Earth and Space 2022: Space Exploration, Utilization, Engineering, and Construction in Extreme Environments 9780784484470

Citation preview

Earth and Space 2022 Space Exploration, Utilization, Engineering, and Construction in Extreme Environments Proceedings of the 18th Biennial International Conference on Engineering, Science, Construction, and Operations in Challenging Environments Denver, Colorado  April 25–28, 2022 Edited by

Christopher B. Dreyer, Ph.D. Justin Littell, Ph.D.

EARTH AND SPACE 2022 SPACE EXPLORATION, UTILIZATION, ENGINEERING, AND CONSTRUCTION IN EXTREME ENVIRONMENTS SELECTED PAPERS FROM THE 18TH BIENNIAL INTERNATIONAL CONFERENCE ON ENGINEERING, SCIENCE, CONSTRUCTION, AND OPERATIONS IN CHALLENGING ENVIRONMENTS

Denver, Colorado April 25–28, 2022

SPONSORED BY

Aerospace Division of the American Society of Civil Engineers

EDITED BY

Christopher B. Dreyer, Ph.D. Justin Littell, Ph.D.

Published by American Society of Civil Engineers 1801 Alexander Bell Drive Reston, Virginia, 20191-4382 www.asce.org/publications | ascelibrary.org Any statements expressed in these materials are those of the individual authors and do not necessarily represent the views of ASCE, which takes no responsibility for any statement made herein. No reference made in this publication to any specific method, product, process, or service constitutes or implies an endorsement, recommendation, or warranty thereof by ASCE. The materials are for general information only and do not represent a standard of ASCE, nor are they intended as a reference in purchase specifications, contracts, regulations, statutes, or any other legal document. ASCE makes no representation or warranty of any kind, whether express or implied, concerning the accuracy, completeness, suitability, or utility of any information, apparatus, product, or process discussed in this publication, and assumes no liability therefor. The information contained in these materials should not be used without first securing competent advice with respect to its suitability for any general or specific application. Anyone utilizing such information assumes all liability arising from such use, including but not limited to infringement of any patent or patents. ASCE and American Society of Civil Engineers—Registered in U.S. Patent and Trademark Office. Photocopies and permissions. Permission to photocopy or reproduce material from ASCE publications can be requested by sending an e-mail to [email protected] or by locating a title in ASCE's Civil Engineering Database (http://cedb.asce.org) or ASCE Library (http://ascelibrary.org) and using the “Permissions” link. Errata: Errata, if any, can be found at https://doi.org/10.1061/9780784484470 Copyright © 2023 by the American Society of Civil Engineers. All Rights Reserved. ISBN 978-0-7844-8447-0 (PDF) Manufactured in the United States of America.

iii

Earth and Space 2022

Preface The first American Society of Civil Engineers (ASCE) Aerospace Division (ASD) International Conference was held in Albuquerque, New Mexico, United States in 1988 under the name “Engineering and Construction in Space (Space 88)”. The Conference has been held every two years since. In 2004 the name was changed to “Engineering, Construction, and Operations in Challenging Environments (Earth & Space 2004),” and in 2008 “Science” was added, to reflect the ever-expanding scope of the exploration of space and of challenging regions of the Earth. The ASCE ASD was due to hold its 17th Biennial International Conference on Engineering, Science, Construction, and Operations in Challenging Environments (Earth & Space 2020) in Seattle Washington, United States from Monday through Thursday, April 20–23, 2020. However, the global COVID-19 pandemic had other plans. As the world began to shut down in an effort to stop the spread of the COVID-19 virus, the United States followed suit in March of 2020, and the Earth and Space Conference was postponed until 2021. As the pandemic continued, the decision was made to move forward with a virtual conference in April of 2021. The conference was successfully completed in April with all elements including the short course, technical presentations, social events, and roundtable discussions held virtually. Shortly after the 2021 conference concluded, the conference committee made a decision to return to an evenyeared in-person conference, and so the 18th Biennial International Conference on Engineering, Science, Construction, and Operations in Challenging Environments (Earth & Space 2022) was planned for April of 2022. With the challenge of a compressed timeline to develop the conference program, the conference committee quickly developed the technical schedule, short course, and all related conference activities. These Proceedings are a reflection of that technical program. This year’s technical program included a six-hour short course entitled “Health Monitoring for Terrestrial and Space Based Environments” developed by the ASCE ASD Dynamics and Controls Technical Committee. This short course was designed for engineers and researchers who are focused on various aspects of health monitoring. This morning session of the short course taught the basics about machine learning algorithms and their applications to detecting structural connection looseness. Structural connections are commonly used in many engineering fields, such as aerospace, mechanical, energy, and civil engineering, among others. The afternoon session was focused on the health monitoring of spacecraft, experiments, human and machine interaction and the human itself. It was intended for engineers working in a wide range of disciplines including spacecraft design and manufacture, human factors, robotics and astronaut physical health. The Conference structure this year was organized into five symposia. Symposium 1 focused on Granular Materials in Space Exploration. Symposium 2 focused on the Exploration and Utilization of Extra-Terrestrial Bodies. Symposium 3 focused on Advanced Materials and Designs for Aerospace Structures and Terrestrial Structures under Extreme Environments. Symposium 4 focused on Structures in Challenging Environments: Dynamics, Controls, Smart Structures, Health Monitoring, and Sensors. Symposium 5 focused on Space Engineering, Construction, and Architecture for Moon, Mars, and Beyond.

© ASCE

Earth and Space 2022

The papers published in the conference proceedings went through three rounds of reviews. The abstract of each paper presented was reviewed and accepted by the Conference Committee before an invitation was issued for a full-length manuscript. The full-length manuscript was peer reviewed by 2 individuals of the Conference Committee, including the symposium chairs. Review comments were provided to authors requesting a final revised full-length manuscript. The final manuscripts were then reviewed by the Editors for content and for compliance with ASCE publication guidelines. Where needed, further revisions were requested from authors. All papers published in the Proceedings are eligible for submission to the ASCE Journal of Aerospace Engineering (JAE) if updated with 51% or more new content and are eligible for ASCE awards. Historically, this conference series has attracted many non-ASCE member participants with multidisciplinary backgrounds. The 2022 Conference is no exception, with paper presenters, invited guest speakers, and conference attendees from all across the globe. The Conference provides an unparalleled opportunity for cross-pollination among civil and aerospace engineering, science, and technology communities. We believe that centuries of experience in civil engineering design, analysis, construction, operations, and management on Earth will significantly contribute to the responsible use of other planetary bodies, including asteroids, the Moon, and Mars, to ensure that humanity not only survives in the long-term, but continues to thrive. As in previous Earth and Space Conferences, Earth & Space 2022 brings together engineers, scientists, educators, and other professionals from many disciplines all around the world to share knowledge and inspiration. The Conference has an extraordinary multidisciplinary flavor that creates a stimulating environment for intellectual exchange by all participants. These Proceedings represent the results of the partnerships, ideas, and goals generated in part during this and previous Earth and Space Conferences. This body of work brings us one step closer to our futures in extreme environments both on this planet, and beyond. Editors : Justin D. Littell Ph.D., M. ASCE ASCE Earth and Space 2022 Technical Chair NASA Langley Research Center Hampton, VA And Christopher Dreyer, Ph.D., M. ASCE ASCE Earth and Space 2022 Conference Chair Colorado School of Mines Golden, CO

© ASCE

iv

v

Earth and Space 2022

Acknowledgments These Proceedings represent the efforts of many people, including the Conference Organizing Committee members, authors, session chairs, abstracts and paper reviewers, and ASCE volunteers and staff. The Conference Organizing Committee members, including Robert Goldberg, Chris Dreyer, Chris Skonieczy, Phil Metzger, Robert Mueller, Kris Zacny, An Chen, Hongyu (Nick) Zhou, Wei Zhang, Arif Masud, Gangbing Song, Ramesh Malla, Melissa Sampson, Alexander Jablonski, and Gerald Sanders, in particular spent many hours discussing, planning, and reviewing, all conference materials under a compressed timeline. Although we cannot list everyone by name, we gratefully acknowledge the efforts of all the dedicated ASCE volunteers from the Aerospace Division. In particular, we offer special thanks to the authors and presenters for their contributions to the body of knowledge and to engineering practice. Without them, we could not have held our Conference. The plenary and special panel speakers all graciously gave their time and energy to inspire conference participants. The pre-conference short course instructors contributed to our understanding of engineering and construction on the moon. We extend our sincere thanks to all members of the ASCE Aerospace Division Executive Committee for their continuous support and encouragement during the long process of conference planning, and for providing the continuing vision that makes this conference possible. We are particularly grateful for the support from Robert Goldberg, Volunteer Liaison between the ASCE Committee on Technical Advancement (CTA) and ASD, as well as ASCE staff Lindsay O’Leary (Director, Technical Advancement), James Neckel (ASCE Staff Liaison to ASD), and the Colorado School of Mines conference organizing staff including Melody Francisco, Becca Guillen, Richard Tyrell-Ead and Julie Gorman-Farquhar. We also greatly appreciate the support from Donna Dickert, ASCE Senior Manager and Acquisitions Editor, Corinne Addison, and the rest of the team at ASCE Publications for their help with the timely publication of these proceedings. The Conference received generous support from Honeybee Robotics and Canadensys Aerospace Corporation. We gratefully acknowledge their support for the Conference. Editors : Justin D. Littell Ph.D., M. ASCE ASCE Earth and Space 2022 Technical Chair NASA Langley Research Center Hampton, VA

Christopher Dreyer, Ph.D., M. ASCE ASCE Earth and Space 2022 Conference Chair Colorado School of Mines Golden, CO

© ASCE

Earth and Space 2022 

vi

Contents Granular Materials in Space Exploration Granular Mechanics of JSC-1 Mars Regolith Simulants............................................................1 Reza Ashtiani, Jesus Baca, Jacob Wessel, and Arash Karimipour Mapping Lunar Lander Plume Ejecta Trajectories to Lunar Surface Elevations.................17 D. Batcheldor and J. Mantovani Geometrical Characteristics of Lunar and Martian Regolith Simulants................................27 Caleb J. Carnes, Reza S. Ashtiani, Joel A. Sloan, Melissa S. Beauregard, and Kimberly D. De la Harpe Experimentally Evaluating Granular Scaling Laws for Predicting Lunar-Gravity Wheel Performance in Cohesive Regolith..................................................................................35 Adriana Daca and Krzysztof Skonieczny Uniform Dust Deposition System for Dust Tolerance Studies..................................................48 Stephen Gerdts, Nathan Jimenez, and Patrick H. Dunlap Demonstration of Capability to Simulate Particle Irregular Shape and Poly-Disperse Mixtures within Lunar Lander Plume-Surface Interaction.............................64 Peter A. Liever and Jeffrey S. West Applicability of Simulants in Developing Lunar Systems and Infrastructure: Geotechnical Measurements of Lunar Highlands Simulant LHS-1........................................76 Jared Long-Fox, Michael P. Lucas, Zoe Landsman, Catherine Millwater, Daniel Britt, and Clive Neal Lunar Dust Simulants and Their Applications..........................................................................86 A. Madison, Z. Landsman, J. Long-Fox, A. Metke, K. Krol, P. Easter, C. Sipe, L. Weber, and D. Britt Traction of Interlocking Spikes on a Granular Material..........................................................95 Volker Nannen and Damian Bover Static and Kinetic Friction Coefficients for Regolith Delivery into a Molten Regolith Electrolysis Reactor.....................................................................................................106 Jason Bendixen Noe, Paul van Susante, Laurent Sibille, and Jafet Pinto-Reveggino

© ASCE

Earth and Space 2022

vii

Power Measurements to Excavate Lunar Soil Simulant GRC-3B Using Arc Backhoe Trajectories..................................................................................................................120 Margaret P. Proctor and Fransua Thomas Magnetic and Microwave Susceptibilities of Lunar Simulants and Their Constituents for Use in Creating Building Materials via Beneficiation.................................130 D. M. Sapkota, P. T. Metzger, A. St. John, M. Phillippe, and E. Franks The Specialized Penetrometer Instrument: SAMPLR and Beyond......................................137 B. C. Thrift, C. B. Dreyer, M. A. Seibert, S. P. Dougherty, and B. A. Cohen Discrete Element Modeling of Drilling Process into the Lunar Regolith Simulants of JSC-1A..................................................................................................................150 Lei Wang, Jesus Badal, Qiushi Chen, Pablo Sanchez Guerrero, and Quan Sun RTG Radiator Efficiency in the Presence of Lunar Dust........................................................160 M. M. Wittal, S. A. Miaule, M. A. Guerrero Nacif, A. M. Swanger, and J. G. Mantovani Exploration and Utilization of Extraterrestrial Bodies Penetration Analysis of High-Frequency Vibro-Based Probes in Granular Media Using the Discrete Element Method..............................................................................176 Mahdi Alaei Varnosfaderani, Pooneh Maghoul, and Nan Wu SPARTA—A New Geotechnical Tool for Subsurface Exploration.........................................184 Robert C. Anderson, Keith Chin, James Dohm, Luke Sollitt, and Kris Zacny Analysis of Sintered Hawaiian Basalt Building Blocks for Landing Pad Use and Recommendations for Improvement........................................................................................199 Chase Dickson and Patrick Suermann Some Key Explorations in Planetary Rover Autonomy for ISRU Roles on the Moon................................................................................................................................207 A. Ellery Discrete Element Method and Multi-Body Dynamics Co-Simulation Framework for Regolith-Tool Interaction Modeling....................................................................................223 Daniel Gaines, Qiushi Chen, Laura Redmond, and Thomas Delvaux Repurposing Drilling Control Diagnostics for Subsurface Edge Detection and Boundary Advisement during Planetary Drilling....................................................................232 B. Glass, T. Stucky, S. Seitz, A. Dave, and R. Haynes Ablative Arc Mining for In Situ Resource Utilization.............................................................243 Amelia D. Greig

© ASCE

Earth and Space 2022 

Testing of a Bucket Ladder Excavation Mechanism for Lunar Applications.......................255 Marcello C. Guadagno, Paul J. van Susante, George Johnson, Zach Crook, Isaac Genther, Ted Gronda, Declan King, Emily Ladensack, Tyler Lupinski, Tyler Rahkola, Colin Schaefer, Grace TenBrock, and Erik VanHorn Rock Breaking Techniques Using High Concentrated Energy Sources for Space Mining Applications...................................................................................................................266 Bharat K. Jasthi, James Tomich, Purushotham Tukkaraja, Prasoon Diwakar, Gokce Ustunisik, and Matthew J. Dietz Method for Thermal Modeling and Volatile Measurement of Lunar Regolith..................273 George Johnson, Travis Wavrunek, Anurag Rajan, Paul J. van Susante, Timothy Eisele, and Jeffrey S. Allen Field Testing of Simulated Lunar Ice Characterization Using Ground Penetrating Radar Technology..................................................................................................281 Caleb J. Kaminski, Paul J. van Susante, and Timothy Scarlett Leaching of Lunar Regolith for Synthetic Phyllosilicates on the Moon................................291 David Karl, Kevin M. Cannon, and Aleksander Gurlo Local Resource Utilization of Lunar Regolith for Manufacturing at the Point-of-Need of Metal Matrix Composites.............................................................................298 Jessica J. Lopez, Malcolm B. Williams, Timothy W. Rushing, J. Brian Jordon, J. A. Cartwright, Gregory B. Thompson, and Paul G. Allison NASA Science Technology Development Programs for Ocean Worlds Exploration............308 Carolyn R. Mercer, Ryan A. Stephan, Mary A. Voytek, Curt Niebur, Lucas Paganini, Mitchell D. Schulte, Quang-Viet Nguyen, K. Michael Dalal, and Hari D. Nayar A Review of Extra-Terrestrial Regolith Excavation Concepts and Prototypes....................321 Robert P. Mueller Lateral Stability of Vehicle with Interlocking Spikes..............................................................332 Volker Nannen and Damian Bover Liberation of Mineral-Bound Water of the Meridiani Planum Driven by Process Heat from Carbonylation Steel-Making and Concentrated Photovoltaic Electricity Generation................................................................................................................343 Rif Miles Olsen Redwater: Extraction of Water from Mars’ Ice Deposits.......................................................355 Joseph Palmowski, Kris Zacny, Boleslaw Mellerowicz, Brian Vogel, Andrew Bocklund, Leo Stolov, Bernice Yen, Dara Sabahi, Lilly Ware, David Faris, Albert Ridilla, Huey Nguyen, Paul van Susante, George Johnson, Nathaniel E. Putzig, Michael Hecht, and Hari Nayar

© ASCE

viii

Earth and Space 2022

Rapid Extraction of Volatiles from Excavated Icy Regolith Using a Rotary Extraction Drum........................................................................................................................363 C. A. Purrington, D. Purcell, Jon Schmit, and B. Thrift Volatile Prospecting through Thermal Properties of Subsurface Icy Regolith.....................374 C. A. Purrington, C. Dreyer, and P. Abel Site Characterization for the RedWater ISRU System...........................................................386 A. T. Russell, N. E. Putzig, and K. Zacny ISRU Pilot Excavator: Bucket Drum Scaling Experimental Results....................................394 Jason Schuler, Andrew Nick, Kurt Leucht, Austin Langton, and Drew Smith Practical Space Resource Utilization at the Hundred Megatonne Scale: Enabling a Planetary Sunshade to Reverse Global Warming................................................408 Liz Scott, Dorian Leger, David Borncamp, Scot Bryson, Peter E. Corwin, Maxwell C. Sissman, and Ross Centers SMART: Instrumented Drill for ISRU Investigations on the Moon....................................423 Leo Stolov, Kris Zacny, Jennifer Heldmann, Kathryn Bywaters, Sofia Kwok, Carter Fortuin, Anthony Colaprete, Arwen Dave, Richard Elphic, Dayne Kemp, and Keith B. Chin Influence of Ice Distribution on Thermal Mining Performance and Strategies to Counter Sublimation Lag..........................................................................................................437 T. Gordon Wasilewski Commissioning and Testing a New Dusty Thermal Vacuum Chamber with Inclusion of Icy Regolith............................................................................................................444 Benjamin David Wiegand, Marcello Guadagno, and Paul van Susante Molten Regolith Electrolysis Using Concentrated Solar Heating..........................................454 Hunter Williams, Timothy Newbold, Kevin Grossman, Evan Bell, Elspeth Petersen, Jaime Toro Medina, Jeff Dyas, and Laurent Sibille TRIDENT Drill for VIPER and PRIME-1 Missions to the Moon.........................................465 Kris Zacny, Philip Chu, Vince Vendiola, Paul Creekmore, Phil Ng, Sam Goldman, Emily Seto, Kathryn Bywaters, Ezra Bailey, Raymond Zheng, Lilly Ware, Ash Rashedi, Phil Beard, Paul Chow, Stella Dearing, Amelia Grossman, Robert Huddleston, Kevin Humphrey, Anchal Jain, David Lakomski, Zach Mank, Gale Paulsen, Sara Martinez, Tom O’Bannon, Aayush Parekh, Jeff Shasho, Alex Wang, Jack Wilson, Helen Xu, Jackie Quinn, Amy Eichenbaum, Janine Captain, and Julie Kleinhenz

© ASCE

ix

Earth and Space 2022 

x

Advanced Materials and Designs for Aerospace and Terrestrial Structures under Extreme Environments A Novel Partitioned Approach for Reduced Order Model—Finite Element Model (ROM-FEM) and ROM-ROM Coupling......................................................................475 Amy de Castro, Paul Kuberry, Irina Tezaur, and Pavel Bochev Lessons Learned and Best Practices for Utilizing a Generalized Composite Impact Model..............................................................................................................................490 Robert K. Goldberg and Trenton M. Ricks Modeling and Analysis of a Nonlinear Locally Resonant Metamaterial with Inductance Shunt........................................................................................................................504 Arun Malla, Mohammad Bukhari, and Oumar Barry Ballistic Impact Simulations of a Titanium 6Al-4V Generic Fan Blade Fragment on an Aluminum 2024 Panel Using *MAT_224 in LS-DYNA................................................516 Chung-Kyu Park, Kelly Carney, Paul Du Bois, Cing-Dao Kan, and Daniel Cordasco Prediction of Fracture Location of Duplex Stainless Steel Welds..........................................529 C. Payares-Asprino Study of Aircraft Structural Response and Occupant Loading during a Water Ditching Event Utilizing LS-DYNA Simulation.......................................................................542 Jacob B. Putnam and Karen E. Jackson A Compact Delayed Photocurrent Model Based on a Reduced Order Data-Driven Exponential Time Integrator...............................................................................556 K. Chad Sockwell, Pavel Bochev, and Biliana Paskaleva Turk Salty Concrete (TSC) Can Isolate the Freshwater Interface against the Sea Water Intrusion and Salty Formations..............................................................................571 Afshin Turk Structures in Challenging Environments: Dynamics, Controls, Smart Structures, Health Monitoring, and Sensors Risk-Based Structural Optimization Framework for Connected Structural System Subjected to Extreme Events........................................................................................587 William Hughes and Wei Zhang Viability of Construction Material within an Extraterrestrial Environment.......................602 Linda E. Kuster and Justin D. Delorit

© ASCE

Earth and Space 2022

Detection of Corrosion-Induced Damage in Bolted Steel Structure Using Piezoceramic Transducers........................................................................................................616 Wen-I Liao, Chien-Kuo Chiu, Rei-Ching Huang, and Meng-Xun Lu Seismic and Resilient Property Analysis of SMA-Based Replaceable BRBs........................628 Q. Y. Pan, S. Yan, and X. Su Seismic Evaluation of Lava Tubes Subjected to Moonquakes................................................641 Hamed Seifamiri, Pooneh Maghoul, Richard Boudreault, Najib Bouaanani, and Roberto de Moraes Examination of Smart Sandbags for Semi-Permanent Structures on the Lunar Surface..................................................................................................................652 Yinan Xu, Jiawei Qiu, Virupakshan Vilvanathan, Athip Thirupathi Raj, and Jekan Thangavelautham Automatic Reading Method for Pointer Meter Based on Computer Vision.........................669 Weijin Xu, Weihua Zhang, Liang Xing, Hongjun Lu, Dongyou Li, and Yang Du Identification Method for Displacement of Substation Structure Based on Machine Vision...........................................................................................................................681 Weijin Xu, Weihua Zhang, Liang Xing, Hongjun Lu, Dongyou Li, and Yang Du Space Engineering, Construction, and Architecture for the Moon, Mars, and Beyond Applying Architectural Design and Construction Principles to Lunar and Martian Construction.........................................................................................................688 Erin Brayley|Werkema and Patrick Suermann A Summary of Technical Requirements, Environmental Factors, and Loading for Lunar Infrastructure............................................................................................................701 Nerma Caluk and Atorod Azizinamini Stone, Brick, and Concrete Masonry on Mars........................................................................717 Peter Carrato, Keith Kennedy, Cory Brugger, Jennifer Heldman, and Darlene Lim Structural and Durability Properties of MgO-Al2SiO3 Concrete for ISRU Martian Construction..............................................................................................................729 Milap Dhakal, Allan N. Scott, Rajesh P. Dhakal, Christopher Oze, and Don Clucas Lunar Demandite—You Gotta Make This Using Nothing but That.....................................743 A. Ellery Is In Situ Electronics Fabrication Feasible on the Moon?......................................................759 A. Ellery

© ASCE

xi

Earth and Space 2022 

Solar Power Satellites—Rotary Joints, Magnetrons, and All—From Lunar Resources?.....................................................................................................................................773 A. Ellery In Situ Lunar Launch and Landing Pad Construction with Regolith-Thermoset Polymer Composite Materials..................................................................................................789 Nathan J. Gelino, Robert P. Mueller, Laurent Sibille, Kyle L. Dixon, Jonathan Gleeson, and Brad Buckles Microwave Sintering of a Lunar Regolith Simulant for ISRU Construction: Multiscale Characterization and Finite Element Simulation.................................................804 Shayan Gholami, Xiang Zhang, Young-Jae Kim, Yong-Rak Kim, Bai Cui, Hyu-Soung Shin, and Jangguen Lee Planetary Construction 3D Printing Using Lunar and Martian In Situ Materials..............817 Ilerioluwa Giwa, Dharla Moore, Michael Fiske, and Ali Kazemian Design, Development, and In Situ Testing of Lunar Technologies.........................................832 Allison Goode Investigation of Martian Concrete: Pumpability and Strength for Extraterrestrial 3D Printing......................................................................................................846 Courtney Keys, Noah McCorkhill, and Yoon-Si Lee Lunar Base Construction Planning...........................................................................................858 Robert P. Mueller Metamodels for Rapid Analysis of Large Sets of Building Designs for Robotic Constructability: Technology Demonstration Using the NASA 3D Printed Mars Habitat Challenge.......................................................................................................................871 Naveen Kumar Muthumanickam, José Pinto Duarte, Shadi Nazarian, and Sven G. Bilén Mixed Reality (XR) as a Validation Method for Digital Modeling of Space Habitats........885 Vittorio Netti, Olga Bannova, Jasleen Kaur, and Richard Spolzino A Review of Additive Manufacturing Technologies for Planetary Construction.................893 Vittorio Netti and Tara Bisharat Systems Engineering of Using Sandbags for Site Preparation and Shelter Design for a Modular Lunar Base.........................................................................................................904 Athip Thirupathi Raj, Jiawei Qiu, Virupakshan Vilvanathan, Yinan Xu, Erik Asphaug, and Jekan Thangavelautham Playing with DIRT: Building the Framework for a Comprehensive Materials Database.........920 S. J. Seitz, R. Haynes, and B. J. Glass

© ASCE

xii

Earth and Space 2022

Minimal Lunar Infrastructure to Facilitate the Construction of Sustainable Structures.........935 Richard Spolzino, Jasleen Kaur, Vittorio Netti, and Olga Bannova Air Pressure and Temperature Distribution within a Dome Habitat Structure on the Lunar Surface.......................................................................................................................946 Jeffrey T. Steiner, Sachin Tripathi, and Ramesh B. Malla Turkish Lunar Regolith Simulant TBG-1................................................................................956 Y. Cengiz Toklu, Nurcan Çalış Açıkbaş, Gökhan Açıkbaş, Ali Erdem Çerçevik, and Pınar Akpınar Diurnal Temperature Variation on an Intact and Damaged Lunar Habitat Structure.......966 Sachin Tripathi, Jeffrey T. Steiner, and Ramesh B. Malla Mars Dune Alpha: A 3D-Printed Habitat by ICON/BIG for NASA’s Crew Health and Performance Exploration Analog (CHAPEA).....................................................976 M. Yashar, C. Glasgow, B. Mehlomakulu, J. Ballard, J. O. Salazar, S. Mauer, and S. Covey

© ASCE

xiii

1

Earth and Space 2022

Granular Mechanics of JSC-1 Mars Regolith Simulants Reza Ashtiani1; Jesus Baca2; Jacob Wessel3; and Arash Karimipour4 1

Visiting Professor, United States Air Force Academy (USAFA) Graduate Research Assistant, Univ. of Texas at El Paso (UTEP) 3 Graduate Research Assistant, Univ. of Texas at El Paso (UTEP) 4 Graduate Research Assistant, Univ. of Texas at El Paso (UTEP) 2

ABSTRACT Any Martian mission entailing establishment of a habitat or vehicles operating on planetary surfaces requires proper characterization of native soils. Planetary geology along with regolith properties greatly influence the construction parameters, such as excavation mechanisms, and mechanical interactions between structures and the regolith platform. The primary objective of this research was to mechanistically characterize the physical and mechanical properties of the JSC Mars-1 regolith simulants to better understand the physical, mechanical, and physio-chemical properties of the native soils and its relevance to planetary construction. To achieve these objectives, a comprehensive experiment matrix was devised to simulate the compaction characteristics of unbound particulate materials in the laboratory. The resilient properties, strength parameters, and deformation potential were of primary interest in this effort. Additionally, the strength and deformation potential of the JSC-1 Martian simulants were contrasted with a conventional terrestrial construction material for comparative purposes. The study also investigated the contribution of the magnitude and method of application of the compaction energy on the dilatancy behavior of particulate materials in the laboratory. The laboratory results showed appreciable strain rate dependency of the strength parameters for simulants at multiple relative compaction levels. This underscores the significance of nonlinear and anisotropic regolith modeling for proper determination of orthogonal strength characteristics, settlement potential, and stability of platforms for planetary construction. INTRODUCTION Mars has a diverse geology with district geotechnical characteristics, for instance the core of Mars is primarily made up of metallic iron and nickel encased in a less dense silicate mantle and crust (Nimmo and Tanaka, 2005). Data obtained from previous missions show the primary chemical elements on Mars’s crust are silicon, oxygen, iron, magnesium, aluminum, calcium, and potassium (Dhir et al., 2018). These chemical elements are main constituents of both native regolith and other relevant studies at the Viking 1, Viking 2, and Pathfinder landing sites. The data gathered from Viking 1, Viking 2, and Pathfinder landers enabled the determination of regolith chemical compositions, particle-size distributions, and porosity characteristics at each landing site (Dhir et al., 2018). The physical characteristics such as particle size, and particle size distributions combined with the mineralogy studies, were the basis for the development of the regolith simulants for further studies. (Fackrell 2018). JSC Mars-1 was developed using weathered volcanic ash, sourced from the Pu'u Nene cinder cone on the island of Hawaii, and characterized at the NASA Johnson Space Center (JSC). JSC Mars-1 is a mixture of ash particles with alteration rinds of various thicknesses (Dhir et al., 2018). The ash is composed of fine crystalline structure and glassy

© ASCE

Earth and Space 2022

particles of Hawaiite composition (Fackrell 2018). Previous research provided comparisons between the JSC Mars-1 simulant mineral composition and the regolith sampled by the Viking 1, Viking 2, and Pathfinder landers (Fackrell,2018). Some researchers elaborated on the role of the particle size, particle size distribution and crystalline structure of the regolith simulants, and concluded that the similarity between the JSC-1 Mars simulants and the data collected from the Viking 1, Viking 2, and Pathfinder landing sites make this simulant suitable for further geotechnical characterization (Gabasova and Kite 2018) As stated earlier, due to the diversity of the Mars geology, experimental results from a single regolith simulant cannot be extrapolated to diverse regions to completely characterize the physical and mechanical properties of the Martian surface (Scott et al., 2017). Therefore, in the effort to delineate the Martian regions that can be represented by JSC Mars-1, the locations corresponding to the Viking 1, Viking 2, and Pathfinder landing sites have been studied in terms of geological and topographic conditions. Field data collected from Viking, Pathfinder, and the Mars Exploration Rover (MER) Mission shows that most of the studied areas on Mars consisted of particles smaller than 50 μm. The researchers used image analysis techniques from previous missions to define the particle sizes of the fine fraction of the surface regolith. The authors concluded that the primary fraction of the fine particles were smaller than 45 μm, while fewer fine sands had a particle size in approximate range of 130-160 μm (Zimmerman, 2016). It’s important to note that the particle size and mineralogy are merely the initial steps in geotechnical characterization of particulate materials. This information should be supplemented by the particle geometrical characteristics such as flat and elongation of particles, angularity, and surface macro-texture to properly assess the interlocking effect, inter-particle frictional properties, and orthogonal load bearing capacity of the platform. Additionally, the density of the regolith medium plays a crucial role in distribution of the load, settlement characteristics, dust lift off, and plume formation in low gravitational environments. This was the motivation to develop an experiment design that incorporates the geometrical features of the particles and different compaction methods for more realistic simulation of the unbound granular native soils in this study. RESEARCH SIGNIFICANCE Unbound granular soils are anisotropic, nonlinear, and stress-dependent materials. The orthogonal load bearing capacity of particulate frictional materials, such as regolith simulants, is greatly influenced by the complex nature of the stress paths, relative compaction in the zone of influence, moisture state, and rate of the application of loads (Ashtiani et al. 2018). For this investigation, the influence of the surface tension properties at the menisci is overlooked as there’s no evidence of the presence of unbound moisture on Mars. In this study, in addition to particle size and particle size distribution, several other parameters and properties were considered for the estimation of the strength parameters, and deformation characteristics of the regolith platform subjected to static and dynamic loading in low gravitational environments. Geometrical properties such as particle form, angularity, and surface textures are of paramount importance for the anisotropic characterization of the regolith simulants. Therefore, a particle imagining system setup was used to properly characterize the geometrical features of the particles in this study. Several methods of application of compaction energy were incorporated in this study to better understand the influence of the particle packing on the strength and deformation potential of the regolith simulants. The information regarding the strength parameters at loosest and densest states

© ASCE

2

Earth and Space 2022

are crucial for proper analysis of the stability of the platform, deformation under the rover wheels and stability of sand dunes in operational environments. The compaction characteristics and its influence on the settlement and stability of the regolith media is often overlooked in regolith mechanics literature. The authors believe it is imperative to incorporate the densification mechanism and compaction energies relevant to planetary conditions for proper simulation of the regolithstructure interactions. The current study explores different methods of application of compaction that result in variable particle orientations and anisotropic particle contact distributions. On planetary surfaces, densification processes of regolith can be attributed to meteorite impacts, seismic activity, and anthropogenic effects (e.g., rover traverses, footprints, etc.). The purpose of exploring different compaction techniques is to determine the influence that the resulting density has on the mechanical properties of JSC Mars-1. The variability of the levels of confining pressure (σ3) exerted on regolith simulants in the laboratory is instrumental for representing the densification due to external loading conditions on the Mars surfaces. Once the strength parameters such as angle of internal friction is determined in the laboratory, the results can be contrasted with available imagery information, such as free-standing slopes for verification of the results. For instance, the angle of repose measured at the Hesperian geological units that are associated with the subject landing sites may provide an indication of the in-situ performance of the mechanical properties of these regolith materials. The information can provide insight on the stability of slopes and likelihood of collapse of the dunes due to vibrations imparted by rovers operating in the proximity of the Aeolian formations. LABORATORY PROGRAM A comprehensive laboratory experiment design was developed to understand the complex geotechnical characteristics of the JSC Mars-1 regolith simulants. Due to particle crushing potential and subsequent alteration of the mechanical properties, the regolith specimens were not reused in destructive mechanical tests in this study. The main testing categories can be summarized as: - Index properties: Particle size distribution and characterization of the fine fraction of the postulate mix. - Particle morphology: angularity and sphericity of particles at several size fractions. - Electrical resistivity - Density and compaction: using raining, impact, gyration, and vibration methods. - Mechanical properties: compressibility, shear strength and dilation behavior. RESULTS AND DISCUSSION The following section provides the laboratory results and post processed data for each category of the tests outlined in previous section. a. Particle Size Distribution Figure 1 provides the particle-size distribution curve for JSC Mars-1 developed in this study. The coarse portion was characterized by performing a mechanical sieve analysis (ASTM C136 and C117). Subsequently, the gradation parameters corresponding to the fine portion of the material were determined by following the hydrometer analysis (ASTM D422). For JSC Mars-1, the gradation parameters that characterizes the shape and scale of the cumulative

© ASCE

3

4

Earth and Space 2022

distributions, namely Cc and Cu, were determined to be 1.23 and 4.31, respectively. Based on these values, and the grain size distribution curve Figure 1, the material is classified as poorly graded sand with silt (SP-SM), according to the Unified Soil Classification System (USCS; ASTM D2487). Mechanical Sieve Analysis

100

Hydrometer Analysis

% Passing

80

60 40 20 0 0.075 Particle Size (mm)

0.001

Figure 1. Particle Size Distribution Curve for JSC Mars-1 b. Characterization of the Fine Fraction of JSC Mars-1 Simulants In contrast with coarse-grained particles, the fine fraction of the particulate mix is characterized by the moisture adsorption capacity and percent fines present in the mix. Due to the absence of unbound moisture on Mars, the plasticity information was not relevant in this study, however, the traditional ASTM D4318 procedure resulted in a negligible Plasticity Index (PI). The particle size distribution analysis for the fine fraction of the mix was carried out using the hydrometer test. This test is primarily based on the Stoke’s law of sedimentation as outlined in ASTM D422. Hydrometer analysis, in combination with mechanical sieve analysis, revealed that most fine particles in JSC Mars-1 are silt-sized 72.9% silt with the remainder 27.1% as clay particles. Table 1 provides the summary of the results obtained from the mechanical and hydrometer sieve analysis on the entire particulate mixture. Table 1. JSC Mars-1 Composition by Soil Group Soil Group Concentration (%)

Gravel (G) 0

Sand (S) 93

Silt (M) 5.1

Clay (C) 1.9

c. Particle Geometry The geometrical characteristics of the coarse portion of JSC Mars-1 were analyzed using a state-of-the-art Aggregate Image System (AIMS) (Lees, 1964). The inter-locking effect of particles is associated with high angularity index values, which results in a higher angle of repose and dilatancy behavior of granular soils (Sibille et al., 2006). AIMS results revealed that the angularity index increases with the diameter of the particles as evidenced in Figure 2. In other words, larger particles exhibited more pronounced jagged fractured faces and broken edges in the analysis. This phenomenon is associated with the crushing behavior and crystalline structure of the parent rock. The particle crushing analysis also showed the susceptibility of the larger

© ASCE

5

Earth and Space 2022

particles to break into fractures faces with a high angularity index. In fact, particle breakage, which is defined as lateral shift in the gradation curve and generation of particles smaller than 0.075 mm, was more pronounced for the coarse fraction of the mix. According to Figure 2, particle angularity is greatly influenced by particle size, with particles larger than 0.149 mm exhibiting higher angularity compared to particles smaller than 0.149 mm. Based on Krumbein angularity index analyses on over 200 particles, approximately 67% of particles in JSC Mars-1 are larger than 0.149 mm in diameter, exhibited high and extremely high angularity indices. High angularity indices range between 3,975 and 5,400 in the Krumbein scale, whereas extreme angularity indices range between 5,400 and 10,000. In other words, the concentration of JSC Mars-1 particles with low angularity ranges in distributions increases as particle diameter decreases. Conversely, the concentration of particles with high and extremely angularity indices tends to increase as particle diameter increases. 100 90 80 70 60 50 40 30 20 10 0

Cumulative Particles (%)

Sieve #30 (0.595 mm) Sieve #50 (0.297 mm) Sieve #100 (0.149 mm) Sieve #200 (0.074 mm)

300

3000 AIMS Angularity Index

Cummulative Particles (%)

Figure 2. Characterization of angularity index for JSC-1 Mars 100 90 80 70 60 50 40 30 20 10 0

Sieve #30 (0.595 mm) Sieve #50 (0.297 mm) Sieve #100 (0.149 mm) Sieve #200 (0.074 mm) 3

AIMS 2D Form Index

30

Figure 3. Particle sphericity distribution. Sphericity is a relative indication of how round or elongated a particle is compared to a perfect sphere. High values of the sphericity index indicate elongated particles with an ellipsoidal

© ASCE

6

Earth and Space 2022

shape, while, low values of the sphericity index indicate more rounded geometries. Unlike the angularity index, the variability of the sphericity index over different particle sizes were in a relatively tight range. The high sphericity values for different particle sizes is primarily an indication of significant different in aspect ratios of particles. In other words, particulate systems consisted of large amounts of flat and elongated particles tend to exhibit high directional dependency of material properties and therefore high anisotropic behavior. Therefore, it is important to consider the variability of the stiffness properties in orthogonal planes for proper characterization of the load bearing capacity of regolith platforms. d. Electrical Resistivity Electrical resistivity is defined as the ability of a material to resist electric current. In the construction industry, the electrical resistivity of soils is used to provide insight on issues pertaining to corrosion and presence of anomalies or discontinuities in the strata (Wadell, 1933). Corrsivity is the relationship between electrical resistivity and moisture content at a single degree of relative compaction. A Miller box and the Vibroground Model 293 assembly was used to asses the electrical resistivity of regolith simulants following the TEX-129-E protocol (TxDOT, 2017). The experiment was repeated twice for different levels of compaction to better understand the synergistic influence of the compaction level, moisture content and electrical resistivity of the regolith medium. The results indicate that the electrical resistivity of JSC Mars-1 asymptotically approaches 0 𝛺m for moisture contents more than 30%. e. Density of Simulants The minimum index density was determined by following the ASTM D4254 test procedure, a method commonly known as the raining technique. The relationship between the minimum and maximum index density test results for JSC Mars-1 are shown in Figure 4.

Density (kg/m3)

1100 1050 1000 950

Dr% at t=8 min. 900 850 0

10

20

30

40

50

60

70

80

90

100

Rheostat Capacity (%)

Figure 4. Maximum index density values achieved by vibration at different rheostat capacities, with respect to minimum index density (Dr%). The most common testing method for laboratory determination of the maximum dry density is the modified Proctor test in geotechnical practice (ASTM D1557). This method consists of developing a moisture-density relationship that provides an indication of the amount of moisture

© ASCE

7

Earth and Space 2022

required to reach a maximum packing of granular materials. In the densification process by using the modified Proctor test, moisture acts as a lubricating agent and facilitates particle rearrangement when subjected to impact forces. This rearrangement reduces the void ratio and therefore, produces an increase in the wet density of the compacted specimen. The achieved dry density is then measured by subtracting the amount of moisture required to achieve the given wet density. However, due to the absence of hygroscopic moisture on Mars, and to replicate the densification processes by impact on the red planet, the variability of the density with moisture was disregarded and all tests were performed at zero moisture content immediately after 24 hours of oven drying. The achieved dry density of JSC Mars-1, from the modified Proctor test was of 1,158.2 + 16.0 kgm-3. Additionally, for reference purposes, the measured minimum index density is included in this plot with a rheostat capacity ranging from 20% to 100% in the shake table tests. i. Vibratory Compaction Regolith materials subjected to vibration may reduce their void ratio by the collapse of pore spaces, and therefore, an increase in its dry density. The compaction process by vibration is replicated in the laboratory with the maximum index density test (ASTM D4253, also known as the shake table method. This test procedure utilizes a surcharge load applied at the top of the soil specimen, which is confined in a calibrated mold. The test is performed while the entire assembly is subjected to vertical vibration for specified time. The densification of JSC Mars-1 was also studied by using the British vibratory hammer method (ASTM D7382). In contrast to the shake table, the British vibratory hammer method does not apply a dead load at the top of the cross sectional area of the specimen. The British vibratory hammer approach depends primarily on densification by vibration, but with a higher efficiency of transference of the compaction energy than the traditional shake table. The compaction energy applied by the British vibratory hammer can also be quantified according to Arcement & Wright (2001) procedure. A higher density was achieved for JSC Mars-1 using the British vibratory hammer compared to the shake table. A possible explanation for this is the variability of the respective compaction energy (CE) values. Table 2 provides a summary of the achieved dry densities in relation with the applied compaction energy values. As shown in Table 2, the achieved dry density tends to increase as the compaction energy increases. Table 2. Summary of Results for Compaction by Vibration Methods Compaction Method

Rheostat Capacity (%)

Dry Density (kg m-3)

Shake Table

20 35 50 75 100

1049.9 1062.3 1066.8 1074.9 1081.4

Standard Deviation (kg m-3) 0.9 0.5 3.9 3.5 4.4

British Vibratory Hammer

100

1124.5

20.8

Compaction Energy (kN-m/m3) 4068.3 7119.4 10170.6 15255.9 20341.3 31947.0

ii. Gyratory Compaction Gyratory compactor utilized combination of static pressure and gyration to shift and readjust the particles into a dense packing. The particle orientation, and therefore directional dependency

© ASCE

8

Earth and Space 2022

of the material properties, using the gyratory compactor is more representative of the manner in which the compaction energy is applied by compaction equipment in the field. Several variables were considered for simulation of the densification of the regolith simulants using the gyratory compactor in this study: Normal pressure (σ1, kPa), the amount of static pressure applied at the top of the soil specimen in a standardized mold, Gyration rate, (Gr, revolutions per minute), and Angle of gyration (Gº) which is the angle of the interior of the mold wall with respect to the bottom plate. A series of gyratory compaction tests were performed on JSC Mars-1 using various values of Gr and σ1. The results are summarized in Table 3. Table 3. Gyratory Compaction Test Results Normal Pressure (kPa)

Gyration Rate (RPM) 5.0 10.0 30.0 60.0 5.0 10.0 30.0 60.0 5.0 10.0 30.0 60.0

206.8

503.3

799.8

Change in Height ∆H (mm.) 4.38 10.08 8.14 7.36 9.374 10.50 9.52 9.39 10.56 12.76 11.91 11.76

Dry Density (kN/m3) 932.3 975.5 964.3 946.7 1023.6 1052.4 1047.6 1030.0 1090.9 1124.5 1118.1 1111.7

Compaction Energy (kN-m/m3) 13 23 25 30 76 85 75 75 141 175 163 160

In total, 24 tests were performed to determine the optimum combination of compaction parameters that result in the highest dry density value of JSC Mars-1. Figure 5 shows the relationship between dry density and Gr, at three levels of static pressure (σ 1). As evidenced in Figure 5, the governing parameter in the densification of JSC Mars-1 by gyratory compaction methods is the magnitude of the static pressure. Each curve represents the relationship of achieved dry density values and gyration rates, at the same level of normal pressure. An increase of 150 kg m-3 in the dry density was achieved by increase of the normal pressure from 206.8 kPa to 799.8 kPa. Conversely, the maximum value of 43.2 kg m -3 was obtained by increasing the dry density value due to the application of different gyration rates. 1150

Density (kg/m3)

1100 1050

σs=206.8 kPa σs=503.3 kPa σs=799.8 kPa

1000 950 900 0

10

20 40 50 60 Gyration Rate 30 (Revolutions per Minute)

70

Figure 5. Dry density (kg m-3) of JSC Mars-1 vs gyration rate (RPM) © ASCE

Earth and Space 2022

The variation in density of JSC Mars-1 as a function of gyration rate is associated with segregation of the fine-grained particles with respect to the coarse-grained particles. As the gyration rate increases above the optimum value, the fine-grained particles tend to migrate to the specimen's outer perimeter due to the centrifugal forces. These forces are reduced as the confinement pressure increases which in turn decreases the amount of segregation. Therefore, the peaks of the curves in Figure 5 are more visible at lower levels of confinement. On the other hand, as the gyration rate decreases below the optimum value, JSC Mars-1 tends to homogenize within the compaction mold. The fine-grained particles generated by crushing, primarily due to the exerted confinement pressure, tend to re-accommodate in the void space between the coarsegrained particles, which in turn results in increase in the dry density of the specimen. This behavior was also observed for the gyration rates that are lower than optimum values for the three curves. iii. Comparison of Compaction Methods In total, 58 tests were performed to simulate the diversity of the compaction characteristics of regolith of surface of Mars. As depicted in Figure 6, the modified Proctor test proved to be the method that resulted in the highest packing characteristics of the simulants. The British vibratory hammer and the gyratory compactor (σ s=799.8 kPa, GR=10 RPM) had comparable density results despite the variation in the method of application of compaction energy in the two methods. Figure 6 clearly shows the nonlinear nature of relationship between density and compaction effort. This is primarily due to the fact that the particulate materials are stress sensitive and the rearrangement of the particles is greatly impacted by the amount and method application of the compaction energy. The results however, show the general directional trend in improvement in density of the particulate assembly with increased compaction energy. The minimum index density, determined through the raining method, is also provided on the plot to provide a visual inference of the loosest state of the particles in relation to the compacted specimen in this study. As shown in Figure 6, the method with the highest energy of compaction is the British vibratory hammer method, however, this method did not yield the highest dry density. The method that resulted in the lowest degree of relative compaction was the shake table, even though it had the second-highest compaction energy. These observations suggest that an increase in the energy of compaction is not correlated with an increase in the maximum achievable dry density of JSC Mars-1. Compaction by gyration is an efficient method to increase the dry density of JSC Mars-1. This method requires a relatively low compaction effort as compared to compaction by vibration and impact. Compaction by gyration required only 6.50% of the energy required by the Proctor method to achieve 97% relative compaction. Another critical behavior associated with each of the compaction methods used for JSC Mars-1 is particle breakage, or generation of fine-grained materials. Previous research by authors shows that the increase in the fines content of the mix results in the higher softening component of the resilient modulus constitutive model, which turn results in the degradation of the stiffness properties in cross-anisotropic granular layers (Ashtiani, 2021). This phenomenon plays an essential role in densification of JSC Mars-1. In contract, the compaction method that generated the smallest percentage of fines was the shake table as shown in Figure 7. The presence of high porous structures is often collocated with lower density of the medium.

© ASCE

9

10

Earth and Space 2022

1200 1150

Dry Density (kg/m3)

1100 1050 1000 950

Gyratory Compaction

900

Shake Table Modified Proctor

850

Vibratory Hammer 800 0.01

0.1

1

10

100

1000

10000

100000

Compaction Energy (kN-m/m3)

Figure 6. Dry density values obtained by using different methods of compaction. A particle crushing potential analysis was performed subsequent to the compaction of the specimen using different method of compaction. The gradation distributions pre-and-post compaction were juxtaposed on same plot to show the particle breakage potential and fines generation manifested by lateral offset of the cumulative distribution curves. As evidenced in Figure 7, specimens prepared with the British vibratory hammer exhibited the highest degree of particle crushing compared to other permutations of the experiment design.

Figure 7. Particle size distribution curves developed with the material tested under different ASTM compaction standards. f. Influence of the Compaction Characteristics on the Deformation Potential Compressibility is defined as the change in the volume of the specimen associated with the change in the internal structure of the particulate assembly due to an external stimuli (Ashtiani, 2020). The primary sources of disruption in the equilibrium of the particulate assemblies are

© ASCE

Earth and Space 2022

external loading conditions such as static habitat loads or dynamic rover-imparted stresses, seismic activities, and vibrations due to the impact of planetary objects to name a few. The densification, and reduction in the volume of regolith assemblies is primarily due to the expulsion of the air voids and improvements in orthogonal distributions of particle contacts. The volumetric reduction of the regolith layer due to loading conditions and settlement characteristics are integral components for proper design and construction of structures. For this reason, the research team prepared nine samples at various compaction levels to study the synergistic influence of the level of compaction and deformation potential of regolith simulants. The loading protocol proposed by (ASTM D2435) was adjusted to account for the gravitational acceleration on Mars for realistic stress path testing of regolith simulants. Figure 8 shows the behavior of JSC Mars-1 subjected to a sequence of incremental static loads at different initial relative compaction levels. Load Increment Ratio (LIR) of 2 was used for this 1-dimensional compressibility test, as specified by ASTM D2435. An initial axial effective stress value of 13.8 kPa was used to simulate the stress induced by the Viking I landing gear footpads (Moore et al., 1987) on the Martian surface.

Figure 8. Compression curves (axial effective stress in logarithmic scale) for JSC Mars-1 corresponding to different initial relative densities (expressed as percent relative compaction). Figure 8 shows the 1D compressibility results for two of the permutations of the experiment design. As evidenced in this plot, particulate assemblies with higher relative compaction exhibited lower deformation potential. The slope of the curves past the maximum past pressure, or the compression index, is an indication of the progression of the plastic deformations when the micro-structure of the soils break. Evidently, the magnitude and rate of the accumulation of deformation is greatly influenced by the compaction characteristics of the specimen as shown in Figure 8. Quantification of the compressibility parameters, such as compression and rebound indices from the 1D compressibility tests are paramount for the design and construction of habitats in future missions. Therefore, the compressibility parameters obtained from these series of tests for different levels of compaction can provide insight on the stability of the platforms and likelihood of nonhomogeneous settlement of foundations.

© ASCE

11

12

Earth and Space 2022

g. Strength Characteristics A series of direct shear tests were conducted in accordance with ASTM D3080. The input parameters used for the characterization of the shear strength properties of JSC Mars-1 are provided in Table 4. Table 4. Direct Shear Test: Input Parameters Initial Relative Compaction (RC%) Deformation Rate (SR, mm min-1)

75.4 0.4

83.6 1.9

91.8 7.6

99.0 -

In total, 18 specimens of oven-dried JSC-Mars 1 were prepared following the raining technique to achieve the minimum index density of 872.7 + 2.2 kg m -3. The JSC Mars-1 material was oven dried at 110+5 Cº for 24 hours prior to specimen’s fabrication in the laboratory. Subsequent to the 18 samples prepared at the loosest state following the raining method, 36 additional specimens with initial relative compaction of 91.8% and 99% were prepared by tamping method until the required relative compaction values were achieved in the laboratory. Considering the replicates, total of 146 direct shear tests were conducted in this study for full factorial testing of the regolith simulants. Subsequent to stress path testing of the specimens, the strength parameters were determined in accordance with ASTM D3080. Table 5 summarizes the Mohr-Coulomb strength parameters, namely cohesion (c) and angle of internal friction (𝜙), for the permutations of the experiment design in this study. The results indicate the expected ascending trends in the friction angle for specimen compacted at higher relative densities. This is primarily due to improvements in the interlocking effects which translates into higher interparticle frictions manifested in higher friction angles. The results provided in Table 5 clearly shows the strain-rate dependency of the strength properties of the Mars 1 regolith simulants. Permutations subjected to lower strain rate protocols exhibited higher frictional properties quantified by the friction angles in Mohr-Coulomb failure envelopes. This is primarily due to the distortion mechanism of particulate materials and rate of imposed deformations along the split planes in the DST tests. Similar trends were observed for the shear strength at the onset of peak state denoted by τ in Table 5. The results indicate that there is an inverse relationship between the strain rate and shear strength of the regolith simulants. This information is instrumental for proper assignment of the shear strength of the platform subjected to dynamic loads such as movement of rovers, as the velocity of the travelling rover is correlated with the rate of imposed dynamic loads on planetary surfaces. Table 5. Direct Shear test: Summary of Results Strain Rate

© ASCE

Relative Density: 75.4%

Relative Density: 83.6%

Relative Density: 91.8%

Relative Density: 99%

SR

𝜙

c

τ

𝜙

c

τ

𝜙

c

τ

𝜙

c

τ

(in./min)

(Deg.)

(kPa)

(kPa)

(Deg.)

(kPa)

(kPa)

(Deg.)

(kPa)

(kPa)

(Deg.)

(kPa)

(kPa)

7.6

46.7

15.2

563.3

47.4

24.8

588.1

50.0

30.3

646.7

51.7

44.1

656.9

1.9

47.0

15.2

570.2

47.6

29.6

595.7

49.5

39.3

644.7

50.7

44.1

635.5

0.4

48.8

2.8

593.6

49.6

13.1

620.5

50.6

29.0

658.4

52.4

46.2

675.1

13

Earth and Space 2022

h. Dilation Behavior Particulate materials subjected to stress path testing protocols exhibit two types of behaviors: dilation and compaction. The angle of dilation (Ψ) is a soil parameter that quantifies the tendency of granular soils to increase their volume as a result of a combination of normal and shear stresses (Chen & Kutter, 2009). Figure 9 provides an overall representation of the dilation (Ψ) behavior of JSC Mars-1 subjected to the combination of normal and shear stresses for a range of initial relative compaction and imposed strain rates. As evidenced in this plot, specimens with high initial density exhibited a high angle of dilation. In addition, the angle of dilation for JSC Mars-1 tends to increase as the strain rate decreases as shown in Figure 9. The dilative behavior of Mars 1 regolith simulants under the imposed stress path tests when combined with high strain rate dependency results shows the importance of nonlinear material modeling for proper characterization of the regolith-structure interactions in low gravitational environments. 35.0

RD: 99%, SR: 7.6 mm./min. RD: 99%, SR: 1.9 mm./min RD: 99%, SR: 0.4 mm./min. RD: 91.8%, SR: 7.6 mm./min. RD: 91.8%, SR: 1.9 mm./min RD: 99%, SR: 0.4 mm./min. RD: 83.6%, SR: 7.6 mm./min. RD: 83.6%, SR: 1.9 mm./min RD: 83.6%, SR: 0.4 mm./min. RD: 75.4%, SR: 1.9 mm./min

30.0

Angle of Dilation

25.0 20.0 15.0 10.0 5.0 0.0 0.0

50.0

100.0

150.0

200.0

250.0

300.0

350.0

Normal Stress (kPa)

Figure 9. Angle of Dilation vs Normal Stress (psi) CONCLUSIONS Total of 345 specimens fabricated using different compaction methods and compaction energies to study the physical indices and mechanical responses of JSC Mars-1 regolith simulants. A full factorial experiment matrix was developed to study the influence of particle size, particle size distribution, regolith geometry, and compaction characteristics on the strength and deformation properties of the Martian regolith simulants. Particle morphology analyses, determined using image analysis techniques on over 1034 particles showed that JSC Mars 1 had substantially higher angularity indices in Krumbein scale compared to typical terrestrial aggregates in geotechnical practice. The results were in line with high friction angle values obtained from the DST strength tests, as particulate assemblies consisted of angular particles provide superior interlocking effects and therefore expected to exhibit improved orthogonal load

© ASCE

Earth and Space 2022

distribution capacity in the field. In addition to the strength and settlement impacts, particle angularity has implications on the longevity of the EVA suits, low cycle fatigue performance of mechanical elements, and durability of the buried inflatable structures. The strength tests also showed high strain-rate dependency of the material properties at various compaction levels. This phenomenon is of paramount importance for the modeling and response calculation of the regolith platform subjected to dynamic loading conditions such as rover operations on planetary surfaces. The dilation angle of the Mars 1 regolith simulants at various strain rates and relative compaction levels were also determined in this study. High dilation angle values, compared to traditional terrestrial construction materials, implies significant nonlinearity in responses upon loading and unloading scenarios, such as rover operations. Therefore, nonlinearity of the responses upon loading should be an integral component of the analysis and design of foundation for planetary construction. The geotechnical parameters and indices obtained from this study can be instrumental for future numerical simulations and sensitivity analysis of the mechanical responses relevant to the strength and settlement parameters in various compaction scenarios outlined in this research. REFERENCES Ashtiani, R. S., “Class Discriminatory Information of Unbound Granular Layers Using Statistical Pattern Recognition Techniques”, In Advances in Transportation Geotechnics IV, Proceedings of the 4th International Conference on Transportation Geotechnics, Chicago, Illinois, Vol. 1, pp. 207-227. Springer, 2021 Ashtiani, R. S., and Asadi, M., “Stability Analysis of Anisotropic Granular Base Layers in Flexible Pavements”, Elsevier Journal of Transportation Geotechnics, Vol. 14, pp. 183189, 2018. Allen, K. P., Johnson, M. L., and May, J. S. (1998) “High Fidelity Vibratory Seismic (HFVS) method for acquiring seismic data” SEG Technical Program Expanded Abstracts, 10, 15-72. doi: 10.1190/1.1820171 Arcement, B. J., and Wright, S. G. (2001). “Evaluation of Laboratory Compaction Procedures for Specification of Densities for Compacting Fine Sands”. Austin. Arabali, P., Lee, S. I., Sebesta, S., Sakhaeifar, M. S., and Lytton, R. L. (2018). “Application of Superpave Gyratory Compactor for Laboratory Compaction of Unbound Granular Materials”. International Conference on Transportation and Development 2018. doi:10.1061/9780784481554.037 ASTM C136, (2014) “Standard Test Method for Sieve Analysis of Fine and Coarse Aggregates”, ASTM International, ASTM Int. ASTM C117, (2013) “Standard Test Method for Materials Finer than 75-μm (No. 200) Sieve in Mineral Aggregates by Washing”, ASTM International, ASTM Int. ASTM D422, (1998) “Standard Test Method for Particle-Size Analysis of Soils”, ASTM International, ASTM Int. ASTM D4318, (2017) “Standard Test Methods for Liquid Limit, Plastic Limit, and Plasticity Index of Soils”, ASTM International, ASTM Int. ASTM D2487, (2017) “Standard Practice for Classification of Soils for Engineering Purposes (Unified Soil Classification System)”, ASTM International, ASTM Int.

© ASCE

14

Earth and Space 2022

ASTM D4254, (2016) “Standard Test Methods for Minimum Index Density and Unit Weight of Soils and Calculation of Relative Density”, ASTM International, ASTM Int. ASTM D1557, (2007) “Standard Test Methods for Laboratory Compaction Characteristics of Soil Using Modified Effort”, ASTM International, ASTM Int. ASTM D4253, (2000) “Standard Test Methods for Maximum Index Density and Unit Weight of Soils Using a Vibratory Table”, ASTM International, ASTM Int. ASTM D7382, (2020) “Standard Test Methods for Determination of Maximum Dry Unit Weight of Granular Soils Using a Vibrating Hammer”, ASTM International, ASTM Int. ASTM D2435, (2004) “Standard Test Methods for One-Dimensional Consolidation Properties of Soils Using Incremental Loading”, ASTM International, ASTM Int. ASTM D3080, (2011) “Standard Test Method for Direct Shear Test of Soils under Consolidated Drained Conditions”, ASTM International, ASTM Int. Ashtiani, R., Little D. N., and Rashidi, M. (2018). “Neural network based model for estimation of the level of anisotropy of unbound aggregate systems”, Transportation Geotechnics 15, 4-12. https://doi.org/10.1016/j.trgeo.2018.02.002 Chen, Y.-R., and Kutter, B. L. (2009). “Contraction, Dilation, and Failure of Sand in Triaxial, Torsional, and Rotational Shear Tests”, Journal of Engineering Mechanics. American Society of Civil Engineers. https://ascelibrary.org/doi/abs/10.1061/ (ASCE)0733-9399(2009)135:10(1155). Coles, K. S., Tanaka, K. L., and Christensen, P. R. (2019). “The Atlas of Mars”. Carr, M. H., and Head, J. W. (2009). “Geologic history of Mars”. https://www.sciencedirect.com/science/article/pii/S0012821X09003847?casa_token=i5p David, L. (2014). “How Wheel Damage Affects Mars Rover Curiosity’s Mission.” Space Insider, 1.6 km/s) do complex gravitational potentials and non-gravitational effects begin to perturb trajectories. This simplification enables the investigation of all possible trajectories for ejecta from plume-surface interactions, and affects whether those ejecta pose a risk to surface and orbital systems. From this it can be determined which ejecta trajectories can be stopped by the natural lunar topography at a chosen landing site location, and whether artificial structures, such as berms, could do the same. Here we build on the general solutions to the maximum altitude, impact position, and times of flight for particles originating from near the lunar south pole at any velocity and angle due to plume-surface interaction. A point-mass gravitational potential is used and coupled to the lunar digital elevation models (DEM) from the Lunar Reconnaissance Orbiter (LRO) and Lunar Orbiter Laser Altimeter (LOLA). We provide an overview of the trajectory calculations, how the trajectories are mapped to the lunar elevation model, and some examples of ejecta rings for given initial conditions around potential landing site locations at the lunar south pole region. APPROACH Earthbound particle trajectories have long been studied. They are at the beginning of all basic physics courses and have obvious sport and military applications. However, while there are complexities in accurately predicting trajectories due to Earth's atmosphere and rotation (drag, buoyancy, Coriolis), and Earth's curvature for significantly long ranges, lunar trajectories are not necessarily so involved due to the lack of atmosphere and Moon’s slow rotation about its axis. The initial conditions for a regolith particle accelerated by a plume-surface interaction are taken to be simply a velocity (v) and angle (ϕ) above the lunar horizon and on a free trajectory. Azimuthal symmetry is also assumed for the ejecta, i.e., no longitudinal direction is preferred. Particles originate at a determinable position on an elliptical lunar orbit with a periapsis less than or equal to the lunar radius. The initial distance from the closest ellipse focus is the lunar radius itself. Given these conditions, the semi-major axis (a) of the orbit is given by a simple

© ASCE

18

19

Earth and Space 2022

rearrangement of the vis-viva equation (Equation 1), where RM is the lunar radius and μ is the gravitational parameter given by GMm: 2

𝑎 = (𝑅 −

−1

𝑣2

𝑀

) 𝜇

(1)

The eccentricity (e) of the orbit is determined by considering the total energy and angular momentum (L) of the orbit for a particle of mass m such that: 𝐿2 = 𝜇𝑚2 𝑎(1 − 𝑒 2 )

(2)

The angular momentum can then be updated to include the initial launch angle (ϕ). As ϕ is the initial angle above the lunar horizon, the component of ϕ perpendicular to the gravity vector is given by the cosine of ϕ as in Equation 3: 𝐿 = 𝑚𝑟𝑣 𝑐𝑜𝑠𝜙

(3)

The eccentricity is therefore determined from a simple rearrangement to give: 1

𝑒 = (1 −

(𝑅𝑀 𝑣𝑐𝑜𝑠𝜙) 2 ) 𝑎𝜇

(4)

With the two major ellipse shape parameters (a, e) the maximum altitude achievable by a particle is simply the difference between the apoapsis of the orbit and the lunar radius: 𝑟𝑚𝑎𝑥 = 𝑎(1 + 𝑒) − 𝑅𝑀

(5)

In addition, the true anomaly at any point in the orbit (θ) can be determined from these same shape parameters, including those points on the ellipse where the distance from the focus is equal to the lunar radius. Thus, the two solutions for Equation 6 can be used to determine the angular separation between the initial ejecta position and its final impact point. 𝑎−𝑅𝑀

𝜃 = 𝑐𝑜𝑠 −1 (

𝑎𝑒

)

(6)

While the range and impact latitude are trivial to determine from Equation 6, the time-of-flight requires a little further action. In particular, the mean anomaly (α) is needed to determine the time since periapsis (τ) given by Equations 7 and 8: 𝛼 = 𝜃 − 𝑒 𝑠𝑖𝑛𝜃

(7)

1

𝑎3 2

𝜏 = 𝛼(𝜇)

(8)

The time-of-flight is then the difference between the total orbital period and the two solutions for Equation 8 at τ(RM).

© ASCE

Earth and Space 2022

To bring each particle to a common position for their initial conditions, the arguments of periapsis for each trajectory are rotated so that the launch point at τ(RM) occurs at the lunar latitude of the chosen landing site location. In addition, the impact latitudes are adjusted for the quadrant of impact to keep solutions between -90 and 90 degrees and in the correct hemisphere. Two example trajectories for a landing site at the lunar south pole are presented in Figure 1. In terms of risk to surface and orbital systems, the major parameters of interest Figure 1: Two example trajectories for particles ejected by (when and where a particle will impact) can now be a plume-surface interaction near at the lunar south pole. determined for all possible The long trajectory is for an initial angle of 30o and a ejecta angles and velocities. velocity of 1.6 km/s. The short trajectory is for 15o and 1.5 However, there are some km/s. physical and practical limits to be considered in both cases. First, velocities below 1 m/s and angles below 0.1 degrees are ignored, as the resultant ranges, maximum altitudes, and times-of-flight are inconsequential; the particles travel only a few centimeters. Second, the escape velocity of the moon (2.38 km/s) provides an upper limit to any initial velocities as particles launched faster than this will not return to the lunar surface. Finally, asid e from the fact that angles of 90 degrees result in zero angular momentum (Equation l), such initial angles are not considered as these particles would be moving directly against the engine plume flow and would simply hit the underside of the descent vehicle in the absence of a plume. This approach to trajectory determination is inversible. For example, it is possible to input a chosen impact point, rather than a landing site location, and determine the initial angles and velocities ejecta at the plume impingement point would need to reach that point. For example, the initial conditions needed for ejecta to reach the Apollo 11 landing site are velocities and angles of greater than 1.58 km/s and 22.7 degrees. Similarly, the velocities and angles of ejecta that could reach chosen altitudes above the lunar surface can also be determined. For example, for ejecta angles below 15 degrees and velocities of greater than 1 km/s are needed for particles to reach the periapsis of LRO’s orbit, and velocities of greater than 2 km/s are needed for particles to reach the periapsis of the Lunar Gateway.

© ASCE

20

Earth and Space 2022

Mapping of the ejecta trajectories to the topographical variations in the lunar surface was accomplished using version 4 of the LRO LOLA DEM for the lunar south pole (87.5 deg, 5-m resolution available from trek.nasa.gov/moon). First, the chosen landing site latitude and longitude was used as the origin of the DEM profile extraction position. QGIS (https://www.qgis.org/) was then used to extract terrain profiles stepping by one degree of azimuth each time. Second, the altitude of the landing site was used to adjust the initial elevation of the ejecta. Third, the initial velocity and angle were chosen, but limited to trajectories in which the particles would not leave the area Figure 2: Matching a calculated trajectory to covered by the DEM (up to 75 km). Finally, the terrain profile extracted from the LOLA the trajectories were sampled every 5-m and DEM. The ejecta is mapped along a 0-degree matched to the terrain profiles. The azimuth from a selected landing site at trajectories were stopped when the height 85.42S, 31.74E. above the lunar surface elevation reached zero, and the appropriate ranges and times of flight were recalculated. An example trajectory, with the corresponding terrain profile, is shown in Figure 2. RESULTS Figure 3 presents the resultant impact ranges and latitudes for all values of v and ϕ when only considering a spherical lunar surface and point-mass gravitational potential. The range continues to increase toward the lunar circumference (10,921 km) as v increases and ϕ tends to zero. The sensitivity of the range to ϕ when v is close to 1.68 km/s is demonstrated by the rapid change in range (color) toward the lower right corner of Figure 3.

Figure 3: Impact range (left) and latitude (right) of particles launched from the lunar south pole at any velocity and angle. Strong inflexions are seen at 1.68 m/s as this is the orbital velocity of a particle around a point mass at the distance of the lunar radius.

© ASCE

21

Earth and Space 2022

The curves in Figure 3 show that below 1.68 km/s there are two values of ϕ that result in the same range. Above 1.68 km/s particles are on trajectories that takes them farther than the north pole and into the opposite hemisphere. This can be more clearly seen in the latitude data in Figure 3 (right panel). Impacts at the north pole are the deep stripe at velocities greater than 1.68 km/s. At velocities greater than 1.68 km/s and for very low angles, particles present a risk of almost making a complete orbit of the moon and impacting within a few hundred meters of the initial engine plume impingement point, i.e., where the descent module landed. Indeed, due to the symmetry in a single point mass gravitational potential, the only solutions for particles to hit precisely at the launch site are those with zero initial angle. Such particles will then return to the launch site at the same launch velocity and at the same launch angle, just simply from t he opposite azimuth. A deviation of only a few thousandths of a percent in the initial angle results in an impact a few hundred meters behind the launch site, and likely outside the footprint of any nearby surface systems. A difference of a few meters per second in initial velocity results in impacts separated by a few tens of meters. For azimuthally isotropic ejecta this result is independent of the Coriolis effect (discussed further below). This result is also independent of the moon’s orbit as the particle dynamics are based on the lunar gravitational potential. However, minor topographical features near the plume impingement point would prevent such low angle particles leaving the landing site altogether. Consequently, the local terrain will influence the likelihood of ejecta returning to the landing site and should be taken into consideration. Figure 4 presents example ejecta rings around a site on the “Connecting Ridge” between Shackleton and de Gerlache craters. Each point represents the impact locations for particles stepping through each degree of azimuth. The range of the particles depends on the topography of the lunar surface. For example, particles that have initial angles high enough to clear a crater rim travel farther than particles that do not. This creates the terrain following effect for each ejecta ring. The azimuths for which there is no data are a result of the slope angle at the initial site along that azimuth. In the case of Figure 4, the slope angle at zero degrees azimuth is 16 degrees, hence the remaining azimuthal gap in the ejecta ring for a 12-degree initial angle.

Figure 4: Example ejecta rings from a site on the “Connecting Ridge” near the lunar south pole at -89.45 S, 222.69 E. [left] Impact locations for ejecta with 400 m/s at 3 degrees. [right] Impact locations for ejecta with 400 m/s at 12 degrees. The gaps in the rings are due to the slope at the initial location being greater than the initial angle.

© ASCE

22

Earth and Space 2022

In the two cases presented in Figure 4 the range of impact distances because of the lunar terrain vary from 18.7 km to 37.9 km for the 3-degree trajectories, and from 44.2 km to 61.6 km for the 12-degree trajectories. And that is only for particles that have enough initial angle to clear the local slope. Consequently, not only can the lunar topography prevent particles leaving the landing site, it can also have a significant influence on the impact locations with ranges potentially varying by over 100 percent! DISCUSSIONS The direct and relatively straightforward approach used here allows for rapid exploration of the velocity and initial angle parameter space populated by plume-accelerated lunar particulates. The resultant trajectories are very sensitive to these initial conditions, especially for high velocities. However, there are influences from other physical processes that could lower the precision of these trajectories. For trajectories that gain significant altitude, the gravitational effects of Earth will eventually require modification to the singular point-mass gravitational potential assumption. Due to the atypical mass ratio of the Earth-moon system, the usual sphere of influence radius for the moon is larger than the distance of the moon from the first and second Earth-moon Lagrange points (EM1 and EM2). Any velocities exceeding 2.27 km/s would reach maximum altitudes greater than 30 percent EM1, but these velocities are only five percent of the cases if all velocities are populated by the plume-surface interaction. Therefore, for most trajectories considered here the singular gravitational potential approximation produces results with an acceptable precision. The assumption of a singular point-mass gravitational potential fails to account for the anisotropic distribution of mass within the moon. However, lunar gravitational anomalies produce an average acceleration variation of 0.0015 m/s2 (less than 0.1 %) over scales between 1.5 km and 78 km (Hirt & Featherstone 2012). Low velocity particles have trajectories shorter than these lengths resulting in minimal deflections, and high velocity particles have trajectories larger than such variations that smooth out deflections during the time of flight (at approximately 485 m/s, particles at 15 degrees launch angle would have a range of 78 km). However, a 0.1 percent gravitational variation over this scale of trajectory would result in a worst-case difference in impact location of less than 100 meters, i.e., less than the recommended artificial boundaries around sites of scientific or historical importance. Non-gravitational perturbations to trajectories could be a result of solar radiation pressure (SRP) and the solar wind. LADEE/LDEX observations have also demonstrated particles in the lunar exosphere to be charge carriers (Horanyi et al. 2015a), therefore the interplanetary magnetic field and Earth's magnetotail could produce perturbations. In addition, the lunar surface at the poles is expected to have a complex electromagnetic environment due to the low altitude of the Sun producing a constant near terminator environment and long shadows from the local topography (significant electric potential gradients are expected across lit and shadows regions). Compared to the lunar gravity, and on the assumption of a 0.7 μm radius spherical perfect absorber experiencing 1,400 W/m2, SRP would impart a force 3-orders of magnitude less. In addition, on the assumption of the 8 nT magnetic field strength reported both in and out of Earth's magnetotail (Xu et al. 2019, Gencturk Akay et al. 2019), charged exospheric lunar regolith would experience an electromagnetic force six orders of magnitude less than the lunar gravity. However, the magnitude of these forces increase as the particle size decreases and altitude and time of flight increases (the time of flight for a particle launched at 2.27 km/s is 25 hours for any

© ASCE

23

Earth and Space 2022

launch angle). At low altitudes, with short times of flight, the complex electromagnetic environment of the lunar surface would be traversed quickly by charged plume ejecta and impart minimal perturbations to particle trajectories. In total, for the expected velocities and angles of ejecta determined from current observations and simulations, most particles accelerated by plume surface interactions will have low initial angles. If confirmed, the consequence for velocities below 1.6 km/s will be trajectories that are well approximated by the simple singular point-mass gravitational potential assumption, and where the resultant precisions are within the recommended artificial boundaries that enable risk assessment to surface sites of interest. Above 1.6 km/s the precision of this approach will drop as the velocities increase. This is not only due to additional gravitational and non-gravitational perturbations, but also due to the sensitivity of the resultant ranges and latitudes on the initial conditions; a small change in angle will result in a large change in the trajectory. In addition, the risk to orbiters begins to increase for these velocities. However, for launch velocities more than 3 km/s these would only be for low mass particles. Indeed, these velocities are still far below the relative velocities of interplanetary dust grains that can have average impact speeds more than 20 km/s (McNamara et al. 2005). However, in the long-term with multiple landings, should plume-surface interactions begin to significantly increase the density of particles in the lunar exosphere above the current background, then a plume-surface resultant dust build up on critical surfaces, i.e., solar panels, radiators, and windows, could begin to introduce some further risk, or loss of efficiency, particularly if those particles carry a charge. Compared to Earth, the lunar sidereal rotation rate is low (2.66 x 10-6 rad/s) leading to a small Coriolis parameter (5.32 x 10 -6 at the lunar poles). This results in a large Rossby number for particles with low velocities. For example, a particle velocity of 500 m/s at 15 degrees results in a longitudinal displacement of only 39 m. However, at 1 km/s the displacement is 580 m, and at 1.5 km/s the displacement is over 8 km. Therefore, while the Coriolis effect can be neglected for low velocities, it must be included in future developments of impact ejecta location predictions. Under the assumption of an azimuthally isotropic outflow from the plume impingement point, and ignoring the lunar topography, the Coriolis effect would have little relevance as all longitudes would still experience ejecta fallout at the same ranges for the given velocities. However, as has been seen, the lunar topography can have a profound effect on the range of a particle. This means that particles with high velocities will be impacting locations at significantly different ranges and longitudes. A given location on the lunar surface could therefore experience a significantly different fallout density, particularly if the lander has multiple engines that result in anisotropic ejecta outflows. If plume ejecta are indeed launched at angles of less than 3 degrees, the majority of trajectories could be mitigated through purposefully designed and constructed structures, or by intrinsic local topography around the plume impingement point. While simulations have shown that a 2 m high berm with a radius of 30 m will stop particles with launch angles of up to 4 degrees (Lane et al. 2010), the local topography at a potential south pole landing site may not prove so effective. Landing sites that maximize the amount of sunlight throughout a lunation, in order to aid in surface operations, will naturally have few surface features that would block the line-of-sight to the Sun, and therefore have a minimal number of natural barriers towards the horizon. Regardless, it is clear from these simple calculations that tightly constraining the launch angle of ejecta from plume surface interactions is critical for both detailed risk assessments and forward planning of long-term surface operations.

© ASCE

24

Earth and Space 2022

CONCLUSIONS A simple point-mass gravitational potential approximation provides the precision needed to quickly explore parameter space and assess the risks to lunar surface and orbital assets from the resultant ejecta of plume surface interactions, particularly when the launch velocities are below 1.6 km/s. Velocities greater than this may achieve altitudes and times of flight that result in a drop in precision due to perturbations from other gravitational and non-gravitational physical processes. What is clear from these results is the importance of constraining further the angles of particles accelerated by a plume-surface interaction and including the lunar topography. The resultant ranges, impact latitudes, times of flight, and maximum altitudes are very sensitive to these parameters especially for the highest velocities. However, should further investigations demonstrate that velocities rarely exceed 1.6 km/s, and are indeed mainly at angles of around 3 degrees, then the risks to both current surface sites, and current or future orbital assets, will be low. Such risks could be mitigated further through careful choice of landing site locations that use lunar topography as a natural ejecta mitigation method. REFERENCES Calle, C. I. 2017, Electrostatic Phenomena on Planetary Surfaces (Morgan and Claypool Publishers), doi: 10.1088/978-1-6817-4477-3. Fenner, M. A., Freeman, J. W., Jr., & Hills, H. K. 1973, in Lunar and Planetary Science Conference, Vol. 4, Lunar and Planetary Science Conference, 234. Garrick-Bethell, I., Head, J. W., & Pieters, C. M. 2011, Icarus, 212, 480, doi: 10.1016/j.icarus.2010.11.036 Gencturk Akay, I., Kaymaz, Z., & Sibeck, D. G. 2019, Journal of Atmospheric and Solar-Terrestrial Physics, 182, 45, doi: 10.1016/j.jastp.2018.11.002 Glenar, D. A., Stubbs, T. J., Hahn, J. M., & Wang, Y. 2014, Journal of Geophysical Research (Planets), 119, 2548, doi: 10.1002/2014JE004702 Glenar, D. A., Stubbs, T. J., McCoy, J. E., & Vondrak, R. R. 2011, Planet. Space Sci., 59, 1695, doi: 10.1016/j.pss.2010.12.003 Halekas, J. S., Mitchell, D. L., Lin, R. P., et al. 2002, Geophys. Res. Lett., 29, 1435, doi: 10.1029/2001GL014428 Hirt, C., & Featherstone, W. E. 2012, Earth and Planetary Science Letters, 329, 22, doi: 10.1016/j.epsl.2012.02.012 Horányi, M., Szalay, J. R., Kempf, S., et al. 2015, Nature, 522, 324, doi: 10.1038/nature14479 Horányi, M., Sternovsky, Z., Lankton, M., et al. 2014, SSRv, 185, 93, doi: 10.1007/s11214014-0118-7 Iglseder, H., Uesugi, K., & Svedhem, H. 1996, Advances in Space Research, 17, 177, doi: 10.1016/0273-1177(95)00777-C Immer, C., Lane, J., Metzger, P., & Clements, S. 2008, in Earth & Space 2008 (American Society of Civil Engineers). Immer, C., Lane, J., Metzger, P., & Clements, S. 2011a, Icarus, 214, 46, doi: 10.1016/j.icarus.2011.04.018 Immer, C., Metzger, P., Hintze, P. E., Nick, A., & Horan, R. 2011b, Icarus, 211, 1089, doi: 10.1016/j.icarus.2010.11.013

© ASCE

25

Earth and Space 2022

Kozai, Y. 1963, PASJ, 15, 301 Lane, J. E., & Metzger, P. T. 2015, Acta Geophysica, 63, 568, doi: 10.1515/acgeo-2015-0005 Lane, J. E., Metzger, P. T., & Carlson, J. W. 2010, in Earth and Space 2010 (American Society of Civil Engineers). Liever, P., Tosh, A., Arslanbekov, R., & Habchi, S. 2012, in 50th AIAA Aerospace Sciences Meeting including the New Horizons Forum and Aerospace Exposition McCoy, J. E., & Criswell, D. R. 1974, Lunar and Planetary Science Conference Proceedings, 3, 2991 McKay, D., Heiken, G., Basu, A., et al. 1991, in Lunar Sourcebook, ed. G. Heiken, D. Vaniman, & B. French (Cambridge Univ. Press), 285-356 McNamara, H., Jones, J., Kauman, B., et al. 2005, Earth, Moon, and Planets, 95, 123, doi: 10.1007/s11038-005-9044-8 Metzger, P. T., Smith, J., & Lane, J. E. 2011, Journal of Geophysical Research, 116, doi: 10.1029/2010je003745 Rennilson, J. J., & Criswell, D. R. 1974, Moon, 10, 121, doi: 10.1007/BF00655715 Stubbs, T. J., Halekas, J. S., Farrell, W. M., & Vondrak, R. R. 2007a, in ESA Special Publication, Vol. 643, Dust in Planetary Systems, ed. H. Krueger & A. Graps, 181, 184. Stubbs, T. J., Vondrak, R. R., & Farrell, W. M. 2006, Advances in Space Research, 37, 59, doi: 10.1016/j.asr.2005.04.048 Stubbs, T. J., Vondrak, R. R., & Farrell, W. M. 2007b, in ESA Special Publication, Vol. 643, Dust in Planetary Systems, ed. H. Krueger & A. Graps, 239, 243. Wittal, M. M., Phillips, J. R. I., Metzger, P. T., et al. 2020, in AAS/AIAA Astrodynamics Specialist Conference, Vol. 20, 511 Xu, X., Xu, Q., Chang, Q., et al. 2019, The Astrophysical Journal, 881, 76, doi: 10.3847/1538-4357/ab2e0a

© ASCE

26

Earth and Space 2022

Geometrical Characteristics of Lunar and Martian Regolith Simulants Caleb J. Carnes1; Reza S. Ashtiani2; Joel A. Sloan3; Melissa S. Beauregard4; and Kimberly D. De la Harpe5 1

Dept. of Civil and Environmental Engineering, United States Air Force Academy, CO. Email: [email protected] 2 Dept. of Civil and Environmental Engineering, United States Air Force Academy, CO. Email: [email protected] 3 Dept. of Civil and Environmental Engineering, United States Air Force Academy, CO. Email: [email protected] 4 Dept. of Civil and Environmental Engineering, United States Air Force Academy, CO. Email: [email protected] 5 Dept. of Physics and Meteorology, United States Air Force Academy, CO. Email: [email protected] ABSTRACT In the next decade, NASA is prioritizing moon missions such as safe human transportation, deployment of instrumentations, establishment of habitats, and exploration of native resources to set the stage for future Mars missions. One of the main challenges to achieve these goals, however, is our lack of understanding of the synergistic interactions between the native soil, lunar vehicles, habitats, and EVA suits. In addition to the particle size distributions, surface properties, and compaction characteristics of the surface regolith, the geometry of the particles is of paramount importance for a proper understanding of the settlement properties, distortion characteristics, and orthogonal strength of the particulate medium, which impacts transportation, habitat development, and the regolith’s potential to degrade EVA suits by abrasion based on the angularity of the particles. Therefore, our multi-disciplinary team in this research envisioned using a state of the art optical profilometer equipment supplemented by image analysis techniques to characterize the geometrical features of a lunar regolith simulant, specifically the CSM-LHT-1 highlands type. The distributions of the particle form, angularity, and surface macro-texture of lunar regolith were in turn contrasted with JSC-1 Mars simulants and a calcareous construction aggregate for comparative purposes in this study. The results will be instrumental to better understand the anisotropic nature of the angular particles and its relevance to the strength and deformation characteristics of native soils on extra-terrestrial planetary surfaces. INTRODUCTION In order to accomplish the establishment of a lasting presence on the lunar surface as well as an eventual presence on the Martian surface, NASA will need to gain a better understanding of the characteristics of the regolith found on each respective surface. Gene Cernan (1973) of the Apollo 17 crew stated in the Apollo 17 debriefing, “I think dust is probably one of our greatest inhibitors to a nominal operation on the Moon. I think we can overcome other physiological or physical or mechanical problems except dust.” It is apparent that the overarching difficulty in the bed down stage of the Artemis program will come from the lunar and Martian regolith.

© ASCE

27

Earth and Space 2022

Agui (2021) writes about the significance of the, “…size and shape, toxicity and reactivity, abrasiveness, and adhesion properties.” This research was done with the particulate environment aboard extended exploration life support systems in mind. As this study established, the elongation of the lunar regolith plays an important role in the long term survivability of life support systems on the lunar surface. This paper presents the particle geometry data for both the CSM-LHT-1 highlands lunar regolith as well as the JSC-1 Martian regolith in order to identify and explain any correlation between the two while highlighting the importance that the particle geometry plays on load bearing capacity, deformation potential, dust lift off, and longevity of EVA suits. BACKGROUND As NASA prioritizes the design of suitable habitats for humans to live in low gravitational environments like the moon and Mars, the importance of the particulate geometry and the anisotropic behavior of the particles cannot be overlooked or oversimplified. Modern design protocols for multi-layer infrastructure facilities such as pavements and airfield runways overlooks the anisotropic behavior through simplifying assumptions that can cause premature structural failure. Further research into the anisotropic behavior of lunar and Martian regolith simulants is necessary for the initial success and cost reduction for future missions and habitant formations on moon and mars. (Ashtiani, 2018). The importance of the anisotropic behavior of the particles in terms of bearing capacity and deformation potential stems from the inevitable vibrations from moving loads such as the lunar rover and other equipment that will be needed in operational environments. Limited research has been conducted on the sintering of lunar and Martian regolith simulants, therefore, similar to terrestrial construction, the primary strength from the soils is a product of the frictional forces between the particles. The random geometry of the regolith results in deviations from the orthogonal particle contact distributions which in turn manifest itself in directional dependency of the material properties in cross-anisotropic particulate medium. This shape-induced anisotropic behavior juxtaposed by the load-induced anisotropy adds an additional layer of complexity in mechanistic characterization of regolith media. In addition to the complex nature of the stress paths and materials behavior, the realistic simulation of the compaction characteristics of the regolith in low gravitational environments greatly impact the mechanical responses under an external stimuli such as habitat static loads or dynamic rover wheel loads. Therefore research must be conducted into the anisotropy of the regolith simulants in order to better understand the effect on these factors as well as the longevity of EVA suits. EXPERIMENTAL DESIGN Regolith morphological properties, namely particle form and angularity, were examined using the Aggregate Imagining System (AMIS) which allows the examination of three components of the geometrical characteristics of particles. However, the surface macro texture was not examined during this study as there’s evidence of lesser impact of surface asperity features of fines on the interlocking effect and inter-particle frictional properties of particulate assemblies. Ashtiani et. al. (2018) describes the physical properties examined as follows, “Aggregate form pertains to the flat and elongation nature of the…aggregate particles. Angularity refers to the broken edges and the roundness of aggregates...” The

© ASCE

28

Earth and Space 2022

AIMS system then produces data which can be analyzed using the Krumbein scale for angularity and also allows for data reduction of form on a scale from 0 to 20. The AIMS system was used to compare angularity and form data between the lunar and Martian regolith simulants. In addition to the AIMS device, an optical profilometer device was used to provide comparative analysis of the particle morphology in this study. The Zygo NewView 8300 is a 3D optical surface profilometer that can be used to provide detailed 3D images of small particles ranging from heights of 1 km) subsurface access under flight -like constraints. In addition, proposals were expected to identify, address, and reduce technical risks for the most promising ice penetration systems so that these systems may eventually be infused into potential future flight opportunities. More specifically, the SESAME technology development solicitation sought to: a) Identify promising cryogenic ice penetration systems capable of facilitating the detection of life, especially extant life, in the ocean worlds of the outer solar system by providing access to subsurface liquid water bodies that may be located hundreds of meters to tens of kilometers below the surface of the ice. b) Identify the technology component(s) that represent the greatest technical risk to the overall penetration system. c) Begin to reduce the key technology risks through an analytical and experimental technology development effort. d) Develop prototype hardware for cryogenic ice penetration system(s); and e) Assess the performance of the prototype hardware through analysis and complementary laboratory experiments. The five activities shown in Table 6 were selected to mature technologies aligned with the program goals. Table 6. SESAME Tasks Principal Investigator / Institution

Task

Kris Zacny/ Honeybee Robotics

SLUSH: Search for Life Using Submersible Heated Drill

William Stone/ Stone Aerospace

PROMETHEUS: nuclear-Powered RObotic MEchanism Technology for Hot-water Exploration of Under-ice Space

Kathleen Craft/ Johns Hopkins University

Europa STI - Exploring Communication Techniques and Strategies for Sending Signals Through the Ice (STI) for an Ice-Ocean Probe

Tom Cwik/ NASA Jet Propulsion Laboratory

Cryobot For Ocean Worlds Exploration

Britney Schmidt/ Georgia Institute of Technology

Vertical Entry Robot for Navigating Europa (VERNE)

SLUSH is a hybrid thermomechanical drill probe system that combines the most efficient aspects of thermal and mechanical penetration techniques to reach Europa’s subsurface ocean. PROMETHEUS advanced a cryobot design that uses closed-cycle hot water drilling as the

© ASCE

Earth and Space 2022

primary means of penetrating the ice to actively control the descent through the ice shell and into Europa’s subsurface ocean. The Europa STI effort tested multiple communication tether designs. The Cryobot task matured a system architecture and associated technologies necessary for accessing Europa’s subsurface ocean, including fluidics and thermal modeling, and testing in cryogenic ice chambers and the Arctic. VERNE is a platform that includes a surface station, self-drilling vehicle with a breakable mechanical/data tether that would deploy acoustic communication stations in the ice during descent, and a base station that would remain in the ice to relay communications with the mobile platform to the surface station. DIRECTED WORK In addition to the competitively awarded efforts described in this paper, NASA also directed funding to advance development of specific technologies and entire lander subsystems. These consisted of three efforts: the Europa Lander pre-project, radiation-hard electronics, and autonomy testbeds. The Europa Lander pre-project led by the NASA Jet Propulsion Laboratory (JPL) defined an end-to-end mission concept, identified the necessary technology developments needed to implement it, and began developing some of the critical and/or long-lead technologies. Specific technologies included a high efficiency antenna for communication with Earth, an end-ofmission terminal sterilization system for planetary protection, batteries tolerant to low temperature and high radiation conditions. Extensive work was conducted to address the challenge of autonomously landing on a poorly characterized surface with unknown hazards (i.e. automated terrain analysis and target selection). The De-Orbit, Descent, and Landing (DDL) system was evolved beyond the system used on the Mars Perseverance mission to include greater autonomy, active hazard detection using an onboard LIDAR, and a dynamic landing gear system to accommodate uneven terrain and boulders. In addition, a variety of surface autonomous sampling techniques and tools were investigated. DDL and surface sampling tool development required multiple testbeds of varying fidelity to test and validate the technologies and systems. The radiation environment in the Jupiter system is the harshest in the Solar system and heavy shielding is required to protect existing flight electronics for long-duration operations there. The development of radiation-hard electronics was funded at the NASA Ames and Glenn Research Centers to reduce – or eliminate – the need for such shielding. Nanoscale vacuum channel transistors that had been developed at Ames for lunar applications, and silicon carbide (SiC) transistors developed at Glenn for long-duration operation on the high temperature Venus surface are also intrinsically tolerant to very high radiation levels. Under PSD funding, both technologies demonstrated excellent radiation tolerance. Nano-vacuum transistors fabricated in both silicon and SiC demonstrated high reliability against gamma and neutron radiation, and 4HSiC junction field effect transistor integrated circuits demonstrated over 7 Mrad total ionizing dose tolerance, with no destructive single-event effect susceptibility. Autonomous operations will be critical on future surface and subsurface missions to ocean worlds because the communication lag to the outer planets is long, the mission life of surface assets is expected to be limited by radiation damage, and traversing vertically through ice will not be conducive to Earth-based control. Because autonomy is a system-level technology and therefore is difficult to develop in isolation, NASA developed two testbeds, one virtual and one physical, to support the development and testing of new autonomy technologies for future lander

© ASCE

316

Earth and Space 2022

missions to ocean worlds. The testbeds provide simulation in hardware and software of a "generic" planetary lander equipped with a seven degree-of-freedom robotic manipulator and a variety of sensors. The Ocean Worlds Autonomy Testbed for Exploration Research and Simulation (OceanWATERS), is a software-based simulator at NASA Ames that emulates surface environmental conditions (e.g., lighting and surface material properties), robotic manipulator operation, and high-level lander systems. The simulator supports injection of faults and provides system introspection capabilities. The simulator is modeled on the Europa Lander pre-project, but can be configured for other lander missions and planetary bodies. The Ocean Worlds Lander Autonomy Testbed (OWLAT), is a hardware-based facility at NASA JPL that emulates planetary bodies with different gravities. The testbed includes a lander deck (with robotic manipulator) mounted on a six degree-of-freedom Stewart platform. The manipulator is equipped with a 6-axis force-torque sensor at its tool interface. Geotechnical instruments and sampling tools are provided as end effectors, which can be attached to the tool interface. Both testbeds have a common software interface for acquiring sensor data and commanding a manipulator arm, the mast, and the sensor suite. Sensor data available from the testbeds include arm joint and end effector position, measured end-effector forces and torques, and stereo camera images. The virtual testbed allows more complete modeling of the surface environment (such as illumination, topography out to the horizon, and surface albedo) as well as of the lander’s external features (in particular a scale model of the complete lander) and its subsystems (such as a power system with onboard battery prognostics). The virtual testbed will also allow rapid and broader modification of environmental and spacecraft parameters. The physical testbed provides higher-fidelity terrain interaction with tools and instruments than the virtual testbed. The physical testbed will also enable more realistic sensing constraints and higher-fidelity dynamics of manipulation / sampling to be studied. AUTONOMOUS ROBOTICS RESEARCH FOR OCEAN WORLDS (ARROW) The ARROW solicitation was released in 2019 for the development of functional and systemlevel autonomous capabilities for the surface exploration of ocean worlds, such as Europa, Enceladus, and Titan. The goal of the program is to develop autonomy software technologies to significantly increase the robustness and productivity of future ocean worlds lander missions including those to destinations with surface conditions and phenomena that may be largely, or completely, unknown a priori at the time of landing. Two projects were funded under this solicitation as shown in Table 7. The objective of the first task, Raspberry SI, is three-fold: to develop machine learning models to discover mission intent and enable transfer learning and online learning during a mission; to develop a model compression technique to compress large neural networks into an accurate and efficient tiny model for deployment; and to develop a probabilistic planner to enable run-time adaptation based on the state of the spacecraft, environment assumption, and resource availability. The objective of the second task, Robust Autonomy for Planetary Sampling, is to develop a novel extension of NASA’s open-source PLEXIL (Plan Execution Interchange Language) execution technology with stochastic decision-making capability to enable dynamic selection of optimal lander procedures.

© ASCE

317

318

Earth and Space 2022

Table 7. AISR:ARROW Autonomy Tasks Principal Investigator / Institution

Task

Pooyan Jamshidi/ University of South Carolina

Resource Adaptive Software Purpose-Built for Extraordinary Robotic Research Yields Science Instruments (RASPBERRY SI)

Jonathan Bohren/ Honeybee Robotics

Robust Autonomy for Planetary Sampling

Both of these tasks will be tested and evaluated on NASA’s OCEANWaters virtual testbed at NASA Ames and are scheduled to conclude in 2022. In addition, Raspberry SI will also be evaluated on NASA’s OWLAT physical testbed at NASA JPL. CONCEPTS FOR OCEAN WORLDS LIFE DETECTION (COLDTech-20) After the successful completion of the SESAME program, specific technology gaps remained. These included technologies to enable communication through many kilometers of ice thickness; radiation-hard digital devices for surface operations and/or an orbiting relay; and autonomy as described in the ARROW section above. The goal of COLDTech-20 is to develop these technologies to acquire samples beneath the surface and process them, including providing access to subsurface oceans or other bodies of liquid water. Table 8. COLDTech Autonomy Tasks Principal Investigator / Institution

Task

Eric Dixon/ Lockheed Martin Melkior Ornik/ University of Illinois at Urbana-Champaign Joel Burdick/ California Institute of Technology Jay McMahon/ University of Colorado

Causal and Reinforcement Learning (CARL) Adaptive, Resilient Learning-Enabled Ocean World Autonomy (DRILLAWAY) Robust, Explainable Autonomy for Scientific Icy Moon Operations (REASIMO) Expert-Informed Autonomous Science Planning for In-situ Observations and Discoveries

Four tasks are funded for autonomy as listed in Table 8. The first three listed tasks are developing advanced task planners to automated surface operation in the presence of unexpected errors. The fourth task focuses on the interface between the scientists and the remote autonomous system to maximize science return in the face of unexpected results. Five tasks were selected to develop through-ice communications technology as shown in Table 9. Both the SLUSH-related and STI Tech tasks are developing tethers to robustly survive the dynamic ice shell on Europa. STI Tech, PARTI Pucks, and the hybrid RF/MI tasks are developing radio frequency transmission concepts, and finally, the hybrid RF/MI task is exploring the use of magneto-inductive transceivers to communicate through ice with unknown and varying conductive properties.

© ASCE

319

Earth and Space 2022

Table 9. COLDTech Communications Tasks Principal Investigator / Institution Kris Zacny/ Honeybee Robotics William Stone/ Stone Aerospace Kathleen Craft/ Johns Hopkins University Michael Cheng/ NASA Jet Propulsion Laboratory Yoseph Bar-Cohen/ NASA Jet Propulsion Laboratory

Task Tether based communication system for the Search for Life Using Submersible Heated drill (SLUSH) Europa probe Puck-Based Data Transmission Using Adaptive Radio Modems for Through-Ice Communications (PARTI Pucks) Technology for Sending Signals Through the Ice on Ocean Worlds (STI Tech) Hybrid Radio Frequency (RF) and MagnetoInductive (MI) Transceiver for Europa Sub-Ice Communications Maturation and demonstration of technology to enable through deep-ice communication (CryoComm) on Europa

Two tasks are being funded to develop radiation hardened electronics as shown in Table 10. The first task will develop and demonstrate silicon-germanium electronics to operate robustly under the combined high radiation and low temperatures of ocean worlds, developing device models for circuit design and a full suite of digital, analog, and RF circuit building blocks to create a large-scale integrated circuit prototype. The second task will develop a radiation-hardened-by-design (RHBD) analog-to-digital converter by exploiting a state-of-the-art fabrication technology combined with RHBD mitigation techniques, and applying them to single-chip, low power consumption device. Table 10. COLDTech Radiation Hard Electronics Tasks Principal Investigator / Institution John Cressler/ Georgia Institute of Technology Tim Holman/ Vanderbilt University

Task Environmentally-Invariant Silicon-Germanium Electronics for On-Surface Ocean Worlds Exploration A Radiation-Hardened Analog-to-Digital Converter for Outer Planetary Missions

Each of these COLDTech-20 tasks began in the summer of 2021; the autonomy tasks will last for two years and the others will continue for three years. ASTRODYNAMICS In addition to the technologies needed for in situ exploration of ocean worlds, it is also challenging to develop mission trajectories to efficiently access the outer planets. To address the shortfalls in our abilities, in 2018 PSD released the Astrodynamics in Support of Icy Worlds Mission solicitation. This program solicited the formulation, maturation, and validation of astrodynamics analysis tools to uncover new mission concepts, motivate entirely new classes of missions that may not have been previously considered, improve the efficiency of missions, and/or extend mission life. The tasks selected under this program are listed in Table 11. © ASCE

320

Earth and Space 2022

Table 11. Astrodynamics Tasks Principal Investigator / Institution

Task

Ryan Russell/ University of Texas at Austin Ossama Abdelkhalik/ Iowa State University Rohan Sood/ University of Alabama Daniel Scheers/ University of Colorado

Falcon: Automating the Search and Design Process for Icy World Trajectories Enhancements to The NASA Evolutionary Mission Trajectory Generator ASSET: Astrodynamics Software and Science Enabling Toolkit Quasi-periodic orbit computational tools for mission design

CONCLUSION The exploration of ocean worlds in the outer solar system could provide unprecedented insight into humanity’s quest for knowledge of the existence of life beyond Earth. Over the past decade NASA’s Planetary Science Division has funded technology development to enable future missions to explore these ocean worlds. These efforts include the development of seismometers, imagers, spectrometers, and organic analyzers, and platform technologies including advanced landing systems, drills, melt probes, through-ice communications, radiation-hard electronics, and autonomy for surface operations. These efforts are establishing a firm foundation to support future missions through advanced technology development that will serve to mitigate risks and enable the exploration of enticing ocean worlds. REFERENCES NASA (2020). “Science 2020-2024: A Vision for Scientific Excellence,” https://science.nasa.gov/science-red/s3fs-public/atoms/files/2020-2024_Science.pdf (accessed Dec. 28, 2021). Neveu, M., Hays, L.E., Voytek, M.A., New, M.H., and Schulte, M.D. (2018). “The Ladder of Life Detection,” Astrobiology, Vol 18, No. 11, pp. 1375-1402. See also https://astrobiology.nasa.gov/research/life-detection/ladder/ (accessed Dec. 28, 2021) Hendrix, A.R., Hurford, T.A., Barge, L.M., Bland, M.T., Bowman, J.S., Brinckerhoff, W., Buratti, B.J., Cable, M.L., Castillo-Rogez, J., Collins, G.C., Diniega, S., German, C.R., Hayes, A.G., Hoehler, T., Hosseini, S., Howett, C.J.A., McEwen, A.S., Neish, C.D., Neveu, M., Nordheim, T.A., Patterson, G.W., Patthoff, D.A., Phillips, C., Rhoden, A., Schmidt, B.E., Singer, K.N., Soderblom, J.M., and Vance, S.D. (2019). “The NASA Roadmap to Ocean Worlds,” Astrobiology, Vol 19, No. 1, pp. 1-27. NASA (2012). “Europa Study 2012 Report,” https://europa.nasa.gov/resources/63/europastudy-2012-report/ (accessed Dec. 28, 2021). NASA (2017). “Seven SMD-supported instruments to search for evidence of life on Europa.” Phys Org, https://phys.org/news/2017-03-smd-supported-instruments-evidence-lifeeuropa.html (accessed Oct. 30, 2021). NASA (2016). “Europa Lander Study 2016 Report,” https://europa.nasa.gov/resources/58/europalander-study-2016-report/ (accessed Dec. 28, 2021)

© ASCE

321

Earth and Space 2022

A Review of Extra-Terrestrial Regolith Excavation Concepts and Prototypes Robert P. Mueller, M.ASCE1 1

Swamp Works, Exploration and Research Technologies, Kennedy Space Center, National Aeronautics and Research Administration (NASA), FL. Email: [email protected] ABSTRACT Regolith is present on many extra-terrestrial bodies, and the crushed rock material it is made of contains many of the resources that are enabling for in situ resource utilization (ISRU). When extracted, these resources can be used to provide consumables such as rocket propellant, human life support, working fluids and gases for industrial processes, and feedstocks for manufacturing. In addition, the regolith can also be very beneficial for construction purposes as an aggregate which can be used for construction materials and shielding for radiation protection and micrometeorite impact. Binders for regolith concrete may also be made from geopolymers that may be in the regolith. The regolith can be melted and drawn out into glass fibers and used as reinforcements in a metal, polymer, or concrete matrix. In addition, there is tremendous scientific and geological knowledge that can only be obtained by studying samples of the regolith. However, none of these valuable activities can proceed without first acquiring the regolith granular material with some type of excavation device and method. Excavation is in the critical path of many workflows that will make up the capabilities required to establish a human and robotic presence in our solar system. While scientific in situ sampling of regolith in small quantities has been achieved since the dawn of the space age in the 1960s, large scale excavation for mining and construction on extra-terrestrial bodies has only been contemplated, for many decades, but serious development and prototyping of excavation technologies for use in reduced gravity space environments was only started in the late 1990s. This paper will review and document the evolution of extra-terrestrial excavation concepts and prototypes based on the available literature and the personal experience of the author who has been working on regolith excavation technology development since 1998. INTRODUCTION Current NASA policy aims to use space resources on the Moon to ensure a sustainable future. On December 11, 2017, United States (US) space policy evolved with the signing of Space Policy Directive 1 which provides for a US led integrated program with private sector partners for a human return to the Moon followed by missions to Mars and beyond. Notably, it directs NASA to pursue human expansion across the solar system (Hill, 2018). This goal will take many decades to achieve at the current rate of NASA human space missions but there is a way to accelerate such an effort, so that humanity and associated industry could expand to the asteroid belt and beyond in this century. Space resources and advanced technologies can be combined to create an affordable, rapid bootstrapping of space industry and solar system civilization (Metzger et al, 2013). The resources on the Moon are, to a large degree, contained in the photonic energy from the Sun, as well as minerals and volatiles in the lunar regolith. In order to acquire the regolith, robotic excavation technologies will be necessary, and these robotic excavators will be very different from terrestrial excavators. The reduced gravity and harsh environment on the

© ASCE

Earth and Space 2022

Moon create unique excavation equipment requirements and there are severe mass and volume limitations that are imposed by the space transportation launch vehicles. Launching from the Earth’s gravity well requires significant energy, so everything brought from Earth must be small and low mass, to avoid high costs and logistics burdens. The NASA Artemis program is focused on landing humans at the South Pole of the Moon by the middle of the 2020’s and the reason that the South Pole was chosen as the destination is that there is evidence of significant water ice deposits as well as other volatiles, while there are also regions of favorable solar illumination, which provides reduced thermal extremes and allows harvesting of solar energy on an almost continuous basis. These water and other volatiles including light hydrocarbons, sulfur-bearing species, and carbon dioxide, as well as mercury, magnesium, calcium, silver, and sodium were discovered by the NASA Lunar Crater Observation and Sensing Satellite (LCROSS) mission (Colaprete et al, 2010). While the geotechnical properties of the water and volatiles ices are unknown, there will be a need for acquiring these regolith resource feedstocks, and excavation is a candidate technology for doing so, which will be evaluated and compared to other competing technologies such as thermal mining. In thermal mining, ice is sublimated by applying heat directly to the surface and the near subsurface of the permanently shadowed region (PSR), the vapor is captured under a dome-like tent, then directed toward cold traps where it refreezes for transport to a processing system (Sowers et al, 2019). If water ice cannot be found, then oxygen can still be extracted from the silicates contained in the regolith. At the lunar south pole these silicates are found in Anorthosite rock, an igneous rock characterized by its composition: mostly plagioclase feldspar (90–100%), with a minimal mafic component (0–10%). The plagioclase feldspar; is calcium-sodium aluminosilicate [(CaAl,NaSi)AlSi2O8] which can be found in the highlands granular rock materials that were crushed by high energy impacts from meteorites, comets and other space objects in the past 4.5 billion years. Excavation is required to acquire and deliver this regolith to an In-Situ Resource Utilization (ISRU) processing plant for oxygen extraction. Chemical engineering processes such as carbothermal reduction can be used to process and extract the oxygen. REGOLITH ON THE MOON While most lunar exploration architectures are focused on ISRU for propellant production, due to the large mass savings that can be achieved by avoiding the launch of propellant for the journey from the Moon back to Earth, there are many other uses of regolith which can be customers for lunar excavation providers. A list of some of the possible uses for regolith and volatiles derived resources is provided: Science investigations Geology investigations Propellant Oxidizer (O2) Extraction from silicates Water Extraction (H2O) for industrial consumables H2/O2 propellant Water (ice or liquid) radiation shielding Human life support consumables Plant growth consumables Fuel cell consumables Other volatiles extraction ( He3, H2, CH4, CO, etc.) Metals extraction for manufacturing

© ASCE

322

Earth and Space 2022

Mineral glass fibers for manufacturing Regolith bulk aggregate (Berms, Contours, sandbags) Radiation bulk shielding for human health: Solar Particle Events (SPE) & Galactic Cosmic Rays (GCR) Nuclear power plant bulk regolith radiation shielding Construction materials feedstocks (Concrete, bricks, pavers, etc.) Industrial processes (solvents, reactant, etc.) Solar photo voltaic arrays manufacturing for electrical power Thermal wadi’s for heat energy storage The most critical regolith and volatiles related functions in these applications are regolith and ice excavation, transportation to the end user, delivery/emplacement and removal of tailings. Regolith excavation and transportation also plays a critical role in site preparation and construction. An example of the workflow associated with ISRU is shown in Figure 1.

Figure 1. ISRU Workflow showing the Role of Regolith (G. Sanders, NASA) Similar functions and workflows will also be necessary on Mars and at Asteroids, but the respective different environments will dictate bespoke solutions. MINING REGOILITH Regolith excavation and hauling is a large part of mining operations. On Earth the technology has evolved to include autonomous mining machines. The advantages of using autonomy include: Increased safety and improved working conditions for personnel Improved utilization by allowing continuous operation during shift changes

© ASCE

323

Earth and Space 2022

Improved productivity through real-time monitoring and control of production loading and hauling processes Improved draw control through accurate execution of the production plan and collection of production data Lower maintenance costs through smooth operation of equipment and reduced damage Remote tele-operation of equipment in extreme environments Deeper mining operations with automated equipment Lower operation costs through reduced operating labor Reduced transportation and logistics costs for personnel at remote locations Control of multiple machines by one tele-operator human supervisor Lessons learned from terrestrial mining can be applied to extra-terrestrial mining, but the machines themselves will be very different, and will require a high degree of innovation and technology development. Some of the distinguishing features are: Lunar excavation requirements are different than terrestrial excavation Launch mass and volume limitations Low reaction force excavation in reduced and micro-gravity Operating in regolith dust Fully autonomous operations Encountering unknown sub surface rock obstacles Unknown water ice / regolith composition and deep digging Operating in the dark cold traps of permanently shadowed craters Unknown soil mechanics in polar regions Extreme access and mobility Slopes >35 degrees Extended nighttime operation and power storage Electrical power storage with high power density Thermal management in temperature extremes Robust “line of sight” RF or laser communications Long life and reliability Long term maintenance & life cycle Relevant technology development efforts have proceeded over the past 20 years, mostly for lunar regolith mining which can be extrapolated to Martian regolith mining. Asteroid regolith mining requires a completely different approach due to the very low 1/1000th G gravity field and the friable nature of the boulder and rubble piles found in asteroids. This paper will be limited to lunar excavation technologies and a future paper will address Martian and asteroid excavation for mining resources. Site preparation and construction are also a large customer for regolith excavation and transportation, but very little research has been performed in these domains, so this topic is also deferred to a future paper. HISTORY OF LUNAR EXCAVATOR PROTOTYPES The development of lunar regolith excavation robot prototypes was not seriously attempted until 2001, when Dr. Michael B. Duke, a former NASA geologist developed a bucket wheel excavator prototype with a team at the Colorado School of Mines (Muff et al, 2004) which is shown in Figure 2. Some testing was done using regolith simulants in collaboration with

© ASCE

324

325

Earth and Space 2022

NORCAT in Sudbury, Canada (Boucher, 2004). The bucket wheel was then adapted by Larry Clark and Tim Muff, while working for Lockheed Martin inc., and transformed into a bucket drum with inherent digging, storage and dumping qualities (Clark et al, 2009). After ISRU field tests in an analog test site on Mauna Kea volcano in Hawaii revealed excavator traction issues, a variety of other designs were also attempted and tested in regolith simulant bins. The NASA Centennial Challenge for Lunar Regolith Excavation (Everingham et al, 2008, Comstock et al, 2009) created a large pool of different designs that were tested from 2007-2009 in JSC-1A regolith simulant during the competition. The winning design was a bucket ladder style excavator which is advantageous for high speed mining but could have reliability issues due to the complexity of the mechanisms. While the competition only had a 30 minute run period, actual lunar operations will require years of operation, making reliability and repairability the primary concern. In 2010, the NASA Swamp Works team at Kennedy Space Center recognized that limited reaction force was the primary issue for lightweight lunar excavation, so they developed a counter-rotating bucket drum excavator which provides zero-horizontal reaction forces. Subsequent testing has shown this to be a very feasible design with good excavation results achieved during gravity offloading tests (Figure 4) in Black Point 1 (BP-1) lunar regolith simulant (Schuler et al, 2019).

Figure 2. Colorado School of Mines Bucket Wheel Excavator Other prototypes were tested during the NASA Desert Research & Technology Studies (RATS) analog field tests. In 2009-2011 a dozer blade was tested on a 1,000 kg lunar rover prototype called “Chariot” with good results (Mueller et al, 2009), The All-Terrain Hex-Limbed Extra-Terrestrial Explorer (ATHLETE) robot from the Jet Propulsion Lab (JPL) demonstrated a bucket excavator mounted on two of its limbs, and an actuated boom mounted bucket from Glenn Research Center (GRC) and Kennedy Space Center (KSC) on a Johnson Space Center (JSC) Centaur rover was also tested (Johnson et al, 2012; Bauman, 2016). The Canadian Space Agency also funded field testing prototypes based on a Load-Haul-Dump (LHD) design using a vertical actuation mechanism similar to a forklift design, created by the Northern Centre for Advanced Technology (NORCAT), now known as Deltion, inc. The LHD implement was mounted on a Ontario Drive & Gear (ODG) inc. mobility platform developed with Neptec, inc. Test campaigns were completed in Canada and Hawai’i (Visscher et al, 2014).

© ASCE

Earth and Space 2022

Figure 3. NASA Chariot rover with a bulldozer blade attached to it during Desert RATS field tests on sand dunes in Moses Lake, Washington.

Figure 4. NASA Regolith Advanced Surface Systems Operations Robot (RASSOR) excavator undergoing gravity off-loading tests The NASA Lunabotics Regolith Mining Competition (RMC) held at KSC has provided over 50 excavator prototypes each year since 2010 (Mueller et al, 2021), providing a wide variety of design ideas. Other efforts funded by the NASA Small Business Innovative Research (SBIR)

© ASCE

326

327

Earth and Space 2022

program produced notable prototypes including the application of pneumatic excavation on a micro-rover by Honeybee Robotics, inc. (Zacny et al, 2009) A list of extra-terrestrial robotic excavator prototypes is shown in Table 1. International projects for lunar excavation that are known, have been limited to documented Canadian and very recent Australian efforts, but more international efforts could exist and are not addressed in this paper. Table 1. History of NASA lunar excavator prototypes Dates

Excavator Prototype

1989-91

NASA Space Exploration Initiative (SEI) Eagle Engineering inc. Concept Studies Colorado School of Mines (Dr. Mike Duke research initiative) NASA GRC Cratos Scraper Excavator NASA Centennial Challenge for Lunar Excavation Lockheed Martin Bucket Drum - Mauna Kea, Hawai’i NASA Field Tests Canadian Space Agency, Mauna Kea, Hawai’i ISRU Tests (NORCAT# / Ontario, Drive & Gear (ODG) inc./Neptec inc. Juno Rover) NASA KSC LANCE* Dozer blade & JSC Chariot Mobility Platform JPL ATHLETEϮ hexapod robot with bucket implement Caterpillar inc. Multi Terrain Loader Tele-Operations at JSC SysRand inc. Moonraker bucket chain excavation implement Honeybee inc. Pneumatic PlanetVac Micro Excavator NASA JSC Space Exploration Vehicle (SEV) & LANCE NASA Lunabotics Robotic Mining Competition Honeybee inc. Planetary Volatile EXtractor (PVEX) Astrobotic inc. Polaris Bucket transverse bucket wheel excavator NASA JSC/GRC/KSC Centaur+ APEX^ + Badger bucket NASA KSC Swamp Works RASSORℇ NASA KSC Swamp Works ISRU Pilot Excavator (IPEx)

2001-2011 2007 2007-2009 2008 2008-2012 2009-2010 2009-2011 2009-2010 2009-2010 2009-2015 2010-2012 2010-2022 2010-2012 2010-2012 2013-2019 2010-2019 2019-2022 #

Northern Centre for Advanced Technology (NORCAT) now Deltion, inc. * Lunar Attachment Node for Construction Excavation (LANCE) Ϯ All-Terrain Hex-Limbed Extra-Terrestrial Explorer (ATHLETE) ^ Advanced Planetary Excavator (APEX) ℇ Regolith Advanced Surface Systems Operations Robot (RASSOR)

TAXONOMY OF LUNAR REGOLITH EXCAVATOR DESIGNS During the last 23 years, many excavator designs have been prototyped, primarily for competitions. The rules of these competitions reward fast results with excavation rates as high as 1,000 kg/hour having been demonstrated. Actual missions to the Moon will reward reliability and there will be more time to excavate, so that the design criteria will be different. Nevertheless, a large variety of robotic excavator designs have been proposed and prototyped which can inform future design efforts. The taxonomy of these designs, based on the NASA Lunabotics Robotic Mining Competition entries by university teams, has been cataloged (Mueller et al, 2021) and is shown in Tables 2, 3 & 4.

© ASCE

328

Earth and Space 2022

Table 2. Most popular excavation and regolith transportation mechanisms # sys 101 37 29 27 17 15 12 8 8 8 7 4 4 4 3 2 2 2 2 2

regolith excavation mechanism bucket ladder front end loader bucket belt bucketwheel bucket drum snow blower (auger or brush) auger backhoe bulldozer scraper large single scoop dual auger dual bucket wheel rotating brush excavating wheels claw/gripper scoop dual bucketladder dual counter rotating bucketdrums large bulldozer scoop paddle conveyor

# sys 103 40 22 21 15 11 8 7 7 6 6 4 4 4 4 3 2 2 2 2

regolith transportation mechanism bucketladder in scoop conveyor belt bucketbelt auger Over shoulder dump into hopper chute for guiding regolith bucketdrum drum bucketwheel impeller bucket rim bucketwheel discharge through bottom in bucket rotate scoop to slide simulant in hopper throw from impeller bucketwheel with side discharge paddle conveyor raising scraper with chute thrown from brush up ramp

Table 3. Most popular regolith storage and regolith dumping mechanisms

© ASCE

329

Earth and Space 2022

Table 4. Most popular robot movement mechanisms # sys 173 73 21 10 6 5 4 3 2 2 2 2 2 2 1 1 1 1 1 1

robot movement mechanism 4 fixed wheels tracks 6 fixed wheels 4 steerable wheels with custom profile two auger drums to propel stationary with swivel 4 fixed track wheels 4 digging wheels 3 wheels (2 driven, one steering) 4 six-legged wheels 4 wheels with suspension each of two robots hase 4 fixed wheels with grousers four individual steerable tracks three robots working together, two transport, one excavator, each with 4 fixed wheels 3 fixed wheels (front wheel swivels freely) 3 large wheels (2 with grousers, third with scoops) 4 medium and 2 large front wheels 4 wheels (two steerable coupled) with grousers 4 wheels with grousers, two of which have buckets to fill with regolith to increase counterweight 4 wheels, of which 2 steerable rear wheels

CONCLUSIONS Regolith excavation and transportation form the basis of regolith mining and also have critical applications in site preparation and construction activities. The unique extra-terrestrial environments (Moon, Mars, Asteroids) where excavation for ISRU and construction are likely to be needed are so different from terrestrial environments that completely new methods and devices must be invented and developed to meet these future needs. Logistics and space transportation are also difficult and expensive, so reducing mass and volume are important considerations for excavation equipment. Since 2001, many robotic excavator prototypes for extra-terrestrial uses have been attempted. However, the fidelity of the testing environments remains at a low technology readiness level (TRL) of 3 or 4, and the reliability of the proposed designs has not been demonstrated. Since these excavation robots will be operating in a much harsher and regolith interactive manner than current Mars and Moon scientific rovers, new ways of dealing with regolith dust, rocks, cryogenic icy regolith, extreme terrain, very cold operating temperatures, lunar nights, shadowed regions and more must be developed, before the NASA Artemis program, and others, can successfully use ISRU and perform in-situ construction. REFERENCES Hill, B. (2018). 45th Space Congress “The Next Great Steps”: Space Policy Directive-1. Muff, T., Johnson, L., King, R., & Duke, M. B. (2004, February). A prototype bucket wheel excavator for the Moon, Mars and Phobos. In AIP conference proceedings (Vol. 699, No. 1, pp. 967-974). American Institute of Physics. © ASCE

Earth and Space 2022

Boucher, D. S., & Richard, J. (2004). Report on the construction and testing of a bucket wheel excavator. P3N 1L1. Clark, D. L., Patterson, R. R., & Wurts, D. W. (2009, September). A novel approach to planetary regolith collection: the bucket drum soil excavator. In AIAA space 2009 conference & exposition (p. 6430). Metzger, P. T., Muscatello, A., Mueller, R. P., & Mantovani, J. (2013). Affordable, rapid bootstrapping of the space industry and solar system civilization. Journal of Aerospace Engineering, 26(1), 18-29. Everingham, M. R., Pelster, N., Mueller, R. P., & Davidian, K. (2008, January). Preparation and Handling Large Quantities of JSC-1A Lunar Regolith Simulant for the 2007 Regolith Excavation Challenge. In AIP Conference Proceedings (Vol. 969, No. 1, pp. 268-273). American Institute of Physics. Comstock, D. A., Petro, A. (2009). NASA’s Centennial Challenges contributions to ISRU. In:47th AIAA Aerospace Sciences Meeting Including the New Horizons Forum and Aerospace Exposition, 1205. Schuler, J. M., Smith, J. D., Mueller, R. P., Nick, A. J. (2019). RASSOR, the Reduced Gravity Excavator. Lunar ISRU 2019, Developing a New Space Economy through Lunar Resources and Their Utilization, 5061 (abs.), Lunar Planetary Institute, Workshop, July 15-17, 2019, Columbia, Maryland Mueller, R. P., Smith, J. D. (2019). NASA Kennedy Space Center Swamp Works: capabilities and facilities. In: Lunar ISRU 2019, Developing a New Space Economy through Lunar Resources and Their Utilization, 5069 (abs.). Lunar Planetary Institute, Workshop, July 1517, 2019, Columbia, Maryland Sowers, G. F., & Dreyer, C. B. (2019). Ice mining in lunar permanently shadowed regions. New Space, 7(4), 235-244. Mueller, R., Schuler, J., Nick, A., Wilkinson, A., Gallo, C., & King, R. (2009, September). Lightweight bulldozer attachment for construction and excavation on the lunar surface. In AIAA SPACE 2009 conference & exposition (p. 6466). Johnson, K., Creager, C., Izadnegahdar, A., Bauman, S., Gallo, C., & Abel, P. (2012). Development of field excavator with embedded force measurement. In Earth and Space 2012: Engineering, Science, Construction, and Operations in Challenging Environments (pp. 365-374). Bauman, S., Newman, P., Izadnegahdar, A., Johnson, K., & Abel, P. (2016). A Basic Robotic Excavator (the “Glenn Digger”): Description, Design, and Initial Operation. National Aeronautics and Space Administration, Glenn Research Center. Mueller, R. P., van Susante, P., Reiners, E., & Metzger, P. T. (2021). NASA Lunabotics Robotic Mining Competition 10th Anniversary (2010–2019): Taxonomy and Technology Review. Earth and Space 2021, 497-510. Boucher, D. S., Atwell, J. T., Theiss, R., Armstrong, R., Benigni, S. (2011). Development and testing of an autonomous regolith excavation and delivery system. In: 49th AIAA Aerospace Sciences Meeting Including the New Horizons Forum and Aerospace Exposition, vol. 431. Graham, L. D., Graff, T. G. (2013). Rover-based instrumentation and scientific investigations during the 2012 analog field test on Mauna Kea Volcano, Hawaii. In: Lunar and Planetary Science Conference.

© ASCE

330

Earth and Space 2022

Visscher, P. D., & Woolley, D. B. (2014). Lunar rover analogue mission deployments. In 7th Symposium on Space Resource Utilization (p. 0686). Zacny, K., Mueller, R., Galloway, G., Craft, J., Mungas, G., Hedlund, M., ... & Fink, P. (2009). Novel approaches to drilling and excavation on the moon. In AIAA SPACE 2009 Conference & Exposition (p. 6431).

© ASCE

331

332

Earth and Space 2022

Lateral Stability of Vehicle with Interlocking Spikes Volker Nannen1 and Damian Bover2 1

Sedewa, Finca Ecologica Son Duri, Vilafranca de Bonany, Spain. Email: [email protected] 2 Sedewa, Finca Ecologica Son Duri, Vilafranca de Bonany, Spain. Email: [email protected] ABSTRACT The interlock drive system generates traction by penetrating narrow articulated spikes into the ground and by using the strength of the deeper soil layers to resist horizontal draft forces. The system promises good tractive performance in low gravity environments where tires generate little traction due to low vehicle weight. Possible applications include heavy-duty vehicles for civil engineering tasks like earthmoving or mining excavation. Safe vehicle operation in complex terrain geometry requires lateral vehicle stability to prevent vehicle rollover. Good lateral stability is a particular requirement for excavation and piling operations where the margins of safety define the terrain geometry that can be worked in, and it is a major constraint in operational planning. An earthmoving vehicle that can operate at a high roll angle reduces the need to maintain ramps in pits and on piles and can shorten and simplify the paths for individual maneuvers. Here we report on several field trials on the lateral stability of an earthmoving vehicle that uses the interlock drive systems. We find that the vehicle can work well at roll angles of up to 20°, but that it needs further improvement if work at roll angles of 30° or more is required. INTRODUCTION Renewed government and commercial interest in lunar and planetary mining and construction pose the challenging engineering question of how to excavate, transport, and tip granular materials—a field long dominated by heavy machinery—in the low gravity environments of the Moon, Mars, and the larger asteroids. Mueller and King (2008) identify the following basic requirements for lunar excavators: navigate the lunar surface without getting stuck, avoid rocks, traverse 20° slopes when fully loaded, minimize total power and peak power, operate reliably over long periods. Just et al. (2020) identify excavation rate, traverse speed, power consumption, and simulant properties as the key performance metrics for the evaluation of regolith excavation techniques. Excavation rate, traverse speed, and power consumption depend on the terrain and the geometry of the deposit. In terrain that is not flat and does not have dedicated access roads, a vehicle that can move or work at high pitch or roll angles can travel shorter distances, which increases traverse speed and reduces vehicle power consumption. Reducing the need to create and maintain a system of access roads also simplifies the logistics and reduces the overall power consumption of mining and other civil engineering projects. The ability of a machine to move and work on a slope angle of at least 20° is therefore a key parameter for the selection of suitable excavation techniques. Tires generate traction on granular material through friction with and between the upper particles of the material on which they move. On terrestrial soils, tires achieve an optimal tractive

© ASCE

Earth and Space 2022

efficiency near a pull/weight ratio of 0.4 (Zoz and Grisso, 2003). In a lunar environment, Wilkinson and DeGennaro (2007) expect a lower pull/weight ratio of 0.21. This poses a serious challenge to the design of civil engineering vehicles like rippers, scrapers, and crawlers. If they rely on tires for traction, every kg of launch mass is expected to generate a tractive force of only 0.35 N. An alternative traction method is to interlock spikes with the ground (Bover 2011; Nannen et al. 2016, 2017). As described by Nannen and Bover (2021), efficient and reliable ground penetration can be achieved if narrow spikes are attached to a lever arm that is attached to a hinge close to the ground, see Figure 1. A backward force on such a spike drives it into the ground to a depth where the lateral strength of the ground equals the draft force, in a selfregulating manner. A forward force pulls the spike out of the ground. Such interlocking spikes can be integrated with a push-pull vehicle (Creager et al., 2012) where alternating frames push or pull tools like rippers or blades from the anchored spikes.

Figure 1: Crawler with spikes for traction. Top: demonstrator on beach sand. The spikes are painted red. In an alternating motion pattern, the two spikes on the rear frame on the left push the blade forward, and the two spikes attached to the blade on the right pull the rest of the vehicle forward. Bottom: schematic drawing. Only one of the spikes that push the blade is shown. Important design parameters for the spike are the diameter of the spike, the rake angle α between the soil surface and the spike, and the thrust angle γ between the soil surface and a line from the spike tip to the hinge. γ in turn depends on the spike radius r, which is the distance between spike tip and hinge, as well as the elevation of the hinge above the ground. Spike diameter and rake angle determine the resistive force of the soil. The difference α – γ defines the penetration force and is best kept in the range 15° < α – γ < 35°. See Table 1 for a list of symbols.

© ASCE

333

334

Earth and Space 2022

Table 1: List of symbols α FD FL r

rake angle, inclination from horizontal of the spike thrust angle, inclination from horizontal of the line from hinge to spike tip draft force at the blade lift force at the hinge spike radius, the distance between spike tip and hinge

Since a horizontal draft force F D creates a lift force FL = FD × tan γ at the hinge, keeping the thrust angle small is important for vehicle stability. For a vehicle with an intended pull/weight ratio of 2, the thrust angle γ should not exceed 26.5°. With spikes as described here, if more tractive force is required from a vehicle with a given weight, the pull/weight ratio can be improved by increasing the spike radius. There is no need to add ballast. To test the ability of a push-pull vehicle that uses interlocking spikes for traction to move and work on a slope of granular material, we performed field trials with a crawler in two relevant environments: a quarry and a beach. The task of the crawler was defined as path clearing, i.e., to clear away ridges and fill gaps in the ground, so that a smooth path is created. EXPERIMENTAL SETUP We conducted the field trials on fine-medium beach sand of bioclastic origin at the Bay of Palma, and on fine to coarse granular material at a stone quarry near the town of Petra, both on the island of Mallorca, Spain. The crawler was small and light enough to be deployed manually. The blade was 80 cm wide and 40 cm high, forming the widest and highest part of the vehicle. The blade was welded to the crawler, with a fixed vertical position and angle. The total vehicle weight was 40 kg. The total vehicle length without spikes was 2.2 m. The center of mass of the crawler was about 10 cm above the ground. The crawler consisted of two alternating frames. The first frame consisted of a 2 m long central bar, with the blade at its front and two small spikes to the rear of the blade, 60 cm apart. The second frame consisted of a 60 cm bar orthogonal to and centered at the long bar, to which an electric motor with a gearbox and two large spikes were attached, 55 cm apart. The motor drove a cogwheel that meshed with a motorcycle chain welded to the top of the central bar. By moving the second frame back and forth along the central bar, the two pairs of spikes were pushed into and pulled out of the soil in an alternating motion pattern, pushing the crawler forward. The second frame traveled 1.15 m along the central bar. The crawler was powered via cable by solar cells, backed up by a car battery for stability. Peak power was 250 Watt. The motor was set to move at a constant speed, which was 0.1 m/s in most cases, except when we tried the crawler uphill on a 20° slope, where it was set to 0.07 m/s. At this level of power and a speed of 0.1 m/s, we expected the crawler to develop a tractive force of up to 1.5 kN, for a maximum pull/weight ratio of 4. The spikes of the first frame were only meant to pull the second frame forward and were comparatively small, designed for a maximum penetration depth of up to 15 cm. Their spike radius was 60 cm, and their spike diameter was 12 mm. The spikes of the second frame, intended to push the first frame, the blade, and the granular material forward, were designed for a maximum penetration depth of 50 cm, had a spike radius of 135 cm, and a spike diameter of 21 mm. Previous tests on beach sand had demonstrated that these spikes can easily sustain a draft of 2 kN and more. The beach at the Bay of Palma had a well-defined slope along the waterline which increased from 12° at the waterline to 16° at the top edge, from where it continued inland almost flat for

© ASCE

Earth and Space 2022

25–30 m. The edge was 1.7 m above the waterline. The horizontal distance from the waterline to the edge was 6.8 m. The sand was moist, especially close to the waterline. No rain had been recorded in the week before the trials. At the quarry, stone is broken, sorted by size, and stockpiled at various locations in the quarry. To allow access with heavy machinery, the larger stockpiles have an access ramp of 17°–20° on one side. All other sides form at the angle of repose of the material as it is dumped from the top. We tested on two different stockpiles: the stockpile with the finest material and another stockpile of coarse sand mixed with small stones, apparently what had been scraped off the ground after clearing another stockpile. The stockpile with the finest material had a 20° access ramp, which was uneven because of traffic by heavy machinery, and a smooth wall at an angle of repose of 40°. The pile of coarse sand had a slope angle of 30° and no ramp. It had an undulating surface because every dump of new material formed a new cone that partly overlapped with the old ones. The material at the quarry was moist but well-drained, 7 days or more after the last recorded rainfall, except for two trials: uphill on the 20° slope and contour following on the 40° slope were done one day after the last rain. We tried the crawler in three different orientations along the slope of the different faces: following the contour, moving up diagonally to the line of steepest ascent, and moving uphill along the line of steepest ascent. When following the contour, vehicle roll matched the terrain slope while vehicle pitch was 0°. Moving uphill, vehicle roll was 0° while vehicle pitch matched the terrain slope. When moving diagonally uphill, both values were somewhere in between. We tried to let the crawler move 10 m or more during each trial. This was not always possible. For example, when moving uphill at the beach, the maximum possible path length was 7 m. Table 2 offers a complete overview of all trials. Results were recorded on paper, in photos, and in videos. Vehicle angles were measured with the vehicle at rest. Electrical power was recorded at the power source. The trials were evaluated for whether the spikes penetrated the ground reliably, whether the crawler consistently moved forward, whether it moved straight (vehicle steering was not part of the trial, but a vehicle that won’t move straight by itself tends to be more difficult to steer), whether the crawler successfully pushed sand, and whether it successfully cleared a path. RESULTS On the beach, the crawler performed mostly as intended. The spikes entered the sand reliably. The blade pushed sand at the intended peak force of 1.5 kN and cleared a path. When following the contour and when moving diagonally uphill, the vehicle veered slowly downhill, see Figure 2. When moving uphill, the path was straight. At the quarry on the 20° ramp of fine material, the spikes entered the material reliably. When following the contour and when moving diagonally uphill, the blade pushed the material at the intended peak force of 1.5 kN and cleared a path, see the upper photo in Figure 3. The vehicle might have veered downhill, but the terrain was too uneven to say so for certain. When moving uphill on the 20° ramp of fine material at the quarry, we decreased the motor speed to 0.7 m/s and increased the volume of material in front of the blade, increasing the draft to an estimated 2.5 kN, for a pull/weight ratio of 6. When the terrain was even, the crawler pushed the sand and cleared the path. When the terrain was uneven, one of the large spikes reached a thrust angle of 25° and lifted the vehicle. It could not continue. When we reduced the volume in front of the blade to an estimated draft of 2 kN and a pull/weight ratio to 5, the crawler resumed operations. Except for the described effect at an increased pull/weight ratio, the spikes penetrated the material reliably on a 20° slope, the material was pushed forward, and the path was cleared.

© ASCE

335

336

Earth and Space 2022

On the 30° stockpile of coarse material, the crawler did not move straight but slowly veered downhill. The large rear spikes penetrated reliably and pushed the blade with material forward. However, the small spikes in front repeatedly lifted the blade, such that the crawler could not continue, see Figure 4. On the 40° wall of fine material, the crawler quickly veered downhill without pushing any material, see Figure 5. Table 2. Trials with main observations. For the slope, the average inclination is given, which might not agree with the measured vehicle orientation. Terrain slope is given in degrees and percent. The main text uses degrees only. Location

beach, flat beach, 12° = 21% slope beach, 12°–16° = 21%–29% slope

Vehicle path, nr. of trials flat, 1 trial contour, 3 trials

Vehicle rotation

Observations

pitch: 1° roll: 0° pitch: 0° roll: 12°

Reliable spike penetration. The crawler moved straight, pushed all sand forward, and cleared a path. Reliable spike penetration. The crawler pushed all sand forward and cleared a path. The vehicle had a slight tendency to veer downhill.

pitch: 8° roll: ~12° pitch: 12°–16° roll: 0° quarry, fine contour, pitch: 0° material, 1 trial roll: 20° 20° = 36% diagonal, pitch: N/A slope 2 trials roll: 7° uphill, pitch: 20° 2 trials roll: 0°

quarry, coarse material, 30° = 58% slope

diagonal, 5 trials uphill, 3 trials

contour, 2 trials diagonal, 1 trial

pitch: 0° roll: 30° pitch: 12° roll: 27°

uphill, 1 trial

pitch: 29° roll: 10°

quarry, fine contour, material, 40° 1 trial = 84% slope

© ASCE

pitch: 5° roll: 40°

Reliable spike penetration. The crawler moved straight, pushed all sand forward, and cleared a path. Reliable spike penetration. The crawler moved mostly straight (difficult to assess in uneven terrain), pushed all sand forward, and cleared a path. Reliable spike penetration. The crawler moved straight. On even terrain, at a draft of ~ 2.5 kN, the crawler pushed all sand forward and cleared a path. On uneven terrain, at a draft of ~ 2.5 kN, one large spike lifted the crawler, and it could not move forward. At a draft of ~ 2 kN, the crawler pushed sand forward and cleared a path. Reliable penetration of the large spikes. The crawler did not move straight but slowly veered downhill. When the small spikes in front worked well, the crawler pushed all sand forward and cleared a path. Eventually, the small spikes in front lifted the blade such that the crawler could not continue. Reliable penetration of the large spikes. The crawler moved straight, pushed all sand forward, and cleared a path. The small front spikes lifted the blade, but the crawler could continue. The crawler veered quickly downhill.

Earth and Space 2022

Figure 2. Path clearing at the beach. Top: contour following. Bottom: diagonally uphill.

© ASCE

337

Earth and Space 2022

Figure 3. Path clearing on the 20° ramp at the quarry. Top: contour following. The vehicle has successfully broken through a large accumulation of sand. Bottom: uphill against a draft of 2.5 kN.

© ASCE

338

Earth and Space 2022

Figure 4. Path clearing on a 30° slope at the quarry. Top: the crawler moves material. Bottom: the small spikes lift the blade; the vehicle cannot move forward.

© ASCE

339

Earth and Space 2022

Figure 5. On the 40° slope at the quarry, the crawler quickly veers downhill.

© ASCE

340

Earth and Space 2022

DISCUSSION AND CONCLUSION We tested the performance of a crawler with spikes for traction on a side slope in two relevant environments: a beach and a quarry. While normal tires achieve an optimal tractive efficiency at a pull/weight ratio of 0.4 and cannot exceed a pull/weight ratio of 0.9, the tested crawler was designed to move earth at a pull/weight ratio of 4. A requirement for such a high pull/weight ratio is, that the thrust angle—the angle between the horizontal and a straight line between spike tip and hinge—does not exceed 15°. Otherwise, the pull on the spikes creates a lift at their hinges which exceeds vehicle weight. During most trials, the thrust angle did not exceed 15°, and the vehicle safely reached the intended peak power and peak force. On the 20° slope we decreased the vehicle speed while keeping the power supply constant and increased the load at the blade, thereby increasing the pull/weight ratio to 6. The crawler performed well when the terrain was even. When the terrain was uneven, the thrust angle of one spike increased beyond 15° and lifted the machine, even though the other spike had the intended thrust angle of 15°. We conclude that on granular materials the calculated pull/weight ratio can be exceeded, though not by much. We assume that the reason for this is, that in granular materials with little or no cohesion the center of force of the spike is not at the tip but further up, decreasing the effective thrust angle. We suggest that an actuator that prevents one spike to penetrate much deeper than other spikes could improve vehicle stability in uneven terrain. We observe that the crawler performed generally well on slope angles of up to 20°. The spikes penetrated reliably, the crawler pushed material as intended and cleared a path. When following the contour or moving up diagonally, the vehicle veered slowly downhill, which requires adaptive path control. At a 30° slope angle, the vehicle in its present configuration failed to operate as intended, chiefly because the front spikes were too small for the task (or the front of the vehicle was too light). The vehicle also tended to veer downhill. With some improvements to the design, the vehicle might be able to traverse a 30° slope angle. Whether it can also perform useful work while moving along the contour of or diagonally up a 30° slope needs to be further investigated. While we have previously demonstrated that a similar vehicle can climb a 40° slope angle (Nannen et al., 2016), we find that in the present configuration, the vehicle cannot safely traverse a 40° side slope angle. ACKNOWLEDGEMENT We thank the owners of Gravillera Son Chibetli, S.L. for their generous permission and support while working at their quarry. REFERENCES Bover, D. (2011). Autonomous Self-Actuated Ploughing Implement. US Patent 9,144,188, see http://sedewa.com/patents.html Creager, C.M., Moreland, S.J., Skonieczny, K., Johnson, K., Asnani, V. & Gilligan, R. (2012). “Benefit of ‘push-pull’ locomotion for planetary rover mobility.” Earth and Space 2012, Pasadena, CA.

© ASCE

341

Earth and Space 2022

Just, G.H., Smith, K., Joy, K.H. & Roy, M.J. (2020), “Parametric review of existing regolith excavation techniques for lunar In Situ Resource Utilisation (ISRU) and recommendations for future excavation experiments”, Planetary and Space Science, 180 Mueller R.P. & King, R.H. (2008), “Trade Study of Excavation Tools and Equipment for Lunar Outpost Development and ISRU”, AIP Conference Proceedings 969, 237–244, 10.1063/1.2844973 Nannen, V., Bover, D., Zöbel, D., Parma, F., Marascio, K. & McKenzie, B.M. (2016). “UTOPUS Traction Technology: A New Method for Planetary Exploration of Steep and Difficult Terrain.” Proc. ISTVS 8th Americas Conf., Detroit, MI. Nannen, V., Bover, D. & Zöbel, D. (2017). “A Novel Traction Mechanism Based on Retractable Crampons to Minimize Soil Compaction and Reduce Energy Consumption.” Technical Report 2017-03, Sedewa. Nannen, V. & Bover, D. (2021). Interlocking Spikes for Extreme Mobility. Proceedings of the Earth & Space Conference 2021, pp 521–530. Söhne, W. (1969). “Agricultural engineering and terramechanics.” J. Terramechanics 6, 9–30. Zoz, F.M. & Grisso, R.D. (2003). Traction and Tractor Performance (ASAE Publication Number 913C0403). Amer. Soc. Agric. Biol. Eng.

© ASCE

342

343

Earth and Space 2022

Liberation of Mineral-Bound Water of the Meridiani Planum Driven by Process Heat from Carbonylation Steel-Making and Concentrated Photovoltaic Electricity Generation Rif Miles Olsen1 1

Two Planet Life and Two Planet Steel. Email: [email protected]

ABSTRACT A proposal is made to liberate water from an extensive reservoir of water bound to hydrated sulfate minerals in a geological structure under the Meridiani Planum called the Burns Formation. This formation of sediments spans the equator, is over 500 km wide, has an area larger than Lake Superior, and has depths typically between 200 m and 1,000 m. About 3 kJ of enthalpy must be transferred into the sediments for each 1 gram of water vapor liberated by dehydrating the sulfates. Relatively large energy requirements are needed to liberate water by the tonne and the kilotonne. A crucial part of the proposal is to link water liberation to power tower concentrated photovoltaic electricity generation and carbonylation steel-making, both of which produce a lot of process heat. This process heat is at suitable temperatures for dehydrating the main sulfate minerals in the Burns Formation. Carbonylation steel-making benefits from an easy-to-use source of hematite that lies in a thin layer on top of soils overlying the Burns Formation. Carbonylation steel-making produces steel powder. Powder sintering to sheet steel plus robotic cutting, welding, and bending then provide a manufacturing capability. Simple electricity generation plants and water liberation plants are described that can be manufactured locally. This makes possible a large power generation capacity on the Meridiani Planum, including a large process heat-generating capacity and a deployment of water liberation equipment that scales with the available process heat capacity. INTRODUCTION The extraction of water using Rodriguez Wells (Rodwells) from underground glaciers in mid-latitude locations of Mars, such as Utopia Planitia, is now an area of active research (van Susante et al., 2021; Mellerowicz et al., 2021). A high priority use for this water is rocket propellant production to return spacecraft, such as the SpaceX Starship, from Mars to Earth (Musk, 2018), although water has many uses. This paper proposes an alternative to Rodwell extraction from underground glaciers. There is a great deal of bound water under the approximately 500 km wide plain called the Meridiani Planum. This plain straddles the equator and is the flat top to a large geological structure of sediments called the Burns Formation (Edgett and Parker, 1997;Grotzinger et al., 2005; Arvidson et al. 2006). Most bound water is in hydrated magnesium, calcium, and ferric sulfates in the formation’s sediments. These sulfates release water vapor when heated to modest temperatures. The proposed water liberation method is to heat these sulfates with process heat from carbonylation steel-making and concentrated photovoltaic (CPV) electricity generation. A feature of this proposal is that CPV electricity generation, steel-making, and steel manufacture can be mutually-self-reinforcing activities. Since most of the plant needed for power tower CPV generation is made from steel while electricity drives both the steel-making and robotic steel manufacture. Self-reinforcing activity leads to exponential growth. If exponential

© ASCE

Earth and Space 2022

growth in CPV electricity generation and steel manufacture is achieved, then any of water we might want on Mars in the next twenty-five years can be liberated from the hydrated sulfates in the Burns Formation. Another feature of this proposal is that lying on top of the Burns Formation is the easiest-touse source material for steel-making anywhere on Mars (or on Earth), i.e., loose spherules of hematite called blueberries (Olsen, 2021). This paper will: 1. Review the relevant literature on the Meridiani Planum and the Burns Formation. 2. Describe small Australian power-tower CPV plants and their process heat. 3. Outline carbonylation steel-making and its process heat. 4. Describe the basics of heating Burns Formation sediments and describe water liberation in ovens and water liberation in-place in the sediments. 5. Give concluding comments. THE MERIDIANI PLANUM AND THE BURNS FORMATION The Mars Odyssey spacecraft went into orbit around the Red Planet in October 2001. It carried a neutron detector/spectrometer capable of measuring water equivalent hydrogen (WEH) in surface material (Boynton, 2004). Mars Odyssey quickly produced a global map of WEH, Feldman, (2003), which was subsequently refined (Prettyman et al., 2004, 2009; Feldman et al., 2004, 2007, 2008a, 2008b; Diez et al., 2008; Pathare et al., 2018). This map shows that the highest concentrations of WEH are in polar regions; however, it also shows two large, transequator regions with high WEH levels: The Meridiani Planum forms the western end of one of these regions. The 2018 maps show weight percentages of WEH ranging smoothly from 10% up to 14% (Pathare et al., 2018) across the plain. The penetration depth of the orbiting detector was around 1 m, so it measured near-surface WEH (Feldman et al., 2004). Between January 2004 and February 2018, NASA’s surface rover Opportunity conducted many experiments on the sediments of the Burns Formation and the soil lying on top of it. The WEH found in Opportunity’s ground measurements is bound in hydrated sulfate minerals in the sediments of the Burns Formation (Squyres et al., 2004; Reider et al., 2004; Grotzinger et al., 2005; Arvidson et al., 2006). The near-surface measurements of the orbiting neutron detector may under-estimate the water content of the Burns formation since a lot of the material measured by that detector was topsoil, and this soil is depleted in sulfate minerals (Golombek et al., 2006) that the underlying sediments are rich in (Squyres et al., 2004; Reider et al., 2004). Further, the levels of hydration may increase with depth into the sediments. The area of the Burns formation is between that of Lake Superior and two Lake Superiors (Christensen et al., 2000; Christensen and Ruff, 2004). Sediment depth usually varies between 200 m and 1000 m (Edgett 1997; Grotzinger et al. 2005). Its flat top, the Meridiani Planum, is relatively easy for spacecraft to land on (Christensen et al., 2005). The micro-scale texture of the formation’s sediments has four main elements: medium-tocoarse grained sand, nearly spherical (1 –8 mm) diagenetic concretions, angular vugs (cavities), and fine-grained cement embedding everything else (Squyres et al., 2004). The embedded concretions are rich in grey hematite (Klingelhöfer et al., 2004; Christensen et al., 2004; Morris et al., 2006; Calvin et al. 2009). They are widely called blueberries. (They are grey but look blue against the ubiquitous rusty colors of Mars). The blueberries only make a small weight percentage of the sediments (Golombek et al., 2006; Calvin et al., 2009).

© ASCE

344

Earth and Space 2022

The sediments started forming more than 3.7 billion years ago when large amounts of basaltic material were transported to the region by rivers (Hynek and Phillips, 2001). In the period between 3.7 billion years and 3.5 billion years ago, in a gradual drying period, the waters became acidic, the water levels varied, and diagenetic processes altered the sediments producing the sulfates and the blueberries (Squyres et al., 2004; Grotzinger et al., 2005; McLennan et al., 2005; Hurowitz et al., 2010). The water content in hydrated magnesium sulfates can go all the way up to that of Meridianiite (MgSO4.11H20). Meridianiite was named in 2007 after the Meridiani Planum and, in particular, after what Peterson et al. (2007) think originally filled the vugs described by Squyres et al. (2004). There is a small scientific field concentrating on how the hydration levels of magnesium sulfate salts vary with Martian conditions (see, for example, Vaniman and Chipera, 2006; Chou and Seal, 2007; Peterson et al., 2007; Grevel et al. 2012; Okhrimenko et al., 2017). Somewhere between 3.5 billion years ago and 3.0 billion years ago, the plain became very arid, and change became very slow (Hartmann and Neukum, 2001; Bibring et al., 2006; Golombek et al., 2006). For the past 3+ billion years, the top sediment layers were changed by slow meteoritic bombardment and wind erosion (Golombek et al., 2006, 2014). This disintegrated the top sediment layers and left larger basaltic sand particles as soil on top of the remaining sediments. This soil covers a significant fraction of the plain, but satellite and rover images also show widespread top sediment outcrop, principally in between soil bedforms called plains ripples (Golombek et al., 2014; Fenton et al., 2015). These ripples occur in giant fields covering many square kilometers (Fenton et al., 2015). The soil also forms into smooth sheet bedforms. These also cover many square kilometers but with little or no sediment outcrop (Soderblom et al., 2004; Golombek et al., 2014; Fenton et al., 2015). Soil bedforms are shallow, usually less than 0.3 m deep; the maximum depth found is close to 1 m (Golombek et al., 2014; Fenton et al., 2015). Erosion and size-sorting loosened and concentrated formerly embedded blueberries (and blueberry fragments) into a thin (1 cm) top layer to the soil bedforms (Soderblom et al., 2004; Arvidson et al., 2006; Calvin et al., 2009; Sullivan et al., 2011; Fenton et al. 2015). The surface densities of top, loose blueberries, and blueberry fragments are large; total numbers are enormous (Calvin et al., 2009; Olsen, 2021). Fenton et al. (2015) and Golombek et al. (2014) made strong arguments that most of the topsoil material on the Burns Formation sediments is not material transported onto the plain by wind but, rather, eroded material from the sediments themselves. Although, there are also meteoritic components (Yen et al., 2006) and small amounts of crater ejecta (thrown long distances onto the plain). An important part of the sediments not retained in the topsoils was the hydrated sulfate minerals. The sulfate levels in the soils are far below the sulfate levels in the sediments (Golombek et al., 2006; MER APXS team, 2016). The lost sulfates were first turned into dust (by bombardment and erosion) and then transported off the plain by wind (Golombek et al., 2006). The composition of blueberries and their fragments make them excellent source material for carbonylation steel-making (Klingelhöfer et al., 2004; Morris et al., 2006; Olsen (paper submitted)). Robotic harvesters can simply lift loose blueberries off the tops of soil bedforms (Olsen, 2021). This harvesting does not require any blasting, drilling, or digging. Mostly autonomous operation of robotic harvesters on smooth sheet bedforms will be readily achievable due to (I) the remarkable smoothness of these bedforms (Soderblom et al. 2006; Golombek et al. 2014; MER PanCam Team, 2016); (II) the thin, top-layer location of blueberry material; and, (III) the years-long global efforts in the automobile and related industries to develop self-driving vehicles.

© ASCE

345

Earth and Space 2022

SMALL, POWER TOWER CPV PLANTS PRODUCING ELECTRICITY AND PROCESS HEAT The Australian company Raygen Resources Pty Ltd has matured a combination of technologies into a new, solar-driven power plant. This is developed enough that it attracts corporate investment and a new plant is now under construction at a larger scale than a previous plant financed by the Australian government. Variations of Raygen’s power plant are suitable for operating on Mars. The sun-facing side of a Raygen plant consists of one or more small, power tower format solar concentrators, each one with a solar receiver made from cooled, concentrated photovoltaic (CPV) panels. With solar concentration and receiver cooling, Raygen CPV receivers run with power intensities 4000X larger than standard solar panels. This higher power intensity is a feature on Mars because of the transport benefit it produces; that is, it reduces the mass of solar panels that need transport to Mars in a spacecraft. The main components in a power tower plant are (a) movable heliostats, (b) the tower, and (c) pipes and tanks used for moving fluids up and down the tower – all of these components can be robotically manufactured from sheet steel using cutting, bending and welding. There are very few other components (beyond (a), (b), (c), and the solar receiver) needed to construct a power tower plant. Once steel-making and sheet-steel manufacture are in place, many new power tower CPV plants can be built on the Meridiani Planum, made mainly from blueberry source material. This is certainly a feature. It can lead to growth on the plain and more of everything, including more electricity generation, more steel-making and manufacture, and more water liberation. CPV panels are about twice as efficient as standard photovoltaic (PV) panels without concentration at converting light energy to electrical energy. According to Wikipedia, the record conversion efficiency is a jaw-dropping 46%. Raygen claims 32% for its CPV panels, while PV panel efficiencies range from 15% to 18%. Another feature of CPV plants like Raygen’s, with concentrators in the power tower format, is that a lot of the energy not turned directly into electricity can be captured as process heat in fluids at temperatures well above ambient (around 95 oC in Raygen’s case). Fluids at such temperatures are hot enough to drive some useful tasks, including (i) (surprisingly) driving a low-temperature Rankine engine for generating electricity at night (this is a particular feature of Raygen plant, pulled-off with some cleverness, and a reason for investor interest in Raygen); (ii) heating spacecraft, buildings, and equipment to protect against the cold of Martian nights; and (iii) liberating water by dehydrating hydrated sulfates in Burns Formation sediments. In contrast, the extra energy from standard PV plant (not turned into electricity) is nearambient temperature heat. This heat cannot drive useful processes. The most significant change from a Raygen power tower CPV plant operating on Earth to a power tower CPV plant operating on the Meridiani Planum would be to change the system fluid for receiver cooling and for transporting/storing process heat. On Mars an alternate system fluid is pressurized carbon dioxide. By Mars exploration standards, a single small Raygen CPV power tower plant produces a lot of electrical and thermal power. The three small power tower plants financed by the Australian government, and operating since 2015, with short tower heights of around 12 m, each has 250 kW peak electricity outputs and 500 kW peak thermal outputs. One such small power tower unit will produce about 4000 MWh of process heat in one Earth/Mars synodic period of 780 days.

© ASCE

346

347

Earth and Space 2022

CARBONYLATION STEEL-MAKING AND PROCESS HEAT The core processes of carbonylation steel-making convert raw source materials into steel powder of mostly iron, small amounts of carbon, and nickel (if nickel is in the source material, as it is in blueberries). It is straightforward to separate the nickel from the iron during an intermediate step. Such nickel/iron separation is important for making high purity powders of interest to the production of cathodes for lithium-ion batteries. Starting steel-making on Mars using carbonylation steel-making makes sense for the following reasons: 1. All equipment needed for an end-to-end process can be scaled to fit SpaceX Starship payload size constraints. 2. It uses relatively low temperatures, so no ceramics or refractories are needed in any equipment, and no water is used for cooling. 3. Following (2), all equipment is made of just metals and some rubber so that this equipment will easily pass vibration/stress tests for spacecraft transport, and it can survive a fairly bumpy landing on Mars – it is tough. 4. Following (2), there are no plant shutdowns for ceramic/refractory repairs/replacement. 5. Beyond the source blueberries, it has only one consumable, easily obtainable atmospheric carbon dioxide. No graphite electrodes, slagging agents, coke, oxygen, or refractories are needed. 6. Carbonylation cleanly separates iron and nickel from other materials. 7. It can work with low- or high-quality source material. 8. No complicated or high-energy beneficiation preprocesses are needed. 9. Robots can operate carbonylation steel-making. 10. Its outputs are steel powder, which can be turned into sheet steel with sintering methods and small parts with 3D metal printing. Many other steel-making and metallurgical methods will likely be used on Mars at some point in the future. The great benefits of carbonylation steel-making are (A) that it is relatively easy to initiate on the Meridiani Planum early (see the list above), and (B) its output is flexible and useful for basic tasks such as making more electricity generation capacity (see the previous section), water liberation (see the next section), making storage tanks, making landing pads, and making laboratories. The main three steps in source material processing are (1) low temperature, direct reduction of hematite (Fe2O3) and nickel oxide (NiO, Ni2O3) to metallic iron and nickel, (2) separation from impurities by carbonylation, (3) carbonyl decomposition to make steel powder. The carbonylation reactions producing iron carbonyl (Fe(CO) 5) and nickel carbonyl (Ni(CO)4) are as follows: Fe(s) + ImP(s) + 5CO(g) →Fe(CO)5 (vapor) + ImP(s) Ni(s) + ImP(s) + 4CO(g) →Ni(CO)4 (vapor) + ImP(s) In these equations, ImP(s) stands for solid impurities. One of the two crucial practical benefits of these carbonylation reactions is that the metal carbonyls are separated from the solid impurities. Since these carbonyls leave the reactor as vapor mixed in with the rest of the carbonylation gas (primarily an excess of CO), while the impurities leave the reactor through a separate outlet as solids. Carbonyl decomposition reactors carry out the reverse of the above equations (without the impurities present). This decomposition makes iron and nickel in a useful

© ASCE

Earth and Space 2022

powder form. (Nickel can also be vapor deposited onto molds and pellets). Decomposition takes place simultaneously with small amounts of soot production so that some carbon mixes with the iron and nickel. There are six required or recommended auxiliaries to the main processes: the electrolytic splitting of carbon dioxide and/or water to produce different streams of pure oxygen and reducing gas (some mixture of CO, H 2, CO2, and H2O); reduction reactor top-gas processing (includes dust and SO2 removal); recycling of cleaned top gases (in particular, CO 2 and H2O) to electrolysis; a dual process of extracting process heat and condensing iron carbonyl and nickel carbonyl to liquids from the carbonylation reactor exit gases; recompression of carbon monoxide exiting the decomposition reactor (so that the pressurized CO can be recycled back to the carbonylation gas); extraction of ~50oC temperature process heat during staged CO compression. Inputs of electricity are needed to drive the electrolysis units, the compressors, and the heaters used in carbonyl decomposition. The iron carbonylation reaction is the primary source of process heat for water liberation from steel-making. This reaction is exothermic, and the changes in enthalpy at various reasonable reactor conditions are around -150 kJ/molFe(CO)5. Lower temperature (under 100 oC) process heat is extracted from old industrial carbonylation reactors. However, a recent carbonylation reactor design should make 150oC to 200oC process heat extraction practical. This will be tested in 2022 and described after testing. The amount of process heat available from carbonylation steel-making will, in most scenarios, be several times less than the process heat from the receivers of power tower CPV electricity generation. WATER LIBERATION FROM BURNS FORMATION SEDIMENTS Basics The temperature of the available process heat is well suited to liberating water in the Burns Formation. Even the lowest temperature (~50 oC) process heat can dehydrate some of the forms of most common sulfate, i.e., magnesium sulfate, including MgSO 4.nH2O, for n = 11, 7, 6, 4, and even n = 3 or 2 (at the ambient pressure ~ 6 mbar) (Okhrimenko et al., 2017); while the highest temperature (~150–200oC) process heat comfortably turns all hydrated magnesium sulfates into anhydrous magnesium sulfate (Okhrimenko et al., 2017). Gypsum (CaSO 4.2H2O), bassanite (CaSO4.0.5H2O), and jarosite ((X)Fe3(SO4)2(OH)6, where X is usually K or Na) are less important because of lower hydration levels and lower abundances. Gypsum dehydrates to bassanite at around 100oC, and bassanite dehydrates to anhydrous CaSO 4 at a little under 200oC (Shen et al., 2019). Jarosite dehydrates so slowly at temperatures under 200oC (Desborough et al., 2006) that it will remain mostly unchanged when heated with the available process heat. The change in enthalpy in the dehydration of hydrated magnesium sulfates is between -53 and -57 kJ/molwater (Okhrimenko et al., 2017; Grevel et al., 2012), while for hydrated calcium sulfates, it is close to -52.5 kJ/molwater (Robie and Waldbaum, 1968). Both enthalpy changes are close to –3kJ/gwater. This enables a simple calculation to find the amount of heat of dehydration that needs to be supplied to liberate enough water to make propellant to fill a SpaceX Starship. This propellant production needs slightly less than 600 metric tonnes of water. So, ~500 MWh of process heat is required as heat of dehydration to liberate water to add to this basic propellant production task. This is much less process heat than the ~4000 MWh of process heat that one small Australian CPV plant will produce in one Earth/Mars synodic period.

© ASCE

348

Earth and Space 2022

Another feature of the Burns Formation sediments that is important for making water liberation from them practicable is that these sediments are very soft and friable (Christensen et al. 2000; Squyres et al. 2004; Grotzinger et al., 2005). This soft friability will lower the energy required to excavate and drill into the sediments. It also means that special hardened steels are not needed for drill bits and excavation buckets – these can both be made locally from the simple steels made by sintering powders from carbonylation steel-making. Water Liberation in Ovens and In-Place Water liberation from the Burns Formation needs both heat exchange from the fluid carrying the process heat to the formation’s sediments and also needs transport so that the hot fluids and sediment are bought into contact. The fluids and sediment can be bought together in ovens or inplace. The first requires moving sediment material into ovens. he second requires moving hot fluids into the sediment formations. A water liberation oven is simple in design and operation. The oven has a main compartment that is loaded with pulverized, hydrated sediment material and then closed shut. Flanged heat exchange pipes cross the interior of the main oven compartment in an arrangement designed so that the loaded sediment material makes good thermal contact with the flanged heat exchange pipes. Hot, high-pressure carbon dioxide (i.e., the carrier of process heat from either electricity generation or carbonylation) flows inside the heat exchange pipes to heat up the sediments in contact with the exterior of the flanged pipes. This hot, high-pressure gas is sealed from the interior of the main oven compartment containing the sediment material. Near ambient (low) pressure atmospheric carbon dioxide is pumped through the bed of sediment material to remove liberated water vapor from this bed. This flow of water-vapor-laden carbon dioxide is outlet from the oven’s main compartment and sent to a cooling chamber to freeze the water out of the outlet gas into frost ice. The interior of this cooling chamber can be forced to a low temperature well below 0oC by making good thermal contact between the cooling chamber and the Burns formation sediments, i.e., a large, cold, local heat sink. Note, the triple point pressure of water is 6.11 mbar, i.e., very close to the average ambient pressure on Mars, while the freezing temperatures of water at these ambient pressures are still close to 0 oC, which is hotter than the heat sink. The cooling chamber should simultaneously be a storage tank. Liquid water is occasionally removed from the storage tank by (a) closing its valves to the oven’s main compartment, (b) raising the pressure a little in the storage tank (to something like 15 mbar), (c) raising the temperature inside the tank to liquefy some ice, and (d) draining off water. One way to raise the temperature inside the storage tank is to concentrate extra sunlight onto the tank walls by aiming some of the solar heliostats for a power tower. The pressure inside the closed tank will rise simply by heating it (net sublimation to gaseous water will occur until this extra gas raises the pressure enough for liquid to form). Almost all parts of a liberation oven and storage tank can be made locally by robots from sheet metal using cutting, bending, and welding. Robots can, of course, operate both ovens and tanks. Sediment for oven liberation can be obtained early on by collecting loose, surface sediment blocks (crater ejecta from meteorite impacts). Another way to obtain sediment material early on is to excavate or drill sediments for reasons of scientific exploration beyond water liberation; for example, underground excavation to build a laboratory protected from radiation. Boring holes should precede such excavation into the sediments to record their stratigraphy, better understand the water environment deep in the sediments, and look for signs of life.

© ASCE

349

Earth and Space 2022

In-place water liberation can start with a multi-megawatt electricity and steel infrastructure established and a green light for larger-scale activity. In-place water liberation involves drilling a large cluster of closely spaced boreholes into the Burns Formation sediments. (The material excavated from these boreholes would be sent to liberation ovens.) Then, a hairpin-type, looping heat exchange pipe is inserted into each borehole, and process heat is sent down into these boreholes in flowing, pressurized carbon dioxide. The hairpin heat exchange pipes should be flanged to make good thermal contact with the sediments. An advantage of in-place water liberation is that it minimizes the amount of material excavated, transported to ovens, loaded into and unloaded from ovens, and transported back to a tailings pit. A disadvantage of in-place water liberation is that energy and some liberated water leaks radially away from the boreholes. So large clusters of boreholes need to be heated simultaneously to lower the fraction of lost heat and liberated water. CONCLUDING COMMENTS One more proposal feature is that it reduces the amount of water needed for rocket propellant production. It does this because it makes many things on Mars out of steel that no longer need transport from Earth. The paper points out the usefulness of steel on Mars. What about the steel available in the structure of the first SpaceX Starship to land on Mars? This first spacecraft steel can be reused to build the first landing pad and the first power tower CPV plant. Both are high-value reuses of the first spacecraft. Planning for such steel reuse lowers the bar on the required capability of the first landed Starship, allows more time where rocket propellant production is not the first priority, and allows more time to learn from the first landing and improve for the next landing. When it comes to extracting or liberating water on Mars, the place where it happens makes a difference. The Meridiani Planum is one of the easiest places on Mars to land. Its underlying Burns Formation binds an extensive reservoir of water. It has a relatively warm, unchanging equatorial climate. The rover Opportunity survived, traversed, and operated on this plain, on its own, without a need to hibernate from cold, for over fourteen years. In the top layer of soils spread across this plain is a bounty of crystalline grey hematite in small blueberry spheres. Robotic harvesters can pick loose blueberries up off the top of this soil. The ability to make things from hematite blueberries extends the range of feasible engineering and scientific activities. These include local manufacture of most of the equipment needed to carry out water liberation driven by process heat and local manufacture and construction of additional plants for electricity generation and steel-making. Expanding power, infrastructure, and equipment capacity will supply more water for all sorts of exploration activities. The Meridiani Planum is an excellent place to establish the first permanent exploration station on Mars. REFERENCES Arvidson, R. E., Poulet, F., Morris, R. V., Bibring, J. P., Bell III, J. F., Squyres, S.W., et al. (2006). “Nature and origin of the hematite-bearing plains of Terra Meridiani based on analyses of orbital and Mars Exploration rover data sets.” J. GeoPhys. Res.: Planets, 111, E12S08. https://doi.org/10.1029/2006JE002728.

© ASCE

350

Earth and Space 2022

Bibring, J. P., Langevin, Y., Mustard, J. F., Poulet, F., Arvidson, R. E., Glendrin, A., et al. (2006). “Global mineralogical and aqueous Mars history derived from OMEGA/Mars Express data.” Science, 312(5772), 400–404. https://doi.org/10.1126/science.1122659. Boynton, W. V., Feldman, W. C., Mitrofanov, I. G., Evans, L. G., Reedy, R. C., Squyres, S. W., et al. (2004). “The Mars Odyssey Gamma-Ray Spectrometer Instrument Suite.” Space Sci. Rev., 110 (1–2), 37. https://doi.org/10.1023/B:SPAC.0000021007.76126.15. Calvin, W. M., Shoffner, J. D., Johnson, J. R., Knoll, A. H., Pocock, J. M., Squyres, S. W., et al. (2009). “Hematite spherules at Meridiani: Results from MI, Mini-TES, and Pancam.” J. Geophys. Res., 113, E12S37. https://doi.org/10.1029/2007JE003048. Christensen, P. R., Bandfield, J. L., Clark, R. N., Edgett, K. S., Hamilton, V. E., Hoefen, T., et al. (2000). “Detection of crystalline hematite mineralization on Mars by the Thermal Emission Spectrometer: Evidence for near-surface water.” J. Geophys. Res.: Planets, 105, 9623–9642. https://doi.org/10.1029/1999JE001093. Christensen, P. R., and Ruff, S. W. (2004). “Formation of the hematite-bearing unit in Meridiani Planum: Evidence for deposition in standing water.” J. Geophys. Res.: Planets, E08003, https://doi.org/10.1029/2003JE002233. Christensen, P. R., Wyatt, M. B., Glotch, T. D., Rogers, A. D., Anwar, S., Arvidson, R. E., et al. (2004). “Mineralogy at Meridiani Planum from the Mini-TES Experiment on the Opportunity Rover,” Science, 306, (5702), 1733–1739. https://doi.org/10.1126/science.1104909. Christensen, P. R., Ruff, S. W., Fergason, R., Gorelick, N., Jakosky, B. M., Lane, M. D., et al. (2005). “Mars Exploration Rover candidate landing sites as viewed by THEMIS.” Icarus, 176(1), 12–43. https://doi.org/10.1016/j.icarus.2005.01.004. Chou, I. M., and Seal, R. R. (2007). “Magnesium and calcium sulfate stabilities and the water budget of Mars.” J. Geophys. Res., 112, E11004. https://doi.org/10.1029/ 2007JE002898. Desborough, G. A., Smith, K. S., Lowers, H. A., Swayze, G. A., Hammarstrom, J. M., Diehl, S. F., et al. (2006). “The use of synthetic jarosite as an analog for natural jarosite.” In 7th Int. Conf. on Acid Rock Drainage (ICARD), 458-474. https://doi.org/10.21000/ JASMR06020458. Diez, B., Feldman, W. C., Maurice, S., Gasnault, O., Prettyman, T. H., Mellon, M. T., Aharonson, O., and Schorghofer, N. (2008). “H layering in the top meter of Mars.” Icarus, 196(2), 409–421. https://doi.org/10.1016/j.icarus.2008.02.006. Edgett, K. S., and Parker, T. J. (1997). “Water on early Mars: Possible sub-aqueous sedimentary deposits covering ancient cratered terrain in western Arabia and Sinus Meridiani” Geophys. Res. Let., 24, 2897–2900. https://doi.org/10.1029/97GL02840. Feldman, W. C. (2003). “The global distribution of near-surface hydrogen on Mars.” In Sixth Int. Conf. Mars, Lunar and Planet. Inst., Pasadena, CA, (20–25). Feldman, W. C., Prettyman, T. H., Maurice, S., Plaut, J. J., Bish, D. L., Vaniman, D. T., et al. (2004). “Global distribution of near-surface hydrogen on Mars.” J. Geophys. Res.: Planets, 109(E9). https://doi.org/10.1029/2003JE002160. Feldman, W. C., Mellon, M. T., Gasnault, O., Diez, B., Elphic, R. C., Hagerty, J. J., et al. (2007). “Vertical distribution of hydrogen at high northern latitudes on Mars: The Mars Odyssey Neutron Spectrometer.” Geophys. Res. Let., 34(5). https://doi.org/10.1029/ 2006GL028936.

© ASCE

351

Earth and Space 2022

Feldman, W. C., Bandfield, J. L., Diez, B., Elphic, R. C., Maurice, S. and Nelli, S. M. (2008a). “North to south asymmetries in the water-equivalent hydrogen distribution at high latitudes on Mars.” J. Geophys. Res.: Planets, 113(E8). https://doi.org/10.1029/2007JE003020 Feldman, W. C., Bourke, M. C., Elphic, R. C., Maurice, S., Bandfield, J., Prettyman, T. H., Diez, B. and Lawrence, D. J. (2008b). “Hydrogen content of sand dunes within Olympia Undae.” Icarus, 196(2), 422–432. https://doi.org/10.1016/j.icarus.2007.08.044. Fenton, L. K., Michaels, T. I., and Chojnacki, M. (2015). “Late Amazonian aeolian features, gradation, wind regimes, and Sediment State in the Vicinity of the Mars Exploration Rover Opportunity, Meridiani Planum, Mars.” Aeolian Res., 16, 75–99. https://doi.org/10.1016/j.aeolia.2014.11.004. Frost, R., Wills, R. A., Kloprogge, T., and Martens, W. (2006). “Thermal decomposition of hydronium jarosite (H 3O)Fe3(SO4)2(OH)6.” J. Thermal Anal. and Calorimetry. 83(1), 213–218. https://doi.org/10.1007/s10973-005-6908-0. Golombek, M. P., Grant, J. A., Crumpler, L. S., Greeley, R., Arvidson, R. E., Bell III, J. F. et al. (2006). “Erosion rates at the Mars Exploration Rover landing sites and long-term climate change on Mars.” J. Geophys. Res.: Planets, 111, E12S10. https://doi.org/10.1029/2006JE002754. Golombek, M. P., Warner, N. H., Ganti, V., Lamb, M. P., Parker, T. J., Fergason, R. L., and Sullivan, R. (2014). “Small crater modification on Meridiani Planum and implications for erosion rates and climate change on Mars.” J. Geophys. Res.: Planets, 119, 2522–2547. https://doi.org/10.1029/2014JE004658. Grevel, K. D., Majzlan, J., Benisek, A., Dachs, E., Steiger, M., Fortes, A. D., and Marler, B. (2012). “Experimentally Determined Standard Thermodynamic Properties of Synthetic MgSO4.4H2O (Starkeyite) and MgSO 4.3H2O: A Revised Internally Consistent Thermodynamic Data Set for Magnesium Sulfate Hydrates.” Astrobiology, 12(11), 1042– 1053. https://doi.org/10.1089/ast.2012.0823. Grotzinger, J. P., Arvidson, R. E., Bell III, J. F., Calvin, W., Clark, B. C., Fike, D. A. et al. (2005). “Stratigraphy and sedimentology of a dry to wet eolian depositional system, Burns formation, Meridiani Planum, Mars.” Earth and Planetary Sci. Let., 240, 11–72. https://doi.org/10.1016/j.epsl.2005.09.039. Hartmann, W. K., and Neukum, G. (2001). “Cratering chronology and evolution of Mars.” Space Sci. Rev., 96, 165–194. https://doi.org/10.1023/A:1011945222010. Hurowitz, J. A., Fischer, W. W., Tosca, N. J., and Milliken, R. E. (2010). “Origin of acidic surface waters and the evolution of atmospheric chemistry on early Mars.” Nature Geoscience, 3, 323–326. https://doi.org/10.1038/NGEo831. Hynek, B. M., and Phillips, R.J. (2001). “Evidence for extensive denudation of the Martian highlands.” Geology, 29(5), 407–410. https://doi.org/10.1130/0091-7613(2001):029 2.0.CO;2 . Klingelhöfer, G., Morris, R. V., Bernhardt, B., Schörder, C., Rodionov, D. S., da Souza, P. A., et al., (2004). “Jarosite and Hematite at Meridiani Planum from Opportunity’s Mössbauer Spectrometer.” Science, 306 (5702), 1740–1745. https://doi.org/10.1126/ science.1104653. MER APXS Team (2016). MER APXS Derived Oxide Data Bundle. PDS Geosciences (GEO) Node. https://doi.org/10.17189/1518973.

© ASCE

352

Earth and Space 2022

MER PanCam Team (2016). MER1 Pancam Science Derived IOF Data Bundle. PDS Geosciences (GEO) Node. https://doi.org/10.17189/1519273. McLennan, S. M., Bell III, J. F., Calvin, W. M., Christensen, P. R., Clark, B. C., de Souza, P. A., et al. (2005). “Provenance and diagenesis of the evaporite-bearing Burns formation, Meridiani Planum, Mars.” Earth and Planetary Sci. Let., 240(1), 95–121. https://doi.org/10.1016/j.epsl.2005.09.041. Mellerowicz, B., Zacny, K., Palmowski, J., Bradley, B., Stolov, L., Yen, B. et al. (2021). “RedWater: Extraction of Water from Mars’ Ice Deposits.” In AIAA ASCEND 2021, 4037, Las Vegas, Nevada & Virtual. https://doi.org/10.2514/6.2021-4038. Morris, R. V., Klingelhöfer, G., Schröder, C., Rodionov, D. S., Yen, A., Ming, D. W. et al. (2006). “Mössbauer mineralogy of rock, soil, and dust at Meridiani Planum, Mars: Opportunity’s journey across sulfate-rich outcrop, basaltic sand and dust, and hematite lag deposits.” J. GeoPhys. Res. Planets, 111, E12S15. https://doi.org/10.1029/ 2006JE002791. Morris, R. V., Klingelhöfer, G., Schröder, C., Rodionov, D. S., Yen, A., Ming, D. W. et al. (2006). “Mössbauer mineralogy of rock, soil, and dust at Meridiani Planum, Mars: Opportunity’s journey across sulfate-rich outcrop, basaltic sand and dust, and hematite lag deposits.” J. GeoPhys. Res., 111, E12S15. https://doi.org/10.1029/2006JE002791. Musk, E. (2018). “Making life multi-planetary.” New Space, 6(1), 2–11. https://doi.org/ 10.1089/space.2018.29013.emu. Okhrimenko, L., Favergeon, L., Johannes, K., Kuznik, F., and Pijolat, M. (2017). “Thermodynamic study of MgSO 4—H2O system dehydration at low pressure in view of heat storage.” Thermochimica Acta, 656, 135–143. https://doi.org/10.1016/j.tca.2017.08.015. Olsen, R. M. (2021). “Iron Oxide Harvesting on Mars.” In AIAA ASCEND 2021, 4037, Las Vegas, Nevada & Virtual. https://doi.org/10.2514/6.2021-4037 Pathare, A. V., Feldman , W. C., Prettyman, T. H., and Maurice, S. (2018). “Driven by excess? Climatic implications of new global mapping of near-surface water-equivalent hydrogen on Mars.” Icarus, 301, 97–116. https://doi.org/10.1016/j.icarus.2017.09.031. Peterson, R. C., Nelson, W., Madu, B., and Shurvell, H. G. (2007). “Meridianiite: A new mineral species observed on Earth and predicted to exist on Mars.” Am. Minerol., 92(10). https://doi.org/10.2138/am.2007.2668. Prettyman, T. H., Feldman, W. C., Mellon, M. T., McKinney, G. W., Boynton, W. V., Karunatillake, S., et al. (2004). “Composition and structure of the Martian surface at high southern latitudes from neutron spectroscopy.” J. Geophys. Res.: Planets, 109(E5). https://doi.org/10.1029/2003JE002139. Prettyman, T. H., Feldman, W. C., and Titus, T. N. (2009). “Chracterization of Mars’ seasonal caps using neutron spectroscopy.” J. Geophys. Res.: Planets, 114(E8). https://doi.org/10.1029/2003JE003275. Rieder, R., Gellert, R., Anderson, R. C., Bruckner, J., Clark, B. C., Dreibus, G. et al. (2004). “Chemistry of Rocks and Soils at Meridiani Planum from the Alpha Particle X-ray Spectrometer.” Science, 306 (5702), 1746–1749. https://doi.org/10.1126/science.1104358. Robie, R. A., and Waldbaum, D. R. (1968). “Thermodynamic Properties of Minerals and Related Substances at 298.15oK (25oC) and One Atmosphere (1.013 Bars) Pressure and

© ASCE

353

Earth and Space 2022

at Higher Temperatures.” Geological Survey Bulletin 1259, U.S. Department of the Interior. Available from https://pubs.usgs.gov/bul/1259/report.pdf. Shen, L., Sippola, H., Li, X., Lindberg, D., and Taskinen, P. (2019). “Thermodynamic Modeling of Calcium Sulfate Hydrates in the CaSO 4−H2O System from 273.15 to 473.15 K with Extension to 548.15 K.” J. Chem. Eng. Data, 64, 2697–2709. https://doi.org/10.1021/acs.jced.9b00112. Soderblom, L. A., Anderson, R. C., Arvidson, R. E., Bell III, J. F., Cabrol, N. A., Calvin, W., et al. (2004). “Soils of Eagle Crater and Meridiani Planum at the Opportunity R over Landing Site.” Science, 306(5702), 1723–1726. https://doi.org/10.1126/science.1105127. Squyres, S. W., Grotzinger, J. P., Arvidson, R. E., Bell, J. F., Calvin, W., Christensen, P. R., et al. (2004). “In Situ Evidence for an Ancient Aqueous Environment at Meridiani Planum, Mars.” Science, 306(5702), pp. 1709-1714. https://doi.org/10.1126/ science.1104559. Sullivan, R., Anderson, R., Biesiadecki, J., Bond, T., and Stewart, H. (2011). “Cohesions, friction angles, and other physical properties of Martian regolith from Mars Exploration Rover wheel trenches and wheel scuffs.” J. Geophys. Res.: Planets, 116, E02006. https://doi.org//10.1029/2010JE003625. van Susante, P. J., Zacny, K., Johnson, G., and Zerbel, S. M. (2021). “Melting Ice under Martian and other Environmental Conditions for ISRU.” In AIAA ASCEND 2021, 4037, Las Vegas, Nevada & Virtual. https://doi.org/10.2514/6.2021-4036. Vaniman, D. T., and Chipera, S. J. (2006). “Transformations of Mg- and Ca-sulfate hydrates in Mars regolith.” Am. Minerol., 91(10), 1628–1642. https://doi.org/10.2138/am.2006.2092. Yen, A. S., Mittlefehldt, D. W., McLennan, S. M., Gellert, R., Bell III, J. F., McSween, H. Y., et al. (2006). “Nickel on Mars: Constraints on meteoritic material at the surface.” J. Geophys. Res., 111, E12S11. https://doi.org/10.1029/2006JE002797.

© ASCE

354

355

Earth and Space 2022

Redwater: Extraction of Water from Mars’ Ice Deposits Joseph Palmowski1; Kris Zacny2; Boleslaw Mellerowicz3; Brian Vogel4; Andrew Bocklund5; Leo Stolov6; Bernice Yen7; Dara Sabahi8; Lilly Ware9; David Faris10; Albert Ridilla11; Huey Nguyen12; Paul van Susante13; George Johnson14; Nathaniel E. Putzig15; Michael Hecht16; and Hari Nayar17 1

Honeybee Robotics, Altadena, CA Honeybee Robotics, Altadena, CA. Email: [email protected] 3 Honeybee Robotics, Altadena, CA 4 Honeybee Robotics, Altadena, CA 5 Honeybee Robotics, Altadena, CA 6 Honeybee Robotics, Altadena, CA 7 Honeybee Robotics, Altadena, CA 8 Honeybee Robotics, Altadena, CA 9 Honeybee Robotics, Altadena, CA 10 Honeybee Robotics, Altadena, CA 11 Honeybee Robotics, Altadena, CA 12 Honeybee Robotics, Altadena, CA 13 Michigan Technological Univ., Houghton, MI. Email: [email protected] 14 Michigan Technological Univ., Houghton, MI 15 Planetary Science Institute, Lakewood, CO. Email: [email protected] 16 MIT Haystack Observatory, Westford, MA. Email: [email protected] 17 Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA. Email: [email protected] 2

ABSTRACT Honeybee Robotics has designed, built, and tested a technology readiness level (TRL) 4/5 system known as RedWater, intended to drill into the surface of Mars and melt/extract water from locations identified by the Shallow Radar (SHARAD). RedWater combines proven terrestrial technologies to extract water from the subsurface Martian ice. Rodriguez Wells, or Rodwells, are a type of water well employed in Antarctica to maintain large pools of liquid water within an ice sheet and pumping water to the surface while heating and recirculating a portion to facilitate continuous well growth. RedWater also repurposes coiled tube drilling technology, which uses a thin-walled metal or composite tube to drive a bottom hole assembly into a borehole; the coiled tube itself is wound onto a drum and deployed by an injector system which transmits the required drilling forces through the tube as it is driven down. The combination of these two technologies with Honeybee’s existing rotary percussive drilling and pneumatic transport technologies make for an efficient means of producing large quantities of liquid water on Mars. Honeybee is currently working on evolving this technology to TRL 6 and will be conducting end-to-end TVAC testing in 2022. INTRODUCTION Orbital measurements from a host of spacecraft and instruments have revealed that a third of the Martian surface may contain shallow ice (see map in Figure 1). Data from the Mars

© ASCE

356

Earth and Space 2022

Reconnaissance Orbiter Shallow Radar (SHARAD) sounder have also revealed the presence of debris-covered glaciers up to 1km thick, as well as buried ice sheets in the Arcadia Planitia (30°N to 45°N, -140°E to -170°E) and Utopia Planitia (38°N to 48°N, -70°E to -130°E) that could be up to 170 m thick and nearly pure ice. Current limitations of the measuring techniques constrain the top surface of the ice to < ~20m deep, however there is a push to conduct an international Mars Ice Mapper mission which would likely result in the detection of the top of ice within a few meters below the surface. In addition to the presence of very promising volumes of subsurface ice, Arcadia is also a region located in the mid latitudes of Mars with low elevations and low rock abundance, making it a great combination of solar energy exposure and optimal landing terrain for future ISRU and manned-mission efforts [Putzig, et al 2022].

Figure 1. Map of features of relevance to interpreting special regions of Mars [Rummel, et al 2014]

Figure 2. Redwater mission architecture The discovery of these thick, nearly pure ice deposits allows for the implementation of two proven terrestrial technologies: coiled tubing (CT) for drilling and Rodriquez well or Rodwell for water extraction. CT rigs use a continuous length of tubing (metal or composite) that is flexible enough to be

© ASCE

357

Earth and Space 2022

wound on a reel and rigid enough to react drilling forces and torques. The tube is pushed downhole using an injector - a set of actuated rollers that grip the tube and advance it downward. The end of the tube has a bottom hole assembly (BHA) containing a motor and a drill bit for penetrating the subsurface. To clear cuttings, compressed air is continuously pumped down the tube and into the borehole. For the Redwater architecture, these concepts can be integrated onto a mobility platform powered by either an onboard nuclear source or a series of solar arrays, as shown in Figure 2. REDWATER OVERVIEW Redwater consists of several major subsystems: the bottom hole assembly (BHA), the coiled tubing (CT), the CT injector assembly, and the CT drum assembly. Bottom Hole Assembly (BHA). The BHA consists of four major subassemblies: heated auger assembly, auger motor assembly, heated fluid transfer line assembly, and packer assembly (see Figure 3). At the far end (opposite of the drill bit) is the BHA to CT coupler. The current overall length of the BHA is approximately 84 cm and the total mass is approximately 6 kg, but both may change slightly as the technology readiness level (TRL) 6 design is finalized. The heated auger assembly consists of a drill bit, heated auger body, and outer auger shell. The auger is heated using a fiberglass insulated nichrome-60 resistive heating wire coil. K-type thermocouples will be installed along the heated auger for monitoring and controlling temperature. The auger motor assembly consists of a mil-spec connector for electrical and mechanical connection between the heated auger assembly and auger motor assembly. Housed in this section are dual brushless DC motors and a total of three planetary gear reduction stages responsible for driving the rotation of the drill. Also housed in this section is a rotary through-hole slipring for passing electronics from the rotating heated auger assembly to the nonrotating auger motor assembly, and an inline tension/compression load cell for drilling weight-on-bit measurement and control. The heated fluid line assembly is situated on the outside of the auger motor assembly and consists of another fiberglass insulated nichrome-60 resistive heating wire coil, responsible for providing heat to the outer wall of the BHA as well as the fluid lines along the outside in case a freeze-in scenario occurs. Attached to the top of this section is the packer assembly which includes a bilayer packer membrane design, consisting of a 5 mil FEP inner bladder and a 7.5 mil outer Vectran layer coated with a TPU layer. The packer is responsible for sealing off the lower borehole from the Mars atmosphere allowing for a borehole pressure high enough for liquid water to exist.

Figure 3. BHA subassembly overview © ASCE

358

Earth and Space 2022

Coiled Tubing (CT). Inside of the coiled tubing consists of a bundle of electrical and fluid lines required at the BHA. The bundle includes pneumatic lines for supply of the gas required to clear cuttings during drilling and pressurize the borehole during melting/water extraction, hydraulic lines for water extraction from the well, and electrical wiring required for motor and heater power, as well as signals from the load cell and temperature sensors (see Figure 4). The coiled tubing itself is designed such that it can be wound along a relatively compact drum radius while still being large enough for the wiring/tube bundle to fit, and also provide the required strength for drilling weight-on-bit without the tube buckling.

Figure 4. Internal coiled tubing bundle overview CT Injector Assembly. The CT injector assembly consists of dual drive/preload rollers responsible for gripping and pushing the coiled tube downward and generating required weighton-bit forces for drilling. The drive rollers are driven by a brushless DC motor with a planetary gearhead reduction as well as an additional spur gear reduction, resulting in a total continuous output torque rating of 630 Nm. CT Drum Assembly. The drum assembly consists of a passively rotating drum wheel which the coiled tubing is wound onto, as well as a slipring and swivel combination for passing the electronics and tubing from the rotating drum to a stationary bulkhead. Figure 5 gives an overview of these subsystems.

Figure 5. CT injector and drum assemblies © ASCE

Earth and Space 2022

Performance. During the drilling phase of the con-ops, the performance of the Redwater system includes a drilling rate of penetration of 3 m/hr with a total energy requirement of 1.5 kW-hr and a total gas requirement of either 4.1 kg of nitrogen or 5.5 kg of carbon dioxide. The current design allows for up to 35 meters of fully extended coiled tubing. Assuming a rocky overburden extending 10 meters deep, this allows for well growth of up to 25 meters deep. With the current 3 kW peak melting capabilities, the system should be capable of extracting 43 metric tones of water in 300 days, with a total energy required of 21.2 MW -hrs and total gas required of 150 kg of nitrogen or 220 kg of carbon dioxide. This information is summarized in Figure 6.

Figure 6. Redwater system level overview and performance summary TESTING Base Period. End-to-end cold chamber testing was conducted in the base period of the contract using the TRL 4/5 prototype of the Redwater system. During the well formation phase of testing, the auger heater temperature set-point was 60°C and had an average thermal power output of 1155 Watts. During melting, the auger was rotated at a constant speed of 120 RPM to generate convective waves to improve melting efficiency. The total melt duration was 120 minutes and resulted in a final well volume of 12.2 liters and well dimensions of 35 cm diameter and 26 cm height. Figure 7 shows well geometry over the course of the test. TVAC testing was also conducted in the base period. The pressure inside of the chamber was maintained at approximately 5 Torr. For this test, an ice sample was prepared inside of a custom steel drum actively cooled with LN2 coils to maintain an ice temperature of approximately 60°C. Given that it was not feasible to use an uncontained ice sample for this test, it was not

© ASCE

359

Earth and Space 2022

possible to capture well geometry over time by visual observation. Instead, a method was devised that involved embedding an array of thermocouples in strategic locations throughout the ice in order to map out the growth of the well during the test. A summary of the data from the thermocouples is shown in Figure 7. To estimate the final volume of the well, the drum and ice sample were vertically cut in half after the test, exposing the refrozen well (red dye was injected into the well during testing to more precisely delineate the boundaries of the well). Horizontal cuts were also made into the ice to verify the thermocouples’ exact final locations. The well reached a final volume of 6.6 liters with a diameter of 28 cm and height of 22 cm. The total melting time was 156 minutes (the auger was rotated at 120 RPM for only the final 44 minutes, which resulted in over 50% of the total well volume). With no auger rotation, the average heater power draw was 570 Watts and with 120 RPM auger rotation, the average heater power draw was 1,300 Watts, indicating that the convective waves caused by the auger rotation are critical to increasing melting performance.

Figure 7. Horizontal cross section of the ice for verification of thermocouple locations (bottom left), CAD sketch of well to calculate volume more precisely (top left), and ice-embedded thermocouple dataset (right) Option Period. End-to-end testing is currently scheduled to commence in 2022/23. The test plan will be similar to base period testing with an initial end-to-end test planned for inside of Honeybee’s 5m tall cold chamber for the purposes of shaking out any issues with the system, and then end-to-end TVAC testing following that. TVAC testing will again target -60°C and 5 Torr pressure, as these are the expected conditions on Mars. One important change from base period testing will be the addition of thermal control of the ambient temperature via liquid nitrogen shrouds throughout the internal walls of the chamber. Figure 8 gives an overview of the TVAC test setup, as well as the model output predicting the well geometry for option period testing.

© ASCE

360

Earth and Space 2022

Figure 8. TVAC testing setup for option period testing (left), Rodwell thermal model output for well growth prediction in option period testing CONCLUSION The text presented here summarizes a feasible approach to acquire large quantities of water on Mars. End-to-end testing completed in the base period of the project validated the feasibility of combining the concepts of coiled-tube drilling and the Rodwell method of mining water from ice. The additional technology advancement and end-to-end testing in the option period of the project will further validate the robustness of the Redwater approach, this time at a TRL 6 level design and testing fidelity. From the datasets that will be acquired in the option period testing, Honeybee’s Rodwell thermal model can be validated for use and applied to future scaled versions of the Redwater architecture that pose a viable solution to collecting large quantities of Water on Mars required for future ISRU and infrastructure development missions. ACKNOWLEDGMENTS The work has been funded by the NASA BAA Next Space Technologies for Exploration Partnerships -2 (NextSTEP-2), In-Situ Resource Utilization (ISRU) Technology. The authors would like to thank the following individuals for their input and help: Hari Nayar, Julie Kleinhenz, Diane Linne, Jerry Sanders, and Brian Wilcox. REFERENCES Bramson, A.M., Byrne, S., Putzig, N.E., Sutton, S., Plaut, J.J., Brothers, T.C., Holt, J.W. (2015). Widespread Excess Ice in Arcadia Planitia, Mars. Geophys. Res. Lett. 42, doi:10.1002/2015GL064844.

© ASCE

361

Earth and Space 2022

Morgan, G.A., Putzig, N.E., Perry, M.R., Sizemore, H.G., Bramson, A.M., Petersen, E.I., Bain, Z.M., Baker, D.M.H., Mastrogiuseppe, M., Hoover, R.H., Smith, I.B., Pathare, A., Dundas, C.M., Campbell, B.A. (2021). Availability of subsurface water-ice resources in the northern mid-latitudes of Mars. Nat Astron. https://doi.org/10.1038/s41550-02001290-z Petersen, E.I., Holt, J.W., Levy, J.S. (2018). High Ice Purity of Martian Lobate Debris Aprons at the Regional Scale: Evidence from an Orbital Radar Sounding Survey in Deuteronilus and Protonilus Mensae. Geophys. Res. Lett. 45, 11,595-11,604. https://doi.org/10.1029/2018GL079759 Putzig, N.E., Morgan, G.A., Sizemore, H.G., Baker, D.M.H., Petersen, E.I., Pathare, A.V., Dundas, C.M., Bramson, A.M., Courville, S.W., Perry, M.R., Nerozzi, S., Bain, Z.M., Hoover, R.H., Campbell, B.A., Mastrogiuseppe, M., Mellon, M.T., Seu, R., Smith, I.B. (2022 in press). Ice Resource Mapping on Mars. Ch. 16 in Badescu, V., Zacny, K., BarCohen, Y. (Eds.), Handbook of Space Resources, Springer Nature Switzerland AG. Rummel, J.D., Beaty, D.W., Jones, M.A., Bakermans, C., Barlow, N.G., Boston, P.J., Chevrier, V.F., Clark, B.C., de Vera, J.P., Gough, R.V., Hallsworth, J.E., Head, J.W., Hipkin, V.J., Kieft, T.L., McEwen, A.S., Mellon, M.T., Mikucki, J.A., Nicholson, W.L., Omelon, C.R., Peterson, R., Roden, E.E., Sherwood, Lollar, B., Tanaka, K.L., Viola, D., Wray, J.J. (2014). A new analysis of Mars “Special Regions”: findings of the second MEPAG Special Regions Science Analysis Group (SR-SAG2). Astrobiology. 2014 Nov;14(11):887-968. doi: 10.1089/ast.2014.1227. PMID: 25401393. Stuurman, C.M., Osinski, G.R., Holt, J.W., Levy, J.S., Brothers, T.C., Kerrigan, M., Campbell, B.A. (2016). SHARAD detection and characterization of subsurface water ice deposits in Utopia Planitia, Mars. Geophys. Res. Lett. 43, 9484–9491. https://doi.org/10.1002/2016GL070138 Wooster, P. (2018), “SpaceX’s Plan for Mars” 21st Annual International Mars Society Convention.

© ASCE

362

363

Earth and Space 2022

Rapid Extraction of Volatiles from Excavated Icy Regolith Using a Rotary Extraction Drum C. A. Purrington1; D. Purcell2; Jon Schmit3; and B. Thrift4 1

Austere Engineering, Littleton, CO; Center for Space Resources, Colorado School of Mines, Golden, CO. Email: [email protected]; [email protected] 2 Austere Engineering, Littleton, CO; Center for Space Resources, Colorado School of Mines, Golden, CO. Email: [email protected] 3 Austere Engineering, Littleton, CO; Center for Space Resources, Colorado School of Mines, Golden, CO. Email: [email protected] 4 Austere Engineering, Littleton, CO; Center for Space Resources, Colorado School of Mines, Golden, CO. Email: [email protected] ABSTRACT The utilization of lunar water enables spacecraft fuel, water for colonization, and radiation shielding. To make use of the water, an industrial scale volatile extraction process is required. Current literature shows in situ extracting of volatiles from lunar regolith is difficult and doesn’t scale well. Supplying heat to subsurface ice is difficult due to the thermal conductivity properties of regolith. And subsurface water vapor has difficulty moving through dry regolith which has low permeability. These conditions are amplified when the ice weight content is low. This study will show, that extraction of volatiles from excavated icy regolith is simplified through the use of a rotary extraction drum. A rotational system extracts 99% of volatiles, regardless of ice weight content in as little as three hours. This study also quantifies the energy consumed during extraction at varying water concentrations of cryogenic icy regolith simulants. An industrial scaled version is discussed to show how a single rotary extraction drum is able to process 1 mT of icy regolith every three hours. INTRODUCTION Remote sensors and the LCROSS impact study have confirmed the presence of water in various quantities across lunar Permanently Shadowed Regions (PSR). To utilize water, it must first be extracted from icy regolith. The primary extraction method typically involves the application of thermal energy to an icy regolith bed, water ice is sublimated and then water captured typically by a cold trap. Extracting volatiles has proved more difficult than initially proposed. Surface heating has been shown to create a 1cm desiccated layer of regolith (Sowers et al., 2020). Desiccated regolith has an effective thermal conductivity of ~0.003 W/mK (Cremers & Hsia, 1974). This makes supplying thermal energy below the surface difficult. To work around this issue multiple studies have suggested the use of heat rods or heated drills, (Brisset et al., 2020; Sowers et al., 2020). In thermal mining studies this has shown to desiccate regolith in a 1cm radius around the thermal probes. Resulting in the same issue of supplying heat to areas beyond the desiccated layer. The thermal extraction study (Brisset, 2021) has shown that 25 heated drills per square meter improve the yield over surface heating but still leave behind 40% of the volatiles. Thermal mining studies on icy regolith with concentrations of water weight percentages (wt%) of 2wt% or less, yielded negligible escaped water (Sowers et al., 2020). The thermal

© ASCE

Earth and Space 2022

extraction models were done on 1wt% icy regolith, and show nearly 80% of the ice is left behind. The LCROSS impact study (Colaprete et al., 2010), showed a water concentration of 5.6wt%. It is plausible there are large areas of the moons Permanently Shadowed Regions (PSR) with concentrations between 0-5.6wt%. This restricts the above thermal mining methods to ores that have a high concentration of water to at least (>3wt%). One of the touted benefits of thermal mining is little to no excavation and thus less maintenance. But these two studies have shown that if the near surface is desiccated and volatile extraction stops. To gain access to volatiles below the desiccate layer, excavation is required. Both studies have shown improved yields with heat rods or heated drills but drilling is a form of excavation. And a drill density of 25 per square meter is a high quantity of drilling holes. Neither study quantifies the energy required to place heated rods and drills. Both of the thermal mining studies show volatile sublimation rates are time consuming processes. The initial volatile sublimation rates start out high and then gradually fall to zero over time. Specifically in the thermal mining studying (Sowers et al., 2020), with a feedstock of icy regolith at 12wt%, half of the ice, 3.25g was captured in the first 5.0 hours, while the next 20 hours only yielded, 3.45g (Sowers, 2020). The thermal extraction study, (Brisset et al., 2020) has shown that maximum ice yields took multiple days to achieve. The highest yield of 69.71 kg, took 51.3 hours to achieve. Which yields an average of 32.7 kg/day with a feed stock of 10wt% under the best-case conditions. The thermal extraction study showed some of the subsurface ices have reached a high enough temperature to sublimate, however, the water vapor doesn’t completely escape. This is due to some of the water vapor migrating down where icy regolith is still cold and redepositing. And due to the low permeability of dry regolith which is shown to have a value of ~1 − 7x10−12 m2 (LaMarche et al., 2011). So as the desiccated overburdened layer grows on the surface the ability for water vapor to escape is increasingly restricted. These studies show that delivering heat below the surface is difficult due to the properties of dry regolith. As the dry layer grows, the overburn chokes the path for volatiles vapor to escape. Even with the addition of heated rods or drills, the overall extraction efficiency never improved beyond 60%. This study describes a system that works around all of the difficulties described above. The Rotary Extraction Drum (RED), is essentially a drum, rotating on its side. The drum contains fins to help mix regolith and provides large surface areas to heat icy regolith. In addition, the turning of subsurface ice to the surface and large fin surface areas allow a path for sublimated volatiles to escape. A slow rotation rate allows volatiles near the surface to escape and to keep dust kickup low. With a 30% fill volume, we also show that for every 150 degrees of rotation the sample is turned over. Since, this study is proposing excavation as a requirement we also compared a rotary system to a static system without applying thermal energy, to show the benefits of aggravating the regolith bed. Finally, we performed extraction tests on cryogenic icy regolith simulants with varying amounts of water concentrations to prove this technology is viable in the lunar PSR conditions. TEST METHODOLOGY This study includes multiple experiments to explore the affects of rotary motion in volatile extraction. The first experiment conducted is to visualize the granular affects inside of drum with fins. The second experiments compare static vs rotary system extraction in a similar vacuum

© ASCE

364

365

Earth and Space 2022

chamber environment, without heat applied. The fourth experiment finally uses cryogenic icy regolith in a vacuum with heat application to prove the applicability in a lunar PSR environment. Grain Mixing Apparatus. The first apparatus was built to visualize the grain mechanics inside of drum. An aluminum drum 7.6cm deep, with diameter of 20.3cm, was fitted with 2.54cm internal fins as shown in Figure 1. A plexiglass front plate was added to allow visual observations. Colored sand of equal mass, 250g each layer, was added incrementally until the volume was approximately 30% full. The colored sand had a grain size between 500um to 1000um. The drum is then rotated slowly and visually observations were made.

Figure 1. Visually drum apparatus, to view granular mechanics Comparing static vs rotary motion without the addition of heat. The second experiment included two separate apparatuses. A static container 8.89cm deep and 8.89 cm in diameter, was made from 6.35mm thick wall cylinder, with a 10.2cm square base plate, with a thickness of 6.35mm. The Rotary Extraction Drum (RED) apparatus is a 17.0cm diameter and 18.4cm deep drum. With 2.54cm internal fins added every 30 degrees. An aluminum front plate, was secured to the front. The front plate also includes a 4.5cm hole to allow water vapor to escape, shown in Figure 2. Each test was conducted with the same amount of water and regolith, prepared at room temperature. The simulant type was Lunar Highlands Type (LHT) from the Colorado School of Mines (Mines, 2021). LHT is modeled after Highlands regolith from the Apollo 16 landing site. It contains a similar grain distribution and is 99% chemically similar to the Apollo 16 samples (Mines, 2021). Room temperature LHT regolith of 775 grams was mixed with 86 grams of room temperature water. Once, thoroughly mixed the water regolith was placed in one of the containers. A single container was moved to the vacuum chamber. The RED apparatus was rotated 2 degrees every 20 seconds. While static container was left to rest in the chamber. The test was started as soon the vacuum pump was turned on. For this series of experiments no heat was applied. After a specified amount of time the test was terminated. The regolith mixture was removed and a mass measurement taken. Then the regolith mixture was baked at 450K for 2 hours and another mass measurement was taken to determine the quantity of water remaining in the system.

© ASCE

Earth and Space 2022

Figure 2. Static and RED apparatus. 1. Rotary Extraction Drum, 2. Aluminum front plate, 3. Pololu 24DC Motor, 4. Internal Fins, 5. Forward Rollers, 6. Static Apparatus. Not to scale. Cryogenic Icy Regolith Simulant Tests. During the final series of extraction tests, the RED system used a 250W halogen bulb as a heat source. The halogen bulb is mounted and protrudes through the front plate as seen in Figure 3. The RED apparatus rotates around it.

Figure 3. Same RED Apparatus from figure 2, with a halogen bulb protruding through the front plate. 1. Halogen Bulb, 2. Bulb Mount Three extraction tests were conducted on cryogenic icy regolith simulants. All the tests had varying amounts of total ice, 1wt%, 5.7wt% and 10wt%. The total amount of icy regolith for each test was approximately ~864g. With varying mounts of regolith and ice to achieve the same total, as seen in Table 1.

© ASCE

366

367

Earth and Space 2022

Table 1. Cryogenic Icy Regolith Simulant Mass Quantities Water Weight % 1.0 5.7 10.0

Input Regolith (g) 853.76 818.63 776.16

Input Ice (g) 8.64 46.88 86.24

Total Icy Regolith (g) 864.2 865.5 864.2

The LHT simulant regolith, was baked at 450K for two hours. Then sealed and moved to a freezer to chill to 250K. Inside of the freezer ice was disaggregated and sifted to a particle size less than 500 μm. The LHT simulant and ice grains were mixed and placed in container, that rested in bath of LN. After 2 hours the temperature was confirmed to be less than 80K. The Rotary Extraction Drum (RED) container also chilled to 255K. It was not possible to cool the RED container to cryogenic temperatures, as the logistics of handling it during entry into vacuum chamber was not conducive for personal safety. In less than a minute the cryogenic icy regolith was placed in the RED container and moved to a vacuum chamber and the vacuum pump turned on immediately. Once the pressure fell below 133 Pa (1 Torr), the test was considered started and heat was applied. The container wall temperature was monitored and the 250W bulb electrical power was adjusted to maintain a temperature of 325K ±5K at the RED wall. During extraction the pressure increased and the test was terminated once the pressure fell 133 Pa (1 Torr) for at least 15 minutes. The remaining regolith simulant was removed and mass measurements taken. The regolith was then baked at 450K for two hours and a final mass measurement taken again to determine the amount of ice remaining in the regolith. Thermal power input was recorded using a KW47-US energy monitor specifically recording the energy consumption used. While the rotary power input was near constant at 1W. Temperature of the regolith bed test was observed using a, HIKMICRO B1L 160 x 120 IR Resolution Thermal Imaging Camera. RESULTS Rotary Mixing of Grains. The first experiment observed the mixing of color sand in a rotary drum shown in Figure 4. The first image is the first 150 degrees of rotary motion.

Figure 4. Grain mechanics of colored sand in the first 150 degrees of rotation. The next images included in Figure 5, are stopped at various angles to provide a visual of how grains are mixed during a full revolution.

© ASCE

Earth and Space 2022

Figure 5. Granular Mechanics of colored sand in rotary drum, shown are different degrees of rotation. 1. 0°, 2. 150°, 3. 300°, 4. 360° Comparing Static vs Rotary without thermal energy. The data in Figure 6, shows the remaining ice percentage in each sample after the test was terminated at different time intervals. This set of experiments did not have any heat applied. The test sample was started at room temperature and placed under vacuum.

Figure 6. Comparing Static vs Rotary Extraction Drum of ice remaining in samples without adding thermal energy Cryogenic Rotary Test with Heat. Three cryogenic rotary tests were conducted. The following Table 2, contains relevant data specific to the extraction of water and the process rate of icy regolith. While the Figure 7, contains the chamber readings.

© ASCE

368

369

Earth and Space 2022

Table 2. Cryogenic RED extraction results Wt %

Input Ice (g)

Extracted Ice (g)

Extraction Time (hrs)

Extraction Efficiency (%)

Icy Regolith Processing Rate (g/hr)

Ice Extraction Rate (g/hr)

1.0 5.7 10.0

8.64 47.02 86.24

8.64 45.96 86.24

1.16 2.0 3.0

99 98 99

749 431 287

7.5 23.4 28.7

RED - Wt 1% Wall Temp, Pressure, Power Input 350

2.5

2

Wall Temp(K) Power(W)

250

Press(Torr)

1.5

200 150

1

Pressure (Torr)

Temperature (K) and Power (W)

300

100 0.5 50 0

0 0

10

20

30

40

50

60

70

Time(min)

Figure 7. Cryogenic Test, Rotary Extraction Drum with 250 W Bulb, Temperature is RED wall temperature, Power is bulb electrical input, Pressure is the chamber pressure. Energy observations were made during cryogenic test and recorded in Table 3 below. Table 3 Energy Consumed during each Cryogenic Test Wt%

Input Thermal Energy (kWh)

Rotational Energy (kWh)

Energy per Icy Regolith (kWh/kg)

Energy per Extracted Ice (kWh/kg)

1.0 5.7 10.0

0.061 0.109 0.125

0.0016 0.002 0.003

0.072 0.128 0.148

7.53 2.42 1.48

Finally, to verify the internal temperature of the regolith a thermal camera was used to take temperature measurements immediately after a test was terminated and the vacuum chamber opened. © ASCE

Earth and Space 2022

Figure 8. Thermal Camera of Regolith immediately after opening the vacuum chamber. Temperature values in image are in F. Max: (137.1 F) 331.5K Min: (112.8 F) 318K, Cen: (129.3) 327.2K DISCUSSION Visual Drum. The visual drum in Figure 1, is method that can be used to visualize how grains will intermix inside of a drum. The drum was filled to approximately 30% of its volume. This fill rate is an estimate to maximize icy regolith processing but still provide enough room for intermixing of grains. The grains shown in Figure 1, 4, and 5 are not the same grain size as the lunar simulant. All of the colored sand was 500μm to 1000μm, while the bulk of the LHT sample was 500μm or less. In addition, the colored sand had no observed cohesion. While, the LHT simulant was observed to be cohesive and fell or cascaded in clumps as it moved inside of the drum. The visuals in Figure 4 and Figure 5, still provide insight into how grains may mix when filled to the same volume percentage. For example, it can be seen in the Figures 1, 4 and 5, the surface layer would theoretically have the highest rate of sublimation as it is nearest to the surface. Since this layer is actively sublimating ice, it is also losing the most heat due to the heat of sublimation of water. As the drum rotates, the top layer would desiccate and fall down and to the left, into area between fins. The container wall is heated, and this recently chilled regolith would now start to gain heat from the wall. This eventually leads to uniform temperature in the bed. It can also be observed in Figure 4, that every ~150 degrees the entire regolith bed has been turned. So even if ice still remains in the sample, it will have a chance to return to the surface every 150 degrees of rotation. Figure 5, also shows the regolith bed becomes intermixed as the drum has more and more rotations. This is good indication that the entire volume of regolith eventually has some exposure near the surface. Static vs Rotary. Studying the differences between a static system and rotary system is important. As the RED system requires excavated icy regolith. It was plausible that a static system may be sufficient for volatile extraction. However, it is obvious from Figure 6, that a rotary system even without thermal energy being applied allows a nearly 50% increase in total extraction efficiency when compared to the

© ASCE

370

Earth and Space 2022

static system. Posttest the static systems were shown to have portions of the regolith bed to be a solid block of icy regolith, with thermal measurements of ~250K. The water in the regolith froze, possibly closing off pore space preventing water vapor escaping. In contrast, posttest observations from the rotary system, yielded a regolith bed that was clumpy, but disaggregated and measured around ~250K. The cooling observed in both systems is due to the sublimation of water. As the temperature dropped, the vapor pressure of water and effective sublimation rate also drops, (Andreas, 2007). Even though the temperatures were similar posttest, the overburden on the static system likely restricted further sublimation. A rotary system disaggregates and intermixes the regolith, as can be seen in Figure 4 and Figure 5. This allows low temperature icy regolith that makes its way to surface a chance to sublimate more ice. After the 24-hour test, the static sample was poured out to perform the mass measurements. At the center of the sample was a ball of icy regolith as can be seen in Figure 9. The center of a static system is the last to lose its ice. While ice observed in rotary systems posttest were intermixed throughout the bed. If these different systems were scaled up to a more meaningful size such as a cubic meter of material, the heat transfer and overburden problem would increase for a static system.

Figure 9. Ice ball at the center of the sample after a 24-hour static test This data proves that a rotary system is a more efficient extraction structure that yields nearly twice the amount of ice. This is due to the turning of medium that is required to turn the dry overburden into the regolith bed and allows volatiles to come near the surface. Cryogenic Test. Three extraction cryogenic icy regolith tests were conducted. And as can be seen in Table 2, the extraction efficiency is noticeable higher at ≥98%, than all other forms of currently published extraction technologies. It is shown in this study that a drum without thermal energy is already an improvement over a static system. The ability to turn the dry overburden below the surface and pull subsurface ice up, doubles the extraction efficiency as seen in Figure 6. However, when starting at cryogenic temperatures the addition of heat is required to reach sublimation temperatures and then continual heating is required to maintain sublimation temperatures due heat loss from sublimating volatiles as seen in Figure 7. The continual turning of the regolith during this process allows for the icy regolith bed to reach a uniform temperature ensuring all of the volatiles reach sublimation conditions. It is possible to further improve extraction efficiency but it may not be worth the additional time and equipment. As can be seen in Table 2, nearly all of the volatiles were exhausted, with

© ASCE

371

Earth and Space 2022

the lowest extraction efficiency at 98%. It’s plausible by extending the 5.7wt% test in Table 2 by 15 more minutes, the extraction efficiency would have increased to 99%. Using a mass scale with an accuracy of ±0.1g, we have shown that all measurable volatiles have been exhausted in the 1wt% and 10wt% tests. However, it is impossible to get truly 100% of all volatiles. The remaining volatiles would be a function of the vapor pressure within the drum. This experiment used a roughing pump to remove vaporized volatiles and once the pressure fell below 1 Torr, the test was allowed to continue for 15 minutes and then terminated. In Figure 7, the final pressure was 0.3 Torr, using the ideal gas law, this equates to roughly 0.06 grams of remaining water that exists as vapor in the entire vacuum chamber of volume 0.25𝑚3 , the drum was much smaller sitting inside the chamber. It is possible the gas within our chamber is not 100% water vapor but a mixture of remaining air that was present when the chamber was open. So, the true value of remaining ice could be lower. It is possible to improve the extraction efficiency using a turbo pump or simply waiting a longer period of time. However, when used at an industrial scale, this may not be worth the additional time or the addition of another pump. As the gains are below 0.1g of volatiles. Energy Consumption. The electrical energy consumed during each test has been recorded in Table 3. While expected, it can be seen that 98% of the energy was consumed was used to heat the drum, the regolith, the ice and for the phase transition of ice water to vapor. While the remaining energy was used to rotate the drum. The total energy consumed per kg of ice is also noted in the Table 3. And as can been seen, the lower weight percentages are less efficient than higher weight percentages. In all tests energy must be consumed to heat the drum and regolith, while ice mass is the primary difference. It’s reasonable to conclude that more energy is required per kg of ice with a lower wt% feedstock. The energy consumption numbers in Table 3 are useful for a discussion of general energy trends in an extraction system but these numbers likely won’t scale well to industrial proportions. In the this study the drum dimensions are rather small and thus the mass of the drum 2.6kg, outweighed the mass of the icy regolith, 0.864 kg. While an industrial scale steel drums with a volume of 5 𝑚3 , may have a mass around 500kg, but is capable of holding 1000kg of icy regolith. This drum to icy regolith mass ratio will change the total energy consumed at scale. In addition, the LCROSS study, (Colaprete et al., 2010), showed that other volatiles exist besides water. The other volatiles are present in quantities and sublimate at different rates. Regardless, this study still provides a rough order of magnitude of energy consumption per kg of extracted ice. It’s also reasonable to conclude a RED system located on a sunlit region of the moon, could leverage radiation energy from the sun. Since, 98% of the energy consumed was just heat applied to the drum, this technique could dramatically reduce the overall energy demands of volatile extraction. FUTURE WORK The next step is to add additional volatiles, such as carbon dioxide ice mixed in with water ice to simulate extracting different volatiles at different temperatures. This would allow targeting specific volatiles based on temperature and then capture them in different containers. Providing a process that creates products with higher concentrations of target volatiles.

© ASCE

372

Earth and Space 2022

In addition, the team hopes to show case this apparatus at an industrial scale through a NASA SBIR. The team would like demonstrate this technology on a larger drum with a volume of 5𝑚3 . This would provide the capability to process 1mT of icy regolith every hour. Numerous other items are currently being reworked and remodeled to optimize the RED apparatus, fin shapes, drum and fin materials, and drum rotation rates. It is possible to improve extraction rate by increasing mixing quality within the drum and improving heating rates. CONCLUSION It has been shown in previous studies, that removing volatiles using surface heating and heat rods or heated drills are insufficient to exhaust all the volatiles in a given area. And to some degree also require some forms of excavation. These processes also requires multiple days to reach maximum water yield and are not as effective on low wt% volatile ores. This study has shown through experimentally results that a Rotary Extraction Drum nearly exhausts all of the volatiles from its feed stock. This is also accomplished by comparison in a relative short amount of time of no more than 3 hours, not days. If the rotary apparatus is appropriately scaled up to a drum with a 1.5m diameter and 2.85m of depth, with a volume of 5m3 the apparatus would be able to process 1mT of icy regolith every hour. The most important impact this study has on water mining is the ability to exhaust volatiles from low water weight percentage feedstocks. Previous studies show poor extraction efficiency rates of 20% or less on icy regolith with 0.5, Arcadia Planitia covers the greatest extent, followed by Utopia Planitia, Deuteronilus Mensae (Fig. 4), and Eastern Hellas Basin (too sparse to see at this scale). The Arcadia Planitia region had the highest ice consistency scores, and is thought to be indicative of subsurface ice sheets of up to 75% purity 10s of meters thick (Bramson et al., 2015). The Utopia Planitia region includes layered mesas containing subsurface ice up to 85% purity ~ 80-170 m thick, as identified by a previous study (Stuurman et al., 2016). However, a recent study has found high radar loss properties that call into question the validity of the ice purities in Arcadia and Utopia Planitia (Campbell & Morgan, 2018). In addition, the SWIM study found higher radar dielectric constants indicative of lower ice concentrations in these areas. The Eastern Hellas region has less extensive signatures of subsurface ice, but does contain CCF, LVF, and LDA (Holt et al., 2008; Levy et al., 2014) and the SWIM project found high ice consistency (Putzig & Morgan et al., 2022). The region has been extensively targeted by SHARAD, and would be a prime candidate for 3D radar processing to build further confidence in the volume and structure of icy landforms present (Bain et al., 2022). The Deuteronilus and Protonilus Mensae regions include CCF, LVF, and LDA, and represent the most substantial volume of martian midlatitude ice with purity levels > 80% and thicknesses of hundreds of meters (Peterson et al., 2021). A maximum of ~ 20m of supraglacial overburden covers much of these massive ice bodies (Baker et al., 2019). The recent completion of the Deuteronilus Mensae 3D (DM3D) radargram, publicly available at https://sharad.psi.edu/3D, shows that areas of ice up to 1,450 m ice exist (Russell et al., 2021). This 3D radargram represents the state of the art and the highest fidelity product in exploring for deep subsurface ice at the midlatitudes.

© ASCE

389

Earth and Space 2022

Figure 4. The increase in detectability of subsurface ice using 3D radar processing of SHARAD data in Deuteronilus Mensae. The right image shows LDA mapped using standard 2D radar analysis. The left image shows that ice is more detectable using 3D radar analysis at the same location. The colorbar shows the thickness of the ice assuming a dielectric constant of 3.15 (pure water ice). Modified from: Russell et al., 2021. DISCUSSIONS The SWIM 2.0 products indicate the consistency of massive subsurface ice in the regions of Arcadia Planitia, Utopia Planitia, Deuteronilus Mensae, Protonilus Mensae, and Eastern Hellas Basin. The relative thinness of the ice and its questionable purity in Arcadia Planitia makes this region less desirable for deploying the RedWater system. While the ice reported in the Utopia Planitia region is ~ 100m thick and may meet that aspect of the system’s requirements, the question of its purity remains, and does not instill confidence in the site being suitable for RedWater. Although the Protonilus Mensae region likely contains vast quantities of relatively pure ice, the roughness properties preclude radar detection of the base of the ice in the majority of the region (Petersen & Holt, 2021). Thus the confidence in the presence and purity of the ice in this region is reduced. Eastern Hellas Basin is a possible target for the RedWater system, as it contains 100s of meters thick subsurface ice. Further imaging from the MRO High Resolution Imaging Science Experiment (HiRISE), and 3D radar processing will further improve the confidence of successfully deploying the RedWater system to this region. The high purity of the glacial landforms in the Deuteronilus region as evidenced by the high strength of the basal radar reflections and the thickness of the ice measured by the highest fidelity product available (i.e., using the 3D radargram) points to the LDA of this region as a primary target for sending the RedWater system. The study of LDA in the DM3D radargram mapped an area of 31,000 km 2 of subsurface ice (Bain et al., 2022). Imagery from (HiRISE) covers ~ 25% of this area, and is mostly constrained to the periphery of the LDA, which was targeted to characterize their glacial morphologies and origins (Fig. 5). These portions of the LDA are likely rougher than the interior areas, potentially introducing a bias in the assessment of traversability and EDL capabilities with respect to current technologies. An increase in the distribution and coverage of high-resolution imagery across the LDA may result in the discovery of areas of greater feasibility of the traversibilty and EDL accessibility in Deuteronilus Mensae.

© ASCE

390

Earth and Space 2022

Figure 5. HiRISE coverage of Deuteronilus Mensae. The upper left image shows the DM3D study area outlined in black with HiRISE coverage outlines (green) over glacial landforms (blue). It is apparent that the majority of HiRISE images cover the periphery of LDA. The upper right image is a zoomed in view (red box in the upper left image) of an LDA with rendered HiRISE images (IDs are labeled on top of each image) over an area midway between a central massif and the periphery. The lower left and lower right images are zoomed in views of the magenta and yellow boxes, respectively. Both lower images show examples of surface textures indicative of subsurface ice. CONCLUSION The LDA in the Deuteronilus Mensae region are the prime candidates for the RedWater system. Prior studies have demonstrated that the LDA in this region are high-purity ice glaciers that are 100s of meters thick with gently sloping surfaces and that are located equatorward of 50°N latitude. Targeting LDA areas midway between the central massifs and the periphery for © ASCE

391

Earth and Space 2022

HiRise imagery and stereo pairs, for use in DTMs, should be a priority to further analyze the EDL and traversability constraints. A successful deployment of the RedWater system to one of these glacial landforms will demonstrate the extractability of water ice and constrain the scalability and maturation of ice resources to ice reserves. REFERENCES: Bain, Z. M., Perry, M. R., Russell, A. T., Foss, F. J., Putzig, N. E., & Campbell, B. A. (2022). Enhanced subsurface analysis of the martian cryosphere afforded by threedimensional radar imaging. Lunar Planet Sci. 53. Baker, D. M., Head, J. W., & Marchant, D. R. (2010). Flow patterns of lobate debris aprons and lineated valley fill north of Ismeniae Fossae, Mars: Evidence for extensive midlatitude glaciation in the Late Amazonian. Icarus, 207(1), 186-209. Baker, D. M., & Carter, L. M. (2019). Probing supraglacial debris on Mars 1: Sources, thickness, and stratigraphy. Icarus, 319, 745-769. Bramson, A. M., Byrne, S., Putzig, N. E., Sutton, S., Plaut, J. J., Brothers, T. C., & Holt, J. W. (2015). Widespread excess ice in Arcadia Planitia, Mars. Geophysical Research Letters, 42(16), 6566-6574. Bussey, B., & Davis, R. R. (2015). First Landing Site/Exploration Zone Workshop for Human Mission to the Surface of Mars: October 27–30, 2015, Houston, Texas. Campbell, B. A., & Morgan, G. A. (2018). Fine‐scale layering of Mars polar deposits and signatures of ice content in nonpolar material from multiband SHARAD data processing. Geophysical Research Letters, 45(4), 1759-1766. Head, J. W., Mustard, J. F., Kreslavsky, M. A., Milliken, R. E., & Marchant, D. R. (2003). Recent ice ages on Mars. Nature, 426(6968), 797-802. Holt, J. W., Safaeinili, A., Plaut, J. J., Head, J. W., Phillips, R. J., Seu, R., ... & Gim, Y. (2008). Radar sounding evidence for buried glaciers in the southern mid-latitudes of Mars. Science, 322(5905), 1235-1238. Levy, J. S., Head, J. W., & Marchant, D. R. (2009). Concentric crater fill in Utopia Planiti a: History and interaction between glacial “brain terrain” and periglacial mantle processes. Icarus, 202(2), 462-476. Levy, J. S., Fassett, C. I., Head, J. W., Schwartz, C., & Watters, J. L. (2014). Sequestered glacial ice contribution to the global Martian water budget: Geometric constraints on the volume of remnant, midlatitude debris‐covered glaciers. Journal of Geophysical Research: Planets, 119(10), 2188-2196.. Mank, Z. D., Zacny, K. A., Sabahi, D., Buchbinder, M. J., Bradley, B. C., Stolov, L. A., ... & Kleinhenz, J. (2021). RedWater: A Rodwell system to extract water from Martian ice deposits. In Earth and Space 2021 (pp. 471-480). Mellon, M. T., Arvidson, R. E., Sizemore, H. G., Searls, M. L., Blaney, D. L., Cull, S., Hecht, M. H., Heet, T. L., Keller, H. U., Lemmon, M. T., Markiewicz, W. J., Ming, D. W., Morris, R. V., Pike, W. T., Zent, A. P., (2009). Ground ice at the Phoenix Landing Site: Stability state and origin. J. Geophys. Res. 114, E00E07. https://doi.org/10.1029/2009JE003417 Morgan, G. A., Putzig, N. E., Perry, M. R., Sizemore, H. G., Bramson, A. M., Petersen, E. I., ... & Campbell, B. A. (2021). Availability of subsurface water-ice resources in the northern mid-latitudes of Mars. Nature Astronomy, 5(3), 230-236.

© ASCE

392

Earth and Space 2022

Petersen, E. I., Holt, J. W., & Levy, J. S. (2018). High ice purity of Martian lobate debris aprons at the regional scale: evidence from an orbital radar sounding survey in Deuteronilus and Protonilus Mensae. Geophysical Research Letters, 45(21), 11-595. Petersen, E. I., & Holt, J. W. (2021). Surface Roughness Prevents Radar Penetration of Some Martian Debris-Covered Glaciers. IEEE Transactions on Geoscience and Remote Sensing. Plaut, J. J., Safaeinili, A., Holt, J. W., Phillips, R. J., Head, J. W., Seu, R., ... & Frigeri, A. (2009). Radar evidence for ice in lobate debris aprons in the mid‐northern latitudes of Mars. Geophysical research letters, 36(2). Putzig, N. E., Morgan, G. A., Sizemore, H. G., Baker, D. M. H., Petersen, E. I., Pathare, A. V., Dundas, C. M., Bramson, A. M., Courville, S. W., Perry, M. R., Nerozzi, S., Bain, Z. M., Hoover, R. H., Campbell, B. A., Mastrogiuseppe, M., Mellon, M. T., Seu, R., Smith, I. B., 2022 (in press). Ice Resource Mapping on Mars. Ch. 16 in Badescu, V., Zacny, K., Bar-Cohen, Y. (Eds.), Handbook of Space Resources, Springer Nature Switzerland AG. Russell, A. T., Bain, Z. M., Perry, M., Putzig, N. E., Foss, F. J., Morgan, G. A., ... & Holt, J. W. (2021). Analysis of water ice volume from 3D radar imaging of lobate debris aprons on Mars. In AGU Fall Meeting 2021. AGU. Stuurman, C. M., Osinski, G. R., Holt, J. W., Levy, J. S., Brothers, T. C., Kerrigan, M., & Campbell, B. A. (2016). SHARAD detection and characterization of subsurface water ice deposits in Utopia Planitia, Mars. Geophysical Research Letters, 43(18), 9484-9491.

© ASCE

393

Earth and Space 2022

ISRU Pilot Excavator: Bucket Drum Scaling Experimental Results Jason Schuler1; Andrew Nick2; Kurt Leucht3; Austin Langton4; and Drew Smith5 1

Exploration Systems and Development Office, NASA Kennedy Space Center, FL. Email: [email protected] 2 Exploration Systems and Development Office, NASA Kennedy Space Center, FL. Email: [email protected] 3 Exploration Systems and Operations Division, NASA Kennedy Space Center, FL. Email: [email protected] 4 Exploration Systems and Development Office, NASA Kennedy Space Center, FL. Email: [email protected] 5 Exploration Systems and Development Office, NASA Kennedy Space Center, FL. Email: [email protected] ABSTRACT NASA’s Space Technology Mission Directorate (STMD) is funding the development of a robotic excavator called the “ISRU Pilot Excavator” (IPEx) which will be a technology demonstration of excavating and transporting 10 metric tons of lunar regolith on the surface of the Moon with a 30 kg-class robotic excavator. IPEx will be the next generation of robotic excavators to use bucket drums as excavation tools. This is an evolution of the regolith advanced surface systems operations robot (RASSOR) developed at NASA’s Kennedy Space Center (KSC). Bucket drums are hollow cylinders with regularly spaced scoops around the perimeter. The drums rotate in one direction to collect regolith with the scoops. The regolith slides down an internal baffling system inside the drum which prevents the regolith from falling back out of the scoops. The captured regolith can then be transported while held in the drum and then deposited by rotating the drum in the opposite direction allowing the regolith to slide back down the baffling and out of the excavation scoops. Bucket drums were developed by Lockheed Martin in 2008 and used on multiple robotic excavator prototypes ever since. However, the forces on a bucket drum and considerations for scaling have not been measured in detail. Bucket drums are challenging to model using classical blade\bucket equations because of their unique geometry. Therefore, this experiment was performed to measure the forces on three bucket drums of the same geometry at different scales. Small: 9.4” (239 mm) dia. × 8.1” (206 mm) width, medium: 11.6” (294 mm) dia. × 10” (254 mm) width, and large: 17” (432 mm) dia. × 14.1” (358 mm) width. The test stand consisted of an actuated gantry with controlled motion in the vertical (Z) and horizontal (X) axes and a single rotation axis (R). The bucket drums were individually mounted to the rotary axis of the test stand and translated across a prepared bed of B lack Point 1 (BP-1) lunar regolith simulant at a specified linear speed and cutting depth. The test stand was outfitted with a torque sensor in line with the rotation of the drum (R) and a 3 axis (X, Y, and Z) load cell. In addition to the three sizes of bucket drums the linear excavation speed and cutting depth were test variables. The results of these experiments show the relationship between the three scales of bucket drums for factors such as: excavation force, torque due to regolith rotation inside the drum, excavation energy, time to fill, etc. and will be discussed in detail in this paper. This fundamental data will be used in the design of IPEx and can inform the design of future bucket drum excavators.

© ASCE

394

395

Earth and Space 2022

INTRODUCTION The In-Situ Resource Utilization (ISRU) Pilot Excavator (IPEx) Project will develop a 30 kgclass excavator to demonstrate robotic excavation of large amounts (10,000 kg) of lunar regolith. IPEx uses novel excavation tools, called bucket drums (Clark et al (2009)), which are hollow cylinders with scoops staggered around the outside. Regolith is collected with the scoops and flows into the drum where it is captured due to an internal baffle system. The excavator can then transport the regolith in the drum and reverse the direction of the drum rotation to dispense the regolith back out. IPEx uses sets of bucket drums that dig simultaneously in opposing directions Figure 1. Bucket Drum partial (see Figure 2). This combination of bucket drum section view showing internals excavation tools and counter-acting excavation forces (Dickson et al (2016)) enables low mass robotic excavators to effectively dig in reduced gravity environments. This is a significant departure from terrestrial excavators that rely on high mass and weights to produce tractive forces to counteract the forces of excavation. To date, NASA has only excavated tens of kilograms of lunar regolith. Excavation has never been performed by a dedicated excavation technology machine/robot, rather only as a secondary function of an exploration rover or by an astronaut using scooping/sampling methods. IPEx will be NASA’s first lunar surface robot specifically designed with the reliability and efficiency to excavate large quantities of regolith. This capability is critical to sustained lunar mission success. By the end of the decade, the excavation needs increase from sampling levels to tens or hundreds of tons of regolith per year. Full-scale sustained ISRU and construction of infrastructure will increase that amount to thousands of tons of regolith per year.

Figure 2. Counter-acting excavation forces Figure XX:

© ASCE

396

Earth and Space 2022

The IPEx leverages years of design refinement and testing of RASSOR (Mueller et al (2013)). The first generation of RASSOR was a proof-of-concept that demonstrated the ability to do excavation in low gravity with a low mass excavator by using counter-acting excavation tools. The second generation of RASSOR (known as RASSOR 2.0) (Mueller et al (2016)) was a breadboard system built with components that have proven paths to flight, which enabled the team to accurately estimate key metrics such as mass and energy usage for future flight versions of the excavator to inform architecture studies. The IPEx dry mass target is 30kg which requires a reduction in scale from the current RASSOR 2.0 system which has a dry mass of 65kg. Volumetrically, IPEx will be scaled between 50% - 70% of RASSOR 2.0. The bucket drums of IPEx will therefore also be reduced scale, and a method of estimating the forces on the scaled bucket drums during excavation is needed. An abundance of prior work and models exist for prediction of excavation forces (Gallo et al (2010), Wilkinson et al (2007), Zeng et al (2007), and Zelenin et al (1975)) however these models represent conventional excavation tools such as blades and buckets and do not address the unique configuration of a bucket drum. The primary goal of this work is to inform the design of the IPEx by answering the following questions: • • • • •

What is the regolith capacity of the scaled bucket drum? What are the forces (X,Y,Z) and torque due to excavation? How do those forces and torque change with drum rotational speed, linear cutting speed, and cut depth? How much time is needed to fill at various cut depths and speeds? How much energy is used during excavation, and how does it vary with drum rotational speed, linear cutting speed, and cut depth?

EXPERIMENTAL SETUP Three different bucket sizes have been selected for the test (see Figure 3). The range of sizes is to help determine a scaling relationship for different size bucket drums. The largest drum is the RASSOR 2.0 drum. The medium and small drums were scaled to meet the minimum and maximum required sizes for the ISRU Pilot Excavator. The overall dimensions for the bucket drums are as follows (see Figure 4 and Table 1):

© ASCE

Figure 3. Bucket Drum Size Comparison

397

Earth and Space 2022

Figure 4. Bucket Drum Dimensions Table 1. Bucket Drum Dimensions

Small Medium Large

Width 7.83” (198.9mm) 9.69” (246.1mm) 13.88” (352.6mm)

Diameter 9.35” (237.5mm) 11.62” (295.1mm) 17.21” (437.1mm)

Scoop Width 2.03” (51.6mm) 2.5” (63.5mm) 3.56” (90.4mm)

Scoop Height 1.04” (26.4mm) 1.36” (34.5mm) 1.88” (47.8mm)

Bucket drums have many performance-dependent dimensions that effect their digging and collection. The scoop height and width determine the dig depth and flow rate into the drum. Special consideration needs to be made to mitigate bridging of the regolith as it enters the drum. Scoop height also determines the effective internal diameter, which determines the amount of total regolith collected. The baffle follows behind the scoop opening and is required to keep regolith from falling out of the scoop opening during digging and transportation. The baffle geometry can also affect bridging, and total regolith collected. The baffle channel width expands as it approaches the center to reduce bridging and ease the flow of regolith as it enters the drum. Each drum slice, which consists of two scoop openings, has closed section walls to separate the drums slices from one another. The intent of the closed slices is to reduce the amount of regolith that may fall out of a downward facing drum scoop by limiting the exposed volume to one slice instead of the entire bucket drum. The bucket drums are made using two different construction techniques and materials. The large drum was made for RASSOR 2.0 and is a bonded assembly consisting of aluminum, carbon fiber, and stainless steel. The vertical side and section walls are aluminum. The top scoop and baffle sections are thin carbon fiber sheets which are glued to the vertical side and section walls. The cutting-edge teeth are made from stainless steel and are riveted to the carbon fiber sheets and vertical side and section walls. The small and medium bucket drums are a monolithic piece made from 3D printed Selective Laser Sintering (SLS) nylon. The general geometry of all bucket drums is intended to be identical except for the obvious scaling differences. The major differences

© ASCE

398

Earth and Space 2022

between the large and small/medium buckets drums are the sharp corners of the baffle sections and the scoop teeth. The larger bucket drum has perpendicular corners in the baffle sections and the small/medium bucket drums have a radius. The radius is assumed to help reduce friction and aid in reducing bridging. The small/medium bucket drums were printed with a straight edge along the scoop cutting edge whereas the larger bucket drums have a jagged tooth profile. Testing was performed in the KSC Regolith Test Bed (RTB). The RTB is an enclosure with 120 tons of BP-1 lunar simulant corresponding to a volume of 8m x 8m x 1.1m. BP-1, or Black Point 1, lunar simulant is an inexpensive geotechnical lunar regolith simulant sometimes used for excavation and mobility testing. BP-1 is derived from material in the Black Point basalt lava flow in northern Arizona. Because it is derived from basalt, BP-1 is more representative of mare lunar soils than highland lunar soils. The granular size distribution of BP-1 falls within the one standard deviation of actual lunar regolith particle distribution returned by Apollo lunar missions. The bucket drum test stand consists of 2 linear axes, X and Z, and a rotary axis, R (see Figure 5). The linear axes are closed loop controlled and have linear absolute sensors measuring the position of the final stage. The X axis has a maximum speed of 246 mm/sec and a positioning accuracy of 0.25 mm. The Z axis has a maximum speed of 460 mm/sec and a positioning accuracy of 0.25 mm.

Z

X

R ~3.85m excavation area

Figure 5. Bucket Drum Test Stand To prepare the BP-1 simulant, leveling and compacting attachments were designed to attach to the bucket drum test stand. A beam was used to drag across the excavation area using the X axis to create a surface parallel with respect to the test stand (see Figure 8). Relief trenches were dug at the start and end of the excavation area to allow deposition of the surcharge of the regolith collected by the beam. The compacting attachment uses a handheld vibratory concrete compactor with a modified compacting plate (see Figure 9). The compacting plate is wider than the large bucket drum and has bent up edges to allow it to stay above the regolith during operation. The compacting attachment is driven using the X axis. The relief trenches also allow for the bucket drum to start and end excavation without engagement. The end relief trench was also used as the dump location for the bucket drum after excavation. © ASCE

399

Earth and Space 2022

The test is outfitted with two feedback sensors: a torque sensor and a multi-axis 3 Axis Force Sensor force sensor (see Figure 6). The torque sensor is coaxial with the R axis. The torque sensor is an Interface Force sensor part number: 5330-600. This sensor has measuring range of ±600 in·lbf (68 N·m) and a nonlinearity of ±0.1% of full-scale output. The multi axis sensor is an Interface Force sensor part number: 3A120-1kN. This sensor has measuring range of ±225 lbf (±1 kN) and a Torque Sensor nonlinearity of ±0.2% of full-scale output. National Institute of Standards Figure 6: Sensor Layout and Technology (NIST) calibrations were used for both sensors. A calibrated weight was used to verify the sensors in their system configuration. The Data Acquisition System (DAQ) is a National Instruments (NI) cRIO. The software allows for manual and scripted control of the test stand actuation. The scripted control allows for blocking and non-blocking position and velocity commands and time waits for each axis. This ensures repeatable test motion profiles for each experiment. The force and torque sensors are logged at 1 kHz, but the feedback from the motor controllers is logged at 10 Hz. This motor feedback is oversampled to properly align with the force/torque data. The slower rate of motor feedback is due to the limitations of the CAN bus interface used to control the motor controllers. The force/torque sensors are measured using a NI-9237 C Series Strain/Bridge Input Module, which has a 24-bit resolution and a ±25 mV/V input range. The software performs all the unit conversions real-time. The variables that are recorded are as follows: X (lbf), Y(lbf), Z (lbf), Torque (in*lbf), Z position (mm), Z Velocity (mm/sec), X Position (mm), X Velocity (mm/sec), R Position (revolutions), R Velocity (rpm), R (Motor Active Current, Amps), Sample Rate (Hz), Elapsed Time (minutes). EXPERIMENTAL VARIABLES Tests were performed varying the size of the bucket drum, linear speed during excavation, and cut depth. Table 2 lists the experimental variables and their ranges. Table 2. Experimental variables and ranges Test Variable

Range

Bucket Drum Size

Small, Medium, Large

Linear Cut Speed

10mm/s, 30mm/s

Cut Depth

10mm, 40% of scoop height

The combination of these variables resulted in 12 unique tests and each test case was repeated 4 times resulting in a total of 48 tests. The two values for cut depth were chosen to provide

© ASCE

Earth and Space 2022

absolute (10mm) and relative (40% of scoop height) comparisons between the drum sizes. The cut depths were based on prior experience with the RASSOR 2.0 excavation robot. During the many hundreds of hours of testing with RASSOR 2.0, it was observed that when digging at the full depth of the bucket drum scoop the regolith could bridge across the opening and reduce the amount of regolith collected per rotation. The team found that limiting the cut depth to only utilize a maximum of 50% of the bucket drum scoop opening would collect more regolith per rotation. The rotational speed of the bucket drums and the linear cut speeds were intentionally linked during testing to keep a constant cut pitch (the number of scoops per linear distance) across all bucket drum sizes. When the linear cut speed was increased or decreased the rotational speed was changed proportionally. The chosen ratio resulted in a bucket drum tangential velocity that was 8.5x the linear cut speed. This ratio was based upon previous experience and modeling and ensured a relatively smooth cut surface that would benefit subsequent cutting passes (see Figure 7). The linear cut speeds chosen are in the range needed for the IPEx concept of operations. TEST PROCEDURE Prepare/Reset Test Bed: The test bed was carefully prepared and characterized with hand geotechnical tools Figure 7. Bucket Drum Test Bed throughout the testing regimen to ensure being precision leveled consistency and repeatability. The BP-1 lunar regolith simulant requires special considerations when preparing or resetting the test bed. The geotechnical properties of the soil bed can vary wildly based on factors such as humidity, pouring method, compacting method, and surrounding soil densities, just to name a few. To prepare the test bed or to reset it between test regimens, a team member first used a hand shovel to loosen the bed of simulant. This operation involved scooping up the BP -1 simulant to approximately 80cm of depth, and then gently pouring it back into the same location. This operation was performed on the entire width and length of the buc ket drum test bed. The team member then scooped excess material from previous tests to roughly fill in noticeable low spots. After the simulant had been adequately churned up and roughly evened, a wide flat rake was manually pulled across the surface to perform a rough level. Care was taken to not compact the soil during this process. Once the surface was roughly leveled by hand, the Bucket Drum Test Stand was outfitted with a leveling bar and was used to perform final precision leveling and smoothing (see Figure 8).

© ASCE

400

401

Earth and Space 2022

Next, the Bucket Drum Test Stand was outfitted with a vibratory compactor and was used to perform precise and repeatable soil compaction. The compactor was gently lowered onto the surface and dragged across the full length of the test bed for 4 separate passes while being lowered 2mm between each pass (see Figure 9). The bucket drum under test was then spun up by the Bucket Drum Test Stand and gently lowered via manual control to find the exact Z-axis position where the bucket drum just barely touched the surface. This Z-axis

Figure 9. Bucket Drum Test Bed being precision compacted

Figure 8. Bucket Drum Test Bed being precision leveled

position became the zero position for height and all subsequent Z-axis position commands were based off that number. Once the bucket drum test bed was leveled and compacted, hand-held geotechnical tools were used to characterize the soil (see Figure 10). A Humboldt Pocket Shear Vane Tester was used for quick and efficient determinations of shear strength. A Humboldt Soil Penetrometer was used to determine compressive strength of the unconfined soil. Three measurements with each

tool were taken in three separate locations along the length of the bucket drum test bed. These measurements were taken before and after each bucket drum test run to confirm consistency from test to test. During the course of testing the average measurement from the shear vane tester ranged from 27-32 kPa and the penetrometer from 206-226 kPa. After the bucket drum test bed was precisely leveled and compacted, a 5mm skim cut was performed with Figure 10. Team members taking geotechnical the bucket drum under test. This skim measurements of the Bucket Drum Test Bed cut removed the less compacted following a test run topmost layer of simulant and ensured each test began with a regolith bed under the same conditions. All these steps were taken to prepare the bucket drum test bed for the actual bucket drum testing.

© ASCE

Earth and Space 2022

Bucket Drum Testing: The Bucket Drum Test Stand control software was configured for each test run by test personnel. Configurable parameters include bucket drum rotational speed in rpm, X-axis linear speed in mm per sec, Z-axis cut depth in mm, data log filename, load cell zeroing speed in rpm, X-axis start/end/dump positions in mm, and others. Once all the control software parameters were configured per the current line in the test matrix, the operator simply pressed the START TEST button and the Bucket Drum Test Stand control software automatically performed the entire test run. Each automated test run included the following actions: start data logging temporarily spin the bucket drum under test in mid-air and monitor the values of all the load cells calculate each load cell value offset, zero the load cells, and stop drum rotation while still in mid-air, move the bucket drum under test to the configured starting X-axis position lower the drum via Z-axis to the configured cut depth inside the trench so it is still hanging in mid-air spin the bucket drum under test to the configured rotational speed translate the bucket drum under test along the X-axis at the configured linear speed once the bucket drum under test has reached the configured end X-axis position (or user hits the ABORT TEST button), raise the bucket drum in midair, allow the full drum to rotate a minimum of one rotation, halt the bucket drum spinning, move the bucket drum in X-axis to the dump location, and spin the bucket drum backwards to dump out the material Figure 11. Bucket drum (see Figure 11) dumping collected halt data logging regolith After each test run, test personnel used the hand-held geotechnical tools to characterize the soil along the newly revealed surface layer. Then test personnel configured the control software parameters per the next line in the test matrix and another run was performed. As this sequence was repeated, the bucket drum under test dug a step deeper into the prepared regolith simulant during each test run. After several test runs, the Bucket Drum Test Stand reached a depth limit as the structure on the Z-axis approached the uncut top layer of regolith simulant. The test matrix was designed such that test runs were paused prior to this limit and the test bed was then reset per the earlier section before continuing.

© ASCE

402

403

Earth and Space 2022

RESULTS Example Test Output Data In order to properly design IPEx the forces and torques during excavation must be understood. Figure 12 below shows a typical plot of the forces and torques produced during a bucket drum excavation test. The force in the X direction is the primary force of excavation. This force was generally constant during an excavation test because the regolith bed was prepared and leveled, and the cutting depth and speed were constant. The force in the Y direction was negligible in all test data. The force in the Z direction starts from zero and linearly increases as the regolith is captured in the drum. The total torque about the R axis of the bucket drum is a combination of the torque due to the excavation X-force and the torque of the captured regolith recirculating inside the drum. Following a test cut the bucket drum was lifted into the air and rotated multiple revolutions. The torque shown during that process is purely due to the recirculation of the captured regolith as the drum is no longer in contact with the test bed. Similarly, the Z force at that time is constant and equates to the weight of the regolith captured in the drum. Finally, the forces and torques can be seen returning to zero quickly upon reversing the rotation of the drum and dumping the regolith.

Drum rotating in air

Start cut

Dump

Stop cut

Figure 12. Example bucket drum test output data

© ASCE

404

Earth and Space 2022

Excavation Force Sensitivity To Cut Depth and Speed To aide with the design of bucket drums and their actuators a plot was generated that shows the horizontal excavation force per bucket drum width vs cut speed and depth (see Figure 1 3). This plot can be used to provide an estimated horizontal excavation force for various scales of bucket drums, cut depths, and cut speeds. The ranges of the variables in this work envelope the needs for the IPEx concept of operations, however future work could increase these ranges to increase fidelity. The shaded area in the plot shows the tested range of these variables and the resulting trend lines from the data.

Figure 13. Example bucket drum test output data Total Regolith Collected The average mass of regolith collected by each bucket drum is an important metric to properly scale IPEx to meet the mission KPPs. Table 3 below lists the average mass of regolith collected for the three scales of bucket drums tested and corresponding total excavator capacity for a 4-bucket-drum system.

© ASCE

405

Earth and Space 2022

Table 3: Total regolith collected Bucket Drum Size

Avg. Total Regolith Collected Per Drum (kg)

4- Bucket-Drum Excavator Capacity (kg)

Small

3.80

15.21

Medium

7.30

29.20

Large

24.98

99.94

Regolith Collection Rate Similarly, the regolith collection rate of each bucket drum is an important metric to properly scale IPEx to meet the mission KPPs. Figure 14 below shows a plot of the collection rate of a single drum against the drum size, cut speed, and cut depth.

Figure 14. Example bucket drum test output data Excavation Energy Per Mass Of Regolith Collected The energy used to excavate the regolith is also a metric that will inform the IPEx design. Figure 15 below shows a plot of the mechanical energy (from torque and RPM at the bucket drum) per kilogram of collected regolith against the tested bucket drum sizes, cut speeds, and cut depths.

© ASCE

406

Earth and Space 2022

Figure 15. Example bucket drum test output data CONCLUSION The experimental data from the testing described above results in the following conclusions: The horizontal excavation force per unit width of bucket drum increases with increasing linear cut speed and cut depth. The plot of this data can be used to estimate the forces for various scale bucket drums, cut speeds, and cut depths. The medium and small-scale bucket drums tested can collect the necessary amount of regolith at the desired rate to support the IPEx concept of operations. Excavation energy per mass of regolith collected is reduced as cut speed decreases, drum size decreases, and cut depth increases. These data and the corresponding trends will be used to inform the design of the IPEx bucket drum subsystem and can form a baseline for future bucket drum designs. Additional tests should be conducted to increase the data available and the fidelity of the predicted trends. REFERENCES Clark, D., Patterson, R., & Wurts, D. “A novel approach to planetary regolith collection: the bucket drum soil excavator.” AIAA Space 2009 Conference & Exposition. 2009.

© ASCE

Earth and Space 2022

Dickson, D. C., Sibille, L., Galloway, G. M., Mueller, R. P., Smith, J. D., Mantovani, J. G., & Schreiner, S. (2016). Modeling Dynamics of Counter-Rotating Bucket Drum Excavation for In Situ Resource Utilization (ISRU) in Low-Gravity Environments. Earth and Space 2016. Gallo, C. A., Wilkinson, R. A., Mueller, R. P., Schuler, J. M., & Nick, A.J. (2010). Comparison of ISRU Excavation System Model Blade Force Methodology and Experimental Results. NASA/TM-2010-215591. Glenn Research Center, Cleveland, OH. Mueller, R. P., Cox, R. E., Ebert, T., Smith, J. D., Schuler, J. M., & Nick, A. J. (2013, March). Regolith Advanced Surface Systems Operations Robot (RASSOR). In Aerospace Conference, 2013 IEEE (pp. 1-12). IEEE. Mueller, R. P., Smith, J. D., Schuler, J. M., Nick, A. J., Gelino, N. J., Leucht, K. W., Townsend, I. I., & Dokos, A. G. (2016). Design of an Excavation Robot: Regolith Advanced Surface Systems Operations Robot (RASSOR) 2.0. Earth and Space 2016. Suescun-Florez, E., Roslyakov, S., Iskander, M., & Baamer, M. (2014). Geotechnical Properties of BP-1 Lunar Regolith Simulant. American Society of Civil Engineers (ASCE), Reston, VA. Wilkinson, A. & DeGenarro, A. (2007). “Digging and pushing lunar regolith: Classical soil mechanics and the forces needed for excavation and traction.” Journal of Terramechanics 44, 133–152. Zelenin, A., Balovnev, V., & Kerov, I. (1975). Machines for Moving the Earth: Fundamentals of the Theory of Soil Loosening, Modeling of Work Processes and Forecasting Machine Parameters. Amerind Publishing Co. and Mashinostroenie, Moscow. Zeng, X., Burnoski, L., Agui, J., & Wilkinson, A. (2007). Calculation of Excavation Force for ISRU on Lunar Surface. 45th AIAA Aerospace Sciences Meeting and Exhibit, Reno, NV.

© ASCE

407

Earth and Space 2022

Practical Space Resource Utilization at the Hundred Megatonne Scale: Enabling a Planetary Sunshade to Reverse Global Warming Liz Scott1,4, Dorian Leger2, David Borncamp1, Scot Bryson3, Peter E. Corwin1, Maxwell C. Sissman1, and Ross Centers4 1

Center for Space Resources, Colorado School of Mines, 1310 Maple St., GRL 234, Golden, CO 80401; email: [email protected]; [email protected]; [email protected]; [email protected] 2 European Space Agency, Researcher Project, Linder Hoehe, 51147 Köln, Germany; email: [email protected] 3 Orbital Farm, 68 Harvard Ave., Toronto, Ontario; email: [email protected] 4 Planetary Sunshade Foundation, 601 16th Street Suite C #174, Golden, CO 80401; email: [email protected]; [email protected]

Abstract Climate change is the defining issue of our time. Decarbonizing Earth's economy and removing existing carbon from the atmosphere are mandatory steps for addressing climate change, but may not occur quickly enough to avoid catastrophic global warming. Space-based solar radiation management is one technique to address global warming by deflecting a small percentage of incoming solar energy before it enters Earth's ecosphere. The concept of a planetary sunshade is a physical, thin-film structure of about one million square kilometers, placed at Sun-Earth Lagrange 1, to mitigate or even reverse global warming. If the sunshade is made from space resources, it could provide a powerful objective for developing a space resources economy built on megascale construction. The planetary sunshade could require hundreds of megatonnes of material to build; this scale of space resource utilization has only barely been considered. In August of 2021, the Planetary Sunshade Foundation, in conjunction with the Colorado School of Mines, held a workshop to begin development of a space resource utilization plan with an emphasis on the scale needed to construct the sunshade. The workshop had 25 attendees, representing civil space agencies, industry, academia, and researchers from six countries. Working groups focused on resource extraction and processing and logistics for both lunar and asteroidal resources. These groups discussed the current state of knowledge, knowledge gaps, and future work required for extraction and processing techniques and logistics architectures. The purpose of this paper is to describe the key findings and conclusions of that workshop.

© ASCE

408

Earth and Space 2022

1. Introduction a. Planetary Sunshade Climate change is the defining issue of our time (United Nations, 2021). Although the possible impacts of anthropogenic climate change have been understood for decades, mitigatory action has thus far been woefully inadequate. The Sixth Assessment Report from the Intergovernmental Panel on Climate Change (IPCC, 2021) makes clear that very little time remains to reduce emissions enough to avoid catastrophic climate change. On current emissions trajectories, sometime between 2030 - 2050 our atmosphere’s cumulative greenhouse gas burden will drive warming past the 1.5°C - 2.0°C maximum tolerable range according to the Paris Accord. At that point, severe environmental and economic consequences will manifest, and geoengineering to cool our planet could become an important policy option. The energy budget of the Earth is the difference between energy entering and energy leaving the Earth system. Energy enters the Earth system from the Sun. Some energy is reflected back to space by Earth’s atmosphere, clouds, and surface, and some is absorbed by the atmosphere, surface, and oceans. When the energy budget is in equilibrium - the energy entering the system equals the energy leaving the system - global temperatures remain relatively stable. However, carbon dioxide and other greenhouse gases (GHG) have altered the composition of the atmosphere, causing it to absorb additional outgoing thermal infrared energy that would have otherwise radiated back out to space; as a result, the Earth no longer radiates as much energy as it absorbs. This imbalance has caused global surface temperatures to increase by 1.07°C from 1850-1900 to 2010-2019 (IPCC, 2021). Addressing the energy imbalance can be accomplished in two ways. The first approach is to reduce carbon emissions and remove carbon already in the atmosphere; this would decrease the GHG effect and increase the amount of energy leaving the Earth system. This is essential to address the full range of effects of climate change. However, this will be a slow process. Even if carbon emissions ceased today, the carbon already in the atmosphere has committed Earth to about 1.5°C of global warming. Additionally, the IPCC report found that, “If global net negative CO 2 emissions were to be achieved and be sustained, the global CO 2-induced surface temperature increase would be gradually reversed but other climate changes would continue in their current direction for decades to millennia” (IPCC, 2021). Meanwhile, catastrophic effects of global warming are already being felt around the globe and will only worsen as emissions continue to rise (Environmental Protection Agency, 2022). Deliberate intervention may be able to stabilize our climate alongside GHG management. This leads to the second approach to address Earth’s energy imbalance: reducing the incoming solar energy through solar radiation management (SRM). SRM is a type of geoengineering that can be accomplished in two broad ways. The first is to increase the amount of solar energy reflected back to space by increasing Earth’s albedo. This could be accomplished through releasing aerosols into Earth’s atmosphere, seeding marine clouds, or altering land use to increase reflectivity. The second method is to reduce the amount of solar energy that reaches Earth in the first place. This could be accomplished with a planetary sunshade.

© ASCE

409

Earth and Space 2022

Conceptually, a planetary sunshade is a large constellation of thin film structures positioned in space between the Earth and the Sun to reduce incoming solar energy by a few percent. The targeted reduction could vary depending on how long it takes to decrease carbon emissions, how quickly carbon can be removed from the atmosphere, and whether society wishes to merely slow global warming or restore preindustrial temperatures. The ideal location for a sunshade constellation is around Sun-Earth Lagrange Point 1 (SEL-1) (Jehle et al., 2020). SEL-1 is an equilibrium point between the Earth and the Sun, about 1.5M km away from Earth, where the gravitational effects from the Earth and the Sun cancel each other out. Thin film structures are also affected by solar radiation pressure, moving the equilibrium point sunward of SEL-1. Minimum mass architectures orbit as far as 2.36M km from Earth (McInnes 2002, Sanchez 2015). The sunshade could maintain its orbital position through the use of solar sailing, a method of spacecraft propulsion using radiation pressure applied by sunlight (McInnes, 1999). Depending on the desired reduction in solar radiation and the properties of the sunshade itself (areal density, reflectivity, etc), the sunshade would need to have a total area of 1M km2 (about the area of Egypt) to 4M km2 (about half the area of Australia) (Jehle et al., 2020). This is many orders of magnitude larger than anything else humans have built in space. While it is possible to construct a massive structure like the sunshade on Earth and then launch and deploy it in space - especially if SpaceX’s superheavy vehicle Starship achieves its design goals and drastically increases access to space - the use of space resources offers compelling advantages.

b. Space Resources Earth’s surface is at the bottom of the deepest gravity well in cislunar space; it takes a tremendous amount of energy to lift materials from the surface of Earth into space. So far in humanity’s space exploration programs, this energy has come in the form of chemical propellants such as liquid hydrogen and refined kerosene. The amount of energy required to leave Earth is so great that propellants and oxidizers can comprise over 90% of the initial mass of the launch vehicle. When the mass of the vehicle structure is included, the payload mass fraction can be as little as 2-5% of the initial mass of the launch vehicle. This is known as the “tyranny of the rocket equation”, and it is an inescapable physical fact for launch vehicles that use chemical propulsion (Pettit, 2012). New technologies, such as nuclear rockets or space elevators, may free us from this tyranny, but those technologies are far in the future. The other way to circumvent this problem is to use the vast and varied resources already located outside of Earth’s deep gravity well. The resources of space include physical materials such as minerals, metals, ice, and volatiles, but can also include intangible environments such as the presence of vacuum, microgravity, or solar energy. From an industrial perspective, resources have value by their utilization, meaning that anything that can be extracted from space or is present in the space environment is a resource if it is useful in some way. Extracted resources can be used in space (in-situ resource utilization, or ISRU) or returned to Earth. Space resources can be found in deep space and on celestial bodies, such as the Moon, asteroids, and Mars. Some examples of space resources include:

© ASCE

410

Earth and Space 2022

 

Metals extracted from lunar regolith for use in construction Water ice extracted from icy regolith in craters at the lunar poles for use in life support or as propellant  Solar energy, unaffected by day/night cycles or Earth’s atmosphere, collected from space and used to power an orbital manufacturing plant  The view of Earth from low-Earth orbit (LEO), used to entice tourists to visit an orbital hotel  Microgravity, used to create perfect crystalline structures for use in medicine, fiber optics, and other applications To develop a space resources economy, three things must be present: a recoverable resource, the technology to recover it, and a customer. These concepts are represented in the generalized extractable space resources value chain, shown in Figure 1.

Figure 1: Space Resources Value Chain (© Colorado School of Mines, 2020) Some steps in the space resources value chain may differ depending on the resource (for example, microgravity does not need to be mined, processed, manufactured, or delivered - once the environment is characterized and the appropriate materials or equipment are delivered to the environment, it can be utilized), but the overall arc is applicable to most space resources. Suppliers and customers can be found at all levels of space resource value chains. Although the idea of space resource utilization is not new, it received renewed attention in the 2010s as major civil space organizations, such as NASA and ESA, turned their focus towards sustainable space exploration. Sustainable space exploration includes two components: the first is sustaining humans in space for longer periods of time with less need for resupply from Earth, and the second is creating a sustainable space economy that is less susceptible to political shifts. The use of space resources is key to both sustainability pillars. In 1976, physicist Gerard O’Neill published The High Frontier, which outlined a technically feasible and economically sound plan to solve the greatest problems facing humanity at the time through the use of space resources. In the 1970s, the greatest concerns facing humanity centered around the limited supply of resources on Earth. Against the backdrop of Malthusian fears of overpopulation, industrial pollution, and the lingering effects of the energy crisis, O’Neill proposed the construction of space-based solar power (SBSP) satellites that would supply Earth with all the clean power it could ever require. These satellites would be built from space resources by a huge workforce of regular people living in Earthlike comfort on enormous, cylindrical orbital habitats (the concept is now referred to as an O’Neill cylinder). As humans adapt to living in space, the use of O’Neill cylinders would increase, and humans could expand further out into the solar system and beyond. This would ease the burden on Earth imposed by human industrial activity, allowing Earth’s environment to return to Edenic equilibrium. Thus far, O’Neill’s vision has not manifested into reality. Some of the reasons for this were financial: the Space Shuttle achieved its cost goals and new oil

© ASCE

411

Earth and Space 2022

extraction technologies, such as fracking, reduced energy costs below what SBSP could compete with. However, the authors believe that the time is right to reinvigorate his vision of a thriving spacefaring society rooted in environmental stewardship, now focused on managing the consequences of climate change.

c. 100MT Space Resources Workshop In August of 2021, the Planetary Sunshade Foundation, in collaboration with Colorado School of Mines’ Space Resource Program, hosted a workshop to start to address the use of space resources at the scale needed to build a planetary sunshade. The focus of the workshop was on the front end of the space resources value chain: specifically, the extraction, processing, and logistics of providing 100MT (megatonnes, millions of metric tons) of metal and 100MT of silicon to cislunar space (such as Earth-Moon Lagrange point 1 [EML-1]) or SEL-1 between 2040 and 2060. The workshop had 25 attendees representing civil space agencies, industry, and academia (from new graduate students to professors) from six countries. The workshop was held virtually and included a mix of small group discussions, presentations to the whole group, and a roundtable discussion at the end. The attendees were broken into three small groups: lunar resource extraction and processing, lunar logistics, and asteroid resource extraction, processing, and logistics. The lunar resource extraction and processing group focused on how the required materials would be extracted from the Moon and how they would be processed. The lunar logistics team considered how materials would be transported around the lunar surface, off the lunar surface, and how they would be moved in space. The asteroids group combined the considerations of both lunar groups, but applied to near-Earth asteroids (NEAs). To help focus the workshop discussions, several requirements were proposed to the workshop participants. The first requirement limited the participants to only consider resources and locations in cislunar space, in addition to SEL-1. This area comprises the Earth and its orbits, the Moon and its orbits, as well as the Earth-Moon Lagrange points. NEAs were also included in the workshop assessment. Although there are numerous resources on Mars and in main belt asteroids - some of which are rare in cislunar space - those locations are distant out that prospecting, extracting, and returning resources to cislunar space is inefficient and unlikely to happen at scale before the end of the century. The second requirement was that the architectures support the extraction, production, and delivery of 100MT of metal and 100MT of silicon. The main structure of a planetary sunshade constructed from space resources, including thin film membranes and support structures, will likely be composed of metal. The type of metal was left open; iron, aluminum, titanium, or alloys are all possibilities. Silicon would be used to create photovoltaics; if the sunshade were constructed as a space based solar power station, it could generate about 70TW of energy in addition to shading Earth (Fix, 2021). The required mass of these materials is inexact and was selected to emphasize the scale of the space resources operations. Participants were also encouraged to consider other materials at smaller scales, including trace metals for alloys, water for propellant, and consumables for use in material extraction and processing.

© ASCE

412

Earth and Space 2022

The third requirement was for the materials to be delivered to a location in cislunar space. This allows an apples-to-apples comparison between resources extracted from the lunar surface and from NEAs. The most obvious delivery location is SEL-1, where the finished sunshade will be located. EML-1, located between the Earth and the Moon, is another potential depot location and will likely be a hub of activity for all future space activities. The final requirement is that the materials are delivered between 2040 and 2060. The political will and financial ability to begin construction of a planetary sunshade could begin sometime around the middle of the century, if climate trends continue and heavy launch architectures currently in development meet design goals by then. Construction of the sunshade would almost certainly be exponential, meaning that resource extraction, processing, manufacturing, and assembly capabilities will ramp up over time.

2. Lunar Resource Extraction & Processing The overall strategy for extracting metals and silicon from the Moon centers on the use of lunar regolith. Regolith is the layer of unconsolidated rocky material (from dust to boulders) that covers bedrock. Since the Moon lacks air or water needed for erosional processes like Earth, lunar regolith is primarily shaped by meteoroid impacts. Surficial regolith is the top layer of material on the lunar surface; depths range from as low as 4-5m in the mare regions to 20m or more in the oldest highland regions. As a general assumption, the top 10m of regolith can be considered easily accessible; below that depth, particle size and bulk density begin to increase, making regolith extraction more difficult (Weber, 2014). Regolith is composed of silicate minerals and differs between the older highlands material and the newer, volcanic mare material. In the lunar highlands, plagioclase-rich rocks such as anorthosite and norite dominate, while in the mare regions, pyroxene and olivine are most prevalent (Heiken, 1991). These minerals are high in metallic elements. Silicon accounts for about 21wt% of the lunar regolith, although it is higher in certain areas, such as the Gruithuisen Domes. Aluminum contributes about 13wt% in the highlands, but only 5% in the maria. Conversely, iron accounts for about 6wt% in the highlands but up to 15wt% in the maria. Titanium is around 1wt% of highlands regolith and as high as 5wt% in the maria (Heiken, 1991). These elements are tightly bonded with oxygen and other elements to form the silicate minerals that comprise regolith. The first major challenge to producing metal from the lunar regolith is the volume of regolith that must be processed. Unlike Earth, the Moon does not possess high-grade ores or veins of relatively pure metals; although certain metals are more abundant in the highlands or in the maria, extracting metals will simply be a matter of processing bulk regolith. One benefit of this fact of lunar geology is that it essentially eliminates the need for prospecting. From Apollo samples and remote sensing data, we know enough about the materials in lunar regolith to begin extraction and processing as soon as the technologies to do so are ready. Although the relative abundance of certain elements varies between highlands and maria regolith, all the major metallic elements (Si, Al, Fe, Ti) are present in both types of regolith. This means that single-stream processing of bulk regolith could be

© ASCE

413

Earth and Space 2022

used to extract different types of metals from the same bulk material. If a process is developed to extract different metallic elements, heavy metals such as iron could be used to build up manufacturing capabilities on the lunar surface while lighter materials such as aluminum and titanium could be moved off the lunar surface for use in orbital manufacturing of sunshade components. One significant trade study to be performed is whether it makes sense, at the 100MT scale, to have multiple extraction sites that specialize in specific elements, or whether single-stream processing of suboptimal bulk regolith at a single location is more efficient. Highlands regolith contains much more aluminum than maria regolith. Polar craters, where volatiles such as water ice are likely to be found, also contain highlands regolith. The maria are higher in titanium and iron. Locations such as the Gruithuisen Domes could be targeted for their high concentration of silicon. While extracting different metallic elements from different locations is likely more efficient, having multiple launch locations or transporting goods over long distances on the lunar surface may negate that benefit. Single site operations may also enable economies of scale. Extraction of metal from lunar regolith is an active area of research for NASA, ESA, and supporting industries. Current excavation equipment designs, naturally, focus on pilot-scale rovers that can be built on Earth and launched to the lunar surface. However, scaling up to the level needed for production of a structure like the sunshade will require thinking less like aerospace engineers and more like mining engineers. Lightweight, finely tuned excavation rovers would be replaced by large, heavy, relatively simple technology like terrestrial bucket wheel excavators, which will have to be built, in large part, on the Moon. The 100MT Workshop group focused on the extraction of iron, aluminum, and silicon. Iron was identified as being the first metal that will be needed to bootstrap manufacturing capabilities on the Moon, which will then be used to extract silicon and aluminum. Iron can be used to produce larger, heavier extraction equipment, launch facilities to get materials off the surface, and structures to house human workers. The production chain for metallic elements requires excavation of regolith, transportation to a processing station, beneficiation of the regolith to enrich it, a reactor to break the metal-oxide bonds, and a high temperature separation step to extract the desired metal from other metals. There are different technology options available for each step of the process. For example, breaking the metal-oxide bonds could be done with a carbothermal reactor, a hydrogen reduction reactor, or molten salt electrolysis (Rasera, 2020). Depending on the choices made, auxiliary components must be added; as an example, a hydrogen reactor will require a water electrolyzer to regenerate H2. The optimal combination of technologies will depend on the scale of production, feedstock compositions, and operation locations. Workshop attendees performed preliminary estimates to achieve an order-ofmagnitude estimate of the regolith mass required for mining 100MT of iron. Attendees analyzed a five-step production chain that included excavation, transportation, beneficiation, redox reaction, and iron separation. The process began with bulk regolith containing 10vol% ilmenite (FeTiO3), which is a common oxide mineral in mare basalts. They assumed beneficiation efficiency using electrostatic

© ASCE

414

Earth and Space 2022

separation that could yield approximately 45% recovery of ilmenite with 67% ore purity (Agosto, 1984). Furthermore, attendees assumed that if 100% of the ironoxide in ilmenite was successfully reduced in the redox reactor, and 95% of the resulting iron was recuperated by temperature-based separation, the mass of recoverable iron would be 1.6% of the mass of regolith brought into the processing system. The mass flow for this process is shown in Figure 2. To produce 100MT of iron, 6,250 MT of regolith is required. Assuming that only the top 10m of regolith are excavated, this results in a mine size of 373km 2, or a square of 19km on a side.

Figure 2: Iron Extraction and Processing Mass Flow Similar calculations were performed for aluminum and silicon, with simpler assumptions. For aluminum, attendees assumed highlands regolith with 30wt% aluminum oxide (Al2O3). Assuming 100% recovery of aluminum from the regolith, a mine size of about 42km2, or 6.5km on a side, is required to produce 100MT of aluminum if the top 10m of regolith are excavated. For silicon extraction, attendees assumed that regolith from the Gruithuisen Domes, at about 63wt% silicon oxide (SiO2), was used. Assuming 100% recovery of silicon, 100MT of silicon can be extracted from a mine 22.6km2, or 4.75km on a side. Substantial additional work remains to fully characterize the volume of regolith needed to produce 100MT of metal and silicon. For example, assessing whether multiple metals and silicon can be extracted from a single stream of bulk regolith, perhaps through the FFC-Cambridge molten salt electrolysis method, could determine whether a single mine or multiple mines in different locations is most beneficial. These calculations, however, provide a rough order of magnitude of the potential size of regolith mines. A terrestrial comparison can be drawn to the phosphate mines in Florida, which cover approximately 1.3 million acres (5,260 km 2) in central Florida1.

1

Phosphate. Florida Department of Environmental Protection. Available at: https://floridadep.gov/water/mining-mitigation/content/phosphate

© ASCE

415

Earth and Space 2022

3. Lunar Logistics The lunar logistics workshop group focused on moving extracted materials around the surface of the Moon, from the lunar surface into lunar orbit, and from lunar orbit to its final location in space.

a. Lunar Surface Transportation In a large-scale mining operation, there will be substantial material moved around the surface of the Moon. Excavated regolith will need to be moved from the excavation site to a processing facility; this transportation operation will likely be a few tens of kilometers at most. This type of transportation is likely to be done by autonomous vehicles, traversing roads established as part of the initial mine setup. Processed materials, whether in raw metallic form or manufactured components, will need to be moved from the processing or manufacturing site to a launch facility, where they will be moved from the lunar surface into space for further processing or assembly. If multiple mining locations are used to optimize metal yield from raw regolith, launch facilities could be hundreds of kilometers from mine and processing operations. No proven technologies for moving large amounts of material on the Moon exist yet, but some possibilities are magnetic levitation trains, large wheeled vehicles, or suborbital launch and landing craft. Each of these options requires extensive infrastructure to establish; maglev trains require tracks, wheeled vehicles require roads, and suborbital launch and landing craft require pads, dust mitigation structures, refueling apparatus, and a supply of propellant.

b. Lunar Surface to Lunar Orbit There are several options for lifting materials from the surface of the Moon into space. Chemical rocket propulsion was demonstrated on the Apollo missions and several robotic sample return missions; it is also the technology used to loft everything from Earth into space. In addition to high technology readiness level (TRL), chemical propulsion is also relatively gentle; humans and delicate components can survive the loads imparted during launch. Chemical propulsion is well understood, but extremely resource intensive. Water from the lunar poles could be used to produce propellant, but unless abundant water is available at the chosen location on the Moon, it may have better uses in life support or in material extraction and manufacturing processes. Oxygen, produced as part of the process of extracting metals from regolith, could be used as an oxidizer for fuel, such as hydrogen, brought up from Earth. Space elevators are a futuristic technology that shows theoretical promise; however, it is not likely to be developed and constructed in time to support the 20402060 timeframe established in this workshop. Construction of a cable extending from the lunar surface to EML-1, nearly 60,000 km away, could require as much material as the sunshade itself. A more promising option for delivering raw materials is a mass driver, where electromagnetic force is applied to a payload to accelerate it extremely quickly. A coilgun design could provide the 2.6km/s launch velocity necessary to reach EML-2 from the lunar surface, without the need for chemical propellants (Maheswaran, 2021). While theoretically well-suited for the lunar environment, lunar coilguns are still very low TRL concepts. The g-force loading on payloads is also extremely high; humans could not survive launch from a coilgun, meaning that some chemical launch © ASCE

416

Earth and Space 2022

capability will still be required. Use of coilguns at the megatonne scale also create some logistical challenges. The material will likely need to be packaged in some sort of reusable container or “bucket” for launch. If the buckets are simply containers, a collector in lunar orbit would have to catch the buckets, remove the material, compensate momentum, and potentially send the bucket back to the lunar surface. The buckets could also be equipped with propulsion systems and guidance, navigation, and control (GNC) systems to navigate the space environment and deliver materials to a specified destination. This option, however, greatly increases the complexity of the buckets. Regardless of the transportation system selected, there are several possible locations to which the material could be delivered. These locations include SEL-1 (where the sunshade will be located), EML-1, or another Earth-Moon Lagrange point such as EML-2 or EML-5. EML-1 is likely to become a hub of activity in cislunar space due to its central location; as space economic activity increases, transportation depots, refueling stations, manufacturing facilities, human habitats, and other infrastructure are likely to be established there. This infrastructure could be leveraged to build sunshade components, making delivery from the lunar surface to EML-1 a good possibility. If manufacturing capabilities are established at the sunshade’s final destination near SEL-1, materials could be delivered directly there. However, this requires significantly more energy to reach than EML-1, making a direct shot from the lunar surface to SEL-1 impractical with small spacecraft.

c. In-Space Transportation There are several potential options for transporting materials in space. As with chemical launch, chemical space propulsion is well-understood. It does, however, require a source of propellant to refuel. Solar sailing is another option for traversing space; several missions, including the IKAROS probe, have successfully demonstrated solar sailing propulsion. Transport of materials from an EML location, such as EML-1, to SEL-1 will likely utilize a series of reusable tugs. This will reduce the complexity of having to incorporate GNC and propulsion systems onto transport containers and could allow sunshade construction to use fleets of tugs constructed for other cislunar economic purposes.

4. Asteroid Resource Extraction, Processing, and Logistics Extracting materials from asteroids, rather than the Moon, presents a different set of benefits and challenges. One major benefit of asteroid resource utilization is the variety of resources available. The Moon is poor in elements such as carbon, nitrogen, and hydrogen that are necessary for human survival and may be required for certain manufacturing processes. However, these elements can be found in certain types of asteroids, particularly carbonaceous chondrites. Volatiles, such as water, are abundant in carbonaceous chondrites, and metals are much more concentrated in metallic asteroids than they are in lunar regolith. The variety of materials within asteroids could allow spacecraft to use waste material, such as excess oxygen or slag solid material, as fuel for the return trip to cislunar space. The most significant challenge to asteroid resource exploration is that humans have not performed substantial prospecting activities on asteroids. Tens of thousands

© ASCE

417

Earth and Space 2022

of NEAs have been identified, but only a few – Eros, Itokawa, Ryugu, and Bennu by the robotic missions NEAR Shoemaker, Hayabusa, Hayabusa2, and OSIRIS-REx, respectively – have been explored up close. While it is possible to get a general idea of an asteroid’s composition by linking asteroids identified through telescope observations to meteorites found on Earth, it is difficult to be more specific than that. Large scale prospecting surveys can be used to identify candidate asteroids for further prospecting but determining composition with sufficient detail for subsequent mining will be expensive and very high risk. As human economic activity extends into cislunar space, NEAs will undoubtedly be mined for resources, but this is unlikely to occur on a large scale in the 2040-2060 timeframe. An interesting aspect of asteroid mining is the orbital mechanics. Energy requirements to reach many NEAs are very low; some are significantly lower than the energy needed to reach the lunar surface. This allows the use of highly efficient, although potentially very slow, transportation technologies. Opportunities to access certain NEAs can vary greatly. With some NEA orbits, there may be years between opportunities to access cislunar space. This opens up the intriguing possibility of less powerful but highly efficient extraction, processing, and manufacturing occurring at the asteroid itself. However, this aspect of NEAs also makes them likely unsuitable for human exploration; these missions will be entirely robotic, with no prospect of repair by humans if something goes wrong. There are several significant architectural options for asteroid resource utilization, many of which are outlined in Brophy, Friedman, & Culick (2012). The first is whether to extract resources from asteroids at the asteroid’s location or bring asteroids back to cislunar space or SEL-1 for processing. Processing at the asteroid’s location allows for lower mass transfer - only needed materials are brought back to cislunar space - but accumulating the mass necessary for large scale construction could take decades. Bringing a large asteroid to SEL-1 for processing is less mass efficient but could allow human intervention in the extraction process if necessary. The asteroid would have to be very well characterized to ensure it contains enough of the necessary materials. Also, the some of the technologies that could be used to move a large asteroid to SEL-1, such as gravity tugs, are still conceptual. Asteroid resource utilization has synergy with planetary defense efforts. Wide prospecting surveys of NEOs will also identify any objects that are potentially hazardous to Earth. The technologies needed to extract resources or divert asteroids to new orbits could also be used to potentially break up or redirect dangerous asteroids.

5. Roundtable Discussion The final section of the 100MT Workshop was a roundtable discussion with all workshop participants. Three questions were presented to the participants and the discussion flowed from there. The discussions are documented below, edited and organized for clarity and to emphasize prominent themes.

a. Advanced Manufacturing Location The first roundtable discussion question centered on the best place to manufacture components for the sunshade. Is it more efficient to manufacture materials near where the resources are extracted and processed, or should resource

© ASCE

418

Earth and Space 2022

locations be thought of simply as mines? Although the workshop focused on resource extraction and processing, rather than manufacturing and assembly of the final sunshade, manufacturing location has a significant impact on logistics, especially for lunar resources. The answer to this question depends on where the materials come from. For asteroidal materials, it is likely that multiple asteroids will be required to supply enough of the necessary materials to build a sunshade. Building multiple manufacturing facilities at different asteroids would be inefficient, so transporting raw materials to a central facility, likely at SEL-1, seems to be the best approach. The answer is more complicated for lunar materials. Component manufacturing on the lunar surface would result in less mass having to be lifted from the Moon, but some high-g transport options (such as mass drivers) that work for raw materials may damage manufactured components. It is possible that there will be more infrastructure on the lunar surface than in orbital space, at least to start. NASA’s Artemis program will place humans on the Moon and may expand to include commercial and international astronauts as well. This could result in a manufacturing hub on the lunar surface, including power, transportation, manufacturing facilities, and humans to assist in the process, that could be leveraged to build sunshade components. Resource extraction for a planetary sunshade could also help bootstrap a lunar base; if different metals are extracted from regolith, heavy metals such as iron could be used to increase manufacturing capabilities on the Moon while lighter metals such as aluminum and titanium could be used to build the sunshade.

b. Lunar versus Asteroid Resources for Megatonne-Scale Production The second roundtable discussion question looked at whether lunar or asteroid materials are best for megatonne-scale production. For a full-scale cislunar space resources economy, the ultimate answer will be “both.” Some resources necessary for sustained human exploration and certain kinds of manufacturing, such as carbon, are rare on the Moon but abundant in certain types of asteroids. However, while we know a lot about the Moon and its resources, we know much less about asteroids, and prospecting campaigns are likely to be expensive and time-consuming; this may put large-scale asteroid resource extraction out of reach until the end of the century. As mentioned in the previous discussion, the Moon is likely to become a hub of activity in the relatively near future as civil space agencies establish infrastructure for human exploration. We know where the Moon is, what resources are available in certain areas, and have some experience in operating there. However, the lunar surface is a relatively deep gravity well - much less so than Earth, but much more than NEAs - so lifting materials off the Moon will require large scale transportation options that have not been developed yet. Chemical propulsion, possibly using water sourced from the lunar poles, is not likely to support the scale necessary for producing a sunshade, but other transportation methods are only conceptual at this point. Although NEAs are further from Earth than the Moon, the logistics of moving materials from asteroids to SEL-1 are simpler in many ways. The energy required to reach most NEAs is relatively low and there are many deep-space propulsion

© ASCE

419

Earth and Space 2022

systems, such as solar sailing and low-thrust ion engines, that could work. One open question is whether it is more efficient to extract materials from numerous small asteroids or one large one.

c. Sustainable Space Resources Development The third and final roundtable discussion focused on the concept of sustainability. Humans have been poor stewards of Earth’s environment, but we have a chance to do better in space if we consider sustainability from the outset. It would not build confidence in our maturity as a species to save the Earth but “ruin” the Moon in the process. But what does sustainability mean for space? In terrestrial resource extraction, we consider remediation and sustainability to mean that life can return to the area after resources are extracted, so what does that imply for the Moon? As with the previous question, the answer depends on the location of resource extraction. There are millions of NEAs, but there is only one Moon. Humans have a cultural connection to the Moon; it is visible from everywhere on Earth and is considered sacred by many. Would large-scale resource extraction from the Moon “deface” it? Even at the hundred megatonne scale, it’s unlikely that extraction activities would be visible from Earth without a high-powered telescope; however, even though Mt. Everest is not visible for most people living on Earth, many are nonetheless upset by images showing the pollution and environmental devastation caused by human tourists, and it would likely be similar with images showing destructive lunar mining. There is also the perception that resources on the Moon are finite. Although lunar regolith, from which metals will be extracted, is abundant, polar volatiles are a limited commodity that must not be wasted. In modern terrestrial mining, remediation and end-of-life planning are part of the initial planning for the mine and are determined before the mine is even established. This is a good model to follow for space mining. As we start to plan what large-scale lunar mining might look like, we also need to plan for how the environment should look after we’re done. What do we do with old equipment? How should mine tailings be dealt with? To what extent should the environment be restored to its original form when life is not a consideration? Should excavation sites be repurposed for other human activities, or should the environment be restored? Questions such as these should be considered, by academics, industry, governments, and cultural representatives, alongside technical considerations of mining on the Moon.

6. Conclusions As climate change worsens for humans around the globe, the need for technology to ease the impacts of global warming will increase. A planetary sunshade could provide a global, flexible, long-term way to reduce incoming solar energy to control global warming while we work to remove carbon from the atmosphere. The sunshade could also provide a unifying goal to achieve human expansion into space; it connects the dream of expanding into space with the imperative to protect and restore Earth’s environment. With climate change driving environmental challenges not widely foreseen in the 1970s, now is the time to update the O’Neillian vision of using the resources of space to improve life for everyone on Earth. As we take our first steps toward making space resource utilization

© ASCE

420

Earth and Space 2022

technologies viable, we should lay the groundwork to achieve the scale necessary for the grander architectures that can make expansion into space benefit Earth’s environment and the whole of humanity.

7. Acknowledgements The authors would like to thank all the attendees at the 100MT Space Resources Workshop for their time, enthusiasm, and expertise.

8. References Agosto, W.N. (1984) Electrostatic Separation and Sizing of Ilmenite in Lunar Soil Simulants and Samples. Lunar and Planetary Science XV, pp 1-2. https://ui.adsabs.harvard.edu/abs/1984LPI....15....1A Brophy, J. R., Friedman, L., & Culick, F. (2012). Asteroid retrieval feasibility. IEEE Aerospace Conference Proceedings, April, 1–51. https://doi.org/10.1109/AERO.2012.6187031 Environmental Protection Agency (2022). Impacts of Climate Change. https://www.epa.gov/climatechange-science/impacts-climate-change Fix, S. (2021) Feasibility Study of a Sunshade in the Vicinity of the Sun Earth L1 Lagrange Point. IRS-21-S-006. Institute of Space Systems, University of Stuttgart. https://www.planetarysunshade.org/publications Heiken, G. H., Vaniman, D. T., & French, B. M. (1991). Lunar Sourcebook. Cambridge University Press, 58(12), 778. https://doi.org/10.1017/CBO9781107415324.004 IPCC, 2021: Summary for Policymakers. In: Climate Change 2021: The Physical Science Basis. Contribution of Working Group I to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change [Masson-Delmotte, V., P. Zhai, A. Pirani, S.L. Connors, C. Péan, S. Berger, N. Caud, Y. Chen, L. Goldfarb, M.I. Gomis, M. Huang, K. Leitzell, E. Lonnoy, J.B.R. Matthews, T.K. Maycock, T. Waterfield, O. Yelekçi, R. Yu, and B. Zhou (eds.)]. Cambridge University Press. In Press. Jehle, A., Scott, E., & Centers, R. (2020). A planetary sunshade built from space resources. Accelerating Space Commerce, Exploration, and New Discovery Conference, ASCEND 2020. https://doi.org/10.2514/6.2020-4077 Kornuta, D., Abbud-Madrid, A., Atkinson, J., Barr, J., Barnhard, G., Bienhoff, D., ... & Zhu, G. (2019). Commercial lunar propellant architecture: A collaborative study of lunar propellant production. Reach, 13, 100026. Maheswaran, T. (2021). Analysis of Logistical Construction Aspects of a Sunshade Concept in the Vicinity of the Sun Earth L1 Lagrange Point. IRS-21-S-017. Institute of Space Systems, University of Stuttgart. https://www.planetarysunshade.org/publications McInnes, C. R. (1999). Solar Sailing. In Springer-Verlag Berlin Heidelberg GmbH. Springer London. https://doi.org/10.1007/978-1-4471-3992-8 McInnes, C.R. (2002) Minimum mass solar shield for terrestrial climate control. Journal of the British Interplanetary Society, 55, pp. 307-311.

© ASCE

421

Earth and Space 2022

Pettit, D. (2012). The Tyranny of the Rocket Equation. https://www.nasa.gov/mission_pages/station/expeditions/expedition30/tryann y.html Rasera, J. N., Cilliers, J. J., Lamamy, J. A., & Hadler, K. (2020). The beneficiation of lunar regolith for space resource utilisation: A review. Planetary and Space Science, 186(February), 104879. https://doi.org/10.1016/j.pss.2020.104879 Sánchez, J. P., McInnes, C. R., & Marchis, F. (2015). Optimal sunshade configurations for space-based geoengineering near the Sun-Earth L1 point. PLoS ONE, 10(8), 1–25. https://doi.org/10.1371/journal.pone.0136648 United Nations (2021). Climate Change. https://www.un.org/en/globalissues/climate-change Weber, R. C. (2014). Interior of the Moon. In Encyclopedia of the Solar System (Third Edition). Elsevier. https://doi.org/10.1016/b978-0-12-415845-0.000244

© ASCE

422

Earth and Space 2022

SMART: Instrumented Drill for ISRU Investigations on the Moon Leo Stolov1, Kris Zacny1, Jennifer Heldmann2, Kathryn Bywaters1, Sofia Kwok1, Carter Fortuin1, Anthony Colaprete2, Arwen Dave2, Richard Elphic2, Dayne Kemp2, Keith B. Chin3 1

Honeybee Robotics; 2408 Lincoln Ave, Altadena, CA 91001; PH (626) 314-8849; email: [email protected] 2 NASA Ames Research Center, Division of Space Sciences and Astrobiology, Planetary Systems Branch, Moffett Field, CA 94035 3 Jet Propulsion Laboratory, California Institute of Technology; 4800 Oak Grove Drive, Pasadena, CA 91109 ABSTRACT Honeybee Robotics is developing a next generation drilling system called SMART (Sensing, Measurement, Analysis, and Reconnaissance Tool) for lunar in-situ resource utilization (ISRU) applications. SMART is a rotary percussive drill mounted on a linear stage, similar to The Regolith and Ice Drill for Exploration of New Terrains (TRIDENT) that is flying to the Moon in 2022 and 2023. Unlike TRIDENT, which uses the auger to move drill cuttings up to the surface for analysis, the SMART auger and bit assembly is integrated with instruments that can perform analysis in situ. By instrumenting the auger, we are changing the paradigm of exploration – we are bringing an instrument to the sample as opposed to bringing the sample to an instrument. The auger will now be an instrument with sample analysis capabilities. Honeybee Robotics has designed and fabricated a prototype for SMART and is adapting an existing rotary percussive drill called RANCOR to be used to conduct tests in a relevant lunar environment.

© ASCE

423

Earth and Space 2022

INTRODUCTION In-situ resource utilization (ISRU) on the lunar surface is key to a sustained human presence on the Moon. Water can be used directly for agriculture and life support or indirectly through electrolysis into hydrogen and oxygen that can power fuel cells and rocket engines. The presence of water ice and other useful volatiles on the lunar poles has been confirmed by previous missions such as LCROSS and LRO (Colaprete et al., 2020). However, more in-depth knowledge of the location, quantity, and composition of usable resources is vital prior to any mining or ISRU activities. NASA has already planned future missions to lunar poles in the form of Commercial Lunar Payload Services (CLPS) that will fly landers and rovers to the lunar surface. Among these is the VIPER rover, which is flying to the Moon in 2023. VIPER will map and survey the lunar south pole with the use of several science instruments including the TRIDENT drill (Figure 1).

Figure 1. Rendering of the VIPER rover on the Moon with the TRIDENT drill from Honeybee Robotics. Image credits: Honeybee Robotics and NASA TRIDENT excavates a cutting pile that is analyzed by the instruments onboard VIPER. SMART (Sensing, Measurement, Analysis, and Reconnaissance Tool) is the natural evolution of this concept, as it packages the sensors into the auger to bring the instruments to the sample instead of bringing the sample to the instruments. The sensors are miniaturized and selected to characterize the lunar surface, search for water ice, and answer other scientific questions about the lunar poles. SMART allows missions to make educated and expeditious decisions as to whether the downhole soil sample should be delivered to any rover mounted ISRU instruments (e.g., gas chromatograph mass spectrometer) for further analysis or processing. This paper discusses the design of SMART, the existing development of each integrated instrument, and planned future work.

© ASCE

424

Earth and Space 2022

SMART OVERVIEW SMART consists of several major subsystems: an instrumented drill string, a slip ring and optical connector section, a rotary-percussive drill head, a fiber optic rotary joint (FORJ), and a linear stage (Figure 2). The instrumented auger and bit contain a suite of sensors that characterizes the drilled borehole. The electrical and optical signals from the sensors are passed through the slip ring and FORJ sections to an avionics box. The rotary-percussive drill head provides the auger torque and percussion necessary to drill through lunar regolith. Honeybee Robotics has adapted one of its own existing drills called RANCOR to be used as a prototype capable of conducting tests in a relevant lunar environment. The linear stage assembly is a close copy of the TRIDENT linear stage and is used to provide preload and advance the drill into the subsurface.

Figure 2. System diagram of SMART. Not pictured is an avionics box that controls the motors and powers and reads each instrument. SMART is instrumented with five sensors in a 5 cm (2 inch) diameter auger and bit assembly: (1) near infrared spectrometer for volatiles and mineralogical information, (2) neutron spectrometer for hydrogen detection, (3) dielectric spectroscopy probe for electrical properties, (4) temperature sensor and heater for thermal gradient and thermal conductivity measurements, and (5) camera for visible light images and surface texture (Figure 3). The drill is also an instrument, as drilling power and penetration rate can be used to determine regolith strength.

© ASCE

425

Earth and Space 2022

Figure 3. Diagram of the SMART drill bit and auger. Front view of the drill bit (left). Section view of the auger internals (right). At a length of 18 cm (7 inches), the prototype SMART auger houses all the instruments (Figure 4). Future designs can easily add length to the auger depending on the desired drill depth. For ease of assembly, the auger body is split into 2 pieces along the length of the drill. Similar to TRIDENT, the SMART drilling design uses a full face drill bit with a single cutter. The drill bit body mounts to the front of the auger and has 4 windows for the sensors. Each window is extruded slightly above the face of the drill bit to prevent cutting build up.

Figure 4. Photo of the prototype SMART auger and bit during assembly with a quarter for scale. The polycarbonate optical window shown will be replaced with a sapphire window.

© ASCE

426

Earth and Space 2022

SMART INSTRUMENTS Near Infrared Spectrometer. The near infrared spectrometer onboard SMART characterizes the borehole composition by illuminating a target and analyzing the spectra produced by the reflection. This data is used to look for signs of water ice and other volatiles in the lunar regolith. The optical components are selected to function in the visible light and near infrared frequency range (400-2200 nm). Both the light source and reflected signal are passed through a multimode optical fiber that is terminated near the front of the drill bit. The fiber is protected by a sapphire window that slightly protrudes above the drill bit body (Figure 4). The fiber runs through the center axis of the auger, slip ring, and drill bit body all the way to a FORJ, which allows it to rotate with the auger while passing the signal to the static end with minimal attenuation (Figure 5). On the static end of the FORJ, the optical path is divided through a fiber splitter that connects to both a light source and a spectrometer. While half of the returning signal is lost in this configuration, it allows the use of only one fiber for both source and signal that simplifies and compacts the integrated auger design.

Figure 5. Labeled CAD diagram of the SMART prototype. The reflected light from the borehole enters the optical fiber through the window on the drill bit and travels through the FORJ to get to the spectrometer. This is not the first time near infrared spectrometers are being used for space robotic drilling missions. The Near InfraRed Volatiles Spectrometer System (NIRVSS) mounted on the VIPER rover will analyze the cuttings pile generated by the TRIDENT drill (Figure 6). ESA’s ExoMars Rover, which is flying to Mars in 2022, also has a drill integrated near infrared spectrometer called Mars Multispectral Imager for Subsurface Studies (Ma_MISS) (Figure 6). Ma_MISS also passes a fiber optic cable through a drill stem and feeds the optical signal to a rover-mounted spectrometer with the use of a FORJ. There are a couple key difference from the optical design on SMART: (1) the Ma_MISS optical window is mounted to the side of the auger and faces the borehole wall rather than the bottom, and (2) the source light is provided by a separate lamp that is mounted inside the auger body rather than through the same fiber (De Sanctis, 2017).

© ASCE

427

Earth and Space 2022

Figure 6. Labeled image of the Near InfraRed Volatiles Spectrometer System (NIRVSS) that will fly on the VIPER rover to the Moon (left). The Ma_MISS optical window mounted on the ExoMars drill auger. Neutron spectrometer. The drill-integrated neutron spectrometer, which has been developed at NASA Ames, is located in the center of the auger (Figure 3). The spectrometer consists of a scintillator, which emits light when hit by neutrons. Photomultipliers on either end of scintillator pick up the resulting light and convert it to electrical signals. The electrical power and data signals are passed through a slip ring to a power supply and data processing unit. The spectrometer is able to pick up neutrons in the thermal and epithermal range, which are produced in the top few meters of the lunar surface primarily by interaction with galactic cosmic rays (McKinney et al., 2006). A hydrogen rich material, such as lunar regolith containing water ice, will have an increase in thermal neutrons compared to a non-hydrogenated regolith (Feldmann et al., 1998). The auger is made from titanium, a neutron-permeable material that allows neutrons to hit the scintillator. SMART will dwell in the borehole to allow the spectrometer to detect the weight percent (wt%) of water in a soil column. The confidence of the measurement depends on the neutron flux, size of the scintillator, and dwell time. Neutron flux from an Apollo 17 neutronlogging experiment indicates that SMART will need to dwell in the borehole for up to 12 minutes to obtain a measurement with a 3-sigma confidence (McKinney et al., 2006). This drill-based, miniature neutron spectrometer is built on existing work on the larger Neutron Spectrometer System (NSS), which is mounted to the VIPER rover (Figure 7). For comparison, the detectors on the NSS are sized to achieve a detection sensitivity of 0.5 wt% at 3-sigma while roving at 10 cm/s (Colaprete et al., 2020).

© ASCE

428

Earth and Space 2022

Figure 7. Neutron Spectrometer System (NSS) that will fly to the Moon in 2023 mounted on thee VIPER rover. The drill-based design has been successfully tested with a neutron source inside regolith with varying wt% water. In addition, the design has been developed and flown to orbit on a NASA CubeSat program under the name Intrepid (Figure 8).

Figure 8. Intrepid, the compact neutron spectrometer initially developed to fit in a drill auger. The design has been developed for flights on a NASA CubeSat program (D. Kemp, A. N. Nguyen. 2020) Dielectric Spectroscopy Probe. Dielectric spectroscopy uses two electrodes to apply a small electric field across a sample to detect polarizable electrons. An impedance spectrometer measures impedance over a frequency range and relates this to sample permittivity and conductivity. This measurement can be used to quantify and characterize electrochemically active materials such as water ice in regolith. Lab measurements have shown that the technique can measure ice content in a soil greater than 0.5 wt%, though the results depend on sample temperature, particle size, and ion concentration (Buehler et al., 2007).

© ASCE

429

Earth and Space 2022

The SMART dielectric spectroscopy probe consists of two stainless steel electrodes that are mounted to the face of the drill bit (Figure 3). Each electrode is electrically insulated from the drill bit body using a PTFE housing. Signal wires are connected to each electrode and run up through the slip ring to an impedance spectrometer. The design requires laboratory calibration to account for the unique geometry and distance between the electrodes. This technology has been tested in the field onboard the AXEL rover system just south of the town of Mojave in southern California (Figure 9) (Chin et al., 2020). The two prototypes tested (Figure 10) effectively quantified the water content of the field sites. This result shows that the dielectric spectroscopy technique is robust among various geometrical designs.

Figure 9 The AXEL rover system deployed near Mojave, California to test prototype dielectric spectroscopy probe designs.

Figure 10. The two dielectric spectroscopy probes tested in the field onboard the AXEL rover. The donut design prototype (left). The sandwich design (right). Camera. The camera is located alongside the near infrared sensor looking through the same sapphire window (Figure 3). The camera head measures just 4.2 mm in diameter by 13 mm in length and houses the lens, high intensity white LEDs, and an integrated CMOS chip. The power and data signals are passed through the slip ring to a computer, allowing the camera to provide live video and still images inside the borehole. The slip ring is selected to be able to transmit the differential signal over a USB 2.0 protocol. Visible light images of the borehole inform on surface texture, cutting geometry, and give geological context to the data from all the other instruments. While the camera for the SMART prototype was chosen mainly for its size and lower cost, more flight-like miniature cameras already exist, such as the ScoutCam 8.0 HD camera head that was operated successfully on NASA’s Robotic Refueling Mission 3 (Lloyd, Yan, 2020).

© ASCE

430

Earth and Space 2022

Figure 11. Cross Section of the SMART auger and bit, showing the thermal probe, heater, near infrared fiber, and the camera. Temperature Sensor and Heater. The regolith temperature sensor is a cryo-rated platinum RTD mounted inside an aluminum probe that protrudes slightly above the drill bit body (Figure 11). The probe is thermally insulated from the rest of the drill with a G10 housing. The sensor measures the temperature of the lunar regolith at the bottom of the borehole. Independent temperature measurements of the regolith can be used to (1) assess the likelihood of volatile presence in the solid state on the lunar surface, and (2) provide a correction for thermal emissions contributions to the near infrared measurement (Ennico-Smith et al., 2020). Multiple temperature measurements over depth determine the thermal gradient of the top layers of lunar regolith. Another RTD and a 150 W heater are integrated near the center of the auger (Figure 11) to measure thermal conductivity of the regolith. While the auger is fully in contact with the borehole, the heater is turned on for a brief period and the temperature rise is monitored by the RTD. The rate of change is used to determine thermal conductivity, i.e., a slower rate of change indicates a poor thermal path and a lower regolith conductivity (Nagihara, 2012). The integrated heater can also be used to: (1) warm components up to a required operation temperature and (2) can heat the drill out should it get stuck in ice-cemented regolith. The combination of thermal gradient and thermal conductivity measurements are used to characterize the geothermal heat flow of the Moon (Nagihara et al., 2020). Heat flow instruments are not new to lunar missions; Apollo 15 and 17 successfully obtained measurements near the lunar equator. Honeybee Robotics is also developing a heat

© ASCE

431

Earth and Space 2022

flow probe named LISTER, which is flying to the Moon in 2022 (Figure 12). SMART will be able to drill through the ice-cemented regolith expected at the lunar poles to add to these heat flow measurements.

Figure 12. LISTER, a pneumatic drill developed by Honeybee Robotics flying to the Moon in 2022. The tip in instrumented to measure heat flow on the Moon. Photo credit: Honeybee Robotics. Drilling system. The drilling system, which consists of the drill head and linear stage, is also an instrument. Telemetry from the actuators output drilling power, preload, and penetration rate. This data is used to determine regolith strength; e.g., an ice-cemented regolith is harder to drill through than loose powder. For the prototype system, Honeybee has adapted an existing drill called RANCOR to be used as the drill head (Figure 13). The RANCOR drill was originally designed for a Mars sample return mission: it is a coring drill that can cache samples by breaking off the rock core and swapping out the bit (Paulsen et al., 2014). By removing the core breakoff and bit swap capabilities, SMART can pass a fiber optic cable through the center axis without any obstruction to the FORJ mounted at the top of the drill head.

Figure 13. Honeybee Robotics RANCOR drill, developed for Mars Sample Return. The drill head has been adapted for the SMART prototype. The auger and percussion mechanisms are coupled together and driven by the same actuator. The original RANCOR actuator provides up to 3.1 Nm of continuous torque © ASCE

432

Earth and Space 2022

while also running a percussion mechanism sized to provide 0.63 Joules/blow at a rate of 4.6 blows/rotation. Since the drill head was originally sized for a 1.9 cm (0.75 inch) coring drill rather than SMART’s 5 cm (2 inch) full face drill, the drill actuator is swapped from a 100 Watt to a 200 Watt motor. HARDWARE AND FUTURE TESTING Honeybee Robotics is building the SMART prototype to demonstrate instrument functionality (Figure 14). GSE has been put together to support testing (Figure 15). The software used to run the system will reference the lessons learned and operations written for the TRIDENT drill and other drilling systems developed by Honeybee. SMART will first undergo individual instrument tests to show functionality before being tested in a relevant lunar environment using water doped lunar simulant. The near infrared spectrometer will be tested using an existing system called Near Infrared Spectrometer for the Surge Tank (NIRST), which is a portable near infrared spectrometer and light source. The integrated neutron spectrometer will be tested using a neutron source at NASA Ames Research Center.

Figure 14. The SMART prototype drill mounted to the linear stage.

© ASCE

433

434

Earth and Space 2022

Figure 15. The GSE built up around SMART to support testing. CONCLUSION Honeybee Robotics is developing a robotic drilling system with instruments integrated into the drill string for ISRU investigations on the Moon. Sensing, Measurement, Analysis, and Reconnaissance Tool (SMART) is designed to characterize the volatile content on the Moon and answer additional geological questions about the lunar surface and subsurface. SMART is instrumented with five sensors in a 5 cm (2 inch) diameter auger and bit assembly: (1) near infrared spectrometer for volatiles and mineralogical information, (2) neutron spectrometer for hydrogen detection, (3) dielectric spectroscopy probe for electrical properties, (4) temperature sensor and heater for thermal gradient and thermal conductivity measurements, and (5) camera for visible light images and surface texture. The drill is also an instrument, as drilling power and penetration rate can be used to determine regolith strength. SMART allows missions to make to make educated and expeditious decisions as to whether the downhole soil sample should be delivered to any rover mounted ISRU instruments (e.g., gas chromatograph mass spectrometer) for further analysis or processing. SMART can be mounted to a lander, rover, or even be adapted as a handheld system for the high grading on the lunar surface as part of the Artemis program. A prototype for SMART is being assembled with the goal of demonstrating instrument functionality and testing in a relevant lunar environment.

© ASCE

Earth and Space 2022

ACKNOWLEDGMENTS This work is funded by NASA SSERVI under the RESOURCE (Resource Exploration and Science of OUR Cosmic Environment) contract. REFERENCES A. Colaprete, D. Lim, K. Ennico-Smith (2020). “Volatiles Investigating Polar Exploration Rover. NASA Technical Report”. Available at: https://ntrs.nasa.gov/api/citations/20210015009/downloads/20210015009%20 -%20Colaprete-VIPER%20PIP%20final.pdf P. Chu, K. Zacny, V. Vendiola, J. Quinn, J. Kleinhenz and TRIDENT/VIPER team (2020) “TRIDENT Drill for VIPER and PRIME1 Missions. LSIC 2020 Fall Meeting”. Available at: https://lsic.jhuapl.edu/uploadedDocs/posters/401Poster%20PDF_8-Chu.pdf K. Ennico Smith, A. Colaprete, R. Elphic, J. Captain, J. Quinn, and K. Zacny. (2020) “The Volatiles Investigating Polar Exploration Rover Payload”. In LPSC 2020, The Woodlands, Texas. Available at: https://www.hou.usra.edu/meetings/lpsc2020/eposter/2898.pdf De Sanctis, M. C., Altieri, F., Ammannito, E., Biondi, D., de Angelis, S., Meini, M., Mondello, G., Novi, S., Paolinetti, R., Soldani, M., Mugnuolo, R., Pirrotta, S., Vago, J. L., & The Ma_MISS Team. (2017). “Ma_MISS on ExoMars: Mineralogical Characterization of the Martian Subsurface”. Astrobiology, 17(6–7), 612–620. https://doi.org/10.1089/ast.2016.1541 McKinney, G. W., Lawrence, D. J., Prettyman, T. H., Elphic, R. C., Feldman, W. C., & Hagerty, J. J. (2006). MCNPX benchmark for cosmic ray interactions with the Moon. Journal of Geophysical Research, 111(E6). https://doi.org/10.1029/2005je002551 D. Kemp, A. N. Nguyen. (2020) “Intrepid Particle Detector”. Available at: https://flightopportunities.ndc.nasa.gov/media/technology/260/257-summarychart.pdf [Accessed 9 Nov. 2021] Feldman, W. C., Maurice, S., Binder, A. B., Barraclough, B. L., Elphic, R. C., Lawrence, D. J. (1998). “Fluxes of fast and epithermal neutrons from lunar prospector: Evidence for water ice at the lunar poles”. Science, 281(5382), 1496–1500. https://doi.org/10.1126/science.281.5382.1496 M. G. Buehler, K. B. Chin, S. Seshadri, D. Keymeulen, R. C. Anderson and T. A. McCann, "Electrical Properties Probe Measures Water/Ice Content of Martian Soils Using Impedance Spectroscopy". 2007 IEEE Aerospace Conference, 2007, pp. 1-19. https://doi.org/10.1109/AERO.2007.352776 Chin, K. B., Anderson, R. C., Schoelen, D. A., Kerber, L. A., McGarey, P., Paton, M., Brown, T. L., & Nesnas, I. A. (2020). “Enabling new exploration opportunities on planetary surfaces: In situ geochemical characterization in soils by dielectric spectroscopy onboard the Axel Rover System.” Planetary and Space Science, 187, 104948. https://doi.org/10.1016/j.pss.2020.104948 V. Lloyd, I. Yan. (2020) “NASA’s Refueling Mission Completes Second Set of Robotic Tool Operations in Space.”

© ASCE

435

Earth and Space 2022

https://www.nasa.gov/feature/goddard/2020/nasa-s-refueling-missioncompletes-second-set-of-robotic-tool-operations-in-space [Accessed November 10, 2021] S. Nagihara, P. Ngo, V. Sanigepalli, M. Zasadzien, L. Sanasarian, G. Paulsen, D. Sabahi, and K. Zacny (2020) “The Heat Flow Probe for the Commercial Lunar Payload Services Program Of NASA.” In LPSC 2020, The Woodlands, Texas. Available at: https://www.hou.usra.edu/meetings/lpsc2020/pdf/1432.pdf S. Nagihara, K. Zacny, M. Hedlund, and P. T. Taylor. (2012). “Development of a Compact, Deep-penetrating Heat Flow Instrument for Lunar Landers: In-situ Thermal Conductivity System.” International Workshop on Instrumentation for Planetary Missions, 2012. Available at: https://ntrs.nasa.gov/citations/20120013638 Paulsen, G.L., Indyk, S., & Zacny, K. (2014). “Development of the RANCOR Rotary-Percussive Coring System for Mars Sample Return.” Proceedings of the 42nd Aerospace Mechanisms Symposium, NASA Goddard Space Flight Center, May 14-16, 2014 Available at: https://esmats.eu/amspapers/pastpapers/pdfs/2014/paulsen.pdf

© ASCE

436

437

Earth and Space 2022

Influence of Ice Distribution on Thermal Mining Performance and Strategies to Counter Sublimation Lag T. Gordon Wasilewski Astronika Sp. z o.o., Warsaw, PL ABSTRACT The unclear nature of ice deposits within the lunar permanently shadowed regions poses many challenges to prospecting and utilisation of those resources. Recent recalculations of LCROSS data suggest that ice concentration is more likely depth-dependent, which may also be similar to Martian ice tables. Heterogeneous deposits have to be analysed in the frame of thermal mining architecture to inform us about the expected performance and drawbacks of the technology, which currently is the most promising method of ice extraction beyond Earth, able to produce multiple tons of water per extraction. In this paper, a reworked combined heat and mass transfer model is utilized to provide further insights into phase change dynamics, when beamed energy is applied to a heterogeneous instead of a homogeneous deposit. New strategies are shown to counter performance loss due to sublimation lag build-up and bulk thermal conductivity decrease. Strategies include continuous thermal mining and fracking-thermal mining methods. The first method is analysed with a combined heat and mass transfer model and compared to the expected outcomes in the original thermal mining method. Bulk granular ice is also introduced in the modelling to show thermal mining performance in a more likely scenario of porous ice deposit. RESEARCH Ice extraction on the Moon holds the promise for a sustainable human presence in the cislunar space. Tapping to those resources could enable space-derived rocket propellant production, as well as supply of critical mission consumables to lunar and orbital bases. Not only is that important from a mission perspective but also from a commercial one, as lunar thermal mining has a business case (Sowers, 2021), and fulfils all three principles of space resources: i.e. (1) an existing resource with (2) a technology to recover it, and (3) a customer (Duke, 1999). In Wasilewski et al. (2021) and Wasilewski (2021), phase change interface movement is analysed during heating of icy deposits respectively on small experimental, and large theoretical scales. Focus on the interface movement stems from terrestrial resource industry analogies, particularly in oil and gas. In pressure-driven systems on Earth, we are often looking for drainage radius to design, manage and forecast production. In the case of lunar thermal mining, a thermal conductivity-driven system, we should investigate the sublimation front position and movement for similar purposes. This paper expands previous models and investigates production in an even more variable, heterogenous icy deposit in Permanently Shadowed Regions conditions. The first major change compared to Wasilewski (2021) is depth-dependence of ice content (Luchsinger, et al., 2021) and bulk icy regolith density, which adds anisotropy of material properties already in initial conditions. The change is visualised in Figure 1.

© ASCE

438

Earth and Space 2022

Figure 1. High-level comparison of models. The additional difference stems from a major problem identified before. During thermal mining, build-up of a sublimation lag should occur, which should significantly dampen water production. This is because bulk thermal conductivity significantly decreases after sublimation, with sublimation lag acting as a thermal insulator in airless conditions. The solution to this problem might be sublimation lag removal. This paper incorporates continuous lag removal during the heating process, introducing a continuous thermal mining method. Continuous thermal mining allows for tapping to increasingly richer icy deposit, when ice is positively depth-dependent. That way, we can always target a high-thermal conductivity deposit, increasing overall production of water. A possible realisation of this mechanism is shown in Figure 2.

Figure 2. Sublimation lag removal in the model obtained with mesh movement.

© ASCE

439

Earth and Space 2022

A seemingly alternative method to keep high thermal conductivity in the deposit under extraction can be called fracking thermal mining. Such a method should rely on the injection of high-TC materials that do not undergo phase change under thermal mining conditions. An example of such material may be alumina nitrate powder, for which aluminium can be sourced locally from the reduction of anorthite, a mineral abundant in Highlands regoliths. The material could be injected pneumatically or explosively. More research is needed to quantify the results of both methods and analyse their feasibility. In this paper, a special focus is given to continuous thermal mining and increased variability stemming from geologic factors – depth-dependent bulk density, depth-dependent ice content and lowered ice density (granular and excess ice). Table 1 summarizes the main initial parameters, while Table 2 summarizes model assumptions in this paper. Table 1. Initial parameters, baseline in the previous model Parameter Initial temperature Sublimation temperature Transition interval Surface icy density Basalt density Solid ice density Surface ice content Heating rate Extraction area Constant pressure under extraction system

Unit K K K kgm-3 kgm-3 kgm-3 % wt Wm-2 m2

Value 40 273 100 1500 2800 920 5 1368 650

Torr

2·10-12

Similarly to (Wasilewski, 2021), phase change interface position and movement is calculated as: 𝑟(𝑡) = (1 − 𝜃̅) ∙ 𝑑𝑖 where 𝜃̅ is a volume-averaged icy fraction indicator in an icy deposit with thickness d i. The position is calculated through a combined heat and mass transfer FEM model. With continuous lag removal, where spatial and material coordinates do not intersect during the process, the interface position must be corrected to: ̅ ) ∙ 𝑑𝑖 𝑟𝑐𝑜𝑛𝑡 (𝑡) = (𝜃̅ − 𝐽𝑚𝑒𝑠ℎ ̅ where 𝐽𝑚𝑒𝑠ℎ is mesh volume in a deposit with thickness d i. At t=0, its value is 1 and is subject to decrease as phase change interface moves to the subsurface. Derivatives of these equations yield interface velocity and acceleration, which are helpful to investigate and forecast water production. In Wasilewski (2021), interface movement is translated to cumulative production of water with: 𝑄(𝑡) = 𝐴𝑒𝑥𝑡𝑟 ∙ 𝑟(𝑡) ∙ 𝜌𝑖𝑐𝑦 ∙ 𝐶𝑖 with icy regolith density (𝜌𝑖𝑐𝑦 ) and ice concentration (𝐶𝑖 ) having uniform initial values. Heterogeneity is however introduced in this paper through a solution shown below.

© ASCE

440

Earth and Space 2022

Table 2. Model assumptions and scenarios A-F. A. Previous model

B. Add bulk regolith depthdependence

C. Add ice content depthdependence

D. Add lowered ice density

E. Add continuous lag removal

F. Combine all

Time dependent with variability of material properties: ▪ Thermal conductivity (k) as f(T, p, φ, F, θ) ▪ Heat capacity (Cp) as f(T, φ, F, θ) ▪ Permeability (κ) as f(φ, F, θ) Volumetric mixing of ice and regolith in a homogenous deposit.

Initial conditions corrected for exponential increase of bulk density (ρicy):

Initial conditions corrected for exponential increase of ice content (Ci):

Ice density lower than normal i.e., less than 920 kgm-3.

Mesh (material) movement rate with an increased average velocity of phase change interface and limits:

Continuous thermal mining in depthdependent deposit: A+B+C+D+ E

𝐶𝑖 = 𝐶𝑖_𝑠𝑢𝑟𝑓𝑎𝑐𝑒 ∙ 𝑒 𝑧/𝑅𝑐

This may yield pore filling fractions F >> 1, which may need to be corrected (excess ice).

As in (Wasilewski, 2021) and (Wasilewski, et al., 2021)

Similarly to Apollo data for dry regoliths, e.g. Carrier et al. (1991). Hyperbolic or power could also be applied

𝜌𝑖𝑐𝑦 = 𝜌𝑖𝑐𝑦_𝑠𝑢𝑟𝑓𝑎𝑐𝑒 ∙ 𝑒 𝑧/𝑅𝑑 with exponential parameter Rd < 10 to yield close Apollo results, when surface density is 1500 kgm-3, and limit at 3 meters.

with exponential parameter Rc < 2 to yield high concentration ice at depth 3-5 meters starting from 5% ice, and limit at 5 meters. Latent heat of sublimation also scales with depth. A possible realisation of recalculated LCROSS data (Luchsinger, et al., 2021)

This model

𝑣𝑚𝑒𝑠ℎ = 𝑣̅𝑖𝑛𝑡𝑒𝑟𝑓𝑎𝑐𝑒 ∙ 𝐼𝑐 𝑓𝑜𝑟 θ = 0 𝑣𝑚𝑒𝑠ℎ = 0 𝑓𝑜𝑟 θ >0 based on icy fraction indicator θ and assuming we can remove fully desiccated regolith. Interface constant (Ic) higher than 4 yields roughly constant velocity.

This model

This model

With depth-dependence, translation from interface position (baseline or continuous) to cumulative water production requires evaluation over density and ice content at depth r(t): 𝑟(𝑡)

𝑄(𝑡) = 𝐴𝑒𝑥𝑡𝑟 ∙ 𝑟(𝑡) ∫

𝜌𝑖𝑐𝑦 (𝑧)𝐶𝑖 (𝑧)𝑑𝑧

0

Evaluation of the model in such form with scenarios from Table 2 showed interesting results. In timeframes beyond thermal mining dwell time (44 hrs per extraction, ca. 150,000 s), total yields are shown in Figure 3. Significant improvements in cumulative water production are shown across all scenarios compared to baseline scenario A. Both depth-dependence factors and continuous thermal mining method contribute to increased yields. The most surprising results are shown however in scenarios containing elements C (ice depth-dependence) and D (lowered ice density).

© ASCE

441

Earth and Space 2022

Figure 3. Scenarios A-F results in the first iteration. The first one can be easily explained with exponential factor R c, which was intentionally low to show a stark contrast in the results, modelling a large growth of ice concentration at depth. However, analysis of increased yields with porous ice showed a significant flaw of the bulk icy regolith density approach, which was a previous baseline. Lowered densities of ice, combined with high ice concentration growth, result in a significant increase of pore-filling fraction F. And since thermal conductivity linearly increases with F, this skews bulk thermal conductivity to very high values. With F significantly larger than 1, ice should increasingly become the rock matrix, instead of pore-filling, and in that view, bulk density must be corrected. The correction is done through: 𝜌𝑖𝑐𝑦𝑐𝑜𝑟𝑟𝑒𝑐𝑡𝑒𝑑 (𝑧) = 𝜌𝑖𝑐𝑦 (𝑧) − 𝐹(𝑧) ∙ 𝜑(𝑧) ∙ 𝜌𝑖𝑐𝑒 and consequent corrections to parameters 𝜌𝑑𝑒𝑠𝑖𝑐𝑐𝑎𝑡𝑒𝑑 (𝑧), 𝜑(𝑧), and 𝐹(𝑧). The correction does not change the results in baseline scenarios, where pore-filling fraction was always lower than 1 and volumetric mixing of ice and dry regolith was a sustainable model. The change does not allow F >> 1.

Figure 4. Corrected results for scenarios A-F. © ASCE

Earth and Space 2022

Again, results in Figure 4 show significant improvements compared to the baseline scenario. Both in the cases of depth-dependence (B, C) and continuous lag removal (E), the yields are multiple times higher than in A, although about 2 times lower than in their counterparts from Figure 3. This time, however, the ultimate production with lag removal is increased, compared to the same scenarios without it. Some differences exist between scenarios with and without lag removal during the primary production phase. This is thought to be a result of the removal of some heat from the system when scenarios E are applied to remove low thermal conductivity regolith. That said, for some conditions and timeframes, an optimal material removal rate v mesh should exist. The problem shown in this paper requires more research, especially given a more faithful representation of the PSR geology and depth-dependence of its main geological parameters. Deposition of ices within PSR regoliths is primarily impact-based and secondarily diffusionbased. This could lead even to ice-dominated mixtures with regolith impurities in the first mechanism, and regolith-dominated mixtures with ice impurities in the second one, which would lead to different geologic scenarios and thermal mining results based on the natures of intramixture and inter-mixture ices. That is why the true results might be in between the values from Figure 3 and Figure 4. Going back to the modelling outcomes from Figure 4, within the timeframes of thermal mining architecture, the lag removal on average increases yields by 1.6 times across the scenarios. The highest influence had ice depth-dependence, which with Rc = 2 increased yields by about 11-12 times across relevant scenarios compared to baseline. Bulk regolith depth-dependence with Rd = 10 increased yields by roughly 60%. While lowered ice density increased yields by about 6%. Non-linearity of yields is expected based on the effects of initial conditions. That is why more detailed research is needed in that field to see the full variability. CONCLUSIONS This paper investigated the influence of depth-dependent geological parameters on thermal mining yields on the Moon, and a new strategy to increase water production. Based on problems encountered in previous modellings (sublimation lag acting as a thermal insulator), two methods of increasing bulk thermal conductivity were proposed – (1) continuous thermal mining with lag removal during the process (operational), and (2) fracking thermal mining with material injection during the process (engineered). The first method was modelled and showed 100% improvements in cumulative water production during thermal mining proposed by Sowers & Dreyer (2019). The second method is part of future work modelling. The continuous lag removal might have an optimum removal rate when it is feasible. This is especially seen in low improvements of production in combined bulk density and ice concentration depth-dependence. With that in mind, continuous thermal mining is very feasible in low-grade icy deposits and loses its benefits in high-grade deposits. The heat of hot desiccated regolith could be transferred to other systems before deposition of regolith in waste/storage areas to circularize the process. Since thermal mining relies on mirrored sunlight, recirculation of desiccated regolith heat could not increase the efficiency of the process, unless conductive rods or other forms of beamed energy are used as secondary sources of heat. In some cases, an increase in thermal conductivity would not compensate for the loss of thermal mass of desiccated regolith.

© ASCE

442

Earth and Space 2022

In any case, thermal mining shows significant yields of water on the lunar surface, which brings great promise for a sustainable human presence within the cislunar space. REFERENCES Carrier III, W. D., Olhoeft, G. R. & Mendell, W., 1991. Physical properties of lunar surface. In: Lunar Sourcebook: A User’s Guide to the Moon. Cambridge: Cambridge University Press, pp. 475-594. Duke, M. B., 1999. The Development of Space Resources. Golden, CO, s.n. Luchsinger, K. M., Chanover, N. J. & Strycker, P. N., 2021. Water within a permanently shadowed lunar crater: Further LCROSS modeling and analysis. Icarus, Volume 354. Sowers, G. F., 2021. The Business Case for Lunar Ice Mining. New Space, 9(2), pp. 77-94. Sowers, G. F. & Dreyer, C. B., 2019. Ice Mining in Lunar Permanently Shadowed Regions. New Space, 7(4), pp. 235-244. Wasilewski, T. G., 2021. Lunar thermal mining: Phase change interface movement, production decline and implications for systems engineering. Planetary and Space Science, Volume 199. Wasilewski, T. G., Barcinski, T. & Marchewka, M., 2021. Experimental investigations of thermal properties of icy lunar regolith and their influence on phase change interface movement. Planetary and Space Science, Volume 200.

© ASCE

443

Earth and Space 2022

Commissioning and Testing a New Dusty Thermal Vacuum Chamber with Inclusion of Icy Regolith Benjamin David Wiegand1; Marcello Guadagno2; and Paul van Susante3 1

Dept. of Mechanical Engineering-Engineering Mechanics, Michigan Technological Univ., Houghton, MI. Email: [email protected] 2 Dept. of Mechanical Engineering-Engineering Mechanics, Michigan Technological Univ., Houghton, MI. Email: [email protected] 3 Dept. of Mechanical Engineering-Engineering Mechanics, Michigan Technological Univ., Houghton, MI. Email: [email protected] ABSTRACT The Planetary Surface Technology Development Lab (PSTDL) purchased a new testing facility in the form of a dusty thermal vacuum chamber (DTVAC). This facility will be used to test the technology readiness level (TRL) of devices intended for extraplanetary use. Environments such as those found on the Moon and Mars will be simulated in the DTVAC via use of vacuum pumps, thermal shrouds, and simulated regolith. This paper details the commissioning and testing of systems put into place to facilitate TRL testing such as data acquisition systems, test fixtures, and baseline DTVAC performance tests. The PSTDL has purchased two insulated shipping containers that will allow for the creation and reclamation of icy regolith. Icy regolith test beds will be used in the DTVAC to test the TRL of water in situ resource utilization (ISRU) devices. INTRODUCTION With NASA’s return to the Moon with the Artemis program there is a need to increase the Technology Readiness Level (TRL) of devices that will be used for in-situ resource utilization (ISRU). The Planetary Surface Technology Lab (PSTDL) hopes to fulfill these needs with its newly launched Dusty Thermal Vacuum Chamber (DTVAC) testing facilities. The DTVAC consists of a large cubic chamber (127x127x177.8cm) that is pumped to vacuum levels with two vacuum pumps. Thermal shrouds within the chamber can cool and heat it to extreme temperatures, and with the inclusion of simulated regolith the DTVAC can create environments similar to those found on the Moon or Mars. Devices can then be tested in the simulated environment to determine their TRL. Performance of the DTVAC and the devices tested is observed using data acquisition systems and videography/photography devices. The DTVAC facility and its components can be seen in Figure 1. To facilitate testing in simulated regolith, the PSTDL has developed and manufactured its own lunar simulant that is representative of the regolith found on the lunar highlands. This simulated regolith can then be processed in a number of ways to create different desired properties. The main motivation for the PSTDL is to combine the regolith with a certain percentage of shaved ice to create icy regolith for testing devices for water ISRU. Two 40 ft. high cube insulated shipping containers have been acquired to allow the PSTDL to create icy regolith, and then reclaim it once testing is concluded.

© ASCE

444

445

Earth and Space 2022

Figure 1. PSTDL DTVAC Facility with Labeled Components Icy regolith test beds will be created in the freezer container so as to ensure that the regolith is below freezing in temperature while it is mixed with shaved ice. The icy regolith bed will then be transported to the DTVAC and used for testing. The DTVAC will bring the test bed to cryogenic temperatures to ensure that the ice stays frozen. Once the test is concluded, the test bed will be brought to the heater container. The then wet regolith will be unloaded into sheet pans and allowed to dry within the container with the help of heated forced air convection currents. Figure 2 shows the CAD models of the two shipping containers.

Figure 2. CAD Models of Freezer and Heater Containers. Freezer container pictured forward. Heater Container pictured in the rear. Heater container contains sheet pans for drying wet regolith. Freezer container contains a frozen regolith test bed, and cart to move the DTVAC test bed.

© ASCE

446

Earth and Space 2022

DTVAC COMMISSIONING Several test fixtures were created before the DTVAC was used for testing. This included regolith bins and transportation/loading carts, as well as a protective cage for the computer and data acquisition system. A regolith bin being loaded into the DTVAC can be seen in Figure 3.

Figure 3. Regolith Bin on top of Cart Being Loaded into DTVAC A data acquisition system was purchased and configured to obtain data concerning devices and the environment within the chamber. For this, a National Instruments cRIO-9035 DAQ system was used and can be seen in Figure 4. Several Modules for the DAQ system were added to the system to allow for data to be collect from a range of sensor types. These modules’ specifications can be seen in Table 1. With this DAQ system, it is possible to obtain digital and analog signals from within the vacuum chamber, including up to twenty thermocouple signals.

Figure 4. NI cRIO-9035 DAQ with Inserted Modules and Thermocouples Connected

© ASCE

447

Earth and Space 2022

Table 1. NI DAQ Modules in Use and Corresponding Specifications Module NI 9344 NI 9230 NI 9375 NI 9218 NI 9213

Purpose 4-Channel C Series Digital User Interface Module 3-Channel, 12.8 kS/s/channel, ±30 V C Series Sound and Vibration Input Module 30 V, 32-Channel (Sinking Input, Sourcing Output), 7 µs (Input)/500 µs (Output) C Series Digital Module 51.2 kS/s/ch, 2-Channel C Series Universal Analog Input Module 16-Channel, 75 S/s Aggregate, ±78 mV C Series Temperature Input Module

To obtain data about the conditions inside the vacuum chamber, National Instruments LabVIEW was used to communicate with the thermal shroud temperature controllers and pressure sensors. This allows for the continued monitoring of the temperatures and pressures of the DTVAC while testing is in progress. Additionally, functions exist that allow for remote altering of the thermal shroud temperature controllers’ set values. With this function, the temperature controllers can be programmed to follow a set temperature profile, enabling testing of devices’ tolerance to changing temperatures. For acquiring visual data, three systems were configured and deployed. The first is a GoPro that is mounted to the exterior of the vacuum chamber and is able to record video or pictures through the two viewing ports on the chamber. To collect visual data from within the chamber, two open source and inexpensive camera systems are used. An ESP32 camera and a Raspberry Pi with a camera attachment were configured and are used to relay video streams back to the DAQ computer over WIFI. The live streams are then displayed and recorded on the computer. This configuration was chosen due to the extreme environments experienced within the chamber would likely destroy any cameras easily available for purchase, thus, inexpensive and expendable cameras were used. Figure 5 shows an image from within the vacuum chamber that was captured by one of the expendable cameras.

Figure 5. Image Captured by ESP32 Cam During Vacuum Testing © ASCE

Earth and Space 2022

DTVAC CHARACTERIZATION For TRL testing the DTVAC must be able to produce conditions similar to environments found on the Moon and Mars. For testing a device that will be used on the Moon, the device must be able to withstand an ultra-high vacuum and temperatures ranging from -173⁰C to 127⁰C while interacting with lunar regolith (Sharp, n.d.). Devices meant to operate in a Martian environment would have to withstand absolute pressures near 5 Torr and temperatures ranging from -125⁰C to 20⁰C (Mars, n.d.; What Is the Temperature of Mars?, n.d.). The DTVAC was tested to ensure that it could replicate the desired environmental conditions. Two tests were conducted to see the maximum vacuum pressures that could be maintained while heating and cooling the DTVAC to its maximum temperatures. Figure 6 and Figure 7 shows the results of the tests. It can be seen that the DTVAC is capable of maintaining pressures below 1E-5 Torr while reaching a temperature of 150⁰C and nearing -196⁰C. During the cooling test the liquid nitrogen (LN2) valve had to be throttled once the temperature of the DTVAC reached -175⁰C because LN2 was exiting the exhaust outlet without evaporating. Although 196⁰C was not observed in this short duration test, it is possible that temperatures nearing the value could be seen in long duration tests with proper LN2 valve throttling. More testing is needed to confirm the minimum temperature the DTVAC is capable of producing.

Figure 6. Pressure and Temperature Profiles for Heating DTVAC to its Max Temperature During the heating testing it was found that the temperature controller for the shroud walls was not adjusted properly, and frequently overshot the desired temperature. The autotune function of the temperature controller was utilized to achieve the desired response. Figure 8 shows the response of the shroud during and after tuning when the desired temperature was set to 100⁰C and -60⁰C. With the shroud temperature controller properly tuned, the performance of the DTVAC was verified to be ample for TRL testing.

© ASCE

448

Earth and Space 2022

Figure 7. Pressure and Temperature Profiles for Cooling DTVAC to its Minimum Temperature

Figure 8. Response of Shroud Walls from Heating and Cooling after Temperature Controller Tuning ICY REGOLITH CREATION The PSTDL will be testing devices that are designed to sample or collect water/ice from ice deposits underneath the moon’s surface (Benna et al., 2019). As such, the frozen regolith found on the Moon needs to be simulated on Earth to test the TRL of the devices before they are to be

© ASCE

449

Earth and Space 2022

used. The PSTDL has developed a procedure for creating icy regolith batches and a 770kg test batch of icy regolith was created in the winter and tested in the DTVAC. Facilities in the form of two insulated shipping containers have been procured and will allow the PSTDL to create icy regolith batches year-round. The procedure used by the PSTDL to create icy regolith involves mixing dry lunar simulant with a percentage by mass of shaved ice. To make the 770kg test batch of icy regolith, 77kg of ice and 693kg of simulant were combined to form a composition of 10% water by weight. A concrete mixer was used to mix roughly 2kg of ice and 20kg of simulant at a time. The mixture was then checked for homogeneity by ensuring there were no visible clumps of unmixed ice or regolith. The mixture was loaded into the regolith cart and gently compacted. Once 770kg of icy regolith was created, the regolith bin was transported to the DTVAC for testing. The regolith bin was loaded into the DTVAC and 10 thermocouples were placed within the icy regolith at varying depths and locations. Figure 9 shows the bin inside the DTVAC once the thermocouples were installed. The DTVAC was then pumped down and the shrouds were set to their minimum temperature of -196⁰C. The test was conducted for 2 hours and reached pressures of 1E-3 Torr and temperatures of -120⁰C until a mechanical issue with the turbomolecular pump occurred. A pressure of 1 Torr was maintained for another 5 hours until the test was concluded.

Figure 9. Icy Regolith Test Bed inside the DTVAC with Installed Thermocouples Thermocouple data from the test was recorded and the data from the portion of the test in which the shrouds were cooling can be seen in F. Using a linear fit of the average of the retrieved

© ASCE

450

Earth and Space 2022

data, it was estimated that it would take 13.25 hours for the average temperature of the bed to reach -196⁰C, however this estimate is based on limited data and is more useful as an insight into the minimum time necessary for future tests.

Figure 10. Thermocouple Data of 770kg Icy Regolith Test during Thermal Shroud Cooling. Dashed lines represent 30cm depth, and solid lines are 5 cm depths from the regolith's surface. The mechanical issues observed during testing were remedied, however, external temperatures low enough for making icy regolith were not persistent, and so another test batch of frozen icy regolith could not be created. This led to the decision to purchase insulated freezer and heater containers so that icy regolith could be created and reclaimed independently of the outdoor climate. Shown in Figure 2 is a CAD image of the future configurations of the freezer and heating containers. Dry regolith will be brought into the freezer container where it will be combined with chosen percentage by weight of shaved ice. The combination will then be mixed till it is homogeneous and then packed into the regolith bin at varying densities. The sub-freezing temperature of the freezer container will ensure that the ice in the icy regolith does not melt, and that the composition of the shaved ice within the regolith will remain consistent. The test bed will then be moved to the DTVAC where the device to be tested will be put on the test bed along with all the connections necessary for data acquisition and power supply. After testing, the test bed will most likely be thawed, and the regolith bin will be transported to the heater container where the wet regolith will be put onto metal sheets on racks. The heater container will heat the air inside to 100⁰C with convection currents, drying out the wet regolith. The dried regolith will then be tested with a convection oven to ensure that the desired water content has been achieved. The dried, reclaimed, regolith will then be put into the freezer container to be used in icy regolith again. Figure 11 shows the freezer and heater containers situated outside of the PSTDL’s lab space.

© ASCE

451

Earth and Space 2022

Figure 11. Freezer and Heater Containers used for Making and Reclaiming Icy Regolith CONCLUSION The PSTDL has commissioned and tested a DTVAC for use in testing the TRL of devices for use in extraplanetary environments. The DTVAC has an internal volume of 127x127x177.8 cm and is capable of reaching temperatures of -180⁰C to 150⁰C and internal pressures below 1E-5 Torr. Test fixtures such as a regolith cart and bin were created to allow for the testing of devices on dry or icy regolith beds. A data acquisition system was developed to obtain data concerning the simulated environment within the chamber, along with parameters concerning the tested devices. Three camera systems were designed and implemented to allow for visual data collection of conducted tests. The DTVAC’s performance was characterized, and the shroud temperature controllers were tuned to enable good responses to the desired temperature inputs. The PSTDL developed a method for creating icy regolith, and a 770kg test batch was created and tested in the DTVAC. Although the test was concluded early due to mechanical issues, data from the test was used to determine the minimum time of 13.25 hours is necessary for the DTVAC to reach pressures below 1E-5 Torr and temperatures nearing -196⁰C. Two 40 ft. high cube insulated shipping containers have been procured by the PSTDL. The freezer container will be used to make icy regolith test bed batches independently of the outdoor weather. A heater container will be used to reclaim the wet regolith after testing has occurred so that the regolith can be mixed with ice again to create icy regolith. This will allow the PSTDL to test the ISRU of water testing/extraction devices meant for use on the Moon.

© ASCE

452

Earth and Space 2022

REFERENCES Benna, M., Hurley, D. M., Stubbs, T. J., Mahaffy, P. R., & Elphic, R. C. (2019). Lunar soil hydration constrained by exospheric water liberated by meteoroid impacts. Nature Geoscience, 12(5), 333–338. https://doi.org/10.1038/s41561-019-0345-3 Mars. (n.d.). Retrieved June 28, 2021, from https://mars.nasa.gov/MPF/mpf/realtime/ mars2.html Sharp, T. (n.d.). What is the Temperature on the Moon? Space.Com. Retrieved June 28, 2021, from https://www.space.com/18175-moon-temperature.html What is the Temperature of Mars? (n.d.). Space.Com. Retrieved June 28, 2021, from https://www.space.com/16907-what-is-the-temperature-of-mars.html

© ASCE

453

454

Earth and Space 2022

Molten Regolith Electrolysis Using Concentrated Solar Heating Hunter Williams1; Timothy Newbold2; Kevin Grossman3; Evan Bell4; Elspeth Petersen5; Jaime Toro Medina6; Jeff Dyas7; and Laurent Sibille8 1

Honeybee Robotics, Altadena, CA. Email: [email protected] Honeybee Robotics, Altadena, CA 3 NASA Kennedy Space Center, Cape Canaveral, FL. Email: [email protected] 4 NASA Kennedy Space Center, Cape Canaveral, FL 5 NASA Kennedy Space Center, Cape Canaveral, FL 6 NASA Kennedy Space Center, Cape Canaveral, FL 7 NASA Kennedy Space Center, Cape Canaveral, FL 8 NASA Kennedy Space Center, Cape Canaveral, FL 2

ABSTRACT Molten regolith electrolysis (MRE) is a space resource utilization technology with the potential to provide a dependable source of oxygen to astronauts in upcoming Artemis missions. Honeybee Robotics and NASA Kennedy Space Center’s SwampWorks Lab have designed and constructed a technology development test bed for the MRE process. The team has used this test bed to develop subsystems related to introducing granular material, melting regolith with a variety of methods, electrolyzing the melt pool, removing byproducts, and regulating these processes to optimize the electrolysis process. Honeybee has specifically focused on a combined Joule heating and concentrated solar heating process to form the melt pool. In this paper we will discuss the background and relevance of MRE, specific aspects of the test bed’s design, test parameters and results, ongoing work, and conclusions based on the work performed. Of specific interest is the alternating use of Joule heating and solar heating as the melt pool forms, oxygen evolves, and melt pool dynamics drive a changing melt pool resistivity. INTRODUCTION The most readily available oxygen source on the Moon is the Lunar regolith itself. It can be harvested and electrolyzed anywhere on the surface, and the regolith contains 41–45% oxygen by mass. Artemis and future in-situ resource utilization (ISRU) missions will likely perform tests on such electrolysis systems, and subsequent mission architectures may rely on them for Lunar sourced oxygen. Molten regolith electrolysis (MRE) is a front runner for near-future Lunar electrolysis technology, ideal for both early subscale testing and future full-scale operations for its scalability, simplicity, and lack of consumables. To melt the regolith, an MRE system may use inductive or resistive heating elements buried in the surface of the regolith without a crucible, or concentrating lenses at the surface to directly harness the heating energy of the Sun. In partnership with Kennedy Space Center’s Swamp Works, Honeybee is pursuing the development of MRE to turn molten rock, similar to lava on Earth, into breathable oxygen. MRE heats regolith to melting then uses a temperature and corrosion-resistant cathode and anode to separate the oxygen from other elements within the melt pool. Melting the regolith before performing the electrolysis reduces the power and amperage requirements to levels that would be available on current Lunar landers. Because MRE can be performed anywhere on the

© ASCE

455

Earth and Space 2022

Lunar surface during the Lunar day and produces oxygen useful for propellant or life support, it is a strong candidate for use as the first space resource utilization technology. Honeybee has developed a test bed for performing MRE on Lunar regolith simulant under partial vacuum using simulated, concentrated solar radiation. Though a variety of other methods have been proposed in MRE architectures, the team at Honeybee Robotics has pursued concentrated solar MRE for its simplicity and low mass requirement. Sunlight is a readily available, predictable, and easily harvested heat source on the Lunar surface. Using a segmented Fresnel lens of around 1m in diameter, temperatures over 1000°C can be achieved at a spot size 5cm in diameter or more during the Lunar day. The Honeybee testbed recreates a Fresnel lens melting system on the Lunar surface using a vacuum chamber with a 15cm fused quartz window, a bed of regolith simulant (JSC-1A, BP-1, NU-LHT, sand, and various simulants from Exolith Labs), and an 8kW xenon arc lamp.

Figure 1. Test bed components. This paper covers work done in the design and test of the test bed as well as of specific components and materials for use in future flight-forward designs. The team performed a trade study on electrode materials and tested various electrode shapes under different power conditions to determine an optimal design for the electrolysis process and ease of refilling regolith and removing byproducts such as silicon and iron. Partially joule heating the regolith directly in

© ASCE

Earth and Space 2022

conjunction with using non-electrical heat sources allows for more complete electrolysis and melting in a constantly changing melt pool. The team experimented with varying amperage to use joule heating at key times in the test process to accelerate electrolysis or continue electrolysis when geometry, melt pool dynamics, and oxygen production logistics limit the effectiveness of concentrated solar melting. Finally, the team developed and tested a number of regolith introduction subsystems necessary for continuous MRE operation. DESIGN AND TEST SETUP Overall Test Bed Design The test bed (Figure 1) was designed to simulate molten regolith electrolysis within a sealed, cold-wall reactor, drawing off oxygen as it is created and keeping pressure to between 10 and 100 mTorr. A regolith container diameter of only two times the expected melt pool diameter was necessary because of the insulating nature of regolith (and most high-quality simulants) under vacuum. Melt pools form and remain stable with a small heat affected zone (observed to reach less than 0.5cm beyond the melt pool during testing in vacuum) because lunar regolith under vacuum has a thermal conductivity between 0.0001–0.0300 W/mK. With a solar simulator, this makes a predictable, compact melt pool. Though Honeybee has performed tests with direct sunlight using a 49” x 35” Fresnel lens, the length of tests and required solar tracking made testing unwieldy. A solar simulator using a xenon arc lamp and an elliptical reflector was designed so that testing could be performed repeatably and safely.

Figure 2. Left, Zemax plot of focused spot with angular defocus of 1° between bulb and reflector, Center, focused spot showing W/cm (total 1.5kW at surface), Right, Zemax ray trace. Optical design for lamp sizing and elliptical mirror geometry selection was done in Zemax Optic Studio (Figure 2). Tolerancing for lamp misalignment showed that >1cm of axial misalignment between the bulb and reflector was acceptable (being non-imaging optics, focus is only important for total power at surface), but 1° of angular misalignment would render the system unusable. Lamp alignment is challenging in high powered systems where fine

© ASCE

456

Earth and Space 2022

adjustments are necessary but can’t be performed while the lamp is in operation and flexing due to heat has occurred. The team built a laser jig for testing alignment and iteratively adjusted between tests. Glass and mirror absorption as well as estimated solid and molten simulant reflectivity were taken into account to determine temperature. JSC-1A reflectivity was observed in previous projects to spike to over 30% in the visible range, so the lamp was sized to accept an 8kW bulb. Xenon arc lamps require booster / igniter circuits to provide an initial spike in voltage up to 20kV to start the plasma ball between the lamp’s electrodes, then quickly lower the voltage and increase the amperage to sustain it. The Superior Quartz air cooled SQP-SX80001 was selected for this project. A 10kW programmable power supply from MagnaPower was chosen to handle the required power spikes from the booster / igniter circuit. The power supply operated from 24-32 volts and 200-240 amps for an operational power output of 6500 watts. Electrical fluttering and electrode reduction with use will eventually cause the system to require higher power outputs for similar results. Running the higher rated bulb at a lower power allows for greater bulb longevity and a wider range of power if needed. The Optiforms E813 elliptical reflector with a focal distance of 81cm (32”) was chosen based on light cone geometry and reflectivity in the xenon arc lamp wavelength range. A common problem for solar simulators using elliptical reflectors is in restrictive light cone geometry; most commercially available reflectors have focal points that are too short to get to a regolith bed placed in the middle of a commonly constructed chamber. To address this, Honeybee constructed a purpose-built chamber out of a 25cm (10”) diameter CF t-flange with a large diameter sight glass window integrated into an easy access door. This chamber allowed for quick pump down and minimized the distance between the regolith bed and the access door so that images of the melt pool could be captured during operation. A secondary piece of thin, sacrificial glass was set between the window and the simulant surface so that if intense bubbling occurred in the melt pool and it flung our droplets of molten rock, they would not break the glass of the main window. For safety, a 3mm steel sheet box was constructed around the test setup and a secondary steel sheet housing was used for the lamp. Xenon arc lamps can reach multiple atmospheres of pressure under operation, so these precautions were necessary. To prevent eye damage from stray light, welding curtains were hung around the test bed. Because of the high waste heat and ozone produced during lamp operation, a high cfm ventilation system was integrated into the test bed.

Figure 3. Top Left, melt pool in JSC-1A; Top Right, thermal image of melt pool; Bottom Left, frozen simulant formed in vacuum; Bottom Right, frozen simulant formed in atmosphere.

© ASCE

457

Earth and Space 2022

Electrolysis tests were always performed under partial vacuum because of effects observed when melting under partial vacuum vs under ambient pressure (Figure 3). The team performed several simple melting tests in ambient pressure and under vacuum pressure between 10 and 100mTorr. These tests were performed to determine lamp alignment, estimate melt pool temperatures, and qualitatively examine the differences in melt pool behavior under differing pressure. It was found that under ambient pressure bubbles tend to be small and only slightly disturb the flat surface of the melt, while bubbles grew multiple centimeters tall under low pressure. As soon as the heat source was turned off, the bubbles would rapidly refreeze, making frozen melt pools with significantly lower density than those made in air. Because gas movement is a crucial part of the MRE process, this necessitated always testing under vacuum. This issue of rapid refreezing into low-density forms has wide reaching effects to other technologies that use regolith melting (such as 3D printing), and is being studied by the team in work performed in parallel with this work. Electrode Design Considering the stability and size of the melt pool formed with the selected lamp and reflector geometry, the electrolysis subsystem was designed around a 5-7cm diameter, >2cm deep melt pool, reaching temperatures over 1200°C in under 10 minutes. Because oxygen and other trace gases are produced at these elevated temperatures, oxidation and potential corrosion of the electrodes and/or the surrounding assembly was a significant design consideration. The strength of the electrode materials and their thermal and chemical stability in the electrolysis process was also considered. Two trade studies were performed to address these potential issues and successfully produce oxygen. Several refractory metals were considered for electrode materials based on their high melting temperatures and strength. The team elected for platinum plating across all potential refractory metal substrates because they would otherwise undergo varying levels of oxidation. The team compared material properties at temperatures of 1200°C or more and the difficulty of the platinum plating process for each. The same material and platinum plating was chosen for both the anode and cathode for simplicity in sourcing and fabrication, but more importantly because generated oxygen within the melt pool will be in contact with both the anode and the cathode (depending on the system geometry and melt pool dynamics). Specific refractory and transition metals were considered for electrode material with a conductor’s electrical resistance of below 0.1Ω. Fabrication ease was determined by machining ease and the process by which the material must first be cleaned and passivated (necessary steps for platinum plating). Passivation difficulty depended on the acid needed in the process, and cleaning difficulty was evaluated based on the alkalinity of the required cleaning solution. After every experiment electrodes must be inspected and often replated because of plating removed with the glassified, refrozen regolith, making the plating process a non-trivial concern. The following list of materials were considered for this effort in order of easiest hardest materials to plate with platinum: titanium, niobium, tantalum, molybdenum, tungsten. We elected to pick a material as high on the list of easiest metals to platinum plate as permitted by the operating conditions and environment to give the highest possibility that a reliable coat of platinum could be plated onto the surface in-house. Rhenium and iridium were also considered, but because of cost and toxicity concerns they were ultimately eliminated from testing. Titanium was eliminated because of creep concerns at high temperatures. Tungsten and molybdenum were eliminated

© ASCE

458

459

Earth and Space 2022

because of the high acidity needed for plating. Ultimately niobium was selected for its machinability, increasing conductivity with temperature, and ease in plating. TEST Tests were performed first to characterize the test bed, then to develop process parameters for MRE tests while measuring oxygen production rate. The test block diagram (Figure 4) outlines the flow path from within the chamber, where the electrolysis occurs, to the roof where gases are being routed. Vacuum Chamber

Media Filter

Vacuum Pump

Mass Flow Sensor

Exhaust System to Roof

Figure 4. Block diagram of flow path through test setup The mass flow sensor was added to monitor the general flow rate of all gases exiting the chamber while under vacuum. Control tests obtained a baseline for the expected flow rate of all the gases out of the chamber while molten regolith is being created within the chamber and electrolysis is not being performed, which confirmed flow rates between approximately 0.01 and 0.1 SLPM to be expected with the vacuum chamber used for this testing while only producing molten regolith (pre-electrolysis). The team traded between several electrode design features and fabricated 6 electrodes based on this design. The electrodes shown (Figure 5) are waterjet-cut niobium plated with platinum. There are holes in the upper electrode to allow gas escape and simulant addition. The supporting alumina assembly holds the electrodes at a 0.5cm gap to each other, which is less than half of the nominal molten regolith sample thickness measured in experiments done without electrodes. This was chosen to ensure heat and electrical transfer in the melt pool. During testing, the melt pool thickness never extended to a depth beyond the lower electrode in the results below, making this electrode gap just wide enough to conduct electricity through the entirety of the melt pool.

Figure 5. Electrode sub-assembly, before testing.

© ASCE

Earth and Space 2022

After the electrolysis assembly was submerged in regolith, vacuum was pulled on the chamber with a dual rotary vane pump. The concentrated solar simulator was turned on when the vacuum level reached 40mTorr. The two electrical leads from the electrolysis assembly within the chamber route through a feedthrough and to a double pole switch. This switch toggles between leads to an ohmmeter and the electrolysis power supply. Initially This switch is toggled to the position that allows an ohmmeter to read the electrical resistance across the melt pool. Once continuity is sensed and the electrical resistance drops below 500kΩ, the switch is toggled over to the power supply and voltage is applied to the electrodes. Representative test results are shown below (Figure 6). In the first test, the voltage was altered for preliminary investigation while monitoring the current and gas flow out of the system. The team was attempting to determine if and how much of an increase in voltage was necessary to begin electrolysis. Sustaining electrolysis and restarting electrolysis was also part of this group of tests. At higher voltages, approximately above 60 volts, there is the potential for higher current to be drawn through the molten regolith at this electrode gap. Another test shown above was carried out over a similar time period, but while keeping voltage more constant than in the first test at a level just below where the current spiking was observed. A similar level of current and O2 production was seen, however, now without the abrupt spikes in current. Ultimately it appeared that higher voltage lead to current spikes (and corresponding flow rate spikes with voltage drops after due to system power constraints), while steady voltage had fewer spikes and a somewhat downward trending oxygen production rate. In total over the time period that power was applied to the electrodes, approximately 5 minutes, a net gas production gain of about 0.25 liters was seen at these levels of power. Sample results from both tests are shown in the bottom of the figure; a colored orange residue can be seen near the electrode where we believe the build-up of electrolysis byproducts has occurred. This is to be confirmed by upcoming XRF analysis. These first tests have improved our understanding of the process and helped us develop a specific testing procedure for molten regolith electrolysis. Future tests will be performed for a longer duration and with different electrode configurations to improve efficiency. There are several potential reasons for the melt pool dynamics that caused the periodic drastic increases in current. The team assumes there is a single conduction path between the electrodes at any given time where there is the least electrical resistance through the melt pool. At higher voltages, it is likely the molten regolith immediately surrounding this conduction path is joule heated. As the molten regolith temperature rises locally, electrical resistance decreases, which allows more current to pass through and more gas is produced. The current increases rapidly until the produced gas interrupts the conduction path and current drops back down as another conduction path is established somewhere else in the melt pool. The process then repeats in rapid succession. This may explain the succession of current spikes in Figure 6 (which are not caused by noise or power source instability). Joule heating augmented by concentrated solar heating may be the most appropriate method for MRE, controlling the voltage to cause current spikes that serve as a jump starting tool when oxygen p roduction trends downwards.

© ASCE

460

461

Earth and Space 2022

Oxygen Production Test 2: Electrolysis Current at Steady Medium Voltage ( 0, define the reduced basis matrix U corresponding to singular values less than δ. 3. Precompute the ROM matrices: f1 := U e0T M1 U e0 ; M

e 1 := U e0T D1 U e0 ; D

e1 := U e0T A1 U e0 ; A

e 1 := G1 U e0 . and G

Solution of the coupled ROM-FEM system for t ∈ [0, T ] (Online) 1. Choose an explicit time integration scheme, i.e., the operator Dtn (ϕ). 2. For n = 0, 1, . . . use ϕnR to compute the vector e0T f1 − (D e1 + A e1 )ϕnR − U e0T (D1 + A1 )β. fe1n := U 3. Use Φn2 to compute the vector fe2n := f 2 (Φn2 ) = f2 − (D2 + A2 )Φn2 4. Solve the Schur complement system
e1 M f−1 G e T1 + G2 M −1 GT2 λn = G e1 M f−1 U e0T fe1n − G2 M −1 fe2n G 1 2 1 2 e T λn and GT λn . for λn . Compute G 1 2 f1 Dtn (ϕR ) = fen − G e T λn and project the ROM solution to the state space 5. Solve the system M 1 1 e0 ϕR + β1 ; of the full order model: Φ1 := U 6. Solve the system M2 Dtn (Φ2 ) = fe2n + GT2 λn . 4.3

IVR EXTENSION TO ROM-ROM COUPLING

In this section we briefly explain the extension of the IVR scheme to a ROM-ROM case, i.e., when ej,0 and βk,j be the reduced a ROM on Ω1 is coupled to another ROM on Ω2 . For j = 1, 2, let U basis matrix and the vectors (13) constructed on Ωj according to the workflow in Section 4.1. We note here that our framework does not require the two ROMs being coupled to have the same number of reduced basis modes. For simplicity, we shall assume again time-independent Dirichlet boundary conditions, so that the vectors βk,j reduce to a vector βj whose Dirichlet coefficients equal the nodal values of g(x) and the free coefficients are zero. As in Section 4.2, we implement the second stage of the POD-based model order reduction directly in the transformed semi-discrete monolithic problem (7). Specifically, we perform a state e1,0 ϕR + β1 for the first transformation of both subdomain equations using the ansatz Φ1 = U e equation, and the ansatz Φ2 = U2,0 ψR + β2 for the second equation. Then, we multiply the first e T and the second equation by U e T . The resulting ROM-ROM monolithic system equation by U 1,0 2,0 is the basis for the ROM-ROM partitioned IVR algorithm which we state below. Computation of the reduced order models (Offline) 1. For j = 1, 2, use an appropriate FOM to simulate the solution on Ωj and collect samples for the snapshot matrix Xj . Compute the SVD of the adjusted snapshot matrix Xj,0 = T Uj,0 Σj,0 Vj,0 . ej,0 by discarding all columns 2. Given a threshold δj > 0, define the reduced basis matrices U in Uj,0 corresponding to singular values less than δj for j = 1, 2.

© ASCE

483

Earth and Space 2022

3. For j = 1, 2, precompute the ROM matrices: fj := U e T Mj U ej,0 ; M j,0

e j := U e T Dj U ej,0 ; D j,0

ej := U e T Aj U ej,0 ; A j,0

e j := Gj U ej,0 . and G

Solution of the coupled ROM-ROM system for t ∈ [0, T ] (Online) n 1. Choose an explicit time integration scheme for each subdomain, i.e., an operator Dj,t (ϕ), j = 1, 2.

2. For n = 0, 1, . . . use ϕnR to compute the vector e T f1 − (D e1 + A e1 )ϕn − U e T (D1 + A1 )β1 . fe1n := U 1,0 R 1,0 n 3. For n = 0, 1, . . . use ψR to compute the vector T n T e2,0 e2 + A e2 )ψR e2,0 fe2n := U f2 − (D −U (D2 + A2 )β2 .

4. Solve the Schur complement system
e1 M f−1 G eT + G e2 M f−1 G e T λn = G e1 M f−1 U e T fen − G e2 M f−1 U e T fen G 1 2 1,0 1 2,0 2 1 2 1 2 e T λn and G e T λn . for λn . Compute G 1 2 f1 Dn (ϕR ) = fen − G e T λn . 5. Solve the system M 1,t 1 1 f2 Dn (ψR ) = fen + G e T λn . 6. Solve the system M 2,t 2 2 7. Project the ROM solutions ϕR , ψR to the state spaces of the full order models on Ω1 and Ω2 : e1,0 ϕR + β1 ; Φ2 := U e2,0 ψR + β2 . Φ1 := U

5

NUMERICAL EXAMPLES

To evaluate our schemes, we adapt the solid body rotation test for (1) from (LeVeque, 1996). The problem is posed on the unit square Ω = (0, 1) × (0, 1) and the following rotating advection field (0.5 − y, x − 0.5) is specified. The initial conditions for this test problem comprise a cone, cylinder, and a smooth hump, and are shown in Figure 2(a). We impose homogeneous Dirichlet boundary conditions on the non-interface boundaries Γi , i = 1, 2. We consider herein two problem configurations for (1) that differ in the choice of the diffusion coefficient. The “pure advection” case corresponds to κi = 0, and the “high Pecl´et” case corresponds to κi = 10−5 . In the former case we adjust the boundary condition so that the boundary values are specified only on the inflow parts of Γi . In all our tests, we run the simulations for one full rotation, i.e., the final simulation time is set to t = 2π. It can be shown that, for the pure advection variant of this problem, the solution at the final time t = 2π should be the same as the initial solution (LeVeque, 1996). Suppose Ω is divided in half vertically by the line x = 0.5, and let Ω1 and Ω2 denote the left and right side of the domain, respectively, as shown in Figure 2(b). Let γ denote the interface (x = 0.5) between the two sides, and let Γi = ∂Ωi \γ for i = 1, 2. We take nγ to be the unit normal on the interface pointing toward Ω2 . In this section, we present select results for solving the model advection-diffusion interface problem (1) by performing both ROM-FEM and ROM-ROM coupling in the two subdomains, Ω1 and Ω2 . The coupled ROM-FEM and ROM-ROM problems are solved by using the IVR partitioned schemes formulated in Section 4.2 and Section 4.3, respectively. We compare our ROM-FEM and ROM-ROM solutions to results obtained by employing our IVR partitioned scheme to perform FEM-FEM coupling between the two subdomains (see Section 3). For comparison purposes, we also include results obtained by building a global (uncoupled) FEM model as well as a global ROM in the full domain Ω.

© ASCE

484

Earth and Space 2022

(a) Initial conditions

(b) Meshes used to discretize Ω1 (blue) and Ω2 (red)

Figure 2: Initial conditions and domain decomposition/mesh for our model 2D transmission problem

1 For the FEM discretizations, we employ a uniform spatial resolution of 64 in both the x and y directions. The ROMs are developed from snapshots collected from a monolithic FEM discretization of Ω using the approach described in Section 4.1, with snapshots collected at intervals ∆s t = 1.35 × 10−2 and ∆s t = 6.73 × 10−3 for the pure advection and high Pecl´et variants of our test case, respectively. These snapshot selection strategies yield 466 and 933 snapshots for the two problem variants, respectively. All ROMs evaluated herein are run in the reproductive regime, that is, with the same parameter values, boundary conditions and initial conditions as those used to generate the snapshot set from which these models were constructed; predictive ROM simulations will be considered in a subsequent publication. In general, between 20-25 modes are needed to capture 90% of the snapshot energy and between 50-65 modes are needed to capture 99.999% of the snapshot energy for both problem variants, where the snapshot energy fraction is defined as 1 − δ. As noted in Section 4.3, for the ROM-ROM couplings, we allow the bases in Ω1 and Ω2 to have different numbers of modes, denoted by NR,left and NR,right , respectively. Hence, the number of modes required to capture a given snapshot energy fraction varies slightly between the two subdomains. All simulations are performed using an explicit 4th order Runge-Kutta (RK4) scheme with time-step ∆t = 3.37 × 10−3 , the time-step computed by the Courant-Friedrichs-Lewy (CFL) condition for this problem. In the results below, we report for the various models evaluated the following relative errors as a function of the basis size and the total online CPU time:

 :=

||X2π − F2π ||2 . ||F2π ||2

(18)

In (18), X ∈ {R, F F, RF, RR}, where R denotes the global ROM solution computed in all of Ω, F F denotes a FEM-FEM coupled solution, RF denotes a ROM-FEM coupled solution, and RR denotes a ROM-ROM coupled solution. The subscripts in (18) denote the time at which a given solution is evaluated, i.e., RF2π is the ROM-FEM solution at time t = 2π. The reference solution in (18), denoted by F2π , is the global FEM solution computed in all of Ω at time t = 2π. For the pure advection problem, we additionally report: 0 :=

||X0 − X2π ||2 , ||X2π ||2

(19)

for X ∈ {F, R, F F, RF, RR}. As shown in (LeVeque, 1996), for the exact solution to the pure advection problem, 0 is identically zero. First, in Figure 3, we plot the relative error  in (18) as a function of the POD basis size for the various couplings and the two problem variants considered herein. All errors are calculated with

© ASCE

485

Earth and Space 2022

respect to the global FEM solution computed in all of Ω. For the ROM-ROM couplings, the basis size in Figure 3 is obtained by calculating the average of the basis sizes in Ω1 and Ω2 , denoted by NR,left and NR,right respectively. The reader can observe that all models exhibit convergence with respect to the basis size. In particular, the ROM-FEM and ROM-ROM solutions converge at a rate of approximately two. For the pure advection problems, the ROM-FEM and ROM-ROM solutions appear to be approaching the FEM-FEM error with basis refinement, and the ROMROM solution appears to be converging to the ROM-FEM solution. It is interesting to observe that the global ROM solutions achieve a greater accuracy than the FEM-FEM coupled solutions. Moreover, for the high Pecl´et version of the problem, the ROM-FEM coupled solution can achieve an accuracy that is slightly better than the FEM-FEM coupled solution. This behavior is likely due to the fact that the ROM solution was created using snapshots from a global FEM solution, which is more accurate than the coupled FEM-FEM solution.

(a) Pure Advection

(b) High Pecl´ et

Figure 3: Relative errors (18) with respect to the global FEM solution as a function of the POD basis size for different discretizations of the pure advection (a) and high Pecl´et (b) variants of our model transmission problem.

In evaluating the viability of a reduced model, it is important to consider not only the model’s accuracy, but also its efficiency. Toward this effect, Figures 4(a) and (b) show Pareto plots for the models evaluated on the pure advection and high Pecl´et problems, respectively. In these figures, we plot the relative errors (18) as a function of the total online CPU time. As expected, the global FEM and FEM-FEM models require the largest CPU time, followed by the ROM-FEM models, the ROM-ROM models and the global ROM models. It is interesting to remark that the FEM-FEM discretizations are actually slightly faster than the global FEM discretizations. This suggests that, in the case of high-fidelity models, our proposed coupling approach does not introduce any significant overhead. While the global ROM achieves the most accurate solution in the shortest amount of time, we are targeting here the scenario where the analyst does not have access to a single domain solver, and is forced to couple models calculated independently in different parts of the computational domain. The results in Figure 4 show that, by introducing ROM-FEM and ROM-ROM coupling, one can reduce the CPU time by 1-1.5 orders of magnitude without sacrificing accuracy. Turning our attention now to the pure advection problem, we plot in Figure 5 the relative errors 0 in (19) as a function of the basis size. Again, the global FEM model is the most accurate, followed by the FEM-FEM, the ROM-FEM and the ROM-ROM models. It is interesting to observe that the global ROM surpasses the global FEM solution when it comes to accuracy for certain (intermediate) basis sizes. The primary takeaway from Figure 5 is that the ROM-FEM, the ROM-ROM and the global ROM solutions asymptotically approach the global FEM solution as the basis size is refined. This provides further verification for the models evaluated, in particular,

© ASCE

486

Earth and Space 2022

(a) Pure Advection

(b) High Pecl´ et

Figure 4: Pareto plot (relative errors (18) as a function of the total online CPU time) for different discretizations of the pure advection (a) and high Pecl´et (b) variants of our model transmission problem.

for our new IVR coupling approach. Next, in Figure 6, we plot some representative ROM-FEM and ROM-ROM solutions to the high Pecl´et variant of the targeted problem at the final simulation time 2π. Also plotted is the single domain global FEM solution computed for this problem. The reader can observe that all three solutions are indistinguishable from one another. Figure 7 plots the ROM-FEM and ROM-ROM solutions to the high Pecl´et problem along the interface Γ for each of the subdomains at the final simulation time 2π. It can be seen from this figure that the solutions in Ω1 and Ω2 match incredibly well along the interface boundary. This suggests that our coupling method has not introduced any spurious artifacts into the discretization. We omit plots analogous to Figures 6 and 7 for the pure advection problem for the sake of brevity, as they Figure 5: Relative errors (19) as a function of lead to similar conclusions as high Pecl´et problem the POD basis size for different discretizations of the pure advection problem. results.

6

CONCLUSIONS

We presented an explicit partitioned scheme for a transmission problem that extends the approach developed in (K. Peterson et al., 2019) to the case of coupling a projection-based ROM with a traditional finite element scheme and/or with another projection-based ROM. In particular, the scheme begins with a monolithic formulation of the transmission problem and then employs a Schur complement to solve for a Lagrange multiplier representing the interface flux as a Neumann boundary condition. We constructed a ROM from a full finite element solution and then presented an algorithm to couple this reduced model with either a traditional finite element scheme or another reduced model. Our numerical results show that the ROM-FEM and ROM-ROM coupling produces solutions which strongly agree with those produced by a global FEM solver. Additionally, implementing the ROM in one or more subdomains reduces the time and computational

© ASCE

487

Earth and Space 2022

(a) Global FEM

(b) ROM-FEM (NR = 80)

(c) ROM-ROM (NR,left NR,right = 110)

=

112,

Figure 6: Comparison of global FEM, ROM-FEM and ROM-ROM solutions for the high Pecl´et variant of our model transmission problem at the final simulation time t = 2π.

(a) ROM-FEM (NR = 80)

(b) ROM-ROM (NR,left = 112, NR,right = 110)

Figure 7: Comparison of the interface ROM-FEM and ROM-ROM solutions for high Pecl´et variant of our model transmission problem at the final simulation time t = 2π.

cost of solving the coupled system. In principle, this coupling method should extend to other discretizations such as finite volume, and the case of multiple (> 2) subdomains; these scenarios will be studied in future work. Additionally, extensions to nonlinear and multiphysics problems, as well as predictive runs will be considered.

ACKNOWLEDGEMENTS This work was funded by the Laboratory Directed Research & Development (LDRD) program at Sandia National Laboratories, and the U.S. Department of Energy, Office of Science, Office of Advanced Scientific Computing Research under Award Number DE-SC-0000230927 and under the Collaboratory on Mathematics and Physics-Informed Learning Machines for Multiscale and Multiphysics Problems (PhILMs) project. The development of the ideas presented herein was funded in part by the third author’s Presidential Early Career Award for Scientists and Engineers (PECASE). Sandia National Laboratories is a multi-mission laboratory managed and operated by National Technology and Engineering Solutions of Sandia, LLC., a wholly owned subsidiary of Honeywell International, Inc., for the U.S. Department of Energy’s National Nuclear Security Administration under contract DE-NA0003525. This paper describes objective technical results and analysis. Any subjective views or opinions that might be expressed in the paper do not

© ASCE

Earth and Space 2022

necessarily represent the views of the U.S. Department of Energy or the U.S. Government.

REFERENCES Ammar, A., F. Chinestra, and E. Cueto (2011). “Coupling finite elements and proper generalized decompositions”. In: International Journal for Multiscale Computational Engineering 9, pp. 17–33. Antil, H. et al. (2010). “Domain decomposition and model reduction for the numerical solution of PDE constrained optimization problems with localized optimization variables”. In: Computing and Visualization in Science 13, pp. 249–264. Aubry, Nadine et al. (1988). “The dynamics of coherent structures in the wall region of a turbulent boundary layer”. In: Journal of Fluid Mechanics 192, pp. 115–173. doi: 10 . 1017 / S0022112088001818. Baiges, Joan, Ramon Codina, and Sergio R. Idelsohn (2013). “A domain decomposition strategy for reduced order models. Application to the incompressible Navier–Stokes equations”. In: Computer Methods in Applied Mechanics and Engineering 267, pp. 23–42. Banks, J.W. et al. (2017). “A stable partitioned FSI algorithm for rigid bodies and incompressible flow. Part I: Model problem analysis”. In: Journal of Computational Physics 343, pp. 432– 468. issn: 0021-9991. doi: https://doi.org/10.1016/j.jcp.2017.01.015. url: http: //www.sciencedirect.com/science/article/pii/S0021999117300256. Bessette, Greg C. et al. (2003). “Modeling Air Blast on Thin-shell Structures with Zapotec”. In: Proceedings of the 74 th Shock and Vibration Symposium. Bochev, Pavel and R. B. Lehoucq (2005). “On the Finite Element Solution of the Pure Neumann Problem”. In: SIAM Review 47.1, pp. 50–66. doi: 10.1137/S0036144503426074. url: http: //link.aip.org/link/?SIR/47/50/1. Buffoni, M., H. Telib, and A. Iollo (2007). Iterative Methods for Model Reduction by Domain Decomposition. Tech. rep. 6383. INRIA Report. Cinquegrana, D., A. Viviani, and R. Donelli (2011). “A hybrid method based on POD and domain decomposition to compute the 2-D aerodynamic flow field - incompressible validation”. In: AIMETA 2011–XX Congresso dell’Associazione Italiana di Meccanica Teorica e Applicata, Bologna, ITA. Corigliano, A., M. Dossi, and S. Mariani (2015). “Model Order Reduction and domain decomposition strategies for the solution of the dynamic elastic–plastic structural problem”. In: Computer Methods in Applied Mechanics and Engineering 290.C, pp. 127–155. Corigliano, Alberto, Martino Dossi, and Stefano Mariani (2013). “Domain decomposition and model order reduction methods applied to the simulation of multi-physics problems in MEMS”. In: Computers & Structures 122. Computational Fluid and Solid Mechanics 2013, pp. 113–127. issn: 0045-7949. de Boer, A., A.H. van Zuijlen, and H. Bijl (2007). “Review of coupling methods for non-matching meshes”. In: Computer Methods in Applied Mechanics and Engineering 196.8. Domain Decomposition Methods: recent advances and new challenges in engineering, pp. 1515–1525. issn: 0045-7825. doi: http : / / dx . doi . org / 10 . 1016 / j . cma . 2006 . 03 . 017. url: http : //www.sciencedirect.com/science/article/pii/S0045782506002817. Eftang, Jens L. and Anthony T. Patera (2013). “Port reduction in parametrized component static condensation: approximation and a posteriori error estimation”. In: International Journal for Numerical Methods in Engineering 96.5, pp. 269–302. — (2014). “A port-reduced static condensation reduced basis element method for large componentsynthesized structures: approximation and A Posteriori error estimation”. In: Advanced Modeling and Simulation in Engineering Sciences 1.3. Gatzhammer, B. (Sept. 2014). “Efficient and Flexible Partitioned Simulation of Fluid-Structure Interactions”. Doctoral Thesis. Fakultaet fuer Informatik. Informatik 5 – Lehrstuhl fuer Wissenschaftliches Rechnen: Technische Universitaet Muenchen.

© ASCE

488

Earth and Space 2022

Gunzburger, Max D., Janet S. Peterson, and John N. Shadid (2007). “Reduced-order modeling of time-dependent PDEs with multiple parameters in the boundary data”. In: Computer Methods in Applied Mechanics and Engineering 196, pp. 1030–1047. Hoang, Chi, Youngsoo Choi, and Kevin Carlberg (2021). “Domain-decomposition least-squares Petrov–Galerkin (DD-LSPG) nonlinear model reduction”. In: Computer Methods in Applied Mechanics and Engineering 384, p. 113997. Holmes, P., J. Lumpley, and G. Berkooz (1996). Turbulence, Coherent structures, dynamical systems and symmetry. Cambridge University Press. Iapichino, L., A. Quarteroni, and G. Rozza (2016). “Reduced basis method and domain decomposition for elliptic problems in networks and complex parametrized geometries”. In: Computers & Mathematics with Applications 71.1, pp. 408–430. Kerfriden, P. et al. (2013). “A partitioned model order reduction approach to rationalise computational expenses in nonlinear fracture mechanics”. In: Computer Methods in Applied Mechanics and Engineering 256, pp. 169–188. issn: 0045-7825. LeGresley, P. (2005). “Application of Proper Orthogonal Decomposition (POD) to Design Decomposition Methods”. PhD thesis. Stanford, CA: Stanford University. LeGresley, P. and J. Alonso (2003). “Dynamic domain decomposition and error correction for reduced order models”. In: 41st Aerospace Sciences Meeting and Exhibit, Reno, Nevada. LeVeque, Randall J. (1996). “High-Resolution Conservative Algorithms for Advection in Incompressible Flow”. In: SIAM Journal on Numerical Analysis 33.2, pp. 627–665. doi: 10.1137/ 0733033. url: http://link.aip.org/link/?SNA/33/627/1. Lucia, D. et al. (2001). “Reduced order modeling for a one-dimensional nozzle flow with moving shocks”. In: 15th AIAA Computational Fluid Dynamics Conference, Anaheim, CA. Lucia, David J., Paul I. King, and Philip S. Beran (2003). “Reduced order modeling of a twodimensional flow with moving shocks”. In: Computers & Fluids 32.7, pp. 917–938. issn: 00457930. Maday, Y. and E. Ronquist (2004). “The reduced basis element method: application to a thermal fin problem”. In: SIAM J. Sci. Comput. 26.1, pp. 240–258. Maier, I. and B. Haasdonk (2014). “A Dirichlet-Neumann reduced basis method for homogeneous domain decomposition problems”. In: Applied Numerical Mathematics 78, pp. 31–48. Peterson, K., P. Bochev, and P. Kuberry (2019). “Explicit synchronous partitioned algorithms for interface problems based on Lagrange multipliers”. In: Computers & Mathematics with Applications 78 (2), pp. 459–482. Piperno, Serge and Charbel Farhat (2001). “Partitioned procedures for the transient solution of coupled aeroelastic problems – Part II: energy transfer analysis and three-dimensional applications”. In: Computer Methods in Applied Mechanics and Engineering 190.24–25. Advances in Computational Methods for Fluid-Structure Interaction, pp. 3147–3170. issn: 0045-7825. doi: http://dx.doi.org/10.1016/S0045-7825(00)00386-8. url: http://www.sciencedirect. com/science/article/pii/S0045782500003868. Radermacher, A. and S. Reese (2014). “Model reduction in elastoplasticity: proper orthogonal decomposition combined with adaptive sub-structuring”. In: Computational Mechanics 54, pp. 677–687. Randers-Pehrson, Glenn and Kenneth A. Bannister (Mar. 1997). Airblast Loading Model for DYNA2D and DYNA3D. Tech. rep. ARL-TR-1310. Army Research Laboratory. Sirovich, L. (1987). “Turbulence and the dynamics of coherent structures, Part III: dynamics and scaling”. In: Quarterly of Applied Mathematics 45.3, pp. 583–590. issn: 0033569X, 15524485. url: http://www.jstor.org/stable/43637459. Sockwell, K. Chad et al. (2020). “Interface Flux Recovery coupling method for the ocean–atmosphere system”. In: Results in Applied Mathematics 8, pp. 100–110.

© ASCE

489

490

Earth and Space 2022

Lessons Learned and Best Practices for Utilizing a Generalized Composite Impact Model Robert K. Goldberg1 and Trenton M. Ricks2 1

Ceramic and Polymer Composites Branch, Materials and Structures Division, NASA Glenn Research Center, Cleveland, OH. Email: [email protected] 2 Multiscale and Multiphysics Modeling Branch, Materials and Structures Division, NASA Glenn Research Center, Cleveland, OH. Email: [email protected] ABSTRACT A material model which incorporates several key capabilities which have been identified as lacking in currently available composite impact models has been developed. The material model utilizes experimentally based tabulated input to define the evolution of plasticity and damage as opposed to specifying discrete input parameters (such as modulus and strength). It has been implemented into the commercially available transient dynamic finite element code LS-DYNA as MAT_213. The model can simulate the nonlinear deformation, damage, and failure that take place in a composite under dynamic loading conditions. As MAT_213 is now being used by a general user community that did not participate in the model development process, a number of issues have been identified that caused uncertainty in assembling a MAT_213 input deck and conducting a MAT_213 analysis. The overall goal of this effort is to develop a quantified set of lessons learned and best practices which will permit a new user to conduct useful MAT_213 simulations without needing to have detailed expert knowledge of the material model and its theoretical underpinnings. INTRODUCTION As composite materials are gaining increased use in aircraft components where impact resistance is critical (such as the turbine engine fan case), the need for accurate material models to simulate the deformation, damage and failure response of polymer matrix composites under impact conditions is gaining importance. While there are several material models currently available within commercial transient dynamic codes such as LS-DYNA® (Hallquist 2013) to analyze the impact response of composites, areas have been identified where the predictive capabilities of these models can be improved. These limitations have been extensively discussed by Goldberg, et al (2014, 2015a). In particular, one major limitation of the currently existing models is that the input to these models generally consists of point-wise properties (such as the modulus, failure stress or failure strain in a particular coordinate direction) that leads to linear approximations to the material stress-strain curves and simplified approximations to the actual material failure surfaces. This type of approach either leads to models with only a few parameters, which provide a crude approximation at best to the actual material response, or to models with many parameters which require a large number of complex tests to characterize. An improved approach is to use tabulated data, obtained from well-defined set of physically meaningful experiments. Using tabulated data allows the actual material response data to be entered in a discretized form, which permits a more accurate representation of the actual material response.

© ASCE

Earth and Space 2022

To begin to address these limitations in the currently existing models, a new composite material model incorporating deformation, damage and failure has been developed and implemented for use within LS-DYNA and has been given the formal name of *MAT_COMPOSITE_TABULATED_PLASTICITY_DAMAGE as well as the numerical identifier MAT_213. The material model is meant to be a fully generalized model suitable for use with a large number of composite architectures (laminated or textile). The deformation model, described extensively in Goldberg, et al (2015a), is based on utilizing the functional form employed in the commonly used Tsai-Wu composite failure model (Daniel and Ishai, 2006) as an orthotropic yield function. This yield function is combined with a non-associative flow rule to generate a strain hardening plasticity model. To compute the current value of the yield stresses needed for the yield function, tabulated stress-strain curves are used to track the yield stress evolution. The user is required to input nine stress versus plastic strain curves. Specifically, the required curves include uniaxial tension curves in each of the normal directions (1,2,3), uniaxial compression curves in each of the normal directions (1,2,3), and shear stress-strain curves in each of the shear directions (1-2, 2-3 and 3-1). To more precisely characterize stress interaction effects, 45° off-axis tension (or compression) curves in each of the 1-2, 2-3 and 1-3 planes can also be entered as input. Strain rate effects and thermal softening can be accounted for within the material model by inputting a full set of stress strain curves for a variety of strain rates and temperatures. By utilizing tabulated stress-strain curves to track the evolution of the deformation response, the experimental stress-strain response of the material can be captured accurately. Note that for thin shell elements, only the curves relating to the in-plane response are required. For the damage model, described extensively in Goldberg, et al (2015b), a strain equivalent formulation is used in which the deformation and damage calculations can be uncoupled. A significant feature in the developed damage model is that a semi-coupled approach has been utilized in which a load in a particular coordinate direction results in damage (and thus stiffness reduction) in multiple coordinate directions. This semi-coupled approach, while different from the methodology used in many existing damage mechanics models (Goldberg et al, 2015b), has the potential to more accurately reflect the damage behavior that actually takes place, particularly for composites with more complex fiber architectures. If the damage model is activated, damage parameters in each of the coordinate directions, and their variation as a function of strain, are entered in a tabulated fashion into the model input. The damage model can be employed to simulate the nonlinear unloading response that a composite exhibits when loaded to strain levels up to the point that the peak stress takes place. The damage model can also be used to model the stress degradation that takes place after the peak stress in the stress-strain response is obtained. Accounting for this “post-peak stress degradation” permits the correct simulation of the additional deformation and energy absorption that takes place in an actual composite structure when loaded beyond the peak stress during an impact or dynamic crush event. An example of how the post-peak stress degradation can be represented in a shear stress-strain curve is shown in Figure 1. In this figure, after the peak stress is obtained the stress is degraded to a significantly lower equilibrium stress which is held constant until a defined effective strain is reached. As can be seen in the figure, this behavior is accounted for within MAT_213 by rapidly increasing a damage parameter to a defined high, constant, value once the peak stress is reached. Further details on the development and implementation of the ability to model the post -peak stress degradation can be found in Achstetter (2019).

© ASCE

491

Earth and Space 2022

Figure 1. Input shear stress-strain curve for MAT_213 including post-peak stress degradation. A wide variety of failure models, which mark the end of the stress-strain curve, have been developed for composites. In commonly used models such as the Tsai-Wu failure model (Daniel and Ishai, 2006), a quadratic function of the macroscopic stresses is defined in which the coefficients of the failure function are related to the tensile, compressive and shear failure stresses in the various coordinate directions. This model, while mathematically simple and easy to implement numerically, assumes that the composite failure surface has an ellipsoidal (in 2D) or ovoid (in 3D) shape. In reality, composite failure surfaces often are not in the form of simple shapes. More complex models, such as the Hashin model (Hashin, 1980) and the Puck model (Puck and Schurmann, 1998), also utilize quadratic combinations of the macroscopic failure stresses, but utilize only selective terms in the quadratic function in order to link the macroscopic stresses to local failure modes such as fiber or matrix failure. However, in these and other advanced models, very complex tests are often required to characterize the model parameters and the applicability of the models may be limited to specific composite architectures with specific failure mechanisms. None of the existing composite failure models show a complete ability to predict composite failure. This difficulty in simulating composite failure can be related to the fact that, in reality, failure is a highly localized phenomenon dependent on various combination of fiber, matrix and interface failures. Due to these complex and interacting local failure mechanisms, and the fact that these mechanisms can vary based on the constituent materials and fiber architecture, the actual composite failure surface often does not conform to a shape that can be easily simulated using a simple mathematical function. Conversely, attempts to utilize discrete functions to analyze the complex local mechanisms can result in models with a large number of parameters that require a highly complex test program to obtain. For users familiar with and comfortable with the existing failure models, the Tsai-Wu and Puck failure models are available within MAT_213. However, a new failure model, referred to as the Generalized Tabulated Failure Criteria (GTFC), is also incorporated within MAT_213. In the GTFC, an approach is used in which the failure envelope for a composite is defined and entered in a tabulated fashion. Specifically, a stress- or strain-based invariant leading to the initiation of failure is defined as a function of the location of the current stress state in stress space. In this manner, an arbitrary failure surface can be easily defined based on actual experimental data in combination with numerical data obtained using any desired existing failure model. The current approach thus serves as a general

© ASCE

492

Earth and Space 2022

framework which is not limited to any arbitrarily imposed failure surface based on an arbitrarily defined mathematical function. Further details on the GTFC model can be found in Goldberg, et al (2018b). As MAT_213 is now being used by a general user community that did not participate in the model development process, a number of issues have been identified that have caused uncertainty in assembling a MAT_213 input deck and conducting a MAT_213 analysis. In utilizing MAT_213 to simulate a variety of impact, crash and crush problems, this community identified several issues which could inhibit the ability of a general user to successfully utilize the code to analyze their specific problems. A need has been identified to develop a quantified set of lessons learned and best practices which will permit a new user to conduct usefu l MAT_213 simulations without needing to have detailed expert knowledge of the material model and its theoretical underpinnings. A summary of these efforts and some of the identified best practices are presented in this document. Specifically, the goals of this effort are to augment the existing documentation. While a keyword manual for MAT_213 is available in the LS-DYNA user manual (Hallquist 2013), for an inexperienced user this documentation may not be sufficient to allow them to successfully generate and execute a MAT 213 simulation. In this paper (and in a subsequent document which is under development), a specific set of best practices and guidelines for inexperienced users in using MAT_213 will be described. Specifically, in this paper the components of a MAT_213 input deck will be described in detail, including a discussion of recommended methods for characterizing several of the parameters and data required for model input. A recommended verification and validation workflow for determining if the input parameters have been characterized correctly, and for using the characterized input deck to conduct predictive simulations, will be described. DEVELOPING MAT_213 INPUT CARDS A sample MAT_213 keyword input file is shown in Figure 2, with annotated quantities specific to the model. This input file was developed to be compatible with version 1.3.5 of MAT_213, which is consistent with the latest development version of LS-DYNA (revision ≥ 87988) at the time this paper was written. Note that only selected features of the input file are described here. Further details of the input file layout can be found in the LS-DYNA user and keyword manuals. Quantities not highlighted in the figure are similar to those in other LS-DYNA composite material models, and include the material identification number, density, elastic properties, and orientation parameters. In the input file, The Young’s and shear moduli (EA, EB, EC, GAB, GAC and GBC) are only used for model initialization and setup and after the initial setup MAT_213 internally calculates and overwrites these quantities based on information included in the input curves. However, these parameters are required and zeros cannot be entered. In general, the MAT_213 specific input parameters can be classified into one of nine categories. A few parameters control different aspects of the solution and typically should be kept at default values. For instance, TCSYM controls the tension/compression symmetry in the model and can override the input curve behavior. In the testing of actual composites, the tension and compression moduli in some (or all) of the coordinate directions can be different. However, to avoid analysis instabilities, a user may desire to manually impose a defined modulus on the tension and compression response. Using the TCSYM card, the user can specify if the input curves should be manually adjusted such that both the tension and compression elastic moduli are set equal to the tension modulus, the compression modulus, or an average of the two values.

© ASCE

493

494

Earth and Space 2022

Figure 2. Example MAT_213 Input File The flow rule coefficients represent how the material model behaves in the inelastic regime and are assumed to be constant regardless of temperature and strain rate. These coefficients must be carefully chosen to be consistent with the input curves to avoid errors. These parameters cannot be computed directly from the input curves. While all flow rule coefficients are required for solid elements, only H11, H22, H12, H44 are required for thin shell elements due to the plane stress assumptions inherent in the use of thin shells. Additionally, the shear coefficients (H44, H55, H66) have been observed to give similar results regardless of their value. There are several methods that are currently being used to determine the appropriate values for these coefficients, with improved methods under active development. The flow rule coefficient values can be directly computed using plastic Poisson ratios (the ratio of transverse to axial plastic strain). While theoretically consistent, the determination of the plastic Poisson ratio values from experimental data is often complicated and the flow rule coefficients determined using this procedure often yield nonoptimal results. An alternate method of determining the flow rule coefficients is to use an optimization algorithm to provide values that result in an acceptable match to the input curves. A more “brute force” approach is to use a trial-and-error process to

© ASCE

Earth and Space 2022

determine which flow rule coefficient values result in simulated results that correlate best to the structural level response. This approach often relies on making simplifying assumptions based on material architecture. For instance, for unidirectional materials that are approximately linear in the 1-direction, H11 can be set to approximately 0, and H12 would be set to zero. For plain weave materials, H11 and H22 would be equal. Currently, the direct and “brute force” approaches are what are currently being used in practice. However, efforts are currently underway to develop advanced coupon level test methods which could lead to a more physically based approach to determine these coefficients. In MAT_213, the stress-strain test data is converted into a series of curves and added to tables (defining any rate or temperature dependency) that are referenced by the LT1-LT12 parameters. These twelve parameters provide a “pointer” to the twelve tabulated input curves previously defined. For cases where actual test data for a particular load case is not available, suitable approximations or alternate analytical tools (such as micromechanics analysis codes) can be used to generate the required tabulated input. Similar to the flow rule coefficients, some of these parameters correspond to out-of-plane curves and can be omitted when using thin shell elements (i.e., LT3, LT6, LT8, LT9, LT11, and LT12). For modeling the ply-by-ply response of laminated composites, using thin shell elements is recommended. An additional table is defined using the yield strains from the actual input curves. These values act as a switch to activate additional inelastic calculations internally. The actual yield strain values are somewhat subjective, but should be defined in a consistent manner to the corresponding input curve to avoid errors. For example, defining a yield strain well into the softening regime for a particular input curve will result in an error. Two parameters, DCFLAG and DC, control the activation of the damage model and define a table for individual damage parameters, respectively. A total of 84 damage parameters (12 uncoupled, 72 coupled) are available to be utilized based on the availability of test data or applicability and significance of a particular parameter. These parameters are defined through input curves which define the variation of the damage parameter as a function of strain. Primarily, uncoupled damage parameters have been utilized to date. Practically, only a relatively small subset of damage curves will be used for a particular analysis to capture the most critical damage mechanisms. Of the 84 possible input curves (defining the 84 possible damage parameters), only those curves corresponding to the damage parameters used in the analysis need to be included. Two cards are allotted to handle failure and can take on multiple forms depending on the particular failure model considered. Three failure models are currently available: Tsai-Wu, Puck, and Generalized Tabulated Failure Criterion (GTFC). For example, four parameters are needed to define the GTFC as shown in Figure 2. While both damage and failure parameters are assumed to be independent of temperature and strain rate in the current release of the model, this capability is currently being added. Additionally, each of the failure models are optional and do not have to be activated to successfully run *MAT_213. An advanced feature of *MAT_213 is that the energy from plastic work can be converted to temperature using the specific heat capacity (CP) and the Taylor-Quinney coefficient (TQC). Since *MAT_213 can model both strain rate and temperature dependent behavior, the TEMP parameter defines the temperature for a specific material (rather than input it from the applied boundary conditions). While not discussed here in detail for conciseness, the variables labeled “Solution Controls” in Figure 2 control convergence tolerance for the deformation model (PTOL), modulus symmetry considerations for the case where the experimental tensile and compressive moduli

© ASCE

495

496

Earth and Space 2022

differ for a particular material (TCSYM) and the number of increments to be used in plasticity model calculations (PMACC). The variables labeled “Viscoelastic Inputs” describe input required for an optional viscoelasticity model (to account for rate dependence in the elastic response) which is described in detail in Achstetter (2019). The temperature variable “TEMP” defines the temperature that was used to generate the baseline input curves. The variables labeled “Taylor-Quinney Inputs” define variables used to compute the adiabatic temperature rise that can place during dynamic deformation. SINGLE ELEMENT VERIFICATION PROCEDURE SUMMARY A two-step process is recommended to be employed to conduct a verification of the developed input cards. First, a series of analyses should be carried out using single elements. The first step in the single element analysis process is to replicate the loading conditions that were used to provide the input data for the material model, specifically uniaxial tension curves in each of the normal directions (1,2,3), uniaxial compression curves in each of the normal directions (1,2,3), shear stress-strain curves in each of the shear directions (1-2, 2-3 and 3-1), and, if entered as input, 45 degree off-axis tension or compression curves in each of the 1-2, 2-3 and 3-1 planes. By examining the results from these analyses, the ability of the code to accurately replicate the input data can be determined. By using single elements for this first set of verification and validation studies, complexities of the analysis caused by complicated boundary conditions, element interactions, and edge effects can be avoided. In analyzing a realistic structure, simple loading conditions such as uniaxial tension are not present in every element in the finite element model. The goal of the verification process is to confirm that under simple loading conditions the simulated stress-strain curves match the input stress-strain curves. In this manner, a user can confirm that the developed input deck will yield expected results. A sample set of elements used for a single element verification study is shown in Figure 3.

*AWG VER-1 example Figure 3. Single element models used for verification studies.

A series of considerations should be made in setting up and executing the single element verification analyses. First, a set of single element verification studies should be conducted for each combination of temperature and strain rate. A combination of load and displacement control boundary conditions should be imposed in order to examine the ability of the input material parameters to fully replicate the material response. Some loading conditions will be well

© ASCE

Earth and Space 2022

simulated using displacement control boundary conditions but not with load control boundary conditions, and vice versa. To fully verify the damage model input, both monotonic loading conditions and load/unload conditions should be simulated. To ensure that the post-peak stress degradation (if specified in the input deck) is properly implemented, this behavior should be simulated as part of the single element simulations. It is also recommended to develop a deformation-only MAT_213 model first before attempting to implement damage and/or failure. There are several cautions that should be mentioned in terms of conducting single element simulations: 1. Occasionally instabilities will be encountered in single element simulations that will disappear when multi-element models are simulated. 2. Many times verification studies will concentrate on quasi-static room temperature simulations with limited consideration being given to higher rates and temperatures. 3. In examining the results from the single element simulations, the stresses, strains and plastic strains in the directions other than the primary loading direction are often ignored, where if these values are unreasonable higher structural level simulations could yield improper results. 4. The TCSYM parameter should be set to zero in order to match individual input curves. 5. A lower TSSFAC parameter in *CONTROL_TIMESTEP may be required for single elements compared to multi-element models. 6. Sufficient state variables should be allocated using the NEIPH (for solids) or NEIPS (for shells) parameters in *DATABASE_EXTENT_BINARY so that output damage variables can be compared (if included). 7. MAT_213 does not generate mesh objective results particularly when failure is added since no regularization scheme is implemented. The selected element size should be consistent with the approximate element size for the problem of interest. There are several considerations which could result in the single element simulation results not matching the input curves: 1. Shear loading is difficult to implement in a single element simulation, so the shear stressshear strain curves may not be computed correctly. 2. Sharp corners in the input curves may be difficult to simulate numerically. 3. If off-axis tests are simulated, the computed results may not match the provided input curves since MAT_213 often internally adjusts the provided off-axis curves to eliminate irregularities in the computed yield surface. COUPON LEVEL VERIFICATION STUDIES As a next step in the input/model verification process, coupon level models can be constructed and simulated. Specifically, multi-element models with the key features of the actual tests which were utilized to generate the data entered into the input curves can be analyzed. While this step in the process is frequently omitted, and analysts proceed directly from single element verification studies to analyzing full components, there are several important reasons why this step should not be neglected. By carrying out the coupon level simulations, the full stress and strain field generated during the actual characterization tests can be analyzed. Particularly if the full strain field experienced by the tested coupon is captured using d igital image correlation, the strain fields computed by the simulation can be compared to the actual experimental strain fields. In this way, if there are differences between the computed and

© ASCE

497

498

Earth and Space 2022

experimental stress-strain curves, a more in-depth study can be conducted to determine the causes of the discrepancies. For example, the effects of boundary conditions can be examined in detail. An example of a finite element model used for a tension test simulation is shown in Figure 4, and an example of a finite element model of a shear test specimen is shown in Figure 5. Note that in both cases the grips are not modeled, to capture the key portion of the experimental specimens. Both tests are modeled using simple straight sided specimens. For the tensile specimen, the nodes on the left hand side of the model are constrained, and a constant displacement in the x direction is applied to the nodes on the right face. For the shear simulations, the nodes on the bottom surface are constrained and a displacement in the x direction is applied to the top surface.

Figure 4. Finite element model for a tensile test simulation.

Figure 5. Finite element model for a shear test simulation.

© ASCE

Earth and Space 2022

There are several important considerations which should be kept in mind while conducting coupon level verification studies. For example, the boundary conditions applied to the analysis model may not absolutely replicate the boundary conditions applied in the actual test. For example, a tensile load may be applied to the gage section rather than explicitly modeling tabs and transferring load to the gage section through the tabs. Another important factor to consider, particularly when examining the post-peak damage and ultimate failure response, is that values of the degraded stress and effective erosion strain that are calibrated to obtain an optimal match to the coupon level behavior may lead to premature failures when used in the analysis of an actual component. As a result, these values may need to be calibrated based on the actual component response. Research efforts are currently underway to determine a defined methodology for calibrating the post -peak stress degradation response and failure strains based on the results of coupon tests that are also valid when analyzing a full component. VALIDATION STUDY OVERVIEW After an input deck with its associated input curves and characterized numerical parameters is verified, validation studies can take place. In a properly executed verification study, a structural scale analysis will be carried out using the exact input deck developed in earlier steps in the process, with no further calibrations taking place based on the results of the simulations. In this way, ability of the model to correctly simulate a problem of interest can be appropriately determined. Currently, erosion strains may have to be calibrated to a specific test. In studies conducted at the NASA Glenn Research Center, various flat panel impact tests conducted based on the ASTM D8101/D8101M-17 standard (ASTM 2017) were simulated in order to validate the ability of the code to predict the impact response of composite structures under realistic impact loading conditions. For these studies, input decks were constructed and verification analyses were performed using the process described in the previous sections of this paper. Details of the impact simulation studies will be described in the next section . SIMULATION OF FLAT PANEL IMPACT TEST A series of dynamic impact tests on flat plates made of an IM7/8552 composite were performed at NASA Glenn Research Center in accordance with ASTM D8101/D8101M-1732 (ASTM 2017). The techniques used to conduct these tests and a sample of the test results are described in Melis, et al (2018). The projectile was a hollowed out cylinder with a hemispherical nose as shown in Figure 6. The projectile was made of Aluminum 6061 with a nominal mass of 91 g. The composite plate was a rectangular plate with a nominal length and width of 30.5 cm. A 40 ply composite with a laminate orientation of [45/90/-45/0]5s was used, with a ply thickness of 0.0183 cm, leading to a total plate thickness of 0.73 cm. The plate was held in place by being bolted to a circular clamp with a nominal diameter of 25.4 cm. High speed digital image correlation (DIC) was used to accurately measure the backside displacement of the plate during impact. A single stage gas gun was used to propel the projectile such that it impacted the plate normal to the plate plane nominally at the center of the plate. The plates were impacted at velocities ranging from 12.5 m/s to 123.1 m/s. A photo of a typical plate used in the impact tests is shown in Figure 7, including the applied speckle pattern which is required to obtain the high speed DIC data.

© ASCE

499

500

Earth and Space 2022

Figure 6. Projectile used in impact tests.

Figure 7. Composite plate used in impact tests. The plate was modeled using 932,960 constant stress solid elements and the projectile was modeled using 30,496 elements. Only the portion of the plate within the clamping fixture was modeled. The plate was constrained at the bolt locations. A sample mesh of the plate is shown in Figure 8 and a sample mesh of the plate and projectile is shown in Figure 9. The projectile was modeled using a piecewise linear plasticity model with standard Aluminum 6061 properties. The material card input for the composite plate was developed using stress-strain curves obtained from Justusson (2020).

Figure 8. Finite element mesh of composite plate used for impact simulations

© ASCE

Earth and Space 2022

Figure 9. Finite element mesh of composite plate and projectile used for impact simulations. As an example of the types of validation results that were obtained, Figure 10 shows the experimental and simulated backside out-of-plane displacements as a function of time at the center of impact predicted for a case where the projectile impact velocity is low enough that visible damage was not observed in the structure. The simulated displacements corresponded closely to what was observed experimentally. The important point to emphasize is that the structural level response was simulated with no additional correlation beyond that which was executed in the single element and coupon level simulations. Other validation analyses (not discussed here for conciseness) were conducted, such as examining the variati on of the backside displacement over the entirety of the surface of the flat plate, comparisons of the predicted damage patterns to the damage patterns observed in the experiments and the comparison of the predicted exit or rebound velocity of the projectile to that observed in the experiments.

Figure 10. Backside out-of-plane displacement at center of impact as a function of time for an impact velocity below which damage is observed.

© ASCE

501

Earth and Space 2022

CONCLUSIONS A generalized composite model suitable for use in polymer composite impact simulations has been developed. The model utilizes a plasticity based deformation model based on generalizing the Tsai-Wu failure criteria. A strain equivalent damage model has also been developed in which loading the material in a particular coordinate direction can lead to damage in multiple coordinate directions. A tabulated failure model has been developed which facilitates the use of actual composite failure data in an analysis. This material model has been successfully implemented within the commercial transient dynamic finite element code LS-DYNA as MAT_213. To facilitate the ability of the general user community to successfully employ the code to simulate impact, crash and crush problems, a series of best practices and calibration, verification and validation processes and procedures have been developed. Ongoing efforts will involve creating a detailed free-standing document which includes expanded detail and multiple example problems beyond those presented in this paper in order to provide a useful introduction of MAT_213 to the general user community. Validation studies have demonstrated that systematic use of the recommendations presented here can lead to successfully simulating complex dynamic events. REFERENCES Achstetter, T. (2019). Development of a Composite Material Shell-Element Model for Automotive Impact Applications. PhD Dissertation, George Mason University, Fairfax, VA. ASTM (2017). ASTM D8101/D8101M-17. Standard test method for measuring the penetration resistance of composite materials to impact by a blunt projectile. ASTM D8101/D8101M-17. ASTM International, West Conshohocken, PA. Daniel, I.M. and Ishai, O. (2006). Engineering Mechanics of Composite Materials Second Edition. Oxford University Press, New York. Goldberg, R., Carney, K., DuBois, P., Hoffarth, C., Harrington, J., Rajan, S., and Blankenhorn, G. (2014). “Theoretical Development of an Orthotropic Elasto-Plastic Generalized Composite Model.” NASA/TM-2014-218347, National Aeronautics and Space Administration, Washington, DC. Goldberg, R., Carney, K., DuBois, P., Hoffarth, C., Harrington, J., Rajan, S., and Blankenhorn, G. (2015a). “Development of an Orthotropic Elasto-Plastic Generalized Composite Material Model Suitable for Impact Problems,” Journal of Aerospace Engineering, 10.1061/(ASCE)AS.1943-5525.000058004015083. Goldberg, R., Carney, K., DuBois, P., Hoffarth, C., Rajan, S., and Blankenhorn, G. (2015b). “Incorporation of Plasticity and Damage into an Orthrotropic Three-Dimensional Model with Tabulated Input Suitable for Use in Composite Impact Problems,” NASA/TM-2015218849, National Aeronautics and Space Administration, Washington, D.C. Goldberg, R., Carney, K., DuBois, P., Hoffarth, C., Khaled, B., Rajan, S., and Blankenhorn, G. (2018a). “Analysis and Characterization of Damage Using a Generalized Composite Material Model Suitable for Impact Problems,” Journal of Aerospace Engineering, 10.1061/(ASCE)AS.1943-5525.0000854. Goldberg, R., Carney, K., DuBois, P., Hoffarth, C., Harrington, J., Shyamsunder, L., Rajan, S., and Blankenhorn, G. (2018b). “Implementation of a tabulated failure model into a generalized composite material model,” Journal of Composite Materials, 52, 3445–3460.

© ASCE

502

Earth and Space 2022

Hallquist, J. (2013). LS-DYNA Keyword User’s Manual, Version 970, Livermore Software Technology Corporation, Livermore, CA. Hashin, Z. (1980). “Failure Criteria for Unidirectional Fiber Composites,” Journal of Applied Mechanics, 47, 329-334. Justusson, B., Molitor, M., Iqbal, J., Ricks, T.M., and Goldberg, R.K. (2020). “Overview of Coupon Testing of IM7/8552 Composite Required to Characterize High-Energy Impact Dynamic Material Models”, NASA TM-2020-220498, National Aeronautics and Space Administration, Washington, D.C. Melis, M.E., Pereira, J.M., Goldberg, R.K., and Rassaian, M. (2018). “Dynamic Impact Testing and Model Development in Support of NASA’s Advanced Composites Program.” AIAA SciTech Forum, 2018 AIAA/ASCE/AHS/ASC Structures, Structural Dynamics, and Materials Conference, AIAA, Reston, VA. Puck, A., and Schurmann, H. (1998). “Failure Analysis of FRP Laminates by Means of Physically Based Phenomenological Models,” Composites Science and Technology, 58, 1045–1067.

© ASCE

503

Earth and Space 2022

Modeling and Analysis of a Nonlinear Locally Resonant Metamaterial with Inductance Shunt Arun Malla1 , Mohammad Bukhari1 , and Oumar Barry1 1

VibRo Lab, Department of Mechanical Engineering, Virginia Polytechnic University, 301 Durham Hall, Blacksburg, VA, 24060

Abstract This paper investigates a weakly nonlinear metamaterial with electromechanical local resonators coupled to a resistance-inductance shunt circuit. An analytical solution is developed for the system using the perturbation method of multiple scales, and validated through direct numerical integration. Linear and nonlinear band structures are used for parametric analysis of the system, focusing on the effect of system parameters on band gap formation and vibration attenuation. In addition, the effects of nonlinearity and the interaction with shunt parameters are examined. Results describe multiple methods of tuning band gaps and pass bands of the system through various parameters, demonstrating the flexibility and potential of the examined metamaterial.

Introduction Metamaterials are artificially engineered structures that possess properties not found in naturally occurring materials (Hussein et al. 2014). The unusual features of metamaterials make them beneficial for numerous applications including vibration and noise control, energy harvesting, non-destructive testing, and acoustic rectifiers. Metamaterials consist of many unit cells arranged in periodic or aperiodic patterns. It has been observed that periodic structures prevent waves from propagating through the structure at certain frequency ranges, known as band gaps (Kushwaha et al. 1993). Because these band gaps are constrained by the unit cell dimensions, the application of basic metamaterials was limited to large structures (Hussein et al. 2014). To expand the use of metamaterials to smaller components,

© ASCE

504

Earth and Space 2022

Liu et al. (2000) introduced local resonators, showing that locally resonant metamaterials are able to control vibrations at wavelengths much smaller than the lattice constant. Local resonators are also capable of widening the original band gap. Further manipulation of the system’s band structure can also be achieved by introducing multiple resonators (Huang and Sun 2010). Band gaps can also be affected by the incorporation of piezoelectric materials and shunt circuits. Piezoelectric materials have long been utilized in both active and passive methods for vibration control (Hagood and von Flotow 1991), and more recently have been incorporated into metamaterials (Thorp et al. 2001). By including piezoelectric elements in a metamaterial, the mechanical system dynamics can be coupled to an easily modifiable shunt circuit, enabling convenient adjustment of the metamaterial’s overall properties. Incorporating piezoelectric materials and shunt circuits enables techniques such as the use of negative capacitance (Beck et al. 2011) or resonant shunt circuits (Wang and Chen 2015) to control vibrations and create or broaden band gaps. In addition, shunt circuits also offer an avenue for simultaneous energy harvesting. An important parameter in any work involving piezoelectric materials is the system’s electromagnetic coupling factor. This parameter is dependent on the design and material properties of the piezoelectric component (Sugino et al. 2017). Though this piezoelectric coupling coefficient is usually on the order of 10−10 for engineering applications (Erturk and Inman 2011), signifying weak electromagnetic coupling, some features may only be apparent in the case of strong electromagnetic coupling. Combining the two previously discussed methods, researchers have also investigated metamaterials with both local resonators and shunt circuits. Sugino et al. (2017) studied a locally resonant material coupled to a shunt with piezoelectric elements. This work differs from Sugino’s by incorporating the piezoelectric material into the local resonators. In addition, the effects of nonlinearity and potential interactions with shunt parameters are examined. This paper applies analytical and numerical methods to investigate the effect of electromechanical coupling and shunt circuit parameters on wave propagation and energy harvesting in a nonlinear acoustic metamaterial with resistance-inductance shunt. The metamaterial is modeled through a nonlinear system of governing equations, and the perturbation method of multiple scales is utilized to derive an approximate solution. This solution is validated against the direct analytical solution for the linear case, and the directly integrated numerical results for the nonlinear case. The solution from the method of multiple scales is then used to analyze the effect of parameters on the system band structure as well as their interaction with nonlinearity, focusing on the resultant applications for simultaneous vibration control and energy harvesting.

© ASCE

505

506

Earth and Space 2022

Figure 1: Schematic of nonlinear metamaterial with electromechanical resonators.

Mathematical Modeling of the System The system studied in this work is a nonlinear acoustic metamaterial coupled with electromechanical local resonators. The metamaterial consists of a chain of cells connected by nonlinear springs as shown in Figure 1. Each nonlinear spring has linear spring coefficient K and nonlinear spring coefficient α ¯ . Each resonator consists of a substrate covered by a piezoelectric layer, with total effective mass mp and effective linear stiffness kp . The piezoelectric layer is shunted to a resistance-inductance (RL) circuit as shown in Figure 1. This circuit has voltage difference v¯n , resistance R, and inductance L. The piezoelectric layer has capacitance Cp and electromechanical (EM) coupling coefficient θ. The absolute displacement of cell n is u¯n , and the absolute displacement of the attached piezoelectric resonator is y¯n ∗ . Following Bukhari and Barry (2020), we introduce nondimensional variables: un = u¯n /U0 , yn = y¯n /U0 , and vn = v¯n /V0 , where U0 and V0 are initial displacement and velocity, respectively. Here, y¯n = y¯n∗ − u¯n is the relative displacement of piezoelectric local resonator n with respect to cell n. We also introduce nondimensional time τ = ωn t, where q ωn = K/M (1) is the mechanical natural frequency of a unit cell. With these variables, the normalized coupled equations of motion for each cell, local resonator, and shunt circuit can be written for an infinite chain as: ¯ 2 (¨ u¨n + 2un − un+1 − un−1 + α(un − un+1 )3 + α(un − un−1 )3 + kΩ ¨n ) = 0 (2) 1 un + y Ω21 y¨n + yn − α1 (α3 v˙ n + vn ) = −Ω21 u¨n

(3)

Ω22 v¨n + α2 v˙ n + vn + α4 y˙ n = 0

(4)

where α = α ¯ U02 /K, Ω1 = ωn /ωp , ωp = q

q

kp /mp , k¯ = kp /K, α1 = θV0 /kp U0 , α3 =

Lωn /R, Ω2 = ωn /ωe , ωe2 = 1/ LCp , α2 = RCp ωn , and α4 = Rθωn U0 /V0 .

© ASCE

507

Earth and Space 2022

To develop an approximate solution of the nonlinear governing equations of motion Eqs. (2)-(4), the perturbation method of multiple scales is utilized. This method is described in the following section.

Solving With Method of Multiple Scales The perturbation method of multiple scales (MMS) is carried out following the procedure outlined by Nayfeh (2011). First we introduce a small dimensionless parameter ϵ (ϵ ≪ 1) in the governing equations by defining multiple time scales: T0 = τ ,

T1 = ϵτ

(5)

where T0 is the fast time scale and T1 is the slow time scale. The time derivative operators are then perturbed and can be expressed as δ = D0 + ϵD1 + O(ϵ2 ) δτ

(6a)

δ2 = D02 + 2ϵD0 D1 + O(ϵ2 ) (6b) δτ 2 where Dn = δ/δTn . Following this, the solutions of the nonlinear governing equations of motion (Eqs. (2) - (4)) can be expressed as power series in powers of ϵ as un (τ ) = un,0 (T0 , T1 ) + ϵun,1 (T0 , T1 ) + O(ϵ2 ) ,

(7a)

yn (τ ) = yn,0 (T0 , T1 ) + ϵyn,1 (T0 , T1 ) + O(ϵ2 ) ,

(7b)

vn (τ ) = vn,0 (T0 , T1 ) + ϵvn,1 (T0 , T1 ) + O(ϵ2 ) .

(7c)

The governing equations are converted into a weakly nonlinear form by rescaling the parameter α = αϵ. Through introducing this substitution and Eqs. (5)–(7) into Eqs. (2)–(4), then collecting different orders of ϵ we get the linear and nonlinear problems. For O(ϵ0 ) terms, the problem is linear. Thus, the solution can be expressed as: un,0 (T0 ) = Aei(nk−ωT0 ) + c.c.

(8a)

yn,0 (T0 ) = Bei(nk−ωT0 ) + c.c.

(8b)

vn,0 (T0 ) = Cei(nk−ωT0 ) + c.c.

(8c)

where k is the wavenumber and ω is the linear frequency normalized by mechanical natural frequency ωn . A, B, and C are functions of the slow time scale T1 , and c.c. denotes the complex conjugate of the preceding term. Solving the linear problem gives B and C in terms of A: C = Γ1 B

© ASCE

(9)

508

Earth and Space 2022

B = Γ2 A

(10)

as well as the linear dispersion relation: ¯ 2 (1 + Γ2 ) = 0 −ω 2 + 2 − 2cos(k) − Ω21 kω

(11)

where Γ1 and Γ2 are functions of ω 2 and various parameters. Solving Eq. (11) reveals six roots for ω in the form of three complex conjugate pairs. Consequently, the band structure may have up to three pass bands depending on the system parameters. Collecting O(ϵ1 ) terms provides the nonlinear problem, which can be used to determine A(T1 ). The polar form of A(T1 ) is defined as: 1 A(T1 ) = a(T1 )eib(T1 ) 2

(12)

Solving for a and b yields the equations: ga′ + hab′ + f = 0

(13)

−ha′ + gab′ + l = 0

(14)

where g, h, f,and l are functions of ω and system parameters. Here, prime (′ ) denotes the derivative with respect to T1 . Eqs. (13) and (14) can then be solved for the slow flow equations: a′ = c 0 a3

(15)

b ′ = c 1 a2

(16)

where c0 and c1 are functions of g, h, f,and l. By comparing the values of c0 and c1 , it can be observed that the magnitude of c0 is extremely small compared to that of c1 . Therefore, it can be assumed negligible, c0 ≈ 0. This gives a′ = 0 and a(T1 ) = a constant a0 . Thus, the nonlinear frequency correction factor is:

and the nonlinear frequency is:

b′ = c1 a20

(17)

ωN L = ω − ϵb′

(18)

Integrating Eq. (17) yields the approximate solution for b(T1 ): b = c1 a20 T1

(19)

The linear and nonlinear dispersion curves derived by the method of multiple scales are then validated against a numerical solution and used to conduct band structure analysis of the system.

© ASCE

509

Earth and Space 2022

(a)

(b)

Figure 2: Validation of numerical and analytical results; (a) Linear problem, ϵαa20 = 0; (b) Nonlinear problem, ϵαa20 = 0.06.

Band Structure Analysis 0.1

Validation

To validate the linear and nonlinear dispersion relations, they are compared to a numerical simulation of the system. The system is as described in Figure 1, with 500 cells coupled to electromechanical resonators with RL shunt circuits. Cells are connected with both linear and nonlinear springs. Shunt parameter values are: R = 103 Ω, L = 0.2212 H, and Cp = 1.13e-10 F. The mass of each main cell is M = 0.125 kg, and the mass ratio between each piezeoelectric resonator and main cell is mp /M = 0.1. The mechanical and electrical resonance frequencies of the resonator, ωp and ωe , are tuned such that ωn = ωp = ωe = 2e5 rad/s. These parameters are chosen based on the similar system examined by Abdelmoula and Abdelkefi (2015). The case of weak EM coupling θ = 110 N/V is considered. Following the similar procedure described by Bukhari and Barry (2020), the system is validated by exciting the chain with a transient wave packet and numerically integrating the governing equations using the built-in MATLAB solver ode45. For a given wavenumber, the system is simulated for a long time to allow the wave to propagate through the chain. After this, a 2D Fast Fourier transform (2DFFT) is applied to the time data collected in the wavenumber and frequency domains. The transformed data is then used to determine the natural frequency of the system by finding the frequency associated with the maximum power density point. By determining the natural frequencies corresponding to a range of wavenumbers over the first Brillouin zone, the dispersion curves are numerically constructed and compared to the analytical solution in Figure 2, which plots normalized frequency ω against wavenumber k. Both linear and nonlinear dispersion relations are compared in Figure 2. These comparisons show good agreement between analytical and numerical solutions for both the linear and nonlinear dispersion relations. It should be noted that the

© ASCE

510

Earth and Space 2022

numerical solution is unable to capture certain frequencies of the nonlinear solution in the medium and short wavelength limits for both branches. These areas, known as pseudo band gaps, are due to a significant frequency shift associated with transient wave packet excitation (Zhou et al. 2018). When excited within these pseudo band gaps, the solution appears instead at frequencies within the long and short wavelength limits.

(a)

(b)

(c)

(d)

Figure 3: Effect of system parameters on linear band structure, ωe = ωn : (a) Effect of EM coupling, mp /M = 0.1, R = 103 Ω, L = 0.2212 H; (b) Effect of resonator/main cell mass ratio, θ = 10−1 N/V, R = 103 Ω, L = 0.2212 H; (c) Effect of shunt resistance, θ = 10−1 N/V, mp /M = 0.1, L = 0.2212 H; (d) Effect of shunt inductance, θ = 10−1 N/V, mp /M = 0.1, R = 103 Ω.

Linear Band Structure Next, the validated analytical dispersion relations are used to study the effect of selected parameters on the band structure, beginning with the linear dispersion relation calculated using Eq. (11). The parameters examined are EM coupling coefficient θ, the mass ratio between the piezoelectric resonator and main cell mp /M , shunt resistance R, and piezoelectric capacitance Cp . Unless otherwise noted, parameter values are the same as in Section 0.1. Comparisons are also made to the band structure of

© ASCE

511

Earth and Space 2022

the system examined by Bukhari and Barry (2020), which is similar except for the absence of inductor in the shunt circuit. Varied EM coupling is shown in Figure 3(a), with the value of θ ranging from weak EM coupling at θ = 1e-10 N/V to strong EM coupling at θ = 1e-1 N/V. For weak coupling values, θ = 1e-10 to 1e-3 N/V, the coupling has minimal effect on the band structure, which is similar to the case of resistance-only shunt circuit in Bukhari and Barry (2020). However, at θ = 1e-2 N/V, a third mode branch can be observed forming between the acoustic and optical modes. Instead of a single band gap, two smaller band gaps are formed. These effects are more visible for θ = 1e-1 N/V.

(a)

(b)

Figure 4: Effect of system parameters on transmissibility, ωe = ωn , R = 103 Ω, L = 0.2212 H: (a) Effect of EM coupling, mp /M = 0.1; (b) Effect of resonator/main cell mass ratio, θ = 10−1 N/V. Figure 3(b) displays the effects of varying the mass ratio between the piezoelectric resonator and main cell. Here, the main cell mass is kept fixed at M = 0.125 kg and the resonator mass is altered to meet the desired ratio. In addition to band gap location, resonator mass has a clear effect on the width of the band gaps between the modes, with the band gaps broadening as resonator mass increases. This is consistent with the effects of mechanical-only resonators, which are well established in the literature (Inman 1994). These broadening band gaps effect the middle mode branch, which becomes increasingly narrow as mass ratio increases. In addition, the upper boundary of the acoustic mode significantly increases with mass ratio. The effects of shunt resistance are shown in Figure 3(c). It is clear that this value effects the formation of the central mode branch. At low resistance values R = 102 to 103 Ω, all three modes are present, but the central branch merges back into the acoustic and optical modes as resistance increases. This high resistance case is near identical to the band structure with resistance-only shunt studied by Bukhari and Barry (2020). Due to this observation regarding the effect of high shunt resistance, R = 103 Ω for all following analysis. In Figure 3(d), varied shunt inductance q is examined. Due to the electrical resonance frequency being fixed at ωe = 1/ LCp = ωn , increasing L means decreasing Cp and vice versa. Increasing L

© ASCE

Earth and Space 2022

results in the central mode branch becoming broader as the band gaps move apart. To supplement the study of the linear band structure, frequency response functions (FRFs) are also obtained through direct numerical simulation of the 500-cell metamaterial chain system. The transmissibility of the chain for each frequency is determined by comparing the output power harvested by the 500th cell to the input power at the 1st cell: R tf Pout (t)dt Hp = R0tf (20) 0 Pin (t)dt where tf = 3000 s. The resulting FRFs, shown in Figure 4, support the observations drawn from the band structure analysis while also making other details more apparent. Figure 4(a), displaying the effects of θ, shows similar results to the band structure in Figure 3(a), with the band gap remaining consistent from θ = 1e-10 to 1e-3 N/V, and splitting into two at θ = 1e-1 N/V. However, at θ = 1e-2 N/V, the band gap broadens, but the central mode branch is not yet visible. This suggests that the mode branch is too narrow to be visible, or that there is no significant difference in vibration propagation between the mode and the band gaps for this value of θ. Another notable observation is the transmissibility of the central mode branch at strong coupling, θ = 1e-1 N/V. While waves certainly propagate with higher amplitude in the mode than the band gaps, the transmissibility is also significantly less than in the acoustic and optical modes. In addition, transmissibility drops off gradually as the band gaps are approached, unlike the lower-coupling cases. Similar observations are made by varying mass ratio, shown in Figure 4(b)

Nonlinear Band Structure The nonlinear band structure is also studied by taking into account the correction factor defined by Eq. (18). The nonlinear stiffness parameter ϵαA2 is varied to model chains with nonlinear hardening (ϵαA2 = 0.03) and nonlinear softening (ϵαA2 = −0.03). The effects of nonlinear hardening and softening are then combined with the effects of EM coupling, mass ratio, and shunt inductance by varying selected parameters. The effect of nonlinear hardening or softening on the dispersion curves with default parameters is shown in Figure 5(a). As expected, hardening shifts the curves up on the frequency axis, while softening shifts them down. The magnitude of the shift is negligible for small wavenumber (long wavelength limit), and more pronounced for the optical mode branch, especially at large wavenumber (short wavelength limit). For the acoustic mode branch and central mode branch, the effects of nonlinearity are largest around the medium wavelength region, but decrease near the end of each mode. Thus, the effect on bandgap boundaries is present, but minimal. However, the upper bound of the optical mode, where nonlinearity has prominent effect, is significantly shifted.

© ASCE

512

513

Earth and Space 2022

(a)

(b)

(c)

(d)

Figure 5: Effect of system parameters on nonlinear band structure for: (a) Default parameters; (b) Weak EM coupling θ = 10−10 N/V ; (c) Large mass ratio mp /M = 0.5; (d) Shunt inductance L = 2L0 . The cases of linear and nonlinear chains with weak EM coupling are displayed in Figure 5(b). Here, the shifts of the cutoff frequencies due to nonlinear hardening and softening are largely the same as the case of strong coupling. The band structure has two branches rather than three, matching the effects of weak coupling previously established. Again, the bounds at the top of the acoustic mode and the bottom of the optical mode are not significantly affected by nonlinearity, with larger effects visible at the upper bound of the optical mode. Similar results are seen when altering the mass ratio, as displayed in Figure 5(c). As expected from the linear analysis, the central mode branch becomes narrow, and band gaps are broader than the default case. The effects of nonlinearity also become slightly more prominent in the acoustic mode, and remain significant in the optical mode. Finally, the shunt inductance is increased to L = 2L0 in Figure 5(d). As noted in Figure 3(d), this results in the central mode branch occupying a larger frequency range than the default case in Figure 5(a). Due to this, the effects of nonlinearity are more prominent for this mode, especially in the medium wavelength region. However, effects of nonlinearity still decrease in the short wavelength region, reducing the effect on the band gap boundary.

© ASCE

Earth and Space 2022

Conclusions This work contains the analysis of a nonlinear, electromechanical metamaterial coupled to a shunt circuit with both resistor and inductor. The system consists of a chain of cells connected by nonlinear springs, with each cell coupled to an electromagnetic resonator consisting of piezoelectric element and shunt circuit. Both the cells and resonators were modeled as spring-mass systems, with the resonator system coupled to the dynamics of the shunt circuit. The resulting governing equations were solved analytically using the perturbation method of multiple scales, and the approximate solution was validated by direct numerical integration. The validated analytical solutions were then used to examine the system’s band structure to view the effect of electromagnetic coupling and resonator parameters including mass ratio, shunt resistance, and shunt inductance. For the linear case, band structure analysis was also supported by transmissibility diagrams obtained by solution of the linear system equations. Focus was given to manipulating band gaps and the system’s mode branches through resonator parameters, and the resulting effects on vibration attenuation within the chain. In this paper, several notable observations were made concerning shunt parameters. It was shown that using strong EM coupling and low resistance, a band structure with three pass bands can be obtained, rather than the two pass band structure from low EM coupling or resistance-only shunt. The effects of nonlinear springs on the system dynamics, especially the interaction with the resonator and shunt parameters, were also studied. With the insight gained from the parametric study in this work, there is great potential for optimizing both vibration control and energy harvesting in this and similar metamaterials.

References Abdelmoula, H. and Abdelkefi, A. (2015). “Ultra-wide bandwidth improvement of piezoelectric energy harvesters through electrical inductance coupling.” The European Physical Journal Special Topics, 224, 2733–2753. Beck, B. S., Cunefare, K. A., Ruzzene, M., and Collet, M. (2011). “Experimental analysis of a cantilever beam with a shunted piezoelectric periodic array.” Journal of Intelligent Material Systems and Structures, 22, 1177–1187. Bukhari, M. A. and Barry, O. (2020). “Simultaneous energy harvesting and vibration control in a nonlinear metastructure: A spectro-spatial analysis.” Journal of Sound and Vibration, 473, 115215. Erturk, A. and Inman, D. J. (2011). Piezoelectric energy harvesting. John Wiley & Sons.

© ASCE

514

Earth and Space 2022

Hagood, N. W. and von Flotow, A. (1991). “Damping of structural vibrations with piezoelectric materials and passive electrical networks.” Journal of Sound and Vibration, 146(2), 242–268. Huang, G. and Sun, C. (2010). “Band gaps in a multiresonator acoustic metamaterial.” Journal of Vibration and Acoustics, 132(3), 031003. Hussein, M. I., Leamy, M. J., and Ruzzene, M. (2014). “Dynamics of phononic materials and structures: Historical origins, recent progress, and future outlook.” Applied Mechanics Reviews, 66(4), 040802. Inman, D. J. (1994). Engineering vibration. Prentice Hall. Kushwaha, M. S., Halevi, P., Dobrzynski, L., and Djafari-Rouhani, B. (1993). “Acoustic band structure of periodic elastic composites.” Physical review letters, 71(13), 2022. Liu, Z., Zhang, X., Mao, Y., Zhu, Y., Yang, Z., Chan, C. T., and Sheng, P. (2000). “Locally resonant sonic materials.” science, 289(5485), 1734–1736. Nayfeh, A. H. (2011). Introduction to perturbation techniques. John Wiley & Sons. Sugino, C., Leadenham, S., and Ruzzene, M. (2017). “An investigation of electroelastic bandgap formation in locally resonant piezoelectric metastructures.” Smart Materials and Structures, 26, 055029. Thorp, O., Ruzzene, M., and Baz, A. (2001). “Attenuation and localization of wave propagation in rods with periodic shunted piezoelectric patches.” Smart Materials and Structures, 10(5), 979. Wang, G. and Chen, S. (2015). “Large low-frequency vibration attenuation induced by arrays of piezoelectric patches shunted with amplifier-resonator feedback circuits.” Smart Materials and Structures, 25, 331–346. Zhou, W. J., Li, X., Wang, Y., Chen, W., and Huang, G. (2018). “Spectro-spatial analysis of wave packet propagation in nonlinear acoustic metamaterials.” Journal of Sound and Vibration, 413, 250–269.

© ASCE

515

Earth and Space 2022

Ballistic Impact Simulations of a Titanium 6Al-4V Generic Fan Blade Fragment on an Aluminum 2024 Panel Using *MAT_224 in LS-DYNA Chung-Kyu Park1; Kelly Carney2; Paul Du Bois3; Cing-Dao Kan4; and Daniel Cordasco5 1

Center for Collision Safety and Analysis, George Mason Univ., Fairfax, VA (corresponding author). Email: [email protected] 2 Center for Collision Safety and Analysis, George Mason Univ., Fairfax, VA 3 Center for Collision Safety and Analysis, George Mason Univ., Fairfax, VA 4 Center for Collision Safety and Analysis, George Mason Univ., Fairfax, VA 5 Federal Aviation Administration, William J. Hughes Technical Center, Atlantic City International Airport, NJ ABSTRACT As a part of the FAA’s Aircraft Catastrophic Failure Prevention Program, advanced Aluminum 2024 and Titanium 6Al-4V material models utilizing *MAT_224 in LS-DYNA have been developed to improve the numerical modeling of turbine engine blade-out containment tests required for certification of aircraft engines. In this effort, NASA conducted four ballistic impact tests on large flat Aluminum 2024 panels with a blade-shaped Titanium 6Al-4V projectile to provide experimental data to evaluate the numerical material model. These tests were designed to represent a realistic turbine engine fan-blade release event. In this research, ballistic impact tests were simulated using advanced Aluminum 2024 and Titanium 6Al-4V material models to validate the material models under simulated turbine engine blade release event conditions. The research also identifies possible challenges for such a ballistic impact simulation with a bladeshaped projectile that slides, bends (plastically deforms), may fracture, and rotates as it moves in three dimensions. INTRODUCTION A team consisting of George Mason University, Ohio State University, George Washington University, the National Aeronautics and Space Administration (NASA) - Glenn Research Center (GRC), and the Federal Aviation Administration (FAA) - Aircraft Catastrophic Failure Prevention Program (ACFPP) collaborated to develop a new material model in LS-DYNA for metallic materials. The research was directed towards improving the numerical modeling of turbine engine blade-out containment tests required for certification of aircraft engines (Emmerling et al, 2014). In this effort, the LS-DYNA constitutive material model *MAT_224 was applied. *MAT_224 is a general elasto-visco-plastic material model that utilizes a tabulated approach to incorporate arbitrary stress versus strain curves to define material plasticity, including arbitrary strain rate and temperature dependency. The element erosion criterion is the plastic failure strain, which can be defined as a function of the state of stress, strain rate, temperature, and element size. The updated *MAT_224 input parameters (Version 2.2) for Aluminum 2024-T351 alloy (Park et al, 2020a) plates were developed based on tabulated data from several material tests performed by OSU (Seidt, 2014), and was released recently (LS-DYNA Aerospace Working

© ASCE

516

Earth and Space 2022

Group, 2021). It has been validated intensively with three series of ballistic impact tests: (1) ballistic impact tests of a sphere projectile to square Aluminum 2024 plates with various thicknesses (Park et al, 2020a) (Kelley and Johnson, 2006), (2) ballistic impact tests of a cylinder projectile to circular Aluminum 2024 plates with various thicknesses (Park et al, 2020a) (Pereira et al, 2013), and (3) ballistic impact tests of 1/8-inch thick Aluminum 2024 plates with rectangular projectiles, having varying oblique incidence and attitude angles (Pereira et al, 2013) (Park et al, 2020b). Overall, the ballistic impact simulations using the updated Aluminum 2024 *MAT_224 material model show good correlations to the lab tests for a broad range of test conditions. In addition, the *MAT_224 input parameters for Titanium 6Al-4V alloy (Version 1.3) projectile were also developed using a series of material tests conducted by OSU (Haight et al, 2016) (Hammer, 2014) and released (LS-DYNA Aerospace Working Group, 2021). It was validated with dynamic punch tests (Haight et al, 2016) (Hammer, 2014) and ballistic tests of cylinder projectiles impacting circular 1/2-inch-thick Titanium 6Al-4V plates (Pereira et al, 2013) (Haight et al, 2016). Four ballistic impact tests on large flat Aluminum 2024 panels with a blade-shaped Titanium 6Al-4V projectile were conducted by the NASA Glenn Research Center to provide experimental data to evaluate the numerical material model (Pereira et al, 2013). These tests were designed to represent key aspects of a real turbine engine fan-blade release event where a released fan blade tip makes contact with the engine case, bends, may fracture, and then rotates to cause the heavy blade root to impact (and potentially penetrate) the engine case. In all the tests, the projectiles were contained, but the panels were perforated in three tests. The critical velocity of the panel perforation was estimated based on the test data. In this research, these ballistic impact tests were simulated using advanced Aluminum and Titanium material models using *MAT_224 in LS-DYNA. This effort had two objectives: (1) to further validate those material models under more realistic turbine engine blade release event conditions and (2) to identify possible challenges for similar ballistic impact simulations. In the ballistic impact tests, the blade-shaped projectile rotates and moves in three dimensions, has prolonged sliding contact with the target plate, and exhibits significant plastic deformation. An accurate simulation must exhibit all of these characteristics. Comparisons between the tests and the simulations showed both promising results and limitations. NASA BLADE IMPACT TESTS NASA conducted four ballistic impact tests on large flat Aluminum 2024 panels with a simulated blade-shape Titanium 6Al-4V projectile to provide experimental data to evaluate the numerical material model (Pereira et al, 2013). These tests were designed to simulate characteristics of a turbine engine blade release event where a released blade tip makes contact with the engine case, skates (slides), bends, and then rotates to cause the heavier blade root to impact and potentially penetrate the engine case. The projectile developed for this test, called the NASA Generic Fan Blade Fragment (NGFBF), was designed to include some of features of a real fan blade, such as a thin tip and heavy root, while being relatively simple to manufacture and model. It was made from Titanium 6Al-4V and had a nominal mass of 340 grams. Its dimensions are shown in Figure 1. The test panels were 24.0 inches by 24.0 inches with a nominal thickness of 0.25 inches and were made from Aluminum 2024-T351. The panels were held at a 45° angle in a square fixture

© ASCE

517

Earth and Space 2022

with a 20.0-inch by 20.0-inch aperture, as shown in Figure 2. The nominal impact obliquity angle defined by the angle between the projectile flight direction and panel normal was 45° in the ballistic impact tests. The panels were through-bolted with 24 0.5-inch bolts equally spaced around the sides, 1.0 inch inside from the edges.

Figure 1. NGFBF projectile (dimensions in inches) (Pereira et al, 2013). The blade projectile was initially positioned with a 45° angle in a gun barrel, which makes an initial 90° angle between the panel and the blade projectile, as shown in Figure 2. However, actual impact angles between the panel and the blade projectile at the impact moment in the tests were much less (45.1° - 61.3°) because the blade projectiles rotated while they were flying from the gun barrel to the impact point on the panel. Figure 3 shows the two coordinate systems: the global fixed coordinate system and the blade local moving coordinate system. The origin of the global coordinate system is fixed at the center of the impact front face of the test panel. The X-axis of the global coordinate system is parallel to the flying direction of the projectile as indicated by the yellow dashed line, and the Z-axis of the global coordinate system points vertically downward. The origin of the blade local coordinate system is located at a point on the left lower root of the blade, 0.5 inch from the lower root edge and 0.5 inch from the left side edge, as shown in Figure 3. The xb -axis and yb -axis of the blade local coordinate system point toward the left tip corner and the right root corner of the blade, parallel to the left side edge and the root edge of the blade, respectively. The blade local coordinate system is moving and rotating with the blade projectile. A total of 4 ballistic impact tests (LG908, LG909, LG910, and LG911) were conducted with the test setup described above. The 4 tests had the same conditions, except for the

© ASCE

518

Earth and Space 2022

intentionally differing initial impact velocities, and the unavoidably differing blade orientations. Tables 1 and 2 summarize the measured linear velocities, angular positions, and angular velocities of the blade projectile at the moment of initial tip impact. Specifically, the linear velocities of the blade projectile were measured at the origin of the local blade coordinate system (Figure 3) in the global coordinate system. The angular orientation and angular velocity are given as a set of Euler angles with respect to the blade local moving coordinate system. The Euler angles are defined as an xb-yb-zb rotation sequence: roll in the xb-axis, pitch in the yb-axis, and yaw in the zb-axis. The desired orientation (attitude angles) of the projectile at impact was (0°, 45°, 0°). However, actual orientations of the projectile were somewhat different due to the difficulty of precisely controlling orientation in the impact tests. This difficulty is due to unavoidable blade rotations. It should be noted that the values in Tables 1 and 2 are different from those in the report (Pereira et al, 2013) because they were re-computed from the test data, and updated here, to correct faults that were found later in the original computation.

Figure 2. Schematic of the test setup Figure 3. Two coordinate systems on the still image (Pereira et al, 2013). from a high-speed movie of the impact test. The containment and perforation results for the panels in each test are identified in Table 1. While in all tests the projectiles were contained and rebounded from the panel, perforation of the panels occurred in the three higher velocity tests (LG909, LG910 and LG911). There was no panel perforation in the lowest velocity test, LG908. The deformed shapes of the post-tested panels are shown in Figure 4. The progression of the blade projectile impact against the panel was essentially similar in all four tests, with the only exception being whether the blade root perforated the panel. In all cases, the tip of the blade projectile made the first contact on the panel, subsequently started bending, and the root rotated, impacting the panel. The root then scraped the panel surface generating deep gouges and ejecting significant wear debris. Since all tests were contained, the impact velocities in these tests were below the ballistic limit. However, there was perforation damage in some plates. In addition, the onset of perforation damage will

© ASCE

519

520

Earth and Space 2022

be referred to as the perforation critical velocity. When the impact velocity of the blade projectile was over this critical velocity, the panel was perforated at the side edge of the wear mark. The perforation propagated to make a wide opening with a large petal. However, in all four tests the blade projectile bounced back and fell away without penetrating the panel thoroughly. It can be seen that the majority of the damage and failure initiation occur when the heavier root section of the blade impacts the panel. This is consistent with what occurs in an actual turbine engine fan blade out incident. Table 1. Measured linear velocities of the blade projectile at impact moment. Test number LG908 LG909 LG910 LG911

X-vel. (ft/s) 712.8 813.2 760.7 723.3

Linear velocity Y-vel. (ft/s) 7.9 11.5 7.8 2.3

Z-vel. (ft/s) -52.2 -10.7 -11.0 -7.4

Comments Contained. Contained, Perforated Contained, Perforated Contained, Perforated

Table 2. Measured angular orientations and angular velocities of the blade projectile at impact moment. Test number LG908 LG909 LG910 LG911

Angular orientation Roll Pitch Yaw (deg.) (deg.) (deg.) -2.0 89.9 -1.5 2.6 73.7 0.1 -0.7 79.1 2.3 8.2 82.8 1.1

Angular velocity Roll vel. Pitch vel. Yaw vel. (deg/s) (deg/s) (deg/s) -626.4 8,255.4 71.8 -2.5 7,664.4 -55.2 193.0 7,202.3 -669.0 343.5 7,888.1 1,007.3

Based on the photos of the post-test panels of LG909, LG910, and LG911 in Figure 4, the shapes of the opening and petals are very similar to each other even though there were some variations of the 3D movements of the blade projectile among tests, as summarized in Tables 1 and 2. Principally, the X-velocity is a dominant factor in this test because that is the flight direction component of the projectile. The X-velocity of the projectile in LG908 was the lowest and the velocity in LG909 was the highest with about a 100 ft/s difference. There was only about a 10 ft/s difference of the X-velocity of the projectile between LG908 and LG911, while the test results showed a distinct difference in the presence of the perforation. This means that under these impact conditions, the perforation critical velocity is in that range. The Z-velocity, pitch angle, and pitch velocity are also influential factors because they contribute to the rate of projectile root rotation onto the panel. Other components, such as Y-velocity, roll and yaw angles, and roll and yaw velocities, could be influential factors. However, their values and their variation ranges are relatively small in the tests. Their effects will be evaluated by comparing the simulations with half-symmetric and full FE models.

© ASCE

521

Earth and Space 2022

(a)

(b)

(c)

(d)

Figure 4. Post-test panels (Pereira et al, 2013): (a) LG908, (b) LG909, (c) LG910, and (d) LG911. BLADE IMPACT SIMULATIONS Finite Element (FE) models were developed in LS-DYNA to simulate the NASA ballistic impact tests. The newly developed *MAT_224 material parameter set for Aluminum 2024 (version 2.2) (Park et al, 2020a) (Park et al, 2020b) (LS-DYNA Aerospace Working Group, 2021) was used to model plastic material failure behavior for the target Aluminum panel, and the *MAT_224 material parameter set for Titanium 6Al-4V (version 1.3) (Seidt, 2014) (LS-DYNA Aerospace Working Group, 2021) was used for the Titanium blade projectile. *MAT_224 is an elasto-viscoplastic material model that allows arbitrarily defined stress versus strain curves to define material plasticity, including arbitrary strain rate and temperature dependency. Adiabatic heating due to plastic work can cause temperatures to increase and the material to soften. Element erosion is included using plastic failure strain as a criterion and can be defined as a function of the state of stress, strain rate, temperature, and element size. This material model resembles the original Johnson-Cook material model (*MAT_015 in LS-DYNA) using similar separation of parameter dependencies, but with the possibility of general tabulated input parameters. The tabulated input parameters allow for a much closer match to mechanical property test data than the Johnson-Cook model, which is limited by curve fitting of the test data. In addition, *MAT_224 allows for parameter dependency of the Taylor-Quinney coefficient and regularization to reduce the mesh dependency of element erosion, the lack of which also limited the original Johnson-Cook model. In a previous study (Park et al, 2020b), it was shown that the mesh size of a FE model using a *MAT_224 material model needs to be within the regularization range defined in that *MAT_224 input parameter set to get accurate failure behavior in a ballistic impact simulation. The allowable element size in the regularization range of *MAT_224, defined by an LCI table for Aluminum

© ASCE

522

Earth and Space 2022

2024 and Titanium 6Al-4V, is smaller than about 0.01 inch, which makes the full FE model required for this study about 82 million solid elements. In addition, a half-symmetry FE model, which has about 41 million solid elements, was also developed for comparison of simulations to study the effects of other components of the blade motion, such as Y-velocity, roll and yaw angles, and roll and yaw velocities. The reduced integration formulation for solid elements was used for all the simulations to help achieve a reasonable runtime. The FE modeling of contact parameters, contact frictions between the panel and blade projectile, and boundary conditions of the panel were studied by using a coarse half-symmetric FE model which has about 5 million solid elements with a 0.02-inch element size. The details of the parametric FE modeling study are described in the FAA report (Park et al, 2021). LG908. In LG908, the X-velocity of the projectile is 712.8 ft/s, which is the lowest of all four tests. The panel deformation in the LG908 test is shown in Figure 4(a). No perforation of the panel occurred by the impact of the blade projectile in LG908. Instead, the hard contact of the bottom root of the blade projectile to the panel surface made a large wear mark on the panel surface. Figure 5(a) and Figure 5(f) show the panel deformations of LG908 simulations in an effective plastic strain contour using the half-symmetric and full FE models, respectively. Similar to the test, no perforation occurred in either simulation. There is no noticeable difference between the two simulations, except for the slightly rotated impact marks of the blade projectile in the simulation using the full FE model due to the roll and yaw rotations of the blade projectile.

(a)

(b)

(c)

(d)

(e)

(f)

(g)

(h)

Figure 5. Panel deformation in simulations at 2 msec: (a) LG908 using the halfsymmetric FE model, (b) LG909 using the half-symmetric FE model, (c) LG910 using the half-symmetric FE model, (d) LG911 using the half-symmetric FE model, (e) LG908 using the full FE model, (f) LG909 using the full FE model, (g) LG910 using the full FE model, and (h) LG911 using the full FE model.

© ASCE

523

Earth and Space 2022

The main difference between the test and simulations is the vertical length of the wear mark created by the blade root. The wear mark in the test is almost two times longer than the one in the simulations. When the blade projectile scraped on the panel surface in the LG908 test, many debris particles were avulsed. In the simulations, however, all the elements at the wear area are eroded and no element debris falls away. It means that an additional portion of the impact energy of the blade projectile was spent eroding these elements, many of which should have become debris. This may be a reason why the wear marks in the simulation are shorter than one in the test. Additionally, necessarily using contact friction coefficients rather than modeling the actual tribochemical friction process (Dong and Bell, 1999) (Mishra, 2014), unavoidably adds uncertainty and inaccuracy to the analysis. The impact process of the blade projectile on the panel was divided into three stages in order to compare their deformation profiles of the central vertical cross-section of the panel between the test and simulations. The first stage is at the time right before the root of the blade projectile impacted the panel, the second stage is at the time right before a crack was initiated in the test, and the third stage is at the time when the panel deformation reached the maximum. The second stage of LG908 in which perforation does not occur is delineated here in order to compare with LG909. The timings of each stage are very closely aligned between the test and simulations. The comparison of the deformation profiles of the central vertical cross-section of the panel between the test and simulation using the full FE model at each stage of LG908 is shown in Figure 6(a). In Stage1 and Stage2, the deformation profiles of the simulation are very close to the one in the test. In Stage3, the deformation profiles between the test and simulation are similar but show some differences in terms of their peak and peak shape. The peak in the simulation with the full FE model is lower than the peak measured in test. The shape of the deformation profile at the peak response in the test is blunt, but the shapes predicted in the simulation is sharp. This difference could be due to the shorter wear area in the simulation. Had the root of the blade projectile scraped the panel surface long enough, it may have resulted in prediction of a blunt peak deformation profile at the central vertical cross-section of the panel.

(a)

© ASCE

524

Earth and Space 2022

(b)

Figure 6. Comparison of deformation profiles of the central vertical cross-section of the panel: (a) LG908 using the full FE model, and (b) LG909 using the full FE model. LG909. In LG909, the X-velocity of the projectile is 813.2 ft/s, which is the highest in all four tests and about 100 ft/s higher than in LG908. The panel deformation in the LG909 test is shown in Figure 4(b). It can be seen that the panel surface was worn initially by the blade root edge and then eventually perforated. During the wear process much debris was ejected. After that, the cracks propagated downward and the large petal was developed with a big opening in the panel. However, the blade projectile did not fully penetrate the panel. Figure 5(b) and Figure 5(f) show the panel deformations of LG909 simulations in an effective plastic strain contour using the half-symmetric and full models, respectively. The wear mark, perforation and petal on the panel were similar to the test. However, the lengths of the wear mark and cracks are shorter than in the test, which is probably due to excessive erosion of solid elements in the wear area, and the necessary use of contact friction coefficients in the simulations rather than a representation of the actual tribochemical friction process (Dong and Bell, 1999) (Mishra, 2014). Comparing the simulations, the wear mark, perforation and petaling are close, except for the left crack propagation direction. The left crack was initially propagating to the left and then turned downward due to the roll and yaw motions of the blade projectile in the simulation with the full FE model. The comparison of the deformation profiles of the central vertical cross-section of the panel between the test and simulation at each stage of LG909 is shown in Figure 6(b). In Stage1 and Stage2, the deformation profiles of the simulation are very close to the one in the test. In Stage3, the deformation curves are discontinuous due to failure in the middle. The deformation profiles between the test and simulation are very close up to the time when crack initiation occurred. Following crack initiation, however, the deformation in the test is much larger than in the simulations. The final deformed shape of the petal is directly related to the length of the crack. Note that the test crack length in Figure 4(b) is noticeably longer than the simulation crack lengths in Figure 5(b) and Figure 5(f). Crack propagation, final crack lengths and petal deformations are inherently highly varied. Crack propagation requires much less energy than the

© ASCE

Earth and Space 2022

initial crack formation. As a result, crack length and petal deformation are dependent on small differences in energy. They also depend on micro-structures such as grain boundaries in the metal. Crack propagation models exist; however, no attempt to incorporate them into this research has been made. Crack propagation models have not been incorporated into *MAT_224. LG910. In LG910, the X-velocity of the projectile is 760.7 ft/s, which is about 50 ft/s lower than the one in LG909 and also about 50 ft/s higher than the one in LG908. The panel deformation in the LG911 test is shown in Figure 4(c). The panel was perforated, but the blade projectile did not penetrate the panel completely. The panel deformation in LG910 is very similar to that in LG909. The petal size and deformation in LG910 are smaller than in LG909 due to lower X-velocity of the blade projectile. Figure 5(c) and Figure 5(g) show the panel deformations of the LG910 simulations in an effective plastic strain contour using the half-symmetric and full models, respectively. Similar to the simulations of LG909, the lengths of the wear mark and cracks are shorter than those in the test. The simulation results are similar, except for the crack propagation direction. In the half-symmetric model simulation, both the left and right cracks initially propagated slightly toward the side and then turned downward, but in the full FE model simulation both cracks propagated directly downward. LG911. In LG911, the X-velocity of the projectile is 723.3 ft/s, which is just about 10 ft/s higher than in LG908. The panel deformation in the LG911 test is shown in Figure 4(d). The panel was perforated, but the blade projectile did not penetrate the panel thoroughly. The panel deformation in LG911 is very similar to the deformations in LG909 and LG910. Figure 5(d) and Figure 5(h) show the panel deformations for the LG911 simulations in an effective plastic strain contour using the half-symmetric and full FE models, respectively. The deformation in the simulation with the half-symmetric FE model shows small perforations at both sides of the wear mark and those perforations did not propagate to make a wide opening and petal on the panel. In the simulations with the full FE model, the perforation initiated at the right side of the wear area and propagated to the left, but not enough to reach to the left side edge of the wear area to make a complete opening on the panel. The deformations of the panel between test and simulations are quite different in LG911. Based on the test results, it was found that the X-velocity of the projectile in LG911 is close to the critical velocity for panel perforation. For that reason, the panel deformation in LG911 could be very sensitive to small variations in the 3D movements of the blade projectile in the test. If the 3D position and velocity of the projectile at the moment of impact are not precisely identified in the post-test calculations or not accurately implemented in the simulation model, then there may be significant differences between the simulation and test. Also, this difference is significantly influenced by crack propagation mechanics in the test, and the lack thereof in the analysis. Nevertheless, the simulations were able to predict the initiation of the perforation in the panel with the measured impact velocity of the projectile. Summary. The half-symmetric and full FE model simulations showed very similar results to the tests in terms of panel deformation and blade behavior, with the exception of test LG911. The initial setup of the simulation models closely replicated the behavior of the blade projectiles in test and caused similar panel deformations by creating wear marks and perforation on the panel. However, the panel crack and petal characteristics in the simulation of LG911 were not fully developed, unlike what was observed in test. It is important to note that the X-velocity of the blade projectile in LG911 is close to the critical velocity for initiation of panel perforation, so the panel deformation could be very sensitive to small variations of the 3D movements of the blade projectile in that test. In addition, the difference could be also caused by the inherent

© ASCE

525

Earth and Space 2022

limitations of the simulation; such as the use of element erosion to represent rupture, the use of friction coefficients to represent tribochemical physics, and the non-inclusion of crack propagation mechanics. The generic difference between the tests and simulations stemmed from the wear process between the panel and blade projectile, such as the presence of expelled wear debris and the size and length of wear mark on the panel. In all simulations, solid elements in the wear area were eroded, while in the test significant quantities of wear debris fell away. It is unclear if the eroded internal energy in the simulation is larger than or equivalent to the absorbed energy to create the wear debris in the test. Physically however, friction coefficients in conjunction with mechanical wear, and tribochemical wear, are different phenomena and this difference affects the simulation results relative to the test. The static and dynamic contact friction coefficients were set to 0.3 in all simulations for both the half-symmetric and full FE models, and they worked well to create the perforation in the panel in LG909, LG910 and LG911. However, if the level of friction is not selected carefully, it could adversely affect the wear mark size, crack propagation, and petal size in the simulations. And as previously presented, a constant fiction coefficient does not represent that physical dependency on rate, pressure, and mesh size. The simulations using the half-symmetric and full FE models were compared against one another to see the effects of roll and yaw motions for the blade projectile. Overall, they gave similar simulation results. However, the roll and yaw motions of the blade projectile affected the crack propagation pattern leading to development of the petal opening in the panel. The effects of projectile attitude angle variation were studied in previous research (Park, et al., 2020) and demonstrated that the attitude angle variation of the projectile could affect the crack propagation pattern sensitivity. Nevertheless, it appears that for this test series, the range of blade projectile variation in roll and yaw motions is small enough that it is not expected to have significant influence on the overall panel deformation and perforation in the simulations of the halfsymmetric and full FE models. CONCLUSION In this research work, four ballistic blade impact tests were simulated using advanced Aluminum 2024 and Titanium 6Al-4V material models utilizing *MAT_224 in LS-DYNA. These simulations were performed to validate these material models under more realistic turbine engine blade release event conditions than previously modeled ballistic impact tests. The half-symmetric and full FE models were developed with 41 million and 82 million solid elements with a 0.01-inch element size, respectively. The initial condition, contact parameters, contact friction, and boundary conditions of the FE models were studied to implement the 3D movements and rotations of the blade projectile accurately in the simulations. The four ballistic impact simulations showed similar results to the tests in terms of the panel deformation and blade behavior, and were able to predict the panel perforation. The initial stages of the simulation model impact closely replicated the behavior of the blade projectile and caused similar panel deformation by creating wear marks and perforation on the panel. However, the crack and petal in the panel in the simulation of LG911, where the X-velocity of the blade projectile is near the critical velocity of the panel perforation, were not fully developed, unlike in the test. It is noteworthy that the ballistic impact simulations using the *MAT_224 metallic material models demonstrate the ability to predict the onset of perforation.

© ASCE

526

Earth and Space 2022

The fundamental difference between the tests and simulations was from the wear process of the panel by the blade projectile, including the presence of the wear debris and the size of the wear mark on the panel. The generic limitation of the simulation, such as constant friction coefficients and mechanical element erosion, which resulted in the absence of wear debris, probably affected the wear mark size, crack propagation, and petal size in the simulations. Comparing the two simulations using the half-symmetric and full FE models, the small variation of the roll and yaw movements of the blade projectile was not very influential to the overall panel deformation and perforation in the simulations. However, these variations affected the crack propagation pattern to develop the petal and opening in the panel. ACKNOWLEDGEMENTS The authors would like to express our gratitude to Dr. Mike Pereira, Mr. Duane Revilock, and Mr. Charles Ruggeri, Aerospace Engineers at NASA Glenn Research Center, for sharing their insight and expertise along with the test data, photos and videos generated during the tests. Their support greatly assisted with this research. The authors are also grateful to Mr. William Emmerling (retired) at FAA William J. Hughes Technical Center for his constant support and advice on this research, and Dr. Gilbert Queitzsch (retired) at FAA for his helpful discussion and feedback for this research. This research was conducted under FAA cooperative agreement 692M151840003 and sponsored by the Aircraft Catastrophic Failure Prevention Program (ACFPP). REFERENCES Buyuk, M. (2014). Development of a New Metal Material Model in LS-DYNA, Part 2: Development of a Tabulated Thermo-Viscoplastic Material Model with Regularized Failure for Dynamic Ductile Failure Prediction of Structures under Impact Loading. Final Report, DOT/FAA/TC-13/25 P2, Federal Aviation Administration, U.S. Department of Transportation. Dong, H., & Bell, T. (1999). “Tribological Behaviour of Alumina Sliding against Ti6Al4V in Unlubricated Contact.” Wear, 874-884. Emmerling, W., Altobelli, D., Carney, K., & Pereira, M. (2014). Development of a New Metal Material Model in LS-DYNA, Part 1: FAA, NASA, and Industry Collaboration Background. Technical Report, DOT/FAA/TC-13/25 P1, Federal Aviation Administration, U.S. Department of Transportation. Haight, S., Wang, L., Du Bois, P., Carney, K., & Kan, C. D. (2016). Development of a Titanium Alloy Ti-6Al-4V Material Model Used in LS-DYNA. Final Report, DOT/FAA/TC-15/23, Federal Aviation Administration, U.S. Department of Transportation. Hammer, J. T. (2014). Plastic Deformation and Ductile Fracture of Ti-6Al-4V under Various Loading Conditions. Technical Thesis, DOT/FAA/TC-TT14/2, Federal Aviation Administration, U.S. Department of Transportation. Kelley, S., & Johnson, G. (2006). Statistical Testing of Aircraft Materials for Transport Airplane Rotor Burst Fragment Shielding. Final Report, DOT/FAA/AR-06/9, Federal Aviation Administration, U.S. Department of Transportation.

© ASCE

527

Earth and Space 2022

LS-DYNA Aerospace Working Group. (2021). Retrieved from https://awg.lstc.com/tikiindex.php?page=Material+Parameter+Sets LSTC. (2017). LS-DYNA Keyword User’s Manual, Volumes I and II, Version R10.0. Livermore, California: Livermore Software Technology Corporation. Mishra, A. (2014). “Analysis of Friction and Wear of Titanium Alloys.” International Journal of Mechanical Engineering and Robotics Research, 3(3), 570-573. Park, C. K., Carney, K., Du Bois, P., Cordasco, D., & Kan, C. D. (2020). Aluminum 2024T351 Input Parameters for *MAT_224 in LS-DYNA. Final Report, DOT/FAA/TC-19/41 P1, Federal Aviation Administration, U.S. Department of Transportation. Park, C. K., Queitzsch, G., Carney, K., Du Bois, P., Kan, C. D., Cordasco, D., & Emmerling, W. (2020). Aluminum 2024-T351 Input Parameters for *MAT_224 in LS-DYNA, Part 3: Ballistic Impact Simulations of an Aluminum 2024 Panel Using *MAT_224 in LS -DYNA Considering Oblique Incidence and Attitude Angles of a Rectangular Projectile. Final Report, DOT/FAA/TC-19/41 P3, Federal Aviation Administration, U.S. Department of Transportation. Park, C. K., Carney, K., Du Bois, P., Kan, C. D., & Cordasco, D. (2021). Aluminum 2024T351 Input Parameters for *MAT_224 in LS-DYNA, Part 4: Ballistic Impact Simulations of a Titanium 6Al-4V Generic Fan Blade Fragment on an Aluminum 2024 Panel Using *MAT_224 in LS-DYNA. Final Report, DOT/FAA/TC-19/41 P4, Federal Aviation Administration, U.S. Department of Transportation. Pereira, M., Revilock, D., Lerch, B., & Ruggeri, C. (2013). Impact Testing of Aluminum 2024 and Titanium 6Al-4V for Material Model Development. Technical Memorandum, NASA/TM-2013-217869, DOT/FAA/TC-12/58, National Aeronautics and Space Administration. Seidt, J. D. (2014). Development of a New Metal Material Model in LS-DYNA, Part 3: Plastic Deformation and Ductile Fracture of 2024 Aluminum under Various Loading Conditions. Final Report, DOT/FAA/TC-13/25 P3, Federal Aviation Administration, U.S. Department of Transportation.

© ASCE

528

529

Earth and Space 2022

Prediction of Fracture Location of Duplex Stainless Steel Welds C. Payares-Asprino1 1

Mechanical Engineering Dept., Norwich Univ., Northfield. Email: [email protected]

ABSTRACT The influence of heat input on the tensile-test fracture location of bead-on-plate welds duplex stainless using an automatic MIG welding machine was investigated. An experiment-based thermal analysis has been performed to obtain the thermal histories, which can be applied to determine the peak temperature for different welding conditions. The temperature distribution during the welding process significantly affects the mechanical and metallurgical properties of a weldment. The changes in microstructure in a weldment are very dependent on the temperature distribution. Based on these facts, this research aims to develop an analytical model for predicting the tensile-test fracture location of the welds varying heat input during the welding process. For this study, a 3×3 matrix with a low, medium, and high level for each of the welding parameters will be applied, giving a total of 27 weld bead plates. Temperature distribution curves have been drawn. Two tensile specimens were manufactured from each welded plate, resulting in 54 tensile specimens for testing. The results show that the fracture location measurements increase until a maximum of HI of 1.267 kJ/mm then decrease. INTRODUCTION For years, the go-to material for aerospace applications was aluminum. However, aerospace manufacturers have started investigating alternatives to aluminum in recent years, one being steel. Stainless steel is an alloy made of steel and chromium. The strength of the stainless steel correlates directly with the amount of chromium contained in the alloy. The higher the chromium volume, the stronger the steel. Although stronger than aluminum, stainless steel is typically relatively heavier, hindering its widespread use in aerospace applications. However, stainless steel parts possess two main advantages over aluminum: 1. It is highly resistant to corrosion. 2. It is stronger and more resistant to wear and tear. 3. Stainless steel parts can handle scratches/impact damage much better than aluminum parts. Other advantages of this material include its superb heat and fire-resistant, aesthetic appearances, and excellent hygiene qualities. It is also easily fabricated, an essential consideration with all the various shaped parts of an aircraft that must be welded, machined, or cut to precise specifications. We see that airplane manufacturers are increasingly relying on stainless steel for specific parts with higher performance requirements. For example, stainless steel is the material used for the fuel tanks of many airplanes since they are exposed to highly corrosive materials and must withstand high temperatures and protection from structural damage. Modern aerospace vehicles are much more likely to be built with stainless steel airframes or fuselages. They are much

© ASCE

Earth and Space 2022

stronger than aluminum, and depending on the grade used, can still offer an excellent strength to-weight ratio. Duplex stainless steel present a combination of good mechanical properties as such as high tensile strength, high fatigue strength, good toughness, and excellent corrosion resistance. Although DSS offer excellent properties some concern exist regarding effects of welding on the microstructure and properties in the heat affected zone (HAZ). The excellent corrosion resistance (e.g. stress corrosion cracking and pitting) result from the almost equal amount of ferrite (a) phase and austenite (g) phase present in this steel. The presences of these two phases combine the attractive properties of austenitic and ferritic steels: Robert (1997) and Chen and Yang (2002). However, the solidification of duplex stainless steel welds does not always produce near equal amounts of ferritic (a) and austenitic (g) phases, as occurs in the parent metal; thus deteriorating the mechanical properties and corrosion resistance of the weld joint. The microstructures developed in the weld’s fusion zone and the heat-affected zone (HAZ) also have a significant influence on the mechanical properties and corrosion resistance of duplex stainless steels, Liou et al (2002) and Miranda et al (2005). For insuring the excellent combination of properties DSS, it is essential to maintain a ferrite-austenite ratio close to 50:50. This phase balance however is upset during welding because of the rapid cooling involved in most thermal cycles. This can result in weld ferrite contents much in excess of 50%. In order to restore the phase balance, weld filler materials are usually over alloyed with 2-4% more Ni than in the base metal, Liljas (1994) and Stephenson (1981). Hsieh et al (2001) reported that the austenite contents lower than 25% are unacceptable for most industrial application. Norsok (2004) standard recommended that a minimum is austenite content of 30% is required to accept the welded joint in piping inspection. Various researchers reported that controlled heat input was the most important factor to maintain the phase balance. Several researchers, Baxter et al (1993) and Lippold and Kotecki (2005), reported that the parameter for controlling the phase balance in weldments is the cooling rate, which dictates the heat input range to be used. Giridharan and Murugan (2009) divulged that heat input has a significant impact on the welds' bead geometry, metallurgical, mechanical, and corrosion resistance properties. Similarly, Karunakaran (2012) published that the rate of heat input during welding followed by the nature of cooling has a strong influence on the grain size and phase formation. Super-duplex steels must be welded with moderate heat inputs depending on thickness and joint geometry to obtain the beneficial outcomes, ASM (1994). It is well inferred from the open literatures that a low heat input and faster cooling rates result in Cr2N precipitation and higher heat inputs lead to the formation of X or σ which were reported as deleterious phases and the occurrence of these phases tend to deteriorate the mechanical properties as investigated by several researchers, Muthupandi et al (2003) and Yousefieh et al (2009). Hence, it is necessary to achieve an optimal heat input to control these phases. It was reported by Ozlati et al (2018) that the strength of resistance weld of martensitic stainless steel to duplex stainless steel rod decreases with welding current. σ phase formation is most rapid between 850ºC and 900ºC, and preferred formation site are austenite/ferrite phase boundaries, Pohl et al (2017). Intermetallic phase is formed if the cooling rate is too slow. Only a small volume fraction of σ phase can reduce impact toughness, Karlsson and Tolling (2006) and corrosion resistance of the weld drastically, Wang (2005). Nishimoto et al (2006), simulated the microstructural evolution during reheating with a particular focus on the sigma phase formation. They produced a map predicting the content of the sigma phase to the number of welds passes and the thermal cycles and indicated that the region of the HAZ heated up repeatedly in the range, 827 to 1027 °C is the most favorable area for sigma phase

© ASCE

530

531

Earth and Space 2022

precipitation. Wang (2005) conducted an experiment gathering data from the HAZ during welding by attaching thermocouples 3 to 6 mm from the weld in a V-grooved pipe. He measured the cooling time between 800 and 500 °C in SDSS weldments at different heat inputs and concluded that the cooling rate plays an essential role for the phase balance and formation of secondary phases. Terasaki and Kitamura (2004) proposed the numerical model to predict the static fracture strength of a laser lap-welded specimen. Also analyses, by using finite element analyses, the ductile fracture initiation and necking phenomena were simulated during tensileshear test of high strength low alloy steel sheets, Asim et al (2012) and Lee et al (2011). Ha and Hu (2013) developed a failure criterion under combined normal and shear loading conditions for laser weld by experimental approaches and numerical analyses. Ma et al(2014) numerically simulated the tensile-shear test of the high strength dual phase 980 steel weldments considering the hardened and the softened zones within the specimens. Kang et al. (2018) developed an analytical model to predict the tensile-shear fracture location without destructive testing in HighStrength Steel Laser Overlap welds. The experimental work aims to optimize welding parameters to ensure that the fracture location of the tensile test specimen occurs in the base metal. EXPERIMENTAL PROCEDURE: Welding Parameter. The material used for this research was a 6 mm thick plate of Duplex Stainless Steel SAF 2205. The chemical composition of the sample was obtained conducting a spectrographic chemical analysis SPECTROLAB 5L and following ASTM A1016 standards. Table 1 shows the chemical composition of SAF 2205 DSS and filler metal ER-2209 (electrode) of 1 mm diameter as recommended by ASTM A815 (2018) and ASTM A789 (2018) procedures GMA welding process. Table 1. Chemical composition of the DSS SAF 2205 plate and ER 2209 filler metal Material %C SAF 2205 0.045 ER 2209 0.015

%Si 0.32 0.54

%Mn 1.41 1.87

%P 0.030 0.023

%S Cr Ni 0.020 22.32 5.31 0.006 23.31 9.81

Mo 3.34 3.77

N 0.08 0.14

The mechanical properties of the base metal are given in Table 2. Table 2. Material Properties of the SAF DSS 2205 [AVESTA 2020] Yield Stress [MPa] 450-620

Tensile Stress [MPa] 620-810

Elongation [%] 15.8

Young’s Modulus [GPa] 200

Poisson’s ratio 0.3

Automatic gas metal arc welding process (GMAW) manufactured the single bead-on-plate welds on DSS under different welding conditions. Bead-on- plate welding was conducted in this research to avoid distortion, obtain suitable geometry, get rid of clamping, and eliminate the

© ASCE

532

Earth and Space 2022

adjustments of gap and mismatch of welds. The recommended values for arc energy (HI) ranged between 0.5 and 2.5 kJ/mm for stainless steel duplex SAF 2205 by AVESTA (2020), and Equation 1 calculated the HI values. The bead-on-plate (BOP) welds were manufactured using the conditions shown in Table 3. Figure 1 shows a schematic drawing of BOP welds and dimensions of the plate. Equation 1 is used for heat input (HI) calculations (IMOA (2009)). HI =

I x E x 60

(1)

v x 1000

Where HI (

kj ) , I: Arc Current (A), E: Arc Voltage (V), v: welding speed (mm/min). mm

Figure 1. Schematic drawing of the Bead on Plate (BOP) welds. As a research methodology the Factorial Design Experiment was used. This design allows the researchers to observe how multiple factors affect a dependent variable, both independently and together. For this study a 33 matrix with a low, medium and high level for each of the welding parameters was applied. The levels for each welding variable can be observed in Table 3. Results in factorial design involved 27 welding conditions as shown in Table 3. Table 3. Welding Conditions for each sample Sample 1 2 3 4 5 6 7 8 9

© ASCE

HI (kJ/mm) 0.501 0.588 0.676 0.539 0.633 0.727 0.576 0.677 0.778

E (volt) 26.5 26.5 26.5 28.5 28.5 28.5 30.5 30.5 30.5

I(A) 200 235 270 200 235 270 200 235 270

v (mm/min) 635 635 635 635 635 635 635 635 635

533

Earth and Space 2022

10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27

0.684 0.804 0.923 0.735 0.864 0.993 0.787 0.925 1.063 1.078 1.267 1.455 1.159 1.362 1.565 1.241 1.458 1.675

26.5 26.5 26.5 28.5 28.5 28.5 30.5 30.5 30.5 26.5 26.5 26.5 28.5 28.5 28.5 30.5 30.5 30.5

200 235 270 200 235 270 200 235 270 200 235 270 200 235 270 200 235 270

465 465 465 465 465 465 465 465 465 295 295 295 295 295 295 295 295 295

Figure 2 shows the tensile specimen mechanized after welding according to ASTM A370 code (2019). A mechanical testing machine brand MTS 810 model 976.04-14 with a 10 mm/min test speed was used. Two tensile specimens from the base metal and two for each welding condition were prepared following ASTM E 8/E 8M-08 (2018) (Figure 3), resulting in 54 tensile samples tested.

Figure 2. Specification of Mechanized Tensile Test Specimen. Peak Temperature Measurement in Heat Affected Zone Measurements of temperature in the HAZ were used to determine the magnitude and location of peak temperature within the heat-affected zone. Four K-type and R-type thermocouples were attached to drilled holes at different distances from the weld line. Type K is available in the -260°C to 1370°C range, and Type R is available in the -50°C and 1760 °C range. The range of the thermocouple used for the experimentation was from 1°C to 900°C. Eight thermocouples were connected to a data logger to collect the time-temperature data during welding. The Picolog Software recorded all temperature-time data continuously.

© ASCE

534

Earth and Space 2022

Preparation of Test Specimen for temperatures measurement

Axis X Weld Bead

Axis Y

The thermocouples were placed using a Cartesian coordinate system (X, Y) represented by the length and width respectively of the plate (Figure 3).This arrangement of the thermocouple was made to provide the information about the temperature’s ranges.

Figure 3 Arrangement of coordinate plane (X, Y) for measurements of temperature. Heat flow during welding will predominantly be in Y direction. Therefore, thermocouples were placed at three distances away from the weld bead (in the Y direction) to determine the thermal gradient and extension of the HAZ. To determine the thermal gradient in the weld bead (X-axis), eight (8) thermocouples were placed in each hole and connected to the eight entrance channels of the equipment (Figure 4). The thermocouples connected to the measurement equipment and the computer gave the thermal data. Table 4 shows the specifications of the thermocouples and their position from the fusion line.

Figure 4 Position of the channel, which were assigned to each thermocouple. Measurements of the Distances between Thermocouple and Welded Centre Line It was necessary to determine the distance between the peak temperature location and the center of the weld bead to evaluate the influence of the welding parameters on the HAZ. The distance between each thermocouple hole and the weld bead center line was measured by optical microscopy, as shown in Figure 5.

© ASCE

535

Earth and Space 2022

Fusion line

Figure 5. Measured distance between thermocouple hole and Weld Bead Centre Line Table 4. Specification and position of the thermocouple during welding. Thermocouple Type K K R R K R K R

Channel 1 2 3 4 5 6 7 8

Thermocouple diameter (mm) 1.0 3.2 3.2 3.2 1.0 3.2 3.2 3.2

Distance from Fusion line (mm) 12 5 5 7 12 5 7 5

RESULTS AND DISCUSSION Thermal Cycle During welding, the thermocouples measured the temperature at different points as a function of time. These readings of temperature are helpful to draw temperatures histories. Thermal histories play an essential role in finding the peak temperatures in the HAZ for different ranges. Figure 6 shows the history of the weldment, welded at low (sample 4), medium (sample 19), and high (sample 26 and sample 27) heat input. The peak temperature in the HAZ increases with increasing of heat input. The maximum peak temperature in the HAZ was ~ 800°C for the weld condition using a higher heat input of 1.675 kJ/mm. The cooling rate is slower with the higher heat input and vice-versa, Kumar et al (2014). 330 297 264 231 198 165 132 99 66 33 0

Channel 1 Channel 2 Channel 3 Channel 4 Channel 5 Channel 6 Channel 7 Channel 8

Temperature(ºC)

Sample 4

0

100

200

a)

© ASCE

300

Time (sec.)

400

500

536

Earth and Space 2022

Sample 19 Channel 1 Channel 2 Channel 3 Channel 4

500 450 Temperature (ºC)

400 350 300 250 200 150 100 50 0 0

100

200

300

400

Time(sec.)

500

b) Sample 26 680 612

Channel 1 Channel 2

544

Channel 3

476

Temperature[ºC]

Channel 4 Channel 5

408

Channel 6 Channel 7

340

Channel 8

272 204 136 68 0 0

100

200

300

400

500

Time (sec.)

c) Sample 27

Channel 1 Channel 2 Channel 3 Channel 4 Channel 5 Channel 6 Channel 7 Channel 8

800

Temperature(ºC)

700 600 500 400 300 200 100 0 0

100

200

300

400

500

600

Time (sec.)

d) Figure 6 Temperature vs. Time of DSS SAF 2205 weld sample a) HI = 0.539 kJ/mm; b) HI= 1.09 kJ/mm; c) HI = 1.455 kJ/mm; d) HI= 1.675 kJ/mm.

© ASCE

537

Earth and Space 2022

Table 5 presents the HAZ peaks' temperature reach for the samples welded at different heat inputs. Peak temperatures in the HAZ increase when HI also increases, giving a maximum peak temperature of 799.5 (°C) using a HI of 1.675 kJ/mm. Table 5. Peak temperature for each distance from fusion line Sample

Distance (mm)

Peak temperature in the HAZ (°C)

4 19 26 27

7 5 7 5

324 494.5 679.3 799.5

Effect of Heat Input on Fracture Weldments Figure 7 shows the distance measurement between the fracture location (Xfr) and the center of the fusion zone of the tensile test specimen.

Sample 4

Figure 7 Measurement of distance between the tensile test fracture location [Xfr] and the center of the fusion zone. Figure 8 shows the relationship between Xfr and HI for all welding conditions (Table 3). It can be observed in this figure that the fracture location of the tensile test specimens tends to increase to a maximum of 31 mm when raising the HI to 1.271 kJ/mm and then it starts to decrease at higher heat input values. A polynomial regression method gave the correlation which predicts the dependence variable Xfr to HI with a coefficient of determination (R 2) of 90%. Equation 2 presented the best correlation between Xf and HI. 𝑋𝑓𝑟 = −11.232 𝐻𝐼 2 + 30.16 𝐻𝐼 + 10.441

(2)

Figure 9. shows the fracture location for duplex stainless-steel welds at different heat inputs, and the HAZ in the welds increases when the HI input rises. In samples 4 and 19, welded with HI ≤ 1.267 kJ/mm failed in the BM, while samples 26 and 27 using a HI≥1.267 kJ/mm, failed in the HAZ. This reduction of the Xfr at higher values of heat input (≥1.267 kJ/mm) might be due to the decreases of the ferrite content on the HAZ reported by Kumar, et al (2014) and Ozlati, and Movahedi,(2018)) and the formation of the intermetallic σ phase present in the HAZ as a consequence of higher peak temperatures and a longer cooling time, Hosseini et al (2016). σ phase is a nonmagnetic intermetallic phase rich in iron, chromium, and sometimes molybdenum, presenting a complex tetragonal crystalline structure. Its presence affects the weldability of duplex stainless steels, Pohl et al (2017).

© ASCE

538

Earth and Space 2022

34 32

Xfr (mm)

30 28 26 24 22 20 0.500

0.670

0.840

1.010

1.180

1.350

1.520

1.690

HI (kJ/mm)

Figure 8. Fracture location [Xfr] vs. Heat Input (HI) for all welding condition.

Figure 9. The fracture location of welds: a visual relationship between the heat input, the distance of the fracture location, and the heat-affected zone (HAZ) for samples 4, 19, 26, 27 welded at HI of 0.539 kJ/mm, 1.09 kJ/mm, 1.455 kJ/mm, 1.675 kJ/mm respectively. © ASCE

Earth and Space 2022

CONCLUSIONS -

-

The peak temperature increased in the HAZ with increasing heat input on DSS AUTOMATIC GMAW welds. HAZ in the welds increases when the peak temperatures rise. The fracture location was in the base metal for lower and medium HI of 0.539 kJ/mm and 1.09 kJ/mm, respectively. The peak temperatures in the HAZ were ≤ than 500 °C. The fracture location was in the HAZ for the higher HI of 1.455 kJ/mm, since the peak temperatures reached in this zone were higher than 699 °C during the welding process. The fracture location measurements increase until a maximum of HI of 1.267 kJ/mm then decrease. This research established the relation between the welding parameters and the fracture location of the joints and found the weakest site of the Duplex Stainless steel of the welds. Fracture Locations of the joint are significantly affected by the welding process variables.

ACKNOWLEDGMENT: The author would like to thank the Norwich University Faculty Development Funding for the Charles A. Dana Research Fellowship AY19-20 and the resourceful contribution of the Kreitzberg Library. REFERENCES ASM Speciality Handbook on Stainless Steels (1994). American Society for Metals; Ohio. Asim, K., Lee, J. and Pan, J. (2012). “Failure mode of laser welds in lap-shear specimens of high strength low alloy (HSLA) steel sheets”. Fatigue Fract. Eng. Mater. Struct., 35, 219–237. ASTM A789/A789M-18 (2018). Standard Specification for Seamless and Welded Ferritic/Austenitic Stainless Steel Tubing for General Service. ASTM A815/A815M-18 (2018). Standard Stainless Steel Specification for Wrought Ferritic Ferritic/Austenitic, and Martensitic Stainless Steel Piping Fittings. ASTM A370 (2019). Standard Test Methods and Definitions for Mechanical Testing of Steel Products, Edition. ASTM E8/E8M (2018). Standard Methods for Tension Testing of Metallic Materials. AVESTA 2205. Basic.Böhler welding by Voestalpine, 2020: https://www.voestalpine.com. Baxter, C.F.G, Irwin J., Francis, R. (1993). “Proceedings of Int. Offshore and Polar Eng. Conf.” Conf. on Duplex Stainless Steel on Glasgow, 2, 401-407. Chen, T.H., and Yang J.R. (2002). “Microstructural characterization of simulated heat affected zone in a nitrogen-containing 2205 duplex stainless steel.” Mater. Sci. Eng. A 338 166-181. Giridharan, P.K., and Murugan, N. (2009). “Optimization of pulsed GTA welding process affected zone in a nitrogen-containing 2205 duplex stainless steel.” Mater. Sci. Eng. A 338 166-181. Ha, J., and Huh, H. (2013). “Failure characterization of laser welds under combined loading conditions”. Int. J. Mech. Sci, 69 40–58.

© ASCE

539

Earth and Space 2022

Hsieh, R.I., Liou, H.Y, and Pan,Y.T. (2001) “Effects of cooling time and alloying elements on the microstructure of the gleeble-simulated heat-affected zone of 22% Cr duplex stainless steels”. J. Mater. Sci. Perfor., 10 526. Hosseini, V. A., Valiente Bermejo, M.A., Gårdstam, J., Hurtig, K., Karlsson, L. (2016). Influence of Multiple Thermal Cycles on Microstructure of Heat-Affected Zone in TIGWelded Super Duplex Stainless Steel. Welding in the World, Le Soudage Dans Le Monde 60(2) 233-214). DOI: 10.1007/s40194-016-0300-5. International Molybdenum Association (IMOA) (2009). Practical Guidelines for the Fabrication of Duplex Stainless Steel, Pergamon Press, London. Kang, M., Jeon, I-H, Nam Han, H. and Kim, C. (2018). “Tensile–Shear Fracture Behavior Prediction of High-Strength Steel Laser Overlap Welds”. Metals, 8, 365-380. Kumar R., Arya H.K. and Saxena, R.K. (2014). “Experimental Determination of Cooling Rate and Its Effect on Microhardness in Submerged Arc Welding of Mild Steel Plate (Grade c-25 as per IS 1570)”. J. Mater. Sci. and Eng. 3(2). DOI: 10.4172/21690022.1000138 N. Karunakaran (2012). Effect of Pulsed Current on Temperature Distribution, Weld Bead Profiles and Characteristics of GTA Welded Stainless Steel Joints, International Journal of Engineering and Technology 2(12) 1908-1916. Karlsson, L., and Tolling, J. (2006). “Experiences and New Possibilities in Welding Duplex Stainless Steels”. Proceedings of IIW Regional Congress on Welding and Related Inspection Technologies, South Africa. Lee, J., Asim, K., Pan, J. (2011). “Modeling of failure mode of laser welds in lap-shear specimens of HSLA steel sheets”. Eng. Fract. Mech. 78 374–396. Liljas, M. (1994). Proceedings of the Fourth International Conference of Duplex Stainless Steels, Glasgow, Scotland, 7(2) 113-116. Liou, H.Y., Hsieh, R.I., Tsai, W.T. (2002). “Microstructure and stress corrosion cracking in simulated heat-affected zones of duplex stainless steels”. Corro. Sci. 44 2841-2856. Lippold, J.C., Kotecki, D.J. (2005). “Welding Metallurgy and Weldability of Stainless Steel”. New York: Willer Inder Science Publication. Ma, J., Kong, F., Liu, W., Carlson, B., and Kovacevic, R. (2014). “Study on the strength and failure modes of laser welded galvanized DP980 steel lap joints”. J. Mater. Process. Technol. 214 1696–1709. Miranda, M.A., Sasaki, J.M., and Tavares, S.S. (2005). “The use of X-ray diffraction, microscopy, and magnetic measurements for analyzing microstructural features of a duplex stainless steel”. Mater. Mater. Character, 54, 87-393. Muthupandi, V., P. Bala, Srinivasan, Seskadri, and Sundaresan, S. (2003). “Effect of weld metal chemistry and heat input on the structure and properties of duplex stainless steel welds”. Mat. Sci. Engineering A 358 9-16. Norsok Standard M601-94. (2004). Welding and Inspection Piping. Lysaker Norway: Standard Norway. Nishimoto K., Saida K., and Katsuyama O. (2006). “Prediction of Sigma Phase Precipitation in Super Duplex Stainless Steel Weldments”. Weld World 50(3-4), 13–28. Ozlati, A., and Movahedi, M. (2018). “Effect of welding heat-input on tensile strength and fracture location in upset resistance weld of martensitic stainless steel to duplex stainless steel rods”. Journal of Manuf. Processes 35 517-525.

© ASCE

540

Earth and Space 2022

Pohl, M., Storz, O., Glogowski, T. (2017). Effect of Intermetallic Precipitations on the Properties of Duplex Stainless Steel. Materials Characterization 58(1), 65-71. DOI:10.4172/2169-0022.1000138. Robert, N.G. (1997). Duplex Stainless Steels Microstructure, Properties and Applications, Abington Publishing, Cambridge, 2841-2856. Stephenson, N. (1981). Welding status of duplex stainless steels for offshore application– Part I. Welding and Metal Fabrication, 5 159-164. Terasaki, T., and Kitamura, T. (2004). “Prediction of static fracture strength of laser-welded lap joints by numerical analysis”. Weld. Int. 18 524–530. Yousefieh, M., Shamanian, M., Saatchi, A. (2011). “Influence of Heat Input in Pulsed Current GTAW Process on Microstructure and Corrosion Resistance of Duplex Stainless Steel”. Welds Journal Iron Steel Rese Int. 18(9) 65-69. Wang, H.-S. (2005). “Effect of welding variables on cooling rate and pitting corrosion resistance in super duplex stainless weldments”. Mater Trans. 46(3) 593–601.

© ASCE

541

542

Earth and Space 2022

Study of Aircraft Structural Response and Occupant Loading during a Water Ditching Event Utilizing LS-DYNA Simulation Jacob B. Putnam1 and Karen E. Jackson2 1 2

NASA Langley Research Center, Hampton, VA. Email: [email protected] National Institute of Aerospace, Hampton, VA. Email: [email protected]

ABSTRACT In this study, the capability of finite element model (FEM) analysis to assess aircraft structural response and occupant injury risk during aircraft water ditching conditions was evaluated. A FEM of the Fokker F28 aircraft, previously validated against full-scale crash test data, was simulated in a realistic water impact environment utilizing recorded conditions from the ditching of US Airways Flight 1549 into the Hudson River. The FEM of the F28 aircraft, was developed from a NASTRAN loads model provided by the vehicle manufacturer, calibrated using component fuselage drop-tests, and validated through test-analysis comparisons from a full-scale crash test of the F28 onto soil. Simulation results are presented in several categories: vehicle kinematics, structural accelerations, damage, occupant responses, and injury risk predictions. Results were consistent with the general outcomes of Flight 1549, indicating that the developed vehicle FEM, validated against land impact test data, may provide an effective tool for evaluating water ditching events. INTRODUCTION On January 15, 2009, a US Airways Airbus A320 transport aircraft crashed under controlled impact conditions into the Hudson River in New York City after hitting a flock of birds, which caused failure of its two engines. The plane carried 150 passengers and 5 crew members, all of whom survived the impact, egressed the aircraft, and were rescued by river ferries and the Coast Guard (Figure 1). Flight 1549 took off from LaGuardia Airport and was on its way to Charlotte, North Carolina when the bird strike occurred, shortly after take-off. One news report of the incident stated that “…successful emergency landing in water is among the rarest and most dangerous feats in commercial flying” (Credeur 2009).

Figure 1. Post-impact photographs of Flight 1549 depicting passenger egress.

© ASCE

Earth and Space 2022

The Flight 1549 event provides a unique case study with which to examine a survivable ditching into water condition utilizing simulation. Successful ditching into water is rare, in a study of water impacts events the FAA found less than 10% of water impacts between 19591975 could be considered ditching events (Patel 1996). In addition, a detailed post-crash investigation was performed by the National Transportation Safety Board (NTSB) which provides recorded impact conditions, information on occupant outcomes, and a documented assessment of structural damage to the airframe (NTSB 2009). This data allows the ditching impact condition to be replicated through simulation and for generalized comparisons to be made in terms of airframe and occupant response predictions. Though the outcomes of the Flight 1549 crash can be used to make general conclusions on the adequacy of the simulated vehicle, the goal of this study was not to precisely predict the response of the Airbus A320 aircraft from Flight 1549, but to evaluate tools for simulating and analyzing water impacts utilizing a realistic aircraft ditching event. Evaluating the crashworthiness of a full aircraft structure, through test, is a time intensive and costly process. Evaluating the crashworthiness of these structures in a water-based impact becomes even more difficult. Finite element models (FEMs) provide an alternative to testing, allowing evaluation of impact environments and conditions not feasible through test. Though they have significant potential to aid in crashworthiness evaluation, FEM accuracy must be fully characterized before it is used to quantify the crashworthiness of an aircraft. Prior to this study, a FEM of a Fokker F28 aircraft was developed, calibrated, and validated against component and full-scale land-based crash test data generated under a joint research program between the National Aeronautics and Space Administration (NASA) and the Federal Aviation Administration (FAA) (Littell 2018, Jackson 2018, Littell 2020, Jackson 2020). The current study utilizes this aircraft FEM, simulated in the recorded flight 1549 impact conditions, to assess the capability of vehicle models developed and validated using land-based crash test data to be utilized in water impact analysis. Characterizing the effectiveness of a vehicle FEM, validated in land landing conditions, used in water impact analysis will help define capabilities and limitations of certification by analysis methodologies for the aerospace crashworthiness field. METHODS – WATER IMPACT MODEL To simulate the previously developed FEM of a Fokker F28 aircraft in the Flight 1549 ditching conditions an Arbitrary Lagrangian-Eulerian (ALE) model was developed to simulate the water environment (Figure 2). The impact environment was modeled as a solid Eulerian mesh sized 80 ft. x 220 ft. x 20 ft. (Length x Width x Height). The top ¼ of this mesh was utilized to represent a void space above the water surface to allow wave formation and fluid movement. The water environment mesh size along the primary path of impact was 3 in. x 3 in. Mesh size was progressively increased moving towards the outer edges of the water block to optimize computation cost. The water properties, material coupling, and contact parameters utilized in this model were previously developed by NASA and verified against test data for prediction of aircraft and spacecraft structural responses during dynamic water impacts (Fasanella 2005, Jackson 2008, Vassilakos 2017). The F28 aircraft model consisted of 745 parts, 75 material cards, 1,299,220 nodes, 26,026 beam elements, 81,093 shell elements, 19,682 solid elements, 34 element masses, and, 76 Constrained Nodal Rigid Bodies (CNRBs). Together, the water and void blocks added an additional 1,808,882 solid elements.

© ASCE

543

544

Earth and Space 2022

(a) Side view.

(b) Front view.

Figure 2. Side and front views of the F28 model and water block. To replicate the ditching event the aircraft model was assigned impact conditions matching those found in the Flight 1549 NTSB report (NTSB 2009). Impact velocity was: Vx (horizontal) = 211 ft/s. and Vz (vertical) = 12.5 ft/s. The impact attitude was: Roll = 0.0°, Pitch (nose-up) = 9.5°, and Yaw = 0.0°. These impact conditions are within the flight performance specifications of the Fokker F28 aircraft which has a landing speed of 210 ft/s. The model was simulated using LS-DYNA Symmetric Multi-Processing (SMP) version R11.1.0 double precision on 8 processors of a Linux-based workstation computer and required 1,902 hours of clock time to reach normal termination of 1.0 seconds. One feature of the simulation that requires additional explanation is the long runtime. The high computational cost of this simulation is attributed to the rate at which the ALE water mesh was re-mapped to the impacted environment, referred to as advection. The original the model had advection set at every 5 time step cycles. Though this had a reduced clock time of 424 hours it resulted in significant mesh distortion. The LS-DYNA User’s Manual recommends that the advection cycle be set to 1.0, meaning that the mesh is advected every time step (Hallquist 2006). To evaluate the tradeoff between computation time and mesh stability a sensitivity study was performed varying advection between 1 and 5 time step cycles. Although performing advection at each time step was computationally expensive, it was selected for the reported results as it was the only value to remove simulation instabilities. The kinematics of the vehicle model as it traveled through the water and structural damage predictions were studied to understand the capability of the previously developed vehicle FEM to be used in water impact simulation. Vehicle kinematics were evaluated both in terms of general motion of the vehicle through the water as well as acceleration time histories recorded at several seat base and frame locations throughout the aircraft (Figure 3). All acceleration time histories were recorded in local coordinate systems defined at each accelerometer location and filtered using a low-pass 50 Hz 4-pole Butterworth filter. Water kinematics were also evaluated to identify the stability of the water-vehicle coupling and contacts defined within the model. Predicted damage to the F28 model was compared to fuselage damage on the Flight 1549 aircraft recorded through post-recovery photographs. Although a different aircraft, the results were

© ASCE

545

Earth and Space 2022

compared to provide a general assessment of the vehicle model’s ability to predict damage consistent with that seen in a traditional aluminum bodied transport category aircraft during a water ditching event.

Figure 3. Schematic of the F28 aircraft accelerometer locations. METHODS - OCCUPANT FEM DEVELOPMENT To evaluate occupant responses for the Fokker F28 aircraft simulated under the Flight 1549 ditching event conditions, occupant models were developed for each of the seating locations previously evaluated in the F28 land landing test and simulation (Putnam 2020). Occupants were represented by a variety of Anthropomorphic Test Device (ATD) FEMs: (1) Humanetics® Hybrid III FAA 50th percentile version 1.2.3 (H3 FAA 50th) (Humanetics 2018) (2) Humanetics® Hybrid III 5th percentile version 7.0.5 (H3 5th) (Humanetics 2013) (3) LSTC® Hybrid III 95th percentile beta version 3.03 (H3 95th) (Guha 2015) (4) Test device for Human Occupant Restraint (THOR) version 2.1 publicly released by the National Highway Traffic Safety Administration (NHTSA) with modifications made by NASA to improve accuracy under vertical loading (Panzer 2015, Putnam 2014, Putnam 2015). The distribution of simulated occupants throughout the aircraft is provided in Figure 4.

Figure 4. Schematic of position and configuration of ATDs simulated. For each seating group, an individual occupant breakout model was created. The occupant breakout model was made up of the seated ATD model, a representative belt model, and the forward row of seats with which the ATD may interact during the impact event. Nine occupant breakout models were simulated, made up of five Starboard and four Port side seat rows. Nine

© ASCE

Earth and Space 2022

FAA H3 50th, one H3 5th, one H3 95th, and one THOR ATD model were simulated. This arrangement included one FAA H3 50th in the braced position in seat 6D. Two occupant breakout models did not include a forward seat, Row 1 Starboard and Row 9 Port. The development of these occupant breakout models, including seat and belt model generation as well as ATD positioning is described in full detail in Putnam (2020). To simulate each occupant breakout model in the crash event, the linear acceleration and rotational velocity predicted at the seat base accelerometer locations that were generated by the F28 ditching simulation were applied to the occupant breakout models at nodes matching the accelerometer locations. This loading condition was applied in a local coordinate system, matching how it was measured in the full-vehicle simulation. The occupant breakout simulations were performed for 0.3 seconds, which encompassed peak acceleration input from the vehicle impact, as well as completion of the contact event with the forward seatback in applicable seating configurations. All simulations were performed on a Linux computer cluster using LS-DYNA SMP Version R10.1.0 double precision with four processors. Simulation runtimes ranged from 26 to 106 hours depending on the quantity and variant of ATDs in the occupant breakout model. Occupant responses in the F28 ditching simulation were quantified through four different injury metrics: Head Injury Criteria (HIC-15), HIC-36, Neck Injury Criteria (Nij), and lumbar load. The HIC and lumbar load injury metrics are currently defined by the FAA for certification of occupant safety under dynamic loading conditions (FAA 1988). Nij requirements are not defined by the FAA but are instead established as a requirement for automotive occupant safety defined by NHTSA (Kleinberger 1998). Although not directly defined for aircraft certification, the Nij metric provides applicable assessment of neck injury that is not quantified by the HIC and lumbar load metrics. Each metric provides insight into occupant risk for different injury mechanisms potentially induced during a crash event. HIC quantifies head injury risk, namely skull fracture risk, and is primarily driven by any contact between the occupant head and the surrounding environment. Nij quantifies injury risk to the cervical spine due to flexion-extension and tension-compression of the neck during impact. This risk is induced through both inertial loading of the head-neck as well as contact loading of the head into the neck. The lumbar load criteria quantify injury risk to the lumbar spine, namely vertebral fracture, due to spinal compression. This risk is primarily induced through vertical acceleration experienced during impact being transferred through the seat and into the pelvis of the occupant causing compression of the spine into the upper body mass. HIC and lumbar load injury metrics were calculated for all simulated ATDs. Appropriate responses were compared to the limits defined for the FAA Hybrid III 50th ATD as FAA certification requirements for these metrics currently exist only for this ATD configuration. The Nij metric calculation was performed using formulation specific to each ATD configuration according to the certification standards defined by NHTSA (Klenberger 1998, NHTSA 2015). RESULTS - VEHICLE SIMULATION PREDICTIONS Evaluation of the F28 water ditching impact simulation showed realistic water impact kinematics, based on similarities to those witnessed in flight 1549. This impact sequence shown in Figure 5, demonstrates the general motion of the aircraft as it impacts into the water as well as the response of the water environment. The tail of the vehicle impacts the water surface first due to the nose-up pitch attitude. Most of the vertical impact energy is taken out by this initial impact resulting in material failure to the fuselage skin at this section of the aircraft. The vehicle

© ASCE

546

547

Earth and Space 2022

then rotates forward about the impact point, reaching a 0-degree pitch at approximately 0.4 second’s post-impact. After this time, the vehicle continues to rotate forward causing the nose of the aircraft to “dig” into the water, as the tail of the aircraft rises. The fuselage skin in the front section of the aircraft was maintained during this secondary impact. During the impact simulation the contact between water and vehicle model was maintained. No water intrusion into the aircraft cabin was predicted and realistic flow into the damaged aft section was observed. Water intrusion was quantified through visual evaluation of the aircraft interior during simulation.

Figure 5. F28 water impact simulation sequence. Acceleration time histories predicted within the aircraft along the horizontal direction at the seat base locations of Row 2 (forward cabin), Row 6 (mid-cabin), and Row 12 (aft cabin) are plotted in Figure 6. The horizontal acceleration time histories closely match between the three locations measured along the length of the aircraft cabin. Acceleration time histories between the port and starboard side of the aircraft are very similar, this was expected with the 0° yaw/roll condition simulated. There are two distinct peaks in the horizontal acceleration direction, one occurring at approximately 0.2 seconds and one occurring at 0.6 seconds. This acceleration shape lines up with the kinematics observed in the simulation impact sequence. The first peak occurs with the initial impact of the aft section of the aircraft and the water. As the vehicle rotates about the impact point the water-surface contact area is decreased until the nose of the aircraft plows into the water resulting in the second increase in horizontal acceleration. Horizontal acceleration magnitude is minimal, at or below 10 g, throughout the impact event.

© ASCE

Earth and Space 2022

Figure 6. Horizontal acceleration responses of the seat bases of Rows 2, 6, and 12: Starboard (left) and Port (right). Acceleration time histories predicted within the aircraft along the vertical direction at the seat base locations of Row 2 (forward cabin), Row 6 (mid-cabin), and Row 12 (aft cabin) are plotted in Figure 7. Vertical acceleration time histories were disordered throughout the impact event. Energy in the vertical direction was minimal due to the low vertical descent velocity at impact, because of this there is not a clearly defined vertical acceleration pulse into the body of the aircraft. All shown data was filtered at 50 Hz as described in the methods section. Due to the large oscillations in the vertical direction, vertical accelerations were also evaluated using a 20 Hz filter but showed little change in overall response. Although the data is highly oscillatory two distinct events can be observed in the results. As the aircraft initially impacts the water vertical acceleration is highest in the aft section of the aircraft. The pitch up angle of the aircraft at impact results in increased vertical acceleration in the local coordinate system of the aft seats while the forward seats see reduced acceleration due to the downward rotation of the vehicle during initial impact. At 0.4 seconds, the forward section of the aircraft impacts the water resulting in the large spike in Row 2 seat vertical acceleration. These trends are consistent between the starboard and portside of the vehicle. During the initial impact the Row 12 seat base acceleration spikes to 15 g but throughout most of the impact event the seat base accelerations oscillate around 5 g in the vertical direction.

Figure 7. Vertical acceleration responses of the seat bases of Rows 2, 6, and 12: Starboard (left) and Port (right).

© ASCE

548

549

Earth and Space 2022

A comparison of airframe damage between the F28 model that was used to simulate the Flight 1549 ditching conditions and the recorded damage on the A320 aircraft from that flight is shown in Figure 8. The F28 model predicted significant failure to fuselage skin from the wingbox to the tail section. The majority of this damage occurred on initial impact in which this section took the primary load due to the vehicle pitch. Similar failure of the fuselage skin along the vehicle bottom and crushing along the lower sides of the aircraft were observed on the A320 vehicle. The similarity in vehicle damage gives confidence in the initial conditions applied, the water-vehicle coupling defined, and the material models utilized in the ditching simulation.

Figure 8. Side view comparisons of the F28 model and actual A320 damage. RESULTS - OCCUPANT INJURY RISK PREDICTIONS Simulation of the occupant breakout models predicted injury risk values well below the defined safety limits for dynamic loading certification (Table 1). These results indicate that the simulation predicts low injury risk in the ditching event. This finding is consistent with the Flight 1549 ditching event. The NTSB report of the ditching event indicates there were five serious injuries that potentially occurred during the impact. None of these injuries were reported to be to the head, neck, or spine of the occupants (NTSB 2009). Overall, the occupant models predicted the highest risk of injury in the cervical spine region, with Nij values within 45% of the defined limit. Skull injury risk was predicted to be very low across the evaluated occupant positions with the highest HIC value at 11% of the certification limit. Table 1. Computed Injury Metric Responses Injury Metric Response

© ASCE

Seat: ATD

HIC 15 (1000)

HIC 36 (1000)

Nij (1)

Lumbar Load (1500)

Seat 1A: FAA H3 50th Seat 3A: FAA H3 50th Seat 3D: FAA H3 50th Seat 3E: THOR Seat 5A: H3 95th Seat 5B: H3 5th Seat 6A: FAA H3 50th Seat 6B: FAA H3 50th*

7.7 104.2 107.7 5.5 48.8 15.4 105.4 2.5

15.8 104.2 107.7 10.7 56.2 23.1 105.4 3.5

0.14 0.32 0.3 0.15 0.38 0.4 0.35 0.22

398 122 128 360 381 174 236 50

550

Earth and Space 2022

Seat 6D: FAA H3 50th Seat 9D: FAA H3 50th Seat 10C: FAA H3 50th Seat 12A: FAA H3 50th

72.9 5.9 29.2 68.3

72.9 3.2 31.1 68.3

0.45 0.07 0.15 0.23

210 311 150 250

*Braced position

The majority of occupant breakout models were simulated in the front section of the aircraft. In this section of the aircraft, Nij was the dominant injury driver in most of the simulated occupants, with Nij responses showing a slightly increasing trend toward the rear of the front section (Figure 9). There are two outliers in which lumbar load is the dominant injury risk. The FAA H3 50th in seat 1A and the THOR ATD in seat 3E. Seat 1A faced a bulkhead with no forward seatback, which meant that all neck loads were due to inertial force, resulting in decreased loading of the neck compared to the ATD’s which experienced seatback contact. The THOR in seat 3E also did not experience seatback contact due to differences in its kinematic response compared to the FAA H3 50 th, which was seated in the adjacent seat 3D. The FAA H3 50th in seat 6D, which was in the braced posture, also predicted different injury responses compared to the upright FAA H3 50 th positioned adjacent.

*Braced position

Figure 9. Normalized injury metric responses predicted in the front section of the aircraft (top) and reference schematic of ATD positioning (bottom).

© ASCE

Earth and Space 2022

Detailed comparisons of the FAA H3 50 th in seat 3D and the THOR in seat 3E show differences in kinematic response of the torso and neck between the two ATDs. The H3 FAA 50th ATD was found to bend at the lumbar spine, the only flexible component of the H3 thoracic-lumbar spinal column, resulting in significant forward flexion of the head and torso. The THOR ATD exhibited more of a curling response, with significant flexion in the cervical spine and minimal flexion within the thoracic-lumbar region in the upper and lower flex joints (Figure 10). The difference in spinal flexibility of the THOR ATD prevented contact with the forward seatback, which is a significant change in its interaction with the environment and altered the predicted risk of injury for a 50 th percentile occupant between these two ATD configurations.

Figure 10. Row 3 Starboard ATD kinematic response at completion of simulation: FAA H3 50th versus THOR. The braced FAA H3 50th ATD seated in seat 6B predicted lower HIC, Nij, and lumbar load values than the upright FAA H3 50 th in the adjacent seat 6A. As the head of the braced ATD was closer to the forward seatback, contact occurred earlier during the impact event. This contact resulted in a smaller differential between head and seat velocity, less time for the ATD to accelerate with respect to the seat prior to contact, and thus a lower transfer of force into the head-neck region of the braced ATD on contact. This effect can be seen in the kinematic response of two ATDs (Figure 11), with the braced ATD exhibiting reduced neck flexion during head-toseat contact than the nominally positioned ATD.

Figure 11. Row 6 Port ATD kinematics: braced versus nominal position.

© ASCE

551

Earth and Space 2022

It should be noted that two of the injuries occurring in the Flight 1549 ditching event were associated with occupants taking the braced position. These injuries were to the shoulders of the occupant (NTSB 2009). The lower injury risk values demonstrated by the ATD in a braced position within this study apply only to skull and vertebral injuries and should not be extended to any determination of full body injury risk. The H3 5th ATD predicts the second highest injury risk in the front section of the aircraft. The Nij value predicted by the H3 5 th ATD is higher than the H3 95 th ATD seated adjacent. In this condition Nij values calculated are primarily driven by compression of the neck as the head impacts the forward seatback. Although the H3 95 th measures a larger peak compressive force, namely due to the larger upper body mass behind the head contact, a higher Nij value is predicted for the H3 5th. This finding is because the intercept values used to calculate Nij are lower for the 5th to account for higher susceptibility for injury under equal load for the smaller anthropometry. The calculated lumbar load risk is higher in the H3 95 th than the H3 5th because a limit based on the 50th percentile occupant size is used for both ATDs, the FAA does not define limits for the other two ATD sizes. The larger H3 95 th mass results in larger compressive forces on the lumbar spine. This calculation does not account for increased musculature and vertebral strength in the larger occupant which may offset the increased load. Application of anthropometry specific lumbar load limits such as those developed for military rotorcraft (Bolukbasi 2011) may provide improved injury prediction across the 5 th-95th anthropometric range in general aviation safety analysis. A single occupant simulation was performed over the wing-box of the aircraft. Seat 9D was evaluated as a representative exit row configuration. With no forward seatback, lumbar load was shown to dominate injury risk predictions (Figure 12). Predicted landing loads were not significant enough to induce head-to-leg contact thus HIC values were negligible and inertial neck loading produced minimal Nij response. Lumbar load values were slightly higher than those observed towards the rear of the forward section.

Figure 12. Normalized injury metric responses predicted in the wing-box section of the aircraft. Occupant injury risk predicted in the aft section of the aircraft exhibited similar distribution of risk regions to those in standard seating configurations throughout the aircraft (Figure 13). The FAA H3 50th in seat 10C exhibited the lowest overall injury risk with respect to all other occupant models simulated. These results are consistent with results observed in previous crash testing and simulation of this seating configuration (Jackson 2018, Putnam 2020). ATDs seated in seat C which is overhanging, i.e. it does not have direct floor support below the seat , have

© ASCE

552

Earth and Space 2022

exhibited reduced loading during the crash event compared to those in other seating positions. The lack of support below seat C allows increased deformation over the inner two seats, which causes it to absorb a larger portion of the crash energy and reduce loads transferred to the occupant.

Figure 13. Normalized injury metric responses predicted in the Aft section of the aircraft. DISCUSSION The horizontal seat base and airframe accelerations predicted for the F28 aircraft simulated under conditions of the US Airways Flight 1549 ditching event were low, generally less than 10 g. The vertical accelerations were also low, generally less than 15 g. The low acceleration response predicted within the cabin of the F28 aircraft resulted in minimal loading of the occupant models evaluated. Overall injury risk prediction was low, with all injury metrics calculated falling below 50% of the defined vehicle certification limits. The neck injury risk metric, Nij, exhibits the highest value with respect to injury limits, followed by lumbar load. HIC values are negligible throughout the occupant models evaluated. The increased spinal flexibility of the THOR ATD over the Hybrid III ATD is shown to result in a difference in injury metric predictions, as different kinematic response leads to a difference in interactions with the forward seatback during the impact event. The braced position is shown to result in reduced injury metric values compared to an upright posture as it reduces closing velocity with the forward seatback as well as changes the orientation of load into the lumbar spine. Of note, two passenger injuries recorded during the Flight 1549 ditching event were shoulder injuries associated with passengers taking the braced position. Although this study indicates reduced neck and spinal injury metrics in the braced position, a more detailed study of the full body response in the braced position would be required to make a determination on its effectiveness in reducing total injury risk during a ditching event. The lack of head, neck, and spinal injury predicted by the F28 occupant models is consistent with the fact that no injuries in these body regions were reported in the Flight 1549 ditching event. This qualitative agreement between the F28 predicted occupant injury risk and the Flight 1549 injury report provides increased confidence in the use of these tools in predicting occupant injury risk during future ditching events.

© ASCE

553

Earth and Space 2022

CONCLUSIONS A previously validated finite element model of a Fokker F28 Fellowship aircraft was used to perform LS-DYNA simulations for water ditching conditions consistent with US Airways Flight 1549 ditching event. The F28 model was simulated for the Flight 1549 impact conditions to evaluate its ability to predict structural and occupant responses in a realistic water ditching environment. Impact conditions recorded from the Flight 1549 event were used to drive a FluidStructure Interaction simulation of the event using the Arbitrary Lagrangian-Eulerian (ALE) capability within LS-DYNA . Several categories of predicted vehicle structural responses were evaluated: impact kinematics, airframe acceleration responses, and airframe damage. In addition, occupant breakout models were developed and simulated using the seat base accelerations predicted by the F28 vehicle model. These occupant breakout models were used to predict occupant injury risk across a variety of seat locations, occupant positions, and ATD types. The simulation-predicted vehicle acceleration in both the vertical and forward directions was low, less than 15 g, resulting in a benign impact event. Likewise, simulations of the occupant breakout models predicted head, neck, and spinal injury metric values well below the defined safety limits for dynamic loading certification. These results were consistent with the lack of head, neck, or spinal injuries recorded in the ditching event. The LS-DYNA simulation predicted damage to the lower rear portion of the airframe, as a result of the 9.5° pitch nose-up initial impact. The damage predicted by the simulation is similar to that seen on the Flight 1549 aircraft post-recovery. These results indicate reasonable prediction of a ditching event by the F28 model using ALE capability within LS-DYNA and lend confidence for future use of water impact simulation utilizing vehicle models validated using land impact test data. REFERENCES Bolukbasi A., Crocco J., Clarke C., Fasanella E., Jackson K., Keary P., et al., “Full Spectrum Crashworthiness Criteria for Rotorcraft,” U.S. Army Research, Development and Engineering Command, Ft. Eustis, VA RDECOM TR 12-D-12, 2011. Credeur M. J., and Schlangenstein M., “US Airways Pilot Averts Tragedy in ‘Miracle’ Landing,” www.Bloomberg.com, New York, January 16, 2009. Fasanella E. L., Jackson K. E., Sparks C. E., and Sareen A. K., “Water Impact Test and Simulation of a Composite Energy Absorbing Fuselage Section,” Journal of the American Helicopter Society, Vol. 50, No. 2, April 2005, pp. 150-164. Federal Aviation Administration (FAA), “14 CFR Part 25.562 emergency landing dynamic conditions,” Federal Aviation Regulation. Volume 53, Amdt. 25-64, 1988. Guha S., “README LSTC H3_95TH_DETAILED Scaled. 151214.V3.03_BETA,” Livermore Software Technology Corporation, MI, 2015. Hallquist J. Q., “LS-DYNA Keyword User’s Manual,” Version 971, Livermore Software Technology Company, Livermore, CA, August 2006. Humanetics, “User’s Manual FAA Hybrid III 50th Male Dummy LS-DYNA Model Version 1.2.3,” Humanetics Innovative Solutions, Plymouth, MI, 2018. Humanetics, “Hybrid III 5 th Percentile Female Dummy LS-DYNA Model Version 7.0.5 User Manual.” Humanetics Innovative Solutions, Plymouth, MI, 2013.

© ASCE

554

Earth and Space 2022

Jackson K. E., and Fuchs Y. T., “Comparison of ALE and SPH Simulations of Vertical Drop Tests of a Composite Fuselage Section into Water,” Proceedings of the 10 th International LS-DYNA Users Conference, Dearborn, MI, June 8-10, 2008. Jackson K. E., Littell J. D., Annett M. S., and Haskin I. M., “Finite Element Simulations of Two Vertical Drop Tests of F-28 Fuselage Sections,” NASA Technical Memorandum NASA/TM-2018-219807, February 2018. Jackson K. E., and Putnam J. B. “Simulation of a Full-Scale Crash Test of a Fokker F28 Fellowship Aircraft,” NASA Technical Memorandum NASA/TM-2020-220435, January 2020. Kleinberger M., Sun E., Eppinger R., Kuppa S., and Saul R., “Development of Improved Injury Criteria for the Assessment of Advanced Automotive Restraint Systems, ” National Highway Traffic Safety Administration, Washington, DC, 1998. Littell J. D., “A Summary of Results from Two Full-Scale Fokker F28 Fuselage Section Drop Tests,” NASA/TM-2018-219829, May 2018. Littell J. D., “A Summary of Airframe Results from a Fokker F-28 Full-Scale Crash Test,” NASA Technical Memorandum, NASA/TM-2020-220572, March 2020. National Transportation Safety Board (NTSB) Report, “Loss of Thrust in Both Engines after Encountering a Flock of Birds and Subsequent Ditching on the Hudson River US Airways Flight 1549 Airbus A320‐214, AAR-10/03, January 2009. National Highway Traffic Safety Administration (NHTSA), “Docket No. NHTSA-20150119.” New Car Assessment Program (NCAP), 2015. Panzer M. B., et al., “THOR 50th Male Finite Element Model User Manual,” Model Version 2.1 for LS-Dyna,” 2015. Patel A. A., and Greenwood R. P., “Transport Water Impact and Ditching Performance,” DOT/FAA/AR-95/54, March 1996. Putnam J. B., et al., “Development, calibration, and validation of a head–neck complex of THOR mod kit finite element model,” Traffic Injury Prevention 15(8): 844-854, 2014. Putnam J. B., et al., “Development and evaluation of a finite element model of the THOR for occupant protection of spaceflight crewmembers,” Accident Analysis & Prevention, 82: 244-256, 2015. Putnam J. B. “Occupant Response Analysis of a Full-Scale Crash Test of a Fokker F28 Fellowship Aircraft,” NASA Technical Memorandum NASA/TM-2020-220571, March 2020. Vassilakos G. J., and Mark S. D. “Boilerplate Test Article (BTA) Water Impact Test Correlation,” NASA Contractor Report, NASA/CR-2017-219792, November 2017.

© ASCE

555

556

Earth and Space 2022

A Compact Delayed Photocurrent Model Based on a Reduced Order Data-Driven Exponential Time Integrator* K. Chad Sockwell1 , Pavel Bochev2 , and Biliana Paskaleva3 1 Sandia

National Laboratories, [email protected] National Laboratories, [email protected] 3 Sandia National Laboratories, [email protected]

2 Sandia

ABSTRACT Analysis of radiation effects on electrical circuits requires computationally efficient compact radiation models. Currently, development of such models is dominated by analytic techniques that rely on empirical assumptions and physical approximations to render the governing equations solvable in closed form. In this paper we demonstrate an alternative numerical approach for the development of a compact delayed photocurrent model for a pn-junction device. Our approach combines a system identification step with a projection-based model order reduction step to obtain a small discrete time dynamical system describing the dynamics of the excess carriers in the device. Application of the model amounts to a few small matrix-vector multiplications having minimal computational cost. We demonstrate the model using a radiation pulse test for a synthetic pn-junction device.

1

INTRODUCTION

When a semiconductor device is exposed to a pulse of ionizing radiation, it produces excess carriers that are not present in normal environments. The photocurrents generated by these carriers can alter the behavior of the device and present threats for microelectronic components operating in radiation environments such as navigation and communication satellites. These threats can be analyzed by performing circuit simulations using simulators such as Xyce, described in Keiter et al. (2020). This, however, requires computationally efficient compact photocurrent models. * Sandia National Laboratories is a multimission laboratory managed and operated by National Tech-

nology and Engineering Solutions of Sandia, LLC., a wholly owned subsidiary of Honeywell International, Inc., for the U.S. Department of Energy’s National Nuclear Security Administration under contract DE-NA-0003525.

© ASCE

557

Earth and Space 2022

Broadly speaking, for the past six decades formulation of such models has followed two distinct development paths. The first one simulates the device response by using equivalent electrical circuits; see, e.g., the work in Caldwell et al. (1963), Raymond et al. (1968), Raymond and Krebs (1972). The second approach uses analytic solution techniques, such as Laplace transforms, described in Wirth and Rogers (1964) or Enlow and Alexander (1988) or Fourier transforms, described in Wunsch and Axness (1992) or Axness et al. (2004), to obtain closed form solutions of the Partial Differential Equations (PDEs) used to model the device physics. The Wirth-Rogers model presented by Wirth and Rogers (1964) is perhaps the earliest example of such compact analytic models, that also established the setting that has been used in the development of virtually all subsequent compact analytic models. This setting comprises a one-dimensional pn-junction device divided into a depletion region Ωd , surrounding the pn-junction, and quasi-neutral regions Ωn and Ωp , adjacent to Ωd . The photocurrents in each region are modeled independently and the total photocurrent I(t) is given by their sum, i.e., I(t) := IΩd (t) + IΩn (t) + IΩp (t) , where IΩd (t), IΩn (t), and IΩp (t) are the photocurrents collected from Ωd , Ωn , and Ωp , respectively. To estimate IΩd (t) one invokes the depletion approximation; see, e.g., Axness et al. (2004), which leads to the simple, closed form expression Id (t) = qg(t)Aw,

(1)

where g(t) is the generation rate, q is the electron charge, A is the area of the pnjunction, and w is the width of the depletion region Ωd . Formula (1) implies that the depletion region responds instantly to the radiation source g(t). For this reason Id (t) is often referred to as the prompt photocurrent. Following Axness et al. (2004), the photocurrents in the quasi-neutral regions are usually estimated by invoking the charge neutrality and the congruence assumptions along with restricting g(t) to low injection rates. Under these assumptions, the dynamics of the excess carrier concentrations in Ωn and Ωp can be modeled by a constant coefficient version of the Ambipolar Diffusion Equation (ADE), described in McKelvey (1986). Compact analytic models then proceed by solving the two ADEs, in Ωn and Ωp , analytically and using the resulting closed form solutions to estimate IΩn (t) and IΩp (t). Because the dynamics of the excess carriers in the quasi-neutral regions are governed by parabolic PDEs, the response of Ωn and Ωp lags the radiation source g(t). Thus, IΩn (t) and IΩp (t) are usually called delayed photocurrents. Deriving exact solutions of the ADE is often the main challenge in the development of traditional analytic models and provides an opportunity to demonstrate the potential of numerical, data-driven techniques. Accordingly, in this work we shall adopt the above setting and focus on the development of compact numerical models for the delayed photocurrents IΩn (t) and IΩp (t). Our strategy will be to combine a system identification step with a projection-based model order reduction step to obtain small discrete time dynamical systems describing the dynamics of the excess carriers in Ωn and Ωp . These systems, along with linear current readout maps define our numerical compact photocurrent models. The first step in our approach utilizes the data-driven exponential time integrator (ETI) model formulated in Bochev and Paskaleva (2021). The ETI model uses the ADE © ASCE

558

Earth and Space 2022

as a physics basis to facilitate the learning of its dynamic and spatial operators from synthetic and/or experimental training data sets. This step can be viewed as a physicsinformed data-driven discovery of a discrete time dynamical system. The accuracy of the ETI model per degree of freedom is roughly the same as that of a piecewise linear finite element discretization of the ADE. As a result, the ETI model learned at the first step may become too large for circuit simulations involving more than just a few devices. To resolve this issue, at the second step we treat the learned ETI model as a fullorder model (FOM) and apply projection-based model order reduction (MOR) to obtain a reduced order model, ETI-ROM model, with a much smaller dimension. However, unlike typical MOR, which finds a reduced basis from snapshots collected by running the FOM, here we determine this basis by reusing the training datasets from the first, system identification, step in our model development workflow. The ETI-ROM formulated in this paper falls between the purely data-driven dynamic mode decomposition-based model in Hanson et al. (2021) and the ROM model in Hanson et al. (2022), which uses a finite element discretization of the ADE as a full order model. ETI-ROM combines some of the advantages of these models and provides further evidence that numerical and data-driven techniques can serve a viable alternative to traditional analytic model development approaches. The rest of the paper is organized as follows. Sections 2 and 3 describe the two steps of our compact model development strategy, i.e., the formulation of the full order ETI model (ETI-FOM) and the subsequent application of projection-based MOR to that model. Since the development of ETI-FOM and ETI-ROM follows the same steps on Ωn and Ωp , Sections 2-3 present the details only for Ωn . Section 4 demonstrates the compact ETI-ROM model by using it to simulate the response of a synthetic pnjunction device to a short radiation pulse.

2

FULL ORDER ETI PHOTOCURRENT MODEL (ETI-FOM)

For completeness, we briefly review the formulation of the data-driven exponential time integrator model for Ωn . We refer to Bochev and Paskaleva (2021) for additional details about the model and the generation of the training datasets. Without a loss of generality we assume that Ωn is to the left of Ωd and set Ωn = (0, wn ). Thus, the boundary between Ωn and Ωd is x = wn . We then consider the parabolic ADE ut = Dp uxx − µp E(t) · ux − u/τp + g

in Ωn × T ,

(2)

where u is the excess hole carrier concentration in Ωn , Dp , µp and τp are the diffusion, mobility and lifetime of the holes, E(t) is the applied electric field in Ωn , and T = (0, t∗ ) is the desired time interval for the simulation. We augment (2) with homogeneous initial and boundary conditions, i.e., u(x, 0) = 0 for all x ∈ Ωn and u(0, t) = u(wn , t) = 0 for all t ∈ T . We will transform (2) into a continuous time dynamical system by applying the method of lines. To that end, recall the standard weak form of (2): seek u ∈ C 1 (T ; H01 (Ω)) such that (ut , v) + a(u, v) = (v, g) ∀v ∈ H01 (Ω) , ∀t ∈ T , (3) © ASCE

559

Earth and Space 2022

where H01 is the Sobolev space of order one whose elements vanish on the boundary points x = 0 and x = wn , (·, ·) is the L2 inner product on Ω, and a(·, ·) : H01 (Ω)×H01 (Ω) is a bilinear form given by a(v, w) = Dp (vx , wx ) + µp E(t)(vx , w) + τp−1 (v, w),

∀v, w ∈ H01 (Ω) .

Next, we use Np interior points 0 < q1 < . . . < qNp < wn to divide Ωn into Np + 1 elements Ki = [qi , qi+1 ] , i = 0, . . . , Np , and consider a conforming piecewise linear finite element space V0h defined with respect to this finite element partition. We assume Np that V0h is endowed with the standard Lagrangian basis {vi }i=1 such that vi (qj ) = δij . Let gh (x, t) be the finite element interpolant of the forcing term g with coefficient vector g(t) = (g1 (t), . . . , gNp (t)) ∈ RNp , i.e., gh (x, t) =

Np X

gi (t)vi (x) .

i=1

Replacing g with its interpolant and restricting (3) to V0h then yields the discrete weak problem seek uh ∈ L2 (T ; V0h ) such that ((uh )t , vh ) + a(uh , vh ) = (vh , gh )

∀vh ∈ V01 (Ω) , ∀t ∈ T .

(4)

Let u(t) = (u1 (t), . . . , uNp (t)) ∈ RNp be the coefficient vector of the finite element solution uh , i.e., Np X uh (x, t) = ui (t)vi (x) . (5) i=1

It is easy to see that (4) is equivalent to a system of ODEs for the unknown coefficient vector u(t) ˙ u(t) + Qu(t) = g(t) , (6) where Q = M −1 C, Mij = (vi , vj ), and Cij = a(vi , vj ), i, j = 1, . . . , Np . Eq. (6) embodies the available physics knowledge about our device. We will use its structure to facilitate the discovery of a discrete time dynamical system that approximates the excess carrier dynamics in Ωn using experimental and/or synthetic training data. In this work we shall use synthetic datasets obtained by using (4) to simulate the time-series responses usk , s = 1, . . . , Ns , to a set of forcing terms g s (x, t). This data is then arranged into snapshot matrices ek = [u1k . . . uNs ] , U k

ek = [gk1 . . . g Ns ] , G k

k = 0, . . . , M ,

(7)

see Bochev and Paskaleva (2021) for details. 2.1

EXPONENTIAL TIME INTEGRATOR DISCRETIZATION

To obtain a fully-discrete form of (6), we apply an exponential time integrator technique (ETI) described by Hochbruck and Ostermann (2010). Let the time domain T = [0, t∗ ]

© ASCE

560

Earth and Space 2022

be discretized by points {tk }M k=1 with uniform interval lengths ∆t = tk+1 − tk , k = 1, . . . , M and consider the fundamental solution of the ODE (6): Z tk+1 u(tk+1 ) = exp(∆tQ)u(tk ) + exp(Q(tk+1 − s))g(s) ds . (8) tk

We approximate the source term in (8) as g(s) ≈ gk = g(tk ) to arrive at Z tk+1  u(tk+1 ) ≈ exp(∆tQ)u(tk ) + exp(Q(tk+1 − s)) ds gk .

(9)

tk

The integral in (9) can be evaluated exactly to obtain u(tk+1 ) ≈ exp(∆tQ)u(tk ) + Q−1 (exp(∆tQ) − I)gk .

(10)

The right hand side in (10) then defines the following ETI scheme: uk+1 := uk + ∆tΦ(∆tQ)(gk + Quk ) with Φ(∆tQ) :=

Q−1 (exp(∆tQ) − I), (11) ∆t

where uk ≈ u(tk ). The ETI scheme (11) is completely defined by the function1 Φ(∆tQ) and the matrix Q. We now briefly explain how these terms can be inferred from the dataset (7) and refer to Bochev and Paskaleva (2021) for full details. 2.2

LEARNING THE Q OPERATOR

To infer the linear operator Q = M −1 C note that the structure of C is prescribed by (2) and is given by C = Dp K + µp E(t)N + τ −1 M , (12) p

where Cij = a(vi , vj ), Kij = ((vi )x , (vj )x ), Nij = ((vi )x , vj ) and Mij = (vi , vj ), i, j = 1, . . . , Np . Thus, learning Q boils down to inferring the physical coefficients in (12) from the training data. To that end we shall use a least-squares operatorregression method similar to the one presented by DeGennaro et al. (2019); Li et al. b ≈ C of the form C b := a1 K + (2019). Specifically, we seek an approximation C a2 N + a3 M , where a = (a1 , a2 , a3 ) are the unknown physical coefficients. To set up an overdetermined linear system for these coefficients we first replace the time derivative in (4) by a central finite difference approximation at some time instance tk , with index k > q: q X ˙ k) ≈ u(t αj uk+j . (13) j=−q

The matrix form of the resulting fully discrete Galerkin problem is given by q X



αj vi> M uk+j + Dp vi> Kuk + µp E(t) vi> N uk

j=−q


+ τa−1 vi> M uk = vi> M gk , 1

© ASCE

One can show that the function Φ(∆tQ) has a removable singularity.

i = 1, . . . , Np , (14)

561

Earth and Space 2022

where vi is the coefficient vector of the ith nodal basis function. Eq. (14) should ek and G ek in (7). Treating Dp , µp and hold approximately for the training data in U τp as unknowns then yields the following overdetermined Ns2 -by-3 linear system of equations:


a1 (urk )> Kusk + a2 (urk )> N usk + a3 (urk )> M usk q X r > = (uk ) M gk − αj (urk )> M usk+j ,

r, s = 1, . . . , Ns .

j=−q

The above problem can be written compactly as Y a = b , where Y is Ns2 -by-3 matrix b = Y + b, and a = (a1 , a2 , a3 ). This system can be solved in the least squares sense as a b is where Y + is the Moore-Penrose pseudoinverse of Y . The approximation to Q, Q, −1 b = M C. b obtained by applying the inverse action of M , i.e., Q 2.3

LEARNING THE Φ OPERATOR

b by rewriting (11) as the following input-output We learn an approximation of Φ, Φ, relation: ∆t−1 (uk+1 − uk ) = Φ(∆tQ)(gk + Quk ) , (15) and note that (15) should hold approximately for the data snapshots, i.e., ∆t−1 (usk+1 − usk ) = Φ(∆tQ)(gks + Qusk )

s = 1, . . . , Ns .

(16)

Ns Ns Ns 1 1 s Let Xk0 = [(u1k − u1k+1 ) . . . (uN k − uk+1 )] and Xk = [(gk − Quk ) . . . (gk − Quk )]. Treating Φ(∆tQ) as unknown in (16) we then arrive at an overdetermined system Xk0 = b k , for the approximate dynamics operator with solution ΦX

b = Xk0 X + , Φ k where Xk+ is the Moore-Pensore pseudo-inverse of Xk . Since the problem is linear, any time step k can be used to identify the dynamic operator Φ. It is convenient to choose k = 0, combined with the fact that u0 = 0 to make the procedure independent of Q, leading to the reduction of Xk to Xk = [gk1 . . . gkNs ] . b and Φ b into the ETI scheme (11) yields the Finally, inserting the learned operators Q data-driven ETI-FOM: b k + Qu b k) . uk+1 = uk + ∆tΦ(g

3

(17)

ETI-ROM COMPACT PHOTOCURRENT MODEL

The accuracy of the learned model (17) is comparable to that of the piecewise linear finite element space underpinning the construction of the matrices M , K and N in (12). © ASCE

562

Earth and Space 2022

As a result, this model may still require a large number of spatial degrees of freedom, which would reduce its utility for circuit simulations involving more than just a few devices. The second step in our compact model development approach mitigates this issue by applying a modified projection-based model order reduction (MOR) workflow to obtain a more cost-efficient variant of (17). To explain these modifications recall that in a traditional POD (Proper orthogonal decomposition) MOR workflow the full order model (FOM) would be given by the weak Galerkin problem (4). One uses this FOM to generate snapshots and then obtains a reduced order basis from these snapshots by applying a POD; see, e.g., Liang et al. (2002) to the snapshot matrix. Finally, one replaces the standard finite element basis functions in the weak Galerkin problem (4) by the reduced basis functions to obtain the reduced order model (ROM) for (4). In the present context we are not interested in a ROM for (4), rather our goal is to reduce the size of the learned model (17), i.e., we wish to treat (17) as the FOM. At the same time, being a data-driven model, (17) inevitably contains regression errors incurred while inferring Q and Φ from data. If we were to follow the standard PODMOR workflow and use (17) to generate the snapshots, these regression errors would be propagated to the reduced basis. However, such an error pollution can be avoided by generating the reduced basis directly from the same training set (7) that was used to infer the learned model (17). To implement such a modified POD-MOR workflow, we first rearrange the training ek from (7) into a single Np × M Ns snapshot matrix as follows: samples U s s b = [u11 . . . u1M . . . uN U . . . uN 1 M] .

(18)

b rather than a snapshot matrix generated by the learned model The idea is to use U (17), to obtain the reduced basis. Thus, at the second step of our modified POD-MOR b = W ΣV > , choose workflow we compute the singular value decomposition (SVD) U an integer r < min{Np , M Ns }, and select the first r left singular vectors (columns of W ), associated with the largest r singular values to define the reduced basis. This choice is motivated by the fact that the truncated SVD approximation br = Wr Σr V > , U r

(19)

where Wr , Σr , and Vr are the truncated quantities from the SVD corresponding to the r largest singular-values σj , provides the best, with respect to the squared Frobenius b . Specifically, we have that norm k · k2F , rank-r approximation of U b − ΨΨ> U b ||2F Wr = arg min ||U rank(Ψ)=r

such that Ψ> Ψ = I ,

(20)

where I is the identity operator. This choice of basis, Wr , gives a residual in (20) which equals the sum of the discarded singular values from the SVD: b − Wr W > U b ||F = ||U r

d−r X j=1

© ASCE

σj2 .

(21)

563

Earth and Space 2022

In (21) d is the total number of non-zero singular values. Once the reduced basis Wr = [w1 . . . wr ] has been determined, we project the learned model and solution onto the reduced basis. The solution uk is approximated by a linear combination of the reduced basis bk = uk ≈ u

r X

wr ar (tk ) = Wr a(tk ) =: Wr ak

(22)

j=1

where a(t) = (a1 (t), . . . , ar (t)) is a vector containing the reduced basis coefficients. Inserting (22) into (17), gives an over-constrained relationship b k + QW b r ak ) + R , Wr ak+1 = Wr ak + ∆tΦ(g

(23)

where R is the residual induced by the approximation. We then project (23) onto the reduced space by multiplying on the left by Wr> and use the fact that Wr> R = 0 to obtain b k + QW b r ak ) , Wr> Wr ak+1 = Wr> Wr ak + ∆tWr> Φ(g Since Wr> Wr = Ir , where Ir ∈ Rr×r is the identity matrix, we arrive at the following reduced order version of the learned model (17) b k + QW b r ak ) . ak+1 = ak + ∆tWr> Φ(g

(24)

b = Wr> Φ b QW b r ∈ Rr×r and the vectors fbk = For efficient simulations the operator A b k ∈ Rr can be precomputed and stored before simulation, resulting in the folWr> Φg lowing compact form of the ETI-ROM: b k + fbk ) . ak+1 = ak + ∆t(Aa

(25)

To obtain a delayed photocurrent reading from (25) at t = tk+1 we proceed as b k+1 = Wr ak+1 . follows. First, we project the ROM solution ak+1 to a full state u b k+1 to a finite element function uh (x, tk+1 ) by using the entries of u b k+1 Then, we lift u as coefficients in (5). Finally, we compute the photocurrent at the boundary x = wn between Ωn and Ωd as IΩn (tk+1 ) = qAb a1 ∂x uh (x, tk+1 ) , (26) x=wn

b q is the electron charge and where b a1 is the first coefficient of the learned operator Q, A is the device area. The reproductive error of (25), i.e., its ability to accurately approximate the snapb , can be controlled by choosing r such that the so-called basis energy satisfies shots in U the lower bound r d X X 2 σj / σj2 ≥ 1 − δ , (27) j=1

j=1

where δ > 0 is the desired threshold for the relative reproductive error.

© ASCE

564

Earth and Space 2022

Table 1: Specification of synthetic device for numerical studies. Symbol Dp µp τp Nd wn E

Value 101

1.19 × 4.64 × 102 1.97 × 10−5 1.00 × 1017 3.08 × 10−2 −20 or 0

Units

Description

cm2 s−1

Minority carrier diffusion constant in the n-region at low injection Minority carrier mobility in the n-region Minority carrier lifetime in the n-region at low injection Doping density in the n-region (Donor Concentration) Width of the quasi-neutral n-region Electric field

cm2 V−1 s s cm−3 cm V cm−1

Let us now discuss conditions under which the ETI-ROM model (24) can realize performance gains over its parent ETI-FOM, while retaining acceptable reproductive accuracy. The main cost of ETI-FOM in terms of floating point operations is in matrixvector multiplications involving Np × Np matrices, while the main cost of ETI-ROM b It follows that an is the matrix-vector multiplication involving the r × r matrix A. effective ETI-ROM requires r  Np . At the same time, a general rule of the thumb is that an accurate ROM requires a reduced basis that retains at least 90% of the snapshot energy. To obtain such a basis one should determine r by setting the threshold δ to at most 0.1 in (27). Whether or not the number r of basis vectors selected in this manner will also satisfy r  Np depends b possesses a low-rank structure, identified by a on whether or not the snapshot matrix U b is well approximated by the sharp decrease in the singular-values in Σ. In this case U rank-r truncated SVD (19).

4

NUMERICAL RESULTS

In this section we demonstrate the performance of the ETI-ROM in the predictive regime by solving the ADE with a source term that has not been used in the generation of the solution snapshots (7). To that end we consider a radiation pulse test problem from (Axness et al., 2004, Section B) in which a synthetic device with the parameters specified in Table 1 is irradiated by a constant in space pulse, defined as ( 4.3 × 1022 pairs/cm3 sec if 0 ≤ t ≤ 1µs g(x, t) = . (28) 0 if 1µs < t In contrast, the forcing terms g s (x, t) in (7) are mollifiers with respect to the spatial variable and are locally supported in Ωn . Following Axness et al. (2004) we solve the test problem with applied voltages set to E = 0V and E = −20V. For this study we choose ETI-FOM models with Np = 128, Np = 256, and set ∆t = 0.001 µ sec. To train these models we use synthetic data sets generated by solving s (4) on a mesh with Np = 1024 for source term sets {g s (x, t)}N s=1 , Ns ∈ {10, 15, 24}, comprising mollifiers with radius r = 100µm. Table 2 shows the combinations of Np , Ns and E defining the example configurations. In all cases the ETI-FOM models were trained using a 7-point finite difference rule in (13). We refer to Bochev and Paskaleva (2021) for further details. © ASCE

Earth and Space 2022

Figure 1: Decay of the singular values for select snapshot matrices. Top row: E = −20 V cm−1 . Bottom row: E = 0 V cm−1 . Left column: Np = 128. Right column: Np = 256.

We then arrange the solution data from these synthetic data sets to obtain the snapb shot matrices in (18) and compute their SVDs. We first verify that the matrices U have the low-rank structure necessary to support the construction of an effective and accurate reduced basis. Figure 1 shows logarithmic plots of the singular values for the snapshot matrices in Experiments 2 and 6 (E = −20 V cm−1 ) and Experiments 4 and 8 (E = 0 V cm−1 ). The plots clearly show an exponential decay of the singular values, thereby confirming that the snapshot matrices in these examples do indeed have a lowrank structure. We only show these four cases because the same exponential decay was observed for all eight experiments. For each ETI-FOM example configuration we consider five different ETI-ROMs obtained by using reduced bases that retain 90%, 99.9%, 99.99%, 99.999, and 99.9999% of the snapshot energy, respectively. Table 2 shows the number of left singular vectors in each basis and the corresponding values of δ used in (27). The data in this table reveals that the number of modes necessary to achieve the highest reproductive accuracy does not exceed r = 13. In our study we compare the delayed photocurrent readouts (26) from the five ETIROM models with their ETI-FOM parent and the “best-possible” ETI model defined by taking Q to be the exact discrete Galerkin operator and using the MatlabTM function expm(A) to evaluate the matrix exponential in the formula (11) for Φ(∆tQ). Because the ADE (2) with the source term (28) does not have a closed form analytic solution, to compute the errors of ETI-FOM and ETI-ROM we employ the over-refined solution technique. Specifically, we solve (4) on the same mesh where the synthetic training data was generated and then use this finite element solution as a proxy for the exact solution in (26) to define the “exact” delayed photocurrent reading. © ASCE

565

566

Earth and Space 2022

Table 2: Number of right singular vectors required to obtain energy error tolerance δ for experiments (Ex.) 1 through 8. Dimension r of the reduced basis for a given energy threshold in (27).

ETI-FOM Np

Ns

E

1×10−1

1×10−3

1×10−4

1×10−5

1×10−6

←δ

128 128

10 24

-20 -20

3 3

7 7

9 9

10 11

12 13

← Ex.1 ← Ex.2

128 128

10 15

0 0

3 3

7 7

9 9

10 10

12 12

← Ex.3 ← Ex.4

256 256

15 24

-20 -20

3 3

7 7

9 9

11 11

13 13

← Ex.5 ← Ex.6

256 256

15 24

0 0

3 3

7 7

9 9

10 11

12 12

← Ex.7 ← Ex.8

Our results for the eight experiments are presented in Figures 2–9. The left plot in each figure compares the delayed photocurrents of the five ETI-ROM models, the ETI-FOM, and the “best” ETI with the “exact” delayed photocurrent. In the plots each ETI-ROMs is identified by the size of its reduced basis, and the ETI-FOM is labeled as “ETI”. The plots on the right show the error of each ETI-ROM model, the best ETI model, and the ETI-FOM model compared to the “exact” delayed photocurrent. In all experiments with Np = 128, many of the ROM models exhibited a smaller error than the parent ETI-FOM. In particular, all ETI-ROMs with basis size r > 3 performed especially well when Np = 128. For the experiments with Np = 256, and E = −20V all ETI-ROM models with r > 3 performed very well and the models with r = 13 were at least as good as the parent ETI-FOM. We observed similar results in the pure diffusion case E = 0V with r = 12 producing excellent results. We note that the accuracy of the ETI-ROM must also be weighed against the large increase of performance the ETI-ROM provides. Although the speed up is not listed here, the cost of the most expensive ETI-ROM in Examples 5-6 is proportional to O(132 ) floating point operations (flops), whereas the cost of its parent ETI-FOM is proportional to O(2562 ) flops. Thus, compared to the ETI-FOM, the cost of ETI-ROM is reduced by a factor of almost 400, while its accuracy is comparable to that of the “best” scheme. Preliminary results suggest that the ETI-ROM models will remain accurate for large Np , without a significant increase in r, providing very efficient low cost models.

CONCLUSIONS In this paper, a novel reduced order data-driven exponential time integrator was derived to obtain a compact numerical model for the delayed photocurrents in a pn-junction device. The spatial parameters and dynamical evolution operator were learned from a synthetic training-data set simulating the type of data that can be generated in a facility such as Sandia’s Ion Beam Lab. The learned ETI scheme was then reduced to a highly efficient model using POD-based model order reduction. POD was applied

© ASCE

567

Earth and Space 2022

Figure 2: Experiment 1 (Ex.1). Left: photocurrents. Right: photocurrent error.

Figure 3: Experiment 2 (Ex.2) Left: photocurrents. Right: photocurrent error.

Figure 4: Experiment 3 (Ex.3). Left: photocurrents. Right: photocurrent error.

Figure 5: Experiment 4 (Ex.4). Left: photocurrents. Right: photocurrent error.

© ASCE

568

Earth and Space 2022

Figure 6: Experiment 5 (Ex.5). Left: photocurrents. Right: photocurrent error.

Figure 7: Experiment 6 (Ex.6). Left: photocurrents. Right: photocurrent error.

Figure 8: Experiment 7 (Ex.7). Left: photocurrents. Right: photocurrent error.

Figure 9: Experiment 7 (Ex.7). Left: photocurrents. Right: photocurrent error.

© ASCE

Earth and Space 2022

to the training-data set to obtain the reduced basis, followed by a projection of the ETI scheme onto that basis to obtain an ETI-ROM model. The ETI-ROM models, with varying reduced basis sizes, were run in the predictive regime and compared to the learned ETI-FOM and to a model which used the analytically calculated operators. The ETI-ROM’s largest error was less than 20%, and in some cases less than 1%, even out performing the ETI-FOM model in some cases. This was done with a relatively small basis sizes 3 ≤ r ≤ 13. Overall, the ETI-ROM model is reasonably accurate while possessing far less degrees of freedom than the ETI-FOM model, making the ETI-ROM model better suited for circuit simulations. ACKNOWLEDGMENTS This work was supported by the Sandia National Laboratories (SNL) Laboratory-directed Research and Development (LDRD) program, and the U.S. Department of Energy, Office of Science, Office of Advanced Scientific Computing Research under Award Number DE-SC-0000230927 and under the Collaboratory on Mathematics and Physics-Informed Learning Machines for Multiscale and Multiphysics Problems (PhILMs) project. This paper describes objective technical results and analysis. Any subjective views or opinions that might be expressed in the paper do not necessarily represent the views of the U.S. Department of Energy or the United States Government.

REFERENCES Axness, C. L., Kerr, B., and Wunsch, T. F. (2004). Analytic light—or radiation— induced pn junction photocurrent solutions to the multidimensional ambipolar diffusion equation. Journal of Applied Physics, 96(5):2646–2655. Bochev, P. and Paskaleva, B. (2021). Development of data-driven exponential integrators with application to modeling of delay photocurrents. Numerical Methods for Partial Differential Equations, In print. Caldwell, R. S., Gage, D. S., and Hanson, G. H. (1963). The transient behavior of transistors due to ionized radiation pulses. Transactions of the American Institute of Electrical Engineers, Part I: Communication and Electronics, 81(6):483–491. DeGennaro, A. M., Urban, N. M., Nadiga, B. T., and Haut, T. (2019). Model structural inference using local dynamic operators. International Journal for Uncertainty Quantification, 9(1):59–83. Enlow, E. W. and Alexander, D. R. (1988). Photocurrent modeling of modern microcircuit pn junctions. IEEE Transactions on Nuclear Science, 35(6):1467–1474. Hanson, J., Bochev, P., and Paskaleva, B. (2021). Learning compact physics-aware delayed photocurrent models using dynamic mode decomposition. Statistical Analysis and Data Mining: The ASA Data Science Journal, 14(6):521–535. Hanson, J., Bochev, P., Paskaleva, B., Keiter, E., and Hembree, C. E. (2022). A hybrid analytic-numerical compact model for radiation induced photocurrent effects. IEEE Transactions on Nuclear Science. © ASCE

569

Earth and Space 2022

Hochbruck, M. and Ostermann, A. (2010). Exponential integrators. Acta Numerica, 19:209–286. Keiter, E. R., Russo, T. V., Schiek, R. L., Thornquist, H. K., Mei, T., Verley, J. C., Sholander, P. E., and Aadithya, K. V. (2020). Xyce Parallel Electronic Simulator: Users’ Guide, Version 7.2. Technical Report SAND2020-11842, Sandia National Laboratories, Albuquerque, NM. Li, X., Dirr, N., Embacher, P., Zimmer, J., and Reina, C. (2019). Harnessing fluctuations to discover dissipative evolution equations. Journal of the Mechanics and Physics of Solids, 131:240 – 251. Liang, Y., Lee, H., Lim, S., Lin, W., Lee, K., and Wu, C. (2002). Proper orthogonal decomposition and its applications—part i: Theory. Journal of Sound and vibration, 252(3):527–544. McKelvey, J. P. (1986). Solid State and Semiconductor Physics. R.E. Krieger Publishing Co., Malabar, FL. Raymond, J. P., Chang, W. W., and Budris, R. E. (1968). Lumped-model analysis of microcircuit vulnerability. IEEE Transactions on Nuclear Science, 15(6):271–278. Raymond, J. P. and Krebs, M. G. (1972). Lumped model analysis of semiconductor devices using the net-2 circuit/system analysis program. IEEE Transactions on Nuclear Science, 19(6):103–107. Wirth, J. L. and Rogers, S. C. (1964). The transient response of transistors and diodes to ionizing radiation. IEEE Transactions on Nuclear Science, 11(5):24–38. Wunsch, T. F. and Axness, C. L. (1992). Modeling the time-dependent transient radiation response of semiconductor junctions. IEEE Transactions on Nuclear Science, 39(6):2158–2169.

© ASCE

570

571

Earth and Space 2022

Turk Salty Concrete (TSC) Can Isolate the Freshwater Interface against the Sea Water Intrusion and Salty Formations Afshin Turk1 1

Khuzestan Water and Power Authority, Ahvaz, Khuzestan, Iran. Email: [email protected]

ABSTRACT Saline water causes a number of serious problems for both shoreline constructions and freshwater aquifers; what’s more, coastal aquifers face major problems such as seawater intrusion into the interface and a decreasing in the amount of water resources during dro ught when water allocations from surface resources are lower less than normal. It is proposed that by injecting TSC into the interface area at a specified depth it would be possible to isolate against saline water intrusion. In 2004, huge salt domes were identified in the Gotvand Dam reservoir which could if released affect the freshwater in the Karun River. TSC grout can be used to isolate salt formations in the form of vertical curtain. SEM imaging at a resolution of x2,000 have shown the adhesiveness of fibers in TSC. The images show that fibers and salt concrete blend producing the ancient mummification techniques. TSC can reduce the costs and time of grouting. INTRODUCTION Seawater penetrates concrete coastal structures especially the piles of ports, causing damage such as skin cracking, steel bars sulfating and corrosion which manifest themselves on protective walls along the shoreline, Thompson (1987) and Chegini (2011). TSC was invented as a measure to mitigate salt formations and protect sea shorelines, Turk (2017, 2018a); whatsmore, TSC has the capability to prevent the intrusion of saline water into structures and aquifers offshore. Due to the fact that TSC materials are easily available in nature and can be used as a specific mixture per case, it is an economical solution to resolve the matter of saline water intrusion into the interface of the aquifer. It is believed that TSC can be applied in areas in the ARVAND River Estuary, and along the coastline of the Persian Gulf, conditional to the providing of funds for such projects. Grouting operations at the Interface of freshwater and seawater would then begin from 1m in shallow water (clay silt soil) to 15m in proximity to the shoreline (silt clay soil), KTNSE (2017). In 2017 it was proposed that TSC be used to modify the amount of EC in the GOTVAND reservoir in the SW of IRAN in as such that TSC grouting would be used to create a sealing curtain around the salt domes.. In the First Report of "the Karun River Development plan" which were submitted to the Government by HARZA Consulting Engineering Company in May 1967 it has been stated that the GOTVAND dam site should be located 12.5km upstream of its present site which would thus place the salt domes downstream of the dam. This implies that in the original study mitigatory efforts to prevent salt intrusion into the waterway had been anticipated. By adding fibers into the mixture, the reinforced behavior of TSC samples can be intensified. Tri- axial stress- strain tests show the flexibility behavior of TSC, while complementary research is ongoing regarding the use of GRP, crushed plastic waste bottles, synthetic fibers, natural cotton, date palm tree fronds and Reedy fibers in the mixture design. TSC materials continuities have been present ed

© ASCE

572

Earth and Space 2022

through SEM images, CENLAB (2019, 2020), in which the SEM images have demonstrated fiber mummification and the adhesiveness of waste plastic fibers, Turk (2007, 2008), Saltmen (1993), Waltari (1945). TSC MATERIALS The TSC materials used in this study were a mixture of Salt stone (S), Water (W), Cement type II or V (Cii, Cv), SHUSHTAR Ceramic Clay (C), Bentonite (B) and waste plastic fiber (F). Equation 1 defines ζTurk of TSC which is includes the existing salt water. The values of ζTurk varies from 0.75 to 6.69. Equation 2 explains the figure μTurk which is the index of salt resistance. Figure μTurk is the combination of an axial resistance of S950713 and ζTurk1 at 10 days of curing. Table 1, presents ζTurk, μTurk and Modulus of Elasticity. Figure 1 represents the behavior of materials vs the axial resistance of samples after 10 day of curing. Since 2016, samples have been taken to determine the saturation mode, KTNSE (2017). Although, the samples were broken up, all of them retained a stable shape in water. Although, the samples are broken, all of them have stable shapes in water. Table 1. Materials Percentages of TSC, E (kg/cm²), t (days), KTNSE (2019). i 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

© ASCE

Sample 950713 961222 970216P 970301G 970328P 980625P 980627P 980710P 980720P 980722P 980729P 980924P 981212P 981220P 990706P 990708P

t 26 55 13 7 8 19 11 46 36 12 30 48 8 18 43 41

S 39 40 40 40 41 48 50 54 62 63 67 69 70 71 70 70

Cii,v 22v 19v 18ii 19v 19v 17ii 15ii 16ii 15ii 15ii 14ii 12ii 12ii 11ii 11ii 10ii

B 22 8.0 7.7 8.0 7.7 7.0 7.1 5.7 2.5 2.2 1.6 1.1 1.1 1.3 1.1 0.9

C 0.0 8.0 8.0 8.0 7.7 7.0 7.1 5.7 2.5 2.2 1.6 1.0 1.0 1.0 0.3 0.3

W 17 25 24 25 22 21 20 18 18 17 15 16 16 16 17 18

F (%) 0.0 0.0 2.1 0.0 2.2 0.2 0.3 0.5 0.3 0.4 0.6 0.4 0.4 0.4 0.4 0.3

ζTurk 1.30 1.86 1.77 1.85 1.68 2.21 2.39 2.62 3.90 4.08 4.78 5.65 6.13 6.69 6.78 7.37

E ------437 375 1,523 486 46,000 51,000 14,000 10,000 12,000 15,500 23,000 15,500 21,000 11,200 14,100

µTurk 1.00 0.56 1.01 0.56 1.35 1.60 2.87 1.58 2.39 3.19 3.33 4.08 2.21 3.70 1.39 1.33

𝜁𝑇𝑢𝑟𝑘 = (𝑊𝑊𝑎𝑡𝑒𝑟 + 𝑊𝑆𝑎𝑙𝑡 )/(𝑊𝐶𝑙𝑎𝑦 + 𝑊𝐵𝑒𝑛𝑡𝑜𝑛𝑖𝑡𝑒+ 𝑊𝐶𝑒𝑚𝑒𝑛