c) 24-h test.
regarded as a single influencing parameter since the violent disturbance of flow at high speed will increase the sediment concentration, i.e. they are linked to each other. To investigate the coupling effects of the flow velocity and the sand content on concrete abrasion damage, the relation between the flow velocity and the sand content should be determined first. Using the home-built setup (Fig. 1), the flow velocity used in the test system is adjusted by the large pump frequency. The relation between the pump frequency and the flow velocity is shown in Fig. 4(a), and the relation between the flow velocity and the sand content is shown in Fig. 4(b). Three pump frequencies of 20 Hz, 35 Hz, and 50 Hz are set to evaluate the abrasion damage corresponding to the velocity of 3.81 m/s, 8.01 m/s, and 12.21 m/s, and the sand content of 46 kg/m3 , 99 kg/m3 , and 151 kg/m3 . The coupling effects of the flow velocity and the sand content on the abrasion volume loss of concrete slabs with a strength of 31 MPa are shown in Fig. 5. The abrasion damage morphologies of concrete slabs after the 8-h test are shown in Fig. 6, while the corresponding analyzed contours of the concrete abrasion damage are shown in Fig. 7. It can be found that the highest flow velocity and the largest sand content result in the largest volume loss of 39 cm3 (Fig. 5) and the deepest maximum abrasion depth at 5 mm (Fig. 7(c)). Lengthy and shallow grooves on the concrete surface are observed in Fig. 6(c). Besides, the total removal of aggregates is also investigated because the
196
Q. Liu and M. Wu
Fig. 4. (a) Relation between the pump frequency and the velocity; (b) Relation between the flow velocity and the sand content.
Fig. 5. Coupling effects of the flow velocity and the sand content on abrasion volume loss.
mortar surrounding the aggregates is gradually peeled away under the repeated impacting loading. On the contrary, the volume loss of the concrete slab is only 0.9 cm3 (Fig. 5) and the maximum abrasion depth is only 0.3 mm (Fig. 7(a)) when the velocity is 3.81 m/s and the sand content in the jet is 46 kg/m3 . There exists no obvious abrasion damage on the concrete surface, which means that only slight scouring marks can be investigated on the mortar layer in Fig. 6(a). This may be because the hydraulic pressure induced by the sediment-laden flow impacts is below the critical pressure for hydraulic erosion to occur. This critical pressure is about 30 times the tensile strength of the concrete specimen [16]. When the flow velocity is 8.01 m/s and sand content achieves 99 kg/m3 , the abrasion volume loss keeps at a medium level and the surface abrasion can be investigated but without a deep abrasion groove on the surface. And the maximum abrasion depth is also at a medium level of 4.5 mm (Fig. 7(b)). This is because the mortar phase is the weakest part of the concrete material which is easily abraded away. After that, aggregates are subjected directly to the sediment-laden flow load. But when the flow velocity is about 8.01 m/s, the ability of the flow to take out the aggregates is reduced as compared with the case under the largest loading velocity of 12.21 m/s.
Experimental and Numerical Investigations
197
Fig. 6. Abrasion damage morphologies of concrete slabs: (a) 20 Hz; (b) 35 Hz; (c) 50 Hz.
Fig. 7. Analyzed concrete abrasion damage contours: (a) 20 Hz; (b) 35 Hz; (c) 50 Hz.
198
Q. Liu and M. Wu
From the perspective of energy conservation, a part of the kinetic energy acting on the sediment-laden flow will be absorbed by the concrete slabs by repeated impaction, whereas kinetic energy characterizes the intensity of dynamic load. So sediment-laden flow has higher kinematic energy and hence larger intensity when flow velocity and sand content increase [7]. It will cause considerable abrasion damage because the cracks will initiate and propagate rapidly in a short time.
3 Numerical Analysis 3.1 3D Meso-Scale Concrete Model In this study, 3D mesoscale concrete models are developed based on Voronoi tessellation techniques [19] to investigate the concrete abrasion behaviors at the mesoscale level. The cube representative volume element with dimensions of 50 mm × 50 mm × 50 mm is separated by Voronoi cells, whereas each cell is corresponding to each seed point placed in advance randomly in the volume region. Then the Voronoi cells are scaled through a shrinkage algorithm based on the center positions of Voronoi cells to generate separated polyhedron particles. The mortar phase is generated by cutting the volume of polyhedron particles from the cube. Based on the generated polyhedron particles and mortar, the interfacial transitional zones (ITZs) are created through the shrinking of the polyhedron boundary inward by the thickness value of ITZs. The gravity-deposition in the discrete element method is applied in this study to obtain the close-packed spheres within the volume region as shown in Fig. 8(a). The sphere centers are selected as seeds of Voronoi cells and the volume of each sphere is obtained for the calculation of scaling ratios of Voronoi cells. To avoid infinity points in the generation process of Voronoi cells, we enlarge the volume range of seed points to 70 mm × 70 mm × 70 mm. But the Voronoi cells are only collected corresponding to these seeds in the cube volume region with the dimensions of 50 mm × 50 mm × 50 mm. The part of Voronoi cells protruding out of the cube with a side length of 50 mm will be cut off from the cube boundary.
Fig. 8. Model generation: (a) The gravity-deposition method to obtain closely packed spheres; (b) The shrinking process for separating Voronoi cells to obtain the polyhedron particles.
Experimental and Numerical Investigations
199
The scaling ratios are calculated to separate Voronoi cells into independent particles. It should be noted that the shrinking factor should be the same for one Voronoi cell, otherwise the convexity of the polyhedron will be destroyed. The shrinking factor for various Voronoi cells is normally less than 1 (in some cases the factor may be greater than 1, and the ratio would be set to 0.95 which is also for reserved space of ITZs). The shrinking process of one Voronoi cell is shown in Fig. 8b. The polyhedron particles are imported into the software ABAQUS and then the mortar phase generation is finished through the Boolean technique by subtracting the volume of polyhedron particles from the total volume of the cube. The generated geometry model is meshed by tetrahedral elements (C3D4), as a result, the polyhedron particles and mortar share nodes at their boundaries. Polyhedrons-shrinking technique is applied to create wedge elements about 0.1 mm that represent ITZs, thin layers in mesoscale concrete as shown in Fig. 9. The generated mesoscale concrete is modeled as a three-phase material composed of mortar, aggregates, and ITZs, as shown in Fig. 10.
Fig. 9. Wedge elements to represent ITZs.
Fig. 10. 3D mesoscale concrete model.
200
Q. Liu and M. Wu
3.2 Material Property The linear elastic material properties of coarse aggregates are considered in normalstrength concrete. Concrete damage plasticity (CDP), available in the ABAQUS material library, is used to describe mortar and ITZ behaviors. Two main failure mechanisms described in the CDP model are tensile cracking and compressive crushing where tensile damage and compression damage are used to consider stiffness degradation. The main material property parameters assigned to each phase of concrete are listed in Table 2. Table 2. Material properties. Property
Aggregates
Mortar
ITZs
Elastic modulus E (MPa)
70000
50000
20000
Poisson’s ration ν
0.2
0.2
0.2
Tensile strength (MPa)
–
6
3
Compressive strength (MPa)
–
40
20
Density ρ (kg/m3 )
2600
2000
2000
Fracture energy Gf (N/mm)
–
0.035
0.02
3.3 Loads and Boundary Conditions The continuous impact loads induced by the sediment-laden flow can be assumed as the sequential arrival of uniform flow layers [20]. The uniform sediment-laden layer can be defined as the equivalent rigid body and the abrasion behaviors of the mesoscale concrete can be simulated under the repeated impact of the rigid body through finite element simulations with ABAQUS/Explicit. Numerically, the boundary conditions on the X-direction and Y-direction are fixed. The velocity load of 8 m/s in the Z-axis direction is applied to the rigid body. 3.4 Simulation Analysis Results The damage results under a single impact are shown in Fig. 11. The elements in red with damage index SDEG ≥ 0.9 represent abrasion damage regions that can be visualized by deleting the corresponding elements. After the scratching damage of the concrete surface, cracks will appear firstly at the ITZs which are the weakest part of the concrete model and then accumulate to the case that most proportion of ITZs will damage. When the loss of mortar reaches a certain depth comparable to the size of the aggregate, the total removal of the aggregate particle may occur. This is consistent with the phenomena investigated in the experimental tests discussed in Sect. 2.4 in which the ITZs are damaged first and then the aggregates are taken out totally.
Experimental and Numerical Investigations
201
Fig. 11. Abrasion damage results obtained by numerical simulation.
4 Conclusions This work presented a sediment-laden flow impact method for simulating the abrasion damage of concrete for hydraulic structures together with a new abrasion measurement and analysis method based on 3D scanning. In addition, a mesoscale concrete model was developed for investigating the abrasion behaviors of concrete. From the results and discussions presented above, the following can be concluded: 1. The effects of key parameters closely related to concrete abrasion damage (e.g. abrasion time and flow characteristics) were investigated with some important results obtained. It was demonstrated that the abrasion material loss increased with an increase in the abrasion time. High-speed flow with strong disturbance would increase the sediment content contained inside of the flow, resulting in significant abrasion damage. As the flow velocity increased from 3.81 m/s to 12.21 m/s and sand contents in the flow correspondingly increased from 46 kg/m3 to 151 kg/m3 , the concrete volume loss and the maximum abrasion depth increased from 0.9 cm3 to 39 cm3 , and 0.3 mm to 5 mm, respectively. 2. The simulation results indicated that the surface layer mortar was peeled away first and then cracks propagated further along the ITZs. This agreed with the observed experimental phenomena. The numerical investigations are still at a preliminary stage. Future work will focus on simulating the repeated impacts exerted by the sediment-laden flow and the numerical analyses will be compared with experimental observations for validation. Acknowledgments. The authors wish to thank the COWI Foundation for the support on part of this work and Prof. Kenny Kataoka Sørensen (Department of Civil and Architectural Engineering, Aarhus University) for the use of the 3D scanner.
202
Q. Liu and M. Wu
References 1. Makarova, N.V., Polonik, M.V., Mantsybora, A.A.: Abrasive wear of cemented granular composites: experiments and numerical simulations. IOP Conf. Ser. Mater. Sci. Eng. (2018) 2. Momber, A.W.: Effects of erodent flow energy and local exposure time on the erosion of cement-based composites at high-speed hydro-abrasive flow. Wear 378–379, 145–154 (2017) 3. Momber, A., Kovacevic, R.: Fundamental investigations on concrete wear by high velocity water flow. Wear 177, 55–62 (1994) 4. Wu, Z., et al.: Coupling effect of strain rate and specimen size on the compressive properties of coral aggregate concrete: a 3D mesoscopic study. Compos. Part B Eng. 200 (2020) 5. Bitter, J.G.A.: A study of erosion phenomena Part I. Wear 6, 169–190 (1963) 6. Zarrabi, N., Moghim, M.N., Eftekhar, M.R.: Effect of hydraulic parameters on abrasion erosion of fiber reinforced concrete in hydraulic structures. Constr. Build. Mater. 267, 120996 (2021) 7. Padhy, M.K., Saini, R.P.: Effect of size and concentration of silt particles on erosion of Pelton turbine buckets. Energy 34, 1477–1483 (2009) 8. Horszczaruk, E., Brzozowski, P.: Effects of fluidal fly ash on abrasion resistance of underwater repair concrete. Wear 376–377, 15–21 (2017) 9. Ramesh Kumar, G.B., Sharma, U.K.: Abrasion resistance of concrete containing marginal aggregates. Constr. Build. Mater. 66, 712–722 (2014) 10. Liu, Y.W., Yen, T., Hsu, T.H.: Abrasion erosion of concrete by water-borne sand. Cem. Concr. Res. 36, 1814–1820 (2006) 11. Liu, Y.W., Lin, Y.Y., Cho, S.W.: Abrasion behavior of steel-fiber-reinforced concrete in hydraulic structures. Appl. Sci. 10 (2020) 12. García, A., Castro-Fresno, D., Polanco, J.A., Thomas, C.: Abrasive wear evolution in concrete pavements. Road Mater. Pavement Des. 13, 534–548 (2012) 13. Siddique, R.: Effect of fine aggregate replacement with Class F fly ash on the abrasion resistance of concrete. Cem. Concr. Res. 33, 1877–1881 (2003) 14. Shi, Z.Q., Chung, D.D.L.: Improving the abrasion resistance of mortar by adding latex and carbon fibers. Cem. Concr. Res. 27, 1149–1153 (1997) 15. Liu, Q., Li, L., Andersen, L.V., Wu, M.: Studying the abrasion damage of concrete for hydraulic structures under various flow conditions. Cem. Concr. Compos. 135, 104849 (2023) 16. Hocheng, H., Weng, C.H.: Hydraulic erosion of concrete by a submerged jet. J. Mater. Eng. Perform. 11(3), 256–261 (2002). https://doi.org/10.1361/105994902770344033 17. Kiliç, A., et al.: The influence of aggregate type on the strength and abrasion resistance of high strength concrete. Cem. Concr. Compos. 30, 290–296 (2008) 18. Sosa, C.T.I., Polanco, J.A.P., Setién, J., Sainz-Aja, J.A., Tamayo, P.: Durability of highperformance self-compacted concrete using electric arc furnace slag aggregate and cupola slag powder. Cem. Concr. Compos. 127, 104399 (2022) 19. Naderi, S., Tu, W., Zhang, M.: Meso-scale modelling of compressive fracture in concrete with irregularly shaped aggregates. Cem. Concr. Res. 140, 106317 (2021) 20. Dandapat, R., Deb, A.: A probability based model for the erosive wear of concrete by sediment bearing water. Wear 350–351, 166–181 (2016)
Multi-physics Modelling of Concrete Shrinkage with the Lattice Discrete Particle Model Considering the Volume of Aggregates Yilin Wang1
, Roman Wan-Wendner1(B) , Giovanni Di Luzio2 and Jan Belis1
, Jan Vorel3
,
1 Magnel–Vandepitte Laboratory, Department of Structural Engineering and Building Materials,
Ghent University, 9052 Ghent, Belgium [email protected] 2 Department of Civil and Environmental Engineering, Politecnico di Milano, 20133 Milano, Italy 3 Department of Mechanics, Faculty of Civil Engineering, Czech Technical University in Prague, 16629 Praha 6, Czech Republic
Abstract. Concrete durability plays an important role in the serviceability of reinforced concrete structures. Deformation induced by shrinkage and thermal strains can lead to the initiation of cracks which in turn may develop into structural damage during several decades of service life. It is time-consuming and impractical to experimentally investigate the long-term mechanical behavior considering environmental influence factors. Hence, a state-of-the-art numerical simulation framework combining the Lattice Discrete Particle Modelling (LDPM) with a multiphysics framework is applied, coupling the mechanical behavior and chemical mechanisms of concrete at an early age and beyond. Based on an equivalent rheological model, the overall age-dependent deformations of concrete can be split into contributions from different physical phenomena assuming the additivity of strain and strain rate in the sense of one-way coupling. The hygro-thermal-chemical model describing the moisture transport, heat transfer and curing reaction drives the development of mechanical properties due to ongoing curing but also thermal and hygral eigenstrains. LDPM reflects the inherently heterogeneous nature of the material concrete at the mesoscale consisting of aggregates and mortar. The effect of aggregate volume and stiffness on concrete shrinkage is investigated by a newly proposed formulation for drying shrinkage of concrete. The results give robust predictions of macroscopic shrinkage for concretes with different mix proportions. A well-established experimental test campaign is selected to calibrate and validate the numerical model, which shows a good agreement and offers promising new insights into the cracking behavior of heterogeneous materials with acceptable computational cost. Keywords: Shrinkage · Lattice Discrete Particle Modelling · Concrete · Multi-physics · Aggregates
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 A. J˛edrzejewska et al. (Eds.): SynerCrete 2023, RILEM Bookseries 43, pp. 203–213, 2023. https://doi.org/10.1007/978-3-031-33211-1_19
204
Y. Wang et al.
1 Introduction Concrete is experiencing volume changes and time-dependent deformations throughout the service life. One of the types of time-dependent deformations is concrete shrinkage, which significantly influences the long-term mechanical behavior of concrete structures. The volume changes in concrete due to shrinkage potentially initiate internal microcracks [1] and may even lead to crack propagation and failure of concrete, especially when it comes to restrained conditions [2]. The autogenous and drying shrinkage are two main causes that result in the early degradation of concrete structures due to volume changes. Drying shrinkage is defined as the reduction in concrete volume over time due to external water loss, while autogenous shrinkage is defined as the decrease in concrete volume due to cement hydration. Several models have been proposed and developed for the reproduction and prediction of concrete shrinkage. Typically, concrete is treated as a homogenous material the behaviour of which can be described by a set of (time-dependent) concrete properties. This approach is typically chosen for structural applications making use of the Finite Element Methods (FEM) or analytical tools. Concrete can also be modelled considering it to be a multiscale material, for example, consisting of coarse aggregates, mortar matrix, Interfacial Transition Zone (ITZ) at the mesoscale [3]. Also in this case the continuum method [4, 5] and FEM analysis can be used if the constituents and ITZ are explicitly modelled. In this case, a very fine mesh is required, considerably increasing the computational resource requirement. The Discrete Element Method (DEM) using spheres elements [6] and lattice models using beam elements [7] are better methods to represent the heterogenous nature of the concrete and local phenomena related to specific meso-structures with simple constitutive laws. Although the concrete deformation due to shrinkage can be predicted by a coupled multi-physics framework and some plausible results are given in previous numerical studies [8, 9], the local interaction of reactive mortar and non-reactive aggregates have never been realistically studied in this context. In this paper, an extension of the mesoscale numerical model, the Lattice Discrete Particle Model (LDPM) [10], is proposed and used to predict the mechanical behavior of concrete [11, 12] subject to drying shrinkage. The formulations of the hygro-thermochemical (HTC) [13] are applied on the Flow Lattice Element (FLE) network [14] to describe the moisture transport and hydration reaction. The Multi-physics Lattice Discrete Particle Model (M-LDPM) combining these two features and in particular employing the newly proposed formulation for drying shrinkage are able to capture the drying shrinkage behavior of heterogenous cementitious materials in a realistic manner.
2 Multi-physics Framework 2.1 Mechanical Model The mechanical behavior of concrete is able to be captured and reproduced by LDPM at the mesoscale, where the coarse aggregates are assumed approximately as spheres and randomly placed inside of the generated concrete structure as show in Fig. 1. The typical particle size distribution is governed by the classical Fuller curve with optimal packing properties. A 3D tessellation (Fig. 1) is generated by connecting edge points, face points
Multi-physics Modelling of Concrete Shrinkage
205
and tet point based on Delaunay tetrahedralization in each tetrahedron, which is formed by four adjacent particles. Each particle is therefore surrounded by these collected facets of tessellations. The potential damage and fracture are assumed to take place in the mortar between aggregate pieces, so the LDPM formulations are performed in the mortar as well, i.e., the facets of tessellations in mesoscale.
Fig. 1. LDPM ( with tessellation of a typical LDPM tetrahedron) and FLE systems.
The details of LDPM formulations can be found in [10]. Here a brief summary of the formulation is given: A displacement jump at the centroid of each facet is defined though the rigid body kinematics. The strain components can be formulated as follows: (1) where n, m and l are unit vectors at local reference system; is the interparticle distance. The elastic behavior LDPM constitutive law is defined according to the assumption that the stresses are proportional to the corresponding strains: σN = E0 εN ; σM = αE0 εM ; σL = αE0 εL . Where the normal modulus E0 and shear-normal coupling parameter α
206
Y. Wang et al.
have relationships with Young’s modulus E and Poisson’s ratio ν, respectively. One can write E0 = E/(1 − 2ν) and α = (1 − 4ν)/(1 + ν). The inelastic behavior is described as three kinds of concrete failure mechanisms. Fracturing behavior can be described by the relationship between the effective strain and 2 2 2 2 2 stress in formulations of ε = εN + α εM + εL and σ = σN + σM + σL2 /α when it comes to σ > 0. The effective stress yields 0 ≤ σ ≤ σbt (ε, ω), which can be formulated as: σbt (ε, ω) = σ0 (ω)exp[−H0 (ω)εmax − ε0 (ω)/σ0 (ω)] and σ0 (ω) can be expressed as: σ0 (ω) = σt −sin(ω) + sin2 (ω) + 4α cos2 (ω)/rst2 / 2α cos2 (ω)/rst2
(2)
(3)
and rst is the ratio between shear and tensile strength; ω is the degree of interaction between normal and shear; x = max(x, 0); ε0 (ω) and εmax refer to the elastic limit and maximum effective strain, respectively; H0 (ω) = Ht (2ω/π)nt , where Ht and nt are the softening modulus and softening exponent under pure tension. The second mechanism is used to describe pore collapse and material compaction when it comes to εN < 0. The compressive normal stress yields −σbt (εD , εV ) ≤ σN ≤ 0, where εD and εV are the volumetric strain and deviatoric strain, respectively. For strain-dependent boundary, one can write: ⎧ ⎨ σc0 if − εDV ≤ 0 (4) σbt (εD , εV ) = σc0 + −εDV − εc0 Hc (rDV ) if 0 ≤ −εDV ≤ εc1 ⎩ σc1 (rDV ) exp{[(−εDV − εc1 )Hc (rDV )]/[σc1 (rDV )]} otherwise in which σc0 is the mesoscale yielding compressive stress; εDV = εV +βεD , β is material parameter; Hc (rDV ) is the hardening modulus; σc1 (rDV ) = σc0 + (εc1 − εc0 )Hc (rDV ). Finally, the frictional effects occur due to the shear stress under compression states. The incremental shear stresses are formulated as follows:
˙p ˙ − εM (5) σ˙M = ET εM
p σ˙L = ET ε˙L − ε˙L
(6)
˙p = λ˙ ∂ϕ/∂σ , ε˙p = λ˙ ∂ϕ/∂σ and λ is the plastic multiplier. The plastic where εM L M L 2 2 potential yields ϕ = σM + σL σbs (σN ), where the shear strength yields: σbs (σN ) = σs + (μ0 − μ∞ )σN 0 − μ∞ σN − (μ0 − μ∞ )σN 0 × exp(σN /σN 0 )
(7)
in which σs refers to cohesive strain; μ0 and μ∞ are the initial and final internal friction coefficients, respectively; σN 0 is the normal compression stress relevant to friction coefficient transition.
Multi-physics Modelling of Concrete Shrinkage
207
2.2 Hygro-Thermo-chemical (HTC) Model A Flow Lattice Element (FLE) network [14] is used to analyse moisture transport, heat transfer and hydration in concrete respecting the mesoscale structure, i.e. particle placement. Each typical FLE is generated by connecting the centroids T1 T2 of two adjacent tetrahedrons, as shown in Fig. 1. In this case, a comprehensive HTC model is utilized here to simulate the concrete shrinkage considering the temperature T , relative humidity h, and cement hydration degree αc . The moisture mass balance and enthalpy balance equations can be written as [13]: ∇ · (Dh ∇h) −
∂we ∂h ∂we ∂we − α˙c − α˙s − w˙n = 0 ∂h ∂t ∂αc ∂αs
∂T Qc∞ + α˙ s s Qs∞ = 0 + α˙ c c ∂t −1 Ead D1 Ead D1 × 1 + Dh (h, T ) = exp − − 1 (1 − h)n RT0 RT D0 ∇ · (λ∇T ) − ρct
(8) (9)
(10)
where c = cement content; s = silica content;αs is silica fume reaction degree; we = Qs∞ are the cement evaporable water; ρ = mass density; λ = heat conductivity; c Qc∞ and hydration enthalpy and latent heat of silica fume. The cement hydration and silica fume reaction can be expressed as [15]: α˙c = Ac1
−1 −η α /α ∞ Ac2 ∞ b c c c 1 + − ah) α e − α e−Eac /RT (a c αc∞ + αc c α˙s = As1
As2 ∞ ∞ αs − αs e−ηs αs /αs e−Eas /RT + αs
αs∞
(11) (12)
where ηc , ηc , Ac1 , Ac2 , ηs , As2 are material parameters. More details of this model can be found in [12]. 2.3 Improved Shrinkage Model The HTC model is implicitly implemented in the FLE system [16, 17], which is coupled with LDPM for achieving the multi-physics simulation at the mesoscale. In the FLE system, each FLE conduit is formed by the volume that the element FLE12 occupies (Fig. 1), in which the mortar mass is filled. For simplicity, the two-dimension (2D) diagram is show in Fig. 2. For the original model (named 1-phase model, see Eq. 1) [8, 9], it can be seen that the calculation of shrinkage is based on the total (sum of mortar and aggregate) volume (concrete) while the improved formulation (named 2-phase model, see Eq. 2) considers the local contributions of shrinking mortar and elastic aggregate. As a result, the former assumes the behavior of concrete as a homogeneous material at the macroscale. However, that is contradictory with the assumption that only mortar is involved in the chemical reaction, and aggregates remain non-reactive. Therefore, the 2-phase model was proposed here to exclude the volume (P2 A1 A2 & P3 A3 A4 ) of
208
Y. Wang et al.
aggregates in mesoscale calculation of shrinkage strains, potentially leading to more accurate predictions of shrinkage strain fields and in turn cracking. c c ˙ ε˙ sh = αsh h
(13)
Vagg m m˙ ε˙ sh = αsh h 1− Vtot
(14)
c and ε˙ m represents the shrinkage strain rate calculated in 1-phase and 2-phase where ε˙ sh sh c and α m are the shrinkage coefficients of concrete and mortar, models, respectively; αsh sh respectively; Vagg and Vtot are the aggregate volume and total volume, respectively.
(a) 1-phase model
(b) 2-phase model
Fig. 2. 2D diagram for FLE conduit in 1-phase model and 2-phase model
2.4 One-Way Coupling Based on the equivalent rheological model, a one-way coupling method is employed to describe the overall time-dependent deformations of concrete due to different physical phenomena by assuming the additivity of strain and strain rate at the mesoscale. One ˙ , where elastic strain rate ε˙e , damage strain rate can write: εtot ˙ = ε˙sh + ε˙th + ε˙e + εdam εdam ˙ and shrinkage strain rate can be obtained by the formulations of LDPM explained in Sect. 2.1 and shrinkage models in Sect. 2.3, respectively; the thermal strain rate ε˙th can be expressed as: ε˙T = αth T˙ , αth is the coefficient of thermal deformations.
Multi-physics Modelling of Concrete Shrinkage
209
3 Simulations of Experimental Campaign To demonstrate the capability and applicability of proposed shrinkage model in the M-LDPM framework, an experimental campaign was selected for simulation, calibration/validation and prediction. 3.1 Experimental Description A series of experimental tests on the shrinkage behavior of specimens with different sizes of concrete members were carried out by Bryant et al. [18]. The cement content of concrete is 390 kg/m3 . The water-to-cement and aggregate-to-cement ratios are 0.47 and 5.1 respectively. The maximum size of aggregate is 24 mm, and a minimum aggregate size of 10 mm was assumed in the simulation work. The reported prism compressive strength at 28 days is 50.1 MPa. The specimens were cured at a constant temperature of 20 ◦ C and 60% RH, then kept in controlled conditions for a period of 6 days in the same temperature and 95% RH, followed by a drying test at 60% RH. The two ends of the square prism were sealed with aluminium foil. Four mechanical gauges of 200 mm length were bonded on the sides to measure the longitudinal strains. Two geometries of 150 × 150 ×600 mm and 300 × 300 × 1200 mm were selected here to numerically simulate and reproduce the shrinkage process. Fig. 3 presents the coarse aggregate distribution and cell elements.
(a) Aggregate distribution
(b) Cell elements
Fig. 3. LDPM discretization of shrinkage specimens
3.2 Calibration of Model Parameters In the M-LDPM simulation used in this paper, a set of material parameters has to be calibrated from experimental data. The geometry-dependent parameters of LDPM can
210
Y. Wang et al.
be obtained by the concrete mix design directly. And the LDPM parameters relevant to the mechanical model are identified firstly. A series of experimental tests, including unconfined compression, biaxial loading test, triaxial compression, cyclic uniaxial compression etc., are required to determine all parameters. More details concerning the parameter identification process can be found in [19]. All mechanical parameters used in this paper are listed in Table 1 and have been obtained from published calibrations on similar concretes [19]. The second set of parameters controlling the moisture transport and heat transfer comes from the HTC model. As before, the HTC model requires a substantial number of parameters. Part of parameters have already been calibrated and validated in [20, 21], also including reasonable assumptions. The remaining parameters can be obtained by fitting an autogenous shrinkage curve and drying shrinkage curve. The sealed test should be normally used to calibrate the moisture diffusion related parameters of the HTC model first. But the experimental data was not ideal due to the possible loss of sealing during the test. Therefore, the drying test of 150 × 150 × 600 mm specimen was also selected for calibration. The identified parameters can be found in Table 1. The Table 1. Calibrated parameters of LDPM, HTC and shrinkage model. Parameter
Unit
Description
Value
dmin
mm
Minimum aggregate size
10
dmax
mm
Maximum aggregate size
25
nF
–
Fuller coefficient
0.6
E0
MPa
Effective normal modulus
47630
σt
MPa
Tensile strength
4.4
lt
mm
Tensile characteristic length
211
Ac1
108 /h
Cement hydration parameter
4.255
Ac2
10–2
Cement hydration parameter
3.75
ηc
–
Cement hydration parameter
9.3
ct
kJ/kg ◦ C
Isobaric heat capacity
1.1
λt
W/m ◦ C
Heat conductivity
2.3
D0
Moisture permeability in dry situation
5.3
D1
10–11 kg/mm h 10–7 kg/mm h
Moisture permeability in saturated situation
6.5
n
–
Exponent parameter
6.5
g1
–
Material parameter
1.75
g2
–
Material parameter
0.025
c αsh m αsh
10–3
Shrinkage coefficient of concrete
1.92
10–3
Shrinkage coefficient of mortar
2.61
αth
10–5
Thermal expansion coefficient
1
Multi-physics Modelling of Concrete Shrinkage
211
same parameters were used in the simulation of 300 × 300 × 1200 mm specimen for validation because of the same water-to-cement ratio. Finally, the parameters relevant to the shrinkage and thermal deformations need to c and α m can be identified through the sealed be calibrated. The shrinkage coefficient αsh sh and drying tests. The calibrated αsh in both 1-phase and 2-phase models are listed in Table 1. For thermal coefficient αth , it always remains 1 e−5 1/◦ C based on [20, 21].
4 Results and Discussion Figure 4 plots the numerical and experimental results of shrinkage strains over the drying time. It can be seen that the simulations fit the experimental data well. The results of 300 × 300 × 1200 mm specimen are used for the validation of the 2-phase model, and the consistency between the numerical and experimental data proves the model’s predictability. Additionally, a prediction of 75 × 75 × 300 mm specimen is shown and the maximum shrinkage strain reaches approximately 700 μm/m, similar to the other two sizes. Generally, the smaller the specimen, the higher the slope of the curve and the faster the drying process. For comparison, the shrinkage strain over time calculated by the 1-phase model was also plotted in Fig. 4 for the calibration specimen and showed an evolution comparable to that of the 2-phase model. However, the local shrinkage strains of the 1-phase and 2-phase models are completely different. The contour plot shown in Fig. 5 presents the strain comparison in mid-section. It can be seen that the shrinkage strain distribution in case of the 1-phase model is more or less uniform contrary to that of the 2-phase model. This demonstrated the advantage of the 2-phase model which is able to distinguish reactive and non-reactive materials and capture local differences in the strain fields such as the so-called wall effect. Consequently, the surface cracks calculated by 2-phase model are obviously different from that of the 1-phase model, as shown in Fig. 6. It should be noted that for the 1-phase model, the cracks occurred at early age but close after a period of
Fig. 4. Numerical and experimental results of shrinkage strains.
212
Y. Wang et al.
drying. This is because the moisture loss started from the outside with the drying front traveling inside of concrete, ultimately leading to a uniform humidity distribution field and thus total shrinkage strain field.
Fig. 5. Comparison of shrinkage strain distribution.
Fig. 6. Comparison of surface cracks.
5 Conclusions A multi-physics framework coupling the LDPM for the mechanical analysis and a FLE system with the HTC model for the calculation of transport was applied to investigate concrete shrinkage. An improved shrinkage model, a 2-phase model, considering the influence of aggregate and mortar contribution was proposed in this paper. This new model enables to capture the local shrinkage strain distribution in a likely more realistic manner and reproduces richer surface crack patterns. An experimental campaign was selected for the calibration and validation. The simulation and experimental results are in good agreement. Importantly, the predictive ability of the new model is verified.
Multi-physics Modelling of Concrete Shrinkage
213
References 1. Bolander, J., John, E., Berton, S.: Simulation of shrinkage induced cracking in cement composite overlays. Cem. Concr. Compos. 26(7), 861–871 (2004) 2. Guo, H., et al.: A quadtree-polygon-based scaled boundary finite element method for imagebased mesoscale fracture modelling in concrete. Eng. Fract. Mech. 211, 420–441 (2019) 3. Unger, J., Eckardt, S.: Multiscale modeling of concrete. Arch. Comput. Methods Eng. 18, 341–393 (2011) 4. Häfner, S., Eckardt, S., Luther, T., Könke, C.: Mesoscale modeling of concrete: geometry and numerics. Comput. Struct. 84(7), 450–461 (2006) 5. Wriggers, P., Moftah, S.: Mesoscale models for concrete: homogenisation and damage behavior. Finite Elem. Anal. Des. 42(7), 623–636 (2006) 6. Qin, C., Zhang, C.: Numerical study of dynamic behavior of concrete by meso-scale particle element modeling. Int. J. Impact Eng 38(12), 1011–1021 (2011) 7. Schlangen, E., Garboczi, E.: Fracture simulations of concrete using lattice models: computational aspects. Eng. Fract. Mech. 57(2–3), 319–332 (1997) 8. Abdellatef, M., Boumakis, I., Wan-Wendner, R., Alnaggar, M.: Lattice discrete particle modeling of concrete coupled creep and shrinkage behavior: a comprehensive calibration and validation study. Constr. Build. Mater. 211, 629–645 (2019) 9. Alnaggar, M., Di Luzio, G., Cusatis, G.: Modeling time-dependent behavior of concrete affected by alkali silica reaction in variable environmental conditions. Materials 10(5), 471 (2017) 10. Cusatis, G., Pelessone, D., Mencarelli, A.: Lattice discrete particle model (LDPM) for failure behavior of concrete. i: theory. Cem. Concr. Compos. 33(9), 881–890 (2011) 11. Lale, E., Rezakhani, R., Alnaggar, M., Cusatis, G.: Homogenization coarse graining (HCG) of the lattice discrete particle model (LDPM) for the analysis of reinforced concrete structures. Eng. Fract. Mech. 197, 259–277 (2018) 12. Alnaggar, M., Pelessone, D., Cusatis, G.: Lattice discrete particle modeling of reinforced concrete flexural behavior. J. Struct. Eng. 145(1), 04018231 (2019) 13. Di Luzio, G., Cusatis, G.: Hygro-thermo-chemical modeling of high performance concrete. i: theory. Cem. Concr. Compos. 31(5), 301–308 (2009) 14. Shen, L., et al.: On the moisture migration of concrete subject to high temperature with different heating rates. Cem. Concr. Res. 146, 106492 (2021) 15. Cervera, M., Oliver, J., Prato, T.: Thermo-chemical–mechanical model for concrete. I: hydration aging. J. Eng. Mech. ASCE 125(9), 1018–1027 (1999) 16. Yang, L., Pathirage, M., Su, H., Alnaggar, M., Di Luzio, G., Cusatis, G.: Computational modeling of temperature and relative humidity effects on concrete expansion due to alkali– silica reaction. Cem. Concr. Compos. 124, 104237 (2021) 17. Zhang, Y., Di Luzio, G., Alnaggar, M.: Coupled multi-physics simulation of chloride diffusion in saturated and unsaturated concrete. Constr. Build. Mater. 292, 123394 (2021) 18. Bryant, A., Vadhanavikkit, C.: Creep, shrinkage-size, and age at loading effects. Mater. J. 84(2), 117–123 (1987) 19. Cusatis, G., Mencarelli, A., Pelessone, D., Baylot, J.: Lattice discrete particle model (LDPM) for failure behavior of concrete. ii: calibration and validation. Cem. Concr. Compos. 33(9), 891–905 (2011) 20. Di Luzio, G., Cusatis, G.: Hygro-thermo-chemical modeling of high-performance concrete. ii: numerical implementation, calibration, and validation. Cem. Concr. Compos. 31(5), 309–324 (2009) 21. Di Luzio, G., Cusatis, G.: Solidification–microprestress–microplane (SMM) theory for concrete at early age: theory, validation and application. Int. J. Solids Struct. 50(6), 957–975 (2013)
Building Information Modelling
Enhanced Interoperability between Geotechnical and Structural Engineering for 3D Building Models Haris Felic1(B) , Dirk Schlicke2 , Andreas-Nizar Granitzer1 , and Franz Tschuchnigg1 1 Institute of Soil Mechanics, Foundation Engineering, and Computational Geotechnics, Graz
University of Technology, Graz, Austria [email protected] 2 Institute of Structural Concrete, Graz University of Technology, Graz, Austria
Abstract. Conventional structural design in building construction is commonly based on the static analysis of the load flow using 2D submodels. Hence, the deformation compatibility and stress redistribution between the structural members of the building are not adequately regarded. In addition, the structural and the subsoil responses are usually assessed based on isolated analyses which use the subgrade reaction modulus as a coupling interface. Many researchers have shown that using the subgrade reaction method can lead to an uneconomic foundation design. Furthermore, the subsoil response generally affects the submodel of the foundation slab or basement, respectively. This leads to a strong simplification of the total stiffness of the building, and results either in conservative or unsafe results depending on the members under consideration. In contrast, the static analysis with a holistic 3D model allows for a more realistic load flow within the building. However, this requires not only an accurate representation of the stiffnesses of the structural members and their connections in the respective limit states, but also an accurate representation of the ground deformation in the calculation model. This contribution aims to enhance the interoperability between geotechnical and structural engineering for 3D building models. The contribution begins with a critical assessment of the subgrade reaction method. The core of this contribution proposes an alternative approach to account for soil-structure-interaction effects for 3D building models by exchanging nodal displacements, instead of using the subgrade reaction modulus. Keywords: holistic 3d calculation models · soil-structure-interaction · enhanced interoperability
1 Introduction In civil engineering, the Finite-Element-Method (FEM) represents a well-established numerical procedure to solve boundary value problems. Structural design and geotechnical investigations are hereby initially carried out in separate calculation models due to the different boundary conditions and model approaches for structures. Nevertheless, a © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 A. J˛edrzejewska et al. (Eds.): SynerCrete 2023, RILEM Bookseries 43, pp. 217–228, 2023. https://doi.org/10.1007/978-3-031-33211-1_20
218
H. Felic et al.
suitable implementation of the settlement trough from the geotechnical model into the structural model is then required. In structural engineering, the conventional procedure usually decomposes problems of interest into 2D submodels to reduce computational effort. Therefore, the load transfer (support loadings) between submodels is assessed manually. Using simplified 2D submodels, however, may produce wrong estimates of the subsoil response. The use of 3D calculation models is increasing due to developments in the runtime of commercial finite element codes, especially for the analysis of complex boundary value problems [1–3]. The analysis of 3D models allows for a stress redistribution between structural members and the deformation compatibility within the entire structure is naturally achieved. The design of large-scale constructions requires a careful consideration of the subsoil response to ensure an economic design that satisfies safety and serviceability requirements. In addition, the structural design depends on the subsoil response which differs for the serviceability limit state (SLS) and ultimate limit state (ULS). However, this is often ignored in engineering practice. Furthermore, the structural and the subsoil response are assessed based on isolated finite element analyses (FEA) in many cases. To couple these isolated analyses, the subgrade reaction method is commonly applied. Therein, the subgrade reaction method is usually computed based on an iterative procedure, see Fig. 1. The structural engineer provides the loadings acting on the foundation from the structural model to the geotechnical engineer. The latter applies the loadings on the foundation in the numerical model, obtains the foundation settlements, and calculates the subgrade reaction modulus based on the effective soil stresses between foundation and subsoil, and the calculated settlements. With the exchanged moduli, the structural engineer calculates again the acting foundation loadings. In theory, this loop is iteratively repeated until the same foundation settlements are obtained in both numerical models [4]. In some cases, it may be sufficient to use constant subgrade reaction moduli obtained by empirical correlations, or by static plate-loading tests, without an iterative procedure [5]. However, the use of the subgrade reaction method yields in inaccurate estimates of internal forces if wrong settlements are obtained [6]. Many researchers have shown that using the subgrade reaction modulus leads to uneconomic slab thicknesses and an overestimation of reinforcements, [7]. In addition, the superstructure considerations of constructions result in different soil-structure-interactions which leads to different soil stresses and subgrade reaction moduli [6]. This contribution aims to enhance the interoperability between geotechnical and structural engineering by exchanging nodal displacements, instead of using the subgrade reaction method for 3D building models, e.g. shown in [8]. The contribution begins with a critical assessment of the subgrade reaction method. The core of the contribution proposes an alternative approach to account for soil-structure-interaction effects for holistic 3D building models by exchanging nodal displacements. Therefore, two different finite element codes (PLAXIS [9], SOFiSTiK [10]) are used to demonstrate the proposed approach. The article concludes with recommendations for the enhanced interoperability between geotechnical and structural engineering for 3D building models. More details related to the FEA studies are presented in [11].
Enhanced Interoperability between Geotechnical and Structural Engineering
219
Fig. 1. Iterative procedure applied in the subgrade reaction method after [4].
2 Benefits and Limitations of the Subgrade Reaction Method The subgrade reaction modulus (kx , ky , kz ) is calculated as the effective soil stress (σx , σy , σz ) acting between the foundation and subsoil divided by the corresponding settlement (sx , sy , sz ) at the position of interest which is commonly written as: kx =
σx sx
ky =
σy sy
kz =
σz sz
(1)
Equation (1) shows that soil stresses and settlements in corresponding directions are considered proportional [12]. These quantities can be obtained from analytical solutions applying linear elastic soil models (e.g. Boussinesq Theory [13]), or can be assessed based on numerical models taking into account a more realistic soil response by means of non-linear constitutive models (e.g. Hardening-Soil model [14]). The latter approach includes accordingly the non-linear soil-structure-interaction effects which are governed by numerous aspects: the soil behaviour, the loading situation, or the foundation geometry. These aspects influence the distribution of the effective normal stresses, and settlements which in contrary also affect the subgrade reaction modulus. The interested reader is referred to [11]. Therefore, the subgrade reaction modulus should not be regarded as a soil parameter [12]. In addition to the multivariate dependency of the subgrade reaction modulus, the use of the modulus is limited to either a constant or variable subgrade modulus along the considered area (3D) or line (2D). Furthermore, subgrade reaction moduli are not fully
220
H. Felic et al.
capable to capture consolidation effects or the plastic deformation behaviour at higher loading stages [12]. A constant subgrade modulus might be suitable for a foundation with a low thickness, in combination with a high soil stiffness, and point loads. However, a constant modulus is less suitable for soft soils, line or surface loading, multi-layered soil strata, and for settlement intersections in the vicinity of constructions. Variable subgrade reaction moduli are recommended for flexible foundation slabs, simple soil layering, and variable loading conditions (surface or line loads). On the contrary, variable modulus are considered inappropriate for rigid foundation slabs, low soil stiffnesses, more complex soil layering (especially for sensitive settlement areas), and for settlement intersections in the vicinity of constructions [12]. The use of the subgrade reaction modulus contains limitations regarding the incorporation of time dependencies such as the generation of excess pore water pressure and consolidation (dissipation of excess pore water pressure over time) of the soil. These effects can become dominant in the structural design in case of fully saturated soil conditions subjected to rapidly applied loadings. Settlements increase along time due to the consolidation of soil which cannot be covered by the subgrade reaction modulus represented by linear elastic spring elements [15]. For higher loading conditions (near to the ultimate limit state), the plastic deformations of the soil body increase significantly. In this case, the settlements are tremendously increasing which results in a rapid decrease of the subgrade modulus [15]. The subgrade reaction modulus is valid for the explicit loading case. This means that strictly speaking no superposition is allowed between load combinations with the same subgrade reaction modulus. To this end, a more accurate alternative is presented in the next section which shows an engineering approach to exchange settlement troughs (3D) and aims to optimize the interoperability between geotechnical and structural engineering.
3 Engineering Approach to Exchange Settlement Troughs This section concerns the numerical analysis of a 3D multi-storey building representing a fictive shopping centre. Furthermore, an engineering approach to exchange settlement troughs (3D) from the FEM software PLAXIS to SOFiSTiK is presented. Contrary to the subgrade reaction method, the soil settlements obtained with the geotechnical FE code PLAXIS are used to prescribe nodal displacements in the structural model in SOFiSTiK. Hence, the nodal displacements in the structural engineering model (SOFiSTiK) are imposed independently from the loadings. 3.1 3D Multi-storey Building The rectangular shaped building consists of five stories, see Fig. 2. The loading combination represents the ULS (ultimate limit state) combination and load components (dead load, live load) according to [16–18]. All structural components are modelled as plate or beam elements with linear elastic (LE) material (equal to C 30/37 stiffness parameters, see [19]). For the sake of simplicity, the consideration of openings such as in walls or
Enhanced Interoperability between Geotechnical and Structural Engineering
221
slabs are ignored in the model. The building foundation lies on a soil volume which consists of four different layers varying from loose clay to dense sand. The Hardening Soil (HS) parameters are adopted from [11]. The width and length of the soil volume is doubled compared to the building geometry to enable an acceptable compromise of the calculation time. However, the compromise affects the displacements located at the corners and edges along the top surface of the soil volume (not shown within this contribution). Regarding the boundary conditions, all outer surfaces are fixed perpendicular, except for the bottom surface which is fully fixed (x-, y-, and z-direction) for the volume elements in PLAXIS. Since only ¼ of the structure is considered in both FEM-codes, the inner surfaces of the structure are perpendicular fixed. The bottom surface is fully fixed in SOFiSTiK. In addition, rotations of the plate edges (elements) are fixed in the out-of-plane direction in PLAXIS and SOFiSTiK; this procedure was considered automatically for the former code. For instance, the rotation of the plate edges in x-direction (see axes in Fig. 2) is fixed in x- and z-direction.
x
y z
x
y z
a) Overview with annotation in PLAXIS
b) Overview in SOFiSTiK
Fig. 2. Geometry overview and mesh discretization.
3.2 Exchanging Nodal Displacements PLAXIS and SOFiSTiK offer different finite elements to discretize the soil domain (volume domain); hence, soil displacements obtained with PLAXIS must be projected onto the SOFiSTIK mesh by employing an interpolation technique. In this case, the programming language Python with the SciPy module (function: interpolate) was used for interpolation. Herein, the interpolation is triangulated, and Bezier polynomial are used for each triangle. By using this procedure, the interpolated data is continuously differentiable, and the curvature of the surface is minimized. The use of this interpolation is recommended for unstructured data [20].
222
H. Felic et al.
Figure 3 illustrates the proposed approach to exchange displacements between FEMcodes. The left part of Fig. 3 illustrates the element nodes of each code whereas PLAXIS nodes contain the information of displacements uz (z-direction). It is important to highlight that the location of nodes is rearranged since both models are created within their global positive axes. The right part of Fig. 3 displays the projection of PLAXIS information onto the nodal position in SOFiSTiK. This procedure is done by using a cubic interpolation method, see Lis. 1. With respect to the element nodes position marked by red circles, the nearest neighbour interpolation (Lis. 2) is applied since a continuous triangulation is not possible for the given discretizations, see right half in Fig. 3. This method is executed for the settlements in x-, y-, and z-direction. The information is gathered and prepared as a text input file in SOFiSTiK, see Lis. 3.
1 2
from scipy import interpolate
3
grid_sofistik_uz = interpolate.griddata((x_plaxis, y_plaxis), uz, (x_sofistik, y_sofistik), method = ‘cubic’) Lis. 1. Python – cubic interpolation
1 2
from scipy import interpolate
3
grid_sofistik_uz = interpolate.griddata((x_plaxis, y_plaxis), uz, (x_sofistik, y_sofistik), method = ‘nearest) Lis. 2. Python – nearest neighbour interpolation
1 2 3 4
knot nr 11 typ wzz , wyy , wxx 1.13437682138610995253 e +02 ,0.0 e +00 ,0.0 e+00 knot nr 12 typ wzz , wyy , wxx 8.80664050933447981606 e +01 ,0.0 e+00 ,4.66655628615825004979 e−02 knot nr 13 typ wzz , wyy , wxx 9.11134914848572066148 e +01 , −4.05397214361099988977e− 01 ,5.64621565843699935172 e−01 ⋯ Lis. 3. Snapshot of input data in SOFiSTiK
Enhanced Interoperability between Geotechnical and Structural Engineering
a) Deformed PLAXIS mesh (uz) and position of SOFiSTiK nodes (uz = 0 m)
223
b) Interpolated settlement at SOFiSTiK nodes
Fig. 3. Interpolation of data.
Figure 4 compares the settlement troughs in z-direction between FEM-codes after the interpolation of nodal displacements in SOFiSTiK. Similar as for the interpolation procedure, the location of the results is rearranged since both models are created within the global positive axes. Thus, the location of SOFiSTiK results is mirrored in the global y-axis for all following figures. Apparently, there is no difference visible between settlements in the figure. This leads to the conclusion that the interpolation procedure is successful. Figure 5 shows the bending moments in x-direction of the foundation obtained with both FEM-codes. The results are in remarkable agreement, except for points located at the edge, corner, and one located within the minimum negative moment. These positions coincide with the red circled point in Fig. 5. For these positions, the nearest neighbour interpolation (instead of cubic interpolation) is applied, except for the one position with the minimum negative moment. Figure 6 presents a cross section along the foundation which compares the bending moment, shear force, and settlement of both numerical models. The results with the Hardening Soil (HS) material show a very good agreement excluding the peak internal forces. By observing the shear forces at the columns, the results indicate a less realistic “jump”. In addition, the location of the maximum shear force is not identical with the dmx peak bending moment dx = vx . A study revealed (not shown in this contribution) that if the mesh is refined around the columns, a more realistic shear force distribution could be achieved.
224
H. Felic et al.
a)
PLAXIS calculation (element nodes as grey nodes in the background)
b) SOFiSTiK calculation (with prescribed displacement, element nodes as grey nodes in the background)
Fig. 4. Comparison of settlement uz between FEM-codes.
Enhanced Interoperability between Geotechnical and Structural Engineering
a)
PLAXIS calculation (element nodes as grey nodes in the background)
b) SOFiSTiK calculation (with prescribed displacement, element nodes as grey nodes in the background)
Fig. 5. Comparison of bending moment mx between FEM-codes.
225
226
H. Felic et al.
Fig. 6. Comparison of FEM-codes results for cross section.
Enhanced Interoperability between Geotechnical and Structural Engineering
227
4 Conclusion The subgrade reaction method has considerable limitations; hence, the method should be applied with care. By using the presented method based on exchanging interpolated nodal displacements rather than exchanging subgrade reaction moduli, the former procedure shows promising results for holistic 3D models to capture a more accurate soil-structureinteraction. Yet, the interpolation contains minor errors for internal forces which most probably are solved with finer mesh discretizations. The process of exchanging and interpolating nodes between FEM-codes should be performed as long as both obtain equally distributed internal forces within the structures; thus, an iterative procedure is still recommended between the geotechnical and structural engineer. The exchanged nodal displacements correspond to an explicit load combination. However, the main benefit of exchanging nodal displacement is that the procedure becomes more independent from the load situation compared to the subgrade reaction modulus since the effective normal stress distribution between foundation and subsoil also depends on the applied load. More research is required whether the interpolation method with Bezier curves is the most suitable method. Regardless which exchange method is employed, the users (geotechnical and structural engineer) must pay attention on the modelling assumptions such as connection to columns, or walls in their models before the exchange.
References 1. Barth, C., Rustler, W.: Finite Elemente in der Baustatik-Praxis: Mit vielen Anwendungsbeispielen. Beuth Verlag GmbH (2013) 2. Granitzer, A., Tschuchnigg, F., Summerer, W., Galler, R., Stoxreiter, T.: Errichtung eines Eisenbahntunnels in Deckelbauweise über dem Hauptsammler West der Stadt Stuttgart/Construction of a railway tunnel above the main drainage tunnel of Stuttgart using the cut-and-cover method. Bauingenieur 96(05), 156–164 (2021). https://doi.org/10.37544/ 0005-6650-2021-05-40 3. Granitzer, A.-N., Felic, H.: Das Embedded Beam Element mit expliziter Interaktionsoberfläche - Optimierte Modellierung linearer Strukturelemente. In: 37. Baugrundtagung. Deutsche Gesellschaft für Geotechnik e.V. (DGGT), pp. 45–54 (2022) 4. Stopp, K.: Trag- und Verformungsverhalten großflächig gegründeter Stahlbetontragwerke unter Berücksichtigung der Boden-Bauwerk-Interaktion. Dissertation, Bergsischen Universität Wuppertal, Wuppertal, Germany (2010) 5. Boley, C. (ed.) Handbuch Geotechnik – Grundlagen - Anwendungen - Praxiserfahrungen, Praxis Vieweg + Teubner/Springer, Wiesbaden (2012) 6. Werkle, H., Slongo, L.: Modellierung des Baugrunds bei der Finite-Element-Berechnung von Bodenplatten. Bautechnik 95(9), 607–619 (2018). https://doi.org/10.1002/bate.201800041 7. Sherif, G., König, G.: Platten und Balken auf nachgiebigem Baugrund / Rafts and beams on compressible subsoil / Radiers et poutres sur sol de fondation compressible / Placas y vigas sobre terrenos compresibles. Springer Berlin Heidelberg, Berlin, Heidelberg (1975) 8. Schwarz, G.: Technisch und wirtschaftlicher Vergleich von Bodenplatten mit unterschiedlichen Gründungssystemen. Master’s Thesis, Graz University of Technology, Graz, Austria (2010) 9. Bentley Systems: PLAXIS Reference Manual – 3D - Connect Edition V21 Ausgabe (2021) 10. SOFiSTiK: Manual 2020 - Basics (2020)
228
H. Felic et al.
11. Felic, H.: Geostructural finite element analysis of soil-structure-interaction problems. Master’s Thesis, Graz University of Technology, Graz, Austria (2022) 12. Fischer, D.: Interaktion zwischen Baugrund und Bauwerk – Zulässige Setzungsdifferenzen sowie Beanspruchung von Bauwerk und Gründung, Schriftenreihe Geotechnik Heft 21. Kassel University Press, Kassel (2010) 13. Lang, H.-J., Huder, J., Amann, P., et al.: Bodenmechanik und Grundbau – Das Verhalten von Boden und Fels und die wichtigsten grundbaulichen Konzepte. Springer (2009) 14. Schanz, T., Vermeer, P.A., Bonnier, P.G.: The hardening soil model: formulation and verification. In: Brinkgreve, R.B.J. (ed.) Beyond 2000 in Computational Geotechnics, pp. 281–296. Routledge (2019) 15. Vogt, N.: Flachgründungen. In: Witt, K.J. (ed.) Grundbau-Taschenbuch. Wiley-VCH Verlag GmbH & Co. KGaA, pp. 1–78, Weinheim, Germany (2018) 16. ÖN EN 1991-1-1:2011 - Eurocode 0 - Grundlagen der Tragwerksplanung (konsolidierte Fassung), 2013-03-15 17. ÖN EN 1991-1-1:2011 - Eurocode 1: Einwirkungen auf Tragwerke, Teil 1-1: Allgemeine Einwirkungen - Wichten, Eigengewicht und Nutzlasten im Hochbau (konsolidierte Fassung), 2011-09-01 18. ÖN B 1991-1-1:2020 - Eurocode 1: Einwirkungen auf Tragwerke, Teil 1-1: Allgemeine Einwirkungen - Wichten, Eigengewicht und Nutzlasten im Hochbau (Nationaler Anhang), 2020-12-01 19. ÖN EN 1992-1-1:2015 - Eurocode 2: Bemessung und Konstruktion von Stahlbeton- und Spannbetontragwerken Teil 1-1: Allgemeine Bemessungsregeln und Regeln für den Hochbau, 2015-02-15 20. Virtanen, P., et al.: SciPy 10: fundamental algorithms for scientific computing in Python. Nat. Methods 17(3), 261–272 (2020. https://doi.org/10.1038/s41592-019-0686-2
Industry 4.0 Enabled Modular Precast Concrete Components: A Case Study Simon Kosse1(B)
, Patrick Forman2 , Jan Stindt2 , Jannik Hoppe2 Markus König1 , and Peter Mark2
,
1 Chair of Computing in Engineering, Ruhr University Bochum, Bochum, Germany
[email protected] 2 Institute of Concrete Structures, Ruhr University Bochum, Bochum, Germany
Abstract. The construction industry faces the challenge of building sustainable and ever faster. The modular construction method with serial precast concrete modules is suitable to achieve this. Here, entire load-bearing structures are segmented into identical or similar modules, which are prefabricated and merely assembled on the construction site. However, the production of precast concrete components has so far by no means been an automated process, as there are hardly any repetition rates. Therefore, Industry 4.0 (I4.0) methods are to be transferred to the prefabrication process, targeted here in line with serial production. The aim is to mass-produce precast concrete modules with high precision and quality assurance in modern production systems based on the I4.0 concept. I4.0 refers to automation through continuous digitalization and networking of production. In the sense of I4.0, smart products seek an optimal path through production systems that interact with machines and processes self-controlled and self-managed. Thereby, the digital twin as a virtual representation for capturing and providing all relevant data is a key component. As a first step towards a holistic digital representation, a high-performance precast concrete module is presented in this work as both a digital twin and a real demonstrator. This Y-shaped module is part of a wall-like honeycomb structure. It is produced using rapid heat treatment and monitored by geometrical and thermal sensors during production and afterwards. The Asset Administration Shell (AAS) as the technical implementation of the digital twin in I4.0 is used to provide suitable methods for communication and interaction. Keywords: Industry 4.0 · Digital Twin · Modular Construction · Precast Concrete Components · Automation
1 Introduction Precast concrete construction is generally considered a suitable approach to meet the growing demand for housing and the main issues of the construction industry, such as long construction times and high costs. Prefabrication, however, implies by no means that the precast modules are manufactured in series with large quantities and a high degree of automation. Instead, they are individual products manufactured in a handcrafted process. Automation is usually only implemented for a few sub-processes. In order to fully © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 A. J˛edrzejewska et al. (Eds.): SynerCrete 2023, RILEM Bookseries 43, pp. 229–240, 2023. https://doi.org/10.1007/978-3-031-33211-1_21
230
S. Kosse et al.
exploit the advantages of prefabrication, precast concrete elements must be produced in series on the one hand and in digitized and automated manufacturing processes on the other. The introduction of Industry 4.0 (I4.0) in the manufacturing industry has demonstrated how end-to-end digitization and interconnection can establish the basis for automated production environments [1]. Applying I4.0 to the industrialized production of precast concrete modules has the potential to increase production efficiency in the construction industry sustainably. The additional requirements placed on the modules by serial production must already be considered in design and construction regarding the subsequent production process. This paper presents an approach to the industrialized series production of precast concrete modules according to the I4.0 model for modular construction. The additional requirements resulting from the serial production and assembly of the modules are specified in the context of design and construction. Furthermore, a concept for a digital twin based on the Asset Administration Shell (AAS) known from I4.0 is proposed. Within the framework of a case study, a prototype of a precast concrete module is presented that meets the requirements for serial production and has a digital twin in the form of an AAS enabling production in a cyber-physical production system (CPPS). The application of the digital twin is exemplarily demonstrated by using heat treatment to rapidly cure a precast concrete module.
2 State of the Art 2.1 Precast Concrete Components and Modular Construction Precast concrete components are manufactured under regulated conditions and thus usually have a better quality than concrete components built on the construction site [2]. This is ensured by, for example, geometrically stable and reusable steel formworks. The individuality of buildings, in contrast, is reflected in a low repetition rate of the concrete components used, which then avoids multiple uses of formworks. However, in industrial buildings that almost consist of the same beam-, plate-, and column elements, the reinforcement ratio or different installation parts for connections also lead to individual design and consequently individual manufacturing. But, manufacturing is most effective when the same element is mass-produced in an automated manner. When automation is incorporated, precast concrete component construction offers the potential to improve sustainability (i.e., environmental, economic, and social) as well as all processes (i.e., design, production, and construction) [3]. To achieve this, a modularization of structures into equal or at least similar modules is reasonable since modular construction aims to industrialize the prefabrication of building elements [4]. This enables mass production techniques up to mass customization when off-site construction is utilized. Hereby, off-site construction refers to partial structures with already installed (non-)structural elements, for example, walls with windows and electric wiring, or whole room modules [5]. However, the benefits of both, modular construction, and off-site construction, are manifold. Since the production is moved from on- to off-site, the production, as well as the construction process, is mainly independent of weather conditions. Building elements can be prefabricated using serial production methods to ensure quality, avoid unnecessary waste, and accelerate – not only – the fabrication process. In addition, the
Industry 4.0 Enabled Modular Precast Concrete Components
231
construction on-site is reduced to “assembly only”. This means that the design of structures must emerge from both, manufacturing and assembly. In architecture, engineering, and construction (AEC) therefore Design for Manufacturing and Assembly (DfMA) [6] originating from the production industry has become an approach to avoid high costs, low productivity, and long delivery time. DfMA follows principles such as minimizing the number of building elements or joints, respectively, choose optimal material, optimize components, and streamline processes. 2.2 Industry 4.0 and Digital Twin Industry 4.0 (I4.0) refers to the digital transformation of the manufacturing industry, enabled in particular by advances in information and communication technology, leading to a convergence of the physical and virtual worlds [1, 7]. In I4.0, production environments form cyber-physical production systems (CPPS) whose participants are interconnected with processes and other participants to exchange data and information to achieve individual goals. Smart products are uniquely identifiable, can be located in the production system at any time, and have information about their history, current state, and next steps to reach their target state. I4.0 enables the integration of individual customer requirements and a lot-size-one production while maintaining the economic conditions of mass production. In addition, last-minute changes, flexible reactions to malfunctions and errors as well as transparency about the manufacturing processes allow for optimized decision-making [1, 8]. The evaluation and analysis of production data in real-time are one of the key concepts of I4.0 in which the digital twin plays a central role [9]. Since its introduction, the digital twin has become a key element in the digital transformation of many industries. As a result, various definitions of the digital twin emerged reflecting industry-specific requirements. In general, a digital twin is a formal virtual representation of an asset, process, or system that captures the entity’s attributes, properties, and behaviours. It provides communication, data management, interpretation, and processing methods in a context-specific frame [10]. The digital twin is often synonymously used in the literature with concepts such as digital model or digital shadow. In their work, Kritzinger et al. [11] delineate the digital twin by its vertical integration. They define the concept by three main components: a physical asset, a virtual representation, and a bidirectional data link that connects the two [11]. In the construction industry, a lack of consensus prevails, particularly regarding the role of BIM in the context of digital twins. The digital twin is understood here either as an extension, complementary, or completely independent of BIM [12]. Delgado et al. [13] see BIM and the digital twin as a response to different industry requirements. According to Boje et al. [14], BIM provides methods, technologies, and data schemas to enable a standardized semantic representation of building structures, while the digital twin provides a more holistic socio-technical and process-oriented representation. Furthermore, BIM lacks methods for integrating dynamic data and their semantic description, which are essential for control systems of any kind.
232
S. Kosse et al.
2.3 The Asset Administration Shell In Industry 4.0, the AAS provides a framework as an industry-neutral standard for implementing modular digital twins [15]. The AAS consists of several submodels, each depicting a closed view of an aspect or use case that combines to form an overall digital representation. Submodels contain all relevant data in the form of static and dynamic attributes and properties stored as literal values or as references to external files such as PDF, IFC, or STEP. For data exchange within a cyber-physical production system, it is of essential importance that the meaning of the data is unambiguously interpretable by both humans and machines. For this purpose, the data stored in the submodels are given a semantic identifier and are unambiguously semantically described by concept descriptions (Fig. 1). The AAS is based on a UML data model for which XML, JSON, and RDF serializations exist. Furthermore, an AAS package container is available, which, in addition to a serialization, contains all other files referenced in the AAS. AASs are divided into three different types. Type 1 is the passive AAS in the form of a file that is used for interoperable data exchange along the value chain. Type 2 is a reactive AAS in the form of a server application whose data content can be accessed via standardized interfaces, e.g., HTTP/REST, OPC UA, or MQTT. Type 3 refers to an active AAS that can initiate communication with other entities of the CPS in addition to type 2 and can thus interact with its environment. A special Industry 4.0 language is currently being developed for active communication [16]. For further information on the AAS please refer to the official documentation [15].
Fig. 1. Concept of the AAS based on [15].
3 Concept The generalized concepts for serial production of precast concrete modules with requirements originating from rapid production and modular construction, as well as for the digital twin are derived being the basis for the case study. 3.1 Serial Production of Precast Concrete Modules The production of precast concrete components can be coarsely divided in the subsequent process steps: shuttering, insert reinforcement, concreting, early strength hardening, and
Industry 4.0 Enabled Modular Precast Concrete Components
233
stripping as well as storing until the strength for assembly is reached. However, serial production of modular concrete components aims for a not only precise but rapid manufacturing. This contrasts with the different production times of the individual process steps. While concreting, for example, takes durations in the range of minutes, the hardening process until stripping exhibits several hours up to days; achieving a final strength sufficient for installation even weeks. Thus, a rapid heat treatment of 80 °C directly after concreting – and therefore non-compliant with standardized heat treatments acc. to [17] – is assumed here as investigated in [18], which only takes about 2–6 h. Thereby, high-performance concrete is used that on the one hand is suitable for these high heating rates and on the other hand exhibits a high strength after heat treatment almost suitable for installation. Since the hardening process still takes several hours, this discrepancy in the time sequence must be adjusted by paralleling or scaling the individual process steps. However, the storage of components can be omitted. Next to accelerated strength development, residual shrinkage deformations are pronouncedly reduced by heat treatment [18] so that time-dependent deformations of the module, except for load-induced creeping, are minimized. This ensures that no undesired shape deviations, which exceed the dimensional tolerances, prevent subsequent assembly. However, resulting tolerances of the individual process steps, e.g., depending on formwork material, manual or automated accuracy, and measurement technique, must comply with the tolerances of the joints and consequently of the building structure [19]. Therefore, all process steps must be controlled by means of quality control so that deviations in the geometrical, thermal, or mechanical parameters of the module or processes, respectively, can be used to adjust the production (or the design). In conclusion, the process steps of serial production of precast concrete modules can be summarized as shuttering, insert reinforcement, concreting, rapid heat treatment, stripping, and (continuous) quality control. However, generalized requirements, e.g., for the dimensional accuracy of modular concrete components cannot be drawn, since they strongly depend on their individual structure, modularization method, and joints, as well as the used materials and production. This underlines the necessity of the holistic digital representation as a digital twin, which is developed conceptually in the following section. 3.2 The Digital Twin of Precast Concrete Modules For precast concrete modules to be produced in a production process based on the I4.0 model, a suitable implementation of a digital twin is required that virtually mirrors the modules and collects, evaluates, and provides all relevant data for the production processes (cp. Fig. 2). In the construction industry, modular digital twin concepts have become established, which can be adapted to the requirements of the individual life phases due to their exchangeability, adaptability, and expandability. The respective modules are globally uniquely identifiable so that data can be referenced and aggregated. The data is accessed via standardized interfaces. The AAS offers an ideal platform for developing digital twins for the industrialized production of precast concrete modules. Due to the data and information structured in submodels, each of which offers a self-contained view of an aspect or use case, the AAS enables a modular and flexible implementation. The digital building models created
234
S. Kosse et al.
as part of the Building Information Modeling (BIM) process, which include the logical building structure and other semantic properties and attributes, contain a subset of the data required to implement digital twins for the industrialized production of precast concrete modules. Integration of data from IFC-based BIM models and other data sources enables a continuous flow of information along the lifecycles and enables rapid deployment of digital twins (cp. Fig. 2) [20]. To build digital twins based on the AAS, specific submodels, and the required data structures must be developed for each aspect or use case of an asset. Standardization of the AAS’s data structure is advisable to ensure exchangeability along the value chain and life phases. The Standardization Council Industrie 4.0 (SCI4.0) is currently, among other things, driving forward the standardization of certain submodels of the AAS. Standardized submodels exist for aspects, such as Identification, Technical data, and Documentation, which provide a basis for developing AASs. Submodels can be composed into AAS templates enabling rapid deployment by connecting, integrating, and evaluating various data sources. In previous work, an AAS template for the industrialized production of precast concrete modules has been created [21]. The template consists of the already standardized submodels as well as additional submodels for Detailed Design, Production, Quality Control, and Requirements (cp. [21, 22]).
Fig. 2. The AAS as the implementation of the digital twin along the value chain.
4 Application Within the scope of a case study, a Y-shaped precast concrete element has been designed that meets the requirements of serial production and is virtually represented by a digital twin in the form of the AAS. Following DfMA and modular construction principles, the module is derived from a honeycomb-like structure minimizing material demands while ensuring high stiffness. It is produced in a rapid, serial production-ready process with integrated quality control. Production relevant data are stored in the AAS and can be used for further communication and interaction in the production.
Industry 4.0 Enabled Modular Precast Concrete Components
235
4.1 Design The Y-shaped module exhibits arm lengths of 1 m. Its cross section is rectangular with rounded edges (radius of 2.215 cm), which results in a module’s height of 4.43 cm defined by the formwork made from steel profiles. The width tapers from 11 cm at the end of the arms to 14 cm in the module’s center. Due to the intended heat treatment, a high-performance concrete based on the binder Nanodur Compound 5941 is used [18]. However, for a standardized design it can be characterized to a C100/115. Two conventional steel bars type B500 with a diameter of 10 mm per module’s arm serve as reinforcement. Thereby, a mean reinforcement ratio of 3.1% results. The minimum concrete cover is 0.5 cm only, which is sufficient due to the dense concrete matrix. The bearing capacity is described by the resistance moment to normal force ratio given by MRd -NRd -design chart, which is derived with respect to the geometry and materials used according to [23]. The conceptual honeycomb structure, the module’s geometry, and an exemplary load-bearing capacity for the center of an arm are depicted in Fig. 3.
Fig. 3. Design of a Y-shaped module for an exemplary honeycomb wall-like structure.
For assembly, the modules exhibit steel plates with holes of 3 cm in diameter for screw connections at the end of their arms, which only transmit forces, not bending moments. The hole clearance for assembly is set to 2 mm since it has been proven to be sufficient for modules with an arm length of 2 m and bolted connections with prealignment of the modules [24]. 4.2 Digital Twin Representation of the Y-Shaped Precast Concrete Module The digital twin representation of the Y-shaped precast concrete module shown in Fig. 3 has been created based on the template for AASs presented in [21]. As data basis serves
236
S. Kosse et al.
an IFC-based BIM that, in addition to the geometric representation of the component, contains all data and information about material and reinforcement. The requirements placed on the component regarding production and quality have been compiled in a Requirements Interchange Format (ReqIF) file, including mapping to the corresponding data in the AAS (cp. [22]). Figure 4 shows screenshots of the resulting AAS of the Y-module.
Fig. 4. AAS of Y-shaped module.
4.3 Production The module is produced in the laboratory KIBKON of the Ruhr University Bochum. The formwork was built up using steel elements consisting of a bottom plate with a thickness of 2 cm on which L-profiles L 50 × 50, t = 4 [mm] with attached semi-circular profiles (Ø = 44.8, t = 2 [mm]) are connected. Thereby, the roundish shape of the module’s arms is ensured. Subsequently the reinforcement as well as the steel connectors were placed and fixed in their position. In addition, in the core of every arm two temperature sensors (edge and center) and in the module’s center are placed for temperature measurement. Also, two strain gauges per arm are attached at the reinforcement bars, also in their center. Figure 5 shows the positioning of the described sensors and of additional deformation sensors at every edge of the arms to control the axial as well as the vertical deviations. For selected sensors exemplary measurement data are depicted. The temperature sensors can be used to monitor the heat treatment. The strain gauges give information about the shrinkage strains, additionally to temperature induced ones. The deformations are critical for the fit accuracy of assembly, i.e., the axial deformations must be within the hole clearance of the connections, whereby the vertical deformations describe unintended “dishing”
Industry 4.0 Enabled Modular Precast Concrete Components
237
effects, likely resulting from uneven temperature distributions during heat treatment, shrinkage, residual stresses or friction between the module and the adjoining formwork. Directly after concreting, a heat treatment of 80 °C for 4 h is conducted using at total of 6 infrared radiators, one each on the upper and the lower side of an arm. The radiators themselves exhibit a temperature range of >100 °C and an electrical power of 3000 W each. The sensor for controlling the temperature was placed in the center, on the top of the module. After heat treatment and cooling of the formwork, the module was stripped and stored under the environmental conditions of the laboratory for further monitoring. To clearly identify the module and access data within the digital twin, a QR code is attached to its surface.
Fig. 5. Sensors of core temperature, strain, as well as deformation and their position in and at the module with exemplary measuring data over time.
The formwork before concreting as well as the concrete module itself has been threedimensionally measured using a 3D scanner type ATOS Compact Scan 8M. Additionally, the strength was evaluated by means of a rebound hammer, which must be adjusted for the used concrete and the heat treatment. Nevertheless, the evaluation of the measurements and their effect on the load-bearing capacity defined by the M-N ratio is still ongoing. In this case study, real-time temperature, and strain monitoring during rapid heat treatment of the module has been implemented. For this purpose, the measured sensor data is transmitted to the AAS, which captures the data in a machine-readable form and makes it available for further applications. For the visualization of the sensor data, an online dashboard based on the open-source visualization platform Eclipse Streamsheets has been developed [25]. Streamsheets allow and analysis of IoT data in an online spreadsheet. The platform supports standard communication protocols such as REST, MQTT, or OPC UA, enabling combination with the AAS. In this study, the HTTP client function of Streamsheets was used to send configurable HTTP requests to an URL and for passing the response as input to Streamsheets. The data collected in the AAS can be
238
S. Kosse et al.
queried and processed regularly via HTTP requests. Figure 6 shows the visualization of the temperature data.
Fig. 6. Plot of the temperature sensors data according to sensors shown in Fig. 5.
4.4 Results The core temperature during heat treatment exhibits values over 90 °C, which can be led back to two effects. First, due to central position of the sensor, the arms are partly exposed to temperatures exceeding the targeted value of 80 °C. Second, heat development from concrete hydration during accelerated curing superimposes. A steering of the heat treatment by means of sensors on the outside and core, controlled by process parameters provided by the digital twin, is adopted in further investigations in order to avoid too high temperatures. The qualitatively course of the strains within the module converges towards approx. 1 mm/m after about a week. Hence, only minor material-dependent strains are to be expected when the module is assembled. Otherwise, the residual deformation must be considered in the design. However, the strains presented are raw data, which still need to be investigated with respect to their material, thermal, and non-identifiable amounts for a better distinction. For the sake of automation, the quality control AAS must be enhanced to directly distinguish and evaluate between the several amounts of strains. The deformation sensors show values for axial and vertical deformations in the same range of dry > 2 days H2 O > 4 days H2 O, which was expected because of the larger scattering cross section of the protium isotope as compared to deuterium. This means that the DFI contrast is proportional to the amount of H2 O in the pore network. The observation that the DFI value for the sample stored in dry condition is lower than the one stored in D2 O, suggests that a certain amount of H2 O remained in the pore network, even if the sample was not immersed in water prior to the measurement. When comparing pairs of samples, stored under the same conditions, with and without the addition of nanoparticles, minor differences are observed. A relative decrease of the DFI value in admixed samples is observed when the overall amount of water is decreased (from dry conditions to storage in H2 O for 4 days). For the samples stored in D2 O, a lower DFI value is observed for the admixed sample. In general, no significant changes in slope with correlation length were observed. Such a change would be expected in the presence of a pore refinement mechanism, associated with a change in nucleation mechanism, which would explain the observed permeability reduction when the Me-S-H nanocomposite is added to the mix.
Assessing Cement Matrix Permeability by Neutron
353
Fig. 6. DFI vs. auto-correlation length plots for paste samples stored in different conditions, in the absence and in the presence of the Me-S-H nanocomposite admixture
One possibility is that adequate auto-correlation length ranges were not probed during the set of experiments performed. To better assess the effect of the Me-S-H nanocomposite on the pore network topology at small scale, we plan to perform further experiments, in which different autocorrelation length values and the effect of different curing regimes will be investigated. Furthermore, we plan to carry out a more quantitative analysis of the experimental data, by modelling the correlation function G (Eq. 1) according to different structural configurations. In conclusion, dark field imaging based on neutron grating interferometry may represent a promising approach to the study of fine details of the pore network topology in cement matrices. In this study, we performed some preliminary experiments by which we could fine-tune some details of sample preparation and experimental setup. Further experimenting and modelling approach to data analysis will be necessary to extract more accurate information, aimed at reconciling the small-scale properties of the pore network with macroscopic properties that control the durability of cement-based materials.
References 1. Artioli, G., Ferrari, G., Dalconi, M.C., Valentini, L.: Nanoseeds as modifiers of the cement hydration kinetics. In: Liew, M.S., Nguyen-Tri, P., Nguyen, T.A., Kakooei, S. (eds.) Smart Nanoconcretes and Cement-Based Materials, pp. 257–269. Elsevier (2020) 2. Valentini, L., Ferrari, G., Russo, V., et al.: Use of nanocomposites as permeability reducing admixtures. J. Am. Ceram. Soc. 101, 4275–4284 (2018) 3. Dal Sasso, G., Dalconi, M.C., Ferrari, G., et al.: An atomistic model describing the structure and morphology of Cu-doped C-S-H hardening accelerator nanoparticles. Nanomaterials 12, 342 (2022)
354
L. Valentini et al.
4. EN 12390-8: Testing hardened concrete – part 8: depth of penetration of water under pressure (2019) 5. Artioli, G., Bravo, A., Broggio, S., et al.: Low-carbon durable concrete for the new Genoa’s San Giorgio bridge. ACI Spec. Publ. SP-355, 1–10 (2022) 6. Moradllo, M.K., Hu, Q., Ley, M.T.: Using X-ray imaging to investigate in-situ ion diffusion in cementitious materials. Constr. Build. Mater. 136, 88–98 (2017) 7. Strobl, M., Grunzweig, C., Hilger, A., et al.: Neutron dark-field tomography. Phys. Rev. Lett. 101, 123902 (2008) 8. Morgano, M., Peetermans, S., Lehmann, E.H., et al.: Neutron imaging options at the BOA beamline at Paul Scherrer Institut. Nuclear Instrum. Methods Phys. Res. Sect. A: Accel. Spectrom. Detect. Assoc. Equip. 754, 46–56 (2014) 9. Kim, Y., Valsecchi, J., Kim, J., et al.: Symmetric Talbot-Lau neutron grating interferometry and incoherent scattering correction for quantitative dark-field imaging. Sci. Rep. 9, 18973 (2019) 10. Strobl, M.: General solution for quantitative dark-field contrast imaging with grating interferometers. Sci. Rep. 4, 7243 (2014)
Dam Concrete in Situ Creep Tests. Experimental Setup and Results from Six Large Concrete Dams Carlos Serra , João Conde Silva(B) , António Lopes Batista , and Nuno Monteiro Azevedo National Laboratory for Civil Engineering, Lisbon, Portugal [email protected]
Abstract. This paper describes the main features of the experimental in situ setups used to estimate the concrete deformability from six large Portuguese dams, all of which were built in the last 25 years. These experimental setups are commonly referred to as creep cells and aim at determining the modulus of elasticity for different ages and estimating the creep curve, under the same thermo-hygrometric conditions as the dam. Creep cells are composed of specimens of concrete with embedded strainmeters, placed in the body of a concrete dam during its construction, with a hydraulic flat-jack underneath. During the elastic modulus and the creep tests, while the jack introduces a given compressive stress in the specimen, the embedded strainmeters measure the subsequent strains, instantly or over time, for the modulus of elasticity or creep assessment, respectively. In recent years, there have been several developments in the design and in the installation methodology of this type of experimental setup, namely in Alqueva, Baixo Sabor, Ribeiradio, Foz Tua, Daivões and Alto Tâmega dams. The article also presents a brief compilation of the most important creep cells results for these dams, which are being used as an input to numerically evaluate the structure’s behaviour over time. Keywords: Dam concrete · In situ experiments · Deformability tests · Creep cells · Creep function
1 Introduction The mass concrete is the main material used in the construction of concrete dams. However, the knowledge of its behavior is still incomplete, mainly with regard to the evolution of its deformability and strength over time. The determination of the concrete deformability is justified by its importance in the dam structural analysis, including the interpretation of the monitoring data, as part of the procedures for the dam safety control. The characterization of the deformability evolution of concrete from dams usually consists of determining the values of the modulus of elasticity at various ages and the estimation of the creep function of the dam and wet-screened concrete, which, together with © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 A. J˛edrzejewska et al. (Eds.): SynerCrete 2023, RILEM Bookseries 43, pp. 355–367, 2023. https://doi.org/10.1007/978-3-031-33211-1_32
356
C. Serra et al.
the characterization of the concrete strength carried out through the standard destructive tests, provide an evaluation of the mechanical properties of the material. The largest aggregates of the dam concrete complicate the logistics associated with the experimental work, which is the reason why the wet-screened concrete is often utilised for laboratory characterization of the concrete used to build dams. The wetscreening procedure consists in removing the larger aggregates of the dam concrete after mixing. Despite being acceptable for controlling the properties of the dam concrete during the construction stage, the wet-screened concrete is not adequate for providing reference values for the properties of the hardened structural dam concrete. A research program has been going on for around half century at the National Laboratory for Civil Engineering (LNEC), aiming at evaluating the instantaneous and delayed concrete deformability of numerous Portuguese dams. This program is based on in situ experimental tests, which have been carried out over a relatively large time span. This implied the installation of unconventional in situ experimental setups, named creep cells, while the dams were still under construction. This article presents a summary of the evolution of the experimental setups adopted for the creep cells in Alqueva dam, between 1999 and 2001 [1–3], Baixo Sabor dam, in 2012 [4, 5], Ribeiradio dam, in 2013 [6], Foz Tua dam, in 2014 [7], Daivões dam, in 2018 and 2019 [8], and Alto Tâmega dam, in 2021 and 2022. A general description of the creep cells is followed by the modifications that have been implemented at different stages. The goal of this paper is to study the influence of each type of creep cell setup. A comparison of the results obtained for each dam is also shown and some final remarks are drawn, with a focus on the advantages and disadvantages of each experimental setup and on the differences between the results obtained for each dam. Based on these results, a dam designer can have an estimate of the range of modulus of elasticity and creep strain for similar dam concretes, considering the conventional construction method.
2 Experimental Setup 2.1 General Aspects Creep cells are in situ experimental setups used for characterizing the concrete deformability. Each cell consists of a cylindrical concrete specimen casted within the dam body, in order to be subject to the same thermo-hygrometric conditions as the surrounding concrete, but isolated from its stress field. A hydraulic loading system injects oil into the flat-jack placed below the specimen, introducing a normal compressive stress into the specimen longitudinal axis, while the corresponding strain is measured through strainmeters embedded in the specimen. This system aims at determining the (instantaneous) elastic modulus and, ultimately, estimating the creep function. The cylindrical specimens are placed with its longitudinal axis in the vertical direction. Underneath the lower flat face of the specimen, a flat-jack is installed to allow the application of a uniform pressure over the entire base. The other end of the specimen (the upper flat face) is in direct contact with the dam body, which supports the specimen and reacts to the applied forces.
Dam Concrete in Situ Creep Tests
357
For each active cell, i.e. cell with a loading system, there is a corresponding correcting cell, i.e. identical cell without loading system, which is solely focused on measuring the free deformations. The loading system associated with each active cell consists of a closed circuit that regulates the pressure applied under the specimen through a steel flat-jack connected by a pipe to the external part of the loading system, located in an accessible niche inside the dam gallery, which comprises a cylinder with hydraulic oil and nitrogen under pressure. The pressure in the system is controlled by adjusting the oil and the nitrogen in the system. Besides allowing the introduction of relatively rapid pressure variations for carrying out the elastic modulus tests, this mechanism allows a constant pressure to be maintained over long periods of time by compensating pressure drops due to concrete creep and small oil leaks. The cells are either composed of dam structural concrete (largest specimens, with maximum aggregate size of 150 mm) or wet-screened concrete (small and medium sized specimens, with maximum aggregate size of either 38 or 75 mm). The dimension of the specimen is function of the maximum size of the concrete aggregates, according to Fig. 1. These criteria are generally in agreement with the recommendations from RILEM TC 107 [9]. Procedures and recommendations for installing creep cells and performing tests are further detailed in [10].
a) Ø480 mm×1500 mm
b) Ø450 mm×1350 mm
c) Ø300 mm×900 mm
for dam concrete
for dam concrete
for wet-screened
for wet-screened
concrete by sieves
concrete by sieve
#76 or #38
#38
d) Ø225 mm× 600 mm
Fig. 1. Dimensions of creep cells and corresponding maximum size of the aggregate (max )
The National Laboratory for Civil Engineering (LNEC) has designed and installed a considerable number of creep cells in Portuguese dams in the last 50 years. Within the last decade, the experimental setups were gradually modified in order to improve its functionality. The most significant changes implemented are related with the type of concrete mould and with the location of the system. The type of strainmeters and the oil piping material were also modified. These modifications are reported below. A summary
358
C. Serra et al.
of the information concerning the creep cells installed in the six dams addressed in the current document are presented in Table 1. Table 1. Features of creep cells designed and installed by LNEC Dam
Installation year
N.º cells
Moulds (material)
Location of specimen
Type of concrete
Number and type of strainmeters per specimen
Alqueva
1999–2001
13
Metallic (Brass)
Above gallery
Dam Wet-scr. #38
One 50 cm Carlson** One 25 cm Carlson**
Baixo Sabor
2012
3
EPS* w/ metallic straps
Above gallery
Dam Wet-scr #38 and #76
Three 50 cm Carlson** One 25 cm Carlson**
Ribeiradio
2013
2
EPS* w/ metallic straps
Above gallery
Dam Wet-scr #38 and #76
Three 50 cm Carlson** One 25 cm Carlson**
Foz Tua
2014
2
EPS* w/ metallic straps
Alongside gallery
Dam Wet-scr #38 and #76
One 50 cm Carlson** One 25 cm Carlson**
Daivões
2018–2019
3
EPS* w/ metallic straps
Above gallery
Dam Wet-scr. #38
One 50 cm Carlson** One 25 cm Carlson**
Alto Tâmega
2021–2022
3
EPS* w/ metallic casing
Above gallery
Dam Wet-scr. #38
One 50 cm VW*** One 25 cm VW***
* expanded polystyrene ** electrical resistance *** vibrating wire
2.2 Moulds Until 2002 only metallic moulds, made of brass, in the form of buckets, had been used for concreting the creep cells specimens. The vertical physical barrier between the specimen and the surrounding dam body was achieved through a double brass sheet, which creates an empty chamber around the specimen, as represented in Fig. 2a. Alqueva was the last dam where these moulds were used. In the following years, these metallic moulds were replaced by expanded polystyrene (EPS) moulds. This improvement was implemented
Dam Concrete in Situ Creep Tests
359
to remove any possible effect of the metallic mould on the stiffness of the specimen, to reduce the costs associated with the construction of the mould and to improve the transmission of the force applied to the specimen. This solution consists in the use of a hollow cylinder made of EPS, as represented in Fig. 2b. The moulds were wrapped with metal straps around their perimeter along the height (spaced at ca. 30 cm) to prevent the moulds radial cut from opening during concreting. Changing the mould implied an adjustment of the loading system. As opposed to the solution with metallic moulds, in which the lower part of the specimen was in direct contact with the base of the mould, the new solution included a set of rigid interface plates, below and above the flat-jack, enhancing the uniformity of the stress distribution on the adjacent concrete (Fig. 2b). Amongst others, this solution was adopted in the Baixo Sabor, Ribeiradio, Foz Tua, Daivões and Alto Tâmega dams. For the Alto Tâmega dam, an external metallic casing replaced the straps in order to improve the effectiveness of the EPS confinement during the casting of the specimens.
a) Metallic (brass) moulds
b) EPS moulds
Fig. 2. Scheme of the specimens used in the creep cells
2.3 Location of the System The creep cells installed in most dams are located two concrete layers above the gallery (2 m per layer) where the visible equipments are located, ensuring that the specimens are surrounded by the dam structural concrete (Fig. 3a). With the objective of replacing the flat-jack and auxiliary parts in case of leakage, the location of the creep cells specimens was moved to beside the gallery, near the external part of the loading system. This type of assemblage was solely used in Foz Tua dam and implied the installation of the entire system in the same concrete layer (Fig. 3b). This solution, which allowed the full access to the loading system, required the construction of a larger niche and allowed the reduction of the strainmeters cables and
360
C. Serra et al.
the oil piping. The fact that oil piping was no longer embedded, allowed the use of a flexible hydraulic pipe rather than a copper pipe for connecting the external part of the loading system to the flat-jack. Between the niche of the gallery and the base of each active cell, a channel was left intentionally open for the passage of the oil piping, allowing the connection between the flat-jack and the external part of the loading system. The flat-jacks and auxiliary parts are exposed to oxygen, hence required corrosion protection that was achieved through a metallization by thermal spray of zinc followed by painting. The change of location of the specimen to near the gallery was not adopted in other dams due to the relatively high influence of ambient temperature, thus it is not adequate for simulating the mass concrete conditions.
a) Above gallery
b) Alongside the gallery
Fig. 3. General scheme of the creep cells systems
2.4 Strainmeters Electrical resistance (Carlson) strainmeters were used in most creep cells, the only exception being the Alto Tâmega dam, in which vibrating wire strainmeters were utilised. The reason for changing the type of strainmeters is justified with the fact that the electrical resistance ones were discontinued. As for the strainmeters size, regardless of the type, for the largest specimens, a ca. 50 cm long device was used, whereas a ca. 25 cm strainmeter was utilized in the medium and small sized ones. In general, only one strainmeter per specimen was used. The only exceptions were the largest specimens in Baixo Sabor and Ribeiradio dams, in which three strainmeters were arranged radially and equally spaced within the cylindrical specimen circle and centered in height. The objective of embedding three strainmeters was to assess the uniformity of the pressure on the specimen. A high uniformity was observed.
Dam Concrete in Situ Creep Tests
361
2.5 Oil Piping The copper pipes, which connect the loading system devices located in the gallery to the flat-jack, are brittle and lack flexibility. The successful experience with the use of a flexible hydraulic hose to inject oil in the Foz Tua dam creep cells resulted in the adoption of a similar material at Alto Tâmega dam. This hose was avoided in previous dams because of the uncertainty concerning its durability when embedded in concrete for long periods of time. Meanwhile, after discussing the properties of this type of hoses with the supplier, a durable flexible hydraulic hose was selected to replace the copper pipe. This hose is easier to install and was a successful alteration.
3 Comparison of the Results from Different Dams The diversity of the concrete composition among the different dams hinder a clear interpretation regarding the relationship between the properties and dosages of the components and the mechanical properties of the hardened concrete for all situations (Table 2). The main differences are the type of aggregate used in Alqueva dam, the fly-ash percentage with respect to the binder content and the water-binder ratio of Daivões dam. Nonetheless, a comparison of the results is presented as it allows to have a general idea of the evolution of the properties over time. Generally, the concrete conditions of each creep cell are similar between them since all dams were built using the conventional construction method and the specimens are embedded inside the dam’s body. The exception being Foz Tua dam creep cells which are closer to the gallery and have higher temperature variations and, eventually, some drying. Despite that, each dam has its specific environmental conditions and had its Table 2. Dam concrete composition Dam
Maximum Type of Binder Water size aggregate Cement Fly-ash aggregate (mm)
Fine Coarse Wateraggregate aggregate binder ratio
(kg/m3 ) (kg/m3 ) (kg/m3 ) (kg/m3 )
(kg/m3 )
Alqueva
150
Green schists
160
40
117
600
1705
0.59
Baixo Sabor
150
Granite
110
110
130
553
1433
0.59
Ribeiradio 150
Granite
95
95
94
540
1600
0.50
Foz Tua
150
Granite
100
100
89
560
1620
0.44
Daivões and Alto Tâmega
150
Granite
180*
116
584
1685
0.64
* Blended cement with fly-ash
362
C. Serra et al.
specific construction scheduling that influence the creep development of the placed concrete. The use of creep cells in situ allows the testing of this type of concrete (with large aggregates) during the real curing conditions. It is known that the temperature dissipation inside dams is slow hence resulting in large periods of higher temperature conditions inside the dam, accelerating the property development at early ages. Figure 4 shows the results of the concrete modulus of elasticity in compression for five of the six dams addressed in this document, the exception being Alto Tâmega dam for which results are still not available. Figure 4 a) presents the results of in situ dam concrete modulus of elasticity yielding the variability of properties of each component and concrete composition, namely the deformability of the aggregates, the type and content of cement and fly ash and the different environmental in situ conditions. For example, the aggregates used in Alqueva dam were green schists, whilst for the other dams the aggregates were granites and the use of fly ash varied from 20% to 50% of the total binder (Table 2). Figure 4 b) shows the correspondent modulus of elasticity 80
In situ dam concrete - E(t) (GPa)
70 60 50 40 30 20
Alqueva Baixo Sabor Ribeiradio
10
Foz Tua Daivões
0 10
100
1000
10000
Age, t (days)
a) Dam concrete creep cells in situ
Laboratory wet-screened #38 concrete - E(t) (GPa)
80 70 60 50 40 30 20
Alqueva Baixo Sabor Ribeiradio
10
Foz Tua Daivões
0 10
100
1000
10000
Age, t (days)
b) Laboratory wet-screened concrete
Fig. 4. Results of modulus of elasticity over time for five dams
Dam Concrete in Situ Creep Tests
363
results obtained from wet-screened concrete specimens kept in controlled conditions, in laboratory. Generally, the in situ dam concrete modulus of elasticity results are higher than the in situ wet-screened concrete (Fig. 5 a)) and higher than in the laboratory wetscreened concrete (Fig. 5 b)). These results can be due to the higher dosage of aggregate in dam concrete and due to the influence of in situ higher temperatures inside the dam over time, particularly for younger ages. The use of green schists in Alqueva dam can explain the larger variability obtained in the modulus of elasticity results (Fig. 4 and Fig. 5), since the aggregate deformability is the main factor for the concrete deformability [11]. The large variability and the occasional unusually high modulus of elasticity results obtained for the Alqueva dam concrete, was actually one of the reasons that lead to the reformulation of the experimental setup, namely the increase of the size of the specimens, the change from metallic to EPS moulds and the use of rigid interface plates in the loading system. 60
50
In situ dam concrete - E(t) (GPa)
In situ dam concrete - E(t) (GPa)
60
40
30
20 Alqueva Baixo Sabor
10
50
40
30
Alqueva
20
Baixo Sabor
Ribeiradio
Ribeiradio
Foz Tua
Foz Tua
Daivões
0
Daivões
10 0
10 20 30 40 50 In situ wet-screened #38 concrete E(t) (GPa)
a) In situ dam concrete vs in situ wetscreened #38 concrete
60
10
20 30 40 50 60 Laboratory wet-screened #38 concrete - E(t) (GPa)
b) In situ dam concrete vs laboratory wetscreened #38 concrete
Fig. 5. Correlation between modulus of elasticity results for five dams
As previously described, the in situ experimental setup allows for the execution of creep tests in the same environmental conditions of the mass concrete. Figure 6 presents two sets of in situ total creep strain results, obtained in Foz Tua (Fig. 6 a)) and Daivões (Fig. 6 b)) dam concretes and the fit to the Bažant and Panula’s creep model (BaP) [12]. In Foz Tua dam, due to the proximity of the creep cells to the gallery, the measured total creep strains yield the influence of temperature variations over time (Fig. 6 a)). In Daivões dam, with creep cells two concrete lifts above the gallery, the development of creep strains has lower variability (Fig. 6 b)). Table 3 presents the parameters of Bažant and Panula’s creep model (BaP) obtained by fitting the model to experimental results of each dam concrete. The fitted parameters show some differences between dams, namely the low value of E 0 for Alqueva dam and the higher value of φ 1 for Foz Tua dam. Despite the good fit to the experimental creep strains in Alqueva dam, the higher values of fitted parameter E 0 results in lower modulus
364
C. Serra et al. 80
Modelo BaP: BaP model: E0 = 44,90 (GPa) 1 = 8,45 m = 0,67 n = 0,21 = 0,00
J(t,t0) (1×10-6/MPa)
60
J(t,t0) - Função de fluência preliminar J(t,t Creep function 0) – Experimental total creep strains – used for Extensão específica experimental - valores utilizados no ajuste fitting
40
Experimental specific experimental total creep strains Extensão específica
1/E(t Instantaneous unit strains 0) – 1/E(t0) - Evolução do módulo de elasticidade correspondente à função de fluência preliminar
20
t0 = 375 days
t0 = 125 days
0 1
10
100 Age (days) Idade (dias)
1000
10000
a) Foz Tua dam concrete 80
BaP model: Modelo BaP:
J(t,t0) - Função de fluência preliminar J(t,t Creep function 0) –
E0 = 58,4 (GPa) 1 = 4,80 m = 0,30 n = 0,54 = 0,01
Experimental specific experimental total creep strains Extensão específica - valores utilizados no ajuste
J(t,t0) (1×10-6/MPa)
60
1/E(t Instantaneous unit strains 1/E(t0) - Evolução do módulo de 0) – elasticidade correspondente à função de fluência preliminar
40
20
t0 =- 35 days CF2 t0=35 dias
t0 =- 96 daysdias CF1 t0=96
0 1
10
100 Age (days) Idade (dias)
1000
10000
b) Daivões dam concrete Fig. 6. Experimental total creep strains obtained from in situ creep cells and fit to BaP model
Dam Concrete in Situ Creep Tests
365
Table 3. Fitted parameters of BaP creep model for each dam concrete Type of aggregate
E 0 (GPa)
φ1
m
n
α
Alqueva
Green schist
38.0
3.9
0.37
0.12
0.05
Baixo Sabor
Granite
50.1
2.2
0.35
0.12
0.05
Ribeiradio
Granite
49.0
1.6
0.37
0.12
0.05
Foz Tua
Granite
44.9
8.5
0.67
0.21
0.00
Daivões
Granite
58.4
4.8
0.30
0.54
0.01
Dam
80.0
Alqueva - fit to exp. results Baixo Sabor - fit to exp. results
70.0
Ribeiradio - fit to exp. results Foz Tua - fit to exp. results Daivões - fit to exp. results
60.0
Alqueva - prediction
E(t) (GPa)
Baixo Sabor - prediction
50.0
Ribeiradio - prediction Foz Tua - prediction Daivões - prediction
40.0 30.0 20.0 10.0 0.0 10
100 Age (days)
1000
a) Modulus of elasticity over time 160.0
Alqueva - fit to exp. results Baixo Sabor - fit to exp. results Ribeiradio - fit to exp. results Foz Tua - fit to exp. results Daivões - fit to exp. results Alqueva - prediction Baixo Sabor - prediction Ribeiradio - prediction Foz Tua - prediction Daivões - prediction
140.0
J(t,t0) (1×10-6/MPa)
120.0 100.0 80.0 60.0 40.0 20.0
t0=28 days
t0=90 days
t0=365 days
t0=1825 days (prediction)
0.0 10
100
1000
10000
Age (days)
b) Creep function for different loading ages Fig. 7. Comparison of BaP models fitted to the experimental results obtained from dam concrete creep cells
366
C. Serra et al.
of elasticity predictions (Fig. 7 a)), which is not in agreement with the experimental results (Fig. 4 a)). The higher φ 1 for Foz Tua dam reflects the higher in situ creep strains at younger ages, compared with the other dams (Fig. 7 b)). This can be explained by the higher temperature variations and, eventually, to the effect of drying creep in this type of creep cell scheme (close to the gallery and with direct access to the flat-jack in the bottom of specimen).
4 Final Remarks This article describes the technological developments carried out in creep cells installed within the last two decades in several Portuguese large concrete dams, namely in Alqueva, Baixo Sabor, Ribeiradio, Foz Tua Daivões and Alto Tâmega dams. The main modifications implemented during this timeframe were the change of specimens’ moulds and the relocation of the specimen and surrounding equipment. The first one concerns the type of moulds used, namely the change from brass moulds (in Alqueva dam) to EPS moulds, with subsequent adjustments to the loading system. The second change concerns the adaptation of the system to make the flat-jacks accessible (in Foz Tua dam). Changing the type of moulds was successful, both in terms of associated costs and in terms of functionality and ease of installation. The results obtained in the creep cells where these moulds were used are consistent and compatible with the expected values for their respective concrete type, which validates the effectiveness of the changes. The change from non-accessible to accessible flat jacks in the Foz Tua dam, despite the undesired greater sensitivity to temperature variations inside the gallery, enables the repair or the replacement of any of the loading system parts in case of malfunction, e.g. leakage. Acknowledgements. Thanks are due to EDIA - Empresa de Desenvolvimento e Infraestruturas do Alqueva, S.A., EDP, Greenvouga, S.A., Movhera - Hidroelétricas do Norte, S.A and Engie Hidroelétricas do Douro, Lda, for the permission for publishing the results presented in this article.
References 1. Serra, C., Batista, A.L., Castro, A.T.: Determinação da função de fluência do betão da barragem de Alqueva. In: Encontro Nacional Betão Estrutural 2010, pp. 1–16. LNEC (2010) 2. Serra, C., Batista, A.L., Castro, A.T.: Caracterização do comportamento diferido do betão de barragens. Aplicação à barragem do Alqueva. In: 8º Congresso Nacional de Mecânica Experimental, pp. 273–274. LNEC (2010) 3. Serra, C., Batista, A.L., Castro, A.T.: Creep of dam concrete evaluated from laboratory and in situ tests. Strain 48(3), 241–255 (2012) 4. Serra, C., Batista, A.L., Gomes, P.G.: Upstream dam from hydroelectric exploitation of Baixo Sabor: creep cells installation and tests at young ages (in Portuguese), LNEC report 311/2014 – DBB/NO. LNEC. Lisbon (2014)
Dam Concrete in Situ Creep Tests
367
5. Serra, C., Batista, A.L., Monteiro Azevedo, N.: Dam and wet-screened concrete creep in compression: in situ experimental results and creep strains prediction using model B3 and composite models. Mater. Struct. 49(11), 4831–4851 (2016). https://doi.org/10.1617/s11527016-0828-2 6. Serra, C.: Ribeiradio dam: Creep cells installation and deformability tests (in Portuguese), LNEC report 396/2018 – DBB/NO. LNEC. Lisbon (2018) 7. Serra, C., Silva, J.C.: Foz Tua dam: Creep cells installation and concrete deformability tests (in Portuguese), LNEC report 344/2020 – DBB/NO, LNEC. Lisbon (2020) 8. Serra, C., Silva, J.C.: Daivões dam: Creep cells installation and concrete deformability tests (in Portuguese), LNEC report 229/2022 – DBB/NO, Lisbon (2022) 9. RILEM TC 107-CSP: Measurement of time-dependent strains of concrete. Mater. Struct. 31(8), 507–512 (1998) 10. LNEC: Procedures and recommendations for the execution of dam concrete deformability tests, in situ and in laboratory (in Portuguese). LNEC report 455/2013 – DBB/NO. LNEC. Lisbon (2013) 11. Alexander, M.: Aggregates and the deformation properties of concrete. ACI Mater. J. 93(6), 569–577 (1996) 12. Bažant, Z.P., Panula, L.: Practical prediction of time-dependent deformations of concrete. Matériaux et Constr. 11(6), 424–434 (1978)
Preliminary Analysis of Non-destructive Test Methods to Evaluate the Self-healing Efficiency on the Construction Site Tim Van Mullem1 , Gerlinde Lefever2 , Arthur Decuypere1 , Erik De Vleeschouwer1 , Yasmina Shields1 , Laurena De Brabandere1 Didier Snoeck3 , Dimitrios G. Aggelis2 , and Nele De Belie1(B)
,
1 Magnel-Vandepitte Laboratory, Department of Structural Engineering and Building Materials,
Faculty of Engineering and Architecture, Ghent University, Technologiepark Zwijnaarde 60, Campus Ardoyen, 9052 Gent, Belgium [email protected] 2 Department Mechanics of Materials and Constructions, Vrije Universiteit Brussel (VUB), Pleinlaan 2, 1050 Brussels, Belgium 3 Department of Building, Architecture and Town Planning (BATir), École Polytechnique de Bruxelles, Université Libre de Bruxelles, Av. F. Roosevelt 50, 1050 Brussels, Belgium
Abstract. In the last decades major advances have been made in the development of self-healing concrete which is able to heal its own cracks without the need for traditional repair interventions, thereby increasing its durability and service life. Recently, more and more self-healing technologies have been applied in demonstrator projects. These demonstrator projects have made it evident that we need to develop new testing methodologies to evaluate the self-healing performance, as many laboratory test methods cannot be applied on structural elements on the construction site. In the current study different non-destructive test methods have been used to analyse the self-healing performance of concrete beams. These beams were cracked in a three-point bending setup. Part of the beams had a cast-in vascular network allowing the crack to be healed via the injection of polyurethane. The healed beams were compared to the reference beams without the vascular network by applying different test methods: microscopy, concrete moisture content, resistivity, air permeability, water permeability, and ultrasound. Based on the results of these methods which were obtained under laboratory conditions, it was found that the concrete moisture content and the resistivity only provided limited value in terms of conclusions for self-healing. All test methods were also applied to concrete walls on site. Based on this last measuring campaign, recommendations are provided for quantification of the self-healing efficiency on the construction site. Keywords: Self-Healing Concrete · Vascular Networks · Non-Destructive Testing
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 A. J˛edrzejewska et al. (Eds.): SynerCrete 2023, RILEM Bookseries 43, pp. 368–379, 2023. https://doi.org/10.1007/978-3-031-33211-1_33
Preliminary Analysis of Non-destructive Test Methods
369
1 Introduction With the rise of the world’s population and globalisation there is a strong increase in our infrastructure and number of buildings. The maintenance of this growing number of concrete elements is becoming a heavy economical burden, especially since many elements need to remain operational after their service life. Tailoring the concrete with self-healing functions allows it to repair its own defects without the need for expensive manual repair operations. Cracks have a detrimental effect on the durability. When they are healed as soon as they are formed, the service life can be extended, thus coming to a more sustainable solution. Currently the research field is moving from laboratory studies to large-scale demonstrator elements. For example, in Belgium, already two demonstrators have been reported. The first one being a roof slab of an inspection pit cast with bacterial concrete [1], and the second one being several retaining walls containing superabsorbent polymers to reduce shrinkage cracking and subsequent healing in case cracks would form [2]. As such, the need has arisen to develop test methods which can be easily executed on site and which give a quick and reliable indication of the degree of self-healing. The aim of the current study was to investigate the effectiveness of several different low-cost test methods in quantifying self-healing. Several concrete beams were cast in the lab and part of them were healed with an embedded vascular network. The beams were analysed with different test methods which are commercially available and which are used for on site inspections (microscopic analysis, moisture content determination, air permeability measurements, water permeability measurements with Karsten tubes and resistivity analysis). In addition, the beams were also analysed with an ultrasonic test setup. With the experience gained from the measurements in the lab, the test methods were also applied on site on a demonstrator element.
2 Materials and Methods 2.1 Vascular Network The relationship of the diameters and branching angles of the vascular network was governed by Murray’s Law, which dictates that the cube of a parent vessel’s radius, r p , is equal to the sum of the cubes of the daughter vessels’ radii, r d (Eq. (1)). This relationship minimizes the power required for fluid transport, and has been previously used for designing networks in self-healing concrete [3, 4]. rp3 = rd3 1 + rd3 2
(1)
For this cube law in Eq. (1), Murray also determined the optimal branching angle to be 75° [5]. The layout of the network is given in Fig. 1. The spacing between the channels from centre to centre was 20 mm. As the beams would be loaded in 3-point bending, the smaller daughter branches were concentrated at the midspan of the beam, where the highest moment occurs. The total length of the smallest daughter branches was limited to the maximum build height that the 3D printer could achieve, namely 200 mm. Starting with the smallest branches, the minimum diameter that can be accurately 3D printed with a standard 0.4 mm nozzle is 3 mm; subsequent internal diameters for the larger branches are thus 4.33 mm and 5.45 mm.
370
T. Van Mullem et al.
Fig. 1. Vascular network configuration.
The wall thicknesses of all the parts are 1.2 mm, which is equal to three perimeter lines for a 0.4 mm nozzle. A ribbing pattern was introduced to promote bonding to the concrete, so the network is guaranteed to fracture upon crack formation to release the healing agent. Ribbing patterns were studied by De Nardi et al. [6], with recommendations to use a spiral ribbing pattern with a 5 mm pitch over an angular ribbing pattern. This was then applied to the current vascular network design (Fig. 2).
Fig. 2. Brittle channels with an included ribbing pattern that will release the healing agent.
The networks were 3D printed using an Ultimaker 2+ fused deposition modelling (FDM) printer, using a 0.4 mm nozzle, a layer height of 0.25 mm, a printing speed of 60 mm/sec, a nozzle temperature of 200 °C and a heated build plate temperature of 60 °C. Each network was printed vertically to the build plate to induce a weaker and more brittle cross-section. Silicone tubes were connected to the ends of the network so the healing agent could be pumped after cracking. These were secured to the 3D printed parts with self-fusing tape to allow for a watertight connection. This tape was also used to connect the individual channels to the branched part of the network (see Fig. 1). The network was set in place with aluminium wire on top of the rebars after its placement in the mould. After the loading regime, water was first injected into the networks to verify network rupture. Subsequently, one end of a network was pumped with a polyurethane healing agent (HA FLEX SLV AF, a 1-component polyurethane injection resin produced by GCP Applied Technologies, Belgium) using a syringe until it was seen coming out of the opposite end. Once the healing agent filled the whole system, a clamp was used to close off the outlet of the network so that an air pressure hose could pressurise the filled network for approximately 20 s or until sufficient fluid was coming out of the cracks. 2.2 Concrete Composition and Specimen Preparation In order to analyse the different test methods, beams were cast with a dimension of 460 × 120 × 120 mm3 and two reinforcement bars with a diameter of 6 mm (concrete cover
Preliminary Analysis of Non-destructive Test Methods
371
30 mm). Five of these specimens were cast with a vascular network, which were termed “VN”. The reference “REF” beams without the vascular network were cast using the same mix design, see Table 1. Note that one of the REF moulds failed, so that only four specimens remained. In order to have a larger database, the test methods were also applied on other reference specimens with the same dimensions but with a different mix design, termed ADD. Cracks were introduced in a three-point bending setup with a CMOD to monitor the crack opening. Subsequently, the different test methods were applied. The VN beams were then healed, see Sect. 2.1. After the polyurethane had cured, the tests were repeated to determine the healed measurements. The REF beams did not undergo healing, meaning that the two measurements are expected to be similar. Table 1. Mix design of laboratory specimens. ADD
REF & VN
CEM I 52.5 N
340.0
337.6
kg/m3
Water
170.0
185.2
L/m3
Sand 0–4 mm
851.0
792.9
kg/m3
Gravel 2–8 mm
430.0
1013.1
kg/m3
Gravel 8–16
566.0
–
kg/m3
Limestone filler
–
58.0
kg/m3
Superplasticizer MG51
1.1
–
kg/m3
Superplasticizer MG27
–
1.39
kg/m3
2.3 Microscopy After cracking, the crack width was determined with a fixed laboratory microscope (Leica DMC 2900). Six pictures were taken along the crack mouth. For each picture 6 measurements were done, resulting in 36 measurements per beam. For part of the ADD beams the crack width was also determined with a portable handheld microscope, see Fig. 3a. The lens magnifies the image with a factor 40 and has an optical measuring grid (division of 20 μm). Nine measurements were done per beam. The REF and VN beams were not measured with the portable microscope. 2.4 Moisture Content The device to determine the moisture content (TQC Concrete Moisture Meter, TQC, the Netherlands) consists out of eight electrodes at the bottom. By pressing the meter down against a surface a low frequency electric field is generated between the electrodes, making it possible to measure the electrical impedance. This is linked to the relative maximal physical water content in the concrete. It is recommended to repeat the measurement in
372
T. Van Mullem et al.
a few points close to each other to compensate for an uneven moisture distribution (in this study 6 repetition points were considered) and to retain the maximum value. The meter has several scale options. The scale used in this study was the “concrete 0–6% H2 O” scale, which considers the relation of the weight of pure water in the material with regards to the material’s dry weight. 2.5 Air Permeability The tester (PermeaTORR, Materials Advanced Services, Argentina) consists out of a two-chamber vacuum cell (concentric inner and outer chamber) and a vacuum pump. The tests start by placing the vacuum cell on the surface and vacuum is produced in both chambers. The pressure in the inner cell is recorded and once a threshold is reached vacuum is cut off from the inner chamber and only the outer chamber is kept in vacuum. At this stage air starts to flow back into the inner chamber through the concrete cover. From the measurements a coefficient of permeability K t can be calculated, as well as the affected depth. In normal operating conditions the air permeability tester is placed on an uncracked surface. However, in the current study the vacuum cell was placed directly on the crack. As such, it was often not possible to obtain a vacuum and determine Kt . Instead, pressure in the inner chamber was recorded 30 s after starting the equipment. 2.6 Water Permeability with Karsten Tubes This is a simple non-destructive method to test the degree of water penetration into concrete. The method is commonly used to test the water penetration in stone and plaster. This test was executed by placing a glass tube on the crack and sealing it with plasticine, after which the tube was filled with water. The Karsten tube allows to measure 4 mL of water penetration with an accuracy of 0.1 mL. While the water tightness is commonly ensured by applying plasticine, it was found that this was rather time-consuming, as it was often required to re-apply the plasticine as a result of leaks. In addition, residue is often left behind from the plasticine. As an alternative double sided vinyl tape was used which enabled a watertight connection without leaving behind residue, see Fig. 3d. 2.7 Resistivity The resistivity was measured with a Wenner probe with four probes (spacing of 50 mm, Resipod, Proceq, Switzerland), see Fig. 1e. Nine measurements were made along the width of the prisms, both on and next to the crack. A current is applied to the two outer probes and the potential difference is measured between the two inner probes. In concrete the current is carried by the liquid in the pores. The presence of reinforcement steel significantly influences the measurement, especially when the cover depth is smaller than 30 mm. A similar test equipment was recently used by Henry and Arnamwat [7] in an effort to quantify the self-healing of mortars with fly ash.
Preliminary Analysis of Non-destructive Test Methods
a)
b)
c)
d)
e)
f)
373
Fig. 3. a) handheld microscope, b) moisture meter, c) air permeability tester, d) Karsten tube, e) resistivity meter, and f) ultrasonic test set-up with one emitter and two receivers.
2.8 Ultrasound The equipment consisted of a wireless acquisition board (micro-SHM), two piezoelectric receivers of type PK15l and one R15α sensor, used as emitter (see Fig. 3f). The resonance frequency of both transducer types was 150 kHz. The emitter was connected to a waveform generator (33220A, Agilent), which sent a single cycle sine wave with amplitude of 5 V and frequency of 150 kHz. Concerning the experimental set-up, ultrasonic surface wave measurements were performed in the uncracked, cracked and healed state. In the uncracked situation, the receivers were positioned onto the concrete beams at a centre-to-centre distance of 5 cm. The emitter was placed outside of this measurement region, but in line with the receivers, and its position was altered from a centre-to-centre distance of 3 cm from the left and right receiver respectively. After cracking, ultrasonic tests were executed in two locations along the crack mouth. Here, the receivers were placed on opposite sides of the crack, while maintaining a distance of 5 cm between them. Similar to the uncracked assessment, the emitter’s position was changed from left to right. A repetition of these tests was performed after healing. A minimum of 5 measurements were conducted per location (2 locations, each 2 cm away from their respective side of the beam) in the cracked and healed state or per specimen in the uncracked situation in order to increase the reliability of the results.
374
T. Van Mullem et al.
From the received waveforms, the longitudinal wave velocity was calculated by dividing the travel distance (centre-to-centre distance between the receivers) by the time difference between the arrivals at the furthest and closest receiving sensor respectively.
3 Results and Discussion of Laboratory Tests Cracking of the beams resulted in 4 REF cracks and 6 VN cracks (one VN specimen had two cracks). 3.1 Microscopy For five of the ADD beams the crack width was measured both with a portable microscope and a fixed laboratory microscope. Table 2 shows the average crack width which was recorded. The average crack width measured with the portable microscope was only 71% of the one measured with the laboratory microscope. It is noted that for the portable microscope the average is calculated based on 9 measurements per beam, while for the laboratory microscope the average is calculated based on 36 measurements per beam. As such, the average of the laboratory microscope can be considered more reliable. Especially considering the fact that the pictures taken with the laboratory microscope can be analysed on a computer, while for the portable microscope the reading needs to be done immediately by repositioning the objective lens until the optical measuring grid coincides perfectly with the crack walls. Also, the optical measuring grid of the portable microscope only has an accuracy of 20 μm. Table 2. Comparison between crack width measurement with a portable microscope and a fixed laboratory microscope on five ADD beams. 1
2
3
4
5
Portable microscope [μm]
96
79
113
173
124
Fixed laboratory microscope [μm]
129
114
182
208
184
Portable/Fixed
74%
69%
62%
83%
67%
It can be reasoned that as crack widths increase, the relative error with a portable microscope will decrease. In addition, this error can be compensated in case cracks are measured before and after healing with the same microscope. As such the accuracy is sufficient for application on the construction site. Nevertheless, this remains a timeconsuming measurement. Efficiency can be increased when working with a microscope which allows taking digital pictures. 3.2 Moisture Content The moisture meter was used to determine the moisture content on the REF and VN beams. The measurements were done both on the crack and next to the crack. For the
Preliminary Analysis of Non-destructive Test Methods
375
beams only a measurement after healing is available. From Fig. 4 no clear trend can be discerned between the REF and VN beams. As such, this method does not seem suitable for quantifying self-healing.
Fig. 4. Moisture content versus crack width for REF and VN beams both on the crack and next to the crack.
3.3 Air Permeability The pressure in the inner chamber was recorded 30 s after starting the equipment. The difference of the nominal atmospheric pressure (1013 mbar) and this value is given in Fig. 5. For the VN beams the measurement was repeated after healing. The red vertical upwards arrow indicates for one of the VN specimens how the beneficial effect of the crack filling results in a larger pressure difference. The pressure difference after healing can be compared to the trend line to come to an equivalent healed crack width. For 3 beams the pressure difference was comparable to the nominal atmospheric pressure, indicating that the crack healing was perfect. Overall, this method seems to be able to quantify crack healing (crack filling) quite well. One attention point is the fact that local discrepancies at the location of the sealing rings can influence the results.
Fig. 5. Pressure difference versus crack width for REF and VN beams. Two sets of ADD prisms are included to come to a more reliable trend line. Red arrows indicate how pressure difference changes after healing and can be compared to the trend line to come to an equivalent healed crack width.
376
T. Van Mullem et al.
3.4 Water Permeability with Karsten Tubes Due to the limited volume of the tubes, it was not possible to record a flow for the REF and VN beams prior to healing; the crack width was too large resulting in a too fast water drop to do an accurate recording. After healing of the VN beams it was possible to record the flow rate, see Table 3. Since no recording could be done for the REF beams, the results are compared to recorded flow rates of five beams of the ADD series which had a smaller crack width than the REF beams. These show that the healing of the VN beams was effective. In fact, for several of the beams the flow rate was comparable to a flow rate of uncracked beams (± 0.013 mL/min). As such it can be concluded that the Karsten tube is an effective method to measure the crack healing, but for larger crack widths (>200–250 μm) it would be better to adopt a system with a larger reservoir like e.g. used by Van Tittelboom et al. [8]. Table 3. Flow rate q and crack width w of the healed VN beams compared to some specimens of the ADD series. VN (healed)
ADD
w [μm]
q [mL/min]
w [μm]
q [mL/min]
355
0.120
140
1.200
343
3.400
188
6.000
330
0.013
189
3.600
108
0.047
78
0.310
155
0.013
205
6.800
257
0.013
3.5 Resistivity The resistivity was measured before and after healing on nine points along the width of the beams. Here only the median of the three middle measurements were considered in order to disregard boundary effects. The electrical resistivity was relatively low (2.4– 6.4 k cm) which can be explained by the high moisture content and the young age of the concrete (measurement after healing was at an age of 15 days). Figure 6 compares the resistivity on the crack after cracking and after healing for the REF and VN beams. The results are normalised by the resistivity next to the crack, as it could be seen that there was an increase in the uncracked zone which can be explained by the ongoing hydration. The REF beams did not undergo healing action. Hence, it would be expected that their results stay approximately constant. This is not the case for two out of four specimens, which might be a result of a different moisture distribution inside the crack. For the VN beams there is an increase after healing. Upon healing the cracks are injected with polyurethane which has a low to negligible amount of moisture, resulting in an increased resistivity.
Preliminary Analysis of Non-destructive Test Methods
377
While it was globally possible to differentiate the healing, this method does not seem optimal for measurements on site. First, one would expect the resistivity to be higher for measurements on the crack compared to measurements next to the crack. This was not always the case, see the bars below the horizontal 100% line in Fig. 6. Secondly, the measurements are very dependent on the degree of moisture in the samples [9], making measurements on site not straightforward in case the exposure conditions change. In addition, the resistivity is also influenced by hydration and pozzolanic reaction [7].
Fig. 6. Median electrical resistivity on the crack normalized over the resistivity next to the crack for REF and VN beams.
3.6 Ultrasound Concerning the uncracked situation, an average longitudinal wave velocity was calculated from all specimens to characterize the intact microstructure of the considered samples. Afterwards, in the cracked and healed stages, the values signify the average per specimen. A comparison between the REF and VN beams reveals a higher wave velocity for the reference concrete (Fig. 7). The reduced velocity upon placement of the vascular systems can be explained by the difference in density and elastic properties of the network compared to the concrete matrix. Also, the concrete casting around the vascular network will introduce a higher porosity, which negatively affects the wave propagation. After cracking, the longitudinal wave velocity decreased significantly for all tested beams, due to the discontinuity between the two receiving sensors. Small variations in the velocity exist due to discrepancies between the crack openings, crack tortuosity, etc. Later on, the wave velocity stayed approximately constant for the REF beams after healing of the VN beams, while for the VN series a clear restoration of the wave velocity was noticed. As such, it is evident that the ultrasonic test setup is able to register the crack healing. 3.7 Summary Table 4 gives a summary of the laboratory evaluation of the different test methods.
378
T. Van Mullem et al.
4 Application of Test Methods on Site The test methods were also evaluated on site. The on-site measurements were done on retaining walls located at Inter-Beton in Bruges, Belgium. These walls (14 m × 2.75 m × 0.80 m) are large-scale demonstrator elements to study the influence of super absorbent polymers on shrinkage cracking [2]. Half of the length of the walls was treated with a curing agent and the other half was left untreated. All test methods could be applied on the treated side of the wall, but the untreated side was too dusty, which made that the Karsten tubes and the air permeability tester could not be attached. This problem could potentially be solved by washing or rinsing (e.g. with a pressure washer) the surface.
a)
b)
Fig. 7. Wave velocity in an uncracked state, after crack introduction and after healing of (a) reference beams (REF) and (b) vascular network beams (VN).
Table 4. Summary of effectiveness of test methods to quantify self-healing. Portable microscope
Time-consuming (digital pictures can increase efficiency), difference with lab microscope due to accuracy discrepancy
Moisture content meter
Not ideal
Air permeability tester
Once calibrated, good in quantifying crack filling
Karsten tubes (water permeability) Good to quantify healing, for wider crack widths a larger reservoir is required Resistivity meter
Not ideal due to other influencing factors
Ultrasound device
Able to register crack healing
5 Conclusion Microscopy remains an essential tool to characterize cracks. Nevertheless, it is important to realize that a handheld microscope might underestimate the crack width compared to a laboratory microscope. Accuracy and measuring time could be improved with a handheld
Preliminary Analysis of Non-destructive Test Methods
379
microscope which allows to take digital pictures. The moisture content and the resistivity were not very promising methods to be used for characterising healing efficiency. Healed cracks and cracks with a small width could be assessed via water permeability tests with Karsten tubes. However, for larger cracks a bigger reservoir needs to be adopted. For the air permeability a new test parameter was defined: the pressure in the inner reservoir 30 s after starting the machine. With this parameter, which is not standard recorded by the equipment, healing could be clearly differentiated and quantified. In addition, with the ultrasound method it was possible to clearly differentiate the crack healing and therefore this seems to be a promising technique as well. Application of the test methods on site learned that the finish of the surface is extremely important; a dusty surface cannot be measured. This could potentially be solved by washing or rinsing the surface, although this should be verified. Acknowledgements. This research was supported by a grant (21SCIP-C158977-02) from the Construction Technology Research Program funded by the Ministry of Land, Infrastructure and Transport of the Korean government. This project has received funding from the European Union’s Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement No 860006.
References 1. Van Mullem, T., Gruyaert, E., Caspeele, R., De Belie, N.: First large scale application with self-healing concrete in Belgium: analysis of the laboratory control tests. Materials 13(4), 997 (2020) 2. Tenório Filho, J.R., Mannekens, E., Van Tittelboom, K., Van Vlierberghe, S., De Belie, N., Snoeck, D.: Innovative SuperAbsorbent Polymers (iSAPs) to construct crack-free reinforced concrete walls: An in-field large-scale testing campaign. J. Build. Eng. 43, 102639 (2021) 3. Li, Z., de Souza, L.R., Litina, C., Markaki, A.E., Al-Tabbaa, A.: A novel biomimetic design of a 3D vascular structure for self-healing in cementitious materials using Murray’s law. Mater. Des. 190, 108572 (2020) 4. Tsangouri, E., Van Loo, C., Shields, Y., De Belie, N., Van Tittelboom, K., Aggelis, D.G.: Reservoir-vascular tubes network for self-healing concrete: performance analysis by acoustic emission, digital image correlation and ultrasound velocity. Appl. Sci. 12(10), 4821 (2022) 5. Murray, C.D.: A relationship between circumference and weight in trees and its bearing on branching angles. J. Gen. Physiol. 10(5), 725 (1927) 6. De Nardi, C., Gardner, D., Jefferson, A.D.: Development of 3D printed networks in self-healing concrete. Materials 13(6), 1328 (2020) 7. Henry, M., Arnamwat, T.: Evaluation of crack self-healing behavior in mortar using electrical resistivity. In: 76th Annual RILEM Week, Kyoto, Japan (2022) 8. Van Tittelboom, K., et al.: Comparison of different approaches for self-healing concrete in a large-scale lab test. Constr. Build. Mater. 107, 125–137 (2016) 9. Polder, R., et al.: Test methods for on site measurement of resistivity of concrete. Mater. Struct. 33(10), 603–611 (2000)
Open-Source EMM-ARM Implementation for Mortars Based on Single-Board Computer Thomas Russo(B)
, Miguel Azenha , and José Granja
University of Minho, Guimarães, Portugal [email protected]
Abstract. The EMM-ARM (Elastic Modulus Measurement through Ambient Response Method) allows the continuous monitoring of the elastic modulus of cementitious materials from early ages. The idea is to subject a beam, made of the specimen in its mould, to an excitation and monitor its response via an accelerometer. The excitation can either rely on naturally occurring vibrations, or on a controlled excitation system creating a signal with the necessary characteristics. The resonant frequency of the tested beam can be assessed with modal identification techniques, whereas the E-modulus of the tested material can be directly calculated with the dynamic equations of motion of the system. The original implementation of EMM-ARM uses specialized devices for the acquisition and excitation systems, which results in a relatively high price, as well as limited options for customization. The software used is based on proprietary systems (LabView), which further brings limitations on sharing for other institutions to use. On the other hand, one should bear into account that cyber-physical systems have shown significant evolutions in the last decade. Open-source platforms are increasingly popular and low-cost single-board computers are becoming widespread (e.g. Raspberry Pi). Electronic components have evolved parallel to these platforms, offering decent performances for low prices nowadays. The work hereby presented took inspiration from these cyber-physical systems to develop an open-source and cost-effective system able to conduct EMM-ARM tests independently from any other computing device. The system will be integrally presented as well as results obtained in comparison with the original implementation of the system. Keywords: Elastic modulus · Cementitious Materials · Cost-Effective · Open Source · Monitoring
1 Introduction The study of elastic modulus (E-modulus) for cementitious materials is of prime importance, especially from a structural analysis point of view. Measuring the E-modulus evolution at very early ages, in the vicinity of the setting process can even bring further valuable information about the performance of a mix [1–3]. For these interests, the work presented hereby focuses on monitoring the E-modulus of mortars. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 A. J˛edrzejewska et al. (Eds.): SynerCrete 2023, RILEM Bookseries 43, pp. 380–391, 2023. https://doi.org/10.1007/978-3-031-33211-1_34
Open-Source EMM-ARM Implementation for Mortars
381
The Elastic Modulus Measurement through Ambient Response Method (EMMARM) is a resonant-based method designed to continuously monitor the E-modulus of a material under test since casting into the testing mould, forming a composite beam. EMM-ARM is based on the determination of the first resonant frequency of the testing beam, which changes over time as the tested material hardens. The tested material Emodulus can be calculated from the identified first resonant frequency using the testing system’s dynamic equations of motion [4]. The current measurement system to conduct EMM-ARM tests is based on specialized equipment, as well as licensed software (MATLAB/LabView). Therefore, a setup for EMM-ARM can be relatively expensive and not adequate to adapt for specific and original applications. The presented work seeks the idea of making developments efficient to deploy and implement for any potential user, without needing a high budget or licensed software. Therefore, all the developments disclosed herein have been made using only open-source software, and cost-effective electronic components. Using open-source software allows the possibility to modify the code, in order to custom design it for other situations, or even components. In this paper, the term “cost-effective” is employed, instead of “lowcost”, because a balance between the price of the components, the performances, and the time invested to get the expected results was made. The system developed could have been made for an even lower price with thorough optimization in terms of electronics, but that was not the point of the research. With open-source and cost-effective developments, the performance requirements to conduct EMM-ARM tests on mortars can be reached, and the system offers the possibility to be custom designed for various applications. For example, additional measurements could be easily made and upgraded in a customized system, in parallel to the measurement of the E-modulus, such as temperature, humidity, gas composition analysis, etc. The number of specimens to monitor can also be adjusted from one unique specimen to 16 or more, using custom-built multiplexers. All these possible customizations are key elements for research, where original measurements are often targeted, but also for the industry that frequently lags behind the capabilities of measurement of research (either due to difficulties in implementing or due to the lack of commercial systems for the purpose at acceptable prices).
2 EMM-ARM Application to Mortars: Original Setup The EMM-ARM was developed in 2008–2009 by Azenha [5] initially to monitor the Emodulus of hardening concrete. Since then, EMM-ARM was applied to a large variety of materials, such as cement mortar, lime-cement blended mortars, cement paste, stabilized soils and even epoxy resins [4, 6–9]. Continuous efforts were produced in order to enhance the performances of EMM-ARM. In its most basic version, EMM-ARM merely relies on the measurement of accelerations at a relevant point in the tested beam (either mid-span for a simply supported setup, or extremity for a cantilevered beam), with the beam solely subjected to the environmental vibrations. This setup, which was the initial implementation and gave its name to EMM-ARM, is currently reserved for cement paste tests only. As a matter of fact, to enhance the reliability of the results for stiffer specimens (e.g. with smaller spans), such as the ones used for concrete, stabilized soils and mortars,
382
T. Russo et al.
an electromagnetic actuator was designed to excite the beam in a controllable way [4]; thus, the tested beam can be subjected to a controlled excitation that over the testing period, with better capacities being attained in terms of modal identification. This setup is currently used for concrete, mortar and stabilized soil specimens. A scheme representing the test setup for such materials can be visualized in Fig. 1. In order to identify the resonant frequency of the testing beam, an excitation signal is sent to an electromagnetic actuator. The response of the beam to this excitation is measured via an accelerometer by an ADC. It can be noticed in the scheme presented in Fig. 1 that a wire links the generated excitation signal to another input of the ADC; indeed, in order to apply the modal parameters identification method, the excitation signal, as well as the accelerations of the beam, have to be measured simultaneously [4].
Fig. 1. Illustration of EMM-ARM setup for mortars.
The original setup for the application of EMM-ARM to mortars is composed of specialized equipment, ensuring laboratory-grade measurements. For the acceleration measurements, the system relies on the PCB TLD352C04, which is a light sensor (5.8 g), perfectly adapted to conduct EMM-ARM tests. The sensitivity of this sensor is 10 mV/g and its typical resolution is rated at 0.5 mg rms. The acquisition system, also in charge of generating the excitation signal is the NI USB-4431. This ADC/DAC is able to acquire 4 channels simultaneously at 102.4 kS/s with a resolution of 24-bits. In the original version of EMM-ARM application for mortar, the data is collected by the ADC and transferred to a computer, where it can be stored and post-processed.
3 Hardware of the New Design 3.1 Computing Device On the market of programmable boards, a wide range of brands and types of boards can be found. It is firstly interesting to distinguish the two principal types of boards available: microcontrollers, and microcomputers. Microcontrollers are the most basic boards, as they simply run a script that is in their memory as soon as they are power supplied. They have the advantage to be, for most models, significantly cheaper than microcomputers,
Open-Source EMM-ARM Implementation for Mortars
383
and to be simple to use: once the script is transferred to the microcontroller, the only action needed is plugging it to a power supply; furthermore, it can be unplugged at any time. They also consume very little amount of energy, which makes them ideal for remote projects, where the use of batteries as a power supply is required. The main drawback of microcontrollers is their low processing power, and often very limited memory, which prevents more complex programs involving a high number of variables to be implemented [10]. On the other hand, microcomputers can usually be found at a higher price, depending on the models. They come with the advantage of having much higher processing power and relatively low energy consumption which therefore makes them more flexible and extends the possibilities compared to microcontrollers. However, their architecture is similar to regular computers, which makes them more complex than microcontrollers. For instance, the time to power on a microcomputer can be approximated to a minute, while it takes merely one second for a microcontroller to start executing its script. A microcomputer, as well as a regular computer, cannot be unplugged from its power source without any risk [11]. Due to the relatively high processing power necessary to perform an EMM-ARM test, in this project, it was decided that the best solution was to use a microcomputer. More precisely, a Raspberry Pi (RPi) 4 model B with 8 GB of RAM was chosen, as it was adequate for the application. The RPi comes with many features, such as WiFi connectivity, 40 global purpose input/output (GPIO) pins, I2C and SPI serial interfaces for peripherical (such as analogue to digital converter). An important feature that comes with RPi is the large community working on it and sharing issues and solutions online, providing relevant resources for initial and intermediate users. Many electronic components were also specifically designed to adapt efficiently to the GPIOs of the RPi (Pi hats) that can extend the functionalities of this microcomputer. With these functionalities and its large RAM, the RPi is suitable to conduct EMM-ARM tests and for post-processing of the data. The RPi runs on a variation of the Debian exploitation system, Raspbian, and offers most of the functionalities of a regular computer, coupled with a development board thanks to its GPIO pins. These pins are used to interact with electronic components and can be controlled via Python programs. 3.2 Excitation of the Specimen In order to excite the beam and monitor its response to the excitation, a predefined signal needs to be generated to the actuator. The signal to be generated is a sine sweep, ranging on the interesting frequency domain that is repeated indefinitely during the measurement period. In practice, to generate an analogue signal, an analogue to digital converter (DAC) is required. A DAC converts discrete digital values into a continuous analogue signal. For this application a 12-bit DAC was selected, the MCP4725 from Adafruit, which uses the communication protocol I2C. This DAC, connected to the RPi, can generate a sine signal at up to 8000 Hz, which is more than required for this case. A common issue with low-cost DACs found on the market is that they are not able to generate negative voltage signals. In the application of this project, it is important to generate a sine signal centred around 0 V, to obtain an average force of excitation
384
T. Russo et al.
null. The chosen DAC suffers from this issue: the sine at its output pin ranges from 0 V to 5 V. The signal needs to be shifted down to be centred around 0 V. Another issue common to most DACs is their low capacity for amperage output. Indeed, when the DAC is connected directly to the electromagnetic actuator, a voltage drop can be observed due to a lack of power. The consequence is that the signal emitted to the actuator doesn’t have constant amplitude along interesting frequency domain. In order to counter these two issues, an amplifying and shifting circuit was designed, and is presented in Fig. 2. The circuit is based on two key components, a transistor and an amplifier operational (op-amp). The transistor allows the draw of electricity from an external power source for supplying the actuator and the op-amp is in charge of shifting down the signal, as well as slightly amplifying it to reach a range of [−4 V; +4 V]. The circuit was implemented into a PCB to simplify the cable management in the device.
Fig. 2. Amplifying and shifting circuit.
An objective of this research was to have the possibility to monitor several specimens during the same test. The strategy chosen to reach this objective, considering the limited capacities of the analogue to digital converter (ADC), is to excite and measure each specimen in sequence. Therefore, the system should have the capacity to activate an actuator when the associated specimen is being measured. Several solutions can be considered to achieve this task, for example, the use of multiplexers. The device presented in this paper relies on a custom-made multiplexer to select the specimen that is being measured. 3.3 Response Acquisition of the Specimen The response acquisition of the specimen corresponds to its acceleration measurement. This measurement is made via an accelerometer. The selected accelerometer is the ADLX203EB from Analog Devices, with a sensitivity of 1000 mV/g, a resolution of 1 mg and a range of ± 5 g. This accelerometer is an analogue sensor, which means that it produces an analogue signal (a voltage), which then needs to be converted to digital data, to be interpreted by the computing device. This conversion is made via an ADC. The RPi does not have any
Open-Source EMM-ARM Implementation for Mortars
385
integrated ADC; consequently, to read the signal of an accelerometer, an external ADC must be added. The principal criteria to choose an ADC are its sampling frequency and resolution. The sampling frequency corresponds to the number of data converted per second. To reliably measure a signal, the Nyquist law states that the sampling frequency should be at least twice the maximum expected frequency [12]. For example, if one expects to measure a signal ranging from 0 Hz to 250 Hz, the sampling frequency should be at least 250*2 = 500 Hz. In practice, and for conservative reasons, factor 3 is commonly used. Another parameter to consider is that the sampling frequency is divided by the number of channels simultaneously measured. In the case of this project, two channels need to be simultaneously acquired, the excitation signal and the acceleration one. Consequently, considering a conservative expected maximum frequency of 250 Hz, the ADC should have a sampling frequency higher than 2*3*250 = 1500 Hz. Regarding the resolution, in order to get the best out of the selected accelerometer, the resolution should allow the reading of 1 mV changes. The ADC chosen for this application is the 24-bit high-precision AD/DA board from Waveshare. This ADC is rated for up to 30 kHz of sampling frequency, even though in practice, with an RPi, the maximum sampling frequency achievable is around 5000 Hz, which is still suitable for the targeted application. Figure 3 presents a scheme showing the different electronic components constituting the developed device and their connections.
Fig. 3. Connections scheme of the developed device components.
386
T. Russo et al.
4 Software of the New Design 4.1 Multiprocessing on Raspberry Pi In this research, as previously mentioned, two functions should take place in parallel two acquisitions and one excitation. To efficiently archive this, the software should be designed in a way that allows it to run these functions in parallel. The coordinated execution of programs by multiple computer processors is known as multiprocessing. As a broad word, “multiprocessing” can refer to either the dynamic assignment of a program to two or more computers working together or to many computers working on the same program concurrently (in parallel) [13]. Multiprocessing is therefore conceived for applications requiring the execution of several functions simultaneously, running on distinct cores of a machine. The RPi used as a computing device integrates four cores. However, for coding simplicity, the designed program does not take benefit from the simultaneous use of more than 1 core. The software will, in the near future, be upgraded with a multiprocessing method. Two techniques are particularly recommended to substitute pure multiprocessing: multithreading and asynchronous methods. It should be clarified though, that the name multithreading refers to a software method, and cannot produce the same results as simultaneous multithread (SMT) hardware (which is absent on RPi), as it is explained below. Synchronous execution is a usual method for coding; a function is entirely executed when called. For the targeted application of this paper, this classic method is not suitable, since channels need to be monitored while generating a signal; if a synchronous method was used, the result would be as illustrated in Fig. 4. The excitation function in charge of generating the excitation signal finishes before the recording is made, which makes it irrelevant.
Fig. 4. Illustration of synchronous programming method.
On the other hand, asynchronous and multithreading programming methods allow the possibility of running parts of different functions in sequence. Due to the speed of the RPi processor, 1.5 GHz, switching between each function part could be, at our scale (measurements at 1.5 kHz), considered instantaneous, making possible the imitation of multiprocessing. When using the multithreading method on a device whose hardware
Open-Source EMM-ARM Implementation for Mortars
387
does not possess SMT, each function will be executed line by line in sequence. The first line of each function will be executed, then the second one, until the last line is executed. The major difference between the multithreading and the asynchronous method is the control of the blocks executed in sequence: in the asynchronous method, the program will execute a block of function until it is indicated to switch to the next function. The asynchronous method consequently offers more consistent control over the timing of execution and was for this reason selected for the development of this project. 4.2 Modal Parameters Identification Method The modal identification process corresponds to the post-processing aspects and consists in extracting relevant information from vibration data. These modal parameters refer to eigenfrequencies, damping ratios, mode shapes and modal participation factors. In this research, a parametric method, named Stochastic Subspace Identification (SSI) method was used to identify the first resonant frequency during the tests made to validate the developed device. This method was initially conceived for Operational Modal Analysis (OMA) when the excitation parameters of the system are unknown. This method allows the identification of the modal parameters in the time domain, which has the advantage of being less sensible to external noises than frequency domain methods, such as peak picking. A thorough presentation of the SSI method can be found in [14]. Even though SSI is a robust method, it does not consider the excitation signal, that is measured during the test. With the excitation signal, it is possible to calculate the Frequency Response Function (FRF), which is defined as: Hij (w) =
yj (w) xi (w)
(1)
where yj (w) and xi (w) are the complex amplitudes of the input and output respectively. For optimal modal parameters identification, the SSI method using FRF data will be implemented in the developed device in the near future.
5 Preliminary Results and Discussions 5.1 Tests Conducted The objective of the conducted tests was to validate the developed device. Consequently, the developed device was systematically compared to the original EMM-ARM setup. To this day, two tests were conducted for the system validation: (i) resonant frequency measurement of a hardened cement paste specimen, with ambient vibration excitation (no use of electromagnetic actuator);, with execution of 6 runs of 10 min-long measurements; and (ii) resonant frequency measurement of a hardened mortar specimen, using the original system excitation system, with execution of 6 runs of 5 min-long measurements. The specimen used for the cement paste test was made according to the latest cement paste mould developments, presented in [4]. The mould is constituted by a 550 mm long acrylic tube with internal and external diameters of respectively 16 mm and 20 mm. The
388
T. Russo et al.
specimen is in cantilever conditions with a span of 450 mm. The accelerometer is placed in the free end of the specimen. Regarding the mortar specimens, the mould used was made according to the tests performed in [9]. The mould is a 550 mm long PVC tube with internal and external diameters of respectively 44 mm and 50 mm. The specimen is simply supported with a span of 500 mm. The accelerometer is placed at midspan. The results of these tests will be detailed in the following subsection. 5.2 Results The results of the conducted tests are presented in Table 1 and Table 2. For each test conducted, the developed device revealed satisfactory results, with an identified resonant frequency in the same range as the one identified with the original setup. The resonant frequencies obtained from the two systems were obtained with the same identification method, SSI. Figure 5 presents the power spectrum that resulted from the experiment on the mortar specimen. The peak exhibited on the graph represents the 1st resonant frequency of the system composed of the specimen in its mould.
Fig. 5. Comparative power spectrum obtained with the mortar specimen.
5.3 Discussions For every test, the developed system identified resonant frequencies within less than 1% of gap compared to the original setup. The standard deviation of the developed device measurements is consistent and in the same order as the one of the original setup. These elements prove that the resonant frequencies identified by the developed device are reliable, and this device can be considered validated for single sample monitoring.
Open-Source EMM-ARM Implementation for Mortars
389
Table 1. Comparative results on 1 hardened cement paste specimen. Cement paste specimen test; ambient vibration excitation Original setup
Frequency (Hz)
Run 1
Run 2
Run 3
Run 4
Run 5
Run 6
Average
Standard deviation
25.36
25.42
25.54
25.36
25.17
25.42
25.4
0.12
25.35
25.21
25.35
25.28
25.21
25.35
25.3
0.07
0.04
0.83
0.74
0.16
0.28
Developed device Frequency (Hz) Gap (%)
0.032
0.39
Table 2. Comparative results on 1 hardened mortar specimen Mortar specimen test; excitation of specimen generated by original setup Original setup
Frequency (Hz)
Run 1
Run 2
Run 3
Run 4
Run 5
Run 6
Average
Standard deviation
87.89
87.89
88.50
88.50
88.50
88.50
88.3
0.32
88.04
88.08
88.04
88.04
88.33
88.33
88.1
0.15
0.17
0.22
0.52
0.52
0.19
0.19
Developed device Frequency (Hz) Gap (%)
0.30
To complete the validation process, the developed device now needs to be tested for multi samples monitoring, as well as for monitoring the full hydration process of mortars. The tests should be conducted using the developed device’s excitation for the electromagnetic actuator. These tests will be performed in the near future. Even though the pure multiprocessing method was not used, the developed device exhibited very reasonable results, which prove the absence of necessity for it to achieve EMM-ARM tests. Some aspects of the developed device should however be considered. Due to its limited acquisition capabilities, the developed device can only monitor a single specimen at a time; when several specimens are to be monitored during the same test, they are monitored in sequence, one after the other. For most applications, this aspect is not an issue considering the time scale of the hardening process of cementitious materials, but it could appear as one for some specific applications. Moreover, as RPi is not specifically designed to be used as a measurement system, the sampling frequency of acquisition can slightly vary from one run to another, which can partly explain the observed gaps with the reference original device in Table 1 and Table 2.
390
T. Russo et al.
6 Conclusions In this research, the aim was to develop an open-source and cost-effective device able to reliably conduct EMM-ARM tests. Therefore, different electronic components such as computing devices, ADC, DAC and accelerometers available on the market were investigated in order to select the most adequate ones. Some issues inherent to low-cost components were overcome, such as the ones related to the DAC, detailed in Sect. 3.2. In conclusion, an open-source and cost-effective EMM-ARM device was developed and presented in this paper. The device exhibited results comparable to those obtained with the reference original setup of EMM-ARM, costing approximately 12 times less. The validation process of the device will be completed with additional tests, to ensure its reliability and a multiprocessing programming method will be implemented. The results obtained comfort the choice of electronic components for the targeted application. The system carries the advantage to be easily implementable and modulable, contributing to the development and dissemination of EMM-ARM. Acknowledgements. This work was partly financed by FCT/MCTES through national funds (PIDDAC) under the R&D Unit Institute for Sustainability and Innovation in Structural Engineering (ISISE), under reference UIDB/04029/2020. This project has received funding from the European Union’s Horizon 2020 research and innovation programme under Marie Sklodowska-Curie project SUBLime [Grant Agreement no. 955986].
References 1. Kar, A., Halabe, U.B., Ray, I., Unnikrishnan, A.: Nondestructive characterizations of alkali activated fly ash and/or slag concrete. Eur. Sci. J. 9(24), 1857–7881 (2013) 2. Singh, G., Siddique, R.: Effect of waste foundry sand (WFS) as partial replacement of sand on the strength, ultrasonic pulse velocity and permeability of concrete. Constr. Build. Mater. 26(1), 416–422 (2012) 3. Rao, S.K., Sravana, P., Rao, T.C.: Experimental studies in Ultrasonic Pulse Velocity of roller compacted concrete pavement containing fly ash and M-sand. Int. J. Pavement Res. Technol. 9(4), 289–301 (2016) 4. Granja, L.J.D.: Continuous characterization of stiffness of cement-based materials: experimental analysis and micro-mechanics modelling. PhD Thesis, University of Minho, Guimaraes, Portugal (2016) 5. Azenha, M.: Numerical simulation of the structural behaviour of concrete since its early ages. PhD Thesis, University of Porto, Porto, Portugal (2009) 6. Benedetti, A., Fernandes, P., Granja, J.L., Azenha, M., Sena-Cruz, J.: Effects of curing temperature on pull-out behavior and stiffness evolution of epoxy adhesives for NSM-FRP applications. In: Proceedings of SMAR 2015 – Third Conference on Smart Monitoring, assessment and Rehabilitation of Civil Structures (2015) 7. Silva, J., Azenha, M., Correia, A.G., Granja, J.: Continuous monitoring of sand-cement stiffness starting from layer compaction with a resonant frequency-based method: issues on mould geometry and sampling. Soils Found. 54(1), 56–66 (2014) 8. Azenha, M., Silva, J., Granja, J., Gomes-Correia, A.: A retrospective view of EMM-ARM: application to quality control in soil-improvement and complementary developments. Proc. Eng. 143, 339–346 (2016)
Open-Source EMM-ARM Implementation for Mortars
391
9. Ramesh, M., Azenha, M., Lourenço, P.B.: Study of early age stiffness development in lime– cement blended mortars. In: Aguilar, R., Torrealva, D., Moreira, S., Pando, M.A., Ramos, L.F. (eds.) Structural Analysis of Historical Constructions. RB, vol. 18, pp. 397–404. Springer, Cham (2019). https://doi.org/10.1007/978-3-319-99441-3_42 10. Mazzei, D., Montelisciani, G., Baldi, G., Fantoni, G.: Changing the programming paradigm for the embedded in the IoT domain. In: IEEE World Forum on Internet of Things, WF-IoT 2015 - Proceedings, pp. 239–244 (2015) 11. Costa, D.G., Duran-Faundez, C.: Open-source electronics platforms as enabling technologies for smart cities: recent developments and perspectives. Electronics (Switzerland) 7(12) (2018) 12. Shannon, C.E.: Communication in the presence of noise. Proc. IEEE 86(2) (1998) 13. Shaik, S., Van Hout, R.: Transparent serverless execution of Python multiprocessing applications. Int. J. Multiphase Flow 104262 (2022) 14. Peeters, B., De Roeck, G.: Stochastic system identification for operational modal analysis: a review. ASME J. Dyn. Syst. Meas. Control 123(4), 659–667 (2001)
Understanding the Degradation of Concrete Structures During the Nitrification Process for the Treatment of Wastewater: A Lab Biological Degradation Test Yasmine Werghi1 , Tony Pons1(B) , Marielle Guéguen Minerbe1 , Marcos Oliviera2 , Sam Azimi2 , Vincent Rocher2 , and Thierry Chaussadent1 1 MAST-CPDM, Université Gustave Eiffel, F77454, Marne la Vallée, France
[email protected] 2 Direction Innovation, SIAAP, 82 Avenue Kléber, 92700 Colombes, France
Abstract. Regulations on wastewater treatment (UWWTD-1991; WFD 2000) have evolved considerably over the past 25 years with the development of increasingly efficient bioprocesses that limit environmental risks. High-performance biophysico-chemical wastewater treatment technologies were implemented for the plants of the main urban areas i.e. biofiltration, membrane bioreactors. Thus, over the last ten years, new types of degradation of concrete structures were observed in wastewater treatment plants, mainly in nitrification basins treating nitrogen compounds (ammonium) in effluents. The exact origins of these degradations are not yet determined. In order to understand the degradation mechanisms of concrete in nitrogen compound treatment basins, mortar specimens based on CEM V and CAC cements were exposed (i) in situ of one nitrogen treatment basin and (ii) in reactor for biological lab test. The design and implementation of the biological test were developed to reproduce were developed to reproduce nitrogen biological treatment and evaluate the degradation of mortars in a more controlled environment than the full-scale process. This laboratory test allows the evaluation of the biotic aspect of nitrification and in particular, the influence of acid production during biotic reactions and the influence of carbonates on mortar specimens. These two approaches allow highlighting the significant impact of the biomass and the modification of the environmental conditions at the biofilm/material interface. Keywords: Nitrification · biological test · mortars · biomass
1 Introduction In wastewater treatment, the application of the Urban Wastewater Directive (UWWD, 1991) and the Water Framework Directive (WFD, 2000) have led to a significant increase in the requirements for the quality of water returned to the natural environment. In response to these requirements, more efficient technologies have been integrated into © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 A. J˛edrzejewska et al. (Eds.): SynerCrete 2023, RILEM Bookseries 43, pp. 392–398, 2023. https://doi.org/10.1007/978-3-031-33211-1_35
Understanding the Degradation of Concrete Structures
393
the treatment processes of wastewater treatment plants, in particular biofiltration on a fixed culture for the treatment of nitrogen. Although these processes are well adapted to reach the regulatory thresholds for residual nitrogen, under certain conditions, premature degradation of the reinforced concrete structures has been observed especially in the nitrification basins where aerobic bacteria oxidize ammonium ions to nitrite (nitritation) and then to nitrate (nitration). Previous studies [1, 2] show that the degradation of these concrete structures is of biological origin and results from acid attack produced by nitrifying microbial activity at the concrete/nitrifying biofilm interface. Abiotic laboratory tests have shown that there is no acceleration of leaching of cementitious materials in the presence of nitrite or nitrate ions resulting from the oxidation of ammonium ions, which means that the two processes during nitrification, nitritation and nitration, are not involved in the deterioration mechanism in terms of nitrite and nitrate ion production [3]. Therefore, in order to preserve the integrity of the structures while ensuring treatment performance, it is essential to be able to test different types of materials. The present study therefore aims to design a biologic laboratory test and to compare degradation of mortar specimens submitted to the laboratory test or exposed in wastewater treatment plants.
2 Materials and Methods 2.1 Materials Description For industrial scale (in situ), the mortars were prepared in the form of cylindrical specimens (height 6 cm, diameter 3 cm) from two types of cement (Portland cement with additions of blast furnace slag and siliceous fly ash CEM V/A and calcium aluminate cement CAC). The mass ratios of water to cement and sand to cement are 0.5 and 3 respectively. These specimens are equipped with a PVC tube allowing them to be placed in support boxes and exposed to immersion in situ. For the biological degradation test in lab, the mortars were prepared as parallelepipeds (2 × 2 × 8 cm) from the same cements with the same mass ratios. Once cured (endogenous cure in plastic bags at 20 °C for 28 days), the materials were sawn to obtain cubic samples (2 × 2 × 2 cm) for laboratory immersions. 2.2 Protocol for in Situ Exposure The support boxes containing the mortar samples were placed in the upper part of an aerated biofiltration basin operating under nitrification in Seine aval’s wastewater treatment plant, in France (Fig. 1). In this nitrification basin, the conditions for daily ammonium ion concentrations are approximately 20 and 0.7 mg N-NH4 + .L−1 at the inlet and outlet respectively. The degradation of the samples was monitored every 6 months for 2 years following (masses, dimensions, surface pH and visual aspect).
394
Y. Werghi et al.
B A
Fig. 1. (A) position of the box in the aerated biofilter. (B) photo of the box before exposure.
2.3 Protocol for Lab Biological Test The schematic bioreactor is shown in Fig. 2, the water used comes from a sewage treatment plant before the nitrification step, it is replaced every fifteen days to limit the concentration of resulting compounds (NOx ). Initially, activated sludge from the water treatment plant is spread on the mortar cube surfaces, a reference cube is kept outside the bioreactor. Each of the two bioreactors is operated in parallel, the evolution of the pH and oxygen content is continuously measured. As the nitrification reaction produces an acidification, the pH was adjusted with a NaOH solution during the first two months and with a carbonate/bicarbonate solution in the following two months. In addition, the nitrite, nitrate and ammonium contents were measured frequently by spectrophotometry, the ammonium content was adjusted by adding NH4 Cl saturated solution. The size and
Fig. 2. Schematic of a nitrification reactor.
Understanding the Degradation of Concrete Structures
395
mass of the mortar cubes are measured at each water change. In addition, a mortar sample is taken once a month to observe chemical variation in depth using a SEM.
3 Results 3.1 Monitoring of in Situ Exposure The monitoring of the mass variation of the samples is shown in Fig. 3. The mass gain in the early stages for the samples corresponds to the filling of the porosity by water and eventually to the biofilm formation on the surface. Then, beginning of mass loss is observed for the CEM V. The pictures show that the CEM V cylinder has rounded corners which are characteristic of a loss of mass, whereas the CAC cylinder still has very sharp corners.
Fig. 3. Mass variation (in %) and photo of CEM V (grey) and CAC (yellow) samples exposed for 26 months in an aerated biological treatment basin.
3.2 Results for Lab Biological Test The variations of the main parameters (ammonium and nitrates) as a function of time for a 20-day cycle are presented in Fig. 4. At the beginning of a cycle the ammonium concentration in the water is about 32 N-NH4 (mg/L) and it decreases reaching zero in 7 days. Then, a doping in NH4 Cl is necessary (blue arrows on Fig. 4.) to induce the nitrification reaction, the decrease in concentration is much faster. Over a 15-day cycle the concentration of nitrate, the product of the nitrification reaction, increases from 1 to 76 NNO3 (mg/L)1 . At the same time the nitrification reaction produces H+ ions which acidify the bioreactor medium to values of 4.5. The pH adjustment was done for two months with a NaOH solution, then for the next two months with a carbonate/bicarbonate solution. The results showed that the pH was stable for longer with the carbonate/bicarbonate solution. This strategy allowed the maintenance of the nitrification process, similarly to the full-scale process. Both bioreactors show the same trends.
396
Y. Werghi et al.
After two months of exposure, mortars with CEM V binder showed a mass loss of 2.5%, while mortars with CAC binder showed a gain of 0.4%. Regarding the dimensional monitoring, the CEM V mortars presented a loss of 3.9% while the CAC mortars had a loss of only 0.7%, which are in agreement with the mass loss.
Fig. 4. Monitoring of NH4 + (N-NH4 , mg/L) and NO3 − (N-NO3 , mg/L) over time for CAC mortar in nitrification bioreactor during 15 days (one cycle).
In order to compare the degradation mechanisms between the samples exposed in situ and those exposed in the laboratory, backscattered electron images and elemental mapping were made with SEM. The results are presented in Fig. 5, after two months of exposure, it can be observed that the degradations are different considering CEM V or CAC mortars. Indeed, the CEM V mortars show a degradation on the first 100 µm whereas the CAC mortars show a degradation on approximately 50 µm. The degraded area has a higher porosity for CEM V mortars than for CAC mortars. In both cases, the degraded areas presented a decalcified zone (Fig. 5. Ca). The degraded zone of the CAC mortars presented an accumulation of aluminum (Fig. 5. Al), this higher aluminum content could play a role of barriers against biological attack.
Understanding the Degradation of Concrete Structures
397
Fig. 5. Backscatter electrons images (top) for CEMV (left) and CAC (right) mortars, as well as elemental mapping of silica (Si), calcium (Ca) and aluminum (Al) after 2 months exposure in bioreactor.
398
Y. Werghi et al.
4 Conclusions Remarks • The chemical monitoring parameters, in particular the decrease in ammonium in favor of the increase in nitrite concentration associated with the pH decrease, clearly confirm that the nitrification reaction is taking place in the bioreactor. • The observations of degradation on a microscopic scale are in line with the observations of samples exposed in-situ. • The first results show that the CAC-based mortar shows a better durability than the CEM V-based mortars to the nitrification reaction by bacteria; these results will be completed by observations at three and four months of exposure. The most probable cause is that the biologically induced nitrification reaction produces H+ ions, and consequently leaching of the cement matrix. The better resistance of the CAC formulation is either that the aluminum phases are more resistant to acidic chemical attack, and/or that aluminum has a negative impact on bacterial proliferation. • In the future, this laboratory test could be used to test different formulations of other types of binders in the laboratory before their possible application in water treatment plants. Acknowledgements. This study is financed by the French National Research Agency, with the WWTConcrete project (ANR-20-CE22-0016) and with the support of the MOdélisation, Contrôle et Optimisation des Procédés d’Epuration des Eaux (MOCOPEE) research programme. The authors thank the technical support of the SIAAP teams.
References 1. Leemann, A., Lothenbach, B., Siegrist, H., Hoffmann, C.: Influence of water hardness on concrete surface deterioration caused by nitrifying biofilms in wastewater treatment plants. Int. Biodeterior. Biodegrad. 64, 489–498 (2010). Author, F.: Article title. Journal 2(5), 99–110 (2016) 2. Leemann, A., Lothenbach, B., Hoffmann, C.: Biologically induced concrete deterioration in a wastewater treatment plant assessed by combining microstructural analysis with thermodynamic modelling. Cem. Concr. Res. 40(8), 1157–1164 (2010) 3. Berrada, L., et al.: Dégradation des ouvrages en béton lors du processus de nitrification pour le traitement de l’azote des effluents d’eau usée en station d’épuration. In: 25th Journées Information Eaux, pp. 551–556 (2022). www.apten.org
A Cost-Effective Micro-controller Based System for EMM-ARM Tests in Cement Paste Renan Rocha Ribeiro1(B) , José Granja1 , Rodrigo Lameiras2 and Miguel Azenha2
,
1 University of Minho, Guimarães, Portugal
[email protected] 2 University of Brasília, Brasília, Brazil
Abstract. The EMM-ARM (Elasticity Modulus Measurement through Ambient Response Method) methodology consists in using ambient vibration tests to monitor the E-modulus evolution of cement-based materials during hydration, right after casting. The dissemination of this methodology may have been hindered by the costs and proprietary platforms associated to the original test system. This work reports the development of a cost-effective and open-source test system to perform EMM-ARM tests, comprised of a data-acquisition system, accelerometer, and post-processing software. The development was supported by the Arduino prototyping platform and the Python program languages, both open-source resources. Being conceived to be an open-source hardware (OSH), full details are given on how to fully replicate the system. Validation was performed by comparing Emodulus estimates obtained from the proposed system and an original system used in previous works. The proposed system is shown to have a satisfactory performance to execute EMM-ARM tests, while being easily replicable, at a considerably lower cost, and offering high potential for further developments and customization due to its open-source nature. Keywords: E-modulus · open-source · data-acquisition · accelerometer
1 Introduction The EMM-ARM (Elasticity Modulus Measurement through Ambient Response Method) methodology was developed initially for monitoring the E-modulus evolution, since the fresh state, of concrete specimens [1]. Since then, it was extended for cement paste, mortar [2], among other materials, and passed through successive optimizations [3]. While being a validated methodology, the EMM-ARM test relies on commercially available general-purpose data-acquisition system and accelerometers that have associated relevant costs and depend on proprietary platforms. This causes difficulties in third-party implementation and widespread dissemination of the methodology. To tackle this, a cost-effective and open-source test system specially customized to perform EMM-ARM tests, comprised of a data-acquisition system, accelerometer, and post-processing software, is presented in this paper. The system is herein disclosed as an © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 A. J˛edrzejewska et al. (Eds.): SynerCrete 2023, RILEM Bookseries 43, pp. 399–412, 2023. https://doi.org/10.1007/978-3-031-33211-1_36
400
R. Rocha Ribeiro et al.
open-source hardware (OSH) and preliminarily validated by comparison to the original test system in a single EMM-ARM test from fresh state to 12 days, which allowed to assess a typical range of stiffness in EMM-ARM tests. Full characterization of the system requires further tests (e.g., more replicates, materials, and test conditions).
2 Original EMM-ARM Test Setup for Hardening Cement Pastes Figure 1 illustrates the principles of an EMM-ARM test and its original setup to study hardening cement pastes. The original proposal of the EMM-ARM test consists in automatically performing measurements sessions to determine the vibration frequency of a tube filled with the material under testing, positioned in known support condition and left to vibrate due to ambient vibrations resembling a white noise load. Using operational modal identification algorithms, such as the SSI-COV method [4], it is possible to estimate the resonant frequency of the specimen. By applying the dynamic equations of motion of a beam [2], all parameters of the problem are known (geometry and mass of specimen can be directly measured, E-modulus of the acrylic of the tube can be obtained in the same setup test with the empty tube) except for the E-modulus of the tested material, which may then be computed.
Fig. 1. EMM-ARM test setup for cement paste specimen.
The test setup in Fig. 1 is the result of successive improvements [3] in relation to the first proposal [2]. It is composed of the specimen (tube filled with investigated material), supports, and a system to acquire acceleration data. In cement paste testing, the tube is made of acrylic, has a total length of 55 cm and 16/20 mm inner/outer
A Cost-Effective Micro-controller Based System
401
diameter, is positioned with a 45 cm free span, closed by two extremity lids, and with an accelerometer placed in its free tip. The support is a clamping device detailed in [3]. The system to acquire acceleration data is comprised by a lightweight analog accelerometer (e.g., PCB 352C34 [5]) with sensitivity at least of 100 mV/g, connected to a 24-bit dataacquisition system (e.g., NI USB-4431 [6]) with ±10 V input, leading to a resolution of 11.9 μg/bit (disregarding noise effects), operated by a portable computer running a LabVIEW software [7] to manage data-acquisition and real-time data post-processing.
3 Proposed System 3.1 Overview Figure 2 presents a block diagram of the main components of the proposed system. The system was conceived to operate independently of a computer. It was built around the 8-bit ATMega328p microcontroller unit (MCU) [8], typical of entry-level Arduino boards, such as the Arduino UNO board [9], which was used as the prototyping platform in the development of the system. The MCU controls three modules: (i) an external 16-bit Analog-to-Digital Converter (ADC) module that improves the resolution over the MCU’s native 10-bit ADC, and acquires data at a maximum of 860 Hz from up to four accelerometers (not simultaneously); (ii) a microSD card module, utilized to save the experimental data; (iii) a real-time clock module, powered with a back-up battery, that time stamps each measurement session to allow tracking the E-modulus evolution. The MCU is also connected to a temperature and humidity sensor and four of its analog input/output pins are made available in panel connectors for future uses (e.g., interfacing with external devices). The user interface of the system is comprised of a 16 characters by 2 lines LCD screen and push buttons for user input.
Fig. 2. Block diagram of the proposed system.
402
R. Rocha Ribeiro et al.
Due to the MCU’s low computing power, the processing of the data is performed in a separate computer, unlike the original system that process data in real-time. After the prototyping, the system was built on a printed circuit board (PCB) to increase the overall reliability. The components have a resilience to degradation factors (e.g., impact, dirty) similar to the original system, being adequate for laboratorial environments. 3.2 Data-Acquisition System Hardware Figure 3 presents the schematics of the data-acquisition system. The components in Fig. 3 are identified by an alphanumerical code (#PXX), further detailed in Table 1. Wherever possible, connections between the PCB and external components fixed to the panel (e.g., buttons, microSD module, connectors, LED screen) were made with headers and receptacles that allow a single connection orientation (e.g., Molex connectors). The MCU’s UART interface (i.e., its RX and TX pins) is also made available for connection to a panel connector, and an auxiliary circuit is prepared for a future UART/USB connection to a computer, which would require a CP2102 module [10]. Figure 4 shows a real-scale view of the PCB, conceived as two-layered board, which means independent circuit tracks can be traced on the top and bottom face of the board. In Fig. 4, top layer tracks are shown in bright yellow, and bottom layer in dark yellow. In the PCB, the alphanumerical code presented in Table 1 is printed near the location of every component. The system was named microEMM-ARM, or μEMM-ARM, and this work addresses its “beta 0.1” version. All the schematics and hardware documentations of this paper are released in GitHub [11]. 3.3 Choice of Accelerometer Besides the data-acquisition hardware, the accelerometer is a crucial component for the performance of the system. The proposed data-acquisition system works with generic analog input signals, so any analog accelerometer (or any other analog sensor) is compatible. For the proposed system, a GY-61 board [12] was chosen. This board includes an ADXL335 [13] accelerometer chip, with sensitivity of 300 mV/g, with capacitor filters in the signal lines and a voltage regulator that allows 5 V powering, as the ADXL335 chip should typically be powered with 1.8 V–3.6 V. Considering the proposed data-acquisition system, this leads to a resolution of 416.7 μg/bit (disregarding noise effects), a value 40 times higher than the original system. A previous work supported the choice of the GY-61 board, indicating enough performance for the intended purposes [14]. It is to be noted that the 0.1 μF capacitors present in GY-61 signal lines are 50 Hz low-pass filter, which, although adequate for EMM-ARM testing of hardening cement pastes (typical frequencies from fresh to hardened state range from 10 Hz to 30 Hz with the setup shown in Fig. 1), may require adjustment for other applications. In such cases, customized ADXL335 boards should be fabricated.
A Cost-Effective Micro-controller Based System
Fig. 3. Schematics of data-acquisition system.
403
404
R. Rocha Ribeiro et al. Table 1. Electronic components of the data-acquisition system.
Id.
Description
Purpose/location
Printed board circuit components and modules 0.1 μF electrolytic capacitor
Noise decoupling at power and serial lines
P18, P32
10 μF electrolytic capacitor
Noise filter
P12, P35, P36
10 k resistor
Pull-up or -down resistors
P09
16 MHz crystal
16 MHz external crystal oscillator
P15
28-pin socket
28-pin socket for the ATmega328p
P17, P23, P26, P30
220 resistor
Resistors to control current drainage
P06, P07
22 pF ceramic capacitor
22pf ceramic capacitor for atmega328p
P19, P20
4.7 k resistor
Pull-up resistors for I2C lines
P01
ADS1115 module
Direct soldering the module can be substituted by 10 female pin headers for ADS1115
P39
DC input jack
Power input for a 9 V–12 V AC-DC voltage
P21a, P21b
DS3231 module
Direct soldering the module can be substituted by 6 female pin headers for DS3231
P25
LM7805
Voltage regulator
P10, P14, P24, P27–31, P33–34, P38, P40
Molex with 2 pins1
Connectors for various panel buttons and connectors
P37, P02–05
Molex with 3 pins1
Connection for temperature and humidity sensor and accelerometer channels
P41
Molex with 4 pins1
Connection for LCD screen module2
P16
Molex with 5 pins1
CP2102 connector for USB/UART communication with ATmega328p
P08
Molex with 6 pins1
Connection for microSD module2
P11, P13, P22
(continued)
A Cost-Effective Micro-controller Based System
405
Table 1. (continued) Id.
Description
Purpose/location
P08b
microSD module
MicroSD module with SPI connectors
P10
PCB push button
Push button to allow atmega328p hard reset
Buttons, connectors, and modules to install in external case P42–45
(On)-Off panel button
Momentary pushbutton, normally open and 12 mm mounting hole diameter, at top panel
P46, P51–54
3.5 mm jack PJ-301M
Stereo audio panel connector with 3 pins for temperature and humidity sensor at top panel and analog input/output at front panel
P47–50
3 way panel connector
Connectors for accelerometer at front panel
P55
5 way panel connector
Connector for a CP2102 module to enable serial communication at back panel
P56
On-Off panel button
Self-lock button, 12 mm mounting hole diameter, at back panel
P57
LED + Panel LED holder
LED to indicate system is powered up
P58
LCD screen module
16 × 2 LCD screen module with I2C interface
P59
microSD card module
MicroSD card module with SPI interface
Notes: 1 – Molex connectors comprise the header, the wire-board receptacle, and crimp terminals 2 – The LCD screen and microSD card modules are not directly soldered on the PCB. Therefore, an extra pair of connectors are needed to connect the PCB to the modules
3.4 Source Code for Data-Acquisition Figure 5 presents the pseudocode of the source code of the data-acquisition system, which summarizes the key points of the source code. The code is comprised of 1838 lines implemented in C++ with the Arduino IDE software [15] and uses 29626 bytes of
406
R. Rocha Ribeiro et al.
Fig. 4. Printed circuit board view.
storage space, which is 91% of the available storage of the MCU. The source code is available in GitHub [11]. The critical point in the code is the strategy for fast data-acquisition and handling the MCU’s low volatile memory. The MCU cannot allocate enough memory to hold all the data collected during a single measurement session. A typical session that lasts 120 s acquiring 2 bytes of data – since ADC has 16-bit resolution – at 860 Hz leads to 206 kB of data. The MCU only has 2 kB total volatile memory, and only 703 bytes available after discounting global variables of the source code. The solution was to use the microSD card to immediately store the acquired data before the volatile memory overflows. Since the microSD card writing also requires volatile memory during its runtime, the only identified strategy was to write every single data point to the card immediately after its sampling. To ensure a stable sampling frequency, the external ADC was used in its continuous mode, which utilizes the MCU’s Interrupt Service Routines (ISR) capabilities to ensure, to the best extent, a stable timed sampling. 3.5 Post-processing Software The post-processing software was developed in the Python language, initially as a Jupyter Notebook. It allows to perform automatic modal identification in all files obtained from
A Cost-Effective Micro-controller Based System
407
an EMM-ARM test and compute the E-modulus from the identified frequencies. A Python library for modal identification named CESSIPy [16] was used to support the
Fig. 5. Pseudocode of the data-acquisition system source code.
408
R. Rocha Ribeiro et al.
modal identification procedures, with minor modifications. The source code is available in GitHub [11]. 3.6 Bill of Materials and Cost Estimation Table 2 presents the bill of materials and cost estimation of a complete system. Datasheets of each component are detailed in the GitHub repository. Costs were estimated from three online retailers via Octopart search engine [17], to ensure quotes referred to same components. Shipping costs were not considered. A minimally operatable system (one accelerometer channel) may cost around e 116. For comparison, a full original system would have a total cost of more than e 12,000 (data-acquisition system, accelerometers, and software) for initial implementation, around 60 times the cost of the proposed system, besides recurrent maintenance costs due to proprietary licences. Additional documentation is available in the GitHub repository [11]. Table 2. Bill of materials and cost estimation for a full system. Qty
Description
Manufacturer/retailer part number
Price (e) Avg. Unit
Avg. Total
3
0.1 μF electrolytic capacitor
MCRH100V104M5X11
0.11
0.33
2
10 μF electrolytic capacitor
ECA1EAK100X
0.22
0.45
3
10 k resistor
MCCFR0W8J0103A20
0.02
0.07
1
16 MHz crystal
A-16.000-18
0.44
0.44
1
28-pin microprocessor socket
1-2199298-9
0.73
0.73
1
ATMega328p
ATMEGA328P-PU
2.93
2.93
4
220 resistor
CFR25J220R
0.05
0.20
2
22 pf ceramic capacitor
S220K25SL0N63L6R
0.39
0.78
2
4.7 k resistor
LR0204F4K7
0.04
0.08
1
DC input jack
DCJ200-10-A-K1-K
0.70
0.70
1
LM7805
MC7805ACTG
0.90
0.90
12
Molex 2 pins 180° header
22-23-2021
0.20
2.44
12
Molex 2 pins 180° receptacle
22-01-3027
0.18
2.15
5
Molex 3 pins 180° header
22-23-2031
0.23
1.13
5
Molex 3 pins 180° receptacle
22-01-2037
0.17
0.87
2
Molex 4 pins 180° header
22-23-2041
0.42
0.85
2
Molex 4 pins 180° receptacle
22-01-2047
0.13
0.25 (continued)
A Cost-Effective Micro-controller Based System
409
Table 2. (continued) Qty
Description
Manufacturer/retailer part number
Price (e) Avg. Unit
Avg. Total
1
Molex 5 pins 180° header
22-23-2051
0.43
0.43
1
Molex 5 pins 180° receptacle
22-01-3057
0.18
0.18
2
Molex 6 pins 180° header
22-23-2061
0.47
0.95
2
Molex 6 pins 180° receptacle
22-01-3067
0.46
0.91
64
KK crimp terminals
08-51-0108
0.08
5.28
1
PCB push button
PHAP5-30VA2B3T2N2
0.13
0.13
4
Momentary (On)-Off panel pushbutton
MCPS25B-3
1.22
4.90
1
Self-locking On-Off panel pushbutton
MCPS25A-3
1.22
1.22
5
Audio jack panel, 3 contacts, 3.5 mm
MJ-355W
2.34
11.72
4
Circular connector, for panel, > 3 ways
PSG01593
1.50
6.01
4
Circular connector, for cable, > 3 ways
PSG01590
1.51
6.02
1
Circular connector, for panel, > 5 ways
PSG01594
0.96
0.96
1
LED 5mm
C503B-RAN-CZ0C0AA1
0.16
0.16
1
Panel LED holder, 5mm
CLP127BLK
0.61
0.61
1
ADS1115 module
ADAFRUIT 1085
13.46
13.46
1
DS3231 module
DS3231 + AT24C32
7.21
7.21
1
microSD card module
PTR006540
4.64
4.64
1
16 × 2 LCD screen I2C module
EF16A0036OK
8.12
8.12
4
ADXL 335 breakout board
PTR004892
15.02
60.09
1
DHT 22 module
PTR002332
10.65
10.65
5
24 AWG cables (meter)
PP002606
1.61
8.07
1
PCB wafer
-
26.34
26.34
1
External box (3D printed)
-
1.61
1.61
Total
e 194.97
410
R. Rocha Ribeiro et al.
4 Validation 4.1 Methodology The validation of the proposed system consisted in performing an EMM-ARM test with a single specimen being measured simultaneously with the previously implemented setup and the proposed system. The previously implemented setup, referred to as “original system”, was comprised of a NI USB-4431 data-acquisition system [6], a PCB 352C34 accelerometer [5], and a portable computer running a LabVIEW software for controlling the devices [7]. The test setup employed was identical to the one presented in Fig. 1. A cement paste, produced with water-to-binder ratio of 0.5 and a CEM II/B-L 32.5 cement [18], was mixed following the EN 196-3:2005 [19] and used to mould the test specimen. The EMM-ARM test started 30 min after the time the cement entered in contact with water, and a fan was used to increase the level of ambient excitation around the test. The original system was configured to sample data at 1000 Hz and the proposed system was configured at its maximum of 860 Hz, both performing 120 s measurement sessions every 5 min. The experimental data was processed with the post-processing software to obtain E-modulus estimates, using the SSI-COV identification method [4] for estimating the natural frequency in each measurement session. 4.2 Experimental Results Figure 6 presents the experimental results obtained with the validation tests. Figure 6a presents the E-modulus evolution. Even though, for better visualization, a continuous line representation is used, the experimental results are, in fact, point estimates associated to each successive measurement session performed. The visual superposition of both curves is an indicative of the validity of the proposed system. A zoomed inspection reveals the proposed systems to have more dispersed results, due to its higher noise. Figure 6b presents estimates of accuracy and precision of both systems in relation to an averaged fitted curve to each data series. This strategy of accuracy and precision assessment is the same as used in Granja [3]. The fitted curve was an exponential law, presented in Eq. (1), that represents well the development of the E-modulus of cementbased materials [3, 20]: τ1 β1
f(t) = α1 · e−
t
(1)
In which α1 , τ1 , and β1 are the fitting parameters, which, respectively, had values of 10.641, 1.069, and 1.092 for the original system, and 10.651, 1.076, and 1.098 for the proposed system. For each system, the relative difference of each experimental data point was calculated in relation to the fitted curve (i.e., the E-modulus difference between fitted and experimental divided by experimental values). For the histogram representation of occurrences (right axis in Fig. 6b), the relative differences were divided in 100 bins in the −0.25 to 0.25 interval. The average and standard deviation of each series of relative differences were used to plot a normal probability distribution curve (left axis in Fig. 6b). It can be observed that, although slightly
A Cost-Effective Micro-controller Based System
(a)
411
(b)
Fig. 6. EMM-ARM results: (a) E-modulus estimates; (b) accuracy and precision.
inferior, the proposed system presented similar precision and accuracy to the original system, considering the sharpness and centralization of its histogram and probability curve around the zero difference value in the x-axis.
5 Conclusions This work presented the development of a cost-effective test system to performing EMMARM tests in cement pastes. The system is disclosed as an open-source hardware (OSH) in GitHub [11], and all details are provided to allow complete replication of the system, including circuit schematics, source code, bill of materials and cost estimation. The proposed system is validated by comparison against the previous implementation of the system (more expensive and less easy to replicate), providing of comparable performance.
References 1. Azenha, M., Magalhães, F., Faria, R., Cunha, Á.: Measurement of concrete E-modulus evolution since casting: a novel method based on ambient vibration. Cem. Concr. Res. 40(7), 1096–1105 (2010). https://doi.org/10.1016/j.cemconres.2010.02.014 2. Azenha, M., Faria, R., Magalhães, F., Ramos, L., Cunha, Á.: Measurement of the E-modulus of cement pastes and mortars since casting, using a vibration based technique. Mater. Struct./Materiaux et Constructions 45(1–2), 81–92 (2012). https://doi.org/10.1617/s11527011-9750-9 3. Granja, J.L.D.: Continuous characterization of stiffness of cement - based materials: experimental analysis and micro-mechanics modelling. Ph.D. thesis, University do Minho (2016) 4. Peeters, B., de Roeck, G.: Stochastic system identification for operational modal analysis: a review. J. Dyn. Syst. Meas. Control Trans. ASME 123(4), 659–667 (2001). https://doi.org/ 10.1115/1.1410370
412
R. Rocha Ribeiro et al.
5. PCB Piezotronics: Model 352C34 - ICP Accelerometer - Revision J (2007). https://www.pcb. com/contentStore/docs/pcb_corporate/vibration/products/specsheets/352c34_j.pdf 6. National Instruments: NI USB-443x Specifications (2018). https://www.ni.com/docs/en-US/ bundle/usb-443x-seri/resource/372485e.pdf 7. National Instruments: What Is LabVIEW? (2022). https://www.ni.com/pt-pt/shop/labview. html. Accessed 21 May 2022 8. Atmel: Data Sheet ATmega328P, pp. 1–294 (2015). http://ww1.microchip.com/downloads/ en/DeviceDoc/Atmel-7810-Automotive-Microcontrollers-ATmega328P_Datasheet.pdf 9. Arduino: Arduino/Genuino Uno Board Anatomy (2019). https://www.arduino.cc/en/guide/ BoardAnatomy. Accessed 06 Aug 2019 10. Silicon Labs: USBXpressTM Family CP2102N Data Sheet (2016) 11. Rocha Ribeiro, R.: microEMMARM repository in GitHub: Synercrete’23 release (v0.1b), GitHub (2022). https://github.com/renr3/microEMMARM/tree/main/V0.1.b-Synercrete23. Accessed 01 Dec 2022 12. Hobby Components: GY-61 ADXL3335 Triple Axis Accelerometer Module (2014). https:// forum.hobbycomponents.com/viewtopic.php?t=1736. Accessed 25 Nov 2022 13. Analog Devices: ADXL335 Accelerometer (2009). https://www.sparkfun.com/datasheets/ Components/SMD/adxl335.pdf 14. Ribeiro, R.R., de Lameiras, R.M.: Evaluation of low-cost MEMS accelerometers for SHM frequency and damping identification of civil structures. Latin Am. J. Solids Struct. 16(7), 1–26 (2019). https://doi.org/10.1590/1679-78255308 15. Arduino: Arduino IDE (2021). https://www.arduino.cc/en/software. Accessed 16 July 2021 16. Carini, M.R., Rocha, M.M.: CESSIPy: a Python open-source module for stochastic system identification in civil engineering. SoftwareX 18, 101091 (2022). https://doi.org/10.1016/j. softx.2022.101091 17. Octopart: Octopart (2022). https://octopart.com/. Accessed 30 Nov 2022 18. CEN: EN 197 -1/ Cement Part 1: Composition, specifications and conformity criteria for common cements (2011) 19. CEN: EN 196-3:2005 +A1:2008 - Methods of testing cement - Part 3: Determination of setting times and soundness (2005) 20. Carette, J.: Towards Early Age Characterisation of Eco-Concrete. Doctoral thesis, Université Libre de Bruxelles (2012)
Analysis of Concrete Transient Thermal Deformation in the Context of Structures Submitted to Various Levels of Temperature and Mechanical Loading Robin Cartier1(B) , Hugo Cagnon1 , Thierry Vidal1 , Jean-Michel Torrenti2 , Alain Sellier1 , and Jérôme Verdier1 1 LMDC, INSA/UPS Génie Civil, Université de Toulouse, 135 Avenue de Rangueil, 31077
Toulouse Cedex 04, France [email protected], [email protected] 2 UMR MCD, Université Gustave Eiffel, Cerema, 77455 Marne-La-Vallée, France
Abstract. During their service life, concrete structures are submitted to various types of boundary conditions among which thermal and mechanical loads are the most common. Thermal variations can be non-negligible in case of specific service conditions of the structures such as the storage of exothermic nuclear waste, or during some accidental conditions. Mechanical loading or moderate temperature can, independently from each other, induce high concrete deformations and cracking leading to a loss of serviceability of the structure. In the case of a simultaneous mechanical loading and heating or a heating of a previously loaded concrete structure, an additional deformation, called Transient Thermal Deformation (TTD), occurs. This phenomenon should be monitored because its kinetics and amplitude are significant. Since many concrete structures may be affected by TTD, investigations are needed to assess the coupled effects of different levels of mechanical loadings and heating on this additional deformation. This study presents a part of a wide experimental program dealing with the containment vessel behaviour of a nuclear power plant in case of a severe accident characterized by an increase in internal pressure and temperature. Concrete specimens were submitted to two levels of sustained stress, 30% and 60% of the compressive strength, using uniaxial compressive creep devices, and heated to various temperatures, 20, 40 and 70 °C under autogenous conditions representative of the core of the structure. TTD and the thermo-mechanical effects were analyzed through the strain evolutions of loaded and unloaded specimens. Keywords: Concrete · Basic Creep · Temperature · Transient Thermal Deformation · Coupling
1 Introduction The work presented in this study was carried out as part of the French national research project MACENA which aims to guarantee the integrity of nuclear reactor containment buildings in the event of a severe accident. In this situation, structural concrete would be © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 A. J˛edrzejewska et al. (Eds.): SynerCrete 2023, RILEM Bookseries 43, pp. 413–424, 2023. https://doi.org/10.1007/978-3-031-33211-1_37
414
R. Cartier et al.
subjected to the coupled effects of a sustained mechanical load and a significant rise in temperature, up to 150 °C. The study of this coupling has shown that it leads to the thermal activation of creep characterized by increased creep strain kinetics [1–3]. Furthermore, when concrete is mechanically loaded then heated, an additional strain called Transient Thermal Deformation (TTD) or transient thermal creep appears [3]. This deformation, which develops rapidly after the concrete is heated, leads to higher creep strains than in the case of a heated and then loaded concrete. However, the occurrence of TTD does not seem to alter the creep kinetics after the initial increase in deformation, resulting in a constant gap between the creep strain curves of a concrete under TTD conditions and of a concrete heated and then loaded [4]. The occurrence of transient thermal deformation was highlighted for both ordinary and high-performance concrete, and it has been shown that this deformation is permanent and can occur only once, even if the favourable conditions are repeated [4]. It has been suggested that the origin of TTD lies in the nanoporous water of C-S-H which expands due to the rise in temperature, resulting into a local overpressure and subsequent movement of water towards the capillary porosity [5]. This temporary displacement of water initially located in C-S-H could weaken the links of their sheets and favour their slidings, leading to their rearrangement and a supplementary contraction deformation. The objective of the current work is to collect data regarding the impact of loading rate and temperature scenario on the TTD and the thermal activation of creep. This was achieved by studying the difference between the basic creep of concrete specimens heated to 40 and 70 °C, under conditions allowing for the occurrence of TTD, and the basic creep of specimens kept at 20 °C. These measurements were conducted at loading rates of 30 and 60% of the compressive strength, which are representative of a common and a more severe loading rate for a concrete structure, respectively. Further understanding of the coupling of mechanical loading with temperature is also needed outside of the scope of nuclear safety. In fact, the conditions necessary for TTD to occur can be achieved for many concrete structures which could lead to excessive unforeseen strains which in turn could result in concrete damage or loss of prestressing in case of prestressed structures, and therefore a loss of serviceability or, worse, a failure.
2 Materials and Methods 2.1 Materials The mix proportions of the concrete chosen for this research project, which is representative of that of containment buildings within the French nuclear infrastructures, are given in Table 1. The samples required for the study came from several batches. To make sure that each batch met the required characteristics, both fresh and hardened concrete samples from each batch were tested. To characterize the fresh state of concrete, slump (Abrams’ cone test) and air content were measured. The slump values ranged from 18 to 23 cm and the air content values from 1.7 to 2.5%. They were therefore consistent between the different batches. The mechanical properties of the hardened concrete were measured on cylindrical specimens of 11 cm in diameter and 22 cm high, randomly selected from the set of specimens that would be used in the experimental program. The values of the
Analysis of Concrete Transient Thermal Deformation
415
Table 1. Mix proportions of the concrete Mix proportions (kg/m3 )
Materials Siliceous limestone rolled sand 0/4
830
Siliceous limestone rolled aggregate 4/11
445
Siliceous limestone rolled aggregate 8/16
550
Cement CEM I 52.5 N CE CP2 NF Gaurain
320
Superplasticizer Techno-80 SikaPlast
2.4
Effective water (Water/Cement = 0.52)
167
compressive strength, Young’s modulus and splitting tensile strength were determined according to the European standard EN 12390. The average values of these test measured 28 days after casting are given in Table 2. Table 2. Mechanical properties of the concrete 28 days after casting Compressive Strength (MPa)
Young’s modulus (GPa)
Split tensile strength (MPa)
46.1 ± 1.2
35.1 ± 0.6
3.9 ± 0.2
The values shown in Table 2 are in line with the expected level of performance for this concrete mix, as shown by previous works within the same project, and the low dispersion shows a good homogeneity between the different batches. 2.2 Experimental Program The experimental program consists of a series of uniaxial compressive creep tests performed on concrete cylinders under several Thermo-Mechanical (TM) conditions. The presented results focus on a campaign of tests conducted in autogenous conditions without moisture exchange. Cylindrical concrete specimens of 11 cm in diameter and 22 cm high were kept in a curing chamber maintained at a relative humidity of 99% and a temperature of 20 °C for 1 day after casting and were then demoulded. The specimens were then wrapped in aluminum foil to prevent moisture exchange and cured under autogenous conditions at 20 °C for a minimum of 2 months before testing to stabilize the hydration of cement. They were kept under autogenous conditions throughout the test program to be representative of the concrete at the heart of a massive structure that will not be affected by moisture transport during the service life of the element [6]. Autogenous shrinkage at 20 °C was recorded on 3 specimens, using a retractometer, from demoulding up to approximately 400 days.
416
R. Cartier et al.
Fig. 1. Evolution of Thermo-Hydro-Mechanical conditions for each uniaxial compressive creep and shrinkage test, t0 date of loading of creep samples
At the end of the curing period, the samples were then subjected to various TM conditions as shown in Fig. 1. They were divided into 3 groups. Groups 2 and 3 were placed in 2 different climatic chambers for thermal testing while the first group was kept in a room at 20 °C. Each location was equipped with several uniaxial compressive creep devices capable of applying the required stress under thermal conditions. In addition, some specimens from each group were kept unloaded throughout the experimental campaign to allow the monitoring of free strains and mass evolutions. The TM tests began by loading the creep specimens in each group to 30 or 60% of the average compressive strength of the concrete. After one day, the specimens from group 2 and 3, stored in the climatic chambers, were heated to 40 °C and 70 °C respectively at a moderate rate of 0.1 °C/min to avoid damage by limiting the thermal gradient between the core and surface of the specimens, as recommended by a previous study [1]. The decision to load the samples before heating was made to be representative of accidental conditions in a nuclear containment building. After heating, the stabilized TM conditions were maintained for approximately a month and a half before mechanical unloading. The heated specimens were then freely cooled to a temperature of 20 °C and the creep recovery phase was monitored for approximately one week. The longitudinal strain of the group 1 specimens, which were kept at 20 °C throughout the experimental campaign, was recorded using LVDT sensors embedded at the center of the concrete cylinders. For samples heated at 40 and 70 °C, it was not possible to use LVDT sensors due to the intrinsic thermal expansion of the instrument. Hence, the longitudinal strain of each heated cylinder was recorded using 3 strain gauges bonded at mid-height on the concrete surface, on the cylinder generators angularly spaced by 120 °C. The intrinsic thermal deformation of the gauges was compensated by tracking the deformation of an additional gauge bonded to a titanium silicate sample selected for its negligible coefficient of thermal expansion (CTE).
Analysis of Concrete Transient Thermal Deformation
417
3 Results and Discussion For all the results, expansion is positive and contraction is negative. 3.1 Free Strains in Autogenous Conditions Free Strains. Autogenous shrinkage and mass loss of sealed specimens at 20 °C were monitored for almost 400 days. The autogenous shrinkage evolution is consistent with other studies on concrete of similar compressive strength class with a value of less than 100 µm/m approximately one year after casting. Since the autogenous shrinkage values remain below 50 µm/m during curing and are relatively small compared to the total deformation, the shrinkage between demoulding and the beginning of the TM test is neglected in this study. Therefore, in the following figures, the shrinkage will start from 0 at the beginning of the TM test. In addition, mass tracking reveals that sealing was effective for almost 400 days, since the mass loss always remained under 0.1%. During the TM test, group 2 and 3 samples were heated to 40 °C and 70 °C while group 1 samples remained at 20 °C. Therefore, these free strains values include both autogenous shrinkage and thermal expansion and contraction (Fig. 2).
Fig. 2. Free strains evolution for unloaded specimens in sealed conditions at the three test temperatures of 20, 40 and 70 °C
Concrete undergoes thermal expansion when heated to 40 °C and 70 °C and contracts when cooled to 20 °C after 52 days. As expected, the samples heated to 70 °C undergo more intense thermal deformation than those heated to 40 °C. In addition to the direct effect of temperature, these curves also show shrinkage for all samples. As discussed earlier, the shrinkage at 20 °C is small for sealed samples, while it is significant for samples heated at 40 °C and even more for those heated at 70 °C. This can be explained by a loss of sealing due to the temperature increase. The mass loss, negligible at 20 °C, respectively reached 0.6% and 3% at the end of the test at 40 °C and 70 °C. As a result, the free strain curves show a parasitic desiccation shrinkage.
418
R. Cartier et al.
Coefficients of Thermal Expansion. The Coefficients of Thermal Expansion (CTE) were determined by calculating the strain difference before and after the heating, and by the same method during the cooling phase. The four CTE values calculated for the test temperatures of 40 °C and 70 °C both during heating and cooling are shown in Fig. 3.
Fig. 3. Coefficients of thermal expansion for unloaded sealed samples heated to 40 and 70 °C
For the heating phase, both coefficients are very close, 12.2 and 12.4 µm/m/°C for 40 °C and 70 °C, respectively, and within the upper range of the values expected for a mature concrete [7, 8]. The CTE is therefore constant with temperature up to 70 °C. Comparison of the heating and cooling phases shows that the CTE is lower during cooling, with this effect being steeper for the specimens kept at 70 °C. A possible explanation for this decrease in CTE could be the parasitic loss of water, especially for specimens heated to 70 °C due to the high CTE of water. It should also be noted that the kinetics during heating and cooling are different, respectively 0.1 °C/min and unmonitored.
3.2 Total Deformation and Basic Creep Strain Total Deformation During TM Test. Fig. 4 shows the evolution of total strain under sustained loading in sealed conditions, for each loading rate and each temperature tested, from the date of loading. Note that the specimens kept at 20 °C were not unloaded at the same time as the others in order to collect additional data on basic creep at 20 °C over a very long period. Upon loading, the elastic deformation was recorded and the deduced Young’s modulus values were found to be close to those measured at 28 days after casting (Table 3). After one day of compressive creep, heating is applied. Thermal dilation is observed at both temperatures, but the amplitudes are less pronounced than those of the unloaded samples (Fig. 2), because the expansion competes with TTD which occurs
Analysis of Concrete Transient Thermal Deformation
419
simultaneously. As expected, the magnitude and kinetics of deformation also increased with loading rate and temperature.
Fig. 4. Evolution of total deformation under load at 30 and 60% of compressive strength for the three tested temperature 20, 40 and 70 °C
At the end of the test, during unloading at the test temperature, elastic deformation is observed, followed by a reversible visco-elastic strain evolution for one week, and finally by a thermal contraction upon cooling at 20 °C. Table 3. Young’s modulus measured upon loading (GPa) 20 °C
40 °C
70 °C
Loaded at 30% of fcm
35.0
35.6
35.5
Loaded at 60% of fcm
35.9
35.1
37.5
Basic Creep Strains. Basic creep curves were obtained by removing elastic deformation and free strains from the total strain under load values. This analysis method is based on the assumption that the compressive loading does not change the CTE value. The resulting curves are shown in Fig. 5.
420
R. Cartier et al.
Fig. 5. Evolution of basic creep strains under load at 30% and 60% of compressive strength for the three tested temperatures 20 °C, 40 °C and 70 °C
Effects of Loading Rate and Temperature. These results highlight the effects of temperature on the basic creep strains. For these Thermo-Mechanical chronologies of test conditions, the following effects of temperature on concrete behaviour occur: the thermal dilation which develops only during the heating, the thermal activation of creep, and the Transient Thermal Deformation (TTD) since the samples are loaded before heating. By subtracting the free strain, thermal dilation has been removed from these plots. Temperature also heavily impacts basic creep strains as even a temperature of 40 °C can drastically increase the extent of basic creep while a temperature of 70 °C is shown to multiply it by a factor of 5 even at a moderate loading rate of 30% as shown in Fig. 5. To further analyse the effect of loading rate on basic creep strains, the strain values are divided by the applied stress level to obtain specific creep values. The resulting curves are shown in Fig. 6. These results highlight the non-linearity of the basic creep between the 30% and 60% loading rates for the 20 °C specimens. This result is in agreement with the findings of other researchers who showed linearity between creep strains and loading rate up to a loading corresponding to 45% of the compressive strength for an ordinary concrete [9] or between 30% and 50% for a high-performance concrete [10]. A loading rate of 60% is likely to induce creep damage within the material which could cause this non-linearity. The same conclusion can be drawn from the curves at 70 °C and the effect is even more significant at 40 °C.
Analysis of Concrete Transient Thermal Deformation
421
Fig. 6. Evolution of specific basic creep for every combination of loading rate and tested temperature
3.3 Coupled Effects of Temperature and Loading Rate As noted above, the TM chronology test conditions, which are representative of accident conditions for a nuclear containment vessel, induce creep thermal activation and TTD. TTD develops rapidly after heating but does not affect the creep rate a posteriori, which means that the creep curves of the specimens with TTD will show a consistent offset with those of specimens heated then loaded [4]. In the present study, the TTD cannot be completely decoupled since no tests were performed on concrete heated and the loaded to evaluate the single thermal creep activation. Therefore, we will study the offset between the curves of basic creep at 20 °C and the ones at 40 °C or 70 °C. It should then be noted that this difference between the two creep curves will be the combination of the steady shift due to TTD and the increased kinetics of delayed strains caused by the thermal activation of creep. The values of the difference between the basic creep at 40 °C or 70 °C and the basic creep at 20 °C increase with time, confirming the coupling between TTD and thermal activation of creep in our data. We chose to highlight the values of this difference at two dates: – 2 days after loading, not long after the end of heating when the TTD should prevail. – 42 days after loading, just before unloading when TTD and thermal activation are fully coupled. The choice of studying the value at 2 days was made in accordance with the findings of Manzoni et al. who proposed a law to evaluate the characteristic time required for the development of the TTD by the evacuation of nanoporous pressure [5]. Putting this relationship into practice for temperatures of 40 °C and 70 °C leads to a characteristic time of up to 0.7 days between the end of the heating phase and the complete development
422
R. Cartier et al.
of the TTD strains. In our study, heating takes place until 1.3 days after the compressive loading, hence the choice to investigate the values of basic creep 1.3 + 0.7 = 2 days after loading. This is also in agreement with the other experimental findings showing that TTD develops quickly during and after heating of a concrete under load [4]. The values of basic creep at 2 and 42 days after loading are given in Table 4 and Table 5. Table 4. Values of basic creep at 2 and 42 days after t0 for a loading rate of 30% (µm/m) 20 °C
40 °C
70 °C
2 days after loading
−44
−229
−561
42 days after loading
−188
−697
−942
Table 5. Values of basic creep at 2 and 42 days after t0 for a loading rate of 60% (µm/m) 20 °C
40 °C
70 °C
2 days after loading
−305
−641
−984
42 days after loading
−841
−2021
−2310
Effects of Loading Rate and Heating Temperature. To investigate the effect of the loading rate and the temperature scenario on basic creep development, the difference between the basic creep at 40 or 70 °C and the basic creep at 20 °C was calculated for both the loading rates of 30 and 60% of fcm , and at 2 and 42 days after compressive creep loading. The values obtained are shown in Fig. 7 and 8. First, the difference between the basic creep of heated and unheated specimens increased between 2 and 42 days, which can be explained by the increased basic creep strain kinetics for heated concrete, leading to an increasing gap between the two curves. Focusing on the results obtained 2 days after loading, when the TTD should prevail over the thermal activation of creep, shows that TTD is influenced by both the level of stress applied and the heating temperature, in accordance with the hypothesis on the origin of these deformations proposed by [5]. Neither of these relationship seems to be linear, since doubling the applied stress leads to strain increases of 82% and 32% at 40 °C and 70 °C respectively, while multiplying temperature increase by 2.5 times between + 20 °C (40–20 °C) and +50 °C (70–20 °C) leads to an increase of strains of 179% and 102% under loading rates of 30% and 60% of fcm , respectively. It can also be noted that the effect of temperature on TTD seems to be higher for the loading rate of 30% while the effect of the loading rate is higher for a concrete heated only to 40 °C, both at 2 and 42 days. This could be explained by the fact that TTD is a potential deformation activated by both stress and temperature, meaning that this additional strain can be triggered either by heating at a high temperature under moderate loading or the opposite scenario.
Analysis of Concrete Transient Thermal Deformation
423
Fig. 7. Difference between basic creep at 40 and 70 °C, and the basic creep at 20 °C for loading rates of 30 and 60% of fcm , 2 days after loading
Fig. 8. Difference between basic creep at 40 and 70 °C, and the basic creep at 20 °C for loading rates of 30 and 60% of fcm , 42 days after loading
The juxtaposition of the results at 2 and 42 days seems to show an interesting trend where temperature appears to be more impactful at 2 days, when the TTD is predominant, while the level of stress applied has more influence on the results at 42 days, when the thermal activation of creep dominates.
424
R. Cartier et al.
4 Conclusion To obtain data on the effect of loading rate and temperature scenario on the Transient Thermal Deformation (TTD) and thermal activation of creep, concrete specimens kept in autogenous condition were loaded using compressive creep devices then heated in order to allow the development of TTD. Two stress levels, 30 and 60% of the compressive strength and 3 temperature 20 °C, 40 °C and 70 °C were applied. Basic creep strains were measured for each scenario for approximately 50 days. The non-linearity of basic creep for higher rate of loading was verified and the effects of temperature on the delayed strains were highlighted. Data on the effect of the stress level and temperature level on TTD and thermal activation of creep were collected at 2 and 42 days after loading. The influence of these two parameters was measured, highlighting the non-linearity of their relationship with TTD for the range of values tested. Finally, it was shown that TTD can lead to non-negligible additional basic creep strains even for common levels of stress and temperature, demonstrating the need for further understanding and awareness of this phenomenon.
References 1. W Ladaoui T Vidal A Sellier X Bourbon 2011 Effect of a temperature change from 20 to 50°C on the basic creep of HPC and HPFRC Mater. Struct. 44 1629 1639 https://doi.org/10. 1617/s11527-011-9723-z 2. S Arthanari CW Yu 1967 Creep of concrete under uniaxial and biaxial stresses at elevated temperatures Mag. Concr. Res. 19 149 156 https://doi.org/10.1680/macr.1967.19.60.149 3. HM Fahmi M Polivka B Bresler 1972 Effects of sustained and cyclic elevated temperature on creep of concrete Cem. Concr. Res. 2 591 606 https://doi.org/10.1016/0008-8846(72)901 13-5 4. H Cagnon T Vidal A Sellier X Bourbon G Camps 2019 Transient thermal deformation of high performance concrete in the range 20 °C–40 °C Cem. Concr. Res. 116 19 26 https://doi. org/10.1016/j.cemconres.2018.11.001 5. F Manzoni T Vidal A Sellier X Bourbon G Camps 2020 On the origins of transient thermal deformation of concrete Cem. Concr. Compos. 107 103508 https://doi.org/10.1016/j.cemcon comp.2019.103508 6. R Mensi P Acker A Attolou 1988 Séchage du béton: analyse et modélisation Mater. Struct. 21 3 12 https://doi.org/10.1007/BF02472523 7. EJ Sellevold Ø Bjøntegaard 2006 Coefficient of thermal expansion of cement paste and concrete: mechanisms of moisture interaction Mater. Struct. 39 809 815 https://doi.org/10. 1617/s11527-006-9086-z 8. CR Cruz M Gillen 1980 Thermal expansion of Portland cement paste, mortar and concrete at high temperatures Fire Mater. 4 66 70 https://doi.org/10.1002/fam.810040203 9. MM SmadiI FO Slate 1989 Microcracking of high and normal strength concretes under short and long-term loadings MJ 86 117 127 https://doi.org/10.14359/2264 10. N Ranaivomanana S Multon A Turatsinze 2013 Tensile, compressive and flexural basic creep of concrete at different stress levels Cem. Concr. Res. 52 1 10 https://doi.org/10.1016/j.cem conres.2013.05.001
Semi-circular Bending Test to Evaluate the Post Cracking Behaviour of Fibre Reinforced Concretes Pedro Paulo Martins de Carvalho1(B)
and Rodrigo de Melo Lameiras2
1 Center of Exact Sciences and Technologies, Federal University of West of Bahia (UFOB),
Barreiras, Brazil [email protected] 2 Department of Civil and Environmental Engineering (ENC), University of Brasília (UnB), Brasília, Brazil
Abstract. The known tests to evaluate the post-cracking behaviour of fibre reinforced concretes (FRC) have some disadvantages. The most representative test of tensile behaviour, the direct tensile test (DT), is difficult to perform. Therefore, the use of indirect tests to evaluate the post cracking behaviour of FRC is common. Some of these tests are carried out in such a way that the samples are molded separately to the structure in prismatic shapes. This geometry makes it difficult to extract samples from the real structure to evaluate the behaviour of the FRC in the structure. Especially for the fibre reinforced self-compacting concretes, the fibre distribution and directions are strongly influenced by the shape of the structure and casting procedures, being important to use structural specimens, representative of the actual structure that the FRC is intended to be applied. This work proposes the use of a semi-circular bending (SCB) test for the characterization of post cracking behaviour of FRC. An experimental program was performed to demonstrate the use of the SCB test to characterize steel fibre reinforced self-compacting concretes (SFRSCC). Samples with small dimensions were extracted by means of a concrete core cutting machine. The results of the SCB tests were presented, and an inverse analysis was performed to obtain the tension behaviour of the SFRSCC. The results, obtained in terms of stress versus crack opening relationship, were compared with results obtained by DT tests. The results showed the potential of SCB test to evaluate the post-cracking behaviour of FRC, and the evaluated test seems to be particularly suitable for structural test specimens, extracted directly from the structure. Keyword: Semi-circular Bending Test · Residual Strength · Post Cracking Behaviour · Steel Fibre Reinforced Concrete · Self-compacting Concrete
1 Introduction The Steel Fibre Reinforced Concretes (SFRC) and Steel Fibre Reinforced SelfCompacting Concretes (SFRSCC) holds particular characteristics that are relevant from a structural point of view, such as the strength capacity after the cracking. This composite © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 A. J˛edrzejewska et al. (Eds.): SynerCrete 2023, RILEM Bookseries 43, pp. 425–435, 2023. https://doi.org/10.1007/978-3-031-33211-1_38
426
P. P. M. de Carvalho and R. Melo de Lameiras
can be used as an alternative of the reinforced concrete in some structures, and is already widely used in industrial pavements and tunnels [1]. However, there is still a significant demand for the improvement of tests capable of accurately determining the structural performance of this type of concrete. The bending tests, as the three point bending test (3PBT) [2–4] and the round panel test (RPT) [5] require the use of samples with large dimensions, in the order of 500 mm, whose handling makes it difficult to perform. Tests that use smaller samples have already been studied, such as the DEWS (Double Edge Wedge Splitting) Test, proposed by [6]. On this test, the maximum dimension of samples is approximately 100 mm. However, the samples in this test are prismatic. This makes it difficult to extract cores from real structures. The knowledge of post-cracking behaviour of SFRC and SFRSCC is very important for the correct use of these materials. The direct tensile test represents very well this behaviour. However, this test is very difficult to perform. In addition, care must be taken to ensure that the specimen is perfectly aligned to the loading direction. The concrete sample must have a suitable geometric shape. Sample bending must be avoided [7]. This work proposes the application semi-circular bending (SCB) test to obtain the post-cracking behaviour of SFRSCC. Four samples of the same type of concrete were tested by the proposed test configuration. Also, a methodology for carrying out the inverse analysis procedure based on the test results was applied to the test results and the direct tensile behaviour of concrete was estimated. An inverse analytical analysis procedure based on constitutive relationships, and usually applied for 3PBT specimens, was adapted for SCB specimens and applied to estimate the direct tensile behaviour of the tested SFRSCC. Furthermore, the effect of using three different values for the nonlinear hinge length in the inverse analysis procedure was investigated.
2 Semi-Circular Bending Test (SCB) Semi-circle-shaped sample test (Fig. 1) is usually employed to determine the mode I fracture toughness of rocks [8]. The test has advantages due to its simplicity, speed of execution, ease of sample preparation, with applicability in quality control [9]. The SCB test samples have reduced dimensions. On the test, the samples are positioned on two supports close to the ends. The load is applied at the centre of the span inducing bending behaviour. A notch is usually made in the middle of the specimen to induce the appearance of a single crack. Vertical displacement and load are monitored over time. These characteristics make the test very similar to the 3PBT. The literature [10] used this type of sample to investigate the effect of loading rate effect on fracture behaviour of fibre reinforced high strength concrete. The properties were studied: ductility, energy absorption, and loading capacity and fibre of the type glass, polypropylene, and steel were used. Among the conclusions, the author highlights the possibility of rapid production of samples and the low consumption of material, which is due to the small size of the sample. In addition, due to the reduced dimensions, it is possible to highlight the possibility of the samples to be extracted by means of a concrete core cutting machine of larger structures. This important advantage, combined with the low material consumption, allows the evaluation of the behaviour of the SFRSCC in the real structure.
SCB Test to Evaluate the Post Cracking Behaviour
427
Fig. 1. Sample illustration of SCB test.
2.1 Inverse Analysis Knowledge of the behaviour of SFRSCC after the first crack is very important. This occurs because after this moment, the fibres begin to act, contributing with strength capacity in this state, implying an ascending stress-strain curve after the first crack [11]. On the indirect tests, as the 3PBT, RPT and SCB and RPT, the sample section is not fully under tensile, but predominantly bending occurs [2] and [5]. Therefore, the results obtained through these tests do not directly represent the tensile behaviour in which the fibers act. So, there is a need for a special analysis of the results of these tests so that tensile parameters can be obtained. This procedure is called inverse analysis. It is the use of analytical or numerical models that consider constitutive relationships to determine stress (σ) – crack width (w) diagrams which represent the tensile behaviour of the studied FRC. The following steps can be performed [12]: – Introduction of input data such as the characteristics of the sample (height, width, length and depth of the notch) and of the concrete (compressive strength and tensile strength of the SFRSCC); – Numerical simulation or analytical approach that provides the force × CMOD response; – Analysis of the accuracy of the response obtained in the previous step; – The fourth level is the output level, in which the parameter values that lead to the smallest error and meet the restrictions imposed for the load limit value and corresponding deflection are obtained. The error can be calculated by comparing the analytical/numerical solution with the experimental one. When carrying out the inverse analysis procedure, it is very usual to use nonlinear hinge models, in which the application of constitutive models for the deformed configuration of the sample during the test is considered. A nonlinear hinge length is usually considered fixed. In the case of 3PBT, the use of values H /2, H and 2H for this length are usual [2, 13–17]. In this work, a study of an inverse analysis procedure for the SCB test is carried out, evaluating the modification of this parameter.
428
P. P. M. de Carvalho and R. Melo de Lameiras
3 Methodology 3.1 Materials and Dosage Definition The dosage of concrete in the samples of this work is represented on Table 1. MACCAFERRI metallic fibers of the type ST-33/44 (FS3N) with a diameter 0.75 mm and a length 33 mm (fibre aspect-ratio: 44) were used. Table 1. Mix proportions of SFRSCC. Cement (kg)
Aggregate (kg)
Sand (kg)
Water (kg)
Superplasticiser (kg)
Weight of fibers (%)
13.75
30.39
17.87
4.69
0.04
0.5
Cylindrical testimonies were extracted from prismatic parts of a specimen previously used to perform 3PBT (Fig. 2a). The 3PBT samples had a central notch of 25 mm and dimensions of 150 × 150 × 550 mm. A bending rupture was performed on the samples (Fig. 2b). The extraction process of SCB specimens was carried out by means of a concrete core cutting machine (Fig. 2c) in each part of the 3PBT specimen. After that, the cylinders were cut in the diametrical direction, aligned to the reference axis, to obtain two samples with a semi-circular section form (Fig. 2d). Then, notches were made with a depth of 10 mm (Fig. 2e). In cases where the samples had an irregular bottom surface, the surface was carefully smoothed to keep it closer to the flat shape and thus avoid stress concentration near the supports. A real illustration of the finished sample is presented (Fig. 2f). From each 3PBT specimen, 4 SCB specimen were obtained. The final SCB sample had height: H = 45 mm and length: LT = 90 mm, as presented in Fig. 2e. The width of the sample corresponds to the height of the extracted cylinder (Fig. 2c), which is b = 150 mm. The notch was 10 mm deep and 2.6 mm wide. The measurements were checked with a caliper and an average was calculated. The maximum deviation from the average values was of the order of 2%. The test setup performed (Fig. 3) was similar to the 3PBT. The sample was supported by two cylinders with a diameter of 12 mm, separated by 80 mm and welded to a metallic support. Another support with a welded cylinder was positioned on top of the sample and over it a load cell with a capacity of 500 kN was positioned. A loading machine with a capacity of 300 kN was used and a constant displacement rate of 0.2 mm/min was established for the test. Two displacement transducers (Linear Variable Differential Transformer - LVDT) with precision of 0.001 mm were used to monitoring the CMOD (Crack Mouth Opening Displacement), one in the front and other in the back face of specimen. The monitoring frequency is 2 Hz. The load cell and the transducers were connected to the same acquisition system so that the monitoring of both parameters was carried out synchronously. The transducers were fixed using screws to aluminium profiles that were glued to the sample using high-adherence glue. An average force × CMOD curve was obtained for each specimen by averaging both CMOD measured in the front and back face of SCB specimen.
SCB Test to Evaluate the Post Cracking Behaviour
429
Part of prism
Part of prism
3PBT notch
(a)
(b)
Part of prism Reference Axis
Sample
hole (c)
(d)
H
LT
SCB notch (e)
Fig. 2. Schematic representation of the process of obtaining samples for the SCB test: (a) initial prismatic 3PBT sample, (b) parts of the prismatic sample after rupture by flexure, (c) extraction of SCB sample from the parts of the prism, (d) cut in the diametral direction of the cylinder, (e) specimen of semi-circular section ready for testing.
3.2 Inverse Analysis Procedure The objective of using the inverse analysis procedure is to obtain the tensile behaviour of the FRC whose σ - w law can be trilinear, quadrilinear, or others. The methodology for the inverse analysis procedure used in this work was model-based [17] and a quadrilinear σ - w opening law was considered for the SFRSCC, which was originally developed for the 3PBT. The main adjustments to transform the inverse analysis algorithm developed for 3PBT so that it can perform the procedure into SCB results were the differentiated geometry of the sample (height, width, length and notch). In this work, the experimental
430
P. P. M. de Carvalho and R. Melo de Lameiras
Load cell
LVTD 2 LVTD 1
Fig. 3. Semi-circular Bending Test setup and instrumentation.
curve experienced in the SCB test is used as a reference to calculate the error and the parameters are chosen to minimize the error between the analytical curve, obtained through inverse analysis, and the experimental curve. [17] used the fixed value equal to the height of the sample (H ) for the nonlinear hinge length (Lh = H ). Three values for this parameter were used in this work (Lh1 = H /2, Lh2 = H , Lh3 = 2H ) with the objective of investigating the effects of its variation. The deduction of the sample geometry after deformation was deduced from the imposition of a rotation θk for every interaction k. θk is obtained by multiplying k by a small constant rate of rotation (θk ). The stretching Dik in a position layer i and deep di was calculated through of Eq. (1). Dik = di − dna θk
(1)
where dna is the position of the neutral line. Each layer considered has a thickness t. The sample width is b. Therefore, the section area of the layer is determined as Eq. (2). A = tb
(2)
SCB Test to Evaluate the Post Cracking Behaviour
431
Applying constitutive models based on stretching Dik , it was determined interactively the opening of cracks wik . Therefore, it was possible to determine Fk for compression or tensile in each layer, as Eq. (3). Fik = σ wik A
(3)
where σ is the compressive or tensile stress depending on the position of the layer. The balance of the section was checked by performing the sum of forces Fik on every layer for every interaction k. The moment-rotation was determined based on the equation of moment equilibrium of the cross section, considering the length of the plastic hinge Lh . Since hooked-end fibers were used, the compression model of [18], modified by [19] for steel fibre reinforced concretes was used. For the tensile behaviour, the multilinear model of [20] was used. The deflection δk was determined analytically based on the principle of virtual work, considering a portion corresponding to elastic deformation and another to plastic deformation, as Eq. (4).
⎛ L − Lh ⎞ ⎧ ⎡ F⎜ ⎪ L 2 ⎟⎠ ⎪ ⎢⎢ .⎨ δk = ⎝ L Lh − 6 Ec I 2 ⎢ ⎛ ⎪ ⎪ ⎢⎣ ⎜⎝ 2 ⎩ 3
3 ⎫ ⎤ ⎪ ⎥ ⎛I ⎞ Lθ ⎥ + ⎜ 1 − 1⎟ ⎬⎪ + k ⎞ ⎥ ⎝ I2 ⎠⎪ 4 ⎟⎥ ⎪ ⎠⎦ ⎭
Elastic deflection
(4)
Plastic deflection
where L is the sample span and Ec is the secant modulus of elasticity of concrete. I1 and I2 are the moment of inertia of the section without and with notch, respectively, as Eq. (5) and Eq. (6). bH 3 12
(5)
b(H − e)3 12
(6)
I1 = I2 =
where e is the notch depth. The tensile-stress versus crack opening curve is obtained through literature [17]. Parameters wki are modified interactively in order to minimize the error between the experimental and analytical force-CMOD response according to a tolerance established by trials. The solution involves solving iteratively the Eq. (7): i) f − σ (w ct ct k − εct,p Lh − wki = 0 (7) Dki + Ec
432
P. P. M. de Carvalho and R. Melo de Lameiras
4 Results and Discussions 4.1 Experimental Results From the responses of the force × CMOD diagrams of the 4 tested samples, a data processing was carried out so that a single representative curve, the average curve, could be determined. For each increment of CMOD in the amount of 0.01 mm, ranging from 0 to 6 mm, an average value of force recorded by the machine during the test was determined for each test result and the average, upper bound and lower bound values of force were obtained. On Fig. 4a and on Fig. 4b results are presented for CMOD in the ranges of 0–6 mm and 0–0.2 mm, respectively. 4
4
Average Force (kN) Lower Bound (kN) Upper Bound (kN)
Average Force (kN) Lower Bound (kN) Upper Bound (kN) 3
Force (kN)
Force (kN)
3
2
1
2
1
0 0
1
2
3
4
5
0 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18 0.20
6
CMOD (mm)
CMOD (mm)
(a)
(b)
Fig. 4. Experimental force × CMOD response for the SCB specimens: average, upper bound and lower bound: (a) CMOD: 0–6 mm and (b) Detail. CMOD: 0–0.2 mm.
3.0
3.0
Fk for Lh=H/2 (kN) Fk for Lh=H (kN) Fk for Lh=2H (kN) Average Fk experimental (kN)
2.5
2.0
2.0
Force (kN)
Force (kN)
Fk for Lh=H/2 (kN) Fk for Lh=H (kN) Fk for Lh=2H (kN) Average Fk experimental (kN)
2.5
1.5
1.5
1.0
1.0
0.5
0.5
0.0 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18 0.20
CMOD (mm)
(a)
0.0 0
1
2
3
4
5
CMOD (mm)
(b)
Fig. 5. Analytical response (Force-CMOD) obtained through inverse analysis and average experimental response for CMOD: (a) 0–0.2 mm and (b) 0.2–5 mm.
SCB Test to Evaluate the Post Cracking Behaviour
433
4.2 Inverse Analysis Results The force-CMOD diagram obtained through the inverse analysis procedure using value Lh fixed in values H /2, H e 2H are shown for the CMOD range of 0–0.2 mm (Fig. 5a) and for the CMOD range of 0.2–5 mm (Fig. 5b). The experimental result is also plotted for comparison purposes. The inverse analysis procedure is performed only for the average response. Results indicate that there are no significant difference when applying the inverse analysis procedure with different values for Lh . This fact can be verified by different authors [2, 13–17], who performed the inverse analysis procedure successfully from the different values of Lh studied in this work. The Table 2 illustrates the value of the areas under the force × CMOD diagrams as well as the percentage difference between the analytical responses and the experimental response. Table 2. Area under force × CMOD diagrams and difference from experimental response. Lh = H /2
Lh = H
Lh = 2H
Experimental answer
Area (kN•mm)
7.34
7.35
7.33
7.29
Difference from the experimental response (%)
0.7
0.8
0.5
-
The inverse analysis procedure provides as a response the σ - w values that represent the tensile behaviour of the studied SFRSCC (Fig. 6). The values of the defining parameters of the σ - w diagram are presented. The results showed good agreement with each other, indicating low variation when Lh is changed considering the values used by the literature (H /2, H , 2H ). This shows a low influence of this parameter on the inverse analysis procedure. This reflects agreement between the force × CMOD and σ w results of the model proposed in this work and a potential use of the tool for the SCB test (Table 3). Table 3. Parameters of the σ - w diagram after inverse analysis procedure. Lh
fct [MPa]
w1 [mm]
σ1 [MPa]
w2 [mm]
σ2 [MPa]
w3 [mm]
σ3 [MPa]
wu [mm]
H /2
2.5
0.21
2.14
1.43
2.03
2.79
0.86
4.12
H
2.5
0.21
2.07
1.36
1.98
2.79
0.92
4.12
2H
2.5
0.21
2.04
1.36
1.99
2.79
0.85
4.12
434
P. P. M. de Carvalho and R. Melo de Lameiras 3
Tensile Stress (MPa)
Results for Lh=H/2 Results for Lh=H Results for Lh=2H 2
1
0 0.0
0.5
1.0
1.5
2.0
2.5
3.0
Crack Opening (mm)
Fig. 6. σ - w results for the analytical responses obtained through the inverse analysis procedure of this work.
5 Conclusions In this work, SCB tests were performed on samples for a given SFRSCC. The results of the experimental force × CMOD diagrams were obtained and compared with the analytical responses obtained through inverse analysis. Results of an inverse analysis procedure for the semi-circular sample bending test with samples moulded in SFRSCC were presented. A usual model from the literature for the inverse analysis of 3PBT test was adapted and used for SCB test. The variation of the length of the plastic label in three different values was evaluated. The comparison between experimental data and the results of the inverse analyses showed that the methodology was satisfactory for determination of the constitutive tensile σ - w opening relationship of SFRSCC. All non-linear hinge lengths employed showed good compatibility with the experimental results. The selected values were compatible with the literature for the three-point bending test. Despite this, this is a preliminary study. Further tests and experimental validation of this procedure are needed, including comparison with tensile tests. Acknowledgements. The authors would like to thank the Universidade de Brasília (UnB) and the Universidade Federal do Oeste da Bahia (UFOB) for their financial and institutional support.
References 1. Figueiredo, A.D.D.: Concreto Reforçado com Fibras. Tese (Livre Docência) - Escola Poitécnica da Universidade de São Paulo. São Paulo (2011) 2. RILEM TC 162-TDF: Test and design methods for steel fibre reinforced concrete. Bending test - final recommendation. Mater. Struct. 35(9), 579–582 (2002) 3. British Standards Institution: Test method for metallic fibered concrete - measuring the flexural tensile strength (limit of proportionality (LOP), residual). London, England. EN 14651 (2005) 4. CNR-DT 204: Guidelines for design, construction and production control of fiber reinforced concrete structures. National Research Council of Italy, Italy (2006)
SCB Test to Evaluate the Post Cracking Behaviour
435
5. Bernard, E.S.: Behaviour of round steel fibre reinforced concrete panels under point loads. Mater. Struct. 3, 181–188 (2000) 6. Di Prisco, M., Lamperti, M.G.L., Lapolla, S.: Double-edge wedge splitting test: preliminary results. In: 7th International Fracture Mechanics of Concrete and Concrete Structures FraMCOS-7, Seogwipo, Jeju, pp. 1579–1586 (2010) 7. van Mier, J.G.M., Mechtcherine, V.: Minimum demands for deformation controlled uniaxial tensile tests. In: PLANAS, J. Report rep039: Experimental Determination of the Stress-Crack Opening Curve for Concrete in Tension. Final report of RILEM Technical Committee TC 187-SOC, Madrid, Spain (2007) 8. Kuruppu, M.D., Obara, Y., Ayatollahi, M.R., Chong, K.P., Funatsu, T.: ISRM-suggested method for determining the mode I static fracture toughness using semi-circular bend specimen. Rock Mech. Rock Eng. 47, 267–274 (2014) 9. Molenaar, A.A.A., Scarpas, A, Liu, X, Erkens, S.M.J.G.: Semi-circular bending test; simple but useful?. In: Proceedings of the Technical Session, pp. 794–815. Asphalt Paving Technology: Association of Asphalt Paving Technologists (2002) 10. Aziminezhad, M., Mardi, S., Hajikarimi, P., Nejad, F.M., Gandomi, A.H.: Loading rate effect on fracture behavior of fiber reinforced high strength concrete using a semi-circular bending test. Constr. Build. Mater. 240, 117682 (2020) 11. Bentur, A., Mindess, S.: Fibre Reinforced Cementitious Composites, 2nd edn. Taylor & Francis Group, Londres e Nova Yorque (2007) 12. Cunha, V.M.D.C.F.: Steel Fibre Reinforced Self-Compacting Concrete (from MicroMechanics to Composite Behaviour). Dissertação (doutoramento em Engenharia Civil) Universidade do Minho. Guimarães, p. 366 (2010) 13. Ulfkjaer, J., Krenk, S., Brincker, R.: Analytical model for fictitious crack propagation in concrete beams. J. Eng. Mech. 1(121), 7–15 (1995) 14. Kooiman, A.G.: Modelling Steel Fibre Reinforced Concrete for Structural Design. Delft University Technology, Delft (2000) 15. Iyengar, K.T.S.R., Raviraj, S., Ravikumar, P.N.: Analysis study of fictitious crack propagation in concrete beams using a bi-linear σ–w relation. In: 3th International Conference on Fracture Mechanics of Concrete and Structure (FRAMCOS III), Japan, pp. 315–324 (1998) 16. Pedersen, C.M.V.: New production processes, materials and calculation techniques for fiber reinforced concrete pipes. Doctoral Thesis, Department of Structural Engineering and Materials. Danmarks Tekniske Universitet. Lyngby, p. 292 (1996) 17. Salehian, H.: Evaluation of the Performance of Steel Fibre Reinforced Self-Compacting Concrete in Elevated Slab Systems; from the Material to the Structure. Tese de doutorado, Universidade do Minho. Guimarães, p. 292 (2015) 18. Vipulanandan, C., Paul, E.: Performance of epoxy and polyester polymer concrete. ACI Mater. J. 3(87), 241–251 (1990) 19. Barros, J.A., Figueiras, J.A.: Flexural behavior of SFRC: testing and modeling. J. Mater. Civ. Eng. 11, 331–339 (1999) 20. Hillerborg, A., Modéer, M., Petersson, P.E.: Analysis of crack formation and crack growth in concrete by means of fracture mechanics and finite elements. Cem. Concr. Res. 6(6), 773–781 (1976)
Development of Damage Monitoring Techniques During Fatigue Compression Test on Concrete Specimen Laura Kerner1(B) , Renaud-Pierre Martin1 , Mezgeen Rasol1 Jean-Claude Renaud1 , and Léopold Denis2
,
1 Univ Gustave Eiffel, MAST-EMGCU, 77454 Marne-la-Vallée, France
[email protected] 2 ENS Paris-Saclay, Gif-sur-Yvette, France
Abstract. The current context of the climate change, exhaustion of natural resources and impetus towards circular economy is leading to an increase in the use of recycled aggregate concrete (RAC) in the construction industry. The French National Project RECYBETON investigated the feasibility of using recycled concrete aggregate (RCA) in concrete through studies of concrete’s properties, including microstructure and durability of RAC. This kind of project, transposed in the standards, allows to promote the advantages of the RCA use and an appropriate substitution rate in the concrete. Based on the previous knowledge, some characteristics of recycled concrete still require further investigation such as fatigue response. In that respect, an experimental protocol for compressive fatigue test has been developed in controlled laboratory conditions. This protocol was designed in order to monitor damage in the experimental samples during the fatigue compression test using various non-destructive methods (Ground-Penetrating Radar, modal analysis and Stiffness Damage Test). As a feasibility study, this protocol has been applied to standard concrete specimens first. The corresponding results obtained for standard concrete specimen submitted to a fatigue test are elaborated in this paper, including a comparative study of the considered non-destructive methods. This methodology not only allows to study mechanical characteristics of recycled concrete but also to monitor damage evolution over various cycles. Keywords: Recycled concrete · Fatigue · SHM · NDT · damage
1 Introduction In a context of natural resources saving needs, the use of Recycled Concrete Aggregates (RCA) in the construction area appears to be a possible solution. This alternative source of aggregates has thus deserved many research efforts over the past years. The use of this resource is nowadays included in standards (e.g. [1]), though with limited contents. Thus, increasing the possible content of RCA in concrete represents a possible improvement of natural aggregates saving. To reach this objective, it is necessary to better understand the behavior of concretes containing RCA: as an example, research projects such as [2, 3] investigated the effect of higher RCA content. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 A. J˛edrzejewska et al. (Eds.): SynerCrete 2023, RILEM Bookseries 43, pp. 436–448, 2023. https://doi.org/10.1007/978-3-031-33211-1_39
Development of Damage Monitoring Techniques
437
The fatigue behavior is one of the concrete characteristics that can be affected by the use of RCA [4, 5]. However, results available in the literature remain limited. In the present research, a testing protocol aiming at investigating the fatigue behavior of concretes containing RCA is developed. As a first attempt, the protocol is tested on a concrete sample containing natural aggregates in order to evaluate the ability of the considered experimental methods to detect damage due to fatigue. As the investigation of fatigue requires long-lasting mechanical tests, the use of NonDestructive Testing (NDT) has the advantage to allow assessment of damage at various stages while keeping the same specimen for further testing. In this work, Ground Penetrating Radar (GPR) is used as a qualitative tool to analyze the damage mechanisms due to fatigue. Additionally, modal analysis is used to quantitatively measure the mechanical damage evolution. Finally, the use of the Stiffness Damage Test (SDT) [6] is considered to evaluate periodically the mechanical behavior of the specimen. Though SDT is well known to evaluate the mechanical effects of internal swelling reactions of concrete such as Alkali-Silica Reaction (ASR), Delayed Ettringite Formation (DEF), Freeze-Thaw (FT), etc. [7], it has been considered in the present research to check its ability to detect quantitatively the damage evolution due to fatigue in compression. The final objective of the research is to assess how complimentary these experimental techniques are. In the following, the testing procedure is first presented, including the methodologies of GPR, modal analysis and SDT. Then, preliminary results on a concrete sample containing natural aggregates and submitted to repeated loading cycles up to 60% of the compressive strength are provided. Finally, a discussion on the possible improvements of the protocol for the future RCA concrete campaign is proposed.
2 Materials and Methods 2.1 Test Setup for Mechanical Testing In this section, the procedure for fatigue compression test including damage monitoring on concrete specimen is introduced. This procedure includes the methodology for the fatigue monitoring and three non-destructive methods: GPR, modal analysis and SDT. The concrete specimen considered in this study has the following dimensions: 10 cm × 10 cm × 100 cm. The concrete mix contains only natural aggregates. This specimen has a length-on-width ratio equal to 10, it can therefore be considered as an Euler-Bernoulli beam so that mainly flexion is considered. The beam was installed in a hydraulic testing machine with a capacity of 5000 kN in compression (see Fig. 1). For the fatigue compression test, the Locati method was considered [8, 9]. This method consists in applying various successive fatigue tests on the same specimen. In this case, the specimen is subjected to 4 stages of fatigue compression loading (see Fig. 2). For each stage, 200 000 loading cycles with a frequency of 1 Hz are applied. The minimum loading is always the same and corresponds to 10% of its compressive strength. The maximum loading is evolving according to the considered stage: it starts from 30% for the first stage up to 60% for the last stage. In order to evaluate the impact of each fatigue compression stage, the damage on the specimen is measured using three non-destructive methods before the first stage, between each stage, and at the end of the last stage.
438
L. Kerner et al.
Accelerometer
Concrete specimen in the 5000kN compression machine
59 cm
Impact
35.5cm
Extensometers on each side of the specimen
Fig. 1. Protocol of fatigue test with damage monitoring
Stage 4: 10% - 60% Stage 3: 10% - 50%
Loading (%F limit)
60% Stage 2: 10% - 40%
50% 40%
Stage 1: 10% - 30%
30%
10% 0
Damage monitoring: GPR, modal analysis and SDT
Fig. 2. Protocol of fatigue compression test with damage monitoring (Flimit = ultimate load in compression)
2.2 Fatigue Monitoring Four companion specimens (11 cm in diameter, 22 cm in length) were submitted to a compression test approximately one year after casting in order to evaluate the mean compressive strength of the concrete (σ s ). The following result was found: σ s = 36.7 MPa. Therefore, the compression stress applied for each stage was calculated and presented in Table 1.
Development of Damage Monitoring Techniques
439
Table 1. Minimal and maximal applied stress for each stage
Stage Minimal compression stress Maximal compression stress
(%) (MPa) (%) (MPa)
1
2
30 11
40 14.7
10 3.7
3
4
50 18.3
60 22
On Fig. 3 is presented the stress implemented by the compression machine on the concrete specimen. A few observations can be made here: – For the minimal compressive stress, the instruction is the same for each stage and equal to 3.7 MPa. The standard deviation increases with stages from 6% up to 17%. – For the maximal compressive stress, the standard deviation is similar for each stage with an average of 6.5%. Investigations are under progress to determine whether this standard deviation is due to the behaviour of the testing machine or to a data logger artefact. Nevertheless, the global result of the fatigue protocol is satisfactory, the compression machine applies the correct loadings with the wanted frequency, though some changes in the controlling software might be needed to reduce the standard deviations in future experiments.
Fig. 3. Compression stress applied for each stage
To evaluate the strain in the concrete specimen during the fatigue test, four extensometers (see Fig. 1), using Solartron AX/1/S sensors with 1mm measuring range and a measurement length of 400 mm, were installed on each side of the specimen. These extensometers were set using metallic anchors embedded in concrete during casting.
440
L. Kerner et al.
2.3 First Non-destructive Method: Ground Penetrating Radar GPR is considered one of the most promising non-invasive method for the monitoring of damages in concrete [10–12]. A high-resolution centre frequency of antenna is highly recommended for the concrete assessment applications. In this study, a commercial GPR system with 2.6 GHz centre frequency ground-coupled antenna is employed for the concrete specimen scanning. As recommended GPR antenna is calibrated prior to the GPR survey [13]. Radar data was acquired along a radar line along the specimen axis, with a 2 mm trace sampling interval. For this study the sampling frequency was 20 GHz (ten times the central frequency), being the time window 7 ns. For this method, all the sensors used during fatigue testing were uninstalled from the concrete specimen which was positioned on the floor with a wooden frame around it. Figure 4 shows the GPR system with the concrete specimen for these measurements. All B-scans in the laboratory experiments were processed with two main postprocessing features: (i) subtract-mean dewow to remove very low frequencies in the signal, (ii) and a background removal filter to eliminate anomalies affecting all the A-scans at a same time wave propagation. The depth of the concrete sample is around 2 ns, which is equivalent to 10 cm of real depth of the concrete sample. Beside that the wave velocity is 10 ns/cm and concrete dielectric constant of 6.5. This work is considered in a controlled laboratory experiment and the focus of the work remain on the depth of concrete sample.
Wooden frame Specimen installed on the floor
GPR system
Fig. 4. GPR system with 2.6 GHz
2.4 Second Non-destructive Method: Modal Analysis Free vibration tests were conducted to measure the first natural frequencies of the specimen. A vibration exciter imposed a brief impact on the specimen and its free decay response was recorded using an accelerometer (Lord MicroStrain G-Link-200-8G) fixed 42 cm from the bottom of the sample (see Fig. 1); on the opposite side from the one that was excited. The extensometers used during fatigue testing are uninstalled for these measurements to avoid added mass on the sample. A systematic and constant compression load is applied by the compression machine equals to 22.5 kN. A preliminary study [14] was conducted to define the configuration allowing to optimize the evaluation of the first natural frequencies (location of the accelerometer, method to apply the impact on the structure, load applied by the compression machine).
Development of Damage Monitoring Techniques
441
The sampling frequency of the accelerometer is 4096 Hz, which allows to evaluate the fist two natural frequencies of the sample. For each configuration (before the fatigue test, between each stage and after the last stage), five free vibration tests were conducted. The first natural frequencies were then extracted with a Fast Fourier Transform (FFT) as shown in Fig. 5.
Fig. 5. (a) response of the accelerometer during a free vibration test and (b) its FFT
2.5 Third Non-destructive Method: Stiffness Damage Test, SDT As presented in [6], SDT consists in applying five compression cycles up to a stress corresponding to 40% of the compressive strength of the material (either at 28 day after casting or determined on sound specimens at the time of SDT investigation, depending on which data can be obtained during the investigation), at a loading rate of 0.1 MPa/s. During the test, the applied force and the longitudinal strain of the specimen are monitored. A detailed description of the method is provided in [15]. The output results defined in [6] and used in the present research are: • the mean modulus of elastic over the 2nd and 3rd load cycles, • the Stiffness Damage Index (SDI) (corresponding to the ratio between the dissipated and provided mechanical energy during the whole test), • the Plastic Deformation Index (PDI) (corresponding to the ratio between the final plastic strain to the longitudinal strain at the maximum load of the final loading cycle). For this test, the setup described in Sect. 2.1 was used. The longitudinal strain was obtained as the mean strain measurement of each face of the concrete specimen.
3 Results 3.1 Fatigue Monitoring The evolution of strain in the concrete sample during the four stages of the fatigue test is introduced in Fig. 6(a). For each stage, one can observe the minimal and maximal compression stress and its associated strain. The cumulative strain is also presented in
442
L. Kerner et al.
Fig. 6(b). In this figure, the cumulative strain clearly increases with the stages. The more loading cycles are applied on the specimen, the more plastic strain is measured.
Fig. 6. (a) Compression stress vs strain during each stage of the fatigue test and (b) cumulative strain
3.2 First Non-destructive Method: Ground Penetrating Radar As a preliminary result, GPR presents limited results as a consequence of the low changes in the subsurface material and changes in background noise amplitude. Figure 7(a) shows a first bench concrete sample with no fatigue cycle applied. During the interpretation of the radargrams, the main hyperbolas are an increase of the amplitude as a response of the metallic anchors embedded in the sample. In addition to that, the most critical section of the concrete sample which could have some changes in the material properties due to the several fatigue cycles over time. For instance, comparing each profile (after stage 1 Fig. 7(b), stage 2 Fig. 7(c), stage 3 Fig. 7(d) and stage 4 Fig. 7(e)) to the first profile which is bench concrete sample radargram. However, for the GPR scanning’s the changes of the damage over time cycles is considered as a crack evolution. The most damaged zones detected with some changes over fatigue cycling process correspond to the zone between 0.5 to 0.6 m and 0.6 to 0.7 m, which is in the middle of the sample. The damage evolution is very slightly detected in the middle of the span, as this could corresponds the fact that mostly the cracking phenomena in concrete spans occurs in the centre -to the right or centre-to the left or both zones near to the middle span of the concrete [16]. Besides that, these results are preliminary and further researches will be conducted over time taking into account different fatigue cycles until damage occurs. This will better support the understanding of the GPR capability for the detection and monitoring of the crack evolution over time in the concrete specimen.
Development of Damage Monitoring Techniques
(a)
(b)
(c)
(d)
443
(e)
Fig. 7. GPR images of the scanned concrete samples (time (ns) vs distance (m) fir the concrete specimen): (a) before the fatigue test, after (b) stage 1, (c) stage 2, (d) stage 3, (e) stage 4
3.3 Second Non-destructive Method: Modal Analysis Before the compression fatigue test and between each stage, free vibration tests were conducted as described in 2.4. For every configuration, the natural frequencies were evaluated using the Fast Fourier Transform. In Fig. 8, the evolution of the second natural frequency of the concrete specimen is presented according to the test stages. The general trend is a decrease of the natural frequency with the increase of the loading cycles. This result is quite expected as the fatigue test is gradually damaging the concrete specimen. It should be noted that this decrease is not linear: for example, the evaluated frequency after the first stage is lower than the one evaluated for the second stage. More compressive fatigue stages would be necessary to confirm the tendency observed in Fig. 8. Moreover, additional investigations would be useful to determine the uncertainty of the identification of the natural frequencies. 3.4 Third Non-destructive Method: Stiffness Damage Test, SDT The stress vs. strain relation during the various SDT tests is presented in Fig. 9. The corresponding output parameters are presented in Table 2. On the one hand, and as expected, the modulus of elasticity is decreasing progressively with the number of cycles applied, suggesting a damage increase during the protocol. After stage 4 (200 000 cycles
444
L. Kerner et al.
Fig. 8. Evolution of the second natural frequency according to the compression fatigue test stages
at a maximum stress of 60% of the compressive strength), the stiffness loss is about 11%: according to the classification of damage degree based on SDT presented in [6], it corresponds to a marginal damage degree. However, one must keep in mind that this classification has been developed for ASR-affected specimens; it could be different for fatigue damage and requires further investigations. On the other hand, based on the same classification, the SDI values obtained suggest a negligible damage level. Finally, PDI does not show any clear trend at this stage of the testing protocol. Table 2. Stiffness Damage Test results Description
SDI
PDI
Modulus of elasticity (GPa)
Before fatigue testing
0.06
0.02
31.7
After stage 1
0.06
0.02
31.1
After stage 2
0.05
0.03
30.9
After stage 3
0.05
0.02
29.8
After stage 4
0.06
0.03
28.1
Development of Damage Monitoring Techniques
445
Fig. 9. Stress vs. strain relation for the different SDT tests
4 Discussion The Ground Penetrating Radar gives qualitative but limited preliminary results. More fatigue stages would be necessary to confirm the first observations. It seems to be a good method to have a qualitative evaluation of the damage on the specimen. The second non-destructive method, the modal analysis, gives interesting results as the second natural frequency of the concrete specimen tends to decrease during the fatigue test. Nevertheless, few observations can be made at this point. As an EulerBernoulli beam, the natural frequencies of the specimen depend on different parameters: its Young modulus, its density, its moment of inertia and its area. During a fatigue test, all these parameters are affected and therefore influence the natural frequencies of the specimen. The evaluation of the modulus of elasticity with the SDT method combined with the evaluation of the natural frequencies up to a large extent of damage could allow us to develop an analytical model to describe the evolution of these parameters during a compression fatigue test. Thus, more results (i.e. additional stages in the process described in Sect. 2.1) are be needed. Concrete is a heterogeneous material, and its heterogeneity tends to increase during a fatigue test. It should be noted that the fatigue test will also increase the asymmetry of the specimen. This makes it difficult to analyze the evolution of the natural frequencies with a free vibration test. The modal analysis will show double or triple peaks near a natural frequency, typical for an asymmetric specimen. A more elaborated data processing is considered in the future using wavelet transform [17]. In addition, the free vibration test can be improved to minimize the white noise and vibration generated by
446
L. Kerner et al.
the compression machine. The response of the concrete specimen could be measured with free-free boundary conditions for example. Regarding SDT results, the modulus of elasticity is so far the most sensitive parameter to the evolution of damage. Figure 10 illustrates the strong relation existing between the modulus of elasticity and the cumulative plastic strain measured during the fatigue test, despite the very early damage stage discussed in Sect. 3.4. Regarding the small SDI and PDI values obtained up to now, no clear trend can be identified so far: these parameters will be further investigated during the next stages of the fatigue protocol (corresponding to higher maximum stress) to check a possible evolution at higher damage levels.
Fig. 10. Evolution of the modulus of elasticity as a function of the cumulative plastic strain
5 Conclusion The objective of the research presented in this paper was to develop a testing protocol to analyze the compression fatigue behavior of concretes containing RCA. The Locati method was used for the fatigue test per se, corresponding to an increasing maximum stress in the specimen. At the end of each stage, the following NDT methods were used: GPR scan, modal analysis, and SDT. As a first attempt, the protocol was applied on a concrete containing natural aggregates. Stages of 200 000 cycles, whose maximum values ranged from 30% to 60% of the compressive strength (with a step of 10%), were applied. So far, (1) all three methods show a qualitative agreement corresponding to an increasing damage and (2) the most sensitive parameter is the modulus of elasticity determined by SDT.
Development of Damage Monitoring Techniques
447
These preliminary results have to be confirmed at higher maximum stress levels (and with increasing number of applied cycles) on the one hand, and compared to measurements obtained with concrete specimens containing RCA. This further work will be performed in a near future.
References 1. AFNOR: NF EN 206+A2/CN : Béton - Spécification, performance, production et conformité - Complément national à la norme NF EN 206+A2 (2022) 2. De Larrard, F., Colina, H.: Le béton recyclé. IFSTTAR (2018) 3. IREX: Comment recycler dans le béton - Recommandations du Projet National RECYBETON (2019) 4. Peng, Q., Wang, L., Lu, Q.: Influence of recycled coarse aggregate replacement percentage on fatigue performance of recycled aggregate concrete. Constr. Build. Mater. 169, 347–353 (2018). https://doi.org/10.1016/j.conbuildmat.2018.02.196 5. Thomas, C., Setién, J., Polanco, J.A., Lombillo, I., Cimentada, A.: Fatigue limit of recycled aggregate concrete. Constr. Build. Mater. 52, 146–154 (2014). https://doi.org/10.1016/j.con buildmat.2013.11.032 6. Sanchez, L.F.M., Fournier, B., Jolin, M., Mitchell, D., Bastien, J.: Overall assessment of Alkali-Aggregate Reaction (AAR) in concretes presenting different strengths and incorporating a wide range of reactive aggregate types and natures. Cem. Concr. Res. 93, 17–31 (2017). https://doi.org/10.1016/j.cemconres.2016.12.001 7. Sanchez, L.F.M., Drimalas, T., Fournier, B., Mitchell, D., Bastien, J.: Comprehensive damage assessment in concrete affected by different internal swelling reaction (ISR) mechanisms. Cem. Concr. Res. 107, 284–303 (2018). https://doi.org/10.1016/j.cemconres.2018.02.017 8. Sainz-Aja, J., Thomas, C., Carrascal, I., Polanco, J.A., de Brito, J.: Fast fatigue method for self-compacting recycled aggregate concrete characterization. J. Clean. Prod. 277, 123263 (2020). https://doi.org/10.1016/j.jclepro.2020.123263 9. Thomas, C., Sosa, I., Setién, J., Polanco, J.A., Cimentada, A.I.: Evaluation of the fatigue behavior of recycled aggregate concrete. J. Clean. Prod. 65, 397–405 (2014). https://doi.org/ 10.1016/j.jclepro.2013.09.036 10. Rasol, M., et al.: GPR monitoring for road transport infrastructure: a systematic review and machine learning insights. Constr. Build. Mater. 324, 126686 (2022). https://doi.org/10.1016/ j.conbuildmat.2022.126686 11. Rasol, M., Pérez-Gracia, V., Fernandes, F.M., Pais, J.C., Santos-Assunçao, S., Roberts, J.S.: Ground penetrating radar system: principles. In: D’Amico, S., Venuti, V. (eds.) Handbook of Cultural Heritage Analysis, pp. 705–738. Springer, Cham (2022). https://doi.org/10.1007/ 978-3-030-60016-7_25 12. Annan, A.P.: GPR—history, trends, and future developments. Subsurf. Sens. Technol. Appl. 3, 253–270 (2002). https://doi.org/10.1023/A:1020657129590 13. Rasol, M., Perez-Gracia, V., Assuncao, S.S.: Analysis and calibration of ground penetrating radar shielded antennas. In: 2018 17th International Conference on Ground Penetrating Radar (GPR) (2018) 14. Denis, L.: Analyse de l’endommagement de matériaux cimentaires soumis à des essais de fatigue en compression - application future aux bétons à base de granulats de béton recyclé ou des granulats alternatifs. 72 p. (2022) 15. Sanchez, L.F.M., Fournier, B., Jolin, M., Bastien, J.: Evaluation of the stiffness damage test (SDT) as a tool for assessing damage in concrete due to ASR: test loading and output responses for concretes incorporating fine or coarse reactive aggregates. Cem. Concr. Res. 56, 213–229 (2014). https://doi.org/10.1016/j.cemconres.2013.11.003
448
L. Kerner et al.
16. Mat Saliah, S.N., Nor, N.M., Kamaruzzaman, K.N., Mohd Mesron, M.A., Abd Rahman, N.: Intensity crack zones of reinforced concrete beam under monotonic loading. In: Journal of Physics: Conference Series, vol. 1349 (2019). https://doi.org/10.1088/1742-6596/1349/1/ 012086 17. Le, T.-P., Argoul, P.: Continuous wavelet transform for modal identification using free decay response. J. Sound Vib. 277, 73–100 (2004). https://doi.org/10.1016/j.jsv.2003.08.049
Comparison of Different Approaches for Quantification of Amorphous Phase in Hydrated Cement Paste by XRD Antonina Goncharov(B)
and Semion Zhutovsky
Technion – Israel Institute of Technology, 3200003 Haifa, Israel [email protected]
Abstract. One of the most important tasks in the investigation of hydrated cement paste is the determination of the amorphous phase content, which is mostly cementitious gel. The determination of the cementitious gel content is important for the evaluation of the hydration process. However, this task is not trivial and it is hard to achieve high accuracy. The major approach for determining the content of the amorphous phase is by the use of an internal or external standard in the x-ray diffraction (XRD) analysis. This study compared two external standards, as well as two procedures of internal standard intermixing on amorphous phase determination accuracy. In addition, a novel method of calibrating the HKL phase for the Partial Or Not Known Crystal Structure method of evaluation of the amorphous phase using an internal standard was applied and evaluated. For this purpose, cement pastes with water to cement ratios of 0.30, 0.35, and 0.40 were prepared, and hydration was stopped using the solvent exchange at the ages of 7 and 28 days. The hydrated cement paste samples were examined using XRD and the hydrates assemblage and amorphous phase content were analyzed using Rietveld refinement. The bound water in hydrated cement paste was studied using thermal analysis. For comparison, the degree of hydration was determined by XRD, thermal analysis, and isothermal calorimetry. The experimentally determined amorphous phase content was compared with theoretical calculations based on the degree of hydration and the recommendations for the best amorphous phase quantification procedure were given. Keywords: Amorphous phase · X-ray diffraction · Cementitious gel · Rietveld analysis · PONKCS
1 Introduction Understanding the hydration process of cement and alternative binders is one of the most important issues, especially since there has been a recent focus on the development of binders with a low carbon footprint [1]. The rate of hydration, quantity and quality of hydration products largely determine the set of mechanical properties at an early age and the durability of concrete at later ages. The proper analysis of hydrated phase assemblage can not be performed without an accurate determination of the amorphous phase, which can be done by using either direct or indirect methods. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 A. J˛edrzejewska et al. (Eds.): SynerCrete 2023, RILEM Bookseries 43, pp. 449–459, 2023. https://doi.org/10.1007/978-3-031-33211-1_40
450
A. Goncharov and S. Zhutovsky
First, the mineral composition of crystalline hydration products has to be determined. One of the most widespread laboratory methods for determining mineral composition is X-ray diffraction (XRD) by using the Rietveld method for quantitative analysis. However, for a correct determination of the mineral composition, it is necessary to accurately identify the amount of the amorphous phase, which is not a trivial issue. The presence of an amorphous phase results in very broad and low-intensity peaks (humps) that cannot be correctly taken into account by Rietveld analysis, and broad peaks may be overlapped by other crystalline phases remaining in the non-reacting cement, such as unhydrated alite or belite [2]. To determine the amorphous phase using XRD, internal or external standard methods are usually used [3–6]. The standard must have known crystallinity. For example, according to the National Institute of Standards and Technology (NIST), their standard reference material corundum has been recertified to be 99.02% ± 1.11%, which makes it possible to evaluate the crystallinity of other standard materials [7]. Also, it is preferred that the standard peaks do not overlap with the main reflection peaks of the phases present in the sample [2]. The method of the internal standard requires the sample to be homogeneously mixed with a known content of a crystalline standard. Homogenization of the sample and standard must be carried out carefully to avoid sample alteration: “Overgrinding may lead to peak broadening and partial amorphisation of softer phases in the sample” [2]. Also, the mass fraction of the standard concerning the sample is of great importance. A mass fraction of the internal standard which is too low does not allow accurate determination of the amount of amorphous phase in the sample, especially if the content of the amorphous phase is relatively low [8]. A mass fraction of the internal standard which is too high can affect the precision of quantification of crystalline phases due to the high dilution of the original sample with highly crystalline material with high peak intensity [2]. The high content of internal standard in a sample can also affect the hydration kinetics of the cement paste through the filler effect, so it is preferable to use a stop hydration before starting the experiment [9]. Thus the optimum internal standard content for hydrated cement paste is around 20% by weight [2]. The external standard method is based on a comparison of the scale factors of a sample to the scale factor of an external standard material measured under the same diffractometer conditions [2, 10]. In this method, the main parameter is the K-factor of the external standard [11]. The obvious advantage of this method is that homogeneously mixing the sample with the internal standard is not needed, which prevents additional sample preparation problems and the possibility of using samples without stopping hydration. However, for comparison, these methods are better to use samples after stopping hydration. For this purpose, it is necessary to take into account the change in the mass of the cement paste after the removal of free water all results must be normalized for which thermogravimetry methods can be used for the investigation of bound water content [2]. Also, isothermal calorimetry is a widely used technic for investigating the degree of hydration for comparison with other methods that use hydration stoppage. Complex use of quantitative phase analysis of hydrated cement pastes by XRD and calorimetry has proven to be a promising way to relate the phase development of cement paste to hydration kinetics [6, 12, 13].
Comparison of Different Approaches for Quantification of Amorphous Phase
451
However, the most problematic part of internal and external methods is the right determination of background, which strongly affects the quantitative analysis. Some researchers used a combination of an analytical background and a broad peaks set to cover the contribution amorphous phase [14]. The visual interpolation of the background can be flexible, which greatly simplifies the task during processing. Nevertheless, the quality of the background is highly dependent on the operator’s skills and is complicated for hydrated cement, which shows a strong overlap of the main peaks. Underestimation or overestimation of the background significantly affects the estimation of peak intensity, especially weak peaks with low intensity, and introduces an error in quantification results. The solution to this problem could be the approach of Partial Or Not Known Crystal Structure (PONKCS) [2, 15]. PONKCS takes the amorphous phase contribution by defining its XRD pattern as a crystalline phase, through calibration of a parameter called “Pseudo Formula Mass” is required for quantification. This calibrated parameter of the amorphous phase can be used in Rietveld refinement in combination with the scale factor to determine its content [2]. The PONKCS approach was successfully used for the quantification of Calcium Silicate Hydrate (CSH) gel in hydrating alite pastes, amorphous supplementary cementitious materials, in anhydrous and hydrated blended cements, and glass powder reaction in blended-cement pastes [5, 16, 17]. However, there is difficulty in obtaining a suitable model for the CSH phase for the quantitive analysis of hydrated cement pastes. It was demonstrated that it is possible to use white Portland cement hydrated for 7 years [16] or pure alite paste [5] for this purpose. Nevertheless, white cement hydrated for many years or pure alite paste is not available in most cases, which makes the calibration of the PONKCS method difficult. It is necessary to create methods for determining the amorphous phase by PONKCS based on available materials, which can be easily produced in a lab. In this study, we evaluated the possibility of using the PONKCS method in combination with an external and internal standard to calibrate the pseudo formula mass of the CSH phase. The CSH phase was modulated by the HKL method based on hydrated ordinary portland cement for 28 days with a water to cement (w/c) ratio of 0.40. Also in this study different ways of mixing the 20% internal standard with the sample were tested: manual mixing and mixing in an XRD mill. Finally, all methods were compared with each other with the results of theoretical calculations based on the degree of hydration of alite and belite. The degree of hydration determined using calorimetry, thermogravimetric analysis (TGA), and XRD was also compared to evaluate the quality of the results. The content of portlandite obtained by XRD and TGA was compared to assess the effect of the preferred orientation.
2 Materials and Methods 2.1 Materials CEM I 52.5 N with a specific surface area of 4650 cm2 /g, produced by Nesher Israel Cement Enterprises, was used for cement paste preparation. The mineral composition investigated by using XRD was: 60.37% alite, 16.14% belite, 10.93% ferrite, and 4.95% aluminate. Loss on ignition of cement between 105 and 1050 °C was 2.99 wt. %. Cement pastes were prepared with w/c ratios of 0.30, 0.35, and 0.40. Cement and water were
452
A. Goncharov and S. Zhutovsky
weighed and mixed in a standard pan mixer and then cast in one-inch cubic molds, which were closed with a greased glass cover. The molded samples are wrapped in plastic covers to prevent carbonation and stored for 24 h in laboratory conditions. After 24 h, the hardened cement cubes were demolded and placed in a saturated Ca(OH)2 solution for curing. The temperature of curing was kept at 21 ± 3 °C. 2.2 Methods Hydration Stoppage. After 7 and 28 days, samples were grounded manually using a pestle and mortar. The size fraction between 2.36 and 4.75 mm was collected and the hydration was stopped by solvent exchange using isopropyl alcohol (IPA). The perform solvent exchange, the ground cement paste was placed in glass beakers and filled with IPA at the solvent-to-sample volume ratio of 50. Every 20 min, IPA was replaced 3 times for a total exchange time of 80 min. Then the samples were placed in a vacuum oven with a pressure of 300 mbar and a temperature of 40 °C, where they were dried for 12 h. After that, all samples were placed in a desiccator with silica and portlandite powder to prevent carbonation and absorption of moisture from the air. It is important to note that by this method of hydration stoppage only free water is removed without removing gel and chemically bound water [18]. Calorimetry. The degree of hydration of cement pastes was determined using the Isothermal calorimeter TAM AIR by measuring the evolved heat of the hydration. The temperature inside the calorimeter was set to 21 °C. The tested samples had a total mass of cement and water of 5 g. The cement and water were mixed for 1 min using the vortex mixer to achieve homogeneity of the sample. After mixing, the sample was sealed in the sample vile and placed in the calorimeter. A reference sample of clean and dry quartz sand with a total mass of 5 g was used. The measurement over the seven days was recorded. The ultimate heat of hydration for the calculation of hydration degree was calculated as 482.85 J/g based on the mineral composition of cement as described by Taylor [19]. Sample Preparation for XRD and TGA. Hydrated cement pastes were manually crushed into coarse powder after stopping hydration and then 8 g of each sample were ground in an XRD mill with 15 ml isopropyl alcohol for 15 min for reducing the preferred orientation in XRD. Then the sample was filtered, washed with diethyl ether, and then dried in a vacuum oven at 300 mbar pressure and 40 °C for one hour. Diethyl ether displaces isopropanol and promotes faster evaporation, thus drying is accelerated. In addition to this, the use of diethyl ether has a positive effect on thermogravimetric analysis [2]. Thermogravimetric Analysis (TGA). Approximately 50 mg of fine ground sample was used for TGA (Netzch STA 449 Jupiter) at a heating rate of 20 °C/min in the temperature range of 40–1050 °C under an inert atmosphere of nitrogen in 150 μl alumina open crucibles. Results were collected using Proteus software. Mass loss for portlandite calculation was taken using the Marsh step between 400 and 530 °C. The mass loss between 105 and 1050 °C was used to normalize the results to the mass of cement (for clinker minerals) and the mass of cement paste (for hydration products and
Comparison of Different Approaches for Quantification of Amorphous Phase
453
CSH). For the calculation of the degree of hydration based on TGA results the bound water content of 0.325 (chemically bound water+gel water) was calculated based on the mineral composition of cement. The calculation of the degree of hydration was based on mass loss between 105 and 1050 °C as well. XRD. The XRD analysis was carried out using a Malvern Panalytical EMPYREAN X-ray diffractometer. Measurement conditions were: CuKα1,2 X-ray source (λ = 1.5408 Å), X-ray generator operated at 45 kV with a current of 40 mA, and a Goniometer radius of 240 mm, x-ray mirror name Bragg-Brentano HD with flat shape and graded type. The detector was a PIXcel 3D detector used in a 1D continuous scan mode. The XRD scans were analyzed using the Panalytical HighScore Plus software using the ICDD PDF-4 Minerals and PAN-ICSD databases. Quantitative analysis was performed by means of Rietveld refinement using the Panalytical HighScore Plus software. For cement phase composition investigation, the incident beam optics included a 1/4° divergence slit, a 10 mm mask, a 0.04 rad Soller slit, and a 1° anti-scatter fixed slit. The diffracted beam optics consisted of an 8 mm anti-scatter fixed slit and a 0.04 rad Soller slit. The scan range was 10 and 70°2θ. A step size was 0.0131°2θ, and a counting time of 58.396 s. For cement paste composition, as well as internal and external standards investigation, the incident beam optics included a 1/8° divergence slit, a 10 mm mask, a 0.04 rad Soller slit, and a 1/2° anti-scatter fixed slit. The diffracted beam optics consisted of a 7.5 mm anti-scatter fixed slit and a 0.04 rad Soller slit. The scan range was 7 and 70°2θ. A step size was 0.0525°2θ, and a counting time of 243.760 s. External Standard. Chemically pure fully crystalline alumina α-Al2 O3 (corundum) was used as external and internal standards. Chemically pure lithium fluoride with the chemical formula LiF was used as an additional external standard. External standards were scanned with the same program as the cement paste samples on the same day. K-factor was determined and applied to quantify the amorphous phase content as described elsewhere [20]. Internal Standard. The internal standard (corundum) was mixed with cement paste at the rate of 20% of the mixture weight in two different ways: by hand, and interground using an XRD mill with 15 ml isopropanol during XRD sample preparation as described above. Hand mixing was done with agate mortar and pestle for 5 min. The scan with internal standard was analyzed using Rietveld analysis and known standard content was used to quantify the amorphous phase content. PONKCS. For the PONKCS method, CSH amorphous phase was defined as a cubic crystal system (space group FM-3M, space group number 225). The unit cell of a = 5.652Å, Caglioti W left = 36.8214, and Peak Shape left = 0.6 fitted on the mix of 20 wt. % crystalline alumina (corundum) and 80 wt. % cement paste sample with a w/c ratio of 0.40 at the age of 28 days. After fitting it as HKL phase, the pseudo formula mass was calibrated so that the content of corundum matched the weighed content of 20%. The fitted profile is shown in Fig. 1. The calibration was performed on all six cement paste samples using the same unit cell and profile variables. The average pseudo formula mass of 11.1234 was obtained with a standard deviation of 0.2433 (which gives the coefficient of variation = 2.19%). The calibrated CSH amorphous phase was used in the Rietveld
454
A. Goncharov and S. Zhutovsky
analysis of cement paste samples without standard, by fitting only the scale factor. The results obtained by this method are denoted as PONKCS.
Fig. 1. Example of calibration of CSH amorphous phase using an internal standard.
3 Results The comparison of the results of the quantification of amorphous phase content by different methods is shown in Fig. 2. It should be noted that all results given in Fig. 2 are normalized to the initial mass of cement paste using the data from TGA. First of all, it can be seen that the hand-mixed internal standard method resulted in the highest scatter of results. In cement paste samples of w/c ratio, hand mixed internal standard resulted in very high deviation – very low for 7 days and very high for 28 days. This is most probably the result of the non-uniform distribution of the standard in the sample. The results obtained using internal standard interground with XRD mill give more consistent results. The results obtained using two external standards are very similar with differences of only around 0.2% between them. The ratio between the internal and external standards is not constant, while some results are higher for internal standards than for external and some are lower for the internal standard. PONKCS method demonstrated results that are consistently slightly higher than external standard. Rietveld analysis also gives us the residual content of clinker phases. Using these data the total degree of hydration as well as the degree of hydration for different phases can be calculated. Considering that only alite and belite produce amorphous CSH gel during hydration, the theoretical CSH content can be calculated. During hydration, each mole
Comparison of Different Approaches for Quantification of Amorphous Phase
455
Fig. 2. Amorphous phase content determined by different methods.
of hydrated alite and each mole of hydrated belite produce one mole of CSH. Taking the molar masses of alite and belite as 228.33 and 172.25 g/mol, respectively, and the molar mass of saturated CSH gel as 227.49 g/mole, the theoretical amorphous CSH gel content was calculated and normalized to the initial weight of cement paste using the TGA data. The comparison of measured and calculated content of the amorphous phase is given in Fig. 3. The solid line designates the equivalency of the results (slope of 1:1) and the dotted lines designate the deviation of 5% to each side. It can be seen that the results of the external standard method are quite consistent. Using internal standard interground with XRD mill gives also very consistent results with low scatter, except one point (w/c = 0.30 at the age of 7 days), while the use of hand-mixed internal standard resulted in a high discrepancy between the measure and calculated results indicating a problem with mixing standard by hand. Figure 4 gives a comparison of the total degree of hydration determined by isothermal calorimetry, TGA, and XRD (internal and external standards). The isothermal calorimetry was performed only for 7 days, so these results seem most consistent giving a monotonic increase of hydration degree with the increase of w/c ratio. The results of XRD show a consistently higher degree of hydration, with the internal standard demonstrating slightly higher values than the external standard method. TGA demonstrated consistent results for 28 days, but for 7 days it can be seen that the degrees of hydration for w/c ratios of 0.35 and 0.40 are too low.
456
A. Goncharov and S. Zhutovsky
Fig. 3. Measure versus calculated amorphous phase content determined by different methods.
Finally, the content of Porlandite is compared between the results of XRD and TGA in Fig. 5. The portlandite crystals have the form of hexagonal plates and, for this reason, are prone to a preferred orientation that may negatively affect the results of XRD. Preferred orientation can be taken into account during Rietveld analysis, though an error still can be introduced because of it. On the contrary, TGA is believed to give a precise estimation of portlandite content because mass loss associated with portlandite decomposition is quite distinctive. For this reason, it is of great interest to compare the content of portlandite obtained by XRD and TGA. It can be seen in Fig. 5 that in most samples portlandite content is slightly overestimated by XRD despite the incorporation of preferred orientation correction into Rietveld analysis. Overestimated values of portlandite may result in the underestimation of clinker phases. This may explain the overestimated values of the degree of hydration by the XRD method. It should be noted that portlandite content obtained by PONKCS method resulted in the most consistent and least scattered results.
Comparison of Different Approaches for Quantification of Amorphous Phase
Fig. 4. Comparison of the degree of hydration determined by different methods.
Fig. 5. Comparison of portlandite content determined by different methods.
457
458
A. Goncharov and S. Zhutovsky
4 Conclusions This study tested various methods for measuring the amorphous phase in hydrated cement paste with different water-cement ratios at 7 and 28 days of age. The internal standard was mixed in various ways, two different crystal materials were used as the external standard. A new approach to the calibration of pseudo formula mass for the HKL phase of amorphous CSH using the internal standard was applied and evaluated. It can be concluded that: • The external standard method demonstrated good and consistent results in the quantification of the amorphous phase. Both standard materials gave the results close to each other with a deviation of only around 0.2%. • Intermixing of the internal standard by hand was proven inappropriate method because of inconsistent results with high scatter. • Intermixing of the internal standard using an XRD mill resulted in good and consistent results. No amorphization effect was observed during the intergrinding procedure. • The proposed novel procedure for calibrating pseudo formula mass for the amorphous CSH phase using the internal standard was proven reliable and demonstrated good and consistent results. • The degree of hydration determined by XRD was slightly overestimated, probably due to the preferred orientation of portlandite. Considering the above, the following recommendations can be summarized: (i) the external standard is the simplest and still reliable procedure; (ii) the internal standard has to be intermixed using an XRD mill; (iii) PONKCS can be reliably used applying calibration based on the internal standard.
References 1. Di Filippo, J., Karpman, J., DeShazo, J.R.: The impacts of policies to reduce CO2 emissions within the concrete supply chain. Cem. Concr. Compos. 101, 67–82 (2019) 2. Scrivener, K.L., Snellings, R., Lothenbach, B.: A Practical Guide to Microstructural Analysis of Cementitious Materials. CRC Press Taylor & Francis Group, Boca Raton (2016) 3. Madsen, I.C., Scarlett, N.V.Y., Kern, A.: Description and survey of methodologies for the determination of amorphous content via X-ray powder diffraction. Zeitschrift Für Krist. 226, 944–955 (2011) 4. Scarlett, N.V.Y., et al.: Outcomes of the International Union of Crystallography Commission on powder diffraction round robin on quantitative phase analysis: samples 2, 3, 4, synthetic bauxite, natural granodiorite and pharmaceuticals. J. Appl. Crystallogr. 35, 383–400 (2002) 5. Bergold, S.T., Goetz-Neunhoeffer, F., Neubauer, J.: Quantitative analysis of C–S–H in hydrating alite pastes by in-situ XRD. Cem. Concr. Res. 53, 119–126 (2013) 6. Jansen, D., Goetz-Neunhoeffer, F., Lothenbach, B., Neubauer, J.: The early hydration of Ordinary Portland Cement (OPC): an approach comparing measured heat flow with calculated heat flow from QXRD. Cem. Concr. Res. 42, 134–138 (2012) 7. Cline, J.P., Von Dreele, R.B., Winburn, R., Stephens, P.W., Filliben, J.J.: Addressing the amorphous content issue in quantitative phase analysis: the certification of NIST standard reference material 676a. Acta Crystallogr. Sect. A Found. Crystallogr. 67, 357–367 (2011)
Comparison of Different Approaches for Quantification of Amorphous Phase
459
8. Westphal, T., Füllmann, T., Pöllmann, H.: Rietveld quantification of amorphous portions with an internal standard—mathematical consequences of the experimental approach. Powder Diffr. 24, 239–243 (2009) 9. Gutteridge, W.A., Dalziel, J.A.: Filler cement: the effect of the secondary component on the hydration of Portland cement. Cem. Concr. Res. 20, 853–861 (1990) 10. Jansen, D., Goetz-Neunhoeffer, F., Stabler, C., Neubauer, J.: A remastered external standard method applied to the quantification of early OPC hydration. Cem. Concr. Res. 41, 602–608 (2011) 11. Suherman, P.M., van Riessen, A., O’Connor, B., Li, D., Bolton, D., Fairhurst, H.: Determination of amorphous phase levels in Portland cement clinker. Powder Diffr. 17, 178–185 (2002). https://doi.org/10.1154/1.1471518 12. Jansen, D., Bergold, S.T., Goetz-Neunhoeffer, F., Neubauer, J.: The hydration of alite: a timeresolved quantitative X-ray diffraction approach using the G-factor method compared with heat release. J. Appl. Crystallogr. 44, 895–901 (2011) 13. Hesse, C., Goetz-Neunhoeffer, F., Neubauer, J.: A new approach in quantitative in-situ XRD of cement pastes: correlation of heat flow curves with early hydration reactions. Cem. Concr. Res. 41, 123–128 (2011) 14. Mertens, G., Elsen, J., Snellings, R., Machiels, L.: Rietveld Refinement strategy for quantitative phase analysis of partially amorphous zeolitised tuffaceous rocks. Geol. Belgica 133, 183–196 (2010) 15. Scarlett, N.V.Y., Madsen, I.C.: Quantification of phases with partial or no known crystal structures. Powder Diffr. 21, 278–284 (2006) 16. Snellings, R., Salze, A., Scrivener, K.L.: Use of X-ray diffraction to quantify amorphous supplementary cementitious materials in anhydrous and hydrated blended cements. Cem. Concr. Res. 64, 89–98 (2014) 17. Mejdi, M., Wilson, W., Saillio, M., Chaussadent, T., Divet, L., Tagnit-Hamou, A.: Quantifying glass powder reaction in blended-cement pastes with the Rietveld-PONKCS method. Cem. Concr. Res. 130, 105999 (2020) 18. Mezhov, A., Kulisch, D., Goncharov, A., Zhutovsky, S.: Effect of soaking time in a solvent on hydration stoppage of cement. RILEM Bookseries 26, 23–27 (2020) 19. Taylor, H.F.W.: Cement Chemistry, 2nd edn. Thomas Telford Publishing, London (1997) 20. Snellings, R.: X-ray powder diffraction applied to cement. In: Scrivener, K., Snellings, R., Lothenbach, B. (eds.) A Practical Guide to Microstructural Analysis of Cementitious Materials, 1st edn, pp. 107–176. Taylor & Francis Group, Boca Raton (2016)
Innovative FWD Testing on Concrete Slabs Rodrigo Díaz Flores1(B) , Valentin Donev2 , Mehdi Aminbaghai1 , Luis Zelaya-Lainez1 , Ronald Blab2 , Martin Buchta3 , Lukas Eberhardsteiner2 and Bernhard L. A. Pichler1
,
1 Institute for Mechanics of Materials and Structures, TU Wien, Vienna, Austria
[email protected]
2 Institute for Transportation, TU Wien, Vienna, Austria 3 Nievelt Labor GmbH, Höbersdorf, Austria
Abstract. Falling Weight Deflectometer (FWD) testing is a non-destructive method used to assess the condition of pavement structures. During FWD tests, a weight is let to freely fall upon the top surface of a slab, while a set of geophones located along the driving direction measure the deflections caused by the impact of the falling weight. These deflections serve as input for the back-calculation of subgrade properties. An unprecedented density of deflection data is obtained by means of multi-directional testing [1]. The present contribution refers to two original contributions. The first one refers to the presentation of experimental data from multidirectional FWD testing of one old and one new slab. The second original contribution refers to a structural analysis of the new slab, using a single slab model. In order to best-fit the measured deflections, a spatially uniform modulus of subgrade reaction and a spatially uniform eigendeflection are introduced as optimization variables. This strategy allows for reproducing the measured deflections very accurately, as quantified by a root mean squared error which is as small as 9.6 µm. This underlines that it is indeed possible to introduce a uniform modulus of subgrade reaction in the context of reliably explaining FWD deflection data, even when resorting to a single slab model, provided that a spatially uniform eigendeflection is accounted for. Keywords: concrete slabs · falling weight deflectometer · FWD · multi-directional testing · asymmetric slab behavior
1 Introduction The need of non-destructive and in-situ tests to characterize rigid pavement structures is well known, as unbound layers often exhibit site-dependent and seasonally varying properties. The popularity of deflection testing increased after a correlation was discovered between deflections and the fatigue response of pavement structures [2]. Falling Weight Deflectometer (FWD) tests are deflection-based tests in which a standardized weight freely falls upon the center of the concrete slab of a rigid pavement. Geophones, usually placed along the driving direction, record the deflection history of different points at the surface. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 A. J˛edrzejewska et al. (Eds.): SynerCrete 2023, RILEM Bookseries 43, pp. 460–472, 2023. https://doi.org/10.1007/978-3-031-33211-1_41
Innovative FWD Testing on Concrete Slabs
461
The identification of layer properties of the pavement structure results from a backcalculation. Pavement properties are taken as optimization parameters to fit measured deflections, assumed to be radial symmetrical. The most common models are the denseliquid model [3–6], assuming plate behavior of the slab on top of a Winkler foundation [7– 9], from which the stiffness properties of the concrete slab and the modulus of subgrade reaction k [6, 10] are back-calculated; and multi-layered half-space models [11–13], from which the stiffnesses of individual layers are back-calculated. Both assume semi-infinite slab sizes, delivering realistic results for some cases [14, 15]. The assumption of radial symmetry of the deflection profile was recently challenged by experimental data from multi-directional FWD tests [1, 16]. The innovative arrangement of geophones captured asymmetries in the deflection basin, clearly showing that old slabs behave in an asymmetric manner, probably because of long-term exposure to traffic loads. The authors also presented a finite-slab-size model. It was used to calculate subgrade stresses and a distribution of the effective modulus of subgrade reaction. The present study extends the database upon which multi-directional FWD tests have been performed. The finite-slab-size model developed in [1] is used for the backcalculation of subgrade properties. This study presents two original contributions: (i) new experimental data from FWD tests on two slabs with different behavior is presented; (ii) in the structural analysis using a single slab model, a uniform eigendeflection w0 has been introduced together with a uniform modulus of subgrade reaction as optimization variables to enhance the accuracy in reproducing the measured data. The study is organized as follows. Section 2 refers to experimental data from two newly characterized slabs: a freshly installed slab, and a 33-year-old slab. Section 3 refers to a structural analysis of the new slab based on the model presented in [1], but accounting for an eigendeflection w0 during the back-calculation. In Sect. 4, conclusions are drawn.
2 Multi-directional FWD Tests Multi-directional FWD tests were performed on two concrete slabs located on the first lane of the A2 highway in Austria: an old slab, A2-54003, and a freshly installed slab, A2-54440. The slabs had a length of 5.60 m and a thickness of 0.22 m. Their width was 3.80 m and 3.50 m, for the old and new slab, respectively. The concrete used for both slabs had a modulus of elasticity E = 36.5 GPa, a Poisson’s ratio ν = 0.2, and a mass-density ρ = 2452 kg/m3 . Therefore, the flexural stiffness of the slabs amounted to K = Eh3 /(12(1 − ν 2 )) = 33.7 MNm. The maximum force resulting from the falling weight was 189 kN for the old slab and 190 kN for the new slab. The slabs, located on the first lane, were symmetrically connected to neighboring slabs through dowels and tie bars. 2.1 Test Protocol Multi-directional FWD testing refers to a set of 27 standard FWD tests performed one after the other. During each test, deflections are measured along one of eight possible directions, starting with the driving direction (N in the local cardinal directional system),
462
R. Díaz Flores et al.
followed by the NE, E, SE, S, SW, W, NW, N. The first and the last set of tests correspond to the driving direction. For every direction, three tests are performed. The angle between subsequent directions is either 38°, between the NE-N, NW-N, SE-S, or SW-S directions, or 52°, for all other directions. Table 1. Radial distances [m] of the geophones (g) from the center of the slab as a function of the measurement direction. Test Directions
g=1 g=2 g=3 g=4 g=5 g=6 g=7 g=8 g=9
N, NE, E, S, W, NW 0.00
0.30
0.45
0.60
0.90
1.20
1.50
1.80
2.10
SE, SW
0.30
0.45
0.60
0.90
1.20
1.50
1.80
2.10
0.00
Deflections were measured by nine geophones along each direction, see Table 1 for their radial distance from the center. Along the E direction, the outermost geophone was no longer located on the tested slab. This was also the case for the outermost two geophones of the W direction. The measurements of these three geophones are excluded from further analysis. A total of 234 deflections were recorded from the 27 tests along eight directions, given that there are nine geophones on every direction, except for the three geophones excluded. The maximum for each of the 234 measured deflection evolutions is taken as representative for the foregoing structural simulations. 2.2 Experimental Data from the New Slab A2-54440 Tests were performed on a new slab. It was not yet open for regular traffic by the time the tests were performed. The 234 deflections measured in this slab are shown as circles and stars in Fig. 1, see also Appendix A. They correspond to six measurements for the geophone locations along the N direction, and three measurements for all other directions. A very satisfactory test repeatability is observed as the circles and stars are hardly distinguishable from each other in Fig. 1. 2.3 Experimental Data from the Old Slab A2–54003 Tests were also performed on an old slab that had already been in service for 33 years. Again, the 234 deflections are portrayed as circles and stars in Fig. 2, see also Appendix A. A very satisfactory test repeatability is observed as the circles and stars are hardly distinguishable from each other in Fig. 2. 2.4 Asymmetries in the Structural Behavior of the Slabs The asymmetry index developed in [1] is used to assess the level of asymmetry present in the structural behavior of both slabs. It reads as 2 wd wδ 1 l ∫0 − Ad ,δ = dr, (1) l wd (r = 0) wδ (r = 0)
Innovative FWD Testing on Concrete Slabs
(a)
463
(b)
Fig. 1. Multi-directional tests on slab A2-54440: 234 deflection maxima and spline interpolations (lines) along the (a) N, S, E, and W directions, and (b) diagonal directions.
(a)
(b)
Fig. 2. Multi-directional tests on slab A2–54003: 234 deflection maxima and spline interpolations (lines) along the (a) N, S, E, and W directions, and (b) diagonal directions.
where wd and wδ refer to the spline interpolation of the deflections measured along the directions d and δ, respectively, see Fig. 1 and Fig. 2; r refers to the radial distance from the center of the slab; and l refers to the radial length of integration, which is equal to 2.10 m, except for the E and W directions, where geophones were excluded from the data. The indexes d and δ run for all of the eight measurement directions. Asymmetries may result from different sources: (i) Slab-to-slab interaction; (ii) eccentric impact of the falling weight; (iii) finite size of the slabs; (iv) asymmetrical long-term service load. From these, only the finite slab size is expected to have an influence on both slabs, as central FWD tests were performed on symmetrically jointed slabs. Furthermore, the asymmetrical long-term load significantly affects the old slab
464
R. Díaz Flores et al.
only. Thus, while the old slab is expected to exhibit large asymmetries in many directions, the new slab is expected to behave mainly in a double-symmetric manner. To test this hypothesis, the asymmetry index was calculated for selected directions, see Table 2. Table 2. Asymmetry indexes for both tested slabs, see Eq. (1)
The average of the eight asymmetry indicators for the old slab is 8.50%. This asymmetry can be explained by the long-term service load that deteriorated the subgrade. It is concluded that the slab behaved in an asymmetrical manner. The average of the eight asymmetry indicators calculated for the new slab is 2.25%. This is by a factor of 3.8 smaller than that for the old slab. No particularly large asymmetries were found on any direction. This underlines that the new slab behaved in a virtually double-symmetric manner.
3 Structural Analysis of the New Slab A structural analysis is performed for the new slab, which behaves double-symmetrically. The new slab is modeled as a single slab with free edges resting on a Winkler foundation. The analysis is performed based on Kirchhoff-Love’s linear theory of thin plates. A Cartesian coordinate system is introduced, where the x-axis is chosen to coincide with the driving direction shown in Fig. 1. The thin-plate theory may be written as a boundary value problem consisting of one field equation, together with boundary conditions. The field equation reads as [1, 8]. 4 ∂ 4 w(x, y) ∂ 4 w(x, y) ∂ w(x, y) + kw(x, y) = p(x, y), (2) +2 + K ∂x4 ∂x2 ∂y2 ∂y4 where K = Eh3 /(12(1 − ν 2 )) = 33.7 MNm is the flexural rigidity of the plate, E is its modulus of elasticity, ν its Poisson’s ratio, w(x,y) is the deflection basin of the plate, p(x,y) the external load, and k is the modulus of subgrade reaction. Free edge boundary conditions are chosen, see [1], following the steps of [8, 9]. This choice of boundary conditions represents an extreme case, as it does not account for slab-to-slab interaction. The commonly used “dense-liquid” model [3–6] and the multi-layered half-space models [11–13], in turn, assume semi-infinite slab sizes. This is equivalent to assuming a completely rigid interaction between slabs, and constitutes the other extreme case. The real behavior of the structure is bounded by both extreme cases represented by the semi-infinite models and the herein presented finite-slab-size model.
Innovative FWD Testing on Concrete Slabs
465
The deflection field is a Fourier series of double-symmetric deflection modes [1], as N N nπ y mπ x cos , (3) Cm,n cos w(x, y) = m=0 n=0 a b where m = 0, 1, 3, 5,…,N and n = 0, 1, 3, 5,…,N refer to the chosen deflections modes. The external loading of the plate, see p(x,y) in Eq. (2), isproduced by the impact of the falling weight with radius rc . Thus, p(x,y) is equal to F/ rc2 π , for any position x and y within the radius rc ; and it is equal to 0 for any position x and y outside of it. 3.1 Identification of a Uniform Modulus of Subgrade Reaction The modulus of subgrade reaction is assumed to be uniform. It is optimized to fit the deflections as accurately as possible within the interval [0.1; 0.5] MPa/mm. The error between calculated and measured deflections, wcalc and wmeas respectively, was quantified by means of the root-mean-square-error (RMSE), as
2 1 9 8 meas RMSE = w xg , yg − wcalc xg , yg , (4) g=1 d =1 69 where xg , yg refer to the coordinates of the g th geophone along the direction d. The 69 geophone positions to which the denominator refers are the nine geophones on each of the eight directions, except for the three geophones excluded in the E and W directions.
Fig. 3. Optimization of k: RMSE between measured and calculated deflections.
The problem is solved numerically and iteratively. The minimum of the RMSE is found with k = 0.16 MPa/mm. The corresponding value of the RMSE amounts to 25.3 µm, see Fig. 3. However, the simulated deflection basin w(x,y) is not satisfactory, as large qualitative differences are found between calculated and simulated deflections, see Fig. 4. 3.2 Extension and Consideration of a Uniform Eigendeflection, w0 The strategy used to improve the reproduction of displacements is to introduce a uniform eigendeflection, w0 , in addition to k. Taking w0 into consideration, Eq. (2) turns into 4 ∂ w(x, y) ∂ 4 w(x, y) ∂ 4 w(x, y) + k(w(x, y) − w0 ) = p(x, y) K + 2 + (5) ∂x4 ∂x2 ∂y2 ∂y4
466
R. Díaz Flores et al.
(a)
(b)
Fig. 4. Optimized k: Measured (symbols) and simulated (lines) deflections, k = 0.16 MPa/mm.
Fig. 5. Optimization of k and w0 : RMSE between measured and calculated deflections.
The values of k and of w0 are optimized such as to reproduce the measured deflections as accurately as possible. The values k = 0.25 MPa/mm and w0 = 0.06 mm are found to minimize the RMSE, see Fig. 5. The smallest achievable error, calculated according to Eq. (4), amounts to 9.6 µm. A very satisfactory qualitative and quantitative agreement between measured and calculated deflections is obtained, see Fig. 6. The effective subgrade stresses acting on the bottom of the plate, considering the dead weight ρgh of the plate, are σzz,eff = ρgh + k(w − w0 ), see also Eq. (5) and Fig. 7. They are double-symmetric with respect to the N-S and E-W axes. The maximum stress at the center of the slab amounts to 73 kPa. The obtained distribution of subgrade stresses is realistic, as it is in equilibrium with the external load and the dead load of the slab, it satisfies the plate’s field equations and free-edge boundary conditions, and it reproduces the measurements very accurately. Notably, the non-linear distribution of subgrade stresses is obtained based on a uniform modulus of subgrade reaction, provided that a uniform eigendeflection is also accounted for.
Innovative FWD Testing on Concrete Slabs
(a)
467
(b)
Fig. 6. Optimized k and w0 : Measured (symbols) and simulated (lines) deflections, k = 0.25 MPa/mm and w0 = 0.06 mm.
Fig. 7. Distribution of effective subgrade stresses in Megapascal: σ zz,eff = ρgh + k(w − w0 ).
4 Conclusions The following conclusions are drawn from the results of the large density of deflection data provided by multi-directional FWD testing on two slabs: • The asymmetry index has been shown to successfully describe the asymmetric behavior of many slabs, supported by the studies in [1, 16].
468
R. Díaz Flores et al.
• Values of the asymmetry index smaller than 7% refer to virtually double symmetric behavior. The asymmetries occur mostly due to the geometry of the slabs and their interaction. Values larger than 7% refer to an asymmetric degradation of the subgrade. The structural analysis of the new slab allows for drawing the following conclusions: • The obtained distribution of subgrade stresses is realistic given that it is in equilibrium with the external load and the dead load of the slab, it fulfils the field equation and boundary conditions, and it reproduces the measurements very accurately. • The subgrade stresses computed with the presented single slab model represent an upper bound, because the used free-edge boundary condition are a lower bound for slab-to-slab load transfer. • Subgrade stresses computed with semi-infinite plate models represent a lower bound, as they assume a rigid connection between neighboring slabs. • It is possible to back-calculate a uniform modulus of subgrade reaction and to reproduce deflections measured during FWD testing reliably, even when using a single slab model, provided that a spatially uniform eigendeflection is also accounted for (Tables A1 and A2). Acknowledgements. Interesting discussions with Franz-Josef Ulm (MIT) are gratefully acknowledged.
Appendix A
Table A1. Maximum deflections measured during 27 FWD tests on the old slab A2–54003. Test Test Geophone Direction Number G = 1 G = 2 G = 3 G = 4 G = 5 G = 6 G = 7 G = 8 G = 9 d = 1 (N) i = 1
0.671
0.624
0.596
0.552
0.473
0.388
0.303
0.228
0.156
d = 1 (N) i = 2
0.676
0.624
0.603
0.556
0.476
0.389
0.304
0.227
0.157
d = 1 (N) i = 3
0.676
0.621
0.604
0.557
0.474
0.391
0.305
0.229
0.156
d= 2(NE)
i=1
0.676
0.630
0.607
0.563
0.482
0.396
0.299
0.233
0.166
d= 2(NE)
i=2
0.677
0.630
0.609
0.565
0.484
0.398
0.300
0.235
0.167
d= 2(NE)
i=3
0.676
0.634
0.607
0.564
0.481
0.396
0.312
0.234
0.167
d = 3 (E) i = 1
0.670
0.630
0.610
0.564
0.475
0.376
0.273
0.174
d = 3 (E) i = 2
0.676
0.636
0.613
0.566
0.476
0.377
0.274
0.176 (continued)
Innovative FWD Testing on Concrete Slabs
469
Table A1. (continued) Test Test Geophone Direction Number G = 1 G = 2 G = 3 G = 4 G = 5 G = 6 G = 7 G = 8 G = 9 d = 3 (E) i = 3
0.680
0.638
0.614
0.566
0.476
0.378
0.275
0.176
d=4 (SE)
i=1
0.670
0.621
0.595
0.548
0.470
0.392
0.310
0.235
0.169
d=4 (SE)
i=2
0.676
0.617
0.596
0.550
0.471
0.394
0.309
0.233
0.168
d=4 (SE)
i=3
0.676
0.620
0.596
0.549
0.472
0.394
0.311
0.234
0.168
d = 5 (S) i = 1
0.672
0.634
0.610
0.569
0.490
0.416
0.333
0.258
0.193
d = 5 (S) i = 2
0.673
0.632
0.616
0.570
0.493
0.418
0.334
0.260
0.191
d = 5 (S) i = 3
0.674
0.632
0.616
0.571
0.493
0.418
0.335
0.261
0.192
d= 6(SW)
i=1
0.672
0.579
0.539
0.491
0.400
0.312
0.228
0.151
0.096
d= 6(SW)
i=2
0.675
0.578
0.540
0.491
0.399
0.314
0.229
0.154
0.097
d= 6(SW)
i=3
0.674
0.577
0.541
0.491
0.401
0.314
0.228
0.155
0.096
d=7 (W)
i=1
0.663
0.586
0.552
0.487
0.369
0.251
0.133
d=7 (W)
i=2
0.668
0.590
0.556
0.491
0.372
0.252
0.136
d=7 (W)
i=3
0.669
0.592
0.556
0.492
0.373
0.253
0.138
d= 8(SW)
i=1
0.665
0.583
0.551
0.495
0.401
0.310
0.222
0.149
0.094
d= 8(SW)
i=2
0.668
0.584
0.554
0.497
0.402
0.312
0.222
0.149
0.094
d= 8(SW)
i=3
0.668
0.585
0.554
0.498
0.403
0.312
0.221
0.149
0.094
d = 9 (N) i = 1
0.660
0.607
0.583
0.538
0.457
0.373
0.297
0.220
0.147
d = 9 (N) i = 2
0.661
0.609
0.587
0.538
0.457
0.372
0.294
0.220
0.150
d = 9 (N) i = 3
0.663
0.609
0.586
0.538
0.458
0.374
0.295
0.219
0.150
470
R. Díaz Flores et al.
Table A2. Maximum deflections measured during 27 FWD tests on the old slab A2–54440. Test Test Geophone Direction Number G = 1 G = 2 G = 3 G = 4 G = 5 G = 6 G = 7 G = 8 G = 9 d = 1 (N) i = 1
0.336
0.284
0.259
0.229
0.178
0.138
0.105
0.078
0.060
d = 1 (N) i = 2
0.337
0.281
0.257
0.227
0.176
0.136
0.106
0.080
0.057
d = 1 (N) i = 3
0.338
0.282
0.257
0.227
0.174
0.136
0.101
0.077
0.060
d= 2(NE)
i=1
0.334
0.278
0.257
0.224
0.174
0.133
0.101
0.075
0.061
d= 2(NE)
i=2
0.336
0.279
0.259
0.225
0.176
0.134
0.102
0.076
0.060
d= 2(NE)
i=3
0.337
0.280
0.257
0.225
0.175
0.135
0.102
0.076
0.060
d = 3 (E) i = 1
0.339
0.275
0.249
0.219
0.167
0.127
0.091
0.063
d = 3 (E) i = 2
0.340
0.276
0.251
0.220
0.167
0.127
0.092
0.063
d = 3 (E) i = 3
0.338
0.276
0.251
0.221
0.167
0.127
0.091
0.063
d=4 (SE)
i=1
0.336
0.254
0.235
0.197
0.158
0.125
0.087
0.067
0.050
d=4 (SE)
i=2
0.338
0.256
0.239
0.201
0.159
0.128
0.085
0.068
0.049
d=4 (SE)
i=3
0.338
0.254
0.242
0.200
0.160
0.127
0.088
0.068
0.049
d = 5 (S) i = 1
0.329
0.276
0.253
0.220
0.179
0.134
0.099
0.081
0.057
d = 5 (S) i = 2
0.334
0.279
0.250
0.221
0.171
0.135
0.109
0.077
0.058
d = 5 (S) i = 3
0.333
0.280
0.249
0.221
0.172
0.132
0.105
0.079
0.058
d= 6(SW)
i=1
0.338
0.251
0.226
0.199
0.150
0.117
0.085
0.063
0.046
d= 6(SW)
i=2
0.338
0.251
0.225
0.199
0.151
0.117
0.084
0.063
0.045
d= 6(SW)
i=3
0.338
0.252
0.224
0.199
0.149
0.117
0.084
0.063
0.045
d=7 (W)
i=1
0.331
0.266
0.241
0.211
0.160
0.117
0.077
d=7 (W)
i=2
0.333
0.269
0.244
0.213
0.161
0.120
0.079
d=7 (W)
i=3
0.331
0.268
0.246
0.212
0.160
0.116
0.075 (continued)
Innovative FWD Testing on Concrete Slabs
471
Table A2. (continued) Test Test Geophone Direction Number G = 1 G = 2 G = 3 G = 4 G = 5 G = 6 G = 7 G = 8 G = 9 d= 8(SW)
i=1
0.334
0.266
0.243
0.213
0.163
0.124
0.089
0.066
0.049
d= 8(SW)
i=2
0.335
0.268
0.243
0.214
0.164
0.126
0.090
0.067
0.049
d= 8(SW)
i=3
0.337
0.269
0.241
0.211
0.165
0.125
0.090
0.064
0.047
d = 9 (N) i = 1
0.334
0.276
0.254
0.221
0.173
0.135
0.102
0.077
0.060
d = 9 (N) i = 2
0.335
0.276
0.257
0.222
0.174
0.139
0.102
0.079
0.059
d = 9 (N) i = 3
0.335
0.277
0.255
0.223
0.174
0.140
0.102
0.079
0.058
References 1. Díaz Flores, R., Aminbaghai, M., Eberhardsteiner, L., Blab, R., Buchta, M., Pichler, B.L.A.: Multi-directional Falling Weight Deflectometer (FWD) testing and quantification of the effective modulus of subgrade reaction for concrete roads. Int. J. Pavement Eng. 1–19 (2021) 2. Hveem, F.: Pavement deflections and fatigue failures. Highway Res. Board Bull. 114 (1995) 3. Westergaard, H.: Stresses in concrete pavements computed by theoretical analysis. Public roads (1926) 4. Westergaard, H.: New formulas for stresses in concrete pavements of airfields. Trans. Am. Soc. Civ. Eng. 113(1), 425–439 (1948) 5. Ioannides, A.: Dimensional analysis in NDT rigid pavement evaluation. J. Transp. Eng. 116(1), 23–36 (1990) 6. Khazanovich, L., Tayabji, S.D., Darter, M.I.: Backcalculation of layer parameters for performance for LTPP test sections. Slab on Elastic Solid and Slab on Dense-Liquid Foundation Analysis of Rigid Pavements, vol. 1. Federal Highway Administration. Office of Engineering (2001) 7. Winkler, E.: Die Lehre von der Elasticität und Festigkeit mit besonderer Rücksicht auf ihre Anwendung in der Technik [Lessons on elasticity and strength of materials with special consideration of their application in technology]. Dominicus, Prague (1867) 8. Vlasov, V.: Beams, plates and shells on elastic foundation. Israel Program for Scientific Translation (1966) 9. Höller, R., et al.: Rigorous amendment of Vlasov’s theory for thin elastic plates on elastic Winkler foundations, based on the principle of virtual power. Eur. J. Mech.-A/Solids 73, 449–482 (2019) 10. Hall, K.T., Darter, M.I., Hoerner, T.E., Khazanovich, L.: LTPP data analysis. Phase I: validation of guidelines for K-value selection and concrete pavement performance prediction. Federal Highway Administration No. FHWA-RD-96-198 (1997) 11. Burmister, D.: The general theory of stresses and displacements in layered systems I. J. Appl. Phys. 16(2), 89–94 (1945) 12. Kausel, E., Roësset, J.: Stiffness matrices for layered soils. Bull. Seismol. Soc. Am. 71(6), 1743–1761 (1981)
472
R. Díaz Flores et al.
13. Pan, E.: Static response of a transversely isotropic and layered half-space to general surface loads. Phys. Earth Planet. Inter. 54(3–4), 353–363 (1989) 14. Setiadji, B., Fwa, T.: Consideration of finite slab size in backcalculation analysis of jointed concrete pavements. Transp. Res. Rec. 2005(1), 124–142 (2007) 15. Liu, W., Fwa, T.F.: Nine-slab model for jointed concrete pavements. Int. J. Pavement Eng. 8(4), 277–306 (2007) 16. Díaz Flores, R., Aminbaghai, M., Eberhardsteiner, L., Blab, R., Buchta, M., Pichler, B.L.A.: Star-shaped Falling Weight Deflectometer (FWD) testing and quantification of the distribution of the modulus of subgrade reaction. In: Computational Modelling of Concrete and Concrete Structures, pp. 284–293. CRC Press (2022) 17. Martin, U., Rapp, S., Camacho, D., Moormann, C., Lehn, J., Prakaso, P.: Abschätzung der Untergrundverhältnisse am Bahnkörper anhand des Bettungsmoduls [Estimation of the subgrade relations on the train body with the help of the modulus of subgrade reaction]. ETR-Eisenbahntechnische Rundschau 5, 50–57 (2016) 18. Murthy, V.: Textbook of Soil Mechanics and Foundation Engineering. CRC Press, Boca Raton (2002) 19. Nielson, F., Bhandhausavee, C., Yeb, K.S.: Determination of modulus of soil reaction from standard soil tests. Highway Res. Rec. 284, 1–12 (1969)
Numerical Simulations for the Determination of Chloride Diffusivity in Reinforced Concrete Under Tensile Load Amandine Asselin1,2 , Jean-Philippe Charron2(B) , Clélia Desmettre2 , Farid Benboudjema1 , and Cécile Oliver-Leblond1 1 Université Paris-Saclay, CentraleSupélec, ENS Paris-Saclay, CNRS, LMPS - Laboratoire de
Mécanique Paris-Saclay, 4 avenue des sciences/8-10 rue Joliot Curie, 91190 Gif-sur-Yvette, France 2 Department of Civil, Geological and Mining Engineering, Research Center on Concrete Infrastructures (CRIB), Group for Research in Structural Engineering, Polytechnique Montréal, succ. Centre-Ville, C.P. 6079, Montréal, Québec H3C 3A7, Canada [email protected]
Abstract. In service conditions, reinforced concrete structures are multi-cracked due to the loads they are submitted to. It enables aggressive agents, such as chlorides, to penetrate the concrete cover and could initiate steel rebar corrosion leading to more structural damage. Some experimental programs focusing on chloride penetration were conducted on plain or cracked concrete, but mainly unloaded and unreinforced concrete specimens were used for the measurement. These tests highlight the linear dependence of chloride diffusivity to crack opening. However, they do not take into account the presence of rebar, and cracks are partially or totally closed up during the test, which is different from service conditions. This research project aims to understand the influence of micro- and macro-cracks on chloride diffusivity in conditions close to the service ones. To achieve this objective, three steady-state accelerated migration tests under electrical field are to be carried out on a same reinforced concrete specimen kept under a tensile load representative of a structural one. This non-standard chloride penetration test requires the adaptation of the experimental protocol. This paper presents preliminary adaptation work using numerical simulations developed with Comsol Multiphysics®. The impact of testing configuration and parameters, as well as cracking was investigated. Comparison of simulations and preliminary experimental results are also given. Keywords: Simulations · Chloride migration · reinforced concrete · tensile load · cracking
1 Introduction Reinforced concrete structures are often multi-cracked because they are submitted to mechanical and environmental loads during their service life. These cracks are a preferential path for aggressive agents carried by water, such as chlorides, to enter the concrete © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 A. J˛edrzejewska et al. (Eds.): SynerCrete 2023, RILEM Bookseries 43, pp. 473–484, 2023. https://doi.org/10.1007/978-3-031-33211-1_42
474
A. Asselin et al.
cover. When chlorides reach steel rebars in sufficient quantity, they initiate rebar corrosion, which progressively damages the structure. Chloride diffusivity measurements allow to quantify the chloride penetration capability and give an indicator of concrete durability. However, chloride diffusivity measurements are often conducted on plain concrete without reinforcement [1] and, in some cases, on cracked concrete without reinforcement. In the latter case, specimens are damaged with a splitting test and then unloaded before the diffusivity measure [2–4]. Through these studies, it has been shown that chloride diffusivity linearly increases with the crack opening until it reaches the chloride diffusivity value in water when the crack opening is over about 80 µm [5]. Nevertheless, these studies do not reflect the reality of the concrete structures in service conditions, which are reinforced, cracked and loaded continuously. This research project aims to understand the impact of the micro- and macro-cracks found in service conditions in reinforced concrete structures on chloride diffusivity. Accelerated migration tests are simultaneously conducted on three different parts of a reinforced concrete tie specimen, which is kept under a service-conditions tension load. The chloride diffusivity measurements are performed under an electrical field in steadystate conditions. Because the thickness of the tie-specimen, representative of concrete cover in real structures, is larger than specimens used in standard migration tests, and also due to the presence of three migration cells on the tie-specimen, the standard migration test had to be adapted for this project. This paper presents preliminary numerical works conducted to evaluate the required adaptations to the NT Build 335 procedure [6]. It investigates variations in the electrical field and the upstream chloride concentration with Comsol Multiphysics® simulations. The effect of proximity between migration cells and the impact of a straight crack is also analysed. Comparisons of simulations and preliminary experimental results are given.
2 Methodology 2.1 Accelerated Migration Test Accelerated migration tests are generally carried out on cylindrical specimens with a thickness of 0.5 to 5 cm [1–7]. In this project, they are conducted on tie-specimens with thickness of 9 cm, offering a classical cover thickness (40 mm around a rebar of 10 mm in diameter). Migration test implies two cells attached to two opposite sides of the specimen. The first cell (upstream) contains a chloride solution, whereas the other one (downstream) contains a solution without chlorides. A constant electrical potential is then applied between these cells to accelerate the migration of chlorides from the upstream to the downstream cells through the concrete. The electrical potential is applied via an electrode placed in each cell: it is called anode in upstream cell and cathode in downstream cell. Figure 1 illustrates the set-up for a preliminary test carried out on a small 9 cm thick concrete prism. At the beginning of a migration test, downstream chloride concentration remains null until chlorides have begun to pass through concrete. Then concentration increases in a non-linear way during the non-steady state. Finally, downstream concentration in chlorides cdown [mmol/L] linearly increases with time t [s] at the steady state, during
Numerical Simulations for the Determination of Chloride Diffusivity
475
Fig. 1. Accelerated chloride migration test device.
which the slope of the chloride concentration evolution is proportional to diffusivity De [m2/s] according to Eq. (1). cdown =
cup ×V ×S×F R×T ×V ×e De t
(1)
where cup [mmol/L] is the upstream concentration, V [V] is the potential difference, S [m2 ] is the surface of the specimen exposed to the diffusion tests, F [C/mol] is the Faraday constant, R [J/mol/K] is the molar gas constant, T [K] is the temperature, V [m3 ] is the downstream cell volume and e [m] is the thickness of the specimen. Equation (1) is derived from the Nernst-Planck equation and is applicable under the following assumptions: – Diffusion and electrical fields are both assumed to be unidirectional; – The concrete is saturated, which means convection is negligible and is not considered in the modelling; – Interactions between chlorides and cementitious matrix are neglected; – Migration of the chlorides under the electrical field is predominant compared to natural diffusion; – The upstream chloride concentration is considered constant. In view of the numerous parameters involved in Eq. (1) and the various assumptions made, several numerical simulations were carried out to study the impact of electrical field and upstream concentration variations. The minimal distance between migration cells and the effect of a crack were also investigated. 2.2 Numerical Simulation The following numerical simulations have been carried out using modules “Transport of Diluted Species in Porous Media” and “Electrostatic” in Comsol Multiphysics®. The aim of this project is to make simultaneous migration tests on three different parts of a tie-specimen with a 9 cm × 9 cm cross-section and 61 cm length. Dimensions were chosen according to a previous project concerning water permeability measures
476
A. Asselin et al.
[8]. A 2D reference simulation has thus been conducted with a concrete specimen with these dimensions (Fig. 2). This reference simulation considers only one migration test carried out at mid-height of the specimen (Fig. 2). The upstream concentration cup (1060 mmol/L) and the electrical ground are imposed on a height of 15 cm corresponding to the diameter of the upstream cell (Fig. 1). It is generally recommended to impose an electrical potential difference of 12 V for 5 cm thick concrete specimens to keep natural diffusion negligible compared to migration transport [5, 6]. That is why the electrical potential, imposed facing on the opposite side, has been chosen equal to 12 V. At the same location, the downstream cell, with dimensions of 15 cm by 15 cm, is modelled and filled with water, of which diffusivity is equal to Dw = 2 × 10–9 m2 /s. The surface average of chloride concentration is evaluated in the latter cell to determine diffusivity. Finally, the chloride flux and the electrical load are set to null on the rest of the specimen boundary. Concrete porosity is not physically considered in the modelling, it is included numerically by setting the chloride diffusivity Dc = 1 × 10–11 m2 /s to the concrete specimen.
Fig. 2. Geometry and parameters of the reference simulation.
Figure 3 gives a 2D representation of the electrical field, and concentration 100 h after the beginning of the simulation. Time-evolution of the chloride concentration in the downstream cell is also given.
Numerical Simulations for the Determination of Chloride Diffusivity
477
Fig. 3. Result of a simulation, a) electrical field (V/m), b) chloride concentration (mol/m3 ) and flow lines for the reference test at t = 100 h and, c) chloride concentration evolution in the downstream cell.
The electrical field is constant between both cells, except on the borders where irregularities can be observed (Fig. 3a). This is linked to the end of the electrode and is expected also present in the experimental test. Indeed, field lines are parallel and regularly distributed in the middle of electrodes. But, on the electrodes rim, field lines are incurved and join the electrodes at the same place, leading to field lines concentration and a higher value of the electrical field. The same phenomenon can be observed with the chloride concentration: a higher value is noticeable close to the electrode ends (Fig. 3b). These perturbations at the extremities lead to a slight decrease in diffusivity. Thus, by evaluating the diffusivity from the evolution of the chloride concentration in the downstream cell, a value of De equal to 0.91 × 10−11 m2 /s is determined (Fig. 3c), which is 9% less than the one imposed for the concrete specimen in the simulation (Dc in Fig. 2).
3 Results of the Sensitivity Studies This section aims to evaluate the impact of some testing parameters by modifying the reference simulation. Each different value of parameters (one point on following figures) corresponds to one simulation. 3.1 Electrical Field In this section, the voltage was changed while keeping the same configuration as the reference simulation to study the impact of the potential difference on the diffusivity. Figure 4 illustrates the modification of the diffusivity when increasing the potential difference.
478
A. Asselin et al.
Fig. 4. Evolution of diffusivity according to the electrical potential difference.
The diffusivity value increases with the potential difference to reach a 27% higher value than the reference when increasing the electrical potential difference to 110 V. It means that the results obtained during migration tests are very sensitive to voltage variations and can be overestimated for high potential differences. Therefore, even if a high potential is a way to achieve test results more quickly, it will be essential to use reasonable value as well as same potential difference to compare results in experimental tests. Considering the previous conclusions as well as the risk of formation of undetectable chlorinated compounds such as dichlore or hypochlorite when the voltage is too high, a potential difference of 58.5 V was chosen for all the experimental tests [9]. This electrical potential value was numerically tested with Comsol Multiphysics® and involved a diffusivity equal to 1.09 × 10−11 m2 /s and a difference of 19% with the reference value, as shown in Fig. 4. 3.2 Chloride Upstream Concentration The influence of upstream chloride concentration has been studied and shown in Fig. 5. The upstream cell chloride concentration does not influence chloride diffusivity in the concrete. However, it is assumed that upstream concentration is constant during the simulation, which is not the case experimentally when the upstream solution is not regularly renewed. Another simulation with an upstream cell filled initially with a chloride solution and free to equilibrate with the system has shown a decrease in concentration with time, as shown in Fig. 6. This decrease reaches 9% of the initial upstream chloride content after 100 h of migration test. Afterwards, this linear decrease was set as a chloride content boundary evolution instead of a constant one. The diffusivity obtained was 0.88 × 10−11 m2 /s, which is only 3% lower than the reference value. This observation means that variation of chloride concentration in the upstream cell should have a negligible effect in experiments.
Numerical Simulations for the Determination of Chloride Diffusivity
479
Fig. 5. Evolution of diffusivity according to upstream concentration.
Fig. 6. Decrease of the chloride concentration in the upstream cell.
3.3 Distance Between 2 Cells In this section, two sets of chloride migration cells are considered in the simulations and the impact of the distance between cells is studied. Cells are located in such a way that geometry is symmetrical to the middle horizontal axis, which means migration phenomena are the same in both migration cells by symmetry. Results of diffusivity are given according to the distance between cells in Fig. 7. When cells are closer than 10 cm, diffusivity is higher than the reference simulation including only one set of migration cells. Indeed, electrical fields of both sets of cells are superposed and locally higher, which drives more easily chlorides and lead to an overestimation of diffusivity. For larger distance between the sets of cells, the diffusivity decreases until it reaches the reference value for a distance of 15 cm. Then, it starts to increase again when the sets of cells become close to the top and bottom specimen border. Indeed, a reflection against the top and bottom surfaces reduces the chloride transit time from the upstream to the downstream cells, artificially raising the diffusivity.
480
A. Asselin et al.
Fig. 7. Impact of the distance between 2 cells on the chloride diffusivity.
The same simulations were thus conducted on a 2 m-long specimen to eliminate the specimen length effect (Fig. 8).
Fig. 8. Impact of the distance between 2 cells on the chloride diffusivity in a 2m-long specimen.
Increasing the specimen length from 0.6 to 2 m brings a very little change in the value of De when in presence of one centred cell (0.92 × 10−11 m2 /s in comparison to 0.91 × 10−11 m2 /s for the 0.61 m-long specimen). The new value of 0.92 × 10−11 m2 /s was then used as the reference in Fig. 8 for proper comparison of the effect of two cells versus one cell on the 2 m-long specimen. As observed in Fig. 7, diffusivity still decreases in Fig. 8 when the distance between cells increases up to 10 cm. However, it then remains constant. This confirms that the increase of diffusivity observed in Fig. 7 when the sets of cells were moved away from each other (for distances superior to 15 cm) is related to the specimen length effect. Consequently, recommendation for the minimal distance between migration sets of cells is 10 cm to avoid spurious effects, whereas the minimal distance to specimen border is 5 cm.
Numerical Simulations for the Determination of Chloride Diffusivity
481
These values have been experimentally verified. A migration test has been carried out on ordinary concrete, reinforced with a 10M epoxy-coated steel rebar. The water/binder ratio was 0.60 and the cement was a standard North American GU type Portland cement. At the end of the test, the specimen was split. A solution of silver nitrate was then sprayed to determine which part of the concrete had been reached by chlorides. The surface of concrete contaminated with chlorides becomes white due to the formation of silver chloride, whereas it stays brown in the absence of chloride [10]. Figure 9 shows that chloride diffused in concrete in a zone wide of 17 cm, corresponding to a distance of 4.5 cm outside the cell. Therefore, sets of cells must be spaced at least by more than 9 cm and located at more than 4.5 cm from the specimen top and bottom surfaces to avoid interferences. This is consistent with numerical results.
Fig. 9. Concrete contaminated with chloride (white surface) and without chlorides (brown) revealed by the addition of silver nitrate.
3.4 Influence of a Straight Crack A straight and constant opening crack filled with water was added at mid-height of the specimen with a unique set of migration cells (Fig. 10a). Several numerical simulations were performed for crack openings ranging from 0.05 mm to 1 mm. A comparison of those simulations with the theory is made. For this, a parallel theoretical model is used: it considers two plain concrete surfaces surrounding a water surface corresponding to the crack. The diffusivity of the cracked concrete is then deduced from the diffusivities of the plain concrete and the water [5] (Fig. 10b).
482
A. Asselin et al.
Fig. 10. Impact of crack, a) cracked specimen simulated, b) decomposition of cracked concrete in plain concrete and crack.
Diffusivity can be obtained from the theoretical model and compared to simulations in Fig. 11. A good consistency between simulation and theoretical values of diffusivity as a function of crack opening in cracked concrete was obtained. This means that Comsol Multiphysics® simulations reproduce correctly the impact of perfect crack (straight crack with smooth surfaces with no chemical and physical interaction with water and chloride).
Fig. 11. Comparison between theoretical and simulated diffusivity values with a crack opening in the specimen. A variability threshold corresponding to 20% of the reference diffusivity has been added.
Numerical Simulations for the Determination of Chloride Diffusivity
483
The variability in diffusivity appears to be between 12.4 and 19.5% [11]. To be sure to quantify the effect of a crack and not the variability, a variability threshold equal to 20% of the reference diffusivity value was considered (Fig. 11). It means the crack opening has to reach 0.1 mm to cause a significant diffusivity variation compared to inherent concrete variability. Experimental loading will be adapted to reach the crack opening.
4 Conclusion This research project aims to understand the impact of micro- and macro-crack on chloride diffusion in reinforced concrete structures in service conditions. To this end, three simultaneous accelerated migration tests under an electrical field are carried out on the same reinforced concrete tie-specimen under tensile load. These non-standard migration tests have required some numerical work to establish the optimal testing configuration. Comsol Multiphysics® was used to model chloride migration in the concrete specimen in different conditions. A set of electrical and transport parameters were imposed to the specimen, whereas the evolution of downstream chloride concentration provides an estimation of chloride diffusivity based on the Nernst-Planck equation. A reference simulation was used to evaluate the impact of the different sets of parameters. The electrical field map in the specimen allows reporting the boundary effect of electrodes with a field lines concentration, leading to a locally higher concentration. Diffusivity found in the simulation is slightly lower than the one imposed, but other diffusivity estimations will be compared to this reference value. Several electrical potential differences in migration test, as well as upstream cell chloride concentrations, were tested. While upstream cell chloride concentration has no impact on the diffusivity, this latter increases with the potential difference. Therefore, keeping the same voltage during the experimental campaign will be mandatory to compare results with each other. The assumption of constant upstream concentration was also verified. Even if the chloride content has decreased by 9% 100 h after the beginning of the accelerated migration test, diffusivity is almost not impacted. It means that the change of upstream cell solution is not necessary during the experimental test. In the final experimental configuration, at least two sets of migration cells will be installed simultaneously on the same specimen. Therefore, the impact of one on the other was verified by simulations. Interferences between the cells are observed when they are separated by less than 10 cm. Furthermore, spurious effects are noticed when sets of cells are too close to specimen extremities. The addition of silver nitrate on a concrete specimen after a migration test showed chloride diffusion up to 4.5 cm around each cell. It corroborates the simulation results. A concrete specimen with a straight crack was also simulated and compared to a theoretical model. Numerical and theoretical results are similar. The conclusion is that the crack opening must reach 0.1 mm to modify significantly the diffusivity in view of the variability of the diffusivity of plain concrete.
484
A. Asselin et al.
References 1. Sanchez, T.: Étude comparative de la diffusion d’espèces anioniques et cationiques dans les matériaux cimentaires: étude expérimentale et numérique. Ph.D. thesis, Université de La Rochelle, La Rochelle (2018) 2. Ismail, M., Toumi, A., François, R., Gagné, R.: Effect of crack opening on the local diffusion of chloride in cracked mortar samples. Cem. Concr. Res. 38(8–9), 1106–1111 (2008) 3. Kessler, S., Thiel, C., Grosse, C.U., Gehlen, C.: Effect of freeze–thaw damage on chloride ingress into concrete. Mater. Struct. 50(2), 1–13 (2016). https://doi.org/10.1617/s11527-0160984-4 4. Fu, C., Ye, H., Jin, X., Yan, D., Jin, N., Peng, Z.: Chloride penetration into concrete damaged by uniaxial tensile fatigue loading. Constr. Build. Mater. 125, 714–723 (2016) 5. Djerbi, A., Bonnet, S., Khelidj, A., Baroghel-Bouny, V.: Influence of traversing crack on chloride diffusion into concrete. Cem. Concr. Res. 38(6), 877–883 (2008) 6. Nordtest method: Chloride diffusion coefficient from migration cell experiments. NT Build No. 335 (1997) 7. Andrade, C., Sanjuàn, M.: Experimental procedure for the calculation of chloride diffusion coefficients in concrete from migration tests. Adv. Cem. Res. 23(6), 127–134 (1994) 8. Desmettre, C.: Contribution à l’étude de la perméabilité du béton armé sous sollicitations statiques et cycliques. Ph.D. thesis, Polytechnique Montréal, Montréal (2011) 9. Asselin, A., Charron, J.-P., Desmettre, C., Simon-Boursier, O., Benboudjema, F., OliverLeblond, C.: Évaluation du coefficient de diffusion des chlorures dans le béton armé sous chargement mécanique. 25e Congrès Français de la Mécanique (2022) 10. Collepardi, M., Marcialis, A., Turriziani, R.: Penetration of chloride ions into cement pastes and concretes. J. Am. Ceram. Soc. 55, 534–535 (1972) 11. Aît-Mokhtar, A., Belarbi, R., et al.: Experimental investigation of the variability of concrete durability properties. Cem. Concr. Res. 45, 21–36 (2013)
Valorisation and Recycling of Non-binder Components of Concrete
Deconstructable Concrete Structures Made of Recycled Aggregates from Construction & Demolition Waste: The Experience of the DeConStRAtion Project Marco Pepe1,2(B)
, Julien Michels3 , Giulio Zani4 and Enzo Martinelli2
, Marco Carlo Rampini4
,
1 TESIS srl, Via Giovanni Paolo II, 132, 84084 Fisciano, (SA), Italy
[email protected]
2 Department of Civil Engineering, University of Salerno, Via Giovanni Paolo II, 132, 84084
Fisciano, (SA), Italy 3 re-fer AG, Riedmattli, 9, CH-6423 Seewen, Switzerland 4 Department of Civil and Environmental Engineering (DICA), Politecnico di Milano, Piazza
Leonardo Da Vinci 32, 20133 Milan, Italy
Abstract. The construction sector is one of the most energy-intensive and rawmaterial demanding human activities and hence contributes a significant share of greenhouse gas emissions. Therefore, making the construction sector greener is one of the main challenges for policy makers, private companies and the scientific community. To this aim, one of the most promising actions is based on recycling Construction and Demolition Waste (CDW) and converting them into secondary raw materials for the construction sector itself. On the other hand, the reduction of the environmental impact can be further amplified through the optimization of the production, assembly and deconstruction/reuse procedures as well as through the maximization of the service life. In this context, the present paper presents the main results of a research project aimed at exploring the possibility to define a concept for deconstructable buildings made by prefabricated modular Recycled Aggregate Concrete (RAC) elements. The concept is based on assembling duly sized and designed prefabricated blocks by means of a prestressing system based on an innovative memory®-steel technique. The target building typology is that of highly modular structures, which generally demand high construction speed and the possibility of reconversion of internal spaces. Several issues are targeted to achieve within the project activities: proper design of mixtures leading to durable RAC members, the most appropriate structural configuration, and assembling technique are considered for achieving deconstruction capabilities. A proof of concept is then designed, produced and tested. Keywords: Recycled Aggregate Concrete · Design for Disassembly · Modular structures
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 A. J˛edrzejewska et al. (Eds.): SynerCrete 2023, RILEM Bookseries 43, pp. 487–499, 2023. https://doi.org/10.1007/978-3-031-33211-1_43
488
M. Pepe et al.
1 Introduction The increase in urbanization has led to an unprecedented generation of Construction and Demolition Waste (CDW), as it is considered as one of the most relevant issues due to its environmental, economic and social impacts [1]. Moreover, there is a growing trend in the consumption of aggregates in construction since global demand is expected to increase from 45 (in 2017) to 66 (by 2025) billion tons [2]. Nevertheless, Recycled Aggregates (RAs) from CDW are still significantly under-exploited and, without a clear entry point into the circular economy model, the comprehensive set of EU policies cannot be achieved [3]. The main issues related to the promotion and the diffusion of RAs is the lack of confidence of consumers and builders regarding the quality of the generated secondary raw materials due to their heterogeneity [4]. Despite the very heterogeneous composition of CDW, there is a high potential for their recycling since around 40% to 85% is composed of concrete, mortar, rocks and ceramic materials [5]. Several research groups worldwide demonstrated the feasibility of using RAs also for structural concrete production: CDW recycling is an opportunity to extract economic and environmental benefits from waste and is therefore a business activity in which it is worth investing [6]. In recent years, many authors have investigated the role of innovative cement-based structural systems in reducing the dead loads, increasing durability and extending the building lifespan through using advanced materials such as ultra-high or very-high-performance cementitious composites, allowing the minimization of the concrete volume [7]. On the other hand, limited attempts to reuse concrete building elements are documented in the literature: current trends, in fact, produce structures devised to maximize the construction speed, ignoring future deconstruction, transformation capacity and reuse [8]. This clearly collides with the evidence that, with reference to concrete bearing members, 75 to 95% of the whole environmental impact is generated in the production phase [9]. In this context, the present study summarizes the results of the DeConStRACtion (Towards Next-generation Deconstructable Concrete Structures made of Recycled Aggregates from Construction & Demolition Waste) project aimed at proposing an innovative strategy for sustainable and durable buildings by maximizing the use of recycled constituents from CDW while optimizing the structural design for “deconstructable” prefabricated concrete elements [10]. This objective contributes to the implementation of the Circular Economy for a sustainable future and to the definition of more inclusive, circular and sustainable socio-economic developmental models for the Construction sector [11]. It contributes to the rearrangement of products manufacturing in order to facilitate the extension of shelf life, recycling, reuse, as well as energy and material recovery over a long-time span. The main novelty in the project is to use RAs to create a modular deconstructable concrete/connector combination to be used a multitude of times, hence significantly enhancing the sustainability idea. Despite many scientific advances in the field of recycled concrete technology, the intensive use at the structural scale and a comprehensive design for disassembly are still missing. The choice of assuming concrete as the only reference material is linked to the promising characteristics of resistance, durability, ability to take any shape, low cost and the possibility of being effectively used in areas of the World characterized by different economic contexts. Moreover, the prefabrication technology, characterized by minimal “wet” operations, can facilitate disassembly of building elements.
Deconstructable Concrete Structures Made of Recycled Aggregates
489
2 Design and Prototyping This section describes the activities conducted in the framework of the research project DeConStRACtion aimed at the prototyping and mechanical testing of innovative precast solutions for flat slabs. 2.1 Design of the Structural Components During the preliminary design phase multiple alternative sections were analysed and evaluated, paying particular attention to ease of prototyping and compatibility of solutions with the available full-scale test infrastructure. Then, the following two alternative solutions were conceived and selected): – Solution A: traditional monolithic solution produced with ordinary concrete and prestressed with high-strength steel bonded tendons; – Solution B: innovative modular solution made with recycled aggregate concrete, post-tensioned with shape memory alloy bars (see Fig. 1).
Fig. 1. Schemes of the innovative modular solution.
In Fig. 1 only 1 m of the monolithic traditional solution is displayed, while in the case of the innovative solution a module of 1 m is reported. 2.2 Prototyping Activities The prototyping activities were defined ensuring production efficiency and adherence to geometric tolerances. More specifically, the following samples were produced: – 2 full-scale traditional shallow beams (0.14 m × 0.60 m × 4.00 m, see Fig. 2) produced with ordinary self-compacting concrete (SCC - solution A), bonded tendons and stirrups (see Table 1); – 10 innovative modules (0.14 m × 0.60 m × 1.00 m, see Fig. 3) produced with recycled aggregate concrete (RAC - solution B), shear keys, stirrups and encased plastic ducts, to be assembled through a subsequent post-tensioning operation. Eight modules were used to make two full-scale shallow beams and the remaining two modules were produced as spare elements (see Table 1).
490
M. Pepe et al.
Fig. 2. Scheme of type A prototypes (traditional): total length 4000 mm.
Fig. 3. Scheme of type B prototypes (innovative): total length 4000 mm (4 × 1000 mm). Table 1. De tails of the two alternative solutions (A traditional and B innovative). Type A
n 1+1
Concrete
Longitudinal reinforcement
Transverse reinforcement
SCC
4 × 0.5 bonded high-strength
4-legged 8/100 mm stirrups
steel B
10
RAC
Post-tensioning re-fer bars
4-legged 8/100 mm stirrups
2.3 Raw Materials and Mixture Proportioning The recycled concrete aggregates (RCAs) used in the present project were obtained from dismissed reinforced concrete elements. Once the “original” waste concrete source was homogenized and separated from possible impurities, the following phases were executed: size reduction, sieving and cleaning. These three phases were executed in one step since the machine available in the powerplant allow to fill the crushing mill with concrete debris which are later cleaned (by washing) and sieved. During this phase, the steel remaining were also removed as well as the fine fraction (i.e., nominal diameter lower than 4 mm) was discharged. The process was set in order to obtain two different class sizes: – coarse RCAs with nominal diameter ranging between 4 mm and 8 mm; – coarse RCAs with nominal diameter ranging between 8 mm and 16 mm. The physical properties of the particles employed herein were evaluated in accordance with the EN 1097-6 [12] for water absorption capacity at 24 h (A) and particle density (γ): Fig. 4 summarizes the results obtained for natural and recycled aggregates. As expected, the natural sand is characterized by a higher water absorption capacity than the natural coarse fraction. On the other hand, the recycled aggregates present a significantly higher water absorption capacity at 24 h moving from 1.20% to 4.14% in the case of 4–8 mm fraction and from 1.00% to 4.33% for the bigger (8–16 mm)
Deconstructable Concrete Structures Made of Recycled Aggregates
491
Fig. 4. Properties of natural and recycled aggregates.
fraction. Contrarily, regarding the particle density of RCA emerges a slightly different result in comparison with the companion natural aggregate. The higher water absorption capacity and slightly lower particle density is due to the composite nature of the RCA which contain natural rocks and “old” mortar derived from the previous concrete (the latter can be defined as Attached Mortar) [13]. It is worth to mention that the produced RCAs stay within the limits proposed by the EN 206 for the use of recycled aggregates (i.e., particle density above 2100 kg/m3 ) [14]. In the last years, Rangel et al. [15] proposed the following classification for identifying the “quality” of the produced RAs: • Class A: RCAs coarse particles characterized by water absorption values (A) ranging between 0% and 3% and an Attached Mortar (AM) content lower than 10%. Obviously, this class includes also natural aggregates; • Class B: RCAs coarse particles characterized by A values ranging between 3% and 6% and an AM content between 10% and 35%; • Class C: RCAs coarse particles characterized by A values ranging between 6% and 9% and an AM content between 35% and 55%; • Class D: RCAs coarse particles A values above 9% and an AM content above 55%. In accordance with this classification, the RCAs produced herein can be classified on Class B clearly indicating a good quality of the produced aggregates (Fig. 5). For the concrete production, two nominally identical concrete mixes were considered: – an ordinary self-compacting concrete (SCC) with a nominal C50/55 strength; – a recycled aggregate concrete (RAC), in which part of the aggregates (the 8–15 mm fraction) was substituted with construction demolition waste materials (Fig. 6). The mixture proportioning of RAC mixes was performed in accordance with the method proposed by Pepe et al. [16]. The prototypes were then produced in two days, with two batches per day (one SCC and one RAC per day). For each batch, a slump flow test was carried out, leading to the following diameters across the spread of concrete: – SCC prototype A01: D1 = 68 cm, D2 = 78 cm; – SCC prototype A02: D1 = 71 cm, D2 = 67 cm; – RAC prototype B01: D1 = 70 cm, D2 = 75 cm;
492
M. Pepe et al. RCA 4-8 mm
RCA 8-16 mm
Rangel et al. 2019
3000
Dry density [kg/m3]
2750 2500 2250 2000 1750 Class D
Class C
Class B
Class A 1500 0
1
2
3
4
5 6 7 8 Water absorp on [%)]
9
10
11
12
Fig. 5. Classification of RAs produced during the DeConStRACtion project.
RCA 8-16 mm
RCA 8-16 mm
NAT 2 NAT 1
NAT 1
Sand
Sand
Reference concrete
NTC 2018 (IT)
RCA 4-8 mm
NAT 2
100 %
RCA 4-8 mm
30 %
RCA 4-8 mm
20 %
RCA 8-16 mm
NAT 2
NAT 1
Sand
EN 206
Sand
DeConStRAC on
Fig. 6. Schematic mixture proportioning of DeConStRACtion concrete.
– RAC prototype B02: D1 = 67 cm, D2 = 75 cm. 10 cubes (0.10 m × 0.10 m × 0.10 m) were produced per each SCC and RAC concrete batch. The results of the uniaxial compression tests conducted on the cubes are reported in the following Table 2. Table 2. Compression test results. Batch
Rcm [MPa] 20 h
7 days
28 days
A.1 (SCC)
23.00
41.80
51.78
A.2 (SCC)
22.20
50.90
65.03
B.1 (RAC)
31.40
57.40
70.51
B.2 (RAC)
22.20
51.10
59.67
Deconstructable Concrete Structures Made of Recycled Aggregates
493
Regarding the reinforcements, three materials were introduced: – B450C steel bars for the stirrups and cage rebar: nominal diameter 8 mm; – 0.5 prestressing tendons: nominal section 93 mm2 and nominal diameter 12.5 mm, class 1670/1860; – shape memory alloy bars provided by the project partner re-fer AG: nominal diameter 18 mm (www.re-fer.eu). 2.4 Casting Procedures The production of type A prototypes comprised the following operations: – equipping of the prefabrication line (formworks, stirrups, spacers, lifting devices); – pre-tensioning of the four high-strength steel tendons (imposed pre-stress force equal to 10 tons per tendon); – casting of self-compacting concrete; – cut of the tendons and demoulding after 20 h of natural curing. The production of type B prototypes comprised (see Fig. 7): – equipping of the prefabrication line (formworks, stirrups, spacers, shear key moulds, lifting devices); – positioning of four plastic ducts (external diameter 25 mm, wall thickness 1.8 mm), kept aligned by using the four tendons of prototype A as jigs (the same production line was employed); – casting of recycled aggregate concrete; – demoulding after 20 h of natural curing.
Fig. 7. Prototypes production of RAC modular elements (type B prototypes).
2.5 Post-tensioning of RAC Modules with Shape Memory Alloy Bars The post-tensioning of prototypes B01 and B02 was carried out according to the following procedure (see Fig. 8): – positioning of 10 displacement sensors (linear variable differential transformers LVDT1-10) aimed at recording the axial strains on the prototype throughout the procedure; – start of data acquisition and preliminary hand-screwing of the end nuts;
494
M. Pepe et al.
Fig. 8. Post-tensioning with shape memory alloy bars: application of the re-bar R18 technology.
– positioning of six temperature sensors (type-K thermocouples T1-6); – power connection and introduction of the current in two steps (firstly on bars 1 and 2, and subsequently on bars 3 and 4). Each heating process lasted around 6 min and 30 s and was then followed by a cooling phase. The time series of temperature and equivalent stress data are provided in the following Fig. 9 for a representative prototype type B. Equivalent axial stresses were calculated considering an equivalent strain (estimated as the ratio between the displacement recorded by the LVDTs and the gauge length of 40 mm), multiplied by the material elastic moduli obtained from the compressive test results through established empirical equations.
Fig. 9. Post-tensioning for representative prototype B: (a) temperature vs. time data and (b) equivalent stress vs. time data in the two central modules.
Deconstructable Concrete Structures Made of Recycled Aggregates
495
3 Full-Scale Flexural Tests 3.1 Description of Test Setup A stiff steel framework and an electro-mechanical screw-jack (maximum load of 1000 kN) were used. The prototypes were subjected to a four-point bending loading. At the two ends the shallow beams were simply supported by rounded steel supports placed at a centre distance of 3.84 m (see Fig. 10 and Fig. 11). The beams were loaded at two loading points, placed 0.5 m from the middle axis and 1.42 m from the lateral supports. Direct contact between the steel supporting/loading knives and the specimens was prevented by inserting 5 mm (supports) to 10 mm (loading points) rubber sheets, aimed at preventing spurious stress localizations.
Fig. 10. Full-scale test setup of prototypes A01 and A02: front, top and bottom sensing.
Fig. 11. Full-scale test setup of prototypes B01 and B02: front, top and bottom sensing.
A set of 11 displacement transducers (LVDTs, axial potentiometers and wire potentiometers) was installed to measure vertical displacements (δ), compressive displacements (COM) and integral crack opening displacements (COD) during the tests, as
496
M. Pepe et al.
shown in Fig. 10 and Fig. 11. In addition, a load cell was used to measure the applied total load throughout the full-scale tests. The transducers were arranged as follows: five (5) wire potentiometers measured the vertical displacements at the central section (i.e., δ1 and δ2 ), under the loading knives (i.e., δ3 and δ4 ) and in the vicinity of one support (i.e., δ5 ), in order to allow the calculation of the total and net beam deflections; three (3) displacement transducers (potentiometers) applied at the bottom face over a gauge length of 1200 mm (COD1 and COD3 ) and 200 mm (COD2 ) measured integral crack opening displacements astride the maximum bending moment region; three (3) displacement transducers (potentiometers and LVDTs) applied at the top face over a gauge length of 600 mm (COM1 and COM3 ) and 200 mm (COM2 ) measured the compressive displacement at the top chord of the section. In case of prototypes B01 and B02, two (2) additional potentiometers were installed at the contact sections subjected to the maximum shear force (SLIP1 and SLIP2 in Fig. 11) to monitor possible relative sliding between the modules. All the tests were carried out at a constant crosshead displacement (stroke) rate of 15 microns per second (Fig. 12).
Fig. 12. Full-scale test setup.
3.2 Test Results The mechanical responses of the four prototypes are compared in Fig. 13 in terms of Load (P) vs average vertical displacement at the middle section. Some observations can be drawn: (i) nominally identical specimens (A01-A02 and B01-B02) showed similar mechanical responses, despite the variability in the mechanical strength of the concretes (Table 2); (ii) traditional specimens (A01-A02) exhibited a remarkable structural ductility, reaching a central deflection larger than 200 mm; (iii) innovative specimens (B01-B02) suffered an early loss of prestress force which, in turn, resulted in reduced mechanical performances, primarily attributable to the crushing of the top concrete cover at the middle contact surface. The observed reduction in strength of the “innovative system” is a consequence of the use of unbonded prestressing reinforcement. However, it is clear that “deconstructability”, which is the main feature of the beams under consideration, is certainly more pronounced in a system with post-installed and unbonded prestressing reinforcement. Therefore, a deconstructable system ought to be based on an unbonded prestressing technology, which might imply some reduction in the structural performance [17]. On
Deconstructable Concrete Structures Made of Recycled Aggregates
497
Fig. 13. Comparison between Type A (traditional, SCC) and Type B (innovative, RAC) prototypes: (a) Load P vs. average vertical displacement at the middle section; and (b) zoom in the vertical displacements range 0–150 mm and in the Load P range 0–40 kN.
the other hand, although the RAC modular system exhibited lower mechanical performances than the traditional one, based on the load capacity of the innovative system it can be safely considered for designing “ordinary” structures (e.g., residential building). In order to improve the capacity of the whole system, next step of the research should be focused on the optimization of the geometric configuration of mid-span cross section in order to exploit the capacity of the re-fer rebars.
4 Conclusion The results presented herein demonstrated that the proposed project is characterized by different aspects of originality and innovation by proposing a comprehensive approach for challenging the most relevant paradigms of the construction sector related to the strategies toward sustainable and durable structures: i. Definition of a protocol processing procedure and performance-based classification of RAs for structural concrete production; ii. Application of an adapted mixture-proportioning method for Recycled Aggregate Concrete (RAC) mixtures; iii. Optimization (geometric, mechanical, durability and sustainability) of prefabricated structural elements for modular and deconstructable buildings; iv. Application of innovative memory®-steel prestressing techniques (re-bar method) leading to an easier and faster assembly procedure of the RAC modules; This approach will allow to mitigate the main issues related to the promotion and the diffusion of recycling and re-use of CDW which is the lack of confidence of consumers and builders regarding the quality of the generated raw materials due to the heterogeneity of the secondary raw materials. Consequently, the project is characterized by a multidisciplinary approach requiring competences on materials science, structural engineering and environmental engineering as well as industrial processing and economy for a proper analysis of the perspective market. It further reinforces already existing ties between the academic research partners as well as the involved SMEs, also laying
498
M. Pepe et al.
the path for future developments in this direction. The conjunction of these aspects will certainly contribute to innovate a relevant portion of the construction section in the view of modular buildings which will result in a significant reduction of the time of building realization while at the same time optimizing the use of raw materials and extending the service life of modular structure which can be easily deconstructed and reassembled in different places. Acknowledgments. This DeConStRAtion project has received funding from the European Union’s Horizon 2020 research and innovation program under grant agreement No 873964 (METABUILDING project).
References 1. Zhang, C., Hu, M., Di Maio, F., Sprecher, B., Yang, X., Tukker, A.: An overview of the waste hierarchy framework for analyzing the circularity in construction and demolition waste management in Europe. Sci. Total Environ. 803, 149892 (2022) 2. Alexander, M., Mindess, S.: Aggregates in Concrete. CRC Press, Boca Raton (2005) 3. Sáez, P.V., Osmani, M.: A diagnosis of construction and demolition waste generation and recovery practice in the European Union. J. Cleaner Prod. 241, 118400 (2019) 4. Khoury, E., Ambrós, W., Cazacliu, B., Sampaio, C.H., Remond, S.: Heterogeneity of recycled concrete aggregates, an intrinsic variability. Constr. Build. Mater. 175, 705–713 (2018) 5. Etxeberria, M., Marí, A.R., Vázquez, E.: Recycled aggregate concrete as structural material. Mater. Struct. 40(5), 529–541 (2007) 6. Colangelo, F., Navarro, T.G., Farina, I., Petrillo, A.: Comparative LCA of concrete with recycled aggregates: a circular economy mindset in Europe. Int. J. Life Cycle Assess. 25(9), 1790–1804 (2020) 7. Zareei, S.A., Ameri, F., Bahrami, N., Shoaei, P., Musaeei, H.R., Nurian, F.: Green high strength concrete containing recycled waste ceramic aggregates and waste carpet fibers: mechanical, durability, and microstructural properties. J. Build. Eng. 26, 100914 (2019) 8. Ding, T., Xiao, J., Zhang, Q., Akbarnezhad, A.: Experimental and numerical studies on design for deconstruction concrete connections: an overview. Adv. Struct. Eng. 21(14), 2198–2214 (2018) 9. fib Bulletin 88: Sustainability of precast structures. State of the art report., Fédération internationale du béton (fib), and Precast/Prestressed Concrete Institute (PCI) (2018) 10. Venkrbec, V., Klanšek, U.: Suitability of recycled concrete aggregates from precast panel buildings deconstructed at expired lifespan for structural use. J. Clean. Prod. 247, 119593 (2020) 11. Lima, L., Trindade, E., Alencar, L., Alencar, M., Silva, L.: Sustainability in the construction industry: a systematic review of the literature. J. Clean. Prod. 289, 125730 (2021) 12. EN 1097-6:2013 Tests for mechanical and physical properties of aggregates - Part 6: Determination of particle density and water absorption 13. Pepe, M., Toledo Filho, R.D., Koenders, E.A.B.: Alternative processing procedures for recycled aggregates in structural concrete. Const. B. Mat. 69, 124–132 (2014) 14. EN 206-1:2000/A2:2005. Concrete - Part 1: Specification performance, production and conformity 15. Rangel, C.S., Toledo Filho, R.D., Amario, M., Pepe, M., de Castro Polisseni, G., de Andrade, G.P.: Generalized quality control parameter for heterogenous recycled concrete aggregates: a pilot scale case study. J. Cleaner Prod. 208, 589–601 (2019)
Deconstructable Concrete Structures Made of Recycled Aggregates
499
16. Pepe, M., Toledo Filho, R.D., Koenders, E.A., Martinelli, E.: A novel mix design methodology for Recycled Aggregate Concrete. Constr. Build. Mater. 122, 362–372 (2016) 17. Nguyen, M.C., Hong, W.K.: Analytical prediction of nonlinear behaviors of beams posttensioned by unbonded tendons considering shear deformation and tension stiffening effect. J. Asian Archit. Build. Eng. 21(3), 908–929 (2022)
Experimental Investigation of the Influence of Hemp Particles on Hydration Kinetics of Multicomponent Mineral Binder Dmytro Kosiachevskyi1,2(B) , Kamilia Abahri1 , Anne Daubresse2 , Evelyne Prat3 , and Mohend Chaouche1 1 Université Paris-Saclay, ENS Paris-Saclay, CentraleSupélec, CNRS, LMPS - Laboratoire de
Mécanique Paris-Saclay, 91190 Gif-sur-Yvette, France [email protected] 2 Centre d’Innovation Parexgroup, 38070 St Quentin Fallavier, France 3 Sika Technology AG, Tüffenwies 16, 8048 Zürich, Switzerland
Abstract. Today, the application of various bio-based low carbon mortars leads to the use of complex multicomponent binding systems. Even if the hydration processes of the binders are known, the presence of organic particles and substances can alter the hydration kinetics and hydration rate of the mineral binder. Understanding and improving the hydration mechanisms of mineral binders in the presence of organic particles has the potential to improve the mechanical properties, hygrothermal performance and durability of bio-based building materials. Currently, there is a lack of knowledge about the effect of organic particles on the hydration of mineral binders, especially in the case of modern multicomponent binder systems. This work is devoted to the study of the effect of hemp particles and their derivatives on the hydration process of mineral binders. In this regard, two parts of experimental studies were conducted. First, the kinetics of hydration of lime-based mineral binder with hydraulic additives was investigated by microcalorimetric analysis. Secondly, a semi-quantitative X-ray diffraction tracking method was used to study the difference in hydration mechanisms of the studied multicomponent binder. The results highlighted the complexity of the influence of organic compounds on the hydration mechanisms of multicomponent binders. The results of the semi-quantitative hydration tracking method demonstrated the difference in the evolution of hydrates and allowed a better understanding of the hydration process. Keywords: Hemp mortar · Hydration kinetics · Multicomponent binder · Microcalorimetric analysis · X-ray diffraction
1 Introduction Climate change is known to be one of the greatest environmental challenges of this century [1]. As for the construction sector, it is responsible for significant amounts of energy consumption, gas emissions and waste generation. In order to effectively address © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 A. J˛edrzejewska et al. (Eds.): SynerCrete 2023, RILEM Bookseries 43, pp. 500–509, 2023. https://doi.org/10.1007/978-3-031-33211-1_44
Experimental Investigation of the Influence of Hemp Particles
501
this problem, it is necessary to introduce changes at different stages of the construction process (from extraction of raw materials to demolition and recycling of materials). The introduction of local biobased solutions in construction is interesting, as it allows not only to reduce gas emissions from the transportation and production of building materials, but also to accumulate carbon in the organic part of the final material [2]. In France, hemp has been used for centuries as a raw material for the textile industry or agriculture [3]. Nowadays, hemp is also well-known as local material for building industry [4]. It can be used as a component in various building materials, or independently as non-load-bearing insulating structures. One such biobased construction application is hemp mortars. Indeed, the hygrothermal properties, economic and environmental advantages of hemp mortar make it interesting for application in the context of retrofit or new construction [5]. Many researchers studied their structural characteristics, mechanical, acoustic and thermal properties [6–9]. However, hemp mortars represent heterogenic and anisotropic behavior due to compositional complexity [10]. In addition, organic hemp particles provoke the hygroscopicity of the material, which directly affects the durability due to dimensional variations under the conditions of the environment [11]. Another important point to consider is the hydration of the binder. For better performance and durability of hemp mortar, all binders must be hydrated. But organic particles, being hygroscopic, adsorb water when hydrated, which reduces the rate of hydration of the solution [11]. Besides, hemp contains various chemical compounds (cellulose, hemicellulose, pectin, lignin and sugars), which can worsen the hydration of the mortar and affect its rheology [7]. For example, Bishop and Barron show that sucrose was able to delay the Portland cement hydration by nucleation poisoning and accelerate the ettringite formation by calcium complexation [12]. In terms of total hemp substances, the study of Delannoy et al. demonstrates correlation between the abundance of hemp shive components and the delay in setting of Portland cement [13]. Furthermore, the use of low-carbon biobased solutions for different building applications (wall, floor and roof insulation) requires the use of different multicomponent binders in order to ensure the suitable installation process and rheology. Nevertheless, there is a lack of information on the effect of organic particles and their extractives on the hydration process of multicomponent binders. Clarification of the main impact of organic compounds of hemp particles on the hydration of lime-based mineral binder with hydraulic additives is the main motivation of this work. Experimentally, the hydration kinetics was studied by different methods. First, microcalorimetric analysis of the hydration process was performed to investigate the hydration kinetics of lime-based mineral binder with hydraulic additives. Second, a semi-quantitative X-Ray Diffraction (XRD) hydration tracking method was used to investigate the difference in hydrate evolution and to better understand the hydration process.
2 Experimental Procedure The scientific interest of this paper consists in the evaluation of the influence of organic particles on the hydration kinetics of mineral binder. This paragraph presents the materials and methodology of the experiment.
502
D. Kosiachevskyi et al.
2.1 Materials Hemp mortars are composed of a mineral binder and hemp shives. The studied mineral binder was a mixture of hydrated lime, Portland cement, anhydrite and other hydraulic cements. In our study, no admixtures were used to minimize their impact on the hydration process. The general composition of the mixture is presented in Table 1. Table 1. Oxide analysis of lime-based binder used with hydraulic cements Oxide
CaO
Al2 O3
Na2 O
K2 O
Fe2 O3
MgO
SO3
Percent by mass
70.62
3.07
0.21
0.38
1.28
0.7
4.31
The hemp shives were originated from France, their particle size distribution is presented in Fig. 1. Since hemp shives are hygroscopic and absorb water from the slurry, they can affect the final results of this study. Therefore, the mineral binder mixture was hydrated with various leachates - demineralized water previously exposed to hemp particles and then filtered. The proportions for preparation of leachate were as follows H/W = 0.1 (hemp/water). In addition, different organic substances have different capacity of being solubilized. That is why, different times of exposure for leachate preparation were used: 30 min (L30m), 1 h (L1h) and 2 h (L2h). All leachates were prepared with the same hemp shives/water ratio: 0.1. The slurry hydrated with demineralized water (DW) was used as a reference for comparing the obtained results. All the hydration waters for the slurries are referenced in Table 2.
Persentage passing (%)
100 80 60 40 20 0 0.1
1 10 Particle diameter, mm
100
Fig. 1. Granulometric curve of hemp shives by image analysis
2.2 Experimental Methods Experimentally, the study consisted of two parts. The first one was represented by the study of the hydration kinetics of prepared slurries to investigate the difference and better understand the influence of hemp substances and time of exposure of leachates on
Experimental Investigation of the Influence of Hemp Particles
503
Table 2. Description of tested hydration waters used for the preparation of slurries Notation
Type of hydration water used for the slurry
Hemp/water ratio during exposure
DW
Demineralized water
0.1
L30m
Demineralized water exposed to hemp for 30 min
L1h
Demineralized water exposed to hemp for 1 h
L2h
Demineralized water exposed to hemp for 2 h
hydration process. This was accomplished by the microcalorimetric study. The second part was dedicated to detailed comprehension of the hydration process. To do this, in-situ XRD measurements were made just after mixing of the studied slurries. It enabled the tracking of the hydrates formation process during the first 24 h. The overall experimental protocol schema is presented in Fig. 2. Isothermal microcalorimetric (IMC) data were recorded with a Thermal Activity Monitor (TAM) Air isothermal Calorimeter from TA Instruments at fixed temperature of 20 °C with an accuracy of ±0.02 °C. The studied slurries were composed of 10 g of binder and 10 g demineralized water or leachate (in function of the sample). They were prepared directly in PEHD ampoules. The reference product used to account for parasitic temperature variations is composed of 10 g of silica sand and 10 g of water. All results were normalized to the mass of binder. Since the mixing was performed outside the instrument, the measurements started approximately 5 min after the powder made contact with the water. In order to study the evolution of hydrates, the in-situ X-Ray Diffraction (XRD) was conducted on the samples directly after hydration for 24 h in endogenous cure. The W/B ratio was 0.52 for all types of hydration liquids (demineralized water of leachates). XRD crystallography patterns were obtained using Cu Kα anode tube (λ = 1.54182 Å) radiation at 40 kV and 20 mA with a Bruker D8 diffractometer (Bruker, Karlsruhe, Germany). The diffractometer scanned from 5° to 60° (2θ°) in step size of 0.015° and the counting time per step was 1.2 s. In order to study the evolution of hydrate formation, the semi-quantitative method proposed by Kedziora [14] was used. Indeed, the main idea of this method is the use of the XRD patterns to measure the signal of the chosen hydrates. Using XRD patterns for all the period of scanning (24 h) we obtain the hydrates evolution profiles, that enable the analysis of the hydration kinetics. From a practical point of view, all the results of this part were represented are related to the intensity of the anhydrite of non-hydrated binder: Semi−quantitative amount of the hydrated phase at term = Intensity of hydrated phase at term ∗ 100 Intensity of anhydrite before hydration
(1)
504
D. Kosiachevskyi et al.
Hydration water preparation Reference sample
Preparation of leachates
Demineralized water
Mixing of hemp particles with demineralized water (H/W = 0.1) for 30 minutes, 1 hour and 2 hours
Filtration Slurry preparation
Mineral binder Four hydration waters: demineralized water (DW), leachate 30 min (L30), 1 hour (L1), 2 hours (L2)
Analysis of hydration kinetics Microcalorimetric analysis of hydration kinetics
Semi-quantitative XRD tracking of hydration
Fig. 2. Experimental protocol schema
Fig. 3. XRD pattern of non-hydrated binder mix: A – anhydrite (CaSO4 ); P – portlandite (Ca(OH)2 ); C – calcite (CaCO3 ); D – dicalcium silicate (Ca2 SiO4 ); T - tricalcium silicate (Ca3 SiO5 ); Ta - tricalcium aluminate (Ca3 Al2 O6 )
The XRD pattern of non-hydrated binder is proposed in Fig. 3. We note that the intensity of anhydrite before hydration was 1705 counts, this value was used later for the semi-quantitative hydration tracking.
Experimental Investigation of the Influence of Hemp Particles
505
3 Results and Discussion In this section, the results of both detailed above experimental parts are presented. First, the results of the microcalorimetric analysis of four types of slurries are presented. Then, the semi-quantitate tracking of hydration process is proposed. 3.1 Hydration Kinetics Analysis of the Binder The hydration process of the studied binder, composed of the hydrated lime and hydraulic cements, was studied by isothermal microcalorimetric analysis. Hydration reactions are most active in the first few days. In our study we were interested in the first 96 h after beginning of hydration of binder. Figure 4 demonstrates four binder hydration patterns with demineralized water (BDW) and different types of leachates exposed to hemp particles during 30 min (BL30m), 1 h (BL1h) and 2 h (BL2h). It is notable that there are two peaks of heat flows in the case of each sample. They can be related to two different hydration reactions: calcium sulfoaluminate (first) and portlandite (second). Also, we note the hydration of the Portland is slower in the case of binder with leachates compared to the binder with demineralized water. This impact of organic extractives on OPC is well known and was demonstrated by Bishop and Barron and Delannoy et al. [12, 13]. Indeed, hemp shive extractives are represented by various reducing or non-reducing sugars, that can complex the calcium ions and thus worsen the hydration reaction. At the same time, the hydration process of aluminates is known to be accelerated by polysaccharide polymers as it was noted in the case of the three binders hydrated with leachates (BL30m, BL1h and BL2h). Different polysaccharide polymers, such as pectin, can provide nucleation sites for hydration reactions and thus accelerate the hydration kinetics [15–17]. Figure 5 shows the curves of cumulative heat release for four types of binders. Here, we note that the total heat release in the case of binder with leachates is lower than with demineralized water. It demonstrates that the amount of hydrated binder for samples with leachates is lower, which indicates a general decrease in the rate of binder hydration. 3.2 Semi-quantitative Methods of Hydration Tracking In the second part of this study, the hydration process and the hydrates formation were investigated by the semi-quantitative X-ray diffraction analysis. It allowed to track the evolution of the amount of hydrates. As explained earlier, all results were related to the intensity of anhydrite in non-hydrated binder. Figure 6 represents the evolution of the different compounds of the binder hydrated with demineralized water (A) or leachates exposed during 30 min (B), 1 h (C) and 2 h (D) as a function of time over 24 h under endogenous curing conditions.
506
D. Kosiachevskyi et al.
Cumulated heat (J/g)
Fig. 4. Heat flow normalized per mass of binder for the first 96 h of hydration for the four types of binders: hydrated with demineralized water (BDW), leachate exposed to hemp particles during 30 min (BL30m), 1 h (BL1h) and 2 h (BL2h) 150 100 BDW BL30m BL1h BL2h
50 0 0
24
48
Time (h)
72
96
Fig. 5. Cumulative heat release per mass of binder for the first 96 h of hydration for the four types of binders: hydrated with demineralized water (BDW), leachate exposed to hemp particles during 30 min (BL30m), 1 h (BL1h) and 2 h (BL2h)
It can be noted that in the case of demineralized water, the consumption of alite and belite is much more active in the first three hours, which indicates the negative impact of organic compounds on the dissolution of alite and belite and coincides with other works on this subject [12, 13]. At the same time, the consumption of anhydrite and ye’elimite is much more active in the first two hours in the case of leachates. It causes the well-known acceleration of aluminate hydration and can be explained by calcium complexation by biopolymers and increasing of the quantity of nucleation sites [15].
Experimental Investigation of the Influence of Hemp Particles
250 200
60
150 40
100 20
50 0
0 4
80
40
100 20
50 0
0 0
4
200 150 40 100 50 0
0 0
4
8 12 16 20 24 Time (hours)
(C)
100
Ettringite Ye'elimite Alite+Belite Anhydrite Portlandite
Intensity of hydrated phases normlized to anhydrate
250
60
20
8 12 16 20 24 Time (hours)
(B) 300 Intensity of portlandite normlized to anhydrate
Intensity of hydrated phases normlized to anhydrate
80
200 150
8 12 16 20 24 Time (hours)
Ettringite Ye'elimite Alite+Belite Anhydrite Portlandite
250
60
(A) 100
300
Ettringite Ye'elimite Alite+Belite Anhydrite Portlandite
80
300 250 200
60 150
40 100
20
50 0
0 0
4
Intensity of portlandite normlized to anhydrate
0
Intensity of hydrated phases normlized to anhydrate
80
100
Intensity of portlandite normlized to anhydrate
300
Ettringite Ye'elimite Alite+Belite Anhydrite Portlandite
Intensity of portlandite normlized to anhydrate
Intensity of hydrated phases normlized to anhydrate
100
507
8 12 16 20 24 Time (hours)
(D)
Fig. 6. Semi-quantitate X-ray diffraction tracking of anhydrous and hydrated compounds of limebased mineral binder in function of time over 24 h under endogenous curing conditions hydrated with demineralized water (a), leachates exposed during 30 min (b), 1 h (c) and 2 h (d)
4 Conclusions In this work, the influence of the hemp particles and their extractives on the kinetics of hydration of multicomponent mineral binder, composed of hydrated lime and hydraulic cements, was investigated.
508
D. Kosiachevskyi et al.
Results revealed the complexity of the hydration process of mineral binder in presence of organic compounds. First, the results of microcalorimetric study demonstrated the presence of two different peaks of heat flow that are related to hydration reactions of calcium sulfoaluminate (first) and portlandite (second). The cumulative heat release showed the negative impact of leachates on total amount of hydrated binder. The semi-quantitative X-ray diffraction allowed to track finely the evolution of different compounds of mineral binder over first 24 h. The obtained results demonstrated the acceleration of the consumption of anhydrite and ye’elimite that can be explained by calcium complexation by biopolymers and increasing of the quantity of nucleation sites. Also, the consumption of alite and belite is less active for the samples hydrated with leachates, that is due to presence of organic particles. These two reactions are in coincidence with the literature but their synergy should be studied in further works. For better understanding of the impact, it is important to analyse the composition of filtrates and study the effect of different organic compounds on the hydration kinetics of the binder mixture and binders separately. Acknowledgments. The authors acknowledge ParexGroup SA for financing this project.
Conflict of Interests. None.
References 1. Zhao, Q., Yu, P., Mahendran, R., Huang, W., Gao, Y., Yang, Z., et al.: Global climate change and human health: pathways and possible solutions. Eco-environment Health 1, 53–62 (2022). https://doi.org/10.1016/j.eehl.2022.04.004 2. Jami, T., Karade, S.R., Singh, L.P.: A review of the properties of hemp concrete for green building applications. J. Clean. Prod. 239, 117852 (2019). https://doi.org/10.1016/j.jclepro. 2019.117852 3. Vandepitte, K., Vasile, S., Vermeire, S., Vanderhoeven, M., Van der Borght, W., Latré, J., et al.: Hemp (Cannabis sativa L.) for high-value textile applications: the effective long fiber yield and quality of different hemp varieties, processed using industrial flax equipment. Ind. Crops Prod. 158, 112969 (2020). https://doi.org/10.1016/j.indcrop.2020.112969 4. Kosiachevskyi, D., Abahri, K., Daubresse, A., Prat, E., Chaouche, M.: Assessment of the hygrothermal, microstructural and chemical evolution of a hemp-based cementitious mortar under ETICS total weathering aging protocol. Constr. Build. Mater. 314, 125471 (2022). https://doi.org/10.1016/j.conbuildmat.2021.125471 5. Bennai, F., Issaadi, N., Abahri, K., Belarbi, R., Tahakourt, A.: Experimental characterization of thermal and hygric properties of hemp concrete with consideration of the material age evolution. Heat Mass Transf. 54(4), 1189–1197 (2017). https://doi.org/10.1007/s00231-0172221-2 6. Gourlay, E., Glé, P., Marceau, S., Foy, C., Moscardelli, S.: Effect of water content on the acoustical and thermal properties of hemp concretes. Constr. Build. Mater. 139, 513–523 (2017). https://doi.org/10.1016/j.conbuildmat.2016.11.018 7. Arufe, S., Hellouin de Menibus, A., Leblanc, N., Lenormand, H.: Effect of retting on hemp shiv physicochemical properties. Ind. Crops Prod. 171, 113911 (2021). https://doi.org/10. 1016/j.indcrop.2021.113911
Experimental Investigation of the Influence of Hemp Particles
509
8. Balˇci¯unas, G., Pundien˙e, I., Lek¯unait˙e-Lukoši¯un˙e, L., V˙ejelis, S., Korjakins, A.: Impact of hemp shives aggregate mineralization on physical–mechanical properties and structure of composite with cementitious binding material. Ind. Crops Prod. 77, 724–734 (2015). https:// doi.org/10.1016/j.indcrop.2015.09.011 9. Sinka, M., Obuka, V., Bajare, D., Jakovics, A.: Durability and hygrothermal performance of bio-based materials in northern European climate (2019). https://doi.org/10.26168/icbbm2 019.53 10. Bennai, F., El Hachem, C., Abahri, K., Belarbi, R.: Influence of hydric solicitations on the morphological behavior of hemp concrete. RILEM Tech. Lett. 4, 16–21 (2019). https://doi. org/10.21809/rilemtechlett.2019.80 11. Kosiachevskyi, D., El Hachem, C., Abahri, K., Bennacer, R., Chaouche, M.: Biomaterials heterogeneous displacement, strain and swelling under hydric sorption/desorption: 2D image correlation on spruce wood. Constr. Build. Mater. 286, 122997 (2021). https://doi.org/10. 1016/j.conbuildmat.2021.122997 12. Bishop, M., Barron, A.R.: Cement hydration inhibition with Sucrose, Tartaric acid, and Lignosulfonate: analytical and spectroscopic study. Ind. Eng. Chem. Res. 45, 7042–7049 (2006). https://doi.org/10.1021/ie060806t 13. Delannoy, G., Marceau, S., Glé, P., Gourlay, E., Guéguen-Minerbe, M., Diafi, D., et al.: Impact of hemp shiv extractives on hydration of Portland cement. Constr. Build. Mater. 244, 118300 (2020). https://doi.org/10.1016/j.conbuildmat.2020.118300 14. Kedziora, C.: Propriétés d’usage et mécanismes d’hydratation du système ternaire [Ciment Alumineux – Sulfate de Calcium – Laitier de Haut Fourneau] à haute teneur en sulfate de calcium: De l’approche expérimentale à la modélisation (2015) 15. Engbert, A., Plank, J.: Templating effect of alginate and related biopolymers as hydration accelerators for calcium alumina cement - a mechanistic study. Mater. Des. 195, 109054 (2020). https://doi.org/10.1016/j.matdes.2020.109054 16. Engbert, A., Gruber, S., Plank, J.: The effect of alginates on the hydration of calcium aluminate cement. Carbohyd. Polym. 236, 116038 (2020). https://doi.org/10.1016/j.carbpol.2020. 116038 17. Engbert, A., Plank, J.: Identification of specific structural motifs in biopolymers that effectively accelerate Calcium Alumina cement. Ind. Eng. Chem. Res. (2020). https://doi.org/10. 1021/acs.iecr.0c01620
Development of Concrete Mixtures Based Entirely on Construction and Demolition Waste and Assessment of Parameters Influencing the Compressive Strength Gurkan Yildirim1,2(B)
, Emircan Ozcelikci2,3 and Ashraf Ashour1
, Musab Alhawat1
,
1 Department of Civil and Structural Engineering, University of Bradford, Bradford, UK
[email protected]
2 Department of Civil Engineering, Hacettepe University, Ankara, Turkey 3 Institute of Science, Hacettepe University, Beytepe, Ankara, Turkey
Abstract. Demolition and reconstruction of degrading structures alongside with the repetitive repair, maintenance, and renovation applications create significant amounts of construction and demolition waste (CDW), which needs proper tackling. The main emphasis of this study has therefore been placed on the development of concrete mixtures with components (i.e., aggregates and binder) coming entirely from CDW. As the binding phase, powdered CDW-based masonry units, concrete and glass were used collectively as precursors to obtain geopolymer binders, which were then incorporated with CDW-based fine and coarse concrete aggregates. Together with the entirely CDW-based concretes, designs were also proposed for companion mixtures with mainstream precursors (e.g., fly ash and slag) occupying some part of the CDW-based precursor combination. Sodium hydroxide (NaOH), sodium silicate (Na2 SiO3 ) and calcium hydroxide (Ca[OH]2 ) were used at various concentrations and combinations as the alkaline activators. Several factors that have impact on the compressive strength results of concrete mixtures, such as mainstream precursor replacement rate, alkaline molar concentrations, aggregate-to-binder ratios and curing conditions, were considered and these were also backed by the microstructural analyses. Our results showed that through proper optimization of the design factors, it is possible to manufacture concrete mixtures entirely out of CDW with compressive strength results able to reach up to 40 MPa under ambient curing. Current research is believed to be very likely to promote more innovative and up-to-date techniques to upcycle CDW, which are mostly downcycled through basic practices of road base/sub-base filling, encouraging further research and increasing the awareness in CDW issue. Keywords: Construction and demolition waste · Geopolymer concrete · Recycled concrete aggregate · Compressive strength
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 A. J˛edrzejewska et al. (Eds.): SynerCrete 2023, RILEM Bookseries 43, pp. 510–520, 2023. https://doi.org/10.1007/978-3-031-33211-1_45
Development of Concrete Mixtures Based Entirely on Construction
511
1 Introduction Urbanization activities such as construction, maintenance, repair, renovation, and demolition are continuously increasing due to growing population. As a result of these activities, tremendous amounts of construction and demolition waste (CDW) are generated all around the world. Correspondingly, CDW occupies the largest portion of all solid waste stream [1] with 10 billion tons of annual generation worldwide [2]. The European Action Plan’s designation of the construction and demolition industry, as one of the five priority sectors within the framework of the circular economy model, makes it quite clear that CDW must be recycled in the most effective way by avoiding being dumped in clean landfills [3]. Although the research on CDW management and reprocessing are fairly active in Europe, recent sources indicate that up to 95% of waste concrete, which accounts for the largest portion of CDW, is downcycled as sub-base fillers [4]. However, to achieve a better position of CDW in the value chain and help close the loop in the construction sector for the purposes of true sustainability, the most efficient recycling/upcycling scenarios need to be identified instead of downcycling/disposing activities. The abovementioned urbanization activities also require the production of high amounts of Portland cement (PC), which is the main binding component of conventional concrete. Despite being an indispensable material for the construction industry for centuries thanks to its outstanding strength and durability characteristics, PC is a troublesome material particularly for the environment. To put it in numbers, it has been reported that 1 ton of PC production requires energy in the range of 3.2–6.3 GJ and releases approximately 1 ton of CO2 to the atmosphere [5, 6]. Research in recent years have therefore concentrated on the development of alternative and more eco-friendly binders than PC, taking the issue of global warming and long-term sustainability into consideration. Geopolymers synthesized by activating aluminosilicate-based materials with alkali hydroxides/silicates are one of such examples that come to the fore recently [7]. Although geopolymers have been developed mostly with industrial by-products/wastes (e.g., fly ash, ground blast furnace slag, silica fume) to date, the limited availability of these materials and their high demands by the PC/concrete industry made it necessary to seek alternative sources for geopolymers. CDW stands out as a suitable resource for geopolymer manufacture due to their aluminosilicate-rich nature and widespread availability. In recent years, studies on the development of geopolymers with CDW have gained momentum, and CDW-based geopolymers reaching compressive strength level of 80 MPa have been successfully developed [8, 9]. Thus, it can be stated that upcycling CDW into novel, eco-friendly and high-performance building materials that can be an alternative to PC via geopolymerization is possible and promising especially for ensuring eco-friendliness and sustainability. Taking the issues related to CDW generation and PC manufacture, this research work concentrated on the development of green geopolymer concretes with entirely CDW-based materials and the examination of the parameters affecting the compressive strength. CDW-based components such as hollow brick (HB), red clay brick (RCB), roof tile (RT), concrete (C) and glass (G) were used in the mixed for as the binder phase of geopolymer concretes, while waste concrete was used as different-size recycled concrete aggregate (RCA). Industrial by-products such as ground granulated blast furnace
512
G. Yildirim et al.
slag (GGBFS) and Class-F fly ash (FA) were also substituted partially in some mixtures to examine their interaction with CDW-based materials. Sodium hydroxide (NaOH), sodium silicate (Na2 SiO3 ) and calcium hydroxide (Ca[OH]2 ) were used in various concentrations as alkali activators. Geopolymer concretes were designed to assess the effects of various parameters such as various combinations of CDW-based precursors, alkali activator concentrations, GGBFS and FA substitutions, and binder/RCA ratios. The mixtures were subjected to ambient and water curing for 7 and 28 days before being tested for compressive strength measurement. Scanning electron microscopy (SEM), energy-dispersive X-ray spectroscopy (EDX) and mercury intrusion porosimetry (MIP) analyses were also performed on the selected mixtures for an in-depth microstructural characterization.
2 Experimental Program 2.1 Materials In this study, CDW-based materials such as hollow brick (HB), red clay brick (RCB), roof tile (RT), concrete (C) and glass (G) were used collectively as the precursor phase in the geopolymer concrete production. These materials, which were of unknown origin, were collected from a demolished residential building and then subjected to a two-step crushing-grinding process separately to obtain adequate fineness for geopolymerization. Images of the CDW-based materials were presented in Fig. 1.
Fig. 1. Images of the CDW-based materials
The chemical compositions of the CDW-based precursors were found by X-ray fluorescence (XRF) analysis and the results were given in Table 1. According to the XRF analysis, clayey components (i.e., HB, RCB, RT) had a similar chemical composition in terms of SiO2 (39.7–42.6%), Al2 O3 (13.8–17.3%) and CaO (7.69–11.6%) content. While C was found to have the highest CaO content with 31.3%, G was found to have the highest SiO2 content with 66.5% and the lowest Al2 O3 content with 0.93% among other CDW-based precursors. Particle size distribution plots of CDW-based precursors subjected to separate crushing and ball mill grinding for an hour were presented in Fig. 2. The particle size distribution of the CDW-based precursors was determined by using
Development of Concrete Mixtures Based Entirely on Construction
513
dry laser granulometry method (Malvern Mastersizer Scirocco 2000 assembled with a hopper instrument). While the clayey CDW-based components exhibited the smallest particle size distribution results, G was found to be the coarsest among all CDW-based precursors. Table 1. Chemical composition of the precursors (units are in %) Materials
SiO2
Al2 O3
Fe2 O3
CaO 11.6
HB
39.7
13.8
11.8
RCB
41.7
17.3
11.3
RT
42.6
15.0
11.6
C
31.6
4.76
G
66.5
0.93
FA
60.1
21.4
7.41
GGBFS
32.1
11.2
0.62
MgO
SO3
TiO2
K2 O
Na2 O
LOIa
6.45
3.40
1.65
1.55
1.45
7.80
6.49
1.41
1.57
2.66
1.15
7.96
10.7
6.26
0.71
1.82
1.60
1.60
7.49
3.50
31.3
5.12
0.92
0.24
0.71
0.25
10.0
3.93
0.24
0.06
0.20
1.82
0.22
0.94
2.91
0.99
2.61
5.64
1.21
1.07
0.83
0.31
9.09
7.69
0.99 36.1
0.45 13.6
20.9 4.15
a Loss on ignition
Fig. 2. Particle size distributions of the CDW-based precursors
As the aggregate phase of the geopolymer concretes, RCA obtained by crushing the CDW-based concrete and sieving it in various sizes were used. The gradation curve of the RCA was presented in Fig. 3. Fine and coarse RCA were blended (50%, by weight each) to obtain mixed RCA. Sodium hydroxide (NaOH), sodium silicate (Na2 SiO3 ) and calcium hydroxide (Ca[OH]2 ) were used as the alkaline activators in the production of geopolymer concretes. Class-F fly ash (FA) and ground granulated blast furnace slag (GGBFS) were also used in some mixtures to observe their effects on the mechanical performance when combined with the CDW-based precursors.
514
G. Yildirim et al.
Fig. 3. Particle size distribution of the CDW-based RCA
2.2 Mixture Design As a result of preliminary studies, 9 mixture designs were made considering the precursor combinations, alkaline solution concentrations and aggregate/binder ratios. Proportions of the mixtures were given in Table 2. In the first three mixtures (GC1, GC2 and GC3), the combination of CDW-based materials in the precursor phase were investigated. In the next three mixtures (GC4, GC5 and GC6), effect of different alkaline activator combinations was examined. In GC7 and GC8 coded mixtures, the effect of fly ash and slag addition at a rate of 20% was investigated, respectively. In the last mixture (GC9), the aggregate/binder ratio was increased to 2 and the effect of increased amount of aggregates was examined. All mixtures were produced according to the following steps: (i) NaOH solution was prepared with the mixing of NaOH pellets and water and left in a room to cool down for 1 day, (ii) Precursors and RCA were mixed for 60 s at 100 rpm and NaOH solution was slowly added to the mixture in 60 s, (iii) during mixing, Na2 SiO3 solution was added to the mixture in 60 s and finally Ca(OH)2 was added to the mixture in 60 s and mixing was kept for 180 s at 150 rpm. 2.3 Curing and Testing For each mixture, three cubic specimens with 100 × 100 × 100 mm dimensions were produced for each testing age (i.e., 7- and 28-day) of compressive strength. After casting, the samples were kept in their mould with their surfaces covered for 24 h under ambient conditions set at average temperature of 23 ± 2 °C and a relative humidity of 50 ± 5%, and then moved into plastic bags having controlled environment set at average temperature of 23 ± 2 °C and relative humidity of 95 ± 5% until the testing date. Additionally, three different specimens for each mixture also left to water curing until the testing age. At the end of 7 and 28 days, the specimens were subjected to compressive strength test. Compressive strength tests were performed at a loading rate of 3.0 kN/s by using a 100ton capacity testing device. The compressive strength result was calculated by averaging the results acquired from three separate specimens tested. In addition, scanning electron
Development of Concrete Mixtures Based Entirely on Construction
515
Table 2. Mixture proportions Materials Precursorsa
Alkaline solutionb
GC1
GC2
GC3
GC4
GC5
GC6
GC7
GC8
GC9
HB
20
23.3
26.7
26.7
26.7
26.7
21.3
21.3
21.3
RCB
20
23.3
26.7
26.7
26.7
26.7
21.3
21.3
21.3
RT
20
23.3
26.7
26.7
26.7
26.7
21.3
21.3
21.3
G
20
15
10
10
10
10
8
8
8
C
20
15
10
10
10
10
8
8
8
FA
–
–
–
–
–
–
20
–
–
GGBFS
–
–
–
–
–
–
–
20
20
RCAb
100
100
100
100
100
100
100
100
200
Water
35
35
35
35
35
35
35
35
35
NaOH
11.2
11.2
11.2
11.2
11.2
16.8
11.2
11.2
11.2
Na2 SiO3
22.4
22.4
22.4
22.4
11.2
–
22.4
22.4
22.4
–
5
5
5
Ca(OH)2 – – – 5 5 a Units are in %, b By the total weight of the precursor (%)
microscopy (SEM), Energy-dispersive X-ray spectroscopy (EDX) and mercury intrusion porosimetry (MIP) analyses were performed on the samples obtained from the selected mixtures.
3 Results and Discussion 3.1 Compressive Strength Compressive strength results of the CDW-based geopolymer concrete after 7 and 28 days of curing under ambient and water curing conditions were given in Fig. 4. 7- and 28-day compressive strengths of ambient-cured GC1 mixture were 9.2 and 24.7 MPa, respectively. The compressive strengths of GC2 mixture, which was produced by increasing the total ratio of clayey CDW-based precursors from 60 to 70% compared to the GC1 mixture, increased to 11.7 and 26.9 MPa after 7 and 28 days of ambient curing, respectively. For the same ambient curing ages, the compressive strengths of GC3 mixture, which included clayey precursors by 80%, were found to be 12.5 and 28.3 MPa, respectively. Considering the results of GC1, GC2 and GC3 coded mixtures, the contributions of increased amounts of clayey precursors on compressive strength may be attributed to higher aluminosilicate content and finer particle size distribution of clayey precursors. During geopolymerization, the aluminosilicate component in the precursor material is dissolved by the alkali solution and releases SiO4 and AlO4 tetrahedral structures to the system, and subsequently, these components are linked and re-precipitated, resulting in the formation of a 3D-cross-linked amorphous geopolymeric gel [8]. Therefore, the high aluminosilicate content and rapid dissolution rate are considered very favourable in determining the performance of final products [10]. On the other hand, the finer grain
516
G. Yildirim et al.
size of the precursors provides a larger surface area to be in contact with the alkaline solution during the reaction and boosts the geopolymerization efficiency, resulting in an improvement in the mechanical performance of the final product [11].
Fig. 4. Compressive strength results of the mixtures
In the next three mixtures where the effect of alkali activator combinations was investigated, the 7-day compressive strength results of the ambient-cured GC4, GC5 and GC6 were found to be 14.6, 11.5 and 6.1 MPa, while the 28-day compressive strength results were 30.2, 27.1 and 20.2 MPa, respectively. It was observed that the addition of 5% Ca(OH)2 increased the compressive strength by 16.8%, which can be explained by the fact that the addition of Ca(OH)2 both enhances the degree of geopolymerization by increasing the alkalinity of the system and promotes the formation of CASH gel, which can contribute to compressive strength as an extra geopolymeric gel, as a result of the release of Ca+2 ions to the system [12]. On the other hand, for GC5 mixture with the NaOH:Na2 SiO3 ratio of 1, a decrease in the compressive strength by 21.2% after 7 days and by 10.3% after 28 days were observed compared to GC4 mixture with the NaOH:Na2 SiO3 ratio of 2. The drop in compressive strength with the reduced Na2 SiO3 ratio can be attributed decrease in the amount of soluble silicates released to the system, decreasing the degree of geopolymerization and therefore the compressive strength of the final product [13]. This was noticeable especially in the case of GC6 mixture, where the compressive strength decreased dramatically. GC6 mixture, in which only NaOH was used as the alkali activator and the NaOH ratio was increased to 16.8% to ensure sufficient alkalinity, showed the lowest strength achievement among all mixtures. NaOH is insufficient on its own for CDW-based materials to participate in geopolymerization at full capacity under ambient curing, especially when these materials are not calcined since NaOH is unable to supply Si and Ca ions to the system for effective geopolymerization, unlike Na2 SiO3 and Ca(OH)2 . In GC7 and GC8 mixtures where the partial substitution (20%, by weight) of FA and GGBFS was made with the CDW-based precursors, GC7 mixture ambient-cured
Development of Concrete Mixtures Based Entirely on Construction
517
for 7 and 28 days reached 15.3 and 32.6 MPa of compressive strength, while the same results for GC8 mixture were 23.8 and 40.1 MPa, respectively. It was observed that the GC7 and GC8 mixtures achieved 7.9% and 32.8% higher 28-day compressive strength results than 100% CDW-based mixture (GC4), respectively. Such a high increase in compressive strength with the addition of GGBFS can be due to that GGBFS acts as an extra calcium source and forms CSH/CASH gel structures that contribute to the final strength by releasing Ca2+ , Si4+ and Al3+ ions into the matrix [14]. In the case of GC9 where the RCA/precursor ratio was increased from 100% to 200%, a drop in compressive strength by 29.2% was observed after 28 days of ambient curing compared to its counterpart mixture (GC8). In general, this decrease can be attributed to the lower strength of RCA compared to clean aggregates as well as the deterioration of the workability of the mixture when utilized at high rates due to its high water absorption capacity, which eventually lowers the compressive strength. Another noteworthy point to mention was that drops in compressive strength in the range of 3.24–16.8% were observed in all mixtures cured in water except 20% GGBFSsubstituted GC8 and GC9 mixtures. This behaviour of GC8 and GC9 mixtures contrary to the overall trend can be explained by the fact that GGBFS hydrates at an early age by showing partial cementitious properties thanks to water curing. On the other hand, the observed compressive strength drops in the rest of mixtures can be attributed to the decrease in geopolymerization efficiency as a result of a decrease in alkalinity in the medium due to water penetration into the samples through the capillary pores. 3.2 SEM/EDX Analysis SEM/EDX analysis of the selected GC8 mixture, which exhibited the highest compressive strength result in the study, was presented in Fig. 5. The SEM images revealed that various types of NASH and CASH type gels were homogeneously and compactly dispersed in the matrix, the defects such as voids and microcracks were also observed in some regions. In the regional EDX analysis, the presence of Ca and Si elements in similar percentages confirms the formation of hybrid geopolymeric gel products [15]. Furthermore, the image acquired from the interfacial transition zone (ITZ) showed a relatively compact and dense microstructure along with small amounts of voids/cracks between the geopolymer paste and RCA. Although heterogeneity was seen throughout the aggregate, ITZ region seemed to have a strong interlocking with the matrix. In addition, the EDX analysis conducted on the marked point in the ITZ region showed that while Si and Al elements were dominant, Ca was quite low compared to the general EDX result of the matrix. This can be due to the accumulation of NASH type gel at the point analysed.
518
G. Yildirim et al.
Fig. 5. SEM/EDX analysis of the selected GC8 mixture
3.3 MIP Analysis MIP analysis results of the selected GC4, GC6 and GC8 mixtures were presented in Fig. 6. Among these three mixtures, the GC4 mixture had the largest cumulative volume intrusion value, which indicates the total porosity of the mixtures, whereas the GC8 mixture had the lowest. The total porosity results were found to be quite parallel with the compressive strength results of the mixtures. This observed correlation between compressive strength and total porosity values has also been documented in the literature [16]. Strength characteristics of the specimens are largely dependent on the pores greater than 50 nm, whereas pores smaller than 50 nm generally affect the creep and shrinkage properties [17]. As shown by the compressive strength results, having less pores greater than 50 nm may be regarded as one of the influencing criteria for achieving a more compact and denser microstructure. Likewise, Fig. 6 clearly shows that GC8 mixture, which exhibited the highest compressive strength results, showed the lowest total porosity and less number of pores with diameters larger than 50 nm among other mixtures.
Development of Concrete Mixtures Based Entirely on Construction
519
Fig. 6. MIP test results of the selected geopolymer concrete mixtures
4 Conclusions Development of entirely CDW-based geopolymer concretes and assessment of various parameters (i.e., precursor combinations, alkaline solution concentrations, FA and GGBFS substitution, and binder/RCA ratio) influencing the compressive strength were aimed in this study. Based on the experimental findings, the obtained results can be summarized as follows: • Clayey CDW-based precursors (i.e., HB, RCB and RT) exhibited better geopolymerization performance than C and glass G due to their higher aluminosilicate content and finer particle size. • Concurrent presence of Na2 SiO3 , which provides soluble silicates into the matrix and Ca(OH2 ), which triggers the formation of extra CASH gels by releasing Ca+2 ions into the matrix, with NaOH seems more effective for geopolymerization than the single presence of NaOH in the mixtures. • Addition of FA and GGBFS was beneficial in terms of compressive strength thanks to yielding formation of extra CASH gel structures. • Due to its low quality and high water absorption capacity, the increase in the RCA ratio had a negative effect on the compressive strength. • The decrease in alkalinity in the matrix due to water penetration during water curing had a negative effect on the compressive strength in general. • SEM/EDX analysis of the geopolymer matrix revealed the simultaneous presence of NASH and CASH gel structures and very compact and dense microstructure. MIP analysis shown clear correlation between compressive strength and total porosity. • Considering all the parameters affecting the compressive strength, green geopolymer concretes based on CDW-based materials and 20% GGBFS substitution reached 40.1 MPa of compressive strength after 28 days of ambient curing. This result clearly demonstrated that, as being more eco-friendly, CDW-based geopolymer concretes can be utilized for structural purposes given the promising compressive strength results.
520
G. Yildirim et al.
Acknowledgements.
This project has received funding from the European Union’s Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement No 894100.
References 1. Eurostat Waste statistics (2015). http://ec.europa.eu/eurostat/statistics-explained/index.php/ Waste_statistics 2. Wu, H., Zuo, J., Zillante, G., Wang, J., Yuan, H.: Construction and demolition waste research: a bibliometric analysis. Archit. Sci. Rev. 62(4), 354–365 (2019) 3. Eur-lex Homepage (2015). http://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:520 15DC0614 4. Zhang, C., Hu, M., Di Maio, F., Sprecher, B., Yang, X., Tukker, A.: An overview of the waste hierarchy framework for analyzing the circularity in construction and demolition waste management in Europe. Sci. Total Environ. 803, 149892 (2022) 5. Rahman, A., Rasul, M.G., Khan, M.M.K., Sharma, S.: Recent development on the uses of alternative fuels in cement manufacturing process. Fuel 145, 84–99 (2015) 6. Akduman, S, ¸ et al.: Experimental investigations on the structural behaviour of reinforced geopolymer beams produced from recycled construction materials. J. Build. Eng. 41, 102776 (2021) 7. Provis, J.L., Lukey, G.C., van Deventer, J.S.: Do geopolymers actually contain nanocrystalline zeolites? A reexamination of existing results. Chem. Mater. 17(12), 3075–3085 (2005) 8. Yıldırım, G., et al.: Development of alkali-activated binders from recycled mixed masonryoriginated waste. J. Build. Eng. 33, 101690 (2021) 9. Ulugöl, H., et al.: Mechanical and microstructural characterization of geopolymers from assorted construction and demolition waste-based masonry and glass. J. Clean. Prod. 280, 124358 (2021) 10. Zakka, W.P., Lim, N.H.A.S., Khun, M.C.: A scientometric review of geopolymer concrete. J. Clean. Prod. 280, 124353 (2021) 11. Komnitsas, K., Zaharaki, D., Vlachou, A., Bartzas, G., Galetakis, M.: Effect of synthesis parameters on the quality of construction and demolition wastes (CDW) geopolymers. Adv. Powder Technol. 26(2), 368–376 (2015) 12. Cheah, C.B., Part, W.K., Ramli, M.: The hybridizations of coal fly ash and wood ash for the fabrication of low alkalinity geopolymer load bearing block cured at ambient temperature. Constr. Build. Mater. 88, 41–55 (2015) 13. Lee, W.K.W., Van Deventer, J.S.J.: Structural reorganisation of class F fly ash in alkaline silicate solutions. Colloids Surf. A 211(1), 49–66 (2002) 14. Alhawat, M., Ashour, A., Yildirim, G., Aldemir, A., Sahmaran, M.: Properties of geopolymers sourced from construction and demolition waste: a review. J. Build. Eng. 50, 104104 (2022) 15. Temuujin, J.V., Van Riessen, A., Williams, R.: Influence of calcium compounds on the mechanical properties of fly ash geopolymer pastes. J. Hazard. Mater. 167(1–3), 82–88 (2009) 16. Kusbiantoro, A., Ibrahim, M.S., Muthusamy, K., Alias, A.: Development of sucrose and citric acid as the natural based admixture for fly ash based geopolymer. Procedia Environ. Sci. 17, 596–602 (2013) 17. Mehta, P.K., Monteiro, P.J.M.: Concrete Structure, Properties and Materials, vol. 2. Prentice Hall, New Jersey (1993)
Utilisation of COVID-19 Waste PPE in the Applications of Structural Concrete Shannon Kilmartin-Lynch1(B) , Rajeev Roychand1 , Mohammad Saberian1 Jie Li1 , and Fangjie Chen2
,
1 RMIT University, Melbourne, VIC, Australia [email protected] 2 ARUP, Melbourne, VIC, Australia
Abstract. There has been an increase in the use of single-use plastic-based personal protective equipment (PPE) since the commencement of the COVID-19 epidemic in late 2019. As a result, clinical waste generation has increased significantly this as well as various environmental strains from excess waste ending up in landfill. Various recycling solutions are needed to reduce the environmental impact of disposal and incineration. This experimental study aims to examine the utilisation of single-use waste PPE generated from the coronavirus pandemic in structural concrete to aid in scaling back the volume of single-use waste ending up in a landfill. Single-use nitrile gloves, isolation gowns and face masks were separately added to aggregates at varying percentages of the volume of concrete. For the purpose of determining the effects of varying concentrations and materials on the mechanical properties and quality of concrete as well as the specific materials bonding performance within the cement matrix, concrete samples were subjected to compression strength and microstructural analysis. Results demonstrate steady trendy development across compressive strength results, with increases of 17%, 20% and 15%, respectively, across varying applications of waste PPE. At the same time, the results of the SEM-EDS analysis present an excellent bond formation amongst the materials utilised and the cement matrix. To the best of the authors’ knowledge there is a lack of existing studies that examine the feasibility of incorporating waste PPE into civil and construction applications, therefore, this study aims to highlight the novelty surrounding the topic for further research. Keywords: COVID-19 · Waste Management · Plastics · Concrete · Polypropylene · Polyethylene · Nitrile
1 Introduction The COVID-19 epidemic has swiftly placed the world on hold since it originally became public knowledge in 2019. Since then, the pandemic has affected economies, health systems and livelihoods globally [1]. Following the pandemic, the World Health Organization (WHO) issued guidelines on the proper use of personal protective equipment (PPE) [2]. This resulted in a sharp incline in the use of PPE, with reports showing approximately 54,000 tonnes of waste PPE was being produced per day on a global scale [3]. A majority © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 A. J˛edrzejewska et al. (Eds.): SynerCrete 2023, RILEM Bookseries 43, pp. 521–527, 2023. https://doi.org/10.1007/978-3-031-33211-1_46
522
S. Kilmartin-Lynch et al.
of personal protective equipment that ends up in landfills or littering streets is single-use face masks, gloves, and isolation gowns [4, 5]. Moreover, waste PPE continues to be observed in various marine environments, posing a new threat to our oceans as a new source of microplastics [6]. Initially designed to protect ground zero workers in medical centres [7], medical waste is now being largely expended by households in response to the COVID-19 pandemic [8]. As a result of the pandemic, household PPE is being used at an increased rate [9], which places increased pressure on waste management systems [10]. Single-use PPE is primarily made up of thermoplastics such as polypropylene (PP) and polyethylene (PE), which are also typical examples of plastics being utilised across multiple products globally [11]. Although the applications of various waste plastics in concrete construction have been extensively studied, to the best of the author’s knowledge the adoption of plastic-based waste PPE in concrete applications have rarely been researched therefore, to address the research gap this experimental study highlights the potential to incorporate single-use PPE into concrete whilst also showing positive effects on the material’s compressive strength.
2 Materials, Mix Design and Methodology The materials utilised in this experimental study include Portland cement (PC), fine and coarse aggregates, potable water, superplasticizer, shredded face masks, shredded nitrile gloves and shredded isolation gowns. Table 1 outlines the chemical composition of PC utilised in the experimental study, and Table 2 details the mix design followed in the experimental study. Figure 1 highlights the mineralogical composition of PC. Table 1. Chemical composition of PC Oxides
Cement
CaO
69.2%
Al2 O3
3.8%
SiO2
18.1%
SO3
2.6%
MgO
1.3%
P2 O5
0.5%
Fe2 O3
3.4%
K2 O
0.5%
TiO2
0.3%
Na2 O
0.2%
MnO
>0.1%
ZnO
>0.1%
Other
>0.1%
Utilisation of COVID-19 Waste PPE in the Applications
523
Table 2. Mix design for experimental study Material (kg/m3 )
Control
Cement
420
Coarse Aggregate
1260
SFM1
Fine Aggregate
630
Water
210
Superplasticiser (mL/m3 )
1260
Material Water Absorption (%)
PPE
3.7 –
8.9
SFM15
5.55
SFM2
7.41
NG1
NG2
NG3
IG1
IG2
IG3
3.7
5.55
7.41
0.084
0.168
0.252
14
1.97
SFM = Shredded Face Masks, NG = Nitrile Gloves, IG = Isolation Gowns
For the purpose of this experimental study, PPE was incorporated at slightly different rates. SFM1 represents that shredded face masks were incorporated into the mix design at a rate of 0.1% by volume of concrete, NG2 represents that nitrile rubber gloves were incorporated at 0.2% by volume of concrete, and IG3 represents isolation gowns being incorporated at 0.03% by volume of concrete. All PPE was shredded into a nominal size of 5mm in width and 20mm in length prior to being incorporated into the concrete mix.
Fig. 1. Mineralogical composition of GPC
524
S. Kilmartin-Lynch et al.
3 Results and Discussion 3.1 SEM-EDS Analysis As shown in Fig. 2, the SEM-EDS analysis shows how shredded NG bond within the cement matrix at 25x magnification. The SEM analysis clearly demonstrates that the shredded nitrile rubber provides an exemplary bond performance within the cement matrix, forming an extremely close bond with the cement matrix on the upright edge of the gloves. The EDS study evidently highlights the single-use gloves from its high concertation of carbon, typically present in organic compounds.
Fig. 2. SEM-EDS analysis of shredded nitrile rubber concrete
Figure 3 illustrates the SEM-EDS breakdown of the cement matrix-shredded isolation gown bond formation as well as highlighting the interfacial transition zone (ITZ) in between the cement matrix and the IG. From the EDS analysis, it is clearly evident Spectrum 1 presents as the shredded IG due to the high level of carbon present, whereas spectrum 2 specifics show higher levels of Ca, O and Si. From the SEM imagery, it can
Utilisation of COVID-19 Waste PPE in the Applications
525
be noted that there is no breach present within the ITZ which evidently reflects the excellent bond performance of the material. Furthermore, the exceptional bond performance between the IG and concrete cement matrix is essential for the crack-bridging effect.
Fig. 3. SEM-EDS analysis of shredded isolation gown concrete
3.2 Compressive Strength Results Figure 4 outlines the 28-day compressive test results for PPE incorporated into concrete. Comparatively to the control mix, concrete containing shredded face masks showed approximately 17% more compressive strength. The increase in the compressive strength is likely caused by the polypropylene in the face masks, attributing to the crack resistance effect of the fibrous material [12]. The inclusion of shredded nitrile gloves shows an increase in compressive strength that is directly related to the increase in the concentration of the nitrile rubber up until 0.2%. In contrast to the control samples, even with 0.3% of nitrile rubber gloves by volume of concrete, compressive strength increases by approximately 20%. The inclusion of nitrile rubber gloves into concrete can therefore be interpreted as improving its compressive strength. There are several reasons for the increase in compressive strength,
526
S. Kilmartin-Lynch et al.
including (a) the strong bond between the shredded nitrile rubber and cement matrix, as shown in Fig. 2, and (b) the high tensile strength of the individual materials. The concrete mix denoted IG3 showed the best results when taking a look at the isolation gowns within the concrete, with the presence of the material showing an increase of 15.5% in compressive strength. Fibre addition in the form of shredded isolation gowns made from a blend of PP/PE highlights positive effects on the compressive strength of concrete. As detailed in previous studies [13], the reinforcement of fibres increases the load-bearing capacity of concrete by confining a simple cement matrix. As shown in Fig. 3, the fibrous material produced a crack-bridging effect as well as an excellent bond, which also contributed to the increased strength.
Fig. 4. 28 – Compressive strength results
4 Conclusions and Recommendations for Future Work C1. Compressive strength of concrete can be increased by the use of single-use face masks at small percentages. C2. The SEM-EDS analysis of nitrile rubber showed a fantastic bond formation within the cement matrix as well as having no identifiable gap within the ITZ, therefore developing a foundation for an enhanced crack bridging effect. However, a more detailed analysis of the ITZ and microstructure of the concrete should be undertaken before being utilised in the field. C3. Based on the studies conducted under the SEM-EDS analysis, the shredded IG and the cement matrix also showed outstanding adhesion, increasing the compressive strength of the sample. C4. Using single-use personal protective equipment in concrete production can result in significant environmental benefits by reducing landfill waste.
Utilisation of COVID-19 Waste PPE in the Applications
527
C5. It is the authors opinion that the use of single-use PPE waste in concrete construction has excellent opportunity to be utilised as a secondary reinforcement as well as reducing significant environmental impacts that would be found if the material were to end up in a landfill. Recommendation 1. Comprehensive long-term durability and mechanical studies which incorporate PPE waste should be undertaken before being utilised in the field. Recommendation 2. Also, the effect of different size materials should be studied on the core mechanical properties and durability of concrete.
References 1. Boroujeni, M., Saberian, M., Li, J.: Environmental impacts of COVID-19 on Victoria, Australia, witnessed two waves of Coronavirus. Environ. Sci. Pollut. Res. 28(11), 14182–14191 (2021). https://doi.org/10.1007/s11356-021-12556-y 2. Adyel, T.M.: Accumulation of plastic waste during COVID-19. Science 369(6509), 1314– 1315 (2020). https://doi.org/10.1126/science.abd9925 3. Purnomo, C.W., Kurniawan, W., Aziz, M.: Technological review on thermochemical conversion of COVID-19-related medical wastes. Resour. Conserv. Recycl. 167, 105429 (2021). https://doi.org/10.1016/j.resconrec.2021.105429 4. Kilmartin-Lynch, S., Roychand, R., Saberian, M., Li, J., Zhang, G.: Application of COVID-19 single-use shredded nitrile gloves in structural concrete: case study from Australia. Sci. Total Environ. 812, 151423 (2021). https://doi.org/10.1016/j.scitotenv.2021.151423 5. Kilmartin-Lynch, S., Roychand, R., Saberian, M., Li, J., Zhang, G., Setunge, S.: A sustainable approach on the utilisation of COVID-19 plastic based isolation gowns in structural concrete. Case Stud. Constr. Mater. 17, e01408 (2022) 6. Anastopoulos, I., Pashalidis, I.: Single-use surgical face masks, as a potential source of microplastics: do they act as pollutant carriers? J. Mol. Liq. 326, 115247 (2021) 7. Ammendolia, J., Saturno, J., Brooks, A.L., Jacobs, S., Jambeck, J.R.: An emerging source of plastic pollution: environmental presence of plastic personal protective equipment (PPE) debris related to COVID-19 in a metropolitan city. Environ. Pollut. 269, 116160 (2021). https://doi.org/10.1016/j.envpol.2020.116160 8. Zambrano-Monserrate, M.A., Ruano, M.A., Sanchez-Alcalde, L.: Indirect effects of COVID19 on the environment. Sci. Total Environ. 728, 138813 (2020) 9. Qian, W., Wang, Z.-J., Li, K.: Medical waste disposal method selection based on a hierarchical decision model with intuitionistic fuzzy relations. Int. J. Environ. Res. Public Health 13(9), 896 (2016). https://doi.org/10.3390/ijerph13090896 10. Ma, Y., Lin, X., Wu, A., Huang, Q., Li, X., Yan, J.: Suggested guidelines for emergency treatment of medical waste during COVID-19: Chinese experience. Waste Disposal Sustain. Energy 2(2), 81–84 (2020). https://doi.org/10.1007/s42768-020-00039-8 11. Jonbi, J., Meutia, W., Tjahjani, A., Firdaus, A., Romdon, S.: Mechanical properties of polypropylene plastic waste usage and high-density polyethylene in concrete. IOP Conf. Ser. Mater. Sci. Eng. 620, 012034 (2019). https://doi.org/10.1088/1757-899X/620/1/012034 12. Nili, M., Afroughsabet, V.: The effects of silica fume and polypropylene fibers on the impact resistance and mechanical properties of concrete. Constr. Build. Mater. 24(6), 927–933 (2010). https://doi.org/10.1016/j.conbuildmat.2009.11.025 13. Farooq, M.A., Fahad, M., Ali, B., El Ouni, M.H., Elhag, A.B.: Influence of nylon fibers recycled from the scrap brushes on the properties of concrete: valorization of plastic waste in concrete. Case Stud. Constr. Mater. 16, e01089 (2022)
Microbial Induced Calcium Carbonate Precipitation (MICP) Treatments for the Reduction of Water Absorption of Recycled Mixed Aggregates Brigitte Nagy(B)
, Johanna Zentner, and Andrea Kustermann
Department of Civil Engineering, Munich University of Applied Sciences HM, Munich, Germany [email protected]
Abstract. A novel treatment method for the improvement of recycled aggregates is MICP (microbial induced calcium carbonate precipitation), which is already used successfully in soil improvement. Recycled mixed aggregates (RMA), in particular, have limited use as they influence the performance of the concrete. The precipitation can be used as a surface treatment for the reduction of the porosity and water absorption of recycled aggregates. This study aims to compare three different MICP treatment methods to investigate their potential for the reduction of water absorption properties of RMA. MICP treatments such as spraying, immersion and an alternative MICP treatment using a Büchner funnel were tested and compared. The results show that all methods used were able to reduce the water absorption of the RMA. It was also found that multiple MICP treatments enhance this effect. In the process, the water absorption of the RMA could be reduced up to 42.8% (from 21% to 12%) after the third MICP treatment when using the immersion method with 24 h intervals. It was found, that due to the MICP treatments, the RMA requires less pre-wetting water for mortar production than untreated RMA. Also, layers of microbial precipitated CaCO3 could be detected on the surface of RMA grains. Keywords: Recycled mixed aggregates · MICP · Biocementation · Biodeposition
1 Introduction The demolition of buildings produces immense amounts of construction waste. The use of recycled aggregate is a sustainable way to save natural resources. Recycled coarse aggregate (RCA) is already successfully used proportionally for the production of new concrete. However, the replacement of natural aggregates (NA) with recycled fine aggregates (RFA) or recycled mixed aggregates (RMA) is often challenging. The attached mortar (AM) on the aggregate surface results in high water absorption and porosity of RCA [1]. Also, the content of non-cement-based materials such as masonry bricks © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 A. J˛edrzejewska et al. (Eds.): SynerCrete 2023, RILEM Bookseries 43, pp. 528–539, 2023. https://doi.org/10.1007/978-3-031-33211-1_47
Microbial Induced Calcium Carbonate Precipitation (MICP) Treatments
529
influences the properties of RMA. Rodríguez et al. [2] found a decrease in mechanical properties with increasing RMA content using 2/6 and 5/12 fractions. For that reason, RMA or RFA are usually disposed of in landfills or used in street construction [3]. A novel treatment method to improve the properties of RMA could be MICP (microbial induced calcium carbonate precipitation) which is already used successfully in soil improvement. In this process, ureolytic bacteria produce CaCO3 in combination with a cementation solution. The precipitation also can be used to produce a CaCO3 layer that fills the pores of the recycled aggregate to reduce the porosity [4]. The preferred treatment method is the immersion of RCA in MICP reagents which has already been tested in several studies with promising results e.g. in [5–8]. Also, spraying methods [9, 10] and alternative treatments [11, 12] with admixtures [13] or combinations with grinding steps [14] have been carried out successfully. Wu et al. [9] achieve similar results in the reduction of water absorption with both the immersion method and spraying. Mistri et al. [10] reduce the water absorption of RCA by 70% using a spraying method. RCA is mainly used as the base material for investigations. However, there are also studies [14–16] using similar mixed materials as is the case in this study to investigate bacterial treatments regarding the reduction of water absorption. García-González et al. [7] were able to reduce the water absorption of coarse mixed and ceramic recycled mixed aggregates by up to 18% with immersion methods. However, previous studies have rarely dealt with the improvement of RMA with particle size under 2 mm and initial water absorption of more than 20%. This study aims to investigate the potential of MICP treatments for the reduction of the water absorption of RMA, especially for the particle size of 0–2 mm in order to reduce the amount of pre-wetting water needed for mortar production. For this purpose, three MICP treatment methods were carried out, spraying, immersion and an alternative developed MICP treatment using a Büchner funnel. From preliminary investigations, the CaCO3 formed on the RMA surface was analysed. The success of the MICP treatments is examined by the reduction of the water absorption measured before and after the treatments of RMA. To determine the improvement effects of MICP treatments on mortar properties, samples of 100% natural, untreated and MICP treated RMA were compared.
2 Material and Methods 2.1 Recycled Mixed Aggregate (RMA) The RMA used in this study comes from a demolition site of a former military building in Munich. The RMA consists of concrete components, masonry bricks and old mortar. The constituents according to DIN EN 933-11 [17] are listed in Table 1. To compare the properties of the RMA with natural aggregate, the grading curve of the CEN reference sand with a grain size of 0–2 mm [18] was used as a reference. The particle size distribution of the RMA was adjusted by interpolation (Fig. 1) to have similar grading curves to eliminate the effect of grain distribution as performed in [19]. The oven-dried density of the RMA (0–2 mm) is 2.45 kg/dm3 and 2.55 kg/dm3 of the CEN reference sand. Similar to the investigations in [5], the water absorption (WA) of the RMA was determined separately for each grain size fraction. The results are shown in Fig. 2. Based on these results it was decided to evaluate the MICP treatments on the
530
B. Nagy et al. Table 1. Constituents according to [17] of the RMA.
Class Components
Proportion [wt.%]
Rc
concrete, concrete products, mortar, concrete bricks
21.9
Ru
natural stone, hydraulic bonded aggregate, unbound aggregate
69.3
Rb
masonry, clinker, limestone bricks, various masonry or roof bricks
8.6
X
wood, plastic, rubber, gypsum
0.2
100% CEN reference sand 90% RMA 80% 70% 60% 50% 40% 30% 20% 10% 0% 0.000 0.063 0.080 0.125 0.160 0.250 0.500 1.000 1.600 2.000
Sieve size [mm] Fig. 1. Particle size distribution of CEN reference sand and RMA. 60%
Water absorption [%]
Passing [%]
grain size fraction 0.125–0.250 mm. It is considered, that the MICP treatments will have a significant effect on fine grains with high water absorption [20].
50%
51.49%
40% 30%
24.95% 21.36% 13.46%
20%
12.70%
8.84%
10% 0% 0.00 0.063
0.063 0.125
0.125 0.250
0.25 0.50
0.50 1.00
1.00 2.00
Grain sizes fraction [mm] Fig. 2. Water absorption (WA) per RMA grain size fraction (n = 2).
Microbial Induced Calcium Carbonate Precipitation (MICP) Treatments
531
2.2 MICP Reagents S. pasteurii (DSM33) was cultivated in a supplemented complex medium, based on Lapierre et al. [21] by the Biotechnology Laboratory of the Department of Engineering and Management. This approach achieves an increased biomass concentration. The optical density of the bacterial cultures used measured at a wavelength of 600 nm (OD600 ) were 11.76 and 13.31. Before use, the culture was diluted and stored at 12 °C. The original OD600 of the bacterial culture was diluted with a NaCl solution to an OD600 of 1. The separate use of bacterial culture dilution and cementation solution in equal volume proportions was carried out for the treatments. The molar concentration in the cementation solution was prepared with 1.68 mol/L calcium chloride and 2.22 mol/L urea. 2.3 MICP Treatment Methods In order to visualise the precipitated CaCO3 coating the grains, preliminary tests were carried out with coarse RMA. Here, the RMA was immersed in MICP reagents, dried after 24 h and then embedded in epoxy resin. Based on these findings, methods of MICP treatment for RMA were developed. Figure 3 shows an overview of the three different MICP treatment techniques and the corresponding procedure, which are defined as one treatment step (bacterial culture and cementation solution). Up to three treatments were carried out. Between the treatments, the RMA was completely oven-dried (110 °C) to ensure optimal conditions for the next MICP treatment and to allow better absorption of the MICP reagents [10]. The MICP treatment effect was determined after each MICP treatment step by testing the water absorption.
Fig. 3. Schematic sketch of the different MICP treatment methods.
Immersion Method. The immersion of RCA in bacterial culture for 24 h was already tested in several studies e.g. in [19]. However, in most cases, the material was then immersed in cementation solution for a longer time. The RMA in this study was first placed in bacterial culture and then stored in cementation solution for 24 h each. When changing from bacteria into cementation solution, the RMA was not dried. It is assumed, that the precipitation process can only proceed in aqueous solution. Spraying Method. During the spraying method, the RMA was first sprayed with bacterial culture and immediately afterwards with cementation solution. The used volume
532
B. Nagy et al.
(bacterial culture and cementation solution) was determined according to the water absorption capacity of the untreated RMA. After treatment, the RMA was surface saturated with MICP reagents. Büchner Funnel Method. This method was performed in a similar way to the method used for the determination of the water absorption [22]. The RMA was immersed in bacterial culture in the Büchner funnel for 5 min. After pumping out the bacterial culture, cementation solution was added and the RMA was further immersed for 5 min. The water absorption test was carried out on the dry RMA as usual between the MICP treatments. 2.4 Preparation of Mortar Samples After evaluation of the three different MICP treatments, it was decided to prepare mortar samples with treated RMA from the Büchner funnel method used. Compared to the spraying method, the Büchner funnel method results in a constant reduction of water absorption of the RMA after each treatment. Although the immersion method also shows a higher treatment effect, it is not easy to implement this method in practice due to the time-consuming procedure of the immersion and drying periods. The mortar was produced according to DIN EN 196-1 [18]. Three mortar (M) samples each with the dimensions of 4 × 4 × 16 mm were prepared by using natural aggregate (NA-M), untreated RMA (RMA-M) and MICP treated RMA (tRMA-M). The mortar mixtures calculated for three mortar samples are shown in Table 2. Table 2. Composition of mortar mixtures. NA-M
RMA-M
tRMA-M
CEM I 42.5 N [g]
450
450
450
Water [g]
225
225
225
Extra water to pre-wet (70% of WA) [g]
–
148
103
Natural aggregate (NA) [g]
1350
–
–
Untreated RMA [g]
–
1506
–
MICP treated RMA [g]
–
–
1476
2.5 Testing Methods Water Absorption. In order to examine the MICP treatment effect on the RMA, water absorption tests according to DIN EN 1097-6 [22] were carried out before and after the MICP treatments. The method with the Büchner funnel and a vacuum pump was used. The immersion and deflation time of 5 min was set. Filter papers type MN 612 1/4 with a filtration speed of 10 s/10 mL were used. Equation (1) defines the water absorption [22]. WA(t) =
M3 (t) − M2 × 100 M2 − M1
(1)
Microbial Induced Calcium Carbonate Precipitation (MICP) Treatments
533
where: M1 = Mass of the Büchner funnel with wet filter paper [g]; M2 = Mass of the Büchner funnel with wet filter paper and dry sample [g]; M3 = Mass of the Büchner funnel with wet filter paper and surface-dry sample after 5 min pumping time [g]. Compressive and Flexural Strength. The compressive and flexural strength of the hardened mortar samples were tested similarly to [18] at an early stage after 7 days. Slump Flow Value. In order to test the improvement effect on the fresh mortar consistence properties, the slump flow value was determined according to DIN EN 1015-3 [23] by flow table. This allows the evaluation of the consistency of the fresh mortar according to DIN EN 1015-6 [24]. Characterization of MICP Treatment. For investigations of CaCO3 formed by MICP treatments, X-Ray diffraction (XRD) and X-Ray fluorescence analysis (XRF) were carried out on RMA before and after MICP treatment. Also, the untreated RMA and MICP treated RMA with Büchner funnel method were investigated by scanning electron microscope (SEM).
3 Results and Discussion 3.1 Preliminary Tests The microscopic evaluation of the MICP treated coarse RMA embedded in greencoloured epoxy resin before and after MICP immersion is shown in Fig. 4. Similar to [11], this method was useful for visualising RMA grains of brick material. CaCO3 precipitate on the surface of the RMA grains due to the MICP treatment can be seen. However, the CaCO3 layer has no equal distribution on the grains.
1000 μm
1000 μm
Fig. 4. Microscopic images of the sample cross section of untreated (left) and MICP treated (right) RMA embedded in epoxy resin.
534
B. Nagy et al.
3.2 MICP Treatments Water Absorption. The results of the water absorption of the RMA before and after the different MICP treatments (immersion, spraying and the Büchner funnel method) are shown in Fig. 5. It can be seen, that all three MICP treatment methods have an effect of water absorption reduction of the RMA, already after the first treatment. After additional MICP treatments, the water absorption of the RMA can also be further reduced. 22%
21%
immersion method
Water absorption [%]
20% 19% 18%
17% 16%
16%
spraying method Büchner funnel method 17% 16% 14%
17% 15%
14% 12% 12% untreated RMA
after 1 MICP treatment
after 2 MICP treatments
after 3 MICP treatments
Fig. 5. Results of the water absorption of RMA before and after different MICP treatments.
Compared to untreated RMA, the MICP immersion method achieves the highest water adsorption reduction down to 12% after the 3rd treatment. That results in a reduction rate of more than 42.8%. Also, the Büchner funnel method shows a continuous reduction after each treatment step of the RMA to 15% after the 3rd treatment. However, with the spraying method, multiple treatments do not seem to have any effect. After the first treatment, the water absorption is constant at 17%. Similar to the results in this study, Jagan et al. [19] were able to reduce the water absorption of RCA after 24 h immersion in Bacteria (and 5 d in medium) by 36.54% (from approx. 6% to 4%). Studies [7, 15, 16] using comparable mixed recycled aggregates as used in this study, have achieved promising results in reduction of water absorption properties (with immersion method) compared to the untreated material up to 18% [7]. However, it should be noted, that none of the studies used RMA with such a high water absorption (21%) as is the case in this investigation. De Belie et al. [15] observed, that more CaCO3 was formed on the RMA than on the RCA and indicate that the high porosity of the RMA could be responsible for this. However, they could not find any proportional correlation between the reduction of water absorption and CaCO3
Microbial Induced Calcium Carbonate Precipitation (MICP) Treatments
535
formation. The successful reduction of water absorption properties by MICP may depend on the composition of the RMA [7]. Characterization of MICP Treatment. SEM imagines of untreated RMA (a–b) and MICP treated RMA (c–d) are shown in Fig. 6. It can be observed that the RMA of the grain fraction 0.125–0.250 mm used as initial material in this investigation, mainly consists of particles as shown in Fig. 6 (a). In detail, a porous surface of the particles can be seen (Fig. 6 (b)). After MICP treatment of the RMA, however, contrary to expectations, rarely calcite was found on the surface. Crystals, which could probably be CaCO3 , could only be found in a few places (Fig. 6 (c–d)). The sample tested was probably not representative of the entire treated RMA sample.
a)
b)
c)
d)
Fig. 6. SEM imagines of untreated (a–b) and MICP treated RMA (c–d).
The results of the XRD and XRF analysis also confirm that after the MICP treatment of the RMA the amount of CaCO3 increased by only 3%. This might be due to the Büchner funnel method used. Probably, the CaCO3 did not coat the RMA grains and instead, CaCO3 precipitated loosely in the solution. Wang et al. [25] describe a shell
536
B. Nagy et al.
effect of CaCO3 and indicate, that the solution pressure plays a role in the growth of the precipitation matrix.
3.3 Mortar Properties Fresh Mortar. The results of the fresh mortar properties containing NA, untreated RMA and MICP treated RMA using the Büchner funnel method are listed in Table 3. It can be observed, that the fresh mortar consistency with untreated RMA, although additional water was used for pre-wetting, does not result in similar consistency of the fresh mortar with NA. The MICP treatment of RMA did not improve the workability of the mortar (tRMA-M). Thus, the use of RMA results in stiff mortars in each case. Table 3. Results of the fresh mortar properties according to [23, 24]. mortar
slump flow [23]
consistence [24]
pre-wetting water
NA-M
154 mm
plastic mortar
–
RMA-M
129 mm
stiff mortar
148 mL
tRMA-M
125 mm
stiff mortar
103 mL
Also, in Wu et al. [5], the fresh mortar properties of mortars (w/c = 0.5) with natural aggregate were found to have a more fluidity consistency than was the case with mortar of RCA but they were able to improve the workability through bacterial treatment. Zhu et al. [6] were also able to improve the consistency of the fresh concrete (w/c = 0.5) with treated RMA compared to untreated RMA, indicating that the reduced water absorption of the treated RMA is responsible for more free water available in the concrete mixture. In [20], however, the flow results are in a similar range and show no significant difference between the mortar of untreated and microbial treated RCA. Although there is no improvement in the workability of the fresh mortar after MICP treatment, it can be observed, that the treated RMA requires less extra water for prewetting than untreated RMA. Hardened Mortar. Compared to the mortar NA-M with 2.14 kg/dm3 , RMA-M (1.86 kg/dm3 ) shows 13% less bulk density. Due to the MICP treatment, the tRMAM has a bulk density of 1.98 kg/dm3 which is 6% higher than the untreated RMA-M. The results of strength tests are shown in Fig. 7. Compared to NA-M, the compressive strength of RMA-M is 47% lower and the flexural strength 31% lower. Also, the tRMAM results no improvement in strength due to MICP treatment. The flexural strength and compressive strength of tRMA-M have even decreased by 23.6% and 5.9%, compared to the RMA-M. Most studies e.g. [5, 26] were able to show that MICP treatment also results an improvement of compressive strength of the concrete when using treated RCA. It was even possible to achieve an improvement compared to samples made of NA [15, 16]. This effect could not be demonstrated in this study. Also, Sonmez et al. [20] found no
Microbial Induced Calcium Carbonate Precipitation (MICP) Treatments
537
flexural strength [N/mm²]
35.0
29.0
30.0 25.0 20.0
NA-M RMA-M tRMA-M
15.3
15.0 10.0 5.0 0.0
5.5
3.8
2.9
14.4
compressive strength [N/mm²]
significant differences in the compressive strength of the mortar samples before and after treatment. Zhao et al. [27] showed that mortar samples with lower w/c ratios had higher flexural strength after 7 or 28 days.
Fig. 7. Results of flexural strength (n = 3) and compressive strength (n = 6) of mortar with natural aggregate (NA-M), untreated RMA (RMA-M) and MICP treated RMA (tRMA-M).
It can not be excluded that the extra water for pre-wetting influences the w/c ratio and thus leads to a reduction in strength. In addition, the improvement potential of compressive strength properties also depends on the type of material of the RMA [10]. MICP treatment is still a surface treatment and therefore has less potential for improvement when applied to RMA with initial adverse properties compared to RCA [10, 16].
4 Conclusions This study shows a comparison of three different MICP treatment methods for the reduction of the water absorption of recycled mixed aggregates (RMA) with a particle size of 0–2 mm. The results show that all methods used were able to reduce the high water absorption of the RMA used. It was also found that multiple MICP treatments enhance this effect. Thereby, the immersion method showed the highest effect and the water absorption could be reduced by 42.8% (from 21% to 12%). Also, the spraying method and the alternative developed MICP method using the Büchner funnel, result in a reduction in water absorption of 19% and 28% after the third MICP treatment. While the MICP treatments show promising results in the reduction of water absorption of the RMA, no improvement in the fresh or hardened mortar properties could be obtained. However, it could be shown that less extra water is needed for pre-wetting due to the MICP treatment of the RMA for similar workability of the mortar. Further investigations are needed to examine the influence of the precipitation process on the surface of the RMA in order to optimise the MICP treatment methods in terms of their improvement effect. It also needs to be determined whether the methods can be applied to other recycled aggregates. According to the results, MICP seems to be a promising approach for the reduction of the water absorption of RMA. This could lead to more acceptance for the use of RMA in construction and thus a higher recycling rate could be achieved.
538
B. Nagy et al.
Acknowledgements. The project “BiCeRcrete” is part of the joint research project “ForCYCLE Technikum” funded by the Bavarian Ministry of the Environment and Consumer Protection (StMUV) and coordinated by the Centre for Basic Materials Efficiency (REZ). The authors thank Dr. Robert Huber, Patrick Hanisch and Frédéric Lapierre for the cultivation and providing the bacterial culture. The authors thank Benjamin Wolf for the support with XRD analysis and Dr. Constanze Eulenkamp for the SEM images.
References 1. Shi, C., Li, Y., Zhang, J., Li, W., Chong, L., Xie, Z.: Performance enhancement of recycled concrete aggregate – a review. J. Clean. Prod. 112, 466–472 (2016) 2. Rodríguez, C., et al.: The incorporation of construction and demolition wastes as recycled mixed aggregates in non-structural concrete precast pieces. J. Clean. Prod. 127, 152–161 (2016) 3. Wang, B., Yan, L., Fu, Q., Kasal, B.: A Comprehensive review on recycled aggregate and recycled aggregate concrete. Resour. Conserv. Recycl. 171, 105565 (2021) 4. Mistri, A., Dhami, N., Bhattacharyya, S.K., Barai, S.V., Mukherjee, A.: Biocement treatment for upcycling construction and demolition wastes as concrete aggregates. Mater. Struct. 55(152), 1–18 (2022) 5. Wu, C.-R., Zhu, Y.-G., Zhang, X.-T., Kou, S.-C.: Improving the properties of recycled concrete aggregate with bio-deposition approach. Cement Concr. Compos. 94, 248–254 (2018) 6. Zhu, Y., Li, Q., Xu, P., Wang, X., Kou, S.: Properties of concrete prepared with recycled aggregates treated by bio-deposition adding oxygen release compound. Materials 12(13), 2147 (2019) 7. García-González, J., et al.: Quality improvement of mixed and ceramic recycled aggregates by biodeposition of calcium carbonate. Constr. Build. Mater. 154, 1015–1023 (2017) 8. Feng, Z., Zhao, Y., Zeng, W., Lu, Z., Shah, S.P.: Using microbial carbonate precipitation to improve the properties of recycled fine aggregate and mortar. Constr. Build. Mater. 230, 116949 (2020) 9. Wu, C.-R., Hong, Z.-Q., Zhang, J.-L., Kou, S.-C.: Pore size distribution and ITZ performance of mortars prepared with different bio-deposition approaches for the treatment of recycled concrete aggregate. Cement Concr. Compos. 111, 103631 (2020) 10. Mistri, A., Dhami, N., Bhattacharyya, S.K., Barai, S.V., Mukherjee, A., Biswas, W.K.: Environmental implications of the use of bio-cement treated recycled aggregate in concrete. Resour. Conserv. Recycl. 167, 105436 (2021) 11. Manzur, T., Rahman, F., Afroz, S., Huq, R.S., Efaz, I.H.: Potential of a microbiologically induced calcite precipitation process for durability enhancement of masonry aggregate concrete. J. Mater. Civil Eng. 29(5) (2017) 12. Singh, L.P., Bisht, V., Aswathy, M.S., Chaurasia, L., Gupta, S.: Studies on performance enhancement of recycled aggregate by incorporating bio and nano materials. Constr. Build. Mater. 181, 217–226 (2018) 13. Rais, M.S., Khan, R.A.: Development of sustainable admixture-based recycled aggregate concrete using ureolytic bacteria. Innov. Infrastruct. Solutions 7(2), 1–27 (2022) 14. Gong, Y., Chen, P., Lin, Y., Wan, Y., Zhang, L., Meng, T.: Improvement of recycled aggregate properties through a combined method of mechanical grinding and microbial-induced carbonate precipitation. Constr. Build. Mater. 342, 128093 (2022) 15. De Belie, N., et al.: Improving the quality of various types of recycled aggregates by biodeposition. In: 14th International Conference on Durability of Building Materials and Components (XIV DBMC), pp. 1–8. RILEM Publications, Paris, France (2017)
Microbial Induced Calcium Carbonate Precipitation (MICP) Treatments
539
16. Wang, J., Vandevyvere, B., Vanhessche, S., Schoon, J., Boon, N., De Belie, N.: Microbial carbonate precipitation for the improvement of quality of recycled aggregates. J. Clean. Prod. 156, 355–366 (2017) 17. DIN EN 933-11:2011-05, Tests for geometrical properties of aggregates - Part 11: Classification test for the constituents of coarse recycled aggregate; German version EN 933-11:2009 + AC:2009 (2011) 18. DIN EN 196-1:2016-11, Methods of testing cement - Part 1: Determination of strength; German version EN 196-1:2016 (2016) 19. Jagan, S., Neelakantan, T.R., Padma Lakshmi, M.: Enhancement on the properties of recycled coarse aggregate through bio-deposition – an experimental study. Mater. Today Proc. 49, 1141–1147 (2022) 20. Sonmez, M., Ilcan, H., Dundar, B., Yildirim, G., Ersan, Y.C., Sahmaran, M.: The effect of chemical-versus microbial-induced calcium carbonate mineralization on the enhancement of fine recycled concrete aggregate: a comparative study. J. Build. Eng. 44, 103316 (2021) 21. Lapierre, F.M., Schmid, J., Ederer, B., Ihling, N., Büchs, J., Huber, R.: Revealing nutritional requirements of MICP-relevant Sporosarcina pasteurii DSM33 for growth improvement in chemically defined and complex media. Sci. Rep. 10, 22448 (2020) 22. DIN EN 1097-6:2022-05, Tests for mechanical and physical properties of aggregates - Part 6: Determination of particle density and water absorption; German version EN 1097-6:2022 (2022) 23. DIN EN 1015-3:2007-05, Methods of test for mortar for masonry - Part 3: Determination of consistence of fresh mortar (by flow table); German version EN 10153:1999+A1:2004+A2:2006 (2007) 24. DIN EN 1015-6:2007-05, Methods of test for mortar for masonry - Part 6: Determination of bulk density of fresh mortar; German version EN 1015-6:1998+A1:2006 (2007) 25. Wang, R., Jin, P., Ding, Z., Zhang, W.: Surface modification of recycled coarse aggregate based on Microbial Induced Carbonate Precipitation. J. Clean. Prod. 328, 129537 (2021) 26. Zhao, Y., Peng, L., Zeng, W., sun Poon, C., Lu, Z.: Improvement in properties of concrete with modified RCA by microbial induced carbonate precipitation. Cement Concr. Compos. 124, 104251 (2021) 27. Zhao, Y., Peng, L., Feng, Z., Lu, Z.: Optimization of microbial induced carbonate precipitation treatment process to improve recycled fine aggregate. Cleaner Mater. 1, 100003 (2021)
Use of Recycled Carbon Fibres in Textile Reinforced Concrete for the Construction Industry Vanessa Overhage(B)
and Thomas Gries
Institut Fuer Textiltechnik of RWTH Aachen University, Aachen, Germany [email protected]
Abstract. The construction industry is the most resource-intensive sector. With a demand of 4.4 billion tonnes annually, concrete consumes, as the most used material in the world, a huge amount of resources. For the year 2050, an increase in demand of a further 25% is already forecast. The main components of concrete are materials whose resources are limited. With the use of textile-reinforced concrete, resources for concrete production can be saved compared to steel-reinforced concrete. The textiles used do not corrode and therefore, in contrast to steel reinforcements, do not have to be protected from external environmental influences with a minimum concrete cover. Thus, a low concrete cover is already sufficient, whereby up to 80% concrete can be saved. A high amount of energy is required to produce the new carbon fibres, so it is desirable to use these already existing fibre materials for as long as possible. It is already technically possible to separate carbon fibres from plastics. Studies show that recycled carbon fibres from carbon fibre reinforced plastic waste still have 80% of the original mechanical properties. A possible field of application for reuse of the recycled material could be in the construction industry. However, the production of inseparable material composites should be avoided. Therefore, the use of recycled fibre materials as textile reinforcement is being investigated in order to enable a separation of the composite components again at the end of the second life cycle of the carbon fibres. Keywords: Recycling · Carbon Fibre · Textile Reinforced Concrete · Mortar · Textile Reinforcement
1 Introduction Concrete is the most commonly used building material and, after water, the most widely used material in the world [1, 2]. It has a high compressive strength, but only a low tensile strength. Therefore, it is reinforced with a material that can absorb the tensile forces. The aim of the work is to investigate the applicability of nonwoven textiles and yarns made of recycled carbon fibres (rCF) as reinforcement in concrete. The recycling of carbon requires less energy than the production of conventional reinforcement from © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 A. J˛edrzejewska et al. (Eds.): SynerCrete 2023, RILEM Bookseries 43, pp. 540–551, 2023. https://doi.org/10.1007/978-3-031-33211-1_48
Use of Recycled Carbon Fibres in Textile Reinforced Concrete
541
steel or carbon filaments from primary raw materials [3]. Therefore, the use of rCF as reinforcement offers the potential to save energy as well as raw materials. Due to the uncorrosive material properties of carbon fibres, concrete components can be made slimmer than components with steel reinforcement, which can reduce CO2 emissions in the production of these components.
2 State of the Art The main components of concrete are cement, sand and water. Large amounts of carbon dioxide (CO2 ) are emitted, especially during cement production. Due to the energyintensive production process, the cement industry causes 8% of the global CO2 emissions [4]. The most commonly used reinforcement material is steel. Steel provides a concrete component with high tensile strength as reinforcement, but has a decisive disadvantage: steel is susceptible to corrosion. A concrete cover layer several centimetres thick creates an alkaline environment that protects the steel from corrosion. Furthermore, this increases the amount of concrete needed. The construction industry, including the maintenance of buildings, alongside manufacturing, transportation and processing of construction materials, is responsible for 38% of the world’s CO2 emissions [5]. The reduction of greenhouse gas emissions is an essential prerequisite for achieving the goal of the Intergovernmental Panel on Climate Change to limit global warming to a maximum of 1.5 °C [6]. Due to its high share in global greenhouse gas emissions, the building industry can contribute significantly to achieving this goal. Due to global population growth and increasing urbanisation, an increasing demand for concrete is to be expected [4]. In the years 2011 to 2013 alone, more concrete was used in China than in the United States of America in the entire 20th century [7]. There are various aspects of how this goal can be achieved at the building component level. This includes, among other things, the reduction of resource consumption, the extension of service life, as well as the use of recycled materials. One approach to cover all three aspects is the use of rCF as an alternative reinforcement. The suitability of carbon as a reinforcement material has been investigated in numerous studies [8]. Carbon is particularly suitable as a reinforcement due to its mechanical properties. Due to its high tensile strength and low density, no more than one-twentieth of the mass of steel is required as carbon fibre to transmit the same tensile force. Carbon is corrosion-resistant, which means that no additional concrete cover is required. This means that up to 80% of the concrete can be saved. [9–11] Furthermore, carbon concrete components are expected to have a technical service life of 200 years, which is more than twice as long. [12, 13]. The production of carbon fibres is also an energy-intensive process. By using rCF, the global warming potential can be reduced by up to 85% compared to virgin fibres [14]. However, the quality of rCF is reduced compared to virgin fibres, as they are no longer available as continuous fibres. Instead, they can be processed into textile semifinished products for further use. The production of yarns and nonwovens from rCF is particularly promising [3].
542
V. Overhage and T. Gries
3 Materials and Methods In this work, flexural tests were carried out on unreinforced, nonwoven and on yarn reinforced concrete. The Four-point bending test is carried out in accordance with DIN EN 1170–5 for measuring the bending tensile strength (Fig. 1).
Fig. 1. Four-Point Bending Test (DIN EN 1170-5)
The elongation was measured during the flexural test by reference of the loading part displacement while conducting the Four-Point bending test. For each test series, samples with the dimension 34 × 10 × 15 cm3 were produced. The used fine concrete mix was developed in sub-project C1 of the special research area (SFB) 532 “Textile-reinforced concrete” at RWTH Aachen University. The concrete mix consists of solid and liquid components. The size of the aggregates are 0.2 - 0.6 mm. The quantity and volume proportions are listed in the following Table 1. Table 1. Fine Concrete Mixture Component
Quantity [g/dm3 ]
Vol.%
Cement CEM I 42.5 R
490
22
Fly ash
175
8
35
3
Quartz powder
500
23
Sand 0.2 - 0.6
713
32
Water
280
11
Silica fume (ElkemMicrosilica® 940U)
Superplasticizer
7
0.3
The produced samples are stored for 24 h in the mould, before the samples were demoulded. Afterwards the samples were stored for approx. 6 days in water and for a further 21 days in room climate before the flexural tests were carried out.
Use of Recycled Carbon Fibres in Textile Reinforced Concrete
543
Various textile reinforcements made out of rCF are being tested as part of this research. The following Table 2 contains the reinforcement materials consisting two different yarns and two different nonwovens. Table 2. Used textile reinforcement. Reinforcement
Material
Production method
Nonwoven 1
Recycled Carbon Fibre
Needling
Nonwoven 2
Recycled Carbon Fibre + PET Binder Thread
Maliwatt
Yarn 1
Recycled Carbon Fibre + Polyamid 6
Friction spun yarn
Yarn 2
Recycled Carbon Fibre + Polyamid 6
Wrap yarn
Yarn 1 is produced as a friction spun yarn. The used material is a combination of rCF and Polyamid 6 (PA6). In comparison, the yarn 2 was produced as a wrap yarn. This means that the core material is wrapped around by a filament. This yarn contains the same material composition as the yarn 1. The yarns were produced within the framework of the CarboYarn project. Both yarns are coated with epoxy resin, before being placed in the concrete. Nonwoven 1 consists of 100% rCF and was not further bonded after needling. The grammage of nonwoven 1 is 100 g/m2 . Nonwoven 2 is additionally bonded in the Maliwatt process. As a result, it consists of 96% rCF and 4% polyethylene terephthalate (PET) through the binder thread. The basis weight remains unchanged at 100 g/m2 . Both nonwovens were produced earlier at ITA Augsburg, Germany.
4 Results
Flexural Stress [MPa]
In the following the results of the four-point bending test are shown in the flexural stress- elongation curves. The results of the rCf nonwovens are shown in Sect. 4.1 and the results of the used rCF yarns are shown in Sect. 4.2. As a comparison for the different rCF reinforcements, an unreinforced series with plain concrete was produced (Fig. 2). 8 6
Sample 1 Sample 2 Sample 3 Sample 4 Sample 5 Sample 6
4 2 0 0
0,1
0,2
0,3
Elongation [%]
Fig. 2. Flexural stress-elongation curves of unreinforced concrete.
544
V. Overhage and T. Gries
Concrete without reinforcement has a flexural strength of around 4 MPa. The elongation is around 0,1%. As the curves shown an immediate failure occurs for all unreinforced samples. 4.1 Recycled Carbon Fibre Nonwoven Reinforcement In this section, the results with rCF nonwoven reinforcement are shown. As mentioned before two different nonwovens were used. Additionally, different variations were made with both nonwovens (Table 3). Table 3. Sample variations of the nonwoven 1. Reinforcement
Variation
Nonwoven 1
Fully pre-impregnated Fully rolled
Nonwoven 1 was used as a fully textile reinforcement. In the first series of tests, the nonwoven was pre-impregnated with the used fine concrete matrix in order to improve the impregnation of the nonwoven. In contrast, in the second series, the nonwoven was placed in the mould on a first layer of concrete and then pressed onto the concrete matrix with a roller. Afterwards, the mould was completely filled with the second layer of concrete. The results of the flexural test are shown in the following two figures (Fig. 3). 12
Flexural Stress [MPa]
10 8 6 4 2
Sample 1 Sample 2 Sample 3
0 0
0,1
0,2
0,3
Elongation [%]
Fig. 3. Flexural stress-elongation curves pre-impregnated nonwoven 1 reinforced concrete.
Concrete with the pre-impregnated nonwoven 1 reinforcement has a flexural strength of around 6 MPa. The elongation is around 0,1%. As the curves shown, the failure occurs
Use of Recycled Carbon Fibres in Textile Reinforced Concrete
545
not directly, but in two or in the case of sample 2 in four stages. Each curve has a stress peak between the beginning of the test and the tensile strength, but then continue to rise with the same gradient. 12
Flexural Stress [MPa]
10 8 6 4 2
Sample 1 Sample 2 Sample 3
0 0
0,1
0,2
0,3
Elongation [%]
Fig. 4. Flexural stress-elongation curves of rolled nonwoven 1 reinforced concrete.
Concrete with the rolled nonwoven 1 reinforcement has a flexural strength of around 8 MPa. The elongation is between 0.1 and 0.2%. As the curves in Fig. 4 show, the failure is not direct but occurs in three steps. The first two peaks have a similar force and the last peak has a lower strength and a lower gradient. Comparing both results of nonwoven 1, the flexural stress-elongation curves of the pre-impregnated nonwoven 1 drop off abruptly and faster than the rolled nonwoven 1 after reaching the tensile strength. The reached tensile strength of the rolled nonwoven 1 is a little bit higher compared to the pre-impregnated nonwoven 1. Nonwoven 2 was also used as a fully textile reinforcement and as a reinforcement made out of stripes (Table 4). Table 4. Sample variations of the nonwoven 2. Reinforcement
Variation
Nonwoven 2
Fully pre-impregnated Fully rolled Stripes rolled
In the first series of tests, the nonwoven 2 was pre-impregnated with the used fine concrete matrix. In contrast, in the second series, the nonwoven was pressed onto the concrete matrix with a roller. In the third variation of nonwoven 2 the nonwoven was
546
V. Overhage and T. Gries
cut into strips of 16 mm. Three stripes were inserted in each sample with a distance of 16 mm in between. Therefore, the reinforcement in the rolled stripes sample contain 50% of reinforcement compared to the other nonwoven samples. The results of the three different variations of nonwoven 2 are shown in the following three figures (Figs. 5, 6 and 7).
Fig. 5. Flexural stress-elongation curves pre-impregnated nonwoven 2 reinforced concrete.
Concrete with the pre-impregnated nonwoven 2 reinforcement has a flexural strength of around 8 MPa. The elongation is around 0.2%. As the curves shown, the failure occurs not directly, but in two steps. Each curve has a stress peak between the beginning of the test and the tensile strength, but then continue to rise with the similar or lower gradient. The yield strength is higher than the tensile strength. The plastic deformation range is more distinctive than with nonwoven 1, especially in Sample 2. Concrete with the rolled nonwoven 2 reinforcement has a flexural strength between 7 and 10 MPa. As the curves shown, the failure occurs not directly. The plastic deformation range is shorter than in the curves of the preimpregnated nonwoven 2. In this case the tensile strength corresponds to the yield strength. Concrete with the rolled nonwoven 2 stripes reinforcement has a flexural strength between 9 and 10 MPa. The elongation with less than 0.1% is lower than in the nonwoven examples before. The yield strength also represents the tensile strength. After the tensile strength has been reached, the curves drop almost vertically. In all three curves, stress around 4–5 MPa is again absorbed, forming a plastic deformation zone with increased elongation until 0.1%.
Use of Recycled Carbon Fibres in Textile Reinforced Concrete
547
Fig. 6. Flexural stress-elongation curves of rolled nonwoven 2 reinforced concrete. 12
Flexural Stress [MPa]
10 8 6 4 2
Sample 1 Sample 2 Sample 3
0 0
0,1
0,2
0,3
Elongation [%]
Fig. 7. Flexural stress-elongation curves of rolled nonwoven 2 stripes as reinforced concrete.
Comparing the results of nonwoven 2, the flexural stress-elongation curves of the pre-impregnated nonwoven 2 shows the highest plastic deformation zone and the yield strength is higher than the tensile strength. In the other curves the drop after reaching the tensile strength is faster and the tensile strength also represents the yield strength. The reached tensile strength of the rolled fully and stripes nonwoven 2 is a little bit higher compared to the pre-impregnated nonwoven 2.
548
V. Overhage and T. Gries
4.2 Recycled Carbon Fibre Yarn Reinforcement In this section, the results with rCF yarn reinforcement are shown. As mentioned before, two different yarns were used. Both yarns were preimpregnated with epoxy resin. The test with the individual yarns are used as preliminary tests for textile reinforcement, as no textile grid structures have yet been produced due to the small quantities of the yarn. The yarns are manually inserted into the tensile zone of the samples (Table 5). Table 5. Sample variations of the Yarn. Reinforcement
Variation
Yarn 1
Epoxy impregnated
Yarn 2
Epoxy impregnated
In the first series of tests the yarn 1, a friction spun yarn made out of rCF and PA6, were impregnated with epoxy resin. Afterwards, the impregnated yarns were inserted with a spacing of 2 cm in the mould, before the concrete mixture was inserted.
Fig. 8. Flexural stress-elongation curves of Yarn 1 reinforced concrete.
Figure 8 shows the flexural stress-elongation curves of the friction spun hybrid yarn. At the beginning, the curve shows an almost linear increase up to the initial crack, which represents the maximum of the force absorption. After reaching its maximum around 12–13 MPa, the flexural stress drops abruptly to a value around 2–6 MPa and forms a local minimum at this point. Thereafter, the stress rises to a local extremum and then falls in steps, so no directly failure occurs. Due to the gradual progression, the influence of the yarn on the strength is clearly recognizable.
Use of Recycled Carbon Fibres in Textile Reinforced Concrete
549
A comparable curve can also be seen in the following Fig. 9. The starting material is identical, but the yarn 2 was produced using the wrap yarn spinning process. At the beginning, the curve shows again an almost linear increase up to the initial crack, which represents the maximum of the force absorption. After reaching its maximum, the flexural stress drops abruptly to a value around 2–3 MPa and forms a local minimum at this point. Thereafter, the stress rises to a local extreme and then progresses in steps. The 3rd local maximum is initially higher than the 2nd local maximum in two of the three samples. But the tensile strength also represents the yield strength. The influence of the reinforcement can also be seen here.
Fig. 9. Flexural stress-elongation curves of Yarn 2 reinforced concrete.
Comparing both results of yarn reinforcement, the flexural stress-elongation curves have a similar gradient and a gradual failure is with all samples recognizable. In both cases the first peak is the highest peak and therefore the tensile strength also represents the yield strength. The curves of the yarn 2 have a higher tensile strength and a more distinctive plastic deformation zone. 4.3 Comparison of Results of Different Recycled Carbon Fibre Reinforcements Concrete specimen, reinforced with rCF showed in all cases an increase in flexural strength. Figure 10 shows the flexural stress-elongation curves of all different variations. From each variation a represented curve was selected for the comparison. The results of the nonwoven 1 showed a gradual failure. In the case of the rolled variation with three peaks around 7–8 MPa. The rolled variation reached with 8 MPa a higher tensile strength. Comparing the results of nonwoven 2, the pre-impregnated nonwoven 2 shows the highest plastic deformation zone and the yield strength is higher than the tensile strength. In the other variations the drop after reaching the tensile strength is faster and the tensile strength also represents the yield strength. The reached tensile strength of the rolled
550
V. Overhage and T. Gries
fully and stripes nonwoven 2 is with 8–10 MPa a little bit higher compared to the preimpregnated nonwoven 2. The highest tensile strength with 9–10 MPa was reached by the rolled stripes. It is noticeable that in this variation only half the amount of nonwoven was used in comparison to the other variations. In contrast the bonding of the concrete layers is presumably better, so that the nonwoven in strip form does not function as a separating layer, whereby the higher tensile strength is achieved. Comparing both results of yarn reinforcement, the flexural stress-elongation curves have a similar gradient and a gradual failure is with all samples recognizable. In both cases the tensile strength also represents the yield strength. The curves of the yarn 2 have, with around 13 MPa, a higher tensile strength and a more distinctive plastic deformation zone. 16
Concrete
Flexural Stress [MPa]
14
Nonwoven 1 preimpregnated Nonwoven 1 rolled Nonwoven 2 preimpregnated Nonwoven 2 rolled Nonwoven 2 stripes Yarn 1
12 10 8 6 4 2
Yarn 2
0 0
0,1
0,2
0,3
Elongation [%]
Fig. 10. Comparison of Flexural stress-elongation curves of of tested variations
5 Conclusion and Outlook Concrete specimen, reinforced with rCF nonwovens or rCF yarns showed an increase in flexural strength. The investigations have shown, that rCF as a nonwoven or a yarn reinforcement are increasing the tensile and yield strength of concrete. Additionally, in most cases a gradual and no direct failure occurs. Therefore, rCF Textile as reinforcement in concrete have a high potential for a more sustainable construction industry. To enable the use of recycled reinforcement materials in the future, further research is required. This includes, among other things, the separability of the materials after reuse and an adaptation of the coating materials used to the material combination is conceivable. Furthermore, the comparable and consistent quality of the recycled reinforcement materials is essential for a possible use in the future.
Use of Recycled Carbon Fibres in Textile Reinforced Concrete
551
Acknowledgments. The work was carried out under the aegis of the PhD-program “Verbund.NRW”, supported by the North-Rhine-Westphalian funding scheme “Forschungskolleg” of the ministry of culture and science. Parts of this investigation are based on the results of the master thesis from Marius Burmester, which I have supervised.
References 1. Gärtner, C.: C3 – Carbon Concrete Composite (2019). https://www.bauen-neu-denken.de/c% C2%B3-projekt/. Accessed 28 July 2022 2. Hensen, F., Kolbmüller, H., Stahr, M., Weber, J., Wild, U.: Beton- und Stahlbeton. In: Stahr, M., Hensen, F. (eds.) Bausanierung. 4., vollst. überarb. und aktualisierte Aufl. Vie-weg + Teubner, Wiesbaden (2009) 3. Meiners, D., Eversmann, B.: Recycling von Carbonfasern. In: Thomé-Kozmiensky, K.J., Goldmann, D. (eds.) Recycling und Rohstoffe. 7. TK Verlag Karl Thomé-Kozmi-ensky, Neuruppin (2014) 4. WWF Deutschland (Hrsg.): Klimaschutz in der Beton- und Zementindustrie Berlin (2019) 5. United Nations Environment Programme (Hrsg.): 2020 Global Status Report for Buildings and Construction Nairobi (Kenia) (2020) 6. Intergovernmental Panel on Climate Change (IPCC) Global Warming of 1.5 °C. - (UK). Cambridge University Press, Cambridge, New York (2018) 7. Swanson, A.: How China used more cement in 3 years than the U.S. did in the entire 20th Century The Washington Post, 24 March 2015 8. Friese, D., Scheurer, M., Hahn, L., Gries, T., Cherif, C.: Textile reinforcement structures for concrete construction applications –a review. J. Compos. Mater. 56, 1–24 (2022) 9. Kimm, M.K.: Ressourceneffizientes und recyclinggerechtes Design von Faserverbundwerkstoffen im Bauwesen-Textiltechnik/Textile Technology. Shaker Verlag, Düren (2021). ISBN: 9783844078268 10. Porta, N.W.: Sustainability and flexural behaviour of textile reinforced concrete, Thesis for the degree of licentiate of engineering thesis for the degree of licentiate of engineering Department of Civil and Environmental Engineering, Division of Structural Engineering, Concrete Structures, Chalmers University of Technology, Gothenburg, Sweden (2012) 11. Aschebrock, G.C.: Textile Reinforced Concrete: A Review, Engineering School. Universidade Federal do Rio Grande do Sul (UFRGS), Porto Alegre, Brazil (2022) 12. Haist, M., et al.: Nachhaltig konstruieren und bauen mit Beton. In: Bergmeister, K., Fingerloos, F., Wörner, J.-D. (eds.) Beton-Kalender. Ernst & Sohn, Berlin (2022) 13. Seifert, W., Lieboldt, M.: Ressourcenverbrauch im globalen Stahlbetonbau und Potenziale der Carbonbetonbauweise Beton - und Stahlbetonbau. Bd. 115 6, S. 469–478 (2020) 14. Jacob, A.: Building confidence in recycled carbon fiber (2019). https://www.compositeswo rld.com/articles/building-confidence-in-recycled-carbon-fiber. Accessed 5 Aug 2022
Valorization of Sulphidic Mine Tailings as Artificial Aggregate: Implementation in Cement-Based Materials Yury Villagran-Zaccardi , Liesbeth Horckmans , and Arne Peys(B) Sustainable Materials Unit, Flemish Institute for Technological Research (VITO), 2400 Mol, Belgium [email protected]
Abstract. Mine tailings are significant environmental liabilities worldwide. This adds up to the increasing depletion of non-renewable resources for use as construction materials. Valorization of such wastes as secondary raw materials is an effective environmental-friendly strategy. This paper presents an investigation on the valorization of low-grade sulphidic mining waste in the shape of coldgranulated artificial aggregates. The study was carried out in the framework of the NEMO project (EU Framework Programme Horizon 2020, Grant Agreement No 776846). Results reveal the efficient immobilization of heavy metals in addition to the possibility of partial replacement of aggregates in mortar mixes. The optimized granulation process, based on intensive mixing with minimal contents of cement as binder, was demonstrated in pilot production, and the artificial aggregate in sizes 0–4 mm and 0–10 mm was afterwards implemented in mortar mixes at replacement ratios of 17 and 32% of natural aggregate. Limited affectation of compressive strength and workability by the substitution with secondary aggregate was resolved through optimization of the superplasticizer dosage. The hydrated cement contained in the granules demonstrated no perceptible interference with the fluidifying action of the superplasticizer. The results reveal that despite their porous nature the granules are feasible to be used in non-structural or low-grade cement-based mixes. A notable contribution is therefore made for the valorization of the mine tailings as artificial granules with a comparable cement consumption to that required in landfilling. Keywords: Sulphidic Mine Tailing · Artificial Aggregate · Cold Granulation · Secondary Raw Materials
1 Introduction Mining of primary natural resources produces a significant volume of waste that have an important geographical footprint and the environmental liability associated with storage facilities. Tailing repurposing is one of the most needed strategies for a more eco-efficient mining industry. Cemented paste backfills, with 3–10 wt% Portland cement content, is a successful valorization method [1], but higher value alternatives deserve to be explored. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 A. J˛edrzejewska et al. (Eds.): SynerCrete 2023, RILEM Bookseries 43, pp. 552–559, 2023. https://doi.org/10.1007/978-3-031-33211-1_49
Valorization of Sulphidic Mine Tailings as Artificial Aggregate
553
The production volume of tailings (~700 Mton/year in Europe [2]) make them suitable for use as natural aggregate replacement. Their high fineness is, however, not appropriate for application as aggregate in mortar or concrete. Cold granulation techniques, which have been applied with success for producing artificial aggregates from other fine materials such as fly ash [3–5], can be implemented for the upcycling of mine tailings. Previous research [6] has optimized the production of artificial granules by cementbased cold granulation. The use of only 5 wt% of cement CEMIII/B (and 95 wt% tailings) showed to be sufficient for the production of aggregates that seem appropriate for implementation in cement-based materials. This is quite encouraging given that with this strategy similar cement contents as those used in backfilling applications can incorporate much higher specific value. Preliminary studies for the application of these granules in mortar mixes showed positive results, with effective control of leaching of heavy metals and values of 44 MPa for the compressive strength when using up to 32 wt% of granules relative to the total sand content [6]. However, given the high porosity of these granules, the main affectation was in the fresh state. The water uptake of the material significantly affected workability and additional optimization of the mixes is needed. Normally, the improvement of the consistency level in cementitious mixes with a superplasticizer depends on the compatibility of the superplasticizer and the cement type in the mix. With the inclusion of cold-granulated artificial aggregates an additional variable is introduced. As these granules contain small amounts of hydrated cement, they can interfere with the adequate fluidification action of the chemical admixture. Therefore, it is necessary to study the capability of superplasticiser to compensate the lack of workability at an affordable cost and without segregation. On this basis, the present study aims for an extended study of the optimization by the use of a superplasticizer of mortar mixes for appropriate strength and workability. Measurements included flow, at 5 and 30 min after mixing, as well as compressive strength, and porosity of the mortar samples containing 17 and 32 wt.% of granules replacing natural sand (CEN sand, with composition and particle size distribution according to EN 196-1). Such replacement would change both composition and particle size distribution is made in terms of the full particle size range of the CEN sand; so only the fraction between 1- and 2-mm size was replaced by granules.
2 Materials and Methods Mine tailing originated from the Boliden plant (Tara, Ireland). Only the fine fraction collected in the tailing pond was used in this study, the coarse fraction is used at the mine site for the production of backfill. Table 1 shows the chemical composition of tailings and CEM III/B and the mineralogical composition of the tailings. The particle size distribution (D10/50/90 ) of the mine tailings and the cement were 1.5/5.9/16.1 and 1.8/10.0/26.0 µm, respectively. Granules were produced in an Eirich R08W high intensity mixer for solids at a ratio of 6 wt.% CEM III/B 42.5N and 94 wt.% mine tailings. The solids were mixed dry, and 16 L of water per 100 kg of solids was slowly added to produce the granules. Mixing at 1500 rpm for the rotor and 25 rpm for the pan was used for homogeneous mixing of the
554
Y. Villagran-Zaccardi et al.
Cumulative percent passing
water and the powders and the formation of microgranules, followed by slower mixing speeds for the rotor (250 rpm) and higher speeds of the pan (50 rpm) for the growing of the granules. The target maximum size for application in mortar was 4 mm, with the particle size distribution presented in Fig. 1. The optimized granules were used as aggregate in mortar mixes. The mortars were optimized for maintaining the consistency and checking the flow loss over time. In addition, effects on compressive strength and porosity were verified. Aggregate packages were prepared by replacing 17 wt.% and 32 wt.% of standard CEN sand by the granules. To maintain a comparable particle size distribution, the replacement was made so CEN sand particles between 1- and 2-mm size were half or totally replaced (17 and 32 wt.% overall replacement ratios, respectively). A reference mix containing 100 wt.% CEN sand was also prepared for comparative purposes. Aggregate-to-cement ratio was 3 and water-to-cement ratio was 0.5. Mixing was done in a Hobart mixer, with the sequence specified in EN 196-1. The consistency of the mortar mixes was measured in the flow table according to EN 1015-3, immediately after finishing mixing (~5 min since the addition of water) and at 30 min after the addition of water in mixing. Before the second flow measurement, the mortar was remixed for 60 s at high speed. Specimens for the assessment of hardened properties were cast after each measurement of consistency. Casting, compaction and curing was done as specified in EN 196-1. 100 90 80 70 60 50 40 30 20 10 0 0.08
0.16 0.25
0.5
1
2 3.35 1.6 2.5 4
6.3
Sieve size (mm)
Fig. 1. Produced granules (size < 4 mm).
Table 1. Chemical compositions of mine tailings and cement and mineralogical composition of tailings. wt.%
CaO
SiO2
SO3
MgO
Al2 O3
Fe2 O3
BaO
CEM III/B
49.5
29.8
3.2
5.9
8.5
1.4
Tailings
27.6
29.0
5.6
1.8
4.7
2.7
wt.%
Others
LOI
/
1.7
/
3.3
2.7
22.6
Calcite Quartz Dolomite Microcline Albite Muscovite+illite Baryte Pyrite
Tailings 45.7
17.2
10.2
7.2
5.1
9.1
3.3
2.1
Valorization of Sulphidic Mine Tailings as Artificial Aggregate
555
Multiple mortar mixes were prepared to achieve the goal consistency level (i.e. a similar value to that of the REF mix), by adding additional water and/or superplasticizer to the mix, as indicated in Table 2. The superplasticizer (SP) was Master Glenium 51 (35% solid residue). The SP is expressed as solid % SP relative to cement. After 28 days of curing time, compressive strength and porosity from water absorption upon vacuum saturation of mortar samples were determined. Table 2. Mortar mixes with consistency optimization. ID
Correction for achieving target consistency
Granules content (wt.%)
REF
None
0
G17
None
17
G17A
+70% of water absorption capacity of granules
17
G17B
+0.06 wt.% of SP
17
G17C
+0.10 wt.% of SP
17
G32
None
32
G32A
+70% of water absorption capacity of granules
32
G32B
+0.10 wt.% of SP
32
G32C
+0.16 wt.% of SP
32
G32D
+50% of water absorption capacity of granules and 0.11 wt.% of SP
32
3 Results and Discussion The consistency of the mortars at 30 min is presented in Table 3. The corresponding flow values are presented in Fig. 2, showing the initial value and the consistency loss at 30 min after the initiation of mixing. Mixes G17C and G32D fulfil the target values for 5 and 30 min. Mixes A demonstrated insufficient fluidity reflecting the limitation of additional water to compensate for the water uptake of the granules. Results of compressive strength are presented in Table 4. A comparison between results of samples cast 5 min after mixing and 30 min after mixing shows a slight increase in the compressive strength for the mortars containing granules (the average increases are 2.179 and 3.41 MPa for mixes with 17 and 32% granules, respectively, with respective 95% confidence intervals of [0.04–4.32] and [0.62–6.20] MPa). The results of porosity of samples (Table 4) show a trend consistent with the one of strength, with average decreases in porosity of 0.93 [0.12–1.74] and 1.71 [−0.10–3.52] percentual points. It is notorious that only the REF sample without granules showed a higher compressive strength for the samples cast immediately after mixing in comparison with the samples cast with 30 min delay. The standard compaction procedure with the jolt
556
Y. Villagran-Zaccardi et al.
device is very energetic (60 jolts for each of the two layers), and this secures a good compaction for all these mixes (which were not always showing adequate workability but were always sufficiently cohesive). Then, the difference in compressive strength is not explained by different compaction degrees, but by the effect of the granules in the fresh mix during the first 30 min. As granules were used unsaturated, they immediately started absorbing water as soon as they were put in contact with the mixing water. This water uptake progresses (especially during the first 10 min [7]), and reduces the waterto-binder in the matrix. When the samples are already cast immediately after mixing, this water uptake would affect the immediate layer of matrix covering each granule particle, and produced a water-to-cement gradient around them. After 30 min, when the mortar is re-mixed, this gradient would disappear as the sample is homogenized. The remaining water uptake is more limited, and the resulting matrix seems to be more homogeneous. This second case is more representative of what happens during production, as 30 min is a relatively normal manipulation time for the mix. It is expected that in practice there would be a more significant impact of the workability of the mix on the compression strength, as the compaction energy is lower than the applied standard procedure. This Table 3. Consistency of fresh mortar mixes.
REF-30 G17-30
G17A-30
G32-30
G32A-30
G32B-30 G17B-30
G17C-30
G32C-30
G32D-30
Valorization of Sulphidic Mine Tailings as Artificial Aggregate
557
seems little relevant as the optimization of the consistency resolves the workability issues and present mixes G17C and G32D as the viable ones for upscaling in concrete mixes (i.e. similar consistency level and loss to that of REF mix).
Fig. 2. Flow values for mortar mixes.
Table 4. Compressive strength and porosity (from water absorption under vacuum) at 28 days (average ± standard deviation). Sample ID
Compressive strength (MPa)
Porosity (%)
5 min
30 min
5 min
30 min
REF
48.1 ± 2.6
45.2 ± 2.4
16.1
15.7
G17
48.9 ± 1.8
50.9 ± 3.1
17.2
15.9
G17A
43.9 ± 0.8
47.1 ± 1.1
18.3
17.7
G17B
44.0 ± 1.0
47.1 ± 1.8
19.1
17.7
G17C
45.5 ± 1.6
45.9 ± 1.8
18.2
17.9
G32
45.5 ± 1.0
45.8 ± 0.5
18.6
18.3
G32A
40.4 ± 1.1
42.7 ± 1.2
20.2
19.3
G32B
46.1 ± 0.9
52.3 ± 1.0
18.2
17.1
G32C
42.4 ± 0.9
48.3 ± 1.0
21.7
17.7
G32D
38.4 ± 1.2
41.7 ± 1.9
22.8
20.6
Thus, the porosity of the granules tends to reduce the compressive strength of the mortar, but the process is more complex than a simple volumetric effect of the porosity of the granules. Decoupling of consistency level and water content leaves only compressive strength as the aspect to put focus. The water uptake by the granules incorporated in
558
Y. Villagran-Zaccardi et al.
unsaturated state and consequent reduction of the water-to-binder ratio is limited. It seems only sufficient to maintain the level of compressive strength when granules content was 17%. This is, in G17C the water uptake compensates partially for the greater porosity of the granule particles when allowing sufficient time for it to occur and applying the remixing afterwards. G32D shows an inevitable reduction in compressive strength when maintaining the same workability level. This was the case because the fluidity loss can be improved with the use of superplasticizer up to a certain point. For the lower granule content, the action of the superplasticizer was sufficient. For the higher granule content, the use of superplasticizer had to be combined with part of the absorption water of the granules, as the full compensation with superplasticizer still led to significant flow loss over time. The granules remain a porous phase in the mortar matrix. As a result, the correlation between porosity and compressive strength among all mortar mixes is the same, irrespective of the content of granules. With this in mind, the strategy of adjusting the consistency with the superplasticizer allowed to maintain a similar consistency of the mix and compressive strength to that of the reference mortar without granules. Important is to say that the whole strategy involves the same cement demand as current practices. First, the cement content used for the production of granules is equivalent to that currently used in backfilling. Second, for 17 wt.% the cement content for equivalent compressive strength of the mortar mix was the same as in the mortar without granules. This advantageous valorisation strategy still requires to be quantified by appropriate life cycle analysis of the processing and implementation in mixes as a whole.
4 Conclusions The consistency of the mixes was efficiently optimized with no effects on the strength or porosity of the mixes. The loss of workability was efficiently controlled with the addition of a superplasticizer for granules content of 17 wt.%, and with the addition of 50% of the water uptake of the granules and a superplasticizer for granules content of 32 wt.%. The technical feasibility of the mixes was demonstrated at an affordable cost. As a result, the correlation between porosity and compressive strength among all mortar mixes is the same, irrespective of the content of granules. The strategy of adjusting the consistency with the superplasticizer allowed to maintain a similar consistency of the mix and compressive strength to that of the reference mortar without granules. The granules remain a porous phase in the mortar matrix. Thus, regardless of the amount of granules included, porosity and compressive strength correlate in all mortar mixtures. For 17 wt.% of granules, the superplasticizer was successfully used to maintain the consistency and compressive strength to match the reference mortar without granules. For 32 wt.%, the effect was only sufficient to lessen the effect. Future research concerning life cycle assessment of this type of granules applied in cement-based mixes is still required to address the impact of the mine waste when valorized with the present strategy. This assessment will necessarily include the determination of the durability performance of the mixes in different exposure conditions. Moreover, the description of the properties of the interfacial transition zone between granules and the matrix may help to explain the performance of the concrete mixes in the hardened state.
Valorization of Sulphidic Mine Tailings as Artificial Aggregate
559
Acknowledgements. This work was carried out in the framework of the NEMO project, funded by the European Union’s EU Framework Programme for Research and Innovation Horizon 2020 under Grant Agreement No 776846. Collaboration of partners within the NEMO Consortium is greatly appreciated.
References 1. Edraki, M., Baumgartl, T., Manlapig, E., Bradshaw, D., Franks, D.M., Moran, C.J.: Designing mine tailings for better environmental, social and economic outcomes: a review of alternative approaches. J. Clean. Prod. 84, 411–420 (2014) 2. European Commission, Directorate-General for Internal Market, Industry, Entrepreneurship and SMEs: Raw materials scoreboard 2018: European innovation partnership on raw materials. EC Publications Office, Brussels (2018). https://data.europa.eu/doi/https://doi.org/10. 2873/08258 3. Janev, J., Koliovski, V.: Light weight aggregates prepared by cold fly ash granulation. Bulletin of the International Association of Engineering Geology - Bulletin de l’Association Internationale de Géologie de l’Ingénieur 30(1), 411–413 (1984) 4. Arslan, H., Baykal, G.: Utilization of fly ash as engineering pellet aggregates. Environ. Geol. 50(5), 761–770 (2006) 5. Joseph, G., Ramamurthy, K.: Influence of fly ash on strength and sorption characteristics of cold-bonded fly ash aggregate concrete. Constr. Build. Mater. 23(5), 1862–1870 (2009) 6. Peys, A., et al.: Transformation of mine tailings into cement-bound aggregates for use in concrete by granulation in a high intensity mixer. J. Clean. Prod. 366, 132989 (2022) 7. Poon, C.S., Kou, S.C., Lam, L.: Influence of recycled aggregate on slump and bleeding of fresh concrete. Mater. Struct. 40, 981–988 (2007)
Influence of the Composition of Original Concrete on the Carbonated Recycled Concrete Aggregates Properties Sandrine Braymand1(B)
, Sébastien Roux2 , Hugo Mercado Mendoza1 , and Florian Schlupp1
1 ICube, Strasbourg University, CNRS, UMR 7357, Strasbourg, France
[email protected] 2 Institut Jean Lamour, Lorraine University, CNRS, UMR 7198, Nancy, France
Abstract. Recycled concrete aggregates (RCA) heterogeneity, leads to different properties compared to the natural aggregates (NA), especially in terms of their water absorption (WA24). Moreover, the variability of these RCA properties is larger than the NA one. This is mainly due to the compositions of original concrete. These disparities in properties and their high variation range limit the reuse of RCA in concrete. In the construction industry, concrete production has a significant environmental impact. Indeed, the cement production induces high greenhouse gas emissions. Accelerated carbonation of RCA can combine the advantages of a capture of CO2 issued from plant and a reduction of water absorption of aggregates. Indeed, the carbonation reaction clogs the capillary networks of aggregates and then reduces the accessible porosity that directly influences the water absorption. To reduce global carbon dioxide emissions and to enhance the recycling of RCA, the French national project FastCarb aims to optimize an accelerated carbonation process at an industrial scale. This work is to study the evolution of RCA properties, issued from several batches, following treatment in a carbonation chamber with defined parameters. This paper analyzes the influence of accelerated carbonation and its efficiency on the evolution of the RCA absorption and its variability. In order to identify the influence of the composition of original concrete, RCA with various original concrete compositions were crushed then tested. It was shown that accelerated carbonation decreases RCA absorption but not its variability. Carbonation (evaluated by mass gain) and absorption reduction efficiencies are not directly correlated. Keywords: Recycled concrete aggregate · Accelerated carbonation · Absorption · Mass gain
1 Introduction Recently, the use of recycled concrete aggregates (RCA) is a common practice in the construction industry. Obtained by crushing old concrete, RCA are composed of natural aggregates and cement paste. This composition leads to some differences in properties © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 A. J˛edrzejewska et al. (Eds.): SynerCrete 2023, RILEM Bookseries 43, pp. 560–569, 2023. https://doi.org/10.1007/978-3-031-33211-1_50
Influence of the Composition of Original Concrete
561
with natural aggregates (NA) and a wide range of variation in these properties, particularly regarding their water absorption (see Fig. 1). These differences are due to the composition of the parent concrete (initial concrete before work demolition) and limit the use of GBR in new concrete at large scale.
Fig. 1. 24h water absorption (WA24) of RCA and NA and their variations [1]
The cement industry contributes 5–7% of the global greenhouse gas emissions. In an attempt to reduce CO2 emissions and to expand the volumes of RCA used in concrete production, the FastCarb project [2] intends to store the CO2 released during cement production in RCA while improving their properties. The carbonation of concrete has been widely studied in the past as a phenomenon that can lead to the corrosion of the rebars in concrete and consequently to the degradation of concrete structures [3]. On the contrary, carbonation is now studied for its positive environmental aspect, which is the mineralization of atmospheric CO2 (in the case of natural carbonation) or of CO2 from industrial processes (in the case of accelerated carbonation) [4, 5]. The objective of the FastCarb project is to propose an accelerated carbonation process on an industrial scale [6]. The efficiency of the different accelerated carbonation techniques is highly dependent on many parameters specific to each one: duration of the treatment, total pressure and relative humidity in the chamber, partial pressure of CO2 , possible flow of gases through the chamber (if CO2 flow), possible movement of the aggregates (by rotation of the chamber for example if dynamic method), etc. [7–10]. The properties of the RCA to be treated and their conditioning before carbonation also have a major influence on the efficiency of the carbonation techniques: previous carbonation state, water saturation degree, grain size, porosity, characteristics of the parent material (nature of the cement, nature of the aggregates), etc. [7, 11]. Some research studies recommend the implementation of pre-treatment methods for recycled aggregates before the application of the carbonation process (vacuum,
562
S. Braymand et al.
water equilibrium vs controlled relative humidity, preliminary immersion in a calcium hydroxide solution, etc.) [7, 12]. These methods could increase the storage of CO2 as well as the beneficial effects on the physical and mechanical properties of the treated aggregates. The composition of the RCA is a major factor of the carbonation reaction potential, especially its cement content. Indeed, the content and type of cement define the quantity of phases that can be carbonated (portlandite Ca(OH)2 , hydrated calcium silicates C-S-H, …) [13, 14]. In the case of carbonation of concrete, the progress of the carbonation front decreases with the cement content of the concrete due to a buffer effect. Also, the rate of carbonation can be correlated with the mechanical properties of the concrete, as high strength material induces a compact microstructure limiting the reactions with the CO2 [15]. The mass gain during accelerated carbonation is an indicator that allows a qualitative evaluation of the process efficiency in some situations, e.g. when selecting the parameters of the accelerated carbonation process. Thus, Djerbi considers that the mass of CO2 fixed during the carbonation is correlated to the dry mass gain. This approach underestimates the rate of CO2 capture by 5 to 10%, but this error is assumed to be systematic for a given recycled aggregate [16]. The evaluation of the influence of mineralised CO2 during carbonation on the properties of RCA, in particular their absorption, requires specific investigation. The objective of this study is to analyze the evolution of the absorption and its ranges of variation after an accelerated carbonation treatment of RCA. The analysis includes the influence of the composition of the parent concrete of the RCA. The parameters of the accelerated carbonation treatment were defined beforehand considering an analysis of literature and a consultation between the research teams of the FastCarb project. A qualitative approach to estimate the efficiency of the treatment, i.e. the quantity of mineralized CO2 , is proposed by measuring the mass gain resulting from the carbonation. Furthermore, a pH test based on phenolphthalein pulverization was performed in order to check if the carbonation of the specimen was homogeneous after the treatment.
2 Materials and Methods 2.1 Materials RCA were obtained by crushing in November 2020 concrete produced in the laboratory between February and March 2019. They were formulated as ordinary concrete. The composition parameters are given in Table 1. Workability of concrete was controlled by slump test. Concrete were tested in compression 28 days of curing. The results are given in Table 2. Concrete were crushed using a laboratory jaw crusher and then separated into 2 granular grades 0/4 mm and 4/10 mm. In this study, only the 4/10 mm RCA were studied and then carbonated. A control of the carbonation depth of the 16 × 32 cm specimen was performed before cruising to assure that a residual potential of carbonation existed.
Influence of the Composition of Original Concrete
563
Table 1. Concrete compositions. Concrete ref Natural Aggregate origin Cement type
W/C ratio Cement content (kg/m3)
BT19-1
Rolled, silico-calcareous
CEM I-52.5
0.66
280
BT19-2
Rolled, silico-calcareous
CEM II-42.5
0.66
280
BT19-3
Rolled, silico-calcareous
CEM III-42.5 0.66
280
BT19-4
Rolled, silico-calcareous
CEM I-52.5
0.58
350
BT19-5
Rolled, silico-calcareous
CEM II-42.5
0.58
350
BT19-6
Rolled, silico-calcareous
CEM III-42.5 0.58
350
Table 2. Compressive strength results, 28 days of age, (MPa). Concrete ref
Comp. Strength
Slump class
BT19-1
34
S4
BT19-2
25
S5
BT19-3
26
S4
BT19-4
37
S3
BT19-5
34
S3
BT19-6
30
S4
2.2 Methods In order to limit damage to the cementitious matrix, the RCA were dried at 60 °C until the mass was completely stabilized, before any characterisation on RCA or RCA carbonated (RCAC), and also before carbonation tests on RCA. A pH test based on phenolphthalein pulverization was performed in order to check if the carbonation of the specimen was homogeneous after the treatment. RCA Characterization. Absorption tests (W24) were performed on non-carbonated RCA and carbonated RCA according to NF EN 1097-6 standard. For each composition and carbonation state, the absorption test was performed for three samples (150g each). Mass increase in terms of gain between non-carbonated RCA and carbonated RCA were measured. Accelerated Carbonation Test. A carbonation test was performed on the RCA with the following carbonation parameters: – Pre-humidification, applied water content: 75% of W24 (measured on an initial batch of non-carbonated RCA) – CO2 content in the chamber: 15%. – Relative humidity set point in the chamber: 60%. – Pressure in chamber: atm – Temperature set point in the chamber: 20 °C
564
S. Braymand et al.
– Carbonation time: 24 h The 6 RCA samples (BT19-1 to 6) were processed during a single carbonation test. For each composition, 1200g sample was submitted to carbonation test. Water content was controlled after the carbonation process. Evolution of the relative humidity set point in the chamber was controlled.
3 Results and Discussion Considering all the specimens the mean absorption is 4.56% ± 0.37 before carbonation (RCA) and 4.20% ± 0.60 after it (RCAC). The decrease of absorption is about 0.37 points (Absolute W24) that represents a relative value of 8% based on the initial value before carbonation (Relative W24%). The increase of the standard deviation (variability of absorption) is 0.22 points (Absolute σ W24) that represents a relative value of 60% based on the initial value before carbonation (Relative σ W24%). Detailed values by composition are given Fig. 2 and Table3.
Fig. 2. Absorption (W24) of RCA (NC) and RCAC (C).
Considering a same cement paste volume (or a same water-to-cement ratio), respectively 28% for samples BT19-1 to BT19-3 and 32% for 4 to 6 samples BT19-4 to BT19-6; in the case of RCA samples, absorption increases with clinker content and mechanical strength. Clinker contents are respectively 98% for CEM I, 86% for CEM II and 37% for CEM III. For RCA samples 4 to 6, increase of absorption could also be attributed to decrease of mechanical strength. In the case of RCAC samples, except for the BT19-5, the same conclusion can be announced. (n.b. BT19-5 C presents a high variability).
Influence of the Composition of Original Concrete
565
Considering comparison between samples BT19-1 with BT19-4, BT19-2 with BT195 and BT19-3 with BT19-6, a higher cement content with a lower water-to-cement ratio (leading to a higher cement paste volume) increases slightly absorption in the case of CEM I and CEM II. As a consequence, the increase of mechanical strength due to higher cement and paste cement content doesn’t lead to a decrease of absorption. For all the RCA, carbonation treatment decreases absorption. On contrary variability (3 repetition tests) is deteriorated for RCAC compared to RCA (Table 3). This contradiction could be explained by a modification of the pore size distribution, despite a global reduction of porosity. Table 3. Absorption and variability evolutions between RCA and RCAC Concrete ref
Absolute W24
Relative W24 (%)
Absolute σ W24
Relative σ W24 (%)
BT19-1
−0.76
−18.33
−0.13
−40.81
BT19-2
−0.25
−5.74
0.03
12.45
BT19-3
−0.27
−5.39
0.26
165.95
BT19-4
−0.36
−8.22
0.28
120.24
BT19-5
−0.75
−16.42
0.24
78.80
BT19-6
−0.08
−1.58
0.07
27.52
As a global result, carbonation leads to a decrease of absorption but to an increase of absorption variability in absolute and relative values. Considering all the specimens the mean mass gain due to carbonation is 5.92‰ ± 2.63, that represents 5.95 g/kg of RCA (Fig. 3). Gain mass seems to be independent from composition, mechanical strength and initial absorption of RCA. Regarding the evolutions of water content of the samples and of the relative humidity point in the chamber during the treatment (up to 90%), it could be considered that a non-conservation of the thermohydric equilibrium contributes to dispersing the results, especially mass gain. Indeed, the mean water content variation is 0.50 ± 0.10 points that represents a relative value of 15.12% based on the initial value before carbonation (Fig. 4). A slight correlation between mass gain and water content evolution is observed: the more the sample loss water, the less gain mass is observed. Furthermore, simultaneous carbonation of many samples issued from different compositions and initial absorptions could lead to a global hydric equilibrium in the chamber. Monitoring the efficiency of the treatment by measuring the mass gain does not allow to dissociate the evolutions of mass due to different origins: release of bound water, reduction of the amount of portlandite, formation of calcite, … Moreover, when gain mass is used to qualify the carbonation of RCA, only the carbonation of portlandite is concerned, for the other hydrates (CSH, ettringite) the water loss balance is difficult to estimate [17].
566
S. Braymand et al.
Fig. 3. Mass gain between RCA (NC) and RCAC (C).
Fig. 4. Pretreatment water content of RCA (NC) and water content of RCAC (C) at the end of the carbonation treatment.
For future tests, gain mass measurement to qualify the efficiency of a carbonation process should only be used for a comparison between RCA whose parent concrete compositions (especially the cement phase) are close [18].
Influence of the Composition of Original Concrete
567
Phenolphthalein test observations confirm that RCA samples were carbonated by the treatment. As illustrated Fig. 5. For BT19-1, crushed RCAC are higher carbonated than RCA.
Fig. 5. Phenolphthalein test of RCA (NC) and RCAC (C) - Aggregate and crushed aggregate
Part of results observed in this study are confirmed by literature. In particular, it was shown in this study that accelerated carbonation decreases RCA absorption for all compositions (parent concrete) and phenolphthalein tests confirm the carbonation of RCA after treatment. This conclusion is commonly accepted and mentioned by Torrenti and al [19]. No notable correlation between decrease of the absorption or gain mass (due to carbonation) and initial composition or properties of the samples were identified. But it is known in literature that content and type of cement define the quantity of phases that can be carbonated [13, 14]. One could expect that compositions with higher cement/clinker content should present a higher decreasing of absorption and a higher mass gain. This no correlation could be explained from one part by initial mechanical strength of RCA which were linked to clinker content and cement paste content. Indeed high strength material induces a compact microstructure limiting the reactions with the CO2 [15]. It is also important to mention that for the same cement paste content and whatever the sample state (carbonated or not), it was shown on this study that absorption is correlated to clinker content but for the same cement type none of cement paste content and compressive strength have any significant influence on absorption. Thus, it could be concluded that efficiency of accelerated carbonation cannot be directly correlated to
568
S. Braymand et al.
initial composition or to initial properties (mechanical strength and absorption) of RCA because influences of these parameters are contradictory. Furthermore, it was concluded that mass gain measurement is insufficient to quantify efficiency of carbonation process mainly because it is based on the carbonation of the only portlandite [17].
4 Conclusion and Perspectives The influence of accelerated carbonation and its efficiency on the evolution of the RCA absorption and its variability were analyzed and deepened. The main conclusions and perspectives of this study are: – It was confirmed that accelerated carbonation decreases RCA absorption and increases carbonation depth but after accelerated carbonation, absorption variability increases for almost all the samples although absorption decreases. – Initial compositions and properties or RCA could lead to opposite influences on the efficiency of accelerated carbonation process (carbonation potential and accessible porosity). – No correlation between mass gain and initial absorption were identified. The change of water content over the carbonation treatment influences the mass gain. This change is linked to the thermo-hydric equilibrium. The release of water during the carbonation of cement hydrates and the drying in the carbonation chamber (HR 60%) disrupt this equilibrium. – As a recommendation for further tests, samples of different compositions and/or different absorptions should be treated separately to avoid mutual interactions. Especially as water content is managed by initial absorption. – For the evaluation of the carbonation efficiency, measurement of CaCO3 content should be more accurate than the one of mass gain, owing to various complex origins of this gain mass. – In order to enhance the comprehension of correlation between absorption and CO2 fixation, more investigations on CaCO3 content and crystallography (calcite, aragonite, vaterite polymorphous) should be performed. Indeed, their formations are influenced by the physic-chemistry of the environment (pH, temperature, etc.). Different shapes of the three CaCO3 polymorphous crystals should lead to different “clogging” of the porous network and as a consequence to different influences on the absorption evolution. Moreover, as mentioned by Thiery [20], the clogging of porous network prevents the ions transfer and as consequence, progression of carbonation. – Current research conducted within the scope of the French national project Fastcarb should contribute to clarify several of these questions.
References 1. Déodonne, K.: Études des caractéristiques physico-chimiques des bétons de granulats recyclés et de leur impact environnemental, Strasbourg, Ph.D. thesis, Strasbourg University (2015) 2. Fastcarb National Project. https://fastcarb.fr/. Accessed 29 Nov 2022
Influence of the Composition of Original Concrete
569
3. Baroghel-Bouny, V., Carpa, B.: La durabilité des armatures et du béton d’enrobage, in La durabilité des bétons. In: Ollivier, J.P., Vichot, A. (eds.) Presse de l’Ecole Nationale des Ponts et Chaussées, Paris, pp. 303–385 (2008) 4. Fernández-Bertos, M.: A review of accelerated carbonation technology in the treatment of cement-based materials and sequestration of CO2. J. Hazard. Mater. 112(3), 193–205 (2004) 5. Jang, J.G.: Review on recent advances in CO2 utilization and sequestration technologies in cement-based materials. Constr. Build. Mater. 127, 762–773 (2016) 6. Sereng, M.: Improvement of recycled aggregates properties by means of CO2 uptake. Appl. Sci. 11(14), 1–22 (2021) 7. Zhan, B.: Experimental study of CO2 curing for enhancement of recycled aggregate properties. Constr. Build. Mater. 67(A), 3–7 (2014) 8. Brück, F.: Accelerated carbonation of waste incinerator bottom ash in a rotating drum batch reactor. J. Environ. Chem. Eng. 6(4), 5259–5268 (2018) 9. Chang, E.-E.: Accelerated carbonation of steelmaking slags in a high-gravity rotating packed bed. J. Hazard. Mater. 227–228, 97–106 (2012) 10. Simoes dos Reis, G.: Effect of the accelerated carbonation treatment on the recycled sand physico-chemical characteristics through the rolling carbonation process. J. CO2 Utilization 39, 1–12 (2020) 11. Kurdowski, W.: Cement and Concrete Chemistry, 1st edn. Springer, Heidelberg (2014) 12. Pan, G.: Effect of CO2 curing on demolition recycled fine aggregate enhance by calcium hydroxide pre-soaking. Constr. Build. Mater. 154, 810–818 (2017) 13. Mistri, A.: A review on different treatment methods for enhancing the properties of recycled aggregates for sustainable construction materials. Constr. Build. Mater. 233, 1–12 (2020) 14. Liang, C.: Utilization of CO2 curing to enhance the properties of recycled aggregate and prepared concrete: a review. Cement Concr. Compos. 105, 1–14 (2020) 15. Shah, V.: Determination of carbonation resistance of concrete through a combination of cement content and tortuosity. J. Build. Eng. 2022(60), 105176 (2022) 16. Djerbi, A.: Stockage du CO2 dans les granulats recyclés: développement des procédés de carbonatation accélérée. Acad. J. Civil Eng. 40(3), 1–19 (2022) 17. Cazagliu, B., Djerbi, A.: Méthode de détermination du CO2 piégé. In : Fast Carb Conférence: Le béton recyclé, un puits de carbone!, Paris, France, 27 Sept 2022 18. Braymand, S.: Carbonatation accélérée de granulats de béton recyclé – Évolution des propriétés selon leur classe granulaire. Acad. J. Civil Eng. 40(1), 76–79 (2022) 19. Torrenti, J.M.: The FastCarb project: taking advantage of the accelerated carbonation of recycled concrete aggregates. Case Stud. Constr. Mater. 17, e01349 (2022) 20. Thiery, M.: Carbonation kinetics of a bed a recycled concrete aggregates: a laboratory study on models materials. Cem. Concr. Res. 46, 50–65 (2013)
Evaluation of Eco-friendly Concrete Release Agents Based on Bio-Waxes Ojas Chaudhari1(B)
, Giedrius Zirgulis1
, Isra Taha2 , and Dag Tryggö2
1 Research Institutes of Sweden, Box 857, 501 15 Borås, Sweden
[email protected] 2 Seelution. AB, Backa Bergögata 5, 422 46 Hisings Backa, Sweden
Abstract. The performance of mould release agent (MRA) has direct effect on the surface properties of concrete, and this eventually affects long-term durability. Currently, the most utilized MRA are from petroleum origin. But they are not ecofriendly and causes health hazard during application. Thus, eco-friendly bio-based wax emulsions are required. They have the potential to mimic the properties of petroleum-based releasing agents and thus could apply as mould-release agent that meets market requirements which has a low health risk and low climate impact. This work examined two bio-wax based MRA (F1 and F2) and compared their performance with commercial bio-based and petroleum-based MRA. The interfacial properties between MRA and formwork (steel, sawn wood and plywood) surfaces were evaluated using contact angle and surface tension measurements. Similarly, capacity of MRA to demould the concrete from formwork (steel and plywood) was measured using controlled tensile pull-off experiments. In addition, influence of MRA on surface appearance of concrete was analysed using image analysis and visual inspection of colour uniformity. The results indicated that for plywood and steel, F2 MRA showed higher adhesion energy compared to F1 MRA suggesting F2 could have good applicability for vertical and horizontal formwork surfaces. Furthermore, F1 MRA and F2 MRA reduced adhesion for low alkaline cement concrete at steel and plywood surfaces, whereas for regular alkaline cement concrete, they effectively reduced adhesion at plywood surface compared to steel surfaces. Keywords: sustainability · calcined clay · SCMs · superplasticizer · rheology
1 Introduction An advantage of concrete as a construction material is the possibility to form components of practically any shape. In ready-mix or prefabricated process, the concrete components are constructed using suitably prepared and reusable mould [1]. The construction process involves of filling the mould with concrete. Once the concrete is placed in the mould, cement hydration continues and process of setting and hardening of concrete begins [2]. The hardened concrete usually adheres firmly to the mould and causes two undesirable effects during demoulding: damage to the surface of the demoulded concrete component or damage to the mould (formwork). Thus, in construction industry, concrete formwork © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 A. J˛edrzejewska et al. (Eds.): SynerCrete 2023, RILEM Bookseries 43, pp. 570–580, 2023. https://doi.org/10.1007/978-3-031-33211-1_51
Evaluation of Eco-friendly Concrete Release Agents
571
removal has always been a complicated part of construction operation [3]. The mould releasing agents (MRA) are used in the mould to avoid damage to concrete surface, during demoulding of concrete component. The MRA reduces interfacial friction and adhesion between concrete and the formwork, resulting good surface quality of concrete and clean formwork for reuse [1]. The common commercial MRA are petroleum based (paraffinic and naphthenic derivatives), mineral oils, greases, petroleum waxes [3]. Petroleum based MRAs such as mineral oils cause environmental pollution and health problems to construction worker. Whereas MRAs based on vegetable oils are more attractive from a sustainability perspective and are increasingly in demand by the market [3]. However, vegetable oils-based MRAs have low performance compared to mineral based MRAs [2]. The ACI report [4] published the list of MRAs based on mineral oils and gives guidelines for selecting the best release agents and it advises the method of application based on the type of the release agents. Similarly, the European Union report [5] suggested guide-line on the use of biodegradable MRAs as well as tests necessary to evaluate their performance. A significant number of studies on bio-based emulsions reported the effective application of vegetable oil as a base for the formulation for MRAs [1, 3, 5–7]. However, there is no scientific explanation related to the role of the bio wax (water-based emulsion) MRA performance in the concrete – formwork (steel, plywood) interfaces. The objective of this research was to compare the performance of bio-wax based MRA on the concrete casted in the plywood and steel mould at laboratory scale. The interfacial properties between MRAs and formwork surfaces were tested using contact angle and surface tension measurements. Also, laboratory prototype set-up was developed to analyze effect of MRA on the adhesion between concrete and formwork (plywood and steel). In addition, effect of MRA on surface appearance of concrete was analyzed using image analysis and visual inspection of color uniformity.
2 Materials 2.1 Mould Release Agent (MRA) Two bio wax based MRA (EmuForm F1 and F2) were provided by Seelution AB (Hisings Backa, Sweden). One commercial bio-based MRA and petroleum-based MRAs were used for the comparison. The details are mentioned Table 1. 2.2 Formwork Surfaces The metal (steel) and plywood formwork are extensively present in the construction industry; it serves to streamline concrete setting and enhance worksite productivity [4]. Three surfaces: two wooden (sawn wood and plywood) and one metal (steel) were selected for the testing. 2.3 Cement To assess effect of alkalinity of cement on the MRA, two types of cements (from Cementa AB., Sweden) were used in the testing. The low alkaline cement - Anläggnings cement
572
O. Chaudhari et al.
42.5N-SR3 MH/LA has alkalinity 0.48% to 0.57%. Whereas regular alkaline cement – SH 52.5R has alkalinity 0.8% to 1.1%. In addition, SH 52.5R cement has rapid hardening properties compared to Anläggnings cement. Table 1. Description and properties of the MRA Name
Description
Origin
Company
Density (g/cm3 )
Viscosity (mPa.s)
F1
Formulation 1
Biowax emulsion
Seelution
0.97
4.20
F2
Formulation 2
Biowax emulsion
Seelution
0.96
4.00
S1
Separol W200
Biological degradable synthetic oil
SIKA
0.7
4.23
M1
Mineral oil
Paraffin based
n/a
0.87
n/a
3 Methods The research was divided into two parts. 3.1 Part 1 Study of Interfacial Properties Between MRA and Formwork Surface The affinity of MRA with formwork surface was analyzed using correlating contact angle of MRA on the formwork surface and surface tension of MRA at liquid/vapor interfaces [3]. Surface tension measurements The liquid/vapor surface (γLV γLV ) tension was measured using Tensiometer Krüss K100SF at 20 °C (Fig. 1). Four beakers for surface tension measurements were cleaned in acetone, IPA, ethanol and finally MilliQ water. Initially, surface tension of MilliQ water was measured and it was observed as 72.55 ± 0.1 mN/m, thus showing that the beakers and the Wilhelmy Pt-plate were clean. Then after, surface tension of MRA was measured in triplicate and average value was considered. Contact angle measurements The contact angle (θ) of MRA to the form-work surface (steel, plywood and swan wood) was determined by a Dataphysics OCA40 instrument. A 3 μL droplet of each MRA was dispensed on the surface and a snapshot was captured after about 30 s, to ensure equilibrium conditions. The contact angles were evaluated using the OCA20 software using ellipse fitting. For each MRA, contact angle was measured for four individual droplets and average was considered. An example of the evaluation is given in Fig. 2.
Evaluation of Eco-friendly Concrete Release Agents
573
Fig. 1. Surface tension measurement device: Tensiometer Krüss K100SF.
Fig. 2. Evaluation of the contact angle using OCA20 software.
3.2 Part 2 Study of Interaction Between Concrete and Formwork Pull-off testing The interaction between concrete surface and the formwork surface (steel and plywood) applied with MRA was tested using pull-off testing. The concrete was casted using concrete mixer at water to cement (w/c) ratio of 0.57 and a gravel to sand (G/S) ratio of 0.56. To avoid any interaction of concrete additives with MRA, concrete was casted without any additional additives (superplasticizer). The mix design is given in Table 2. Table 2. Concrete mix design used in the pull-off testing Number
Components
Amount (kg/m3 )
1
Cement (Anläggnings or SH cement)
320
2
Sand natural (0–8)
1170
3
Gravel natural (8–16)
650
4
water to cement ratio
0.57
The concrete samples were casted into plywood moulds 10 × 10 × 10 cm3 size (Fig. 3a). Before the casting, MRA was applied on the bottom surface (either steel or
574
O. Chaudhari et al.
plywood) and moulds were attached to the test surface base using duct tape (Fig. 3b-c), so that the mould could not move during casting process. The side surfaces of plywood moulds were excessively coated with same MRA as for bottom surface, to eliminate adhesion of side surfaces with concrete. After casting, threaded M8 rod (150 mm length) was embedded into each concrete sample (Fig. 3b-c) and samples were cured under plastic sheet for 24 h. The samples were tested for adhesion after 24 h of casting. The pull-off testing (Fig. 4) was performed using hydraulic press machine and using mass sensor (HBM S40ac3). The test load was kept below maximum limit of mass sensor ( Anläggnings cement 320 m2 /kg) and rapid hydration that resulted in the low surface bleeding and uniform surface [9]. Regardless of cement fineness, M1 and S1 MRA showed higher surface interaction and higher number of surface pores compared to F1 and F2 MRA. The low surface porosity can decrease penetration of water and other aggressive agents (chlorides, sulfates etc.)
Evaluation of Eco-friendly Concrete Release Agents
579
Fig. 10. Pull-off testing of concrete sample from MRA coated steel mould.
into the concrete which can lead to increased durability. In addition, low surface porosity can have positive effect on the compressive strength of the concrete as it improves the bond between the aggregate and hydrated cement [10]. However, influence of surface porosity on the concrete properties is beyond the scope of this paper (see [10]] for an overview).
Fig. 11. Surface pore analysis for concrete surface after demoulded from plywood mould.
5 Conclusion The F1 and F2 bio-wax based emulsion are tested and compared with commercial product S1 and petroleum-based product M1. The comparison of the results obtained from twopart testing allows to explain the performance of MRA according to the base chemistry (bio or petroleum based) and the formwork surface. The two selected criteria, adhesion energy and pull-off energy, revealed excellent parameters to quantify the efficiency of the MRA. Based on the results, F1 and F2 MRA are recommended to use in low alkaline cement concrete at steel and plywood surfaces. In addition, F1 and F2 are recommended to use
580
O. Chaudhari et al.
with regular alkaline and rapid setting cement concrete at plywood surfaces. For steel surfaces more MRA is needed to apply to reduce the adhesion with the surface. The general conclusion obtained from this study is that the technical performance of the F1 and F2 release agents equals and sometimes surpasses the one from conventional mineral oil and commercial product. The investigation is mainly focused on initial development and testing of bio-based MRAs, and it do not include effect of concrete additives, especially superplasticizers on the performance of bio-based MRAs, however this work will be part of the future projects. The benefits in terms of environmental concerns, working conditions and lateral economic consequences are bound to tip the balance in favor of bio-based relasing agents in the near future.
References 1. Djelal, C., de Caro, P., Libessart, L., Dubois, I., Pébère, N.: Comprehension of demoulding mechanisms at the formwork/oil/concrete interface. Mater. Struct. 41, 571–581 (2008) 2. Barnat-Hunek, D., Szafraniec, M.: Influence of biodegradable release oils on the physical and mechanical properties of light-colored architectural concrete. Materials 14, 4630 (2021) 3. Libessart, L., Djelal, C., de Caro, P., Laiymani, I.: Comparative study of the tribological behaviour of emulsions and demoulding oils at the con-crete/formwork interface. Constr. Build. Mater. 239, 117826 (2020) 4. ACI, Guide to Formwork for Concrete, American Concrete Institute (2005) 5. León-Martínez, F.M., Abad-Zarate, E.F., Lagunez-Rivera, L., Cano-Barrita, P.F.D.J.: Laboratory and field performance of biodegradable release agents for hydraulic concrete Mater. Struct. 49, 2731–2748 (2016) 6. de Brito, J., dos Santos, R., Branco, F.A.: Evaluation of the technical performance of concrete vegetable oil based release agents. Mat. Struct. 33, 262–269 (2000) 7. Bouharoun, S.: Characterization of the interface between fresh concrete and formwork. J. Civ. Eng. Manag. 22, 26–37 (2016) 8. de Caro, P., Djelal, C., Libessart, L., Dubois, I.: Influence of the nature of the demoulding agent on the properties of the formwork–concrete interface. Mag. Concr. Res. 59, 141–149 (2007) 9. Higginson, E.C.: Fineness of cement: a symposium presented at the seven-ty-first annual meeting. ASTM International (1970) 10. Neville, A.M.: Properties of Concrete, vol. 4. Longman, London (1995)
Durability Characterization of Concrete Using Seashell Co-products as Aggregate Replacement Camille Martin--Cavaillé1(B) , Alexandra Bourdot1 and Rachid Bennacer1
, Nassim Sebaibi2
,
1 Université Paris-Saclay, CentraleSupélec, ENS Paris-Saclay, CNRS, LMPS - Laboratoire de
Mécanique Paris-Saclay, 91190 Gif-Sur-Yvette, France [email protected] 2 COMUE NU, Builders Ecole d’Ingénieurs, Builders Lab, 1 Rue Pierre Et Marie Curie, 14610 Epron, France
Abstract. To limit the environmental impact of the construction sector, including the need for natural resources, new materials using co-products from other industries need to be investigated. At the same time, the aquaculture industry produces a large number of seashell co-products that need to be reused or discarded. Some researches were carried out on using seashell co-products as aggregate replacement in concrete but mainly focused on the workability and mechanical properties of seashell concrete. Thus, interrogations remain on their durability properties. This paper investigates the durability properties of concrete with a high substitution rate of aggregates by seashell co-products. Concrete with the same mechanical resistance and workability was developed with a replacement ratio of aggregates by oyster shell up to 50%. Two different types of cement were investigated: a reference Portland cement (CEM I 52.5 N) and another cement with a high blast furnace slag content (CEM III 32.5N). Over six months, the evolution of concrete’s chloride resistance was studied using durability indicators such as porosity accessible to water, resistivity and apparent chloride diffusion coefficient. At the same time, the gas permeability of concrete after six months of curing was investigated under different degrees of saturation. The first results show better durability properties for concrete, including oyster shell aggregates and higher gas permeability linked with a higher porosity accessible to water. This durability improvement is increased with cement including blast furnace slag. Keywords: Biobased Concrete · Seashell Co-products · Chloride Diffusivity · Air Permeability · Resistivity
1 Introduction Seashell production originates approximately 16 million tons of waste each year in the world [1] that can be reused, discarded, or incinerated. Those wastes are mainly made of calcareous material that could be used in cementitious materials. Some works were already conducted on the use of seashell coarse aggregates in concrete [2]. Those works proved that using seashell co-products is possible even with a high substitution level © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 A. J˛edrzejewska et al. (Eds.): SynerCrete 2023, RILEM Bookseries 43, pp. 581–592, 2023. https://doi.org/10.1007/978-3-031-33211-1_52
582
C. Martin--Cavaillé et al.
but often comes with a decrease in the mechanical properties. However, as most studies mainly focused on the mechanical strength and workability of concrete produced, very few studies were conducted on the durability properties of concrete in which seashell co-products replaced coarse aggregates. Besides, as the objective is to valorise shell co-products as local resources, seashell concretes are likely to be used in a marine environment, which means it could be relevant to assess their durability against chloride ingress. The objective of this study is to characterize the durability properties of concrete samples in which 50% of coarse aggregates were replaced by crushed oyster shells used as aggregates. Several durability indicators were studied, such as porosity accessible to water, apparent chloride diffusion coefficient, resistivity and gas permeability. As durability is also highly dependent on the cement used in concrete, two types of cement were used: CEM I and CEM III.
2 Materials To define the used products, we will introduce the raw materials followed by the concrete mix design in this section. 2.1 Raw Materials Raw materials used in this study are: • • • •
Cements CEM I 52,5 N and CEM III/C 32,5 N-LH/SR PM; Silico-limestone reference aggregates with a 4/10 mm size designated by RA; Sand with a 0/2 mm granulometry; Crushed oyster shell with a 4/10 mm granulometry designated by OSA.
Oyster Shell Aggregates (OSA) and their microstructure are presented in Fig. 1. Those OSA have a particular shape as they are flat ellipsoids contrary to natural aggregates (RA), which are more rounded. OSA also have a particular microstructure which was studied in [3]: oyster shells are made of two different microstructures, a dense microstructure made of a superposition of calcite lattices and a porous microstructure made of a more random arrangement of the same lattices with a porosity of approximately 35%. This porous microstructure was characterized by Mercury Intrusion Porosimetry (MIP), and results are presented in Fig. 2: OSA bring an additional porosity between 10 nm and 10 µm to the concrete mix. The physical characteristics of raw materials are presented in Table 1 [4]. RA and OSA have very similar bulk and absolute density, but OSA have an absorption coefficient of 6.2%, which is higher than the RA one, probably due to the additional porosity described previously. 2.2 Concrete Mix Design Concretes in this study were developed for the same application: to get similar workability and compressive strength. Each formulation was made with an objective of a 10 cm slump (according to NF EN 12350-2) and a compressive strength between 30
Durability Characterization of Concrete Using Seashell Co-products
583
D
P
Fig. 1. Oyster Shell Aggregates (OSA) (left) and optical microscopy observation (right) of both phases microstructure (D = dense microstructure and P = porous microstructure)
Fig. 2. Pore size distribution of OSA
Table 1. Physical characteristics of aggregates used for concrete formulations [4] Granulometry (mm)
Bulk density (kg/m3 )
Absolute density (kg/m3 )
Absorption coefficient (%)
Sand equivalent (%)
Sand
0/2
1464
2625
1.79
92
Natural aggregate RA
4/10
1531
2677
2.85
/
Oyster shell aggregate OSA
4/10
809
2512
6.20
/
and 33 MPa at 28 days. In those formulations, the volume content of OSA achieve 50% (vt./%) of RA substitution. To ensure the ratio efficient water/cement (Weff /C) was constant between reference concrete and oyster concrete, the absorption of water by
584
C. Martin--Cavaillé et al.
OSA was compensated. Concrete formulations are designated by R or OS if they were made respectively with 100% RA or with half RA replaced by OSA, and I or III if they were made respectively with CEM I or CEM III. Also, to ensure the same compressive strength, concrete formulated with OSA was compensated in cement content. Formulations, slump value, and compressive strengths are presented in Table 2. Concretes were cast and demoulded after 3 days and then cured under endogenous conditions by being wrapped in a sheet of plastic and a sheet of aluminium. Table 2. Concrete compositions R_I
OS_I
R_III
OS_III
Binder (kg/m3 )
350.0
366.0
454.3
475.0
RA (kg/m3 )
1012.8
494.1
949.2
465.9
OSA (kg/m3 )
0.0
463.7
0.0
437.2
Sand (kg/m3 )
718.9
701.5
687.4
661.5
Efficient water (Weff )
209.0
218.5
209.0
218.5
Total water (kg/m3 )
250.1
273.3
247.7
270.2
Weff /Binder ratio
0.6
0.6
0.46
0.46
Compressive strength at 28 days (MPa)
34.7
31.8
28.5
31.0
Slump (cm)
12.5
10.1
9.8
9.5
Density at 28 days (kg/m3 )
2235
2200
2240
2205
Note: Composition with R = Reference aggregates, OS = Oyster Shells, I = CEM I, III = CEM III
3 Methods Three durability indicators were studied to assess the durability of concrete composition: porosity accessible to water, apparent chloride diffusion coefficient, and resistivity of concrete. As those parameters could evolve during the life of concrete samples, their evolution in time was assessed by measuring the parameters at 7, 28, 90, and 180 days of endogenous curing samples. Then, the gas permeability of concrete was studied, and as it is dependent on the saturation degree of concrete, it was measured within samples 180 days after endogenous curing by varying their saturation degree. 3.1 Porosity Accessible to Water Measurement Porosity accessible to water was measured according to modified NF P 18–459 [5]. Samples of concrete were cast into Ø11 × 22 cm and cut into three 5 cm slices. Those slices were imbibed in demineralized water under vacuum for 48h, and their mass was measured thanks to a hydrostatic scale and a classical scale. Samples were then dried in a 60 °C oven until mass stabilisation of the samples.
Durability Characterization of Concrete Using Seashell Co-products
585
3.2 Methods for Chloride Resistance Measurements Apparent Diffusion Coefficient of Chloride. Chloride apparent diffusion coefficient was measured according to modified XP P 18-462 [6]. Samples of same concrete (cast into Ø11 × 22 cm) where cut into three 3 cm slices. Those slices were coated with epoxy resin on the edges, imbibed in NaOH at 0.3 mol/L under vacuum for 72 h, and placed on a cell with two compartments: one with 0.3 mol/L NaOH solution and one with 0.7 mol/L NaCl solution. An electrical current of 20 V was applied for 6 h, samples were then split, and the penetration front of chloride in concrete was measured by AgNO3 pulverisation technic. Electrical Resistivity. Resistivity was measured according to the protocol defined by the Chlortest project [7]. Samples were prepared and saturated in the same way as porosimetry samples. Then, concrete slices were placed between two metallic plates with a tap water-imbibed sponge in the middle, the set being compressed by a 2 kg weight. Then, a continuous tension of 60 V was applied, and the current intensity was measured. 3.3 Method for Gas Permeability Measurements In the state of the art, various experimental protocols were identified to measure the gas permeability of concrete samples. In particular, the importance of the degree of saturation of concrete was demonstrated by [8–10]. The protocol adopted is an adaptation of the protocol performed by Carcasses and Abbas [10]. Concrete samples were cast into Ø15 × 30 cm and cut into three 5 cm slices. The edges were coated with aluminium paper, and samples were imbibed under vacuum with water for 72h. Then samples were placed in an oven at 60 °C (this temperature was chosen to avoid microstructure modifications in concrete), and a test was performed after 1 day, 3 days, 7 days, 21 days of drying and after mass stabilisation. Mass of the samples was monitored at each step to be able to estimate the average degree of saturation for which the test was performed. Before each test, samples were entirely wrapped in aluminium paper in the oven for 2 days to equilibrate the hydric gradient in the sample before the measure is taken. The preparation steps for each permeability measurement is presented in Fig. 3. Each permeability measure was taken with Cembureau test equipment in which a sample of concrete was placed in a chamber with lateral conditioning pressure: an entrance pressure was applied on one side of the sample, and the airflow going through the concrete sample was measured. Apparent permeability can be measured for each applied entrance pressure. Besides, apparent permeability decreases with the entrance pressure applied, as explained by [11]. Apparent permeability is the permeability measured for each value of applied entrance pressure, and theoretically, this apparent permeability represents the accumulation of a viscous flow and a non-viscous flow coming from the gas and is related to the average free path of gas molecules in pores. To eliminate this dependency, intrinsic permeability, which represents the permeability that would be measured if there was only a viscous flow in pore structure, can be calculated based on the Klinkenberg equation [11] Eq. (2) with K int the intrinsic permeability and K app the apparent permeability. However, this intrinsic permeability still depends on the saturation
586
C. Martin--Cavaillé et al.
degree of the sample. In this case, tests were performed with a conditioning pressure of 5 bars, different entry pressure of 0.5, 1, 1.5 and 2 bars and an atmospheric exit pressure. b (1) Kapp = Kint 1 + pmoy
Fig. 3. Pre-conditioning of samples for permeability tests
4 Results and Discussion 4.1 Porosity Accessible to Water Results Porosity accessible to water results presented in Fig. 4 reveal that the addition of OSA led to an augmentation of porosity accessible to water of concrete both on CEM I (with an increase ranging from 14 to 19% depending on the curing time) and CEM III formulations (with an increase ranging from 11 to 14% depending on the curing time). Those results are coherent with results found by [12, 13] while using another type of seashell as coarse aggregate in concrete. Several hypotheses can be formulated to explain such augmentation: this additional porosity could be due to the porosity included inside the seashell itself that was described in [3], but it could also be introduced by the mixing procedure during which the seashell aggregates would play an air-entraining role, or it could be a result of the compacity arrangement of aggregates being modified by the presence of oyster shell. Also, the porosity levels of those concretes are quite high, and their potential durability would have been classified as very weak based on [14]. Finally, results show that the porosity decreases for R_I and OS_I concrete during the first three months after cast and increase after 6 months. For R_III and OS_III, porosity decreases over time which could be due to the evolution of the microstructure in time. Coherently with observations made by [15–17], porosity mostly changes during the first days of hydration and then its evolution is very slow and is higher with slag additions in cement.
Durability Characterization of Concrete Using Seashell Co-products
587
Fig. 4. Porosity accessible to water upon time for concrete samples
4.2 Apparent Chloride Diffusion and Resistivity Results The apparent chloride diffusion coefficient results presented in Fig. 5 first reveal that the apparent diffusion coefficient of R_III and OS_III is much lower than that of R_I and OS_I (by 74 to 89%, depending on the curing time). This is coherent as slag have a higher binding capacity than CEM I due to the presence of high alumina content in slag, which can form more Friedel’s salt in CEM III concrete [18]. The apparent chloride diffusion coefficient obtained would give the concrete formulations potential durability classified as weak for R_I and OS_I and as high for R_III and OS_III at 6 months based on AFGC [14]. Also, the OS_I and OS_III have respectively lower chloride diffusion coefficients than R_I and R_III, which is a positive aspect in terms of durability. This decrease is from 3% to 15% for CEM I and 11% to 60% for CEM III, depending on the curing time. Several hypotheses could explain this phenomenon: either it is entirely due to the different levels of cements in seashell and oyster formulation, or the presence of oyster shells could also play a role. For example, [19] suggested that the particular shape of seashell aggregates can make a barrier to diffusion. Also, oyster shell aggregates are made of calcite that can provide additional nucleation sites for C-S-H at the surface of the aggregates, leading to a potentially modified microstructure around the OSA [20]. For CEM I, the evolution in time shows that this coefficient tends to decrease in time, particularly during the first three months. This could be coherent with the porosity decreasing in parallel. There is a stabilisation of this evolution after 3 months for OS_I and a slight increase for R_I. Nevertheless, a larger standard deviation is observed for the measurements made on R_I at 6 months, which is higher than the variation between measures over time. For OS_III and R_III, measures show that the diffusion coefficients also tend to decrease in time but as chloride diffuse less during the experiment time, the front penetration of chloride is only of few millimetres and the standard deviations on the test on OS_III and R_III are much higher. Those results can be completed with resistivity results obtained in Fig. 6, as [21] showed that resistivity was inversely proportional to the chloride diffusion coefficient independently of the type of cement used. Based on the results presented, the resistivity of OS_III and R_III concrete is much higher than OS_I and R_I (between 7 and 10 times more, depending on the measurement curing time). Resistivity would be classified as very
588
C. Martin--Cavaillé et al.
Fig. 5. Apparent chloride diffusion coefficient evolution upon time for concrete samples with CEM I (left) and CEM III (right)
weak for OS_I and R_I and high for OS_III and R_III based on [14]. However, the effect of aggregates replacement by oyster shells is less visible in OS_I resistivity compared to R_I results as the resistivity value is quite low in CEM I concrete (the impact of oyster shell varies from −3% to + 9%). In CEM III, after 28 days, the resistivity of R_III is a bit higher than the resistivity of OS_III by approximately 18%. The evolution over time of resistivity is coherent as it is increasing in CEM I concrete and R_III coherently with [22] study made on slag concrete with ordinary aggregates. For OS_III, resistivity increases in the first three months and slightly decreases at 6 months (which needs to be balanced by the high standard deviation at 3 and 6 months).
Fig. 6. Resistivity evolution upon time for concrete samples with CEM I (left) and CEM III (right)
4.3 Gas Permeability Gas permeability evolution based on the estimated degree of hydration and entrance pressure applied are presented in Fig. 7. First, values obtained are quite high. However, they are coherent with values obtained by [23] with an ordinary concrete sample with
Durability Characterization of Concrete Using Seashell Co-products
589
similar cement quantities. Similar Weff /C with CEM I. AFGC classification would lead to very weak durability potential [14] which is coherent with what was found for porosity levels. The evolution of apparent permeability with the pressure applied is coherent with Klinkenberg equation Eq. (2), so the intrinsic permeability of concrete in each concrete mix could be calculated according to its saturation level.
Fig. 7. Evolution of apparent permeability and intrinsic permeability of four formulations of concrete according to estimated saturation degree and entrance pressure
To compare the evolution of permeability in different formulations, the apparent permeabilities obtained for a pressure of 2 bars and the intrinsic permeability calculated are presented in Fig. 8. For apparent permeability with a 2 bars pressure, the permeability of OS_I is from 10% to 21% higher than R_I and from 3% to 17% higher for OS_III compared to R_III. This is coherent with porosity accessible to water results presented before. However, R_III permeability is higher than R_I, and OS_I permeability is higher than OS_III. This discrepancy could be due to the fact that permeability is not only linked to the total amount of porosity in concrete but also the pore size distribution. The study by Zhang [24] revealed that the gas permeability coefficient increases with the amount of pore size within 10–1000 nm and 100–1000 nm. This range of porosity corresponds to the additional porosity brought by OSA, as shown in Fig. 2. Zhang [23] also showed that the presence of admixture could change the link between permeability and pore size distribution in concrete. For intrinsic permeability, results show that it increases for OS_I compared to R_I (from 40% to 81% depending on the saturation degree), but the effect in OS_III compared to R_III is not as visible (from −13% to +14%): as intrinsic permeability is calculated from other measurements by taking the coordinate at origin on the Klinkenberg equation, the low precision of the measure could explain those observations. Besides, the apparent and intrinsic permeability of concrete increases while the degree of saturation of concrete decreases for a degree of saturation lower than 60%. However, permeability variations between high level of saturation and low degree of saturation are relatively low (from 2 to 11%) compared to what was found in studies on classical concrete [8], where augmentation could be up to 100 times. Also, when
590
C. Martin--Cavaillé et al.
concrete is drying, the biggest pores are emptied first, followed by the smaller pores. In this study, as drying was controlled by oven temperature of 60 °C and not by relative humidity (varying between 10% and 20% at the final stage of drying), small pores of concrete samples might still be saturated, which would explain why the permeability is not as high as expected when the sample is considered as dry. This could also be explained by the fact that drying would only occur on the surface and thus the permeability would be more controlled by the core which remains practically saturated.
Fig. 8. Comparison of permeability measured for concretes for an entrance pressure of 2bars (left) and for the intrinsic permeability (right)
5 Conclusion This study characterised the durability of concrete with similar workability and mechanical resistance at 28 days and with 50% of aggregates being replaced by crushed oyster shells. Results showed that oyster shell concretes have a higher porosity accessible to water coherently with a higher gas permeability. Those results are coherent because of oyster shell porous microstructure. Those formulations also have a lower chloride diffusion coefficient and a similar resistivity for CEM I and slightly inferior resistivity for CEM III. Potential interactions between oyster shells and cements might explain this difference and a better durability. Thus, besides an increase in porosity, the potential durability of oyster shell concrete relating to chloride diffusion is better with oyster shell aggregates. Further investigations will be performed, including microstructure characterization of concrete over time, chloride binding and microscopic observations to study the interfacial transition zone between oyster aggregate and the cement matrix. Impact of oyster shell and cement compensation will be distinguished by studying concretes in which the overall microstructure is kept similar. Acknowledgements. The authors thank the AUGC (Association Universitaire de Génie Civil), the French civil engineering academic association, for its contribution to funding the travel and registration fees of the author to the Synecrete’23 conference.
Durability Characterization of Concrete Using Seashell Co-products
591
References 1. The State of World Fisheries and Aquaculture 2016 - Contributing to food security and nutrition for all, Rome (2016) 2. Mo, K.H., Alengaram, U.J., Jumaat, M.Z., Lee, S.C., Goh, W.I., Yuen, C.W.: Recycling of seashell waste in concrete: a review. Constr. Build. Mater. 162, 751–764 (2018). https://doi. org/10.1016/j.conbuildmat.2017.12.009 3. Martin--Cavaillé, C., Bourdot, A., Sebaibi, N., Bennacer, R.: Microscopic and chemical characterization of seashell co-products for valorization in cementitious materials. In: 40ème rencontres universitaires de Génie Civl, pp. 383–386. Lille, France (2022) 4. Bourdot, A., Martin-Cavaillé, C., Vachet, M., Honorio, T., Sebaibi, N., Bennacer, R.: Microstructure and durability properties of concretes based on oyster shell co-products. Presented at the 75th Rilem Annual Week and International Conference on Advances in Sustainable Construction Materials and Structures, Merida, Mexico, August 2021 5. NF P 18-459 : Béton — Essai pour béton durci — Essai de porosité et de masse volumique (2022) 6. XP P18-462: Essai sur béton durci - Essai accéléré de migration des ions chlorure en régime non-stationnaire - Détermination du coefficient de diffusion apparent des ions chlorure AFNOR (2012) 7. Tang, L.: Guideline for practical use of methods for testing the resistance of concrete to chloride ingress, CHLORTEST - EU funded research project “Resistance of concrete to chloride ingress - from laboratory tests to in-field performance” G6RD-CT-2002-00855 (2005) 8. Abbas, A., Carcasses, M., Ollivier, J.-P.: Gas permeability of concrete in relation to its degree of saturation. Mat. Struct. 32, 3–8 (1999). https://doi.org/10.1007/BF02480405 9. Sanjuán, M.A., Muñoz-Martialay, R.: Oven-drying as a preconditioning method for air permeability test on concrete. Mater. Lett. 27, 263–268 (1996). https://doi.org/10.1016/0167577X(95)00283-9 10. Carcasses, M., Abbas, A.: An optimised preconditioning procedure for gas permeability measurement, 6 (2002) 11. Klinkenberg, J.: The permeability of porous media to liquids and gases. Am. Petrol. Inst. Drill. Prod. Pract. 2, 200–213 (1941) 12. Khankhaje, E., et al.: Properties of quiet pervious concrete containing oil palm kernel shell and cockleshell. Appl. Acoust. 122, 113–120 (2017). https://doi.org/10.1016/j.apacoust.2017. 02.014 13. Sugiyama, M.: Freeze-thaw test results of porous concrete with crushed scallop shell material added. J. Hokkai-Gakuyen Univ. 120 (2004) 14. Conception des bétons pour une durée de vie donnée des ouvrages - Maîtrise de la durabilité vis-à-vis de la corrosion des armatures et de l’alcali-réaction. Association Française de Génie Civil (2004) 15. Roy, D.M.: National Research Council: Concrete Microstructure /D. M. Roy, Washington (1993) 16. Distler, P., Kropp, J.: Effect of ageing on pore structure and permeability of cementitious materials. In: Jennings, H., Kropp, J., Scrivener, K. (eds.) The Modelling of Microstructure and its Potential for Studying Transport Properties and Durability, pp. 339–350. Springer, Dordrecht (1996). https://doi.org/10.1007/978-94-015-8646-7_17 17. Wioletta, S., Goerget, F., Maraghechi, H., Scrivener, K.: Evolution of microstructural changes in cement paste during environmental drying. Cem. Concr. Res. 134, 106093 (2020). https:// doi.org/10.1016/j.cemconres.2020.106093 18. Yuan, Q., Shi, C., De Schutter, G., Audenaert, K., Deng, D.: Chloride binding of cementbased materials subjected to external chloride environment – a review. Constr. Build. Mater. 23, 1–13 (2009). https://doi.org/10.1016/j.conbuildmat.2008.02.004
592
C. Martin--Cavaillé et al.
19. Martínez-García, C., González-Fonteboa, B., Martínez-Abella, F., Carro-López, D.: Performance of mussel shell as aggregate in plain concrete. Constr. Build. Mater. 139, 570–583 (2017). https://doi.org/10.1016/j.conbuildmat.2016.09.091 20. Wang, D., Shi, C., Farzadnia, N., Shi, Z., Jia, H., Ou, Z.: A review on use of limestone powder in cement-based materials: mechanism, hydration and microstructures. Constr. Build. Mater. 181, 659–672 (2018). https://doi.org/10.1016/j.conbuildmat.2018.06.075 21. Use of electrical resistivity as an indicator for durability. Constr. Build. Mater. 73, 434–441 (2014). https://doi.org/10.1016/j.conbuildmat.2014.09.077 22. van Noort, R., Hunger, M., Spiesz, P.: Long-term chloride migration coefficient in slag cementbased concrete and resistivity as an alternative test method. Constr. Build. Mater. 115, 746–759 (2016). https://doi.org/10.1016/j.conbuildmat.2016.04.054 23. Zhang, J., Bian, F., Zhang, Y., Fang, Z., Fu, C., Guo, J.: Effect of pore structures on gas permeability and chloride diffusivity of concrete. Constr. Build. Mater. 163, 402–413 (2018). https://doi.org/10.1016/j.conbuildmat.2017.12.111
Production Waste Fibres as a Sustainable Alternative for Strengthening Cementitious Composites Ana Bariˇcevi´c(B)
, Katarina Didulica , Branka Mrduljaš , and Antonija Oceli´c
Faculty of Civil Engineering, University of Zagreb, 10 000 Zagreb, Croatia [email protected]
Abstract. The considerable development of the textile industry contributes to the fact that the production waste generated in the manufacture of yarns and fabrics is constantly increasing. These are so-called “pure wastes”, that, as a rule, never reach the consumer. Examples of such waste are weaving machine waste, products with a manufacturing defect, etc. Surveys in the technical textile industry have shown that only a few manufacturers use these wastes in their production, while they mostly end up in landfills. In Croatia alone, 300 tons of these wastes are generated annually, which indicates that the available quantities are sufficient for use in the construction sector and thus represent a valuable resource for the production of construction materials. High-quality fibres can be obtained from these wastes, which can be used as reinforcement in cementitious composites. However, it is critical to analyse the potential waste fibre streams, characterize the available production waste fibres to determine any initial flaws, the surface treatment processes used in relation to their original purpose, and finally the resistance to the alkaline environment. This paper provides an overview of available production waste fibres, their properties, and challenges for future use in cementitious composites. Keywords: Inorganic fibres · Waste streams · Durability of waste fibres · Fibre reinforced cementitious composites
1 Introduction Cracks in concrete occur during construction, but also during service life. Their occurrence and width can be reduced by using fibres that have the ability to bridge cracks and delay failure, i.e., contribute to the ductility of cementitious materials [1]. Various types of fibres are available on the market, and their use in cementitious composites is well known. The use of polymer fibres reduces shrinkage [2] while the use of carbon fibres helps to improve tensile strength and toughness [3]. The use of glass and basalt fibres is limited due to their low resistance to the alkaline environment [4], but can also contribute to composite properties. The high cost of industrially produced fibres limits their everyday use. Therefore, locally available waste fibres, which would otherwise have to be disposed of in a landfill © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 A. J˛edrzejewska et al. (Eds.): SynerCrete 2023, RILEM Bookseries 43, pp. 593–603, 2023. https://doi.org/10.1007/978-3-031-33211-1_53
594
A. Bariˇcevi´c et al.
or incinerated as part of the energy process, provide a sustainable alternative. Waste fibres are generated as production waste or after the recycling process of various polymer composites [5, 6]. The potential of recycled fibres depends on the process. Mechanical recycling usually results in damage to the surface, while the properties of the fibres after heat treatment are reduced, but their use is still possible. Production waste fibres are the clean residues from production that have no further use. Their great potential is recognised because they do not require heat treatment and have retained most of their original properties. However, before using waste fibres, it is necessary to analyse the potential waste streams, define the market demand and availability of sufficient quantities of waste fibres, and study their short and long-term effects on the properties of cementitious composites. The aim of this paper is to characterize two types of waste fibres: a) production waste fibres obtained from production of high-value technical textiles (alkali-resistant glass AR-GF, basalt - BF, carbon - CF) and b) recycled tire polymer fibres (RTPF) produced by mechanical recycling of scrap tires.
2 Waste Fibre Streams Potential industries of interest were analysed to identify local sources of fibre waste. The surveys aimed to collect qualitative and quantitative data on fibre waste streams in the region. The main criterion was location, and only plants close to Croatia were considered. In the end, 10 potential waste generators were identified. The results of the conducted surveys show that the average amount of production waste is 10% of the mass of raw material used. These data vary depending on the manufacturer, production process and final product. Production waste is mostly the ends of roving, netting, etc. This waste is stored in containers and bags or scattered in a dry state. It is disposed of as hazardous waste (carbon) or landfilled as waste from unprocessed textile fibres, while RTPF is used as fuel in cement furnaces, and causes air, soil and water pollution. The cost of disposal varies, but of particular concern is that a very valuable raw material is being disposed of while other factories are simultaneously producing new fibres to reinforce building materials. Even though the lockdown and development in the East have disrupted the growth of several industries that use fibre, the market continues to grow. The global fibre reinforced concrete market was valued at USD 2.13 billion in 2021 and is expected to reach USD 3.86 billion by 2030, growing at a compound annual growth rate (CAGR) of 6.82% from 2022 to 2030 [7]. The global carbon fibre market is expected to reach USD 8.9 billion by 2023, at a CAGR of 8.6% from 2020 to 2031, with the PAN- based carbon fibre segment dominating the carbon fibre market. The construction carbon fibre market is expected to reach USD 531.5 million by 2027, at a CAGR of 10% from USD 330 million in 2022 [8]. Another indicator of the demand for these fibres is the development of the global recycled carbon fibre market, which is expected to reach USD 222 million by 2026, at a CAGR of 12% between 2021 and 2026 [9]. The market for glass fibre reinforced concrete was valued at USD 1.83 billion in 2017 and is expected to reach USD 3.32 billion by 2023, at a CAGR of 10.5% by 2023 [10]. When used in cement composites, the most commonly used glass fibres are AR-GF, i.e., alkali-resistant fibres.
Production Waste Fibres as a Sustainable Alternative
595
At the same time, the global polypropylene fibre market more than USD 3.5 billion in 2020 and is expected to grow significantly at a CAGR of more than 3% from 2021 to 2027. The largest share of polypropylene production in 2017, was used for the production of packaging, 49%, while the share of use in construction is 7% [11], representing more than 5 million tonnes of polypropylene. Every year, 1 billion waste tyres are generated, but the waste tyre recycling industry only processes 100 million tyres per year, which is 10% [12]. Mechanical recycling generates about 10% polymer fibres, which means that 10 million tonnes of polymer fibres are available in the secondary market worldwide every year. Finally, the amount of production waste is lower for basalt fibres than for the other fibres because a larger portion is reused in production. Respondents use basalt with SiO2 content of 45–60%, which puts it in the category of acid basalts, the only ones that meet the requirements for fibre production. A high content of silica is necessary to obtain an atomic network as in glass, i.e. an amorphous structure [13]. If only 10% of the total amount of fibre mentioned above ends up as waste, there are significant amounts of fibres production waste in the construction sector alone. The survey showed that on the territory of the Republic of Croatia alone, an average of 327 tonnes of production waste fibres from the production of high-quality construction textiles and 2,500 tonnes of polymer fibres from the mechanical recycling of used tyres are generated annually.
3 Characterization of Waste Fibres Waste fibres from the production of high-quality technical textiles occur in two forms: a) direct roving, i.e. the end of the winding, and b) selvages with a width of 40 to 60 mm, consisting of roving and thread (Fig. 1). Such fibres have a high degree of purity, but it is necessary to design the processing of the fibres to achieve the desired geometric properties. RTPF fibres, on the other hand, are extremely short and further processing in the form of cutting is neither necessary nor possible. Thus, mortar mixes reinforced with RTPF showed limited post-crack bridging capacity [14]. However, RTPF fibres are heavily contaminated with residual rubber particles, which can affect the properties of the concrete [15], as an improvement of dynamic compressive [16] and tensile properties [17]. A typical sample of the so-called mixed RTPF fibres consists of 15% pure RTPF, 20% rubber-contaminated RTPF, and 65% rubber particles contaminated with very fine fibres [18]. The properties given in manufacturers’ technical data sheets are listed in Table 1 [19–21]. The above fibres are supplied in bundles and the number of individual fibres in the bundle is defined by the ‘number of filaments’. The linear density property, i.e. mass in the longitudinal direction, is defined for a fibre bundle, while the diameter is defined for a filament, i.e. a single fibre. Recycled tire polymer fibres do not have declared properties, but the expected values are known from the literature, Table 1 [16, 22]. 3.1 Methods Waste fibre characterization includes determination of diameter (HRN ISO 137:2016) and length distribution (HRN ISO 6989:2003) using the projection microscope method
596
A. Bariˇcevi´c et al.
a)
b)
c)
d)
Fig. 1. As received production waste fibres available in Croatia: a) end of roving (CF), b) end of nets (AR-GF), c) end of nets (BF), d) RTPF Table 1. Properties of production waste fibres (data from technical data sheets given by producer) Fibre type
AR-GF
BF
CF
RTPF
Number of filaments (yarn)
1 500
1 000
47 000
-
Density (g/cm3 )
2.68
2.67
1.77
0.96–1.16
Linear density (tex)
1 200
600
3 200
-
Sizing (%, type)
0.80 (aminosilane)
>0.40 (silane)
0.95–1.45 (PU)
-
Filament diameter (µm)
19
13
7
8–38
Length (mm)
Variable
Variable
Variable
8–20
Tensile strength (MPa) >1000
>1750
4300
0.1–0.475
Modulus of elasticity (GPa)
-
250
-
72
and individual measurements. The determination of the diameter of waste fibres was performed on a sample of 100 individual fibres, while a sample of 500 individual fibres was used for the length. Tensile strength and modulus of elasticity of the fibres were determined according to ISO 5079:2020 (Table 4). The tests were performed on a sample of 50 fibres. The loading was carried out in two phases: a) prestressing and b) loading at
Production Waste Fibres as a Sustainable Alternative
597
a rate of 10 mm/min. Due to the limited length of the RTPF, they were not tested, as the test clamp spacing requires a fibre sample length of 20 mm or more. The morphology of the fibre surface was determined using a scanning electron microscope (SEM). 3.2 Results and Discussions Surface Morphology Scanning electron micrographs (Fig. 2) show that carbon fibres have the highest surface roughness, which may indicate a better mechanism for transmitting frictional forces in cementitious composites. Although AR-GF, BF, and CF fibres have a protective coating on their surface, only CF fibres exhibit a uniformly grooved surface, which can be attributed to the fibre structure and the application of the coating [23], while small “particles” on the surface, which are particularly pronounced in AR-GF and BF fibres, may be caused by an uneven coating of the sizing [24]. This is consistent with previous studies [25] showing that surface morphology is affected by the type of protective coating. According to the manufacturer’s technical data sheet [19–21], CF fibres have a coating based on PU resins, while AR-GF and BF are based on silane.
a)
b)
c)
d)
Fig. 2. SEM micrographs: a) carbon fibres, b) alkali-resistant glass fibres, c) basalt fibres, d) recycled tire polymer fibres
Geometric Properties of Waste Fibres Two key parameters that ensure the contribution of fibres to stress transfer in cement composites are fibre diameter and length. Production waste glass, basalt, and carbon
598
A. Bariˇcevi´c et al.
fibres occur in lengths that are not optimal for use in cement composites. Additional processing is needed to prepare fibres with required length. Therefore, their length was not determined in the present characterization. RTPF fibres produced by mechanical recycling of scrap tires are very short and further processing in terms of length change is not possible. For this reasons, a test was conducted to determine the length range of RTPF fibres and the results are shown in Table 2. RTPF fibres are variable in length, ranging from 1 to 28 mm, with a mean value of 8 mm and a coefficient of variation equal to 58%, Table 2. In addition to length, diameter is also an important parameter for fibres, Table 3 the diameter of AR-GF ranges from 14.0 to 24.0 µm (xs = 20.4 ± 0.47 µm), BF from 12.0 to 22.0 µm (xs = 17.2 ± 0.41 µm), CF from 6.0 to 8.0 µm (xs = 7.1 ± 0.20 µm), and from 8.0 to 32.0 µm (xs = 18.0 ± 0.45 µm) for RTPF fibres. The greatest variability in the results, as expected, occurred for the RTPF fibres. The diameter of the fibres is consistent with the values reported in the technical data sheet and is within the limits of the diameter values of the most commonly used factory-produced fibres. Table 2. Length distribution of recycled tire polymer fibres Number of specimens, n
Length distribution, Average value, xmin - xmax (mm) xs (mm)
Coefficient of variation (%)
95% confidence interval (%)
500
1.0–28.0
58.0
5.1
8.0
Table 3. Diameter of production waste fibres and RTPF Fibre type
Number of specimens, n
Diameter distribution, xmin - xmax (µm)
Average value, xs (mm)
Coefficient of variation (%)
95% confidence interval (%)
AR-GF
100
14.0–24.0
20.4
11.6
2.3
BF
100
12.0–22.0
17.2
12.1
2.4
CF
100
6.0–8.0
7.1
14.0
2.8
RTPF
500
8.0–32.0
18.0
28.4
2.5
The further use of waste fibres in cementitious composites strongly depends on their geometric properties. From previous studies, the use of macrofibres (l > 30 mm) has a greater effect on the strength and toughness of the material, while the use of shorter microfibres (l ≤ 30 mm) has a greater effect on the shrinkage behaviour [26]. In accordance with the standard HRN EN 14889:2013, all fibres considered in this study belong to the category of microfibres (d < 0.3 mm). However, it should be noted that these fibres are in bundles and the average area of the bundle is 0.52 mm2 for AR-GF, 0.23 mm2 for BF, and 12.2 mm2 for CF.
Production Waste Fibres as a Sustainable Alternative
599
Mechanical Properties Tensile strength and modulus of elasticity, together with geometric properties, are the main requirements for the fibre use in cementitious composites. The relationship between the properties of the fibres and the cement matrix can be observed in two cases: at an early age, when the strength and modulus of elasticity of the matrix are still low, and at a later age, when the matrix properties improve due to curing. The results of testing the tensile strength and modulus of elasticity of the waste fibres are shown in Table 4. Table 4. Tensile strength and modulus of elasticity of waste fibres Fibre type
Elongation at break (%)
Tensile strength (MPa)
Modulus of elasticity (GPa)
AR-GF
3.03 ± 0.12
1849.73 ± 72.14
61 ± 2.4
BF
3.08 ± 0.16
2063.11 ± 105.21
67 ± 3.4
CF
0.63 ± 0.04
828.36 ± 72.6
RTPF
*tests cannot be carried out due to insufficient fibre length
132.5 ± 11.5
The test yielded values indicating that basalt fibres had the highest tensile strength with an average value of 2063.11 ± 105.21 MPa, followed by AR-GF with an average tensile strength of 1849.73 ± 72.14 MPa and carbon fibres with a tensile strength of 828.36 ± 72.6 MPa. In terms of mechanical properties, differences between the waste fibres and the properties given in the technical data sheet are to be expected due to weaving and processing, as well as the declaration of the yarn’s properties. The tensile strength values for glass fibres are about 10% higher, while the values for basalt and carbon fibres are lower compared to the values given in the technical data sheet. The largest difference is observed for carbon fibres, namely a reduction of about 80%. However, it is not clear from the technical data sheets whether the specified properties were obtained for filaments or yarns. When comparing different fibres, the size of the diameter of each fibre should be taken into account. CF has a diameter 2.4 and 2.9 times smaller than basalt and glass fibres, respectively. The modulus of elasticity shows that carbon fibres have the highest modulus of elasticity and it is 132.5 ± 11.5 MPa, while the value of modulus of elasticity for ARGF and BF is similar and it is 61 ± 2.4 MPa and 67 ± 3.4 MPa, respectively. The values of modulus of elasticity for all three types of fibres are lower compared to the values given in the technical data sheet: 15%, 22% and 47% for glass, basalt and carbon fibres, respectively. 3.3 Discussion During production fibres are subjected to strict quality control, but after cutting and processing in the production process of technical textiles, their properties deteriorate. This deterioration depends on the type of fibre, sizing and production processing, so these differences lead to inconsistent quality of waste fibres. The so-called protective
600
A. Bariˇcevi´c et al.
coating, i.e., the “sizing” on the fibre surface plays a key role in the final properties of the fibre reinforced composite. It is responsible for the improved adhesion between the matrix and the fibres and determines the mechanical properties, chemical resistance, and thermal stability of the fibres [27, 28]. Production waste fibres have sizing optimal for their further use in matrix resin and it is therefore questionable how these sizing perform in alkaline environments such as concrete. Exposure of the protective coating to an alkaline environment, may have a negative effect on its properties, i.e., degradation of the coating occurs, but it may also contribute to inhibition of fibre corrosion [29]. Mechanical deterioration of the sizing is also possible during the production process, leading to weak points and accelerating fibre corrosion, but also having negative impact on the residual properties. The stability of sizing is one of the key characteristics that should be considered. In order to compare fibres manufactured exclusively for use in cement composites with waste fibres, a brief review of the literature on the properties of factory-produced fibres was conducted, Table 5. It is clear that both factory-produced and waste fibres have similar geometric and mechanical properties. Depending on the production process and intended use, yarns (individual sizing coated filaments) can be manufactured to disperse back into their filament form (micro fibre) upon contact with water, or they can be manufactured to remain in their integral yarn form (macro fibre). Table 5. Literature review on the properties of factory-produced fibres [26, 30, 31] Fibre type Length (mm) Diameter (µm) Tensile Modulus of Elongation at strength (GPa) elasticity (GPa) break (%) CF
3–10
5–10
1.0–7.0
150–820
0.5–3.2
BF
9–30
6–20
0.87–4.8
40–115
2.4–3.15
AR-GF
6–30
14–20
1–3.5
72–80
2–4
RTPF
8–19
8–38
0.10–0.475
2.1–3.5
-
Polymer
6–12
10–60
0.3–0.7
3–40
3–60
The inconsistent quality of waste fibres is the main challenge for their use in the construction industry. Processing technology, i.e., cutting the fibres to the desired length, is required to prepare production waste fibres and, together with the lack of reliable integration technology, is the main obstacle to the uniform distribution of waste fibres in hardened cementitious composites. At the same time, rubber contamination is the main challenge for the further use of RTPF. These can be partially solved by using a cleaning technique developed at the University of Zagreb [22]. However, although this technique reduces the impurities, some rubber is still present after cleaning and its content is difficult to quantify.
Production Waste Fibres as a Sustainable Alternative
601
4 Conclusions Waste fibre characterization is performed on selected waste fibres classified as alkaliresistant glass (AR-GF), basalt (BF), carbon (CF), and recycled tire polymer fibres (RTPF). Glass, basalt and carbon fibres are certified by the manufacturer and their properties are specified in the technical data sheets. Although there are no major temperature variations, the fibres are damaged during processing and cutting in the weaving machine, which affects the residual properties of the waste fibres. This is reflected in lower tensile strength and modulus of elasticity compared to manufacturers’ technical data sheets. At the same time, recycled polymer fibres obtained by mechanical recycling of waste tyres contain rubber particles, impurities and dust in addition to the fibres. Their geometric properties are not controlled during the production process, as they are a by-product of the factory. Considering that fibre properties and the presence of impurities affect the behaviour of fibre-reinforced cement composites, it is necessary to determine the basic quality parameters of waste fibres before they are used in these materials. In addition, the processing technique, i.e., cutting of production waste fibres or purification of RTPF, is required for the preparation of production waste fibres and, together with the reliable integration technique, is the main obstacle for the uniform distribution of waste fibres in the cement composites. Acknowledgements. The research presented here was carried out as part of the projects “Cement Composites Reinforced with Waste Fibres” - ReWire (UIP-2020-02-5242) and “Career Development of Young Researchers - Training of New Doctors of Science” (DOK-2021-02-4884) at the Faculty of Civil Engineering, University of Zagreb, both funded by the Croatian Science Foundation. Special thanks to Kelteks Ltd., Karlovac, Croatia (www.kelteks.com) and Gumiipex – GRP d.o.o. (www.gumiimpex.com) for providing resources and continuous support in the implementation of the ReWire project.
References 1. Nakagawa, H., Akihama, S., Mahi, F.T.: Fiber reinforced cement composites. Ref. Modul Mater. Sci. Mater. Eng. Epub ahead of print (2017) 2. Wongtanakitcharoen, T., Naaman, A.E.: Unrestrained early age shrinkage of concrete with polypropylene, PVA, and carbon fibers. Mater. Struct. Constr. 40, 289–300 (2007) 3. Kizilkanat, A.B.: Experimental evaluation of mechanical properties and fracture behavior of carbon fiber reinforced high strength concrete. Period Polytech Civ. Eng. 60, 289–296 (2016) 4. Wang, Q., Ding, Y., Randl, N.: Investigation on the alkali resistance of basalt fiber and its textile in different alkaline environments. Constr. Build. Mater. 272, 121670 (2021) 5. Abdou, T.R., Botelho Junior, A.B., Espinosa, D.C.R., et al.: Recycling of polymeric composites from industrial waste by pyrolysis: deep evaluation for carbon fibers reuse. Waste Manag. 120, 1–9 (2021) 6. Haramina, T., Jelavi´c, T., Šoli´c, T., et al.: Analiza mogu´cnosti recikliranja diskontinuiranih E-staklenih vlakana. Polim 32, 52–61 (2011) 7. Straits Research. https://www.globenewswire.com/news-release/2022/09/05/2509962/0/en/ Concrete-Reinforcing-Fiber-Market-Share-is-growing-at-a-CAGR-of-6-82-With-Compet itive-players-Like-BASF-SE-Bekaert-Sika-AG-Nycon-Corporation-etc.html (2022)
602
A. Bariˇcevi´c et al.
8. Global demand for carbon fiber from 2010 to 2022. https://www.statista.com/statistics/380 538/projection-demand-for-carbon-fiber-globally/. Accessed 1 Dec 2022 9. Recycled Carbon Fiber Market (2022). https://www.marketsandmarkets.com/Market-Rep orts/recycled-carbon-fiber-market-212462057.html 10. Glass Fiber Reinforced Concrete Market Size, Share & Trends Analysis Report By Process (Spray, Premix, Hybrid), By Application, By Region, And Segment Forecasts (2020) – 2027. https://www.grandviewresearch.com/industry-analysis/glass-fiber-rei nforced-concrete-market. Accessed 1 Dec. 2022 11. Ian, T.: Market value of polypropylene worldwide from 2015 to 2020, with a fore-cast for 2021 to 2026 (2021). https://www.statista.com/statistics/858659/global-polypropylene-mar ket-value-projections/. Accessed 1 Dec 2022 12. Global Tire Recycling Industry Analysis By Rubber Type, By Product Type, By End User And By Geography & COVID-19 Impact With Market Outlook 2017–2030 (2020) 13. Militký, J., Mishra, R., Jamshaid, H.: Basalt fibers (2018). Epub ahead of print 2018. https:// doi.org/10.1016/B978-0-08-101272-7.00020-1 14. Onuaguluchi, O., Banthia, N.: Value-added reuse of scrap tire polymeric fibers in cementbased structural applications. J. Clean. Prod. 231, 543–555 (2019) 15. Mavridou, S., Oikonomou, N.: Utilization of textile fibres from worn automobile tires in cement based mortars. Glob. Nest. J. 13, 176–181 (2011) 16. Chen, M., Chen, W., Zhong, H., et al.: Experimental study on dynamic compressive behaviour of recycled tyre polymer fibre reinforced concrete. Cem. Concr. Compos. 98, 95–112 (2019) 17. Chen, M., Zhong, H., Wang, H., et al.: Behaviour of recycled tyre polymer fibre reinforced concrete under dynamic splitting tension. Cem. Concr. Compos. 114, 103764 (2020) 18. Baricevic, A., Pezer, M., Jelcic Rukavina, M., et al.: Effect of polymer fibers recycled from waste tires on properties of wet-sprayed concrete. Constr. Build. Mater. 176, 135–144 (2018) 19. Owens Corning CEM-FIL® 5325. https://www.owenscorning.com/en-us/composites/pro duct/cem-fil-5325 20. TEIJIN Tenax(TM) filament yarn (2021). https://www.teijincarbon.com/products/tenaxr-car bon-fiber/tenaxr-filament-yarn 21. Kamenny Vek - Advanced Basalt Fibre (2022). https://basfiber.com/products/roving 22. Bariˇcevi´c, A., Jelˇci´c Rukavina, M., Pezer, M., et al.: Influence of recycled tire polymer fibers on concrete properties. Cem. Concr. Compos. 91, 29–41 (2018) 23. Sharma, M., Gao, S., Mäder, E., et al.: Carbon fiber surfaces and composite interphases. Compos. Sci. Technol. 102, 35–50 (2014) 24. Helebrant, A., Hradecká, H., Holubová, B., et al.: Kinetics of processes modeling corrosion of glass fibers mixed into concrete. Ceram. Silik. 61, 163–171 (2017) 25. Yang, T., Zhao, Y., Liu, H., et al.: Effect of sizing agents on surface properties of carbon fibers and interfacial adhesion of carbon fiber/bismaleimide composites. ACS Omega. Epub ahead of print (2021). https://doi.org/10.1021/acsomega.1c01103 26. Brandt, A.M.: Fibre reinforced cement-based (FRC) composites after over 40 years of development in building and civil engineering. Compos. Struct. 86, 3–9 (2008) 27. Mao, L., Shen, H., Han, W., et al.: Hybrid polyurethane and silane sized carbon fibre/epoxy composites with enhanced impact resistance. Compos. Part A Appl. Sci. Manuf. 118, 49–56 (2019) 28. Thomason, J.L., Nagel, U., Yang, L., et al.: A study of the thermal degradation of glass fibre sizings at composite processing temperatures. Compos. Part A Appl. Sci. Manuf. 121, 56–63 (2019) 29. Scheffler, C., Förster, T., Mäder, E., et al.: Aging of alkali-resistant glass and basalt fibers in alkaline solutions: evaluation of the failure stress by Weibull distribution function. J. Non. Cryst. Solids 355, 2588–2595 (2009)
Production Waste Fibres as a Sustainable Alternative
603
30. Bertelsen, I.M.G., Ottosen, L.M., Fischer, G.: Influence of fibre characteristics on plastic shrinkage cracking in cement-based materials: a review. Constr. Build. Mater. 230, 116769 (2020) 31. Hussain, I., Ali, B., Akhtar, T., et al.: Comparison of mechanical properties of concrete and design thickness of pavement with different types of fiber-reinforcements (steel, glass, and polypropylene). Case Stud. Constr. Mater. 13, e00429 (2020)
Effect of Elevated Temperatures on Concrete Made with Ash from Wood Biomass and Recycled Polymer Fibers from Waste Rubber Marija Jelˇci´c Rukavina1(B)
, Ivan Gabrijel1 , Martina Kozlik2 , Vanja Žvorc3 , and Nina Štirmer1
1 Faculty of Civil Engineering, University of Zagreb, Zagreb, Croatia
[email protected]
2 Vinci Construction Terrassement GmbH, Berlin, Germany 3 Medimurje ˇ PMP, Cakovec, Croatia
Abstract. With the increasing awareness of global warming, largely due to the emission of large amounts of CO2 , some attempts have been made to develop strategies to help combat the growing trend of global warming. To reduce the harmful effects of concrete production, efforts are being made to reduce the amount of cement used by replacing the cement with mineral additives to maintain or improve the properties of the concrete. In addition, the use of waste materials as substitutes for cement and other concrete constituents is being promoted and investigated. This paper presents an experimental study on the influence of ash from wood biomass (used as a cement substitute up to 30%) and the addition of recycled polymer fibers from waste rubber in an amount of 2 kg/m2 (used as a substitute for industrial polypropylene fibers) on the residual mechanical and durability properties of concrete exposed to high temperatures up to 600 °C. In addition, reference concretes were prepared using only cement and industrial polypropylene fibers for the comparison purposes. A total of 5 different mixes were tested. The following tests were performed on the concrete specimens before and after thermal treatment: compressive strength, static modulus of elasticity, ultrasonic pulse velocity and gas permeability. These properties represent the input parameters for the evaluation of the condition of reinforced concrete elements after exposure to fire. The obtained results show that the use of the investigated waste materials in concrete leads to comparable or slightly better properties after fire exposure compared to the reference concrete. Keywords: wood biomass ash · recycled polymer tyre fibers · elevated temperatures · mechanical · durability properties
1 Introduction In view of the problems of global warming and dealing with non-renewable energy sources, there is a growing interest in the research of the waste materials as new additives for concrete production. The use of ecological components can not only have a positive © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 A. J˛edrzejewska et al. (Eds.): SynerCrete 2023, RILEM Bookseries 43, pp. 604–611, 2023. https://doi.org/10.1007/978-3-031-33211-1_54
Effect of Elevated Temperatures on Concrete Made with Ash
605
effect on reducing the energy demand in cement production and reducing harmful CO2 emissions, but also reduces the cost of waste disposal. For several years, the possibility of using waste ash generated from wood biomass energy production (WBA) as a substitute for cement in concrete has been investigated [1–3]. Considering the growing number of biomass power plants, the potential for using waste generated during energy production is increasing. Taking into account the whole process from energy consumption and environmental impact during the production of the material to installation, maintenance and finally disposal, it can be said that concretes containing ecological components have a positive impact on the environment [4]. On the other hand, reinforced concrete structures are exposed to the risk of fire daily. Many fire accidents have shown that reinforced concrete structures do not collapse in most cases and that strengthening, and rehabilitation can be performed. Therefore, knowledge about the properties of cementitious composites after a fire is very important taking into an account the fact that a large number of innovative cementitious materials are used nowadays [5]. In this work, the effects of two ecological components were analyzed, namely wood biomass ash (WBA) as a substitute for cement in an amount up to 30% by weight and polymer fibers from recycled car tires (RTPF) as a substitute for polypropylene fibers. The research is focused on experimental investigation of mechanical and durability properties after exposure to high temperatures of 600 °C.
2 Experimental Program 2.1 Materials and Concrete Mixes Cement CEM I 42.5 according to HRN EN 197-1 [6] and wood ash from biomass from the Croatian biomass plant LIKA ENERGO EKO d.o.o. were used as binders for the concrete mixes. Other materials comprised: aggregates with a maximum size of 16 mm were naturally sourced crushed dolomite, a superplasticizer based on modified polycarboxylic ether polymer and water from the general city drinking-water supply. For mixes that comprised fibres, two types of fibers were used: commercial monofilament polypropylene (PP) and recycled tire polymer fibers (RTPF). The mix proportions of the tested concretes are listed in Table 1. All mixes were prepared with the same binder content (475 kg/m3 ) and the same effective water to binder ratio (w/b = 0.4). Mix M0 was prepared as control mix only with cement as a binder. In the mixes, labelled M10 and M30 in Table 1, part of the cement was replaced with WBA in amounts of 10% and 30% by total binder weight, respectively. Mixes M10-2PP and M10-2RTPF beside WBA had 2 kg of fibers, respectively. All mixes were designed to meet consistency class S4 (160–210 mm) which was achieved with certain amount of superplasticizer. 2.2 Specimens – Curing, Dimensions and Heat Treatment The study comprised mechanical (compressive strength and modulus of elasticity) and durability tests (ultrasonic pulse velocity and gas permeability). Apart from permeability testing, the prepared specimens were cylinders with dimensions Ø = 75 mm and L =
606
M. J. Rukavina et al. Table 1. Concrete mix designs
Mix
M0
M10
M30
M10-2PP
M10-2RTPF
Cement [kg]
475
428
333
428
428
v/b ratio
0.4
0.4
0.4
0.4
0.4
Superplasticizer [l/m3 ]
2.4
4.4
4.8
4.8
4.8
WBA
% mb
-
10
30
10
10
kg/m3
-
48
143
48
48
-
-
-
2
-
PP [kg/m3 ] RTPF [kg/m3 ]
-
-
-
-
2
Aggregate 0/4 [kg]
630
625
620
623
623
Aggregate 4/8 [kg]
386
383
380
381
382
Aggregate 8/16 [kg]
753
748
742
745
745
Slump value [mm]
195
200
180
175
175
Density [kg/m3 ]
2.46
2.50
2.48
2.49
2.45
Air content [%]
1.7
1.6
1.3
2
2.9
225 mm (i.e. slenderness equal to 3) as recommended by RILEM Technical Committees [7–9]. Gas permeability was measured with the RILEM-CEMBUREAU method based on the Klinkenberg approach [10] using the specimens Ø = 100 mm and L = 50 mm, cut from cylindres of Ø = 100 mm and L = 50 mm. All specimens were demoulded one day after casting and stored in a curing room for another 27 days at a temperature of 20 ± 2 °C and a relative humidity of 95%. Then the specimens were placed in the normal laboratory conditions until high-temperature exposure. At the age of more than 90 days, the test specimens were exposed to high elevated temperatures of 200 °C, 400 °C and 600 °C in programmable control electric furnace capable to reach the maximum temperature of 1400 °C. The maximum temperature of 600 °C was chosen in order to clearly see the influence of the binder on post-fire residual properties, because decomposition of dolomite into lime (CaO), periclase (MgO) and carbon dioxide (CO2 ) is produced at the temperature of about 700 °C [11]. The thermal treatment of specimens consisted of three phases, as follows: 1. the specimens were heated to the target temperature at slow heating rate of 1 °C/min to avoid unacceptable temperature gradients across the cross-section; 2. the target temperature was kept constant until a uniform thermal condition was reached in the specimen; 3. after the heating process was completed, the furnace was turned off and the specimens were allowed to cool naturally to room temperature in the closed furnace to avoid thermal shock and cracking induced by cooling of the concrete material. Eight cylindrical specimens were used for each temperature cycle, with one specimen equipped with NiCr thermocouples that monitored temperature evolution across the specimens (both surface and in the centre) according to RILEM recommendations [7]
Effect of Elevated Temperatures on Concrete Made with Ash
607
for high temperature testing. After cooling, the specimens were stored in a normal laboratory environment until testing.
3 Results and Discussion Results are presented with average and absolute deviations upon testing of four specimens compressive strength (Fig. 1a) and ultrasonic pulse velocity, UPV (Fig. 2a) and three specimens for modulus of elasticity (Fig. 3a) and gas permeability (Fig. 4a). UPV of tested mixes were assessed upon testing the same specimens before and after thermal treatment, while for gas permeability, compressive strength and modulus of elasticity four groups of specimens were formed in respect to three temperature levels (200, 400 and 600 °C) and non-heated specimens for comparison. Apart from gas permeability, other tested properties are presented in both absolute and relative terms. Relative values of each property present ratio between property obtained on specimens after thermal treatment and the same property of non-heated specimens. 3.1 Compressive Strength Figure 1 shows the results of the compressive strength test in absolute and relative terms.
Fig. 1. Evolution of compressive strength with temperature a) absolute values b) relative values
The initial compressive strength results presented here show that mixes with 10% WBA as binder have 11.6% (M10), 14.5% (M10-2PP), and 3.9% (M10-2RTPF) higher compressive strength, respectively, compared to the control concrete mix. When 30% binder was replaced by WBA, the compressive strength decreased by 8% compared to the control mix. Figure 1b shows the decrease in compressive strength (as a percentage of the initial values) for the studied mixes after exposure within the temperature range considered. Compressive strength decreased for 15–19.6%, 45.1–49.3% and 61–70%, after exposure to temperatures of 200, 400 and 600 °C, respectively. Similar values (70, 70% the aggregates; Sand, Sakcrete Coarse, 30% of the aggregates; PCM, Micronal DS 5008 X, 13.5% by volume of concrete.
White Portland cement was used for architectural purposes, as it allowed an easier adjustment of the final color of the concrete through pigments. A wide range of colors was evaluated, before choosing a light grey color (Fig. 5). 3.2 Mechanical Properties Concrete was mechanically tested according to the ASTM C109/C109M. The addition of PCM was expected to reduce the compressive strength of the concrete, as a consequent effect of the reduced density. With the addition of 20% PCM by volume of concrete, the compressive strength was reduced by 40%. A consistent trend for other percentages of PCM was found, coherently with literature and Fig. 5 3.3 Thermal Properties The thermal properties of the pure PCM were measured using a DSC apparatus. The principle of the DSC test is to keep a temperature equilibrium between a small test sample and a reference sample that is heated or cooled at a constant rate. The excess heat absorbed or emitted by the test sample is recorded as a function of the time. A PCM sample was placed in an aluminum pan. The temperature range from 10 °C to 40 °C was selected, whereas cyclic heating/cooling/heating scans were conducted at a rate of 5 °C/min with 2 min isothermal holds at both minimum and maximum temperatures. The first cycle started with heating up the sample from 10 °C in order to come to the situation in which all wax was solid; for the subsequent cooling and heating cycles, solidification and melting enthalpies were estimated by using a linear approximation both below and above the melting and solidification peaks. The DSC experiment was conducted using a Q20 manufactured by TA Instruments calibrated in temperature and heat flux indium with 99.99% of purity. The sample content was registered by double weighing with a ± 0.02 mg error. The results of multiple cycles showed high reproducibility. Figure 6 reveals that the melting process takes place progressively. This is because some of the chemical components are able to extend the melting range. The differences observed between the measured values were considered normal, due to the nature of the process. Moreover, no significant difference was recorded repeating the measurement at a speed of 0.2 ºC/min. Based on these tests, during solidification (the endothermic process), an enthalpy of 102.8 J/g in a temperature range from 22.1 °C to 9.3 °C was measured. For the melting process (the exothermic process), the enthalpy was 99.7 J/g in a temperature range of from 18.8 to 35.4 °C. The difference between the solidification and melting enthalpies was 3.1 J/g, a negligible effect.
Innovative PCM-Enhanced Concrete Tiles for High Performance Buildings
1215
The TES properties of the PCM-enhanced concrete specimen were determined using the dynamic heat flow meter apparatus (DHFMA) method. Traditionally, a heat flow meter apparatus is used to measure steady-state thermal properties but if the temperatures of the top and bottom plates are kept identical and a step change in specimen temperature is introduced by changing the temperature of both plates by the same amount, then the heat capacity during the temperature change is calculated as the ratio of total heat flow to the temperature step size. Melting and solidification tests on the sample using a FOX 200 HFMA system were performed. The specimen was 30 × 30 cm in area and 3,8 cm thick. The dynamic testing results are shown in Fig. 7. To improve the accuracy and temperature resolution of the measurements, two tests with step size of 1.5 °C and separated by 0.75 °C with each other were performed, and combined as described by Shukla et al. (2012). Figure 6 shows that the melting and freezing curves are very similar with negligible subcooling. The onset, peak and end of the melting process were observed at 18, 22.5 and 24.75 °C, respectively. On the other hand, the onset, peak and end of the solidification process were observed at 24, 22.5 and 18 °C respectively.
Fig. 6. Volumetric heat capacity as a function of temperature for PCM-enhanced concrete.
4 Solatrium House 4.1 Project: The Need for Concrete Thermal Storage Solatrium was a project undertaken to participate to the China Solar Decathlon, a competition that is based on the request to have houses powered only by sun energy. This requirement is generally satisfied by providing photovoltaic systems on the roof. To achieve a low energy demand, the design generally focuses on highly insulated envelopes aiming to isolate the houses from the external environments, and small window-to-wall ratios. On the opposite, one of the goals of Solatrium was to reinterpret this approach and to select a highly transparent facade design.
1216
U. Berardi
The team adopted low-E double glazing for all the vertical windows, with a solar heat gain coefficient of 59%. For the atrium roof windows, Solatrium uses a multi skin acrylic glazing with an exterior solar control films (Fig. 7). Solatrium is primarily constructed of sandwich fiber-reinforce polymer FRP panels with FRP outer skins of glass fibers embedded in a polyester matrix, and connected by glass fiber strands, and a core of polyisocyanurate foam. These panels, 9 cm thick, are particularly light, and serve as the structural shell, insulation layer and finishing surface in the ceiling and walls. The U-value of the walls is 0.33 W/m2 K. As evident, the envelope missed any thermal mass. Meanwhile, more than half of the walls in Solatrium were occupied by floor-to-ceiling windows, with the South façade having a WWR of 90%. Solatrium has a total transparent opening area of about 80 m2 , a quantity prohibitive for typical approaches to zero-energy goal. However, beyond the benefit of significant daylight, solar gains were abundant and the design opted to store in the floor PCM-concrete, the high thermal energy that enters through the windows of Solatrium (Fig. 7).
Fig. 7. Photo taken in the project during the installation of the concrete tiles.
4.2 Energy Saving Energy simulations of Solatrium were performed using EnergyPlus. This tool simulates materials with variable properties such as PCM by using the one-dimensional conduction finite difference, i.e. CondFD, solution algorithm. Then, additional temperaturedependent property information is associated by using a dedicated module. The user can choose between two different models: one is semi implicit model, based on the Crank– Nicholson scheme (Eq. 1), and it is second-order in time, and adopts an implicit finite
Innovative PCM-Enhanced Concrete Tiles for High Performance Buildings
1217
difference scheme, combined with an enthalpy–temperature function, based on assigned input; the other is the fully implicit scheme Eq. 2, which is of first order in time. This last was selected as provide more stable and realistic results. The PCM were modelled using the function MaterialProperty: PhaseChangeHysteresis which allows to take into account that the conductivity, the specific heat and the density in the liquid phase and in the solid phase are different. Moreover, this tool also allows to take into account the phenomena of the hysteresis. In the simulation, the tiles enriched with PCMs have been simulated with two types of PCMs (melting temperature of 21 °C and 24 °C), five mass fraction (0%, 3%, 5%, 10% and 15% of content of PCM into concrete tiles by weight) and five thickness of the tiles (0 cm, 1 cm, 2 cm, 3 cm and 4 cm) for a total of fifty different scenarios. The properties of the PCM are reported in Fig. 8. The models were performed in different cities to be able to compare the effects of the weather too. The results will be shown only for the cities of Toronto (Ontario, Canada), and then also for Rome (Italy) and New York (USA).
Fig. 8. Specific heat and enthalpy in function of the temperature of concrete tiles enriched with 5% (above) and 10% (below) PCM.
1218
U. Berardi
5 Results and Discussion of Energy Saving Figure 9 shows the trend of the heating energy use intensity when the concrete tiles are enriched with PCMs. For the PCM with a phase change temperature of 21 °C, the best configuration is when the floor is made with concrete tiles enriched with a 3% of PCM by mass thick 4 cm. This scenario is characterized by a consumption of 67,3 kWh/m2 . The floor configuration with no tiles has the worst performance (93,4 kWh/m2 ). From this configuration, increasing the thickness of the tiles, the consumptions have a particularly trend. Using tiles with a thickness of 1 cm the consumptions are reduced, from 81,8 kWh/m2 to 74,5 kWh/m2 increasing the content of PCMs into tiles from 3% to 15%. When the thickness tiles values 2 cm, this trend is no more respected, because the
Fig. 9. Energy use intensity in Toronto with concrete tiles enriched with PCM having a phase change temperature of 21 °C a), c) and 24 °C b), d).
Innovative PCM-Enhanced Concrete Tiles for High Performance Buildings
1219
best scenario is not anymore the scenario with the higher content of PCMs. The lowest consumption is with a concrete tile with 10% of PCMs. From 3 cm, the consumptions have the inverse trend compared to the trends having for 1 cm thickness tiles. The lowest values of consumption are for tiles with the smallest content of PCMs and the highest values for tiles with the highest content of PCMs. The values range from 68,8 kWh/m2 to 72,8 kWh/m2 for 3 cm, and from 67,3 kWh/m2 to 73,0 kWh/m2 for 4 cm. Increasing the thickness of the tiles increases the thermal inertia of the house and so decrease the indoor temperature fluctuations as shown in the graph of Fig. 10. In Fig. 11 it is shown that when the wave of temperature is less flattened is easier for it to reach lower temperatures and maybe to reach the phase change temperature. On days characterized by high solar radiation, the characteristics of the house (high WWR and light and insulated walls) mean that the internal temperature reaches very high temperatures. For example, the second day analysed in the graph, shows that a solar radiation of about 900 W/m2 leads to an internal temperature of about 53 °C.
Fig. 10. Cooling energy use intensity using tiles enriched with 15% of PCM with a phase change temperature of 24 °C, Toronto.
Figure 12 shows the internal temperature trends during three typical days of the winter season taken in February with the tiles have a thickness of 1 cm. In the figure, it is clearly shown that as the PCM content increases, there is a decrease in temperature peaks (both the maximum and the minimum). We pass from peaks of maximum 38,7 °C to 27 °C for the cases in which the floor is free of PCM and for the case in which there is 15% of PCM in the tile, respectively.
1220
U. Berardi
Fig. 11. Temperature trends using tiles enriched with 15% of PCM with a phase change temperature of 24 °C, no system, Toronto.
Fig. 12. Winter temperature trends in Toronto with a) 1 cm and b) 3 cm concrete tiles.
6 Conclusions The concrete tiles with PCM on the floor that receive direct solar radiation within an enclosure offer significant advantages as a LHTES approach that can be used as passive thermal systems, stabilizing the internal temperature and increasing the comfort. This paper has presented a comprehensive investigation of a new PCM-enhanced concrete floor tile. It has also shown that high effectiveness is achieved in sunny tiles, while the contribution of the tiles far from the windows is smaller.
Innovative PCM-Enhanced Concrete Tiles for High Performance Buildings
1221
References 1. Berardi, U., Gallardo, A.: Design and control of radiant cooling panels incorporating PCMs. Appl. Energy 304, 117736 (2021) 2. Hun Park, J., Wi, S., Jin Chang, S., Berardi, U., Kim, S.: Energy retrofit of PCM-applied apartment buildings considering building orientation and height. Energy 222, 119877 (2021) 3. Soudian, S., Berardi, U.: Assessing the effect of night ventilation on pcm performance in high-rise residential buildings. J. Building Phys. 43, 3 (2019) 4. Soudian, S., Berardi, U., Laschuk, N.: Development and characterization of a cementitious plaster with phase change materials finished with thermochromic paint. Sol. Energy 205, 282– 291 (2020) 5. Berardi, U., Abdolmaleki, L.: Single and multi-phase change materials used in cooling systems. Int. J. Thermophys. 43, 4 (2022) 6. Naraid, J., et al.: Design and application of concrete tiles enhanced with microencapsulated phase-change material. J. Archit. Eng. 22, 1 (2016)
Phase Change Materials Shape Stabilized in Biochar for Energy Efficiency and Structural Strength Enhancement in Buildings Carolina Santini1 , Claudia Fabiani1,2(B) , Antonella D’Alessandro3 and Anna Laura Pisello1,2
,
1 CIRIAF - Interuniversity Research Center, University of Perugia, Via G. Duranti 63, 06125
Perugia, Italy [email protected] 2 Department of Engineering, University of Perugia, Via G. Duranti 97, 06125 Perugia, Italy 3 Department of Civil and Environmental Engineering, University of Perugia, Via G. Duranti, 93, 06125 Perugia, PG, Italy
Abstract. In this work, a commercial phase change material (PCM) was stabilized in biochar, by vacuum impregnation technique and later incorporated into cement pastes to be used in real building applications. The selected paraffin is characterized by a phase transition temperature of 27 °C, i.e., within the most common thermal comfort conditions in building applications. The obtained compounds were analyzed at various scales of investigation using advanced thermo-chemical techniques, to properly assess the composites’ thermophysical performance and long-term stability. The obtained results highlight the promising thermal buffer capability of the shape-stabilized samples during the early-stage hydration process. In general, all the compounds tend to lose PCM during cycling, but significant leakage was only found after 1000 thermal cycles, suggesting a relatively stable behavior. Adding the shape stabilized PCM lowers the peak temperature by about 4 and 5 °C compared to the normal and the biochar-doped composite with positive consequences regarding compressive strength. Keyword: biochar · phase change materials (PCM) · thermal energy storage · vacuum impregnation · cement composites
Highlights • Production of new cement pastes with PCM-RT27 stabilized in porous matrix added to the mix; • Use of vacuum impregnation technique to directly incorporate PCM into the biochar matrix; • Reduction of the peak temperature during the hydration process of cementitious paste; • Increased compressive strength of PCM-impregnated compound; • Higher heat storage capacity due to latent heat of PCM-RT27 evaluated by DSC analysis. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 A. J˛edrzejewska et al. (Eds.): SynerCrete 2023, RILEM Bookseries 43, pp. 1222–1231, 2023. https://doi.org/10.1007/978-3-031-33211-1_109
Phase Change Materials Shape Stabilized in Biochar
1223
1 Introduction Cement is a widely used material worldwide. In FederBeton’s “Chain 2020 Report” [1] global cement consumption, was estimated at about 4 billion 138 million tons, and distribution channels, among other analyses, were evaluated, noting an increase, especially in building material retailers, construction companies, premixing and precast concrete. Its wide use has led researchers to focus on cement paste properties and implementation. Among the most important characteristics to be considered is the heat that cement releases during the setting process, as it takes on very high values during the initial period (first two-three days) creating non-negligible temperature differences between the various parts of the compound (core and peripheral areas), which can lead to cracking of the concrete, resulting in a decrease in its mechanical strength. Several studies have monitored the hydration process of cement by various means, including the use of calorimeters. Over the years, different types of these instrumentations have become widespread, variations of which, along with their advantages/disadvantages, were analyzed in the article by Forbrich L. R. And Carlson R. W. [2]. Other experimental ways of analyzing hydration-induced reactions have been conducted by Tang S. W. et al. using electrical methods [3] and by Kong Q. and Song G., who, instead, used piezoceramicbased transducers as smart aggregates capable of transmitting and receiving the waves generated during hydration [4]. Piezoelectric transducers have also been used by Qin L. and Li Z. [5] to implement the ultrasonic technique and correlate the results obtained with the properties of the cement hydration process. In addition to developments in the monitoring technique, research has also addressed possible interventions to be applied to be able to mitigate the heat loss induced by the hydration of cement paste. In the study by Bundur Z. B. et al., for example, cement was mixed with a bacterial solution to evaluate the effect on compressive strength after hydration, finding an increase in it compared to that pure material [6]. Czapik P. et al., instead, used cement kiln dust to understand how it reacts with cement paste and water in the initial setting phase of cement by XRD, TG, and SEM analysis [7]. Recently, therefore, studies have been directed toward the type of mixture to be introduced into cement to mitigate hydration heat dissipation and improve mechanical performance. From the research of Zhang Q. et al. [8] it is inferred that using carbon black in percentages of 2% relative to the weight of the initial cement yields good values in terms of compressive strength. Other research has used PCMs as additional materials to exploit their phase transition property as a mode of heat storage in the form of latent heat. Studies using these materials are various and differ in how PCMs are incorporated into the cement paste; Feng Q. et al. [9] have used microencapsulated PCMs dispersed in the paste and performed DSC, FTIR, SEM, and TG analyses to derive the cement-acquired properties while Fabiani C. et al. [10] carried out analyses on the thermal and mechanical behavior of cement with 1% of microencapsulated PCMs, finding maintenance of adequate flexural and compressive strength and a decrease in the temperature differential induced on the compound during the hydration process. Eddhahak A. et al. also focused their experiments on evaluating the effects induced by the use of PCMs in the cementitious compound during hydration, creating a numerical model for calculating heat flow [11].
1224
C. Santini et al.
The work presented aims to implement research in this area by using PCMs as materials capable, when mixed into cement, of decreasing the differential temperature rise during hydration and maintaining adequate mechanical performance. Compared with research to date, the PCM chosen in this study was mixed into cement after being incorporated into a matrix through vacuum impregnation. With a view to environmental sustainability and the reuse of waste materials, moreover, biochar, a by-product from biomass pyrolysis, was chosen as the matrix. In addition, mechanical (flexural and compressive strength after 7 days from production) and thermal (monitoring in the first two days of cement paste hydration) analyses were carried out on samples in pure cement, in cement and non-impregnated biochar, and in cement and impregnated biochar.
2 Materials and Methods In this work, we use PCM shape stabilized in biochar for producing thermally enhanced cement pastes with improved mechanical performance. We carry out extended thermomechanical investigations for evaluating the final performance of the innovative mix design. 2.1 Materials and Sample Preparation The PCM-doped biochar (PCM-BC) was prepared using the vacuum impregnation technique [12] and two basic constituents: the biochar matrix and an RT organic paraffin wax from RUBITHERM® with a nominal melting temperature of 27 °C. The biochar was produced by a commercial gasification plant (located in Perugia, Italy) using a downdraft fixed-bed gasifier to process softwood chips as feedstock. The RT27 paraffin from RUBITHERM® has a heat storage capacity of 189 kJ/kg and a relatively low thermal conductivity (about 0.2 Wm−1 K−1 ). 2.1.1 Production of the Impregnated Biochar Following is the description of the impregnation technique used in this work. A 1:10 matrix-to-PCM ratio was used to guarantee effective impregnation. After drying and weighing the biochar matrix, the following steps were followed: • First the previously conditioned biochar was placed inside a beaker and maintained in vacuum conditions for three hours; • Second, the PCM was liquefied and slowly poured into the beaker while vacuum was till on; • Next, the biochar + PCM mix was mechanically stirred for 24h under controlled temperature and humidity conditions (60 °C and 10%); • Finally, the excess PCM was filtered, and the composite was kept at 70 °C and 10% relative humidity for 8 h. The obtained sample was weighed again after filtering, and the percentage of PCM actually absorbed by the matrix was calculated (results show an impregnation rate of about 66.7%).
Phase Change Materials Shape Stabilized in Biochar
1225
2.1.2 Mix Design and Sample Preparation The shape stabilized PCM producing following the procedure in Sect. 2.1 was used to produce the innovative cement paste with a fixed amount of thermally enhanced additive, i.e., 2% of the total cement weight. A standard and a biochar-added cement paste without the addition of PCM were also prepared for comparative purposes. In all cases, no aggregates were considered, because they might affect the curing process and interact with the PCM. All additives were used in addition to the reference mix design, for maintaining the same water/cement ratio of the original product and, consequently, comparable mechanical results. The three different mix designs are summarized in Table 1. Table 1. Mix designs of the produced cement pastes, i.e., normal paste (NP), paste doped with pure biochar (BCP), and paste doped with PCM shape stabilized in biochar (PCM-BCP). Components
NP
BCP
PCM-BCP
Cement (g)
317.89
316.21
317.10
Water (g)
127.15
126.48
126.84
Biochar (g)
–
6.32
–
PCM-biochar (g)
–
–
6.34
w/c ratio
0.40
0.40
0.40
6 prismatic and 3 cubic samples were produced in total, i.e., 2 beams and 1 cube per mix design, with dimensions of 40 mm × 40 mm × 160 mm and 50 mm × 50 mm × 50 mm, respectively. All samples were produced with the same procedure: • First all dry materials (cement and biochar or PCM-doped biochar) were homogeneously mixed; • Second, water was added, and the compound was carefully blended; • Next, the mixes were poured in oiled moulds and compacted to reduce the voids in the matrix, upon reaching adequate workability; • Finally, all samples were cured within the controlled environment of a Matest climatic chamber at 20 °C and 95% relative humidity. 2.2 Experimental Methodology The previously described samples were investigated by means of a dedicated thermomechanical investigation campaign aimed at quantifying the effect of the shape-stabilized component of the curing process of the mortars. Therefore, the three pastes were thermally monitored during curing and their mechanical performance were later investigated by means of acknowledged mechanical tests. Additionally, the shape-stabilized components were analysed using Differential Scanning Calorimetry (DSC) before and after thermal cycling.
1226
C. Santini et al.
2.2.1 Thermo-physical Characterization of the Shape-Stabilized PCM The shape-stabilized PCM produced following the procedure in Sect. 2.1.1 was thermally cycled for evaluating PCM loss due to leakage processes. To do that, the composite was placed inside 4 plastic cylinders equipped with a small piece of paper towel at their base. Each cylinder was weighed before and after introducing the composite. A T100-BIORAD thermal cycler was used for repeatedly (10 and 1000 times) exposing each sample to the following cycle: • • • •
2 min at 10 °C; temperature increase with the constant rate of 5 °C/min from 10 °C to 50 °C; 2 min at 50 °C; temperature decrease with the constant rate of 5 °C/min from 50 °C to 10 °C.
After the thermal cycling was completed, the samples were weighted for evaluating the mass loos and consequently their leakage performance. For further characterizing the composite, DSC investigations were carried out before and after thermal cycling. The cycling was done at 5 °C min−1 under flowing 80 mL min−1 nitrogen gas. Each test was performed between −20 and 60 °C with a mass sample of around 5 mg located in 40 µL aluminium crucibles. 2.2.2 Thermal Monitoring of the Hydration Process 7 T-type thermocouples, 6 placed at the center of each sample’s face and one at the center of the specimen, were used for registering the temperature gradients produced throughout the whole volume of the samples during the hydration process. The thermal analysis of the cement pastes was carried out inside polyurethane moulds (simulating the occurrence of more massive pouring) during the first 72 h of the curing process since the critical exothermic reaction in mortars takes place in the first hours after casting [10]. A cDAQ-9184 data acquisition system equipped with 2 NI 9213 Spring slots was programmed to read the sensors every 10 s. 2.2.3 Mechanical Characterization The 40 mm × 40 mm × 100 mm prismatic specimens of each sample were first broken in their central part in the flexural test and then each of two halves was used for measuring the compressive strength of the samples. Compressive and flexural strength tests were carried out using a universal testing machine Controls model Advantest 9. The frame for flexural tests had a maximum compressive force capacity of 15 kN. The tests were load-controlled, with a loading rate of 50 N/s, up to failure. The results were obtained from the average calculation of the values. To determine the flexural strength of each type of cement paste the equation presented in Eq. (1) was used: (1) Rf = (1.5 · L · Ff )/ b3 where Rf is the flexural strength (N/mm2 ), Ff is the maximum flexural load (N), L is the distance between the supports in the three-point bending test configuration (100 mm), b is the side of the square section of the prismatic sample.
Phase Change Materials Shape Stabilized in Biochar
1227
The frame for compressive tests had a maximum compressive force capacity of 250 kN. The tests were load-controlled, with a loading rate of 2400 N/s, up to failure. Four cubic specimens were used for this test, and the results were obtained from the average calculation of the values. To determine the compressive strength of each type of cement paste the equation presented in Eq. (2) was used: Fc = F/Ac
(2)
where Fc is the compressive strength (MPa), F is the maximum load applied (N), and Ac is the cross-sectional area of the specimen where the load for the test is applied (i.e., 1600 mm2 ).
3 Results from the Thermal Investigation Figure 1 shows the trends of average temperatures recorded during the first three days of hydration of the three produced cubic samples (NP, BCP, PCM-BCP). The tem-peratures were recorded by the 7 probes used for thermal monitoring carried out inside the climate chamber (with constant ambient conditions of 20 °C. The boundary conditions were having a constant temperature of 20 °C and 95% RH). The test lasted 72 h. The trends are similar and from these three phases can be distinguished: • the heating phase of the samples from room temperature to the temperature of the climatic chamber; • the heating phase related to the cement hydration process, characterized by exothermic chemical reactions; • the cooling phase. The values considered particularly significant (peak temperatures and duration between the slowing down of heating and cooling rates for the most acute phase of the cement hydration process) are given in Table 1. In a previous study from the same authors [10], it was shown that micro-encapsulated PCM could reduce temperature peaks in the curing phase, which, in turn, allows a slight improvement in the mechanical performance of the samples. In the current contribution, we show that by using PCM-impregnated biochar one can maintain and even improve the thermal effect of the latent additive while using a more sustainable and cost-effective PCM integration process. Also, using a char matrix was shown to be more promising in terms of final mechanical performance as shown in Sect. 4. These differences testify that the BCP sample reaches the highest peak temperature. The sample containing PCM-BCP has both a significantly lower peak temperature than the other two and a shorter duration. This behavior is related to the solid/liquid state changes and vice versa of the PCMs contained in the BC that result in heat absorption and heat transfer, respectively, with consequent slowdowns in heating and cooling rates (Table 2 and Fig. 2). The DSC trend of PCM-BC was evaluated as a function of the heating and cooling cycles performed on the impregnated compound. The 3 curves shown in Fig. 3 show the trends according to the number of thermal cycles to which the compound was subjected, namely at 0 cycles, 10 cycles and 1,000 cycles, from which a progressive reduction in
1228
C. Santini et al.
Fig. 1. Average temperature trend of cubic samples during the hydration process Table 2. Peak temperature, duration and density of cubic samples NP
BCP
PCM-BCP
Peak temperature [°C]
30.76
31.39
26.33
Duration [hh]
18.83
16.68
13.68
1.8
1.91
1.88
Density [g/cm3 ]
Fig. 2. DSC analysis of PCM-BC
Phase Change Materials Shape Stabilized in Biochar
1229
latent heats during both melting and solidification of the PCM as the number of thermal cycles increases, due to the dispersion and loss of the PCM itself. The problem of the durability of PCM within the matrices is evident, especially after 1000 cycles. Future studies regarding these innovative materials should therefore focus on solutions to reduce the loss of PCM during the aging of impregnated materials.
Fig. 3. DSC analysis of PCM-BC at 0 cycles, 10 cycles and 1000 cycles
4 Results from the Mechanical Investigation Figure 4 shows the flexural and compressive strength results in relation to the density of each composite sample. Generally, the flexural strength reduces by adding biochar and shape-stabilized composite. On the contrary, the minimum value of compressive strength is registered for the NP sample (24.06 MPa), followed by the PCM-BCP (31.75 MPa) and the BCP sample (31.84 MPa). This can be explained by considering the registered density and the thermal monitoring during curing. Indeed, BCP samples show the highest density and possibly a reduced porosity, which produces a higher and more collimated temperature peak during curing and a steep increase in terms of compressive strength. Using the shape stabilized PCM, on the other hand, limits the density increase to lower values compared to the BCP sample. Despite this, the reduced thermal gradients registered within the sample compensate for this reduced density maintaining the final compressive strength of the paste above 31 MPa. This is a rather good increase (about 31%), highlighting the promising potential of PCM-doped char integrations in concrete matrices. Indeed, if the previous work from the same authors did show a non-negligible increase in the final compressive strength of the samples including micro-encapsulated PCMs, i.e., about 14% [10], this work showed that even greater performance can be achieved when the micro-capsule is replaced by a more sustainable material such as impregnated biochar.
1230
C. Santini et al.
Fig. 4. Flexural strength, compressive strength and density of the different samples.
5 Conclusions In this work, a small amount of PCM shape stabilized in biochar, i.e., 2% in weight, is introduced in a cement paste for reducing the sensible heat released during the hydration process and the consequent thermally induced micro-cracking. Lower temperature gradient within the matrix could indeed improve mechanical performance. The thermal monitoring during the first 72 h of curing showed the significant role of the PCM in reducing the sensible heat released in the first 24 h of the hydration process. In particular, the peak temperature registered for the standard cement paste was lowered by about 4 °C compared to the normal paste and 5 °C to the biochar-doped composite. The latent additives were also shown to positively affect the dynamics of the heating process, producing a more stable temperature profile during the overall hydration process. The positive thermal effect of the shape stabilized PCM was also corroborated by the compressive strength investigation showing an average characteristic strength increase of about 7 MPa compared to the reference cement paste. In conclusion, the thermo-mechanical investigations carried out in this work demonstrated the positive effect deriving from the introduction of a small amount of PCM shape stabilized in a biochar matrix, i.e., 2% of the total cement weight, in common cement-water composites. This innovative solution could represent an effective way to
Phase Change Materials Shape Stabilized in Biochar
1231
improve the final mechanical performance of concrete castings subjected to extensive thermal expansion phenomena.
References 1. Di Filiera, R. (2020) 2. Carlson, R.W., Forbrich, L.R.: Correlation of methods for measuring heat of hydration of cement. Ind. Eng. Chem. Analytical: 382–386 (1939) 3. Tang, S.W., Cai, X.H., He, Z., Zhou, W., Shao, H.Y., Li, Z.J., et al.: The review of early hydration of cement-based materials by electrical methods. Constr. Build. Mater. 146, 15–29 (2017) 4. Kong, Q., Song, G.: A comparative study of the very early age cement hydration monitoring using compressive and shear mode smart aggregates. IEEE Sens. J. 17(2), 256–260 (2017) 5. Qin, L., Li, Z.: Monitoring of cement hydration using embedded piezoelectric transducers. Smart Mater. Struct. 17(5) (2008) 6. Bundur, Z.B., Kirisits, M.J., Ferron, R.D.: Cement and concrete research biomineralized cement-based materials: Impact of inoculating vegetative bacterial cells on hydration and strength. Cem. Concr. Res. [Internet]. 67: 237–245 (2015). https://doi.org/10.1016/j.cemcon res.2014.10.002 7. Czapik, P., Zapała-sławeta, J., Owsiak, Z., Ste, P.: Hydration of cement by-pass dust. Constr. Build. Mater. 231 (2020) 8. Zhang, Q., Luan, C., Yu, C., Huang, Y., Zhou, Z.: Mechanisms of carbon black in multifunctional cement matrix: Hydration and microstructure perspectives. Constr. Build. Mater. 5, 346 (2022) 9. Feng, Q., Liu, X., Peng, Z., Zheng, Y., Huo, J., Liu, H.: Preparation of low hydration heat cement slurry with micro- encapsulated thermal control material. Energy [Internet]. 187, 116000 (2019). https://doi.org/10.1016/j.energy.2019.116000 10. Fabiani, C., Pisello, A.L., D’Alessandro, A., Ubertini Filippo, C.L., Cotana, F.: Effect of PCM on the hydration process of cement-based mixtures: A novel thermo-mechanical investigation. Materials (Basel) 1–17 (2018) 11. Eddhahak, A., Drissi, S., Colin, J., Caré, S., Neji, J.: Effect of phase change materials on the hydration reaction and kinetic of PCM-mortars. J. Therm. Anal. Calorim. 117(2), 537–545 (2014). https://doi.org/10.1007/s10973-014-3844-x 12. Wan, Y.C., et al.: A promising form-stable phase change material prepared using cost effective pinecone biochar as the matrix of palmitic acid for thermal energy storage. Sci. Rep. 9(1):1–10 (2019)
The Effect of Salt-Impregnation on Thermochemical Properties of a Metakaolin Geopolymer Composite for Thermal Energy Storage Lorena Skevi1(B) , Xinyuan Ke1 , Jonathon Elvins2 , and Yulong Ding3 1 Department of Architecture and Civil Engineering, University of Bath, Bath, UK
[email protected]
2 SPECIFIC Innovation and Knowledge Centre, Swansea University, Swansea, UK 3 Birmingham Centre for Energy Storage, School of Chemical Engineering,
University of Birmingham, Birmingham, UK
Abstract. Effective and efficient recovery, storage, and reuse of heat, together with renewable energy, play an indispensable role in decarbonising the built environment. Thermochemical energy storage materials possess the highest volumetric energy density compared to latent and sensible heat storage materials under similar conditions. However, conventional thermochemical energy storage materials face several challenges including high cost, low sustainability, and limited heating power. Alkali-activated metakaolin (geopolymer) containing alkali aluminosilicate hydrates (N-A-S-H) has been shown to have a considerable thermochemical heat storage capacity at medium temperatures (below 400 °C) but is less efficient at low-temperatures (90%, Sigma Aldrich) and sodium silicate (H2 O/Na2 O = 8, SiO2 /Na2 O = 3, Sigma Aldrich). The SiO2 and Al2 O3 content of the metakaolin was found to be 52.6% and 44.4% respectively in previous study [22]. For enabling the formation of beads, a 3.65 wt.% sodium alginate (Sigma Aldrich) aqueous solution was added to the geopolymer mix, serving as an emulsifier and stabiliser [21]. The bead production process is based on the ionotropic gelation of polyelectrolytes – sodium alginate in this case – in the presence of counter ions, provided here by the CaCl2 hydrous solution [20]. The detailed composition of the alkaline activator and the sodium alginate solutions is given in Table 1, while the composition of the geopolymer-alginate (GA) composite slurry is presented in Table 2 and was designed so that a molar ratio Si/Al of 1.5 was acquired. Table 1. Composition of the solutions used for making the GA composite. Sodium silicate (g)
NaOH (g)
Sodium alginate (g)
Distilled water (g)
Alkaline activator solution
200
45.95
-
54.03
Sodium alginate solution
-
3.65
100
-
Table 2. Mix design of the GA slurry composite.
GA
Metakaolin (g)
Alkaline activator solution (g)
Sodium alginate solution (g)
30
46.58
8.67
The Effect of Salt-Impregnation on Thermochemical Properties
1235
Figure 1 presents schematically the sample preparation process and their subsequent treatment. First, metakaolin was mixed with the alkaline activator solution at an electric mixer and stirred for 2 min at 1600 rpm. The sodium alginate solution was then added to the geopolymer mix and the GA slurry was stirred for another minute at 1600 rpm. The beads were prepared following a spherification process, in which droplets of the GA mix were dropped via a 1ml pipette into a 500 ml of 1M CaCl2 .2H2 O (>90%, Sigma Aldrich) solution kept at 80 °C throughout the dripping process. The sodium alginate contained in the GA slurry gels directly when contacting Ca2+ in the CaCl2 bath, allowing the formation of beads as the GA slurry falls in the chloride solution. A total of approximately 50–60 beads were prepared in this way, with their diameters varying between 3–4 mm as shown in Fig. 2. Once they were formed the beads were treated in three different ways as shown in Fig. 1: one group of approximately 15 beads were put directly to the oven to dry at 80 °C (GA-O); another one was first rinsed with distilled water, to remove remains of the CaCl2 from the shell of the beads and were then left to dry in the oven at 80 °C (GA-R); the final group was left to stay in the CaCl2 bath for 1 h before taken to the oven for drying at 80 °C (GA-1h). An overview of the testing methods applied to the geopolymer beads is given in Fig. 1.
Fig. 1. Overview of the sample preparation and the testing process
1236
L. Skevi et al.
Fig. 2. The batch of geo-beads after taken out of the CaCl2 bath (A), with their mean diameter being 3–4 mm (B).
2.2 Methods Phase Assemblage The compositional phases present in the geo-beads after each treatment were characterised using XRD and Attenuated total reflectance Fourier transform infrared (ATRFTIR) spectroscopy. Prior to these tests, 3 oven-dried geo-beads of each type, i.e., GA-O, GA-R and GA-1h, were crushed into powder with pestle and mortar. Transmission powder XRD was conducted with a STOE STADI P (Cu radiation, λ = 1.54 Å) operated at 40 kV voltage and 40 mA current. The diffraction patterns were recorded over the range 5° - 75° (2θ), with a step resolution of 0.015° (2θ), using a double Mythen detector where each detector covers a range of 19° (2θ) and operates at 31.6 s per degree (2θ). ATP-FTIR was conducted with a Perkin Elmer Frontier instrument using the transmission cell. The spectral analysis was performed from wavenumber 400 cm−1 to 4000 cm−1 , with 16 repeated scans and 1 cm−1 resolution for each spectrum acquisition. Heat Storage Capacity To evaluate the thermal energy storage capacity of the geo-beads, first 5 beads per type, i.e., GA-O, GA-R and GA-1h were put in an environmental chamber under controlled temperature and humidity of 20 °C and 90% RH respectively for 15 h. Subsequently, the hydrated beads were crushed into powder with pestle and mortar and their dehydration enthalpy was evaluated using differential scanning calorimetry (DSC) conducted in the DSC Q20 (TA Instruments) in combination with thermogravimetric analysis (TGA) performed in Setsys Evolution TGA 16/18 (Setaram) instrument. The DSC analysis was conducted in two heating ramps, in samples of approximately 5 mg. First, the samples remained at 25 °C for 30 min and then they were heated up to 395 °C at a 10 K/min rate, where they remained for 15 min before cooling down to 25 °C and stabilising there for another 15 min (first ramp). Next, they were heated again up to 395 °C at the same rate of 10 K/min (second ramp).
The Effect of Salt-Impregnation on Thermochemical Properties
1237
For the TG analysis approximately 12 mg of the powdered samples were used, which were heated from 30 °C up to 400 °C at a rate of 10 K/minute under inert Ar atmosphere at 60 ml/minute. The water content of the samples was then calculated as the mass loss (%) that occurred from 30 °C to 400 °C. Water Sorption The water sorption of the geopolymers were assessed with dynamic (water) vapor sorption (DVS) using the DVS Advantage instrument (Surface Measurement Systems). The cyclic sorption performance of the geopolymers was evaluated by 10 cycles of relative humidity changes from 0% to 95% at a constant temperature of 25 °C. Before starting the test, the samples were pre-dried at 200 °C (maximum temperature achievable by the instrument) until a constant weight was obtained, signifying complete dehydration of the geopolymers. No heating stages were applied between the cycles.
3 Results and Discussion 3.1 Phase Assemblage The XRD results of the three examined geo-beads are presented in Fig. 3. The spectra of the raw metakaolin and the sodium silicate that were used for preparing the samples are also given in the same figure. Two main crystalline phases were identified in the metakaolin used in this study; anatase and quartz. These were still present in the geo-beads, meaning that they did not participate in the formation of hydrates in the geopolymer samples. Sodium alginate, has an amorphous structure, having little impact on the diffraction spectra of the geo-beads. In addition to quartz and anatase, portlandite (Ca(OH)2 ) was found in all three geo-beads samples. It is possible that during the spherification process, some Cl− anions were bound with Na+ that is present in the geo-beads, allowing the Ca2+ to hydrate forming portlandite. Indeed, halite (NaCl) was found in the geo-beads that remained in the CaCl2 solution for 1 h after their formation and were not rinsed with water afterwards (GA-O-1h). The geo-beads that were removed from the CaCl2 solution immediately after their formation and were either rinsed with water (GA-R) or not (GA-O) did not show any clear peaks of halite. The FTIR spectra of the geopolymer beads did not reveal the presence of portlandite in any of the geo-beads, as the characteristic peak of portlandite at 3640 cm−1 contributed by the vibration of O-H bond did not appear in any of the spectra (this area of the graph is not shown in Fig. 4). The band at 1458 cm−1 is attributed to a combination of the C-OH deformation vibration and the carboxylate symmetric stretch vibration with the bending of the -CH2 that is found in sodium alginate [20]. This peak was more prominent in the GA-R and GA-O beads, but not in the GA-O-1h. It is also possible that some carbonation occurred during the FTIR analysis resulting in the 1458 cm−1 and the depletion of portlandite. Other than that, all these three samples presented a peak at 995 cm−1 which is attributed to the presence of Si or Al tetrahedral functional groups due to the asymmetric stretching of the Si-O-Si and/or Si-O-Al bonds [17, 23]. Finally, the peak at 1639 cm−1 due to H-O-H bending vibration indicates the presence of water in all samples and overlaps with the asymmetric stretching of the -COO of the alginate [20].
1238
L. Skevi et al.
Fig. 3. XRD spectra of the geo-beads, threated in 3 different ways (GA-R: water rinse, GA-O: no rinse, GA-O-1h: impregnated in CaCl2 for 1h and no rinse) and the metakaolin (MK) and sodium alginate used for preparing the beads.
3.2 Heat Storage Capacity Figure 5 shows the two heating ramps applied for the DSC analysis (Fig. 5a) of the 90% RH saturated GA-R, GA-O and GA-O-1h geo-beads in terms of heat flow over time. Dehydration is completed in the first ramp (up to 395 °C), while the enthalpy during the second ramp reflects the sensible heat of the materials. Since the heat storage capacity of the geopolymers lies in the dehydration/rehydration (desorption/sorption) process, the dehydration enthalpy (hatched area of 1st ramp) for each of the geo-beads was calculated using the Universal Analysis software coupled with the DSC Q20 instrument and is presented in Table 3 as H per mass of sample (J/g), H per mole of H2 O (kJ/molH2O ) and H per dry mass of sample (J/gdry-mass ). The peak dehydration temperature for the examined samples is also given in Table 3. The dehydration enthalpy per mass of sample was found to be increased in the following order: GA-O-1h < GA-R < GA-O being 356.1, 369.6 and 395.7 J/g respectively. Considering the density of the metakaolin based geopolymers approximately 1.6 g/cm3 [24], the volumetric energy density for the GA-O-1h, GA-R, and GA-O geo-beads would be 158, 164 and 176 kWh/m3 respectively. These values are comparable to the volumetric energy densities of ettringite-based [11, 25] and cement-based [13] composites, but are lower than zeolite-based or previously studied geopolymer systems [17]. The
The Effect of Salt-Impregnation on Thermochemical Properties
1239
Fig. 4. FTIR spectra of the geo-beads, threated in 3 different ways (GA-R: water rinse, GA-O: no rinse, GA-O-1h: impregnated in CaCl2 for 1h and no rinse).
relatively low water content present in the GA as occurred from the TGA, between 9.4– 11% (Table 3), as compared to pure metakaolin geopolymers whose water content is approximately 15% [17] seems to be related with the lower energy density in the GA samples. The more crystalline structure of the GA-O-1h due to the presence of NaCl as shown by the XRD results, appears to lead to decreased dehydration enthalpy, which agrees with previous results [17]. The peak temperature increased in the order of GA-O-1h < GA-R < GA-O, following the dehydration enthalpy trend, with the temperatures being 102 °C, 106 °C and 109 °C respectively. These temperatures are lower than the ones found for pure metakaolin activated geopolymers, which was found to be 124 °C. The lower peak temperature designates that lower charging temperatures can be used in the GA samples, making them appropriate for use in waste heat storage applications where lower temperature heat is available. Overall, the prolonged (1 h) salt-impregnation (GA-O-1h) that resulted in higher crystallisation of the geopolymer structure led to lower energy density but also lower charging temperature. The short salt impregnation on the other hand (GA-O sample) favoured the energy storage capacity of the geo-beads samples without significantly increasing the charging temperature.
1240
L. Skevi et al.
Fig. 5. The DSC heating ramps configuration (a) and the DSC results for the geo-beads, threated in 3 different ways (GA-R: water rinse, GA-O: no rinse, GA-O-1h: impregnated in CaCl2 for 1h and no rinse). Table 3. Dehydration enthalpy and onset temperature of the examined geo-beads. Water content (%)
Hdehydration (J/g)
Hdehydration (kJ/molH2O )
Hdehydration (J/gdry-mass )
Peak temperature (°C)
GA-R
11
369.6
60.3
415.4
106.1
GA-O
11.6
395.7
61.7
447.4
109.5
GA-O-1h
9.4
356.1
68.5
392.9
102.2
3.3 Water Sorption The mass increase of the GA-R and GA-O samples that were subjected to cyclic vapor sorption over time is shown in Fig. 6. It is obvious that the water sorption capacity of the samples is decreased over time after the 4th cycle. As a result, the mass increase changes
The Effect of Salt-Impregnation on Thermochemical Properties
1241
from 25% in the 1st cycle to approximately 15% in the 10th , with the GA-O presenting slightly higher stability over time.
Fig. 6. Mass increase of the samples over time during DVS at 10 cycles of 0 to 90% of RH at 25 °C.
4 Conclusions This work presented the preliminary results on the thermal energy storage performance of a geopolymer-sodium alginate composite impregnated in CaCl2 solution. It was shown that although short-time salt-impregnation resulted in enhanced heat storage capacity, the prolonged impregnation in the salt hydrate had the opposite effect. Future work will be conducted for improving the manufacturing process of the geobeads as well as increasing their heat storage capacity by examining the impregnation in combinations of salt hydrates such as MgSO4 /CaCl2 . Further testing of the surface and pore structure properties of the beads could provide valuable information about their sorption/desorption capacity, while their testing in the reactor level will provide a more realistic view of their performance as thermochemical energy storage materials. Acknowledgement. This research was funded by the UK Engineering and Physical Sciences Research Council (EPSRC) through Grant EP/W010828/1.
References 1. Kaufmann, J., Winnefeld, F.: Seasonal heat storage in calcium sulfoaluminate based hardened cement pastes – experiences with different prototypes. J. Energy Storage 25, 100850 (2019) 2. Tatsidjodoung, P., Le Pierrès, N., Luo, L.: A review of potential materials for thermal energy storage in building applications. Renew. Sustain. Energy Rev. 18, 327–349 (2013) 3. André, L., Abanades, S., Flamant, G.: Screening of thermochemical systems based on solidgas reversible reactions for high temperature solar thermal energy storage. Renew. Sustain. Energy Rev. 64, 703–715 (2016)
1242
L. Skevi et al.
4. Zhao, Q., et al.: Optimization of thermochemical energy storage systems based on hydrated salts: a review. Energy Build. 244, 111035 (2021) 5. Cot-Gores, J., Castell, A., Cabeza, L.F.: Thermochemical energy storage and conversion: a-state-of-the-art review of the experimental research under practical conditions. Renew. Sustain. Energy Rev. 16(7), 5207–5224 (2012) 6. Humphries, T.D., et al.: Dolomite: a low cost thermochemical energy storage material. J. Mater. Chem. A 7(3), 1206–1215 (2019) 7. Wang, K., et al.: A review for Ca(OH)2 /CaO thermochemical energy storage systems. J. Energy Storage 50, 104612 (2022) 8. Yang, Y., et al.: Thermochemical heat storage and optical properties of red mud/Mn co-doped high alumina cement-stabilized carbide slag in CaO/CaCO3 cycles. Fuel Process. Technol. 236, 107419 (2022) 9. Johannes, K., et al.: Design and characterisation of a high powered energy dense zeolite thermal energy storage system for buildings. Appl. Energy 159, 80–86 (2015) 10. Ndiaye, K., Ginestet, S., Cyr, M.: Experimental evaluation of two low temperature energy storage prototypes based on innovative cementitious material. Appl. Energy 217, 47–55 (2018) 11. Chen, B., et al.: Investigation on ettringite as a low-cost high-density thermochemical heat storage material: thermodynamics and kinetics. Sol. Energy Mater. Sol. Cells 221, 110877 (2021) 12. Alonso, M.C., et al.: Calcium aluminate based cement for concrete to be used as thermal energy storage in solar thermal electricity plants. Cem. Concr. Res. 82, 74–86 (2016) 13. Lavagna, L., et al.: Cementitious composite materials for thermal energy storage applications: a preliminary characterization and theoretical analysis. Sci. Rep. 10(1), 12833 (2020) 14. Ke, X., Baki, V.A.: Assessing the suitability of alkali-activated metakaolin geopolymer for thermochemical heat storage. Microporous Mesoporous Mater. 325, 111329 (2021) 15. Bell, J.L., et al.: X-Ray pair distribution function analysis of a metakaolin-based, KAlSi2 O6 ·5.5H2 O inorganic polymer (geopolymer). J. Mater. Chem. 18(48), 5974–5981 (2008) 16. Gong, W., et al.: Geopolymer concretes for energy storage applications. United States of America (2020) 17. Ke, X., Baki, V.A.: Assessing the suitability of alkali-activated metakaolin geopolymer for thermochemical heat storage. Microporous Mesoporous Mater. 325, 111329 (2021) 18. Xueling, Z., et al.: Heat storage performance analysis of ZMS-Porous media/CaCl2 /MgSO4 composite thermochemical heat storage materials. Sol. Energy Mater. Sol. Cells 230, 111246 (2021) 19. Walsh, S., et al.: Assessing the dynamic performance of thermochemical storage materials. Energies 13 (2020). https://doi.org/10.3390/en13092202 20. Medri, V., et al.: Metakaolin-based geopolymer beads: production methods and characterization. J. Clean. Prod. 244, 118844 (2020) 21. Papa, E., et al.: Geopolymer-hydrotalcite hybrid beads by ionotropic gelation. Appl. Clay Sci. 215, 106326 (2021) 22. Walkley, B., et al.: Thermodynamic properties of sodium aluminosilicate hydrate (N–A–S–H). Dalton Trans. 50(39), 13968–13984 (2021) 23. Olvianas, M., Widiyatmoko, A., Petrus, H.T.B.M.: IR spectral similarity studies of geothermal silica-bentonite based geopolymer. AIP Conf. Proc. 1887(1), 020015 (2017) 24. Jaya, N.A., et al.: Correlation between pore structure, compressive strength and thermal conductivity of porous metakaolin geopolymer. Constr. Build. Mater. 247, 118641 (2020) 25. Ndiaye, K., Cyr, M., Ginestet, S.: Development of a cementitious material for thermal energy storage at low temperature. Constr. Build. Mater. 242, 118130 (2020)
Hygrothermal Measurement of Heavy Cob Materials Ouellet-Plamondon Claudiane(B)
and Kabore Aguerata
Eole de Technologie Supérieure, Université du Québec, Montreal H3C 1K3, Canada [email protected], [email protected]
Abstract. The need to reduce the energy for heating and cooling is leading to a renewal of interest in geo-sourced materials for modern building construction. These materials offer many advantages in terms of climate change adaptation, and reduction of CO2 emissions from building materials. This study presents the characterization of clay, fiber, and hygrothermal characterization of two types of clay mixtures reinforced with natural vegetable fiber for use in modern construction. Cob is a mixture of clay, water, and vegetable fibers. It is non-load bearing and serves as a filler for the wood frame. The wall design is suited for lightly seismic zones. Cob has great potential for applications in places with high housing and cooling needs. The hygrothermal properties of 30 clay formulations containing 0%, 3%, and 6 wt% wheat fibers were evaluated. At this fiber content, the material is considered a heavy type of cob. The measured thermal and hydric properties of the clay-fiber mixture presented in this paper are thermal conductivity, thermal effusivity and diffusivity, specific heat, water absorption, water vapor permeability, moisture buffer value (MBV), and sorption/desorption isotherms. Wheat fibers used in clay material mixtures improve their hygrothermal properties for sustainable construction. These results allow the development of a model hygrothermal model of wall elements. Keywords: Hygrothermal properties · clay · fiber · cob · geo-sourced materials · sustainable building
1 Introduction The population used ecomaterials such as earth and wood or earth and stone in the construction of buildings. Over the years, the ecomaterials used in construction for decades have been replaced by cement blocks with high thermal conductivity and low thermal inertia. Cement, concrete blocks, and rebar are the main consumers of energy and CO2 emitters [1].The impact of these materials on climate change is a major issue for the construction sector, especially in Africa, where the main construction material is cement. These materials are responsible for more than 50% of greenhouse gas emissions over the entire life cycle of a building [1]. Awareness of energy resource depletion, environmental sustainability issues, and climate change has prompted architects as well as researchers to revisit clay, fiber-reinforced clay, and wood-clay building implementation techniques. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 A. J˛edrzejewska et al. (Eds.): SynerCrete 2023, RILEM Bookseries 43, pp. 1243–1252, 2023. https://doi.org/10.1007/978-3-031-33211-1_111
1244
O.-P. Claudiane and K. Aguerata
Directing the building sector towards the use of these composite materials with a woodframe structure is a major issue in the overall perspective of sustainable development [2]. The construction of wood-frame buildings using clay reinforced with plant fiber as filling materials has a particularly low energy footprint. In addition, this type of construction increases the life of the building by providing them with good seismic resilience due to its wooden structure [3]. However, the phenomena that influence the proper functioning of wood-clay walls are heat and moisture transfer phenomena. The thermal and moisture properties of clay materials reinforced with plant fiber made using the cob technique, are not precisely known and there are few studies present in scientific journals [4]. As a result, it is difficult to predict the thermal and hydric performance of buildings constructed with wood/clay reinforced plant fiber. However, before improving the energy performance of buildings, which are responsible for nearly 50% of GHG emissions, it is necessary to improve the hygrothermal performance of the elements of its envelope even before thinking about improving the mechanical systems [5]. The key factor in the energy efficiency of a building is the quality of the materials used in the design of its envelope [5]. In Africa, the housing deficit stands at about 51 million housing units and is increasing over the years due to high population growth [6]. Woodframe construction using vegetable fiber-reinforced clay as fillers could help reduce the number of housing demands and effectively dispose of the many agricultural wastes. In this article, we are interested in the manufacturing technique of the vegetable fiber reinforced clay material manufactured using the cob manufacturing technique and the hygrothermal properties of the materials obtained after manufacturing. The manufacture of the samples was carried out by research of quality index, cracking rate and shrinkage. This allowed us to find the most appropriate technique for the fabrication of samples for the measurement of hygro-thermal properties.
2 Materials and Methods 2.1 Materials The study focuses on the plant fibers reinforced clay material that will be used for the filling of the wood-frame building structure. The material was manufactured in the laboratory of École de technologie supérieure. Two types of clay were used, a red colored clay containing iron oxide called red clay and another beige colored clay containing more kaolinite called beige clay. The bulk density of the red clay and beige clay is 2.8 g/cm3 and 2.7 g/cm3 with a plasticity index of 15% and 16% respectively. Bulk density was determined according to ASTM D854 [7] guidelines and details can be found in Kabore and al. [8] and plasticity index according to Canadian standard CAN/BNQ-2501-090 [9]. Both clays were sourced from the province of Alberta and the wheat fibers were sourced from the province of Quebec, Canada. Figure 1 shows the clays and fibers used to make the samples.
Hygrothermal Measurement of Heavy Cob Materials
1245
Fig. 1. a) Red clay, b) beige clay and c) fibers from 0.4 to 30 cm long
2.2 Method 2.2.1 Method of Making the Samples Prior to the fabrication of the samples for the choice of the quality index as well as the mode of humidification of the fibers, the clays and fibers were previously characterized. The methods and results of the characterization of clays and fibers are presented in the work of Kaboré and Ouellet-Plamondon [8]. The manufacturing of the samples in clay without fibers was then carried out with a water/clay ratio of 15%, 20%, 25%, and 30% with 3 samples for each water/clay ratio to determine the quality index for the following. The samples have dimensions of 25 cm × 25 cm × 2.5 cm and 36 mm in diameter and 30 mm in height. The samples were made using the adobe method, a mixture of water and clay [10]. After manufacture, they were then dried in an oven at 30 °C for 10 h–11 h to reduce the amount of moisture before being exposed to laboratory temperature (23 °C) for 7 days. Volume shrinkage during drying being one of the most important factors for the manufacture of clay materials, especially adobe materials using a lot of water for manufacturing were therefore evaluated. The three dimensions 25 cm × 25 cm × 2.5 cm of each sample were measured using 0.01 mm precision digital calipers every 24 h for 7 days of drying until the measured dimensions values stabilized. The volume shrinkage was therefore calculated as a function of the values obtained in each measurement of each dimension by the Eq. 1 [11] Rv = ((Vo − Vf )/Vo) ∗ 100
(1)
With Rv the volume shrinkage rate in %, Vo the initial volume in m3 and Vf the final volume in m3 . After obtaining the quality index with the water/clay ratio, the manufacture of samples reinforced with 3% and 6% of fibers were made with unmoistened and moistened fibers. This step consists of manufacturing clay samples reinforced with vegetable fibers with less shrinkage and without cracking. The manufacturing technique used is that of the cob and the manufacture was carried out on the spot in laboratory with the hand. Figure 2 and 3 present the principle of choice of the quality index and the procedure of manufacture of the samples. The choice of the quality index was made considering the plasticity of the mixture for the filling of the frames and for a manufacture of the material in a construction site.
1246
O.-P. Claudiane and K. Aguerata
Fig. 2. Choice of the quality index for the manufacture of samples without plant fibers reinforcement.
This manufacturing technique is very simple and is done in a traditional way (mixing by hand, feet, and shovels for large quantities). At the end of the experiment presented in Fig. 2, the choice of the quality index was the water/clay ratio of 25%. The samples without fibers made with 25% water gave a volume shrinkage between 12 and 13% and did not show any surface cracking. According to Laou, Lamyaa [12] volume shrinkage during of clay materials depending on their nature and clay content should be between 4% and 20%. Lertwattanaruk et al. [13] in their work, obtained volume shrinkage of 30% on adobe bricks of dimension 10 cm × 10 cm × 10 cm. And Emiro˘glu et al. [14] obtained a maximum volume shrinkage rate of 23% for red clay samples and 26% for yellow clay samples of dimension 5 cm × 5 cm × 5 cm. The volume line of the samples with the water/clay ratio of 20% being less than 20%, we proceeded to manufacture samples with unmoistened and moistened fibers presented in Fig. 3. The procedure of fabricating samples with both wetted and unwetted fibers resulted in the selection of samples that did not show cracks. The samples selected for testing were the samples made with wetted and unsaturated fibers (see Fig. 3). After these two steps, the samples with dimensions of 25 cm × 25 cm × 2.5 cm and 36 mm diameter and 30 mm height were again fabricated for thermal property measurements. For measurements of hydric properties, the beige clay samples tested were samples without fibers and samples with 6% fibers. This clay will be used as a fiberless coating. However, it is necessary to observe the influence of the fibers on the hydric properties of the samples made with this clay, so measurements of the hydric properties were made on samples of beige clay alone and reinforced with 6% fibers. Three samples per mix were used for the tests and the results are presented in Sect. 3. 2.2.2 Method for Measuring Hygrothermal Properties Thermal conductivity (λ) and thermal effusivity (E) were measured by the transient plane source method modified using the C-THERM apparatus (see Fig. 4a), thermal
Hygrothermal Measurement of Heavy Cob Materials
1247
Fig. 3. Procedure for making samples
diffusivity (D) and specific heat [15] were obtained by calculation using Eqs. 2 and 3. The C-THERM method uses a one-sided sensor to directly measure the thermal conductivity and effusivity of materials. The samples used for testing were dried in room air (23°C). D = λ2/E2
(2)
Cp = λ/ρ · D
(3)
The water absorption coefficient was determined by the partial immersion capillary absorption method (Fig. 4b) and the water vapor permeability was determined according to ASTME96 [16]. As for the moisture buffer values (MBV), they were measured following the NORDtest protocol [17]. All samples were first sealed and stabilized at 50% relative humidity before testing. Figure 4c shows the principle of sealing the samples according to the Nordtest protocol. The dynamic vapor sorption (DVS) method was used for the sorption tests.
1248
O.-P. Claudiane and K. Aguerata
Fig. 4. a) C-THERM device, b) Principle of water absorption test of cob and earthen samples, and c) Principle of sample sealing for the Nordtest protocol.
3 Results and Discussion 3.1 Thermal Properties The average dry densities of the samples used for thermal properties measurements range from 2016 to 1997 kg/m3 for samples without fibers, 1626 kg/m3 to 1588 kg/m3 for samples with 3% fibers, and 1412 to 1370 kg/m3 for the 6% fibers. As expected, the dry density decreases with increasing fiber content in the mixtures. The results of thermal conductivity, specific capacitance, thermal diffusivity and thermal effusivity of the samples are presented in Fig. 4 and 5. The addition of fibers in the clay mixtures positively affects all the coefficients of thermal properties of the samples.
Fig. 5. Thermal conductivity and specific capacity with increasing fiber content (dried samples at 23 °C)
Hygrothermal Measurement of Heavy Cob Materials
1249
Fig. 6. Thermal diffusivity and Thermal diffusivity with increasing fiber content (dried samples at 23 °C)
The results indicate that increasing the wheat straw fiber content from 0 to 3, or 6% results in a significant increase in the thermal insulation of the material. The increase in specific capacity improves the ability of the material to store absorbed heat. The remarks on the decrease in thermal conductivity have been observed in the work of several authors including Ashour and al. and Ramakrishnan and al. [18, 19] although the technique of fabricating the samples for his work is different from the one used for this work. As for the thermal inertia properties, they decrease with fiber content, which was observed in the work of [20]. Ideal building materials for application in construction in hot regions should have the lowest thermal effusivity, i.e., the lowest capacity to absorb heat from the environment, and/or the highest specific heat capacity, i.e., the highest capacity to store the absorbed heat [21]. This translates into low thermal conductivity and diffusivity and consequently, high thermal inertia. Materials with 3% and 6% fibers would be better materials to use for wood framing in hot regions, however materials with 6% fibers have the best properties because materials made with 6% fibers have high thermal inertia than those with 3% fibers and no fibers. 3.2 Hydric Properties Table 1 shows the results on the average values of water vapor permeability, water absorption and moisture buffer value. The water absorption coefficient increases with the increase of the fiber rate used for the fabrication of the samples. These results were validated with the sorption isotherm curves presented in Fig. 6. The moisture buffer value decreases with the fiber rate. However, all the values of the moisture buffer value remain above 2 (g/ (m2 . %H), thus excellent according to Rode and al. [17]. The hygrothermal properties of the materials such as water vapor permeability, water absorption coefficient and moisture buffer value (MBV) of the plant fiber reinforced clay materials manufactured by the cob manufacturing technique are little studied and few results can be found in the literature, so it is difficult to comment on the hygrothermal properties obtained in this paper with those in the literature. Following the understanding of the hygrothermal performance of wheat fiber reinforced materials, Fig. 6 presents the sorption behavior of samples manufactured with 0%, 3% and 6% fibers. The time evolution of all samples subjected to increasing relative
1250
O.-P. Claudiane and K. Aguerata
Table 1. Average values of water vapor permeability, water absorption coefficient and moisture buffer value of the samples Sample code
Water vapor permeability (kg/(s·m·Pa))
Water absorption coefficient (%)
Moisture buffer value (MBV) (g/ (m2 . %H)
Red clay without fiber
1.5.10–11 ± 5.1.10–13
7
6.1 ± 0.01
Red clay with 3% fibers
1.4.10–11 ± 5.2.10–13
10
5.5 ± 0.1
Red clay with 6% fibers
1.6.10–11 ± 5.4.10–13
22
3.2 ± 0.1
Beige clay without fiber
1.2.10–11 ± 7.8.10–13
7
4.2 ± 0.1
Beige clay with 6% fibers
1.3.10–11 ± 1.2.10–12
24
4.0 ± 0.1
humidity shows that the response of the plant fiber reinforced and non-fiber reinforced samples to a change in relative humidity is somewhat rapid and the water balance is reached in less than 10 days for some humidity intervals. It can be seen from the sorption curve that the equilibrium moisture content increases with increasing relative humidity for all samples but also with increasing fiber content (Fig. 7).
Fig. 7. Sample mass variation during sorption-desorption tests.
Hygrothermal Measurement of Heavy Cob Materials
1251
4 Conclusion This article aims to valorize clay materials in the cob technique by studying their hygrothermal properties. The samples in the study were made with 0%, 3%, and 6% of traditional fibers. The role of the addition of fibers is to improve the hygrothermal properties of the clay materials. This work has shown that the addition of fibers in the matrix of clays have improved all the hygrothermal properties of the studied samples. The thermal conductivity, thermal diffusivity and thermal effusivity of the samples decreased with the increase of the fiber content in the mixtures. Except the heat capacity which increased with the increase of fiber content in the mixtures. The water absorption coefficient increased with increasing fiber content while the moisture buffer values decreased with increasing fiber content. However, all MBV values are above 2 (g/ (m2 . %H) so these samples have excellent moisture capacity. The sorption test also showed that the samples reinforced with the fibers have a capacity to absorb moisture from its environment for more than 10 days without showing mold on their surface.
References 1. Taffese, W.Z., Abegaz, K.A.: Embodied energy and CO2 emissions of widely used building materials: the Ethiopian context. Buildings 9(6), 136 (2019) 2. Umurigirwa, B.S., Élaboration et caractérisation d’un agromatériau chanvre-amidon pour le Bâtiment, Reims (2014) 3. Tomovska, R., Radivojevi´c, A.: Tracing sustainable design strategies in the example of the traditional Ohrid house. J. Clean. Prod. 147, 10–24 (2017) 4. Toure, P.M., et al.: Experimental determination of time lag and decrement factor. Case Stud. Constr. Mater. 11, e00298 (2019) 5. Kabore, A., Maref, W., Plamondon, C.: Hygrothermal performance of the building envelope with low environmental impact: case of a hemp concrete envelope. J. Phys. Conf. Ser. (2021) 6. Bah, E.-h.M., Faye, I., Geh, Z.F.: Housing Market Dynamics in Africa. Springer, London (2018). https://doi.org/10.1057/978-1-137-59792-2 7. ASTM D854-14: Standard Test Methods for Specific Gravity of Soil Solids by Water Pycnomete. ASTM International, 100 Barr Harbor Orive, PO Box C700, West Conshohocken (2014) 8. Kaboré, A., Ouellet-Plamondon, C.: Characterization of the clay and fibres for hygrothermal modelling. In: Proceedings of the Canadian Society of Civil Engineering Annual Conference 2021, CSCE 2021. Lecture Notes in Civil Engineering, vol. 248. Springer, Singapore (2023). https://doi.org/10.1007/978-981-19-1004-3_29 9. CAN/BNQ-2501-090: Sols - Determination de la limite de liquidite a l’aide de l’appareil de Casagrande et de la limite de plasticite, in Standards Council of Canada (2011) 10. Dawood, A.O., et al.: Investigation of compressive strength of straw reinforced unfired clay bricks for sustainable building construction. Civ. Environ. Eng. 17(1), 150–163 (2021) 11. ASTM C157/C157M-17: Standard Test Method for Length Change of Hardened HydraulicCement Mortar and Concrete. ASTM International, West Conshohocken, PA (2017) 12. Laou, L.: Evaluation du comportement mécanique sous sollicitations thermohydriques d’un mur multimatériaux (bois, terre crue, liants minéraux) lors de sa construction et de son utilisation. Université de Limoges (2017) 13. Lertwattanaruk, P., Choksiriwanna, J.: The physical and thermal properties of adobe brick containing bagasse for earth construction. Int. J. Build. Urban Inter Landscape Technol. (BUILT) 1, 57–66 (2011)
1252
O.-P. Claudiane and K. Aguerata
14. Emiro˘glu, M., Yalama, A., Erdo˘gdu, Y.: Performance of ready-mixed clay plasters produced with different clay/sand ratios. Appl. Clay Sci. 115, 221–229 (2015) 15. Dormann, C.F., et al.: Methods to account for spatial autocorrelation in the analysis of species distributional data: a review. Ecography 30(5), 609–628 (2007) 16. ASTME96-95: Standard test methods for water vapor transmission of materials. ASTM committe on Standards, 100 Barr Harbor Drive, West Coshohocken, PA 19428-2959, United States (1995) 17. Rode, C., et al.: Moisture buffering of building materials. Technical University of Denmark, Department of Civil Engineering (2005) 18. Ashour, T., et al.: The influence of natural reinforcement fibres on insulation values of earth plaster for straw bale buildings. Mater. Des. 31(10), 4676–4685 (2010) 19. Ramakrishnan, S., et al.: Adobe blocks reinforced with natural fibres: a review. Mater. Today Proc. 45, 6493–6499 (2021) 20. Mellaikhafi, A., et al.: Characterization and thermal performance assessment of earthen adobes and walls additive with different date palm fibers. Case Stud. Constr. Mater. 15, e00693 (2021) 21. Nshimiyimana, P., Messan, A., Courard, L.: Physico-mechanical and hygro-thermal properties of compressed earth blocks stabilized with industrial and agro by-product binders. Materials 13(17), 3769 (2020)
Novel Cement-Lime Composites with Phase Change Materials (PCM) and Biomass Ash for Energy Efficiency in Architectural Applications C. Guardia1(B)
, A. Guerrero2
, and G. Barluenga1
1 Department of Architecture, University of Alcala, Madrid, Spain
[email protected] 2 Institute of Construction Science Eduardo Torroja, CSIC, Madrid, Spain
Abstract. The improvement of energy efficiency in buildings is the goal of many new European standards and regulations. New building technologies and materials are being designed to achieve this goal by integrating new properties through new dynamic environment responsive and recycled components. In this study, lime cement pastes with phase change materials (PCM), due to their thermal storage capacity, and biomass ashes, because of their good mechanical and physical behavior, were investigated. An experimental program was carried out to assess the synergies and effects of the biomass ashes and PCM on the mineralogical, physical, mechanical and thermal performance of the mixtures. Nine cement-lime pastes were designed, where 10 and 20% cement was replaced by biomass ashes and 10% and 20% of microencapsulated paraffin waxes PCM were incorporated. Bulk density, open porosity, capillary water absorption, compressive and flexural strength, Ultrasonic Young Modulus and thermal conductivity were characterized. It was found that biomass ashes did not modify significantly the mixtures properties while increased material’s sustainability. On the hand, PCM changed the physical, mechanical, and thermal properties of the mixtures that can be advantageous for building applications as mortar renders. The larger the PCM addition, the higher the mixtures properties changes. Keywords: Biomass ash · PCM · cement-lime paste · energy storage · recycling
1 Introduction The energy efficiency guidelines established by the European Commission require the compliance with almost zero energy consumption regulations by both new and existing buildings [1, 2]. To achieve this goal, many studies have worked on the development of new building materials which can help to comply with these requirement [3]. New materials have been developed to get thermal energy storage capacity which could be used as thermal accumulators to improve thermal behavior of buildings. They © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 A. J˛edrzejewska et al. (Eds.): SynerCrete 2023, RILEM Bookseries 43, pp. 1253–1263, 2023. https://doi.org/10.1007/978-3-031-33211-1_112
1254
C. Guardia et al.
could be applied to different building elements, such as the structure [4], cooling and heating active systems [5], or the building enclosure [6]. This is the case with Phase Change Materials (PCM) which can take advantage of the latent heat required to produce their phase change at room temperature, acting as energy accumulators [7]. When PCM are conveniently encapsulated, they can be mixed with other binder materials and constitute a building component [8]. PCMs are able to change their state (from solid to liquid or from liquid to solid) at a certain melting temperature while the material absorbs or release heat remaining the material temperature constant (latent heat) [9]. Microencapsulated paraffin wax (organic PCM) mixed with different binders, such as cement-lime mortars, are showing great improvements in the energy efficiency of buildings [10]. Moreover, energy efficiency of buildings can also be achieved by incorporating recycled materials [11]. The manufacture savings, environmental impact reduction and properties improvement that these additions can provide (physical and thermal depending on their composition and chemical reaction with other components) make recycled materials an excellent choice for their use in sustainable construction [12]. Fly ashes [13] or biomass ashes [14] are good examples of recycled materials. Some studies presented biomass ashes as a good sustainable solution (reducing carbon footprint) as a partial replacement of cement maintaining materials’ initial mechanical or physical properties [15, 16]. The aim of this study was to investigate new low carbon footprint composite materials using biomass ashes and a microencapsulated PCM for improved energy efficiency in architectural applications. The combination of both components in the same mixture is presented as a novel solution of accomplishing the requirements for improving energy efficiency and sustainable cementitious material, achieving the main objective from different points. This work presents the preliminary results of a study where mineralogical (synergies between the cement-lime paste and biomass ashes and PCM), physical, mechanical and thermal properties of cement-lime pastes were characterized and analyzed. These pastes are aimed to be the active phase of rendering mortars with a binder to aggregate ratio of 1:0.5:4,5 (cement: lime: aggregate), where biomass ashes would replace cement and PCM would substitute sand.
2 Materials and Paste Compositions This paper presents a first stage of a larger study on sustainable and energy efficient mortars where nine cement-lime paste where designed. In this stage, only the paste phase of the composite was investigated. This ongoing study will continue with the complete mortar (including siliceous aggregates), although the design of the pastes will respond to the final proportions of the mortar components. The paste components were: • a cement type CEM I – 42,5R (UNE-EN 197-1) supplied by Cementos Portland Valderribas • an air lime class CL90-S, designated according to the European Standard (UNE. EN 459-1).
Novel Cement-Lime Composites with PCM and Biomass Ash
1255
• Biomass ashes with a particle size of 186μm supplied by ENCE Energía & Celulosa located in Mérida, Spain. • a microencapsulated paraffin wax (Phase Change Material) - MICRONAL ® DS 5040X, with a particle size ca. 50–300 μm and a melting point of ca. 23 °C ± 1, supplied by BASF Construction Chemicals Spain. Table 1 summarizes the nine paste compositions used in this study. A reference cement-lime paste was design (C), where 10 and 20% of cement was replaced by biomass ashes (CB10 and CB20, respectively). Afterwards 10% and 20% (by fresh mortar volume, considering a binder to aggregate ratio of 1:0.5:4,5) of PCM was added (C10 and C20 ). Namely, the amount of PCM showed in Table 1 is the same amount of PCM that is going to be added into the mortar. Finally, combinations of cement replacement by biomass ashes and PCM addition in different quantities were designed (CB1010 , CB1020 –CB2010 , CB2020 ). Table 1. Pastes compositions (components in kg). C
CB10
CB20
C10
C20
CB1010
CB1020
CB2010
CB2020
Cem I-42,5R 847
762
677
847
847
762
762
677
677
CL90-S
167
167
167
167
167
167
167
167
85
169
-
-
85
85
169
169
167
Biomass ash PCM
-
-
-
176
400
176
400
176
400
Water*
435
385
400
516
668
484
666
487
668
w/b
0.43
0.38
0.39
0.43
0.47
0.41
0.47
0.41
0.47
Consistency mm
175
170
180
180
180
175
180
185
180
* Liquid water added.
Table 1 shows the water to binder ratio (w/b) adjusted to get a plastic consistency and similar fresh workability for all the mixtures. The paste mixing procedure was composed by two stages. First, the dry components, included the biomass ashes and PCM microcapsules were mixed, and water was added afterwards. In all mixtures, the total mixing time did not exceed 5 min.
3 Experimental Program The experimental program proposed in this work, assessed mineralogical, physical, mechanical, and thermal properties of the nine cement-lime pastes (Fig. 1). 3.1 Mineralogical Characterization Testing Procedure The structural phase composition of the cement pastes was analysed by X-ray diffraction (XRD) with a Philips PW 1730 diffractometer. The experiments were done on the sample
1256
C. Guardia et al.
MIXTURES Cement replacement by Biomass ash Reference (C) (cement/lime/water)
10
%
20 10
PCM
20
CB20
%
1010 Biomass ash 1020 + PCM 2010 2020
CB10
C10 C20 CB1010
%
CB1020 CB2010 CB2020
EXPERIMENTAL PROCEDURE Mineralogical characterization
X-Ray diffranction Bulk density
Physical Characterization
Open Porosity Capillary water absorption Compressive strength Flexural strength
Mechanical Characterization Ultrasonic pulse velocity propagation (UPV) Thermal properties
Thermal conductivity
Fig. 1. Mixtures and Experimental procedures summary.
surfaces using Cu Kα radiation. A recording angle from 5° to 60° was set with a 0.02° step and a scanning rate of 0.5 s per step. Crystallographic Open Database (COD) and Joint Committee on Powder Diffraction Standards (JCPDS) databases were used for the structural phase identification. XRD analyses were also conducted on the reference cement paste surface. 3.2 Hardened Properties Testing Procedures Pastes’ hardened properties were characterized on 40 × 40 × 160 mm specimens demoulded at 24h and cured until tested at 28 days under laboratory conditions, according to the European standard UNE-EN 1015-2.
Novel Cement-Lime Composites with PCM and Biomass Ash
1257
Bulk density and open (accessible to water) porosity were calculated by using a hydrostatic scale (UNE-EN 1015-10). Capillary water absorption coefficient was measured according to UNE-EN 1015-18. Capillary water absorption was measured on 40 × 40 × 160 mm samples in contact with water on their 40 × 40 mm side while the other lateral sides of the samples were covered with a plastic film. Regarding to mechanical properties, compressive and flexural strength (UNE-EN 1015-11:2000) were tested at 28 days on standard specimens with an AUTOTEST 300/200 testing machine. A PUNDIT LAB ® Ultrasonic device equipped with P- and S-wave 250 kHz transducers was used to evaluate Ultrasonic pulse velocity propagation (UPV) of standard paste specimens at 28 days. Transmission times were identified using the Hilbert transform algorithm and Ultrasonic Young modulus (E) was calculated according to Eq. 1 and Eq. 2: 2 ρ · VS2 (1 + υ) (1) E= 106 (2) υ = 0.5 · Vp2 − Vs2 Vp2 − Vs2 where ρ is the apparent (bulk) density (g/cm3 ), ν is the Poisson ratio Coefficient and VP and VS are the P- and S- wave velocity (m/s), respectively. This experimental method has been previously described elsewhere [8]. The thermal conductivity of the nine mixtures under 23 °C, when PCM is under melting temperature and therefore is solid, were measured on cylindrical hardened samples with 10 cm of diameter and 2.5 cm height. A portable heat transfer analyser ISOMET 2114- Applied Precision equipped with a surface probe with a measurement range of thermal conductivity from 0.30 to 3.0 W/mK was used.
4 Experimental Results and Analysis 4.1 Hardened Properties Figure 2 shows the X-ray diffraction patterns pastes after 28 days of hydration. The main compounds found were: P [Ca(OH)2 ]; C [CaCO3 ]; E [Ca6 Al2 (SO4 )3 (OH)12. 26H2 O]; Q [SiO2 ]; A [Ca3 SiO5 ]; B [α and β-Ca2 SiO4 ]; g [of C–S–H gel (Ca1.5 SiO3.5 .xH2 O)]. After 28 days of hydration the main hydrated compounds detected in the reference sample (binder: cement + lime - CaO; C-28d)), were portlandite [Ca (OH)2 ] and ettringite [Ca6 Al2 (SO4 )3 (OH)12.26 H2 O]. Calcite [CaCO3 ] was also detected. Alite [Ca3 SiO5 ] and α and β-Ca2 SiO4 reflections from the anhydrous cement remained. On the other hand, the addition of PCM (10 and 20%) and/or Biomass ash (10 and 20%) to the reference binder doesn’t produce negative effect in the microstructure of reference material. The presence of crystalline phases as ettringite and portlandite at 28d of hydration remain in all the matrices. It is interesting to notice that the addition of 20% Biomass ash and PCM (10 and 20%) increase the hydration reaction of samples. The intensity of anhydrous phases of cement, Ca3 SiO5 and α and β-Ca2 SiO4 decrease and the peak 29.39 28, due to the CaCO3 loss of crystallinity and a widening of 29.01 and 30.01 28 due to the presence of C–S–H gel (Ca1.5 SiO3.5 .xH2 O) amorphous phase.
1258
C. Guardia et al.
Fig. 2. X-ray diffraction patterns pastes after 28 days of hydration.
Table 2. Physical properties of the mixtures. C
CB10 CB20 C10
C20
CB1010 CB1020 CB2010 CB2020
Bulk density kg/m3
1550
1590
960
1210
O.Porosity %
35.99 33.74 35.8
Capillary 0.62 kg/m2 min0.5
0.56
1520
0.7
1220
930
1190
900
26.76 20.91 26.27
22.11
27.9
24.2
0.23
0.18
0.34
0.23
0.15
0.32
Physical properties of the pastes are summarized in Table 2. Bulk density values varied between 900 kg/m3 (CB2020 ) and 1590 kg/m3 (CB10). The reference paste and the mixtures with biomass ashes presented the highest values. However, the addition of 20% of PCM decreased bulk density values. Reference paste (C) was the mixture that presented the highest Open Porosity value (35.99%) while C20 had the lowest with 20.9%. Capillary water absorption values are also recorded in Table 2. These values varied between a maximum of 0.7 kg/m2 min0.5 (CB20) and 0.15 kg/m2 min0.5 for C20 . Figure 3 presents the effect of PCM addition on the capillary water absorption and open porosity values. It can be seen that, the addition of PCM influenced directly on these physical properties of the mixtures, due to its activity as densifier of the paste microstructure and its lack of open porosity. The higher the PCM amount, the lower the capillary water absorption and open porosity values. These results corresponded to the results obtained in a previous study on PCM mortars [8, 10].
Novel Cement-Lime Composites with PCM and Biomass Ash
Capillary water Absorption Kg/m2min0.5
0,8
1259
y = 0,035x - 0,6157 R² = 0,9548
0,7 0,6
PCM0
0,5 0,4
PCM10
0,3 0,2 PCM20
0,1 0 0
5
10
15 20 25 Open Porosity %
30
35
40
Fig. 3. Effect of PCM addition on capillary water absorption and open porosity in cement-lime pastes.
Table 3. Mechanical properties of the mixtures. C Com. Str. MPa Flexural Str. MPa E GPa
CB10 CB20 C10
42.60 54.21 3.04
42.82
C20
CB1010 CB1020 CB2010 CB2020
23.07 12.45 22.52
10.70
20.38
9.30
2.99
3.69
3.37
2.25
4.11
2.76
4.17
2.76
12.86 13.54
11.70
6.42
3.41
6.12
3.08
5.93
2.69
Table 3 presents the mechanical properties obtained of the nine different mixtures. Compressive strength values ranged between a maximum of 54.21 MPa (CB10) and a minimum of 9.30 MPa (CB2020 ). It can be seen that mixtures C, CB10 and CB20, mixtures with de replacement of cement by biomass ashes but without PCM addition did not show significant differences. As expected, when PCM was added, the compressive strength decreased [7], around 50% for 10% of PCM and 80% for 20% of PCM addition. Table 3 also displays flexural strength values at 28 days. CB2010 showed the highest value with 4.17 MPa while C20 presented the lowest one, 2.25MPa. According to what happened with compressive strength, the mixtures without PCM presented the highest values (C, CB10 and CB20). However, in this case, the addition of 10% PCM increased flexural strength regarding to the reference paste or mixtures with biomass ashes: 3.37MPa, 4,11MPa y 4,17MPa for C10 , CB1010 y CB1020 respectively. On the contrary, the addition of 20% PCM reduced flexural strength. Regarding to Ultrasonic Young modulus (E) values (Table 3), CB1020 had the lowest value (3.41 GPa) and CB10 the highest (13.54 GPa). Once again, the biomass ashes addition did not affect significantly E values and mixtures with PCM had lower values. Figure 4 shows the relation between compressive strength and Ultrasonic Young Modulus. It can be observed that the PCM addition produced three groups of pastes:
1260
C. Guardia et al.
Compressive Strength MPa
60
PCM0
y = 3,737x - 0,8521 R² = 0,9798
50 40 30
PCM10
20 10
PCM20
0 0
5
E GPa
10
15
Fig. 4. Influence of PCM on compressive strength and Young Modulus in cement-lime pastes.
without PCM, with 10% of PCM and with 20% of PCM. More specifically, when 20% of PCM was added the compressive strength remainded under 30 MPa and the Ultrasonic Young Modulus under 10 GPa. Therefore, it can be said that the higher the addition of PCM was, the lower were the compressive strength and Ultrasonic Young Modulus of the mixtures at 28 days (Fig. 4). Table 4. Thermal conductivity of the nine mixtures under solid conditions.
λ* W/mK
C
CB10
CB20
C10
C20
CB1010
CB1020
CB2010
CB2020
0.69
0.72
0.63
0.47
0.32
0.57
0.30
0.68
0.33
* Thermal Conductivity was calculated with PCM in solid state.
Table 4 presents thermal conductivity values of the nine pastes measured under 23 °C of temperature, which corresponds to PCM melting point, and therefore PCM was in solid state. Thermal conductivity varied between 0.3 W/mK (CB1020 ) and 0.72 W/mK (CB10). Mixtures with biomass ash but without PCM did not show significant thermal conductivity variations regarding to the reference mixture. It can be said that biomass ashes did not affect the thermal performance of the mixtures for temperature under 23 °C. However, PCM addition decreased thermal conductivity. Pastes with 20% of PCM addition presented the lowest values of thermal conductivity. The difference between C10 and C20 was 30% lower. However, when biomass ash and PCM were mixed, the difference between 10% or 20% of PCM increased. The difference between mixtures with biomass ash and 10% of PCM and 20% increased to a 50%. Figure 5 relates thermal conductivity (PCM in solid state) and open porosity. It can be seen that the higher the PCM addition, the lower the thermal conductivity and the open porosity. Therefore, the reduction of open porosity by the PCM also caused a thermal conductivity decrease of pastes.
λ
W/mK
Novel Cement-Lime Composites with PCM and Biomass Ash
0,8 0,7 0,6 0,5 0,4 0,3 0,2 0,1 0,0
PCM0
PCM10
y = 0,0257x - 0,2026 R² = 0,7417
0
5
10
15
1261
PCM20 20
25
30
35
40
Open Porosity % Fig. 5. Thermal conductivity of pastes vs. open porosity.
5 Conclusions In this preliminary study on the effect of microencapsulated paraffin wax PCM and biomass ashes on cement-lime pastes properties, nine different paste mixtures were investigated. The combined effect of the replacement of different amounts of cement by biomass ashes and the addition of different amounts of a Phase Change Materials in the properties was carried out. The main conclusions of this study are summarized next: • The replacement of cement by biomass ashes did not affect significantly the mineralogical, physical, mechanical and thermal properties of the mixtures. Therefore, the recycling of biomass ashes can be considered a positive effect to obtain sustainable cement-lime mortars. • The addition of PCM influenced the physical, mechanical and thermal properties of the mixtures. • The microcapsules of the Phase Change Materials worked as a densifier of the paste matrix, reducing the pore interconnection and decreasing open porosity and capillary water absorption values of the pastes. • PCM addition reduced paste compressive and flexural strength and Young modulus, due to the effect of microcapsules as microdefects in the paste matrix. The larger the amount of PCM the lower the mechanical properties. • However, PCM addition showed a deep impact on paste thermal conductivity. Mixtures with 20% of PCM, with and without biomass ash and independently to the amount of ash, reduced 50% thermal conductivity regarding reference paste values for pastes. The chemical synergies between all the components still have to be studied. A physical and mechanical characterization of a mortar prepared with the paste presented in this preliminary study is going on. The thermal behavior of the mixtures incorporated in multilayer enclosure applications under different climatic conditions is going to be studied.
1262
C. Guardia et al.
Acknowledgements. Some of the components were supplied by Omya Clariana S.L, BASF Construction Chemicals España S.L, Cementos Portland Vaderrivas and ENCE Energía & Celulosa. The authors acknowledge the financial support provided by the projects: CALTH3D (TED2021-132585B-I00), funded by MCIN/AEI/10.13039/501100011033 and the European Union “NextGenerationEU”/PRTR; the European Union by the action HORIZON-TMA-MSCASE_2021 “BEST-Bio-based Energy-efficient materials and Structures for Tomorrow” (grant number 101086440) and the Research Program for the Promotion of Young Researchers, co-funded by Comunidad de Madrid and the University of Alacla (Spain), as part of the project IndoorComfort (CM/JIN/2019-46). Finally, Cynthia Guardia acknowledge the financial support provided by the University of Alcala (Spain), the Spanish Ministry of Universities and the European Union “NextGenerationEU”- Margarita Salas.
References 1. https://ec.europa.eu/info/energy-climate-change-environment/topics/energy_en 2. https://ec.europa.eu/info/funding-tenders/find-funding/eu-funding-programmes/recoveryand-resilience-facility_en 3. Palomar I., Barluenga, G., Ball, R.J., Lawrence, M.: Laboratory characterization of brick walls rendered with a pervious lime-cement mortar. J. Build. Eng. 23, 241–249 (2019). https://doi. org/10.1016/j.jobe.2019.02.001. ISSN 2352-7102 4. Kim, H.G., Qudoos, A., Jeon, I.K., Woo, B.H., Ryou, J.S.: Assessment of PCM/SiC-based composite aggregate in concrete: energy storage performance. Constr. Build. Mater. 258, 119637 (2020). ISSN 0950-0618. https://doi.org/10.1016/j.conbuildmat.2020.119637 5. Crespo, A., et al.: Thermal performance assessment and control optimization of a solar-driven seasonal sorption storage system for residential application. Energy 263(Part A), 125382 (2023). ISSN 0360-5442. https://doi.org/10.1016/j.energy.2022.125382 6. Mankel, C., Caggiano, A., Koenders, E.: Thermal energy storage characterization of cementitious composites made with recycled brick aggregates containing PCM. Energy Build. 202, 109395 (2019). ISSN 0378-7788. https://doi.org/10.1016/j.enbuild.2019.109395 7. Venkateswara, R.V., Parameshwaran, R., Vinayaka, R.V.: PCM-mortar based construction materials for energy efficient buildings: a review on research trends. Energy Build. 158, 95–122 (2018). https://doi.org/10.1016/j.enbuild.2017.09.098 8. Guardia, C., Barluenga, G., Palomar, I., Diarce G.: Thermal enhanced cement-lime mortars with phase change materials (PCM), lightweight aggregate and cellulose fibers. Constr. Build. Mater. 221, 586–594 (2019). ISSN 0950-0618. https://doi.org/10.1016/j.conbuildmat.2019. 06.098 9. Liu, L., Hammami, N., Trovalet, L., Bigot, D., Habas, J.-P., Malet-Damour, B.: Description of phase change materials (PCMs) used in buildings under various climates: a review. J. Energy Storage 56(Part A), 105760 (2022). ISSN 2352-152X. https://doi.org/10.1016/j.est. 2022.105760 10. Guardia, C., Barluenga, G., Palomar, I.: Evaluation of the energy storage capacity of Phase Change Material cement-lime mortars by using heat flux meters and ultrasonic pulse transmission. J. Energy Storage 50, 104674 (2022). ISSN 2352-152X. https://doi.org/10.1016/j. est.2022.104674 11. Rodríguez, J., Frías, M., Tobón, J.I.: Eco-efficient cement based on activated coal washing rejects with low content of kaolinite. Constr. Build. Mater. 274, 122118 (2021). ISSN 09500618. https://doi.org/10.1016/j.conbuildmat.2020.122118
Novel Cement-Lime Composites with PCM and Biomass Ash
1263
12. Farinha, C.B., De Brito, J., Veiga, R., Fernández, J.M., Jiménez, J.R., Esquinas, A.R.: Wastes as aggregates, binders or additions in mortars: selecting their role based on characterization. Materials 11(3), 453 (2018). https://doi.org/10.3390/ma11030453 13. Marieta, C., Guerrero, A., León, I., Municipal solid waste incineration fly ash to produce ecofriendly binders for sustainable building construction. Waste Manage. 120, 114–124 (2021). ISSN 0956-053X. https://doi.org/10.1016/j.wasman.2020.11.034 14. Munawar, M.A., et al.: Challenges and opportunities in biomass ash management and its utilization in novel applications. Renew. Sustain. Energy Rev. 150, 111451 (2021). ISSN 1364-0321. https://doi.org/10.1016/j.rser.2021.111451 15. Velay-Lizancos, M., Azenha, M., Martínez-Lage, I., Vázquez-Burgo, P.: Addition of biomass ash in concrete: effects on E-Modulus, electrical conductivity at early ages and their correlation. Constr. Build. Mater. 157, 1126–1132 (2017), ISSN 0950-0618. https://doi.org/10.1016/ j.conbuildmat.2017.09.179 16. Farinha, C.B., de Brito, J., Veiga, R.: Influence of forest biomass bottom ashes on the fresh, water and mechanical behaviour of cement-based mortars. Resour. Conserv. Recycl. 149, 750–759 (2019). ISSN 0921-3449. https://doi.org/10.1016/j.resconrec.2019.06.020
Cement Based Materials with PCM and Reduced Graphene Oxide for Thermal Insulation for Buildings Edurne Erkizia1(B) , Christina Strunz2 , Jean-Luc Dauvergne3 , Guido Goracci4 , Ignacio Peralta5,6,7 , Ángel Serrano3 , Amaya Ortega8 , Beatriz Alonso8 , Francesca Zanoni9 , Michael Düngfelder2 , Jorge S. Dolado4 , Juan Jose Gaitero1 , Christoph Mankel5 , and Eduardus Koenders5 1 TECNALIA, Basque Research and Technology Alliance (BRTA), 48160 Derio, Spain
[email protected]
2 NETZSCH Geratebau GMBH, Wittelsbacherstr. 42, 95100 Selb, Germany 3 Centre for Cooperative Research on Alternative Energies (CIC energiGUNE), Basque
Research and Technology Alliance (BRTA), Alava Technology Park, Albert Einstein 48, 01510 Vitoria-Gasteiz, Spain 4 Centro de Física de Materiales (CSIC, UPV/EHU) Materials Physics Centre (MPC), 20018 San Sebastián, Spain 5 Institut für Werkstoffe im Bauwesen, Technische Universität Darmstadt, Germany, Franziska-Braun-Straße 3, 64287 Darmstadt, Germany 6 Centro de Investigación de Métodos Computacionales (CIMEC), UNL-CONICET, Predio CONICET “Dr. Alberto Cassano”, 3000 Santa Fe, Argentina 7 Laboratorio de Flujometría (FLOW), FRSF-UTN, Lavaisse 610, 3000 Santa Fe, Argentina 8 GRAPHENEA SA, 20009 San Sebastian, Spain 9 SPHERA ENCAPSULATION SRL, 37069 Dossobuono, Verona, Italy
Abstract. Energy demand for heating and cooling represents a large part of building´s (residential and non-residential) energy consumption around the world. Development of thermal insulating construction elements with thermal energy storage and release capacity could be one way of reducing this consumption while maintaining thermal comfort inside the buildings. Using phase change materials (PCMs) as thermal storage/release materials for “porous” cement-based construction elements is a possible solution. However, the relatively low thermal conductivity of the cement matrix could impair the efficient transfer of the heat to the PCM reducing its effectivity. Addition of thermal and electrically conductive nanoparticles such as graphene-based particles could improve enough the thermal and electrical conductivity but maintain a good energy storage capacity. In this study the production of cement pastes with different dosage of PCMs (20% and 40% in volume) and reduced graphene oxide will be described. Furthermore, the characterization of their thermal and electrical conductivity, latent heat and thermal diffusivity will also be shown and discussed. Keywords: cement-based composites · thermal energy storage · phase change materials · reduced graphene oxide
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 A. J˛edrzejewska et al. (Eds.): SynerCrete 2023, RILEM Bookseries 43, pp. 1264–1276, 2023. https://doi.org/10.1007/978-3-031-33211-1_113
Cement Based Materials with PCM
1265
1 Introduction Around 40% of the total energy usage is consumed to meet building energy demands for heating and cooling [1]. In order to enhance the energy efficiency of buildings a lot of research has been carried out to improve and develop novel materials and concepts, which have better insulation properties or take advantage of the material’s thermal properties (such as sensible heat and latent heat) [1, 2]. In this regard, researchers have studied the use of concrete due to its high thermal mass [3] and materials such as phase change materials (PCMs) for their latent heat properties [2]. The H2020 project NRGSTORAGE unites both these materials and proposes the development of a PCM-based cementitious product by taking advantage of the porous nature of a foam concrete for insulation capacity and the latent heat of organic PCMs for energy storage and preservation of the indoor temperature. Furthermore, graphene particles have also been added in order to improve the relatively low thermal conductivity of the cement matrix and the electrical conductivity of the composite to make the PCM-based cementitious product an active system via thermoelectrical coupling. In this work, an initial study carried out in the project, which consists in the development of cement pastes with different additions of microencapsulated PCM (Rubitherm24) and reduced graphene oxide (GO-NRG), will be described. Moreover, their morphological and comprehensive thermal characterization by differential scanning calorimetry (DSC), transient plane source method (hot-disk), Laser Flash and electrical conductivity by broadband dielectric spectrometer will be shown and discussed.
2 Experimental Section 2.1 Materials Ordinary Portland cement (CEM I 52.5R) was provided by Cementos Lemona. The metakaolin (Centrilit NCII, an amorphous aluminosilicate) was obtained from MCBauchemie. The admixtures Centrament Rapid 500 (accelerator), Centrament Stabi 520 (stabilizer, water retainer) and Powerflow 3195 superplasticizer were also provided by MC-Bauchemie. The phase change material used was RT24 (paraffin-based wax with melting temperature at 24 ºC and heat storage capacity of 160 kJ/Kg according to the supplier) purchased from Rubitherm GmbH and encapsulated by Follmann GmbH & Co. KG (MPCM). The microcapsules were in a form of a powder and based on the technology SmartCaps, commercialized under the name Folco SmartCaps®. The shell was organic and composed from melamine-formaldehyde obtained through polymerization technique. The reduced graphene oxide (GO-NRG) was provided by Graphenea as a 0.89 wt% water dispersion. First, graphene oxide was produced via Hummers modified method [6]. Then graphene oxide was thermally reduced in a reflux system to produce GO-NRG. 2.2 Paste and Specimen Preparation Different cement pastes with 20% and 40% in volume of RT24 were prepared. Furthermore, pastes containing 0.3 wt% GO-NRG based on weight of binder (bwb) and 20%
1266
E. Erkizia et al.
and 40% in volume of RT24 were also cast. In Table 1 the compositions of the different type of pastes are shown. When adding the microcapsules and GO-NRG, to obtain a workable paste, large amounts of superplasticizer were used (Table 1). The general mixing procedure of the different pastes was as follows. First, the cement and metakaolin (Centrilit NCII) were mixed for 1 min at 300 rpm. Subsequently, water with the superplasticizer was added and mixed for another 1 min at 750 rpm. Next the stabilizer was added by mixing the mixture for 30 s at 750 rpm, followed by the accelerator, mixing the paste for 30 more seconds at 750 rpm. At this point, for the reference specimens the paste was moulded. In the case of the pastes with the microencapsulated RT24, the microcapsules were added after the mixing of the accelerator, and the whole mixture was mixed at 300 rpm for 1 more minute. For the pastes that contain the GONRG, the particles were added as a dispersion in the step of the water addition. In fact, the water needed for the mixture was provided by the GO-NRG dispersion to obtain a 0.3wt% bwb of GO-NRG in the paste. Once all the components were mixed, the pastes were moulded, wrapped with a cling film, demoulded after one day, except for the case of pastes with 40% vol of MPCM which required two days to hardened, wrapped again in cling film and cured for 28 days in the lab (which was at 21.0 ºC and 48.9% relative humidity) after which the different characterizations were carried out. Each batch was large enough to prepare different size specimens to carry out hot-disk and DSC measurements, as well as Laser Flash analysis, porosimetry measurements, and scanning electron microscope (SEM) characterization. Specimens for electrical measurements by broadband dielectric spectrometer were also prepared. Table 2 shows the dimensions of the different specimens, and which characterization they were used for. 2.3 Characterization Techniques Samples were characterized by DSC in order to assess phase transition temperatures and melting enthalpies (latent heat capacity). The used DSC was a NETZSCH DSC 204 F1 Phoenix® device. The measurements were run at two different heating rates, 5 ºC/min and 0.25 ºC/min. Thermal conductivity of the different pastes was characterized by the transient Hot Disk method. The size of the sensor and samples, the heating power and the measurement time were carefully chosen according to the standard Hot Disk DIN-EN-ISO 220072:2021-05 [4]. Thus, the measurements were carried out with a Kapton 5501 sensor (radius 6.403 mm) subjected to a heating power of 80 to 100 mW during 80 s. The MPCM pastes were tested inside a climatic chamber Binder MKFT 115 E5 (±0.1 ºC with a relative humidity fixed at 60%RH for all the experiments). The temperatures at which the thermal conductivities were measured were 5 ºC (T1) and 35 ºC (T2). The thermal diffusivity of the cement pastes, stored under laboratory condition, was measured by using the NETZSCH Laser Flash Apparatus (LFA 467 HyperFlash®). The Flash method is the reference method for determination of the thermal diffusivity of a material. However, as the Hot Disk method allows for measuring of the thermal conductivity and a simultaneous estimation of the diffusivity, with a slightly lower accuracy for the latter (2% to 5% and 5% to 10%, respectively [4, 7]), the results obtained with both approaches were compared and cross-validated at 5 °C and 35 °C.
187.63
438.15
Cem I 52.5R
Centrilit NCII
Water
NRGS-20RT24
112.62
38.37
Stabilizer
23.03
** GO-NRG is a 0.89wt% water dispersion
* The required water (38.44 g for w/b = 0.33) was added with the GO-NRG dispersion
30.78
38.37
6.64
26.55
144.42
6.64
11.03
26.55
15.96
187.63 *
Superplasticizer
351.99
262.85
NRGS-0.3GO-NRG 1140.09
Accelerator
21.2
NRGS-40RT24 683.92
441.92
175.99
350.39
150.12
911.89
GO-NRG 0.3wt%bwb (0.89wt% water dispersion)
MPCM-RT24
REF
1140.09
Components (Kg/m 3 )
Table 1. Composition of the pastes NRGS-20RT24-0.3GO-NRG
30.78
21.25
21.2
353.54
175.99
*
150.12
911.89
NRGS-40RT24-0.3GO-NRG
23,03
178.62
15,96
265.16
351.99
*
112.62
683.92
Cement Based Materials with PCM 1267
1268
E. Erkizia et al. Table 2. Dimensions of the different cast specimens
Specimen dimension Hight (mm)
Characterization Diameter (mm)
3
5
Differential Scanning Calorimetry
20
40
Hot disk, mercury porosimetry, SEM
20
12
Laser Flash technique (LFA)
5
15
Electrical Conductivity measurements
The electrical conductivities were measured by broadband dielectric spectrometer Novocontrol Alpha-A + in the frequency range of 10–1 –106 Hz. MPCM and GO-NRG were measured in powder, sandwiched between parallel gold-plated electrodes, with a diameter of 20 mm and a sample thickness of ~0.5 mm. Prior to data acquisition, the samples were kept inside the instrument at 42 ºC during 10 min. Finally, the isothermal frequency scans were performed on heating every 5 degrees over the 12–42 ºC temperature range. SEM images were obtained with a “FEI Quanta 200 ESEM” microscope. Furthermore, this equipment includes a Fluorescence Dispersive Energy Spectrophotometer (EDAX) to analyze the elemental composition of a sample. The porosity and pore size distribution of the different pastes were characterized by mercury intrusion porosimetry (MIP). The equipment used is an Autopore III from Micromeritics Instrument Corp.
3 Results and Discussion All the pastes were mixed with a water-binder ratio (w/b) of 0.33. This posed a difficulty: when mixing the pastes with the microencapsulated RT24 and GO-NRG, an increasing amount of superplasticizer was required, in order to obtain a good workability of the paste. Moreover, in the case of the pastes with the 40% in volume of MPCM, the addition of the superplasticizer was very large which delayed the hardening of those pastes for demoulding to one more day of curing. Nevertheless, after 28 days all the pastes were hardened. It was also observed that the addition of GO-NRG to the paste (with MPCM or without) reduced the workability slightly and made the paste drier. This is quite usual when nano/micrometer size particles are added due to their large surface area/aspect ratio. 3.1 Morphological Characterization As said previously, the distribution of the microencapsulated RT24 in the paste and its morphology was characterized by SEM. In Fig. 1 the SEM images of the metakaolin (Centrilit NCII) (Fig. 1a), the microencapsulated RT24 (Fig. 1b) and the dispersed GONRG (Fig. 1c) are shown. In the case of metakaolin (Fig. 1a) particles vary widely in size, ranging from a few micrometres to more than a hundred micrometres. Morphology also varies considerably but in general particles show low sphericity with elongated shape. The smallest particles
Cement Based Materials with PCM
1269
Fig. 1. SEM image of metakaolin (Centrilit NCII) (1a), microencapsulated RT24 (2b) and the GO-NRG (1c and 1d).
show the typical flaky morphology of metakaolin. In the case of the microencapsulated RT24, microcapsules smaller than 5 µm can be observed and these microcapsules tend to group forming larger spherical particles. In the case of the GO NRG particles (Fig. 1c and 1d) flake/plate like particles can be observed (Fig. 1c). These plates show a small thickness and several microns in width, but they tend to agglomerate giving larger size particles (Fig. 1d). Figure 2 shows SEM images of the reference paste (Fig. 2a), and pastes with 20% (Fig. 2b) and 40% (Fig. 2c) in volume of microencapsulated RT24. In the latter images, besides the paste, the metakaolin as well as the microcapsules can be observed. The microcapsules can be observed as dark colour spheres. Both pastes are overcrowded with microcapsules, but they seem to be well dispersed and no agglomerations (like the ones seen in Fig. 1b) can be observed in the paste. Furthermore, the microcapsules seem to be embedded within the paste, which indicates a good interaction between the shell and the paste. Elemental analysis (Fig. 3) shows a large carbon element signal which confirms that the microcapsules contain the organic PCM. In the case of pastes which also contain the GO-NRG besides the microencapsulated RT24, SEM images show similar morphology and a good dispersion of the MPCMs at both dosages (Fig. 4). However, we were not able to detect the GO-NRG particles and their dispersion in the paste. Porosity is an important parameter of a cement paste. It can give an indication of the strength of the paste, how the hydration of the cement has been carried out, as well as the permeability to the ingress of deleterious ions.
1270
E. Erkizia et al.
Fig. 2. SEM images of just the paste (scale bar 200 µm) (2a), paste with 20% vol of microencapsulated RT24 (scale bar 20 µm) (2b) and paste with 40% vol MPCM (scale bar 20 µm) (2c).
Fig. 3. Elemental analysis of point 1 in the sample. The analysis shows a large peak of carbon which is due to the paraffin phase change material.
Fig. 4. SEM images of pastes with 20% in volume of RT24 and 0.3 wt% of GO-NRG (a) and with 40% in volume of RT24 and 0.3 wt% of GO-NRG (b).
Furthermore, porosity and pore size distribution can have an effect in the thermal properties of the paste, which is what we are focus on this study, therefore it is an important parameter to characterize. The effect of the different additions on the porosity of the paste was characterized by MIP (Fig. 5). In both types of pastes (with and without GO-NRG) as the volume of the microcapsules is increased, the porosity increases and the pore diameter distribution shifts towards smaller pore sizes in comparison to the reference. This could be due to a bad interaction between the paste and the microcapsules
Cement Based Materials with PCM
1271
although SEM images do not clearly show any space between the polymeric shell and the paste. The raise in porosity could also be due to the large amount of superplasticizer used to obtain a workable paste with the large quantity of MPCMs. Superplasticizers are surfactant-like molecules that are known to form air bubbles. However, the addition of the GO-NRG does not seem to affect the porosity too much.
Fig. 5. Comparison of pore diameter distribution of the reference paste, and pastes with 20% and 40% vol of microencapsulated RT24 (a). Comparison of reference paste, paste with 0.3wt% of GO-NRG and pastes with 0.3 wt% of GO-NRG and 20% and 40% vol of microencapsulated RT24 (b).
3.2 Characterization of Thermal and Electrical Properties Table 3 depicts the onset temperature, peak temperature and melting enthalpy of the microencapsulated RT24, the cement pastes with 20% and 40% vol of MPCM RT24, as well as the same pastes with addition of reduced graphene oxide (GO-NRG). The onset temperature describes, at which temperature the PCM starts to melt while the peak temperature describes the temperature at which all PCM is molten. The higher the heating rate, the peak and onset temperature shift to higher temperatures. Simultaneously, higher heating rates gives higher accuracy in melting enthalpies as these are very sensitive to the signal to noise ratio. For accurate PCM investigation of onset –and peak temperature, it is important to measure closest to the thermal equilibrium. The onset and peak temperatures were evaluated from the DSC curve carried out at 0.25 ºC/min, following the procedure for measuring PCM material described in RAL-GZ 896 and IEA task 42 / Annex 29 Standard [8]. This attempt works quite well for pure RT24, however, for microencapsulated RT24 or cement paste including 20% or 40% MPCM by volume, the signal to noise ratio becomes very poor at lower heating rates. Therefore, it was decided to measure the pastes at 0.25 ºC/min and use both onset and peak temperature for comparison, while the melting enthalpy/ latent heat was investigated at 10 ºC/min for the microencapsulated RT24, and at 5 ºC/min heating rate for the different cement pastes. The results show peak temperatures for all the pastes very close to the ones for the pure microencapsulated RT24 and the values provided by the supplier. However, there is a higher deviation in the onset temperatures caused by the poor signal to noise ratio at low
1272
E. Erkizia et al.
heating rates and a simultaneously high melting range of the MPCM itself. Therefore, focus is placed on the peak temperatures for comparison of the pastes. Regarding the latent heat, the microencapsulated RT24 has a latent heat of 133.9 J/g, slightly lower than 160 J/g of the virgin Rubitherm PCM given by the provider. The latent heat enthalpy obtained from the 40% vol MPCM is double the one obtained with 20% vol MPCM which is the expected behaviour when increasing the MPCM amount twofold. Table 3. Onset temperature, peak temperature and melting enthalpy of the phase change of microencapsulated RT24 and the cement pastes with 20% and 40% vol and GO-NRG. Sample
TOnset 0.25 ºC/min (°C)
TPeak 0.25 ºC/min (°C)
Melting Enthalpy (J/g)
Microencapsulated RubiTherm24
17.80
24.95
133.90 (@10K/min)
NRGS-20RT24
15.97
25.00
13.37 (@5K/min)
NRGS-40RT24
14.77
25.00
33.50 (@5K/min)
NRGS-20RT24–0.3GO-NRG
15.93
24.80
14.37 (@5K/min)
NRGS-40RT24–0.3GO-NRG
15.87
25.00
30.82 (@5K/min)
In Fig. 6, the thermal conductivity of the reference paste, pastes with 20% and 40% vol RT24 and same pastes with 0.3 wt% of GO-NRG at 5ºC and 35ºC are shown. In all cases, the thermal conductivity decreases with the increase MPCM/cement paste ratio as expected since the PCM have lower thermal conductivity than the cement paste (see table 4). Furthermore, the addition of GO-NRG has no significant impact on thermal conductivity of the microencapsulated RT24 cement pastes. Although, it slightly reduces the conductivity of the reference paste. Concerning the impact of the measurement temperature, thermal conductivities decrease when temperature increase as observed in Fig. 6a and 6b (and Table 5). This behaviour, already reported in literature [5], can be related to the similar behaviours of both MPCM and cement paste. Table 4. Thermal conductivities of pure cement paste at w/b ratio of 0.33, reduced graphene oxide and microencapsulated RT24. Material
Thermal conductivity (W/m/ºC) 5 °C
Thermal conductivity (W/m/ºC) 35 °C
Pure cement paste w/b = 0.33
0.9182
0.8655
Microencapsulated RT24
0.145
0.111
GO-NRG
0.518 (@20ºC)
Thermal diffusivity of the reference cement pastes, as well as the cement pastes containing microencapsulated PCM and/or GO NRG at 5ºC and 35ºC, was investigated
Cement Based Materials with PCM
1273
Fig. 6. Thermal conductivity of pastes with 20% and 40% in vol. of microencapsulated RT24 and the same pastes with 0.3wt% bwb GO-NRG measured at 5ºC (a) and 35ºC (b).
Table 5. Thermal conductivity (W/m/ºC) values of the reference and pastes with MPCMs and GO-NRG graphene measured at 5ºC and 35ºC. Thermal Conductivity (W/m/ºC)
REF 5 ºC
NRGS-20RT24 5 ºC
NRGS-40RT24 5 ºC
REF 35 ºC
NRGS-20RT24 35 ºC
NRGS-40RT24 35 ºC
Without GO-NRG
0.918
0.750
0.591
0.866
0.698
0.562
With GO-NRG
0.869
0.723
0.591
0.838
0.698
0.476
via the Laser Flash method (Fig. 7 and Table 6) and compared for validation with the estimations from the Hot Disk method. Figure 7a shows the thermal diffusivity measured at 5 ºC and 7b measured at 35 ºC. For a higher quantity of microencapsulated RT24, the thermal diffusivity of the pastes decreases. Furthermore, in general terms, the addition of reduced graphene oxide does not have much effect in the diffusivity. There is only one exception, the paste with 40% in vol of MPCM, where the addition of the GO-NRG marginally improves the thermal diffusivity (at 35 ºC). Table 6 also shows the thermal diffusivities estimated by the thermal conductivity measurements carried out by the transient hot disk method. This estimation was carried out by following the reference 7. It can be noted that the order of magnitudes and trends derived at hot-disk were confirmed by the Laser Flash results. The differences between the two methods are on average 4% and do not exceed 11%, i.e., in the range of accuracy of the Hot Disk method, which is good results, considering the 2 different measurement methods, different sample dimensions and conditioning. Electrical conductivity of the samples was also characterized in this study by broadband dielectric spectrometer since one of the reasons to add graphene (GO-NRG) was to improve the electrical conductivity of the pastes. In Table 7, the electrical conductivities of the microencapsulated RT24, cement paste and pastes with different volume of microencapsulated RT24 and GO-NRG addition are shown (no data is shown for the paste with 20% vol of microencapsulates RT24 because the sample broke). The GO-NRG particles show an electrical conductivity of 1.0e-2 S/cm at 27 ºC, whereas the cement paste has a much lower conductivity (2.1e-10 S/cm, 8 orders of
1274
E. Erkizia et al.
Fig. 7. Thermal diffusivity measured by Laser Flash method of different pastes at 5 ºC (a) and at 35 ºC (b)
Table 6. Thermal diffusivity of the different pastes at 5 ºC and 35 ºC measured by Laser Flash method and estimated by the thermal conductivity results obtained by Hot Disk. Sample
Therm. Diff. (LFA) (mm2 /s)(5 ºC)
Therm. Diff. (LFA) (mm2 /s) (35 ºC)
Therm. Diff. (HD) (mm2 /s) (5 ºC)
Therm. Diff. (HD) (mm2 /s) (35 ºC)
REF
0.3950
0.3723
0.5042
0.3500
REF-0.3GO-NRG
0.3710
0.352
0.4709
0.3675
NRGS-20RT24
0.2900
0.274
0.3381
0.2921
NRGS-20RT24-0.3GO-NRG
0.2950
0.249
0.3145
0.2789
NRGS-40RT24
0.2217
0.1163
0.2315
0.2578
NRGS-40RT24-0.3GO-NRG
0.2110
0.194
0.2497
0.2148
Table 7. Electrical conductivity of the different samples at 7 ºC (300 K) Sample
Electrical conductivity (S/cm) @ 7 ºC (300 K)
GO-NRG
~1.0e-2
Pure cement paste w/b = 0.33
2.1e-10
Microencapsulated RT24
8.5e-9
NRGS-GO-NRG
5.8e-10
NRGS-20RT24 NRGS-40RT24
1.0e-7
NRGS-20RT24–0.3GO-NRG
7.7e-10
NRGS-40RT24–0.3GO-NRG
1.6e-8
Cement Based Materials with PCM
1275
magnitude lower). The microencapsulated RT24 also shows an electrical conductivity similar to the cement paste. Addition of GO-NRG does not have a strong effect in the electrical conductivity (Table 7). In fact, the conductivity of the pastes with GONRG are quite similar to the reference cement paste. One possible explanation could be that the GO-NRG is not well dispersed in the cementitious matrix and therefore an improvement in the electrical conductivity cannot be obtained. It is well known that micro/nanoparticles tend to agglomerate and that it is difficult to obtain a good dispersion. A slight increase in conductivity can be observed in the pastes with 20% and especially with 40% volume microencapsulated RT24. The electrical conductivity is still low (much lower than the GO-NRG conductivity), and it is hypothesized that is due to ionic movement/conductivity. This improvement in conductivity could be due to the increase in porosity, characterized by MIP, when the volume fraction of MPCM is increased (see Fig. 5). As the porosity increases, and especially if it is interconnected, the ions dissolved in the water could move more freely, increasing the ionic conductivity.
4 Conclusions The casting of cement-based pastes with 20% and 40% in volume of microencapsulated RT24 and 0.3wt% bwb of GO-NRG has been described as well as their morphological, porosity and extensive thermal properties characterization. Next the main conclusions drawn from the study are listed. Scanning electron microscopy images show that the microcapsules, for both volumes, are well dispersed in the cement paste and well embedded in the matrix. Furthermore, most of the microcapsules do not seem to be broken and the elemental analysis by EDAX shows that the phase change material is within the microcapsules. Overall, as the volume of the MPCM addition increases, the total porosity of the paste also increases and the pore size distribution moves to lower size distributions. The porosity could have increased due to bad interaction between the polymeric microcapsules shell and the paste, or it could have also increased due to the large amount of superplasticizer used in both pastes (20% vol and 40% vol RT24) which is known to form air bubbles. Addition of the GO-NRG does not seem to have an effect in the porosity. Analysis by DSC shows that mixing the MPCM and GO-NRG into cement paste has no impact on onset temperature and peak temperature. Concerning the enthalpy, the latent heat in the phase transition is double for the pastes with twice the amount of microencapsulated RT24 as expected. Regarding thermal conductivity and diffusivity, in general, they decrease when the volume fraction of MPCM increases. Addition of GO-NRG has no significant impact. Only in the case of 40% microencapsulated RT24 the addition of GO-NRG increases thermal diffusivity. Concerning electrical conductivity of the pastes, addition of GO-NRG does not seem to have any effect, even though its electrical conductivity is much higher (1.0e-2 S/cm) than the one of the cement paste (2.1e-10 S/cm). These results seem to indicate that there is not a good dispersion of the GO-NRG in the cementitious matrix. A slight increase in conductivity can be observed in the pastes with 20%, and especially with 40% volume
1276
E. Erkizia et al.
microencapsulated RT24. Considering that the pastes were only superficially dried the pores of the pastes are expected to contain water, therefore this increase in porosity could facilitate the movement of ions in the paste. Finally, it could be stated that the addition of GO-NRG within cement pastes has no appreciable effect on the thermal and electrical properties of the composites, which requires to investigate new methods for increasing the relatively low thermal and electrical conductivity of the cement matrix, with the aim of obtaining a PCM-based cementitious product that can work as an active system via thermoelectrical coupling. Acknowledgements. The authors gratefully acknowledge the financial support from the NRGSTORAGE project (870114, 2020–2024, https://nrg-storage.eu/), financed by the European Union H2020 Framework under the LC-EEB-01–2019 call, IA type.
References 1. Balali, A., Yumusa-Kaltungo, A., Edwards, R.: A systemic review of passive anergy consumption optimisation strategy selection for buildings through multiple criteria decision-making techniques. Renew. Sustain. Energy Rev. 171, 113013 (2023) 2. Sawadogo, M., Duquesne, M., Belarbi, R., Hamami, A.E.A., Godin, A.: Review on the integration of phase change materials in building envelopes for passive latent heat storage. Appl. Sci. 11, 9305 (2021) 3. Shafigh, P., Asadi, I., Mahyuddin, N.B.: Concrete as a thermal mass material for building applications - a review. J Build. Eng. 19, 14–25 (2018) 4. DIN EN ISO 22007-2:2021-05 - Plastics - Determination of thermal conductivity and thermal diffusivity - Part 2: Transient plane heat source (hot disc) method (ISO/DIS 22007–2:2021) 5. Kim, K.H., Jeon, S.E., Kim, J.K., Yang, S.: An experimental study on thermal conductivity of concrete. Cem. Concr. Res. 33(3), 363–371 (2003) 6. Zurutuza-Elorza, A., Alonso-Rodriguez, B.: Method for obtaining graphene oxide. European patent No. 3,070,053B1. European Patent Office (2015) 7. He, Y.: Rapid thermal conductivity measurement with a hot disk sensor: Part 1. Theoretical considerations. Thermochimica Acta 436(1-2), 122–129 (2005) 8. Phase Change Material, Quality Assurance RAL-GZ 896. RAL Deutsches Institut für Gütesicherung und Kennzeichung e. V (2009)
Hygrothermal Behaviour and Durability of Bio-aggregate Based Building Materials (TC 275-HBD)
Rilem TC 275 HDB – International RRT on MBV Measurement of Vegetal Concrete Florence Collet1(B) , Stijn Mertens2 , Paulina Faria3 , Sofiane Amziane4 Thibaut Colinart5 , Camille Magniont6 , Sylvie Prétot1 , Romildo Dias Toledo Filho7 , and Méryl Lagouin6
,
1 Univ. Rennes, Laboratoire de Génie Civil et Génie Mécanique, Rennes, France
[email protected]
2 Buildwise, Building Materials Laboratory, Brussels, Belgium 3 CERIS, NOVA School of Science and Technology, NOVA University of Lisbon, Lisbon,
Portugal 4 Univ. Clermont Auvergne, CNRS, Institut Pascal, UMR 6602, Clermont-Ferrand, France 5 Univ Bretagne Sud, UMR CNRS 6027, IRDL, 56100 Lorient, France 6 Laboratoire Matériaux Et Durabilité Des Constructions, Université de Toulouse, INSA, UPS,
Toulouse, France 7 COPPE, Universidade Federal do Rio de Janeiro, Rio de Janeiro, Brazil
Abstract. The RILEM TC 275-HDB - Hygrothermal behaviour and Durability of Bio-based materials aimed to investigate testing methods of hygrothermal and capillary behaviour on vegetal concrete. Usual protocols were considered in Round Robin Tests (RRT) to assess their applicability and to point out warning aspects. The studies were performed on hemp concrete provided by an industrial partner of the TC (VICAT Group), with density about 330 kg/m3 . Regarding the measurement of Moisture Buffer Value (MBV), seven laboratories took part to the study. As the participating laboratories commonly used the NORDTEST protocol to measure MBV, this RRT was performed “as usual”, without additional guideline except initial conditioning of specimens. The results differ more or less between laboratories. Their analysis allows identifying several shortcoming regarding the exchange surface, the moisture flux direction versus the compacting one and the climate chamber technology. A second RRT has to be performed to formulate a more detailed protocol suitable for vegetable concrete and to assess its robustness. Keywords: hemp concrete · hygric behavior · round robin test · moisture buffer value · test protocol
1 Introduction Vegetal concrete are building materials made of vegetal particles glued by a binder paste. The raw materials, as well as the composites, show particular characteristics compared to those of usual concrete; namely vegetal concrete has very high porosity and is highly hygroscopic [1]. It is thus necessary to assess the applicability and the robustness of experimental protocols to characterise them. A first RILEM Technical © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 A. J˛edrzejewska et al. (Eds.): SynerCrete 2023, RILEM Bookseries 43, pp. 1279–1287, 2023. https://doi.org/10.1007/978-3-031-33211-1_114
1280
F. Collet et al.
Committee (TC 236-BBM - Bio-aggregate based Building Materials) investigated the bio-particles characterisation and mechanical behaviour of vegetal concrete. The work based on Round Robin Tests (RRT) led to recommendations on the characterisation methods to determine the initial water content, water absorption, dry density, particle size distribution and thermal conductivity of hemp shiv [2]. It also provided a study on the variability of the mechanical properties of hemp concrete [3]. The RILEM TC 275-HDB (Hygrothermal behaviour and Durability of Bio-aggregate based building materials) is performed in the continuity of the TC 236-BBM. It investigates the characterisation of the hygrothermal behaviour and the durability of vegetal concrete. It includes RRT on the measurement of water vapour permeability, moisture buffer value (MBV), thermal conductivity and capillary water absorption. The RRT are performed on hemp concrete provided by VICAT Group. This paper deals with a first RRT performed on the MBV measurement of vegetal concrete within RILEM.
2 Materials and Methods 2.1 Materials This study is performed on specimens cut from hemp concrete blocks, which were produced by a precast company from VICAT Group (Vieille Matériaux, Besançon - France). The binder used is the prompt natural cement and the hemp shiv is from near Besançon. All the specimens were produced with one batch. Their curing time at the precast company was higher than one month, in a ventilated hall with ambient temperature about 20 °C. According to the tests performed, different shapes and sizes of specimens are considered. For the MBV study, the specimen are 15 cm edge cubes. Such cubes are also used for thermal conductivity and capillary water absorption on other RRT. After stabilisation at 23 °C and 50% RH, the average density from 86 specimens distributed in 6 laboratories is 329 kg.m−3 with a standard deviation of 6 kg.m−3 . This leads to a coefficient of variation lower than 2% in density and underlines the high consistency of the production method. Due to their production method, the specimens have cut and molded surfaces (Fig. 1).
Fig. 1. Cut vs moulded faces of a hemp concrete 15 cm cubic specimen.
Rilem TC 275 HDB – International RRT on MBV
1281
2.2 Methods This study was conducted within a RRT. Seven laboratories participated: Buildwise, formerly the Belgian Building Research Institute (BBR) in Belgium; the Alberto Luiz Coimbra Institute for Graduate Studies and Research in Engineering (COP) in Brazil; the University of South Brittany (UBS) in France; the University of Clermont – Auvergne (UCA) in France; the NOVA University of Lisbon (UNL) in Portugal; the University Paul Sabatier of Toulouse (UPS) in France; the University of Rennes, previously University of Rennes 1 (UR1) in France. The protocol commonly used to measure the MBV was similar in all laboratories: it was based on the NORDTEST protocol [4]. This protocol consists in measuring the mass variation of specimens exposed to daily cyclic variations: 8 h at high relative humidity (75% RH) followed by 16 h at low relative humidity (33% RH) (Fig. 2). The test goes on until the change in mass, m, is the same between the last three cycles with less than 5% of discrepancies. The specimens are sealed on all but one exchange surface, and the MBV is calculated from the last three cycles as Eq. (1): mhigh − mlow MBV = RH high,av − RH low,av A
(1)
being mhigh/low the mass at the end of the adsorption / desorption phase (g) and RHhigh,av/low,av the average value of relative humidity at high/low relative humidity level (%RH).
Fig. 2. Daily cyclic variation of RH and mass during the MBV test following the NORDTEST protocol [4].
The recommendations of the NORDTEST protocol also include a size of the specimen higher than the penetration depth, the stabilisation of specimens at 23 °C and 50%
1282
F. Collet et al.
RH before the test, and an air velocity of 0.1 ± 0.05 m.s−1 near the exchange surface. The test can be performed with horizontal or vertical exchange surfaces. The size of the specimens meets the requirement of the NORDTEST protocol as they are 15 cm in edge and the penetration depth, estimated in [5], is 5.8 cm. For the first step, the participating laboratories performed their measurement as they are used to do it. However, as hemp concrete is a highly hygroscopic material with hysteresis between the adsorption and the desorption curve [1], it was decided to dry the specimen before the stabilisation at 23 °C and 50% RH, in order to begin with the same water content. The specimens were thus conditioned as follows. Firstly, they were dried during one day at 30°, one day at 40 °C, one day at 50 °C, and then at 60 °C (to prevent the damage due to too high temperature increase). Once the dry point was reached, the specimens were stabilized at 23 °C and 50% RH. As well for dry point, the specimens were considered stabilized when the variation of mass was lower than 0.01% between three weights at 24 h time-step. To ensure a high quality of the sealing, dust had to be removed with a light air flow before applying an aluminium tape. Two cross layers were laid with an overlap between the strips and an overlap at the edges (Fig. 3). The air velocity should be measured and controlled to meet the requirement of the NORDTEST protocol (Fig. 4).
Fig. 3. Sealing recommendations for the RRT.
3 Discussion and Results of the Round Robin Test Table 1 lists the test conditions with differences between the different laboratories. All laboratories used a climatic chamber that works with refrigeration group except UBS where the climatic chamber works with Peltier effect. The exposed surface may be horizontal or vertical (or in both directions). Other varying parameters were: test performed on cut surface or moulded surface, exposed surface parallel or perpendicular to compaction direction, and slightly different climatic conditions. Slightly different air velocity were observed, although they meet in general the requirement of the NORDTEST protocol.
Rilem TC 275 HDB – International RRT on MBV
1283
Fig. 4. Example of device used to control air velocity near the exchange surface of specimens.
Figure 5 displays typical ambient conditions for tests conducted with refrigeration technology (UR1 as example) and for test conducted with Peltier effect (UBS as example). The ambient conditions are better respected with the climate chamber with refrigeration technology than with climate chamber with Peltier effect. In particular, with Peltier effect, the change in relative humidity is longer and does not reach the set point in desorption. The MBV results obtained for the different configurations are plot in Fig. 6. The ordinate shows the measured average moisture buffer value with its standard deviation. There is a large variation between the obtained results. The average value from all configurations (shown in Fig. 6 by the solid red bar) is 2.56 g/m2 .%RH with a variation coefficient of 14.9%. The difference between the extremely measured values is 1.51 g/m2 %RH, corresponding to a variation of 59% related to the average value of all configurations. To identify the cause of this large variation, the results were studied in more detail. The results are similar with vertical and horizontal exposition (2.52 and 2.62 g/(m2 .%RH)) with higher standard deviation for vertical exposition (0.46 g/(m2 .%RH)) and lower for horizontal exposition (0.23 g/(m2 .%RH)). Concerning the exchange surfaces, cut surfaces lead to higher values than molded surfaces with average MBV of 2.67 g/(m2 .%RH) for cut surfaces and 2.49 g/(m2 .%RH) for molded surfaces, with standard deviations of 0.58 and 0.20 g/(m2 .%RH), respectively. Regarding the flux direction, the results obtained with flux parallel to the compaction direction are lower to the one obtained with perpendicular flux. They are respectively of 2.52 ± 0.25 g/(m2 .%RH) and 2.61 ± 0.52 g/(m2 .%RH). This first RRT shows that: (i) the exchange surface widely impacted the MBV results, and actually when the surface was cut, the porosity was more open and induced a higher MBV than for molded surface; (ii) the moisture flux direction versus the compaction direction did not show a high impact on the results, as the anisotropy was not so pronounced for the studied hemp concrete; (iii) climate chambers with Peltier technology are less efficient in desorption phase both for decreasing relative humidity rate and for low relative humidity level. Furthermore, the impact of air velocity was not highlighted, as the air velocity was similar in all laboratories.
1284
F. Collet et al.
Table 1. Test abbreviation, climate chamber used and conditions in the different laboratories the first 3 letters indicate the lab, following letters the exchange surface orientation (Vertical or Horizontal) and type (Molded or Cut), and finaly the moisture flux versus compation direction (perpendicular or parallel), Test
Climate chamber
Exchange surface orientation
Exchange surface type
Moisture flux vs compaction direction
BBRV,M,⊥
Vötsch VC 4034 (refrigeration group)
Vertical
Molded
Perpendicular (⊥)
BBRV,M,//
Vötsch VC 4034 (refrigeration group)
Vertical
Molded
Parallel (//)
COPV,C, ⊥
Controls, model 10-D1429/A (refrigeration group)
Vertical
Cut
Perpendicular (⊥)
UBSV,C,//
Memmert HPP Vertical 108 (peltier effect)
Cut
Parallel (//)
UBSH,C,//
Memmert HPP Horizontal 108 (peltier effect)
Cut
Parallel (//)
UCAH,M,//
BINDER MKF Horizontal 240 (refrigeration group)
Molded
Parallel (//)
UCAH,C,//
BINDER MKF Horizontal 240 (refrigeration group)
Cut
Parallel (//)
UPSV,C, ⊥
WEISS WK3–340/0 (refrigeration group)
Vertical
Cut
Perpendicular (⊥)
UPSV,M, ⊥
WEISS WK3–340/0 (refrigeration group)
Vertical
Molded
Perpendicular (⊥)
(continued)
Rilem TC 275 HDB – International RRT on MBV
1285
Table 1. (continued) Test
Climate chamber
Exchange surface orientation
Exchange surface type
Moisture flux vs compaction direction
UR1V,M, ⊥
Vötsch VC 4060 (refrigeration group)
Vertical
Molded
Perpendicular (⊥)
UR1V,M, //
Vötsch VC 4060 (refrigeration group)
Vertical
Molded
Parallel (//)
UR1H,M, ⊥
Vötsch VC 4060 (refrigeration group)
Horizontal
Molded
Perpendicular (⊥)
UR1H,M, //
Vötsch VC 4060 (refrigeration group)
Horizontal
Molded
Parallel (//)
Fig. 5. Comparison of the climatic conditions obtained by a climatic chamber with refrigeration technology (UR1 as example) to those obtained with Peltier technology (UBS).
1286
F. Collet et al.
Fig. 6. Results for the MBV obtained by the different laboratories, with different configurations and test conditions defined in Table 1.
4 Conclusion In the framework of the RILEM TC 275-HDB (Hygrothermal behaviour and Durability of Bio-based building materials), a Round Robin Test was performed on the measurement of Moisture Buffer Value, on the same materials, in seven laboratories. The results show some discrepancies between the laboratories, all with protocols based on the NORDTEST protocol. Their analysis leads to several shortcomings, mainly on the exchange surface and the direction of moisture flux versus compaction direction. Furthermore, the impact of air velocity was not highlighted, as the air velocity was similar in all laboratories and complying with NORDTEST requirements. Therefore, a second Round Robin Test has to be performed with complementary details. It is necessary to ensure that the exchange surface and moisture flux direction are representative of in-situ conditions: the exchange surface should always be molded; and the flux direction should correspond to the one with on-site implementation. This second Round Robin Test should underline the main points to be taken into account and ensure an adapted robust protocol, applicable to vegetal concrete. The application of a robust protocol by reseachers all around the world when testing vegetal concrete will allow comparison of results, quality and improvement of products, and contribute to a wider use of this type of concrete.
References 1. Amziane, S., et al.: Bio-aggregates Based Building Materials. Amziane Sofiane and Collet Florence. Springer, Netherlands (2017). 9789402410303. Accessed 24 Feb 2017 2. Amziane, S., Collet, F., Lawrence, M., Magniont, C., Picandet, V., Sonebi, M.: Recommendation of the RILEM TC 236-BBM: characterisation testing of hemp shiv to determine the initial water content, water absorption, dry density, particle size distribution and thermal conductivity. Mater. Struct. 50(3), 1–11 (2017). https://doi.org/10.1617/s11527-017-1029-3
Rilem TC 275 HDB – International RRT on MBV
1287
3. Niyigena, C., et al.: Variability of the mechanical properties of hemp concrete. Mater. Today Commun. 7, 122–133 (2016). https://doi.org/10.1016/j.mtcomm.2016.03.003 4. Rode, C., et al.: NORDTEST project on moisture buffer value of materials. In: AIVC Conference on Energy Performance Regulation Ventillation Relation Energy Performance and Buildings, Bruxelles, Belgique, pp. 47–52 (2005) 5. Collet, F., Pretot, S.: Experimental investigation of moisture buffering capacity of sprayed hemp concrete. Constr. Build. Mater. 36, 58–65 (2012). https://doi.org/10.1016/j.conbuildmat. 2012.04.139
Rheological Behavior of 3D Printable Bio-Concretes Produced with Rice Husk Matheus P. Tinoco(B)
, Oscar A. M. Reales , and Romildo D. Toledo Filho
Civil Engineering Program, Universidade Federal do Rio de Janeiro, COPPE/UFRJ, Cidade Universitária, Ilha do Fundão, CEP 21941-972, Rio de Janeiro, RJ, Brazil [email protected]
Abstract. This article presents an investigation on the rheological behavior of 3D printable bio-concretes containing rice husk fine particles. This agro-industrial waste was used to adjust the fresh-state properties of the concrete mixture, required for the printing process, while decreasing its environmental impact. In this study, the rice husk content varied from 5 to 12.5%, in mass of solids. The rice husk was submitted to a crushing process and two different particle sizes were investigated. The static yield stress was measured using a rotational viscosimeter equipped with a Vane spindle after different resting times. The results were used to obtain the thixotropic build-up rate and to calculate the printing parameters, such as layer height, time between successive layers and printing velocity. The results showed that the use of rice husk increases both the initial yield stress and the thixotropic build-up rate of the bio-concretes, which leads to an increase of the maximum layer height and the maximum printing velocity. Finally, the results showed the feasibility of using biomass for improving the buildability and printability of the mixtures. Keywords: biomass · rheology · 3D concrete printing · bio-concretes
1 Introduction Recently, 3D concrete printing (3DCP) has appeared as one of the main alternatives to increase the productivity and reduce the use of manual labour in the civil construction sector. The technology is based on the use of robotic device, which is responsible for depositing layer by layer of material, while drawing the outline of the structure to be constructed [1]. Some advantages of this technology are the reduction in waste production and increase in the construction speed. Furthermore, the automated process allows the production of multifunctional and optimized structures, with better thermoacoustic and mechanical performance [2]. Despite the advantages, there are several challenges that must be solved in order to make 3DCP viable on large scale applications. One of the main difficulties is the adjustment of the rheological properties, since the material to be printed must be fluid enough to be pumped, but, at the same time, must be capable to support the weight of successive printed layers [3]. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 A. J˛edrzejewska et al. (Eds.): SynerCrete 2023, RILEM Bookseries 43, pp. 1288–1296, 2023. https://doi.org/10.1007/978-3-031-33211-1_115
Rheological Behavior of 3D Printable Bio-Concretes
1289
In order to adjust the rheological properties and obtain printable mixtures, high Portland cement contents and chemical admixtures are generally used. Furthermore, the use of course and fine aggregates is limited by the capacity of the pumping systems. Thus, the materials adopted in 3DCP generally present high environmental impact, if compared to conventional concrete. In this sense, the present work proposes an alternative way to reduce the cement consumption and adjust the fresh properties of cementitious composites for 3D printing, thought the use of rice husk fine particles. The morphology and chemical composition of the particles was studied through Scanning electron microscopy (SEM). The effect of the natural particles on the hydration was evaluated through isothermal calorimetry. The fresh properties were assessed by rotational viscosimeter. Finally, the results were used to obtain the thixotropic build-up rate and to calculate the printing parameters.
2 Materials and Mixing Protocol 2.1 Materials Under Investigation In this study, rice husk bio-concretes were produced using high-early strength Portland cement, rice husk particles, tap water and viscosity modifying agent (VMA). The Portland cement (Brazilian type CPV-ARI) was produced by LafargeHolcim, in CantagaloRJ. To avoid segregation and exudation during the mixture, a viscosity modifying agent from Basf, MasterMatrix UW 410, was used. The rice husk came from local producers from São Paulo state, in Brazil. To prevent clogging of the pumping system, the rice husk was crushed using a cutting mill from SOLAB. To study the effect of the particles size on the rheology and hydration, the resulting material was submitted to mechanical sieving and divided into two different sizes: rice husk fine particles (RHF), which contains the particles retained in the 0.075 mm sieve, and the rice husk powder (RHP), which is formed by the particles below 0.075 mm. Figure 1a shows the rice husk fine particles, and Fig. 1b presents the rice husk powder, obtained from the crushing process.
Fig. 1. Rice husk used in this study: (a) rice husk fine particles (RHF), (b) rice husk powder (RHP).
1290
M. P. Tinoco et al.
2.2 Mixing Procedure The cementitious composites were produced using a constant 0.45 water-to-cement ratio. The viscosity modifying agent (VMA) was used in a constant fraction of 0.6% (by mass of cement - bmoc). To study the effect of the rice husk content on the rheology, the rice husk fine particles (RHF) were used in different fractions, from 5 to 12.5%, in mass of solids. The rice husk powder (RHP) was used in a constant fraction of 5%, in mass of solids. Table 1 shows the formulations investigated in the present study. The Portland cement and rice husk content is presented in mass of solids. Table 1. Formulations used in the present study. Mixture
Cement
RHF
RHP
W/C ratio
VMA (bmoc)
REF
1
-
-
0.45
0.006
RHF5
0.95
0.05
-
0.45
0.006
RHF10
0.90
0.10
-
0.45
0.006
RHF12.5
0.875
0.125
-
0.45
0.006
RHP5
0.95
-
0.05
0.45
0.006
The mixing protocol was conducted in 5 stages using a 5-L capacity planetary mixer. Firstly, the dry material (cement, rice husk, and VMA) was mixed for 1 min. Next, tap water was added, and the materials were mixed for 3 min. A 1-min stop was performed to mix the remaining material in the bottom part of the mixer. Finally, the material was mixed for 2 min. The resulting material was stored using plastic film until testing.
3 Testing Procedure Methods 3.1 Scanning Electron Microscopy (SEM) In order to study the surface and morphology of the rice husk particles, scanning electron microscopy (SEM) was used. The tests were performed using a TM3000 microscope, from Hitachi. The fine particles (RHF) and powder (RHP) were investigated. Furthermore, the chemical composition of the particles surface was assessed using energy dispersive X-ray spectroscopy (EDS). This technique was used to study the differences in composition between the outer layer and the inner structure of the rice husk and evaluate the presence of silicon dioxide (SiO2 ). 3.2 Isothermal Calorimetry The influence of rice husk particles on the cement hydration was evaluated using isothermal calorimetry. The samples were prepared outside the calorimeter, during a maximum time of 10 min. Samples of 5 g were used. The materials were manually mixed using glass rod and a 150 ml beaker. The tests were performed during seven days and the temperature was set as 25 °C. Deionized water was used as reference material.
Rheological Behavior of 3D Printable Bio-Concretes
1291
3.3 Rheological Tests Rheological tests were performed using a rotational viscosimeter Brookfield DV-III Ultra (Fig. 2a), equipped with a Vane spindle. The static yield stress was obtained applying an increasing shear rate, from 0 to 0.2 s−1 , during 200 s, and was defined as the maximum shear stress measured during this period. The tests carried out at different times: 10, 30, 60, 90, and 120 min, to obtain the structural build-up rate of the composites (Athix ), according to Eq. (1). τ (t) = τ0 + Athix · t
(1)
where τ (t) is the static yield stress at a given time “t”, τ0 is the initial static yield stress, Athix is the structural build-up rate, and t is the resting time. To evaluate the printability of the composites, the mixture RHF12.5, produced with 12.5% of rice husk particles, and the reference mixture were chosen to perform printing tests. The tests were performed using the 3D printer presented in Fig. 2b. In the tests, the nozzle diameter was 6.35 mm, and the printing velocity was 20 mm/s. A cylindrical object, with 150 mm diameter, and 200 mm height, was considered. The layer height was 10 mm, and the layer thickness was 30 mm.
Fig. 2. Effect of rice husk particles on the rheology and printability of the composites: (a) rotational viscosimeter, and (b) 3D printing device.
4 Results and Discussion 4.1 Rice Husk Properties Figure 3 shows the SEM images obtained for the different particle sizes used in this study. Figure 3a presents the rice husk fine particles, and Fig. 3b presents the rice husk powder.
1292
M. P. Tinoco et al.
Fig. 3. SEM images of the crushed rice husk: (a) fine particles (RHF), and (b) powder (RHP).
From the images, it is possible to see that the particles obtained present a wide range of sizes. The crushed particles maintained their porous cell structure, even after the crushing process. The powder (Fig. 3b) is mainly composed by extremely fine particles, below 0.075 mm, which, consequently, increase its surface area. From Fig. 3a, the particles are formed by a lighter outer layer, and an internal and porous structure, which is responsible for the high-water absorption capacity of the natural particles. According to previous studies [4], the outer epidermis is mainly composed by silicon dioxide (SiO2 ), while the internal structure is formed by lignin, cellulose, hemicellulose, and extractives. Figure 4 shows the results obtained from the EDS analysis, where the chemical composition of these structures can be assessed. The presence of silicon and oxygen is more pronounced in the outer layer. The internal structure is poor in silica and presents more carbon, since it presents more complex organic polymers.
Fig. 4. SEM images and energy dispersive spectroscopy (EDS) of the of the rice husk particles: (a) SEM image of the outer epidermis, (b) silicon map, and (c) oxygen map.
4.2 Isothermal Calorimetry Figure 5a shows the heat flow results obtained from the isothermal calorimetry for the composites in study. From the curves, it is possible to see that an increase in the rice husk content reduces the velocity of the cement hydration. The peak of the acceleration period was reached after 11.5h for the reference mixture. For the RHF12.5 mixture, however, this peak was reached after 21.5 h.
Rheological Behavior of 3D Printable Bio-Concretes
1293
According to previous studies [5], the effect of the natural particles on the hydration can be explained by the presence of extractives in the husk, which capture calcium ions from the solution and delay the formation of C-S-H. It can also be seen that the rice husk powder, used in a mass fraction of 5%, was sufficient to delay the heat release peak to times above those studied for the composite RHP5. The result can be explained by the higher surface area of the powder, which potentializes its negative effects on the cement hydration. From Fig. 5b, it can be seen that the rice husk increases the release of heat in the first 6 h, but the cumulative heat is reduced at the end of the tests. From the figure, it is also possible to see the negative effect of the rice husk powder on the hydration. The cumulative flow of the composite RHP5 presented small variation during the first 72 h. According to da Gloria [6], the negative effects of these particles on the hydration can be reduced through the alkaline treatment of the particles, which removes part of the extractives, or by the use accelerators, such as CaCl2 . More information on the use of alkaline treatment can be found in the study by Tinoco et al. [7].
Fig. 5. Results from the isothermal calorimetry: (a) Heat flow results, and (b) cumulative heat.
4.3 Rheological Tests Results Figure 6 presents the results obtained from the rheological tests. Figure 6a shows the shear stress versus time curves obtained after 120 min. These curves were used to obtain the static yield stress at different resting times. Figure 6b shows the static yield stress versus resting time results and the linear fits used to obtain the structural build-up rate (Athix ), from the slope of the curve, and the initial static yield stress (linear coefficient). From the results, it can be seen that the rice husk particles increased the static yield stress if compared to the reference mixture. Due to the high absorption capacity of the rice husk particles, the distance between particles is reduced, which increases both the flocculation and formation of hydrates and the contact points between particles [7]. This
1294
M. P. Tinoco et al.
behavior can also explain an increase in the structural build-up rate (Athix ), as the rice husk content increases. The composite RHP5 presented higher static yield stress if compared to RHF5, since the powder presents higher surface area and, consequently, higher initial water adsorption. The structural build-up rate of RHP5, however, is considerably smaller, since the rice husk powder is extremely harmful for the hydration. Furthermore, the particles present in the powder are less porous, which reduces their water absorption along time. In this sense, the evolution of the rheological properties is considerably affected. The results show that rice husk particles can be used to adjust the rheology and increase both the static yield stress and structural build-up rate of the composites. On the other hand, ultra-fine particles (below 0.075 mm) should be avoided, since they affect both the hydration and rheology.
Fig. 6. Rheological tests results: (a) stress versus time curves at t = 120 min, and (b) static yield stress versus resting time and linear fits to obtain the Athix and initial static yield stress.
4.4 Printing Tests Regarding the 3D printing process, materials with higher static yield stress and structural build-up rate (Athix ) lead to higher maximum printing height and maximum printing velocity [8]. High structural build-up rate values, however, can reduce the intermixing between layers and create cold joints [9]. To evaluate the printability of the composites studied, the reference mixture and the mixture RHF12.5 were chosen to perform 3D printing tests. Figure 7a shows the printing test performed with the reference mixture. It is possible to observe that the reference material did not present adequate rheology and could not maintain its own shape, even at the beginning of the printing process. Figure 7b highlights the plastic collapse of the printed object. Figure 8a shows the 3D printing test performed with the mixture RHF12.5, where the shape stability and extrudability of the printed material can be observed. Figure 7b shows
Rheological Behavior of 3D Printable Bio-Concretes
1295
the final geometry of the printed object. The height of the printed object was 190 mm, which is 10 mm smaller than the target height. This difference can be explained by the deformation of the initial layers during the printing process. The results show that the composite RHF12.5 was capable to support the weight of the successive printed layers without collapsing, which can be explained by its higher static yield stress and structural build-up rate, if compared to the reference mixture. The result indicate that rice husk can be used to produce 3D printable bio-concretes.
Fig. 7. 3D printing tests with the reference mixture (REF): (a) beginning of the printing process, and (b) plastic collapse of the printed object.
Fig. 8. 3D printing tests with the mixture RHF12.5: (a) printing process, and (b) final geometry of the printed object.
5 Conclusions This paper studied the hydration and rheology of cementitious composites produced with rice husk particles. Two different particle sizes and different rice husk contents were evaluated. The rice husk particles present an outer surface mainly composed by
1296
M. P. Tinoco et al.
silica, and a porous internal structure, which leads to high water absorption. From the results, it can be concluded that the rice husk particles delay the hydration of cement. When particles below 0.075 mm are used, even in a small content, the hydration is considerably affected. Further investigation should be done to improve the compatibility between the cementitious matrix and natural particles. From the rheological tests, rice husk particles increase both rheological parameters evaluated. When rice husk powder is used, the structural build-up rate is smaller, since the hydration is extremely affected. Other rheological parameters, such as dynamic yield stress and plastic viscosity, can also affect the printing process, and should be better evaluated in future research. Finally, a printing test was performed, and the results showed the feasibility of using rice husk particles to produce 3D printable bio-concretes. Acknowledgements. This research was financed in part by the Brazilian funding agencies CNPq (Conselho Nacional de Desenvolvimento Científico e Tecnológico), and Fundação Carlos Chagas Filho de Amparo à Pesquisa do Estado do Rio de Janeiro (FAPERJ).
References 1. Wangler, T., Roussel, N., Bos, F.P., Salet, T.A.M., Flatt, R.J.: Digital concrete: a review. Cem. Concr. Res. 123 (2019) 2. Tinoco, M.P., de Mendonça, E.M., Fernandez, L.I.C., Caldas, L.R., Reales, O.A.M., Toledo Filho, R.D.: Life cycle assessment (LCA) and environmental sustainability of cementitious materials for 3D concrete printing: a systematic literature review. J. Build. Eng. 52 (2022) 3. Roussel, N.: Rheological requirements for printable concretes. Cem. Concr. Res. 112, 76–85 (2018) 4. Juliano, B.O., Tuaño, A.P.P.: Gross structure and composition of the rice grain. In: Rice: Chemistry and Technology, pp. 31–33. Elsevier (2018) 5. da Gloria, M.Y.R., Andreola, V.M., dos Santos, D.O.J., Pepe, M., Toledo Filho, R.D.: A comprehensive approach for designing workable bio-based cementitious composites. J. Build. Eng. 34 (2021) 6. Delannoy, G., et al.: Impact of hemp shiv extractives on hydration of Portland cement. Constr. Build. Mater. 244 (2020) 7. Pimentel Tinoco, M., Gouvêa, L., de Cássia Magalhães Martins, K., Dias Toledo Filho, R., Aurelio Mendoza Reales, O.: The use of rice husk particles to adjust the rheological properties of 3D printable cementitious composites through water sorption. Constr. Build. Mater. 365 (2023) 8. Reales, O.A.M., Duda, P., Silva, E.C.C.M., Paiva, M.D.M., Toledo Filho, R.D.: Nanosilica particles as structural buildup agents for 3D printing with Portland cement pastes. Constr. Build. Mater. 219, 91–100 (2019) 9. Roussel, N., Cussigh, F.: Distinct-layer casting of SCC: the mechanical consequences of thixotropy. Cement Concr. Res. 38, 624–632 (2008)
Flax Fabric-Reinforcement Lime Composite as a Strengthening System for Masonry Materials: Study of Adhesion Ali Rakhsh Mahpour1(B) , Josep Claramunt2 , Mònica Ardanuy3 , and Joan Ramon Rosell4 1 Department of Architectural Technology, Universitat Politècnica de Catalunya, Barcelona,
Spain [email protected] 2 Department of Agricultural Engineering, Universitat Politècnica de Catalunya, Barcelona, Spain 3 Department of Materials Science and Engineering, Universitat Politècnica de Catalunya, Barcelona, Spain 4 Department of Building Construction, Universitat Politècnica de Catalunya, Barcelona, Spain
Abstract. The present study was designed to investigate the bond mechanism between flax nonwoven fabric reinforced lime mortars and four different types of masonry stones. The researchers carried out a comprehensive characterization of the mortars and stones, and analyzed the interface between these materials based on bond extension and flexural strength. The results showed that bond strength values were directly related to adherence extension, which was dependent on surface treatment (such as water, lime, and latex). The flexural tests conducted on stones strengthened with the flax nonwoven fabric reinforced lime composite revealed that the latex treatment showed the best performance. This study provides preliminary evidence that the use of flax nonwoven fabric reinforced lime composite may be an effective method for masonry strengthening. However, further research is necessary to reach a comprehensive conclusion. Further studies may include an investigation of the effects of different types of latex, the influence of curing conditions on bond strength, and the long-term performance of the strengthening system. The results of this research may have important implications for the conservation and strengthening of historic masonry structures, as well as for the design of new masonry constructions. Keywords: Masonry stones · Fiber reinforced lime Mortar · adherence extension · vegetable fibers
1 Introduction During the 19th century, before the widespread of cement use, mortar binders were commonly air lime or gypsum, with a well-known use that dates back millennia. In terms of the adherence between materials, during the 19th-century air lime-based mortars © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 A. J˛edrzejewska et al. (Eds.): SynerCrete 2023, RILEM Bookseries 43, pp. 1297–1306, 2023. https://doi.org/10.1007/978-3-031-33211-1_116
1298
A. R. Mahpour et al.
were used for their attachment to the wall, sometimes with the use of additional clay, with provenance from the sand, and material with some pozzolanic behavior [1]. Current research has focused on the study of the constituent materials of this system, focusing on the mortars with fiber and kinds of stones. In the case of historic mortars, besides physical and mechanical characterization, the determinations of binder/aggregate proportion and chemical composition have been undertaken. Various analytical methodologies (manual and mechanical) have been used to find the best adherence [2]. Historical buildings have used multilayer mortar systems both as coatings for masonry work and as substrates for floors [3]. Multilayer mortar typically had two to three layers, with the thickness gradually decreasing toward the surface and the maximum size and percentage of the aggregates decreasing as well. Still, the majority of the aggregates used were natural, and of siliceous origin, and their gradation varied from 2 mm in the outer layers to 8 mm in the interior one [3, 4]. The degradation patterns encountered generally occur at the interface between the layers, leading to detachments, discontinuities, and cracking. Between the mortar layers themselves or between the mortar layers and the brickwork base, detachments can be seen. Therefore the mortar layers application technology is also very significant since it has a direct impact on the durability and adherence of the mortar layers [5, 6]. The prewetting of the substrate (adjacent mortar layer or stone) appears to greatly affect the binding strength of the composite systems, and the individual features observed in each case study are taken into consideration in this regard [3]. In general, it could be said that the adhesion of multilayer mortar systems depends on a variety of elements, including the individual characteristics of the mortar layers (structure, physic mechanical properties), the expansion of the effective contact area (through time), the interval between the application of the layers, and the layering technique used [3–7]. Numerous studies conducted in recent years have demonstrated the improved strength and durability of nano-structured materials [7, 8]. Also during the last decade, steel, glass, and polymers have been tested as predominant fibers for reinforcing cementitious matrices [9]. However, there are some disadvantages related to the use of these fibers, including high cost, and specifically, substantial environmental footprint. Therefore, vegetable and cellulose fibers together with recycled textile waste fiber have attracted significant interest in recent years as sustainable and suitable reinforcements in mortars and composites for low-to-medium-performance structural applications [10– 12]. Non-woven fabrics are fibrous planar engineered assemblages in which the fibers have been physically or chemically entangled to offer a particular amount of structural stability. When compared to open-structure woven textiles normally used for cementitious reinforcement, non-woven textiles display a lower porosity and less strength, but, are much cheaper and their reinforcement capacity is enough for certain applications. Moreover, despite their low porosity, using a proper technique, it is possible to achieve a good penetrability of the cement through the fabric. This makes them an attractive alternative for various applications, including strengthening systems for masonry materials. Furthermore, the use of non-woven textiles made from sustainable and renewable resources, such as flax fibers, can also contribute to reducing the environmental impact of these strengthening systems [13, 14].
Flax Fabric-Reinforcement Lime Composite
1299
Textile Reinforced Mortar (TRM) is a fairly new composite material consisting of inorganic matrices reinforced with high-strength non-metallic fiber grids or technical textiles [14, 15]. Efficient use of this composite material largely depends on the bond between the externally bonded grids and the existing substrate, since debonding may lead to premature failure and, thus, to a low degree of exploitation of the fibrous reinforcement which is the most costly component of the strengthening system. Different types of experimental set-ups have been employed for the investigation of the shear bond capacity of TRM strips externally bonded on masonry (or other) substrates including doublelap/double-prism, single-lap/single-prism, and double-lap/single-prism shear bond tests [16, 17]. Although the common practice for engineers is to design and apply readily available commercial product systems (mortar + textile), deep knowledge of the effect of parameters related to both constituent materials on the performance of the strengthened element is of crucial importance. Nevertheless, such parametric studies are relatively scarce and employ concrete substrates or single masonry units [14, 18]. This paper presents the results of such an investigation comprising an experimental program carried out to assess the effect of (i) the best state and materials in adhesion, and (ii) the best treatment in adhesion on the surface such as water, lime, resin. For this purpose, a variety of stones were used and the results were analyzed for tensile and bending strength.
2 Experimental Procedure 2.1 Materials 2.1.1 Binder A commercial lime (CL 90) supplied by DCAL BY CIARIES, SLU was used for producing the mortars. Metakaolin is supplied by the company “ARCIRESA” with the commercial name METACAOLIN PESER. The best combination of materials was determined through a combination of mechanical, chemical, and rheological evaluations, considering sustainability as a key factor. Mechanical evaluations examine the properties such as strength, toughness, and elasticity of the materials. Chemical evaluations determine the chemical composition and purity of the materials. Rheological evaluations measure the flow properties of the materials such as viscosity and thixotropy. By performing all of these evaluations, a complete and comprehensive understanding of the materials was obtained, considering sustainability criteria. This allowed for the selection of the best combination of materials for the specific application, ensuring both optimal performance and environmental responsibility. The combination that performed the best in all of these evaluations was then chosen as the optimal solution. This approach ensures that the materials used have the appropriate properties for the intended application, are of the highest quality, and are produced in a sustainable manner. It is worth noting that the properties of the binders given by the supplier were reported in previous work [13, 15].
1300
A. R. Mahpour et al.
2.1.2 Nonwoven Fabric Reinforcement The non-woven material was produced using flax fibers with a typical length of 60 mm. The fibers were obtained from a pilot plant named DILO OUG-II-6, which uses a process combining carding and needle-punching. This method has been previously described in other studies [13]. The hornificated flax nonwoven fabrics, exposed to four drying and rewetting cycles (Fig. 1).
Fig. 1. Hornification treatment flax nonwoven fabrics process: (a) water at room temperature for 6 h; (b) drying in an air-circulating oven at 60 ºC for 24
2.1.3 Stones The following stones, see Table 1 and Fig. 2; were used as the masonry unit: Indian sandstone from Rajasthan (smooth surface and curly surface). Table 1. Physical-chemical properties of Indian sandstone from Rajasthan provided by the supplier Physical properties
Chemical Analysis
Appearance/color
Red stone
%SiO2
95–97
Hardness
6.5 to 7 (Mohs scale)
%Fe2 O3
0.5–1
Density
2.3 to 2.4 kg/m3
%Mgo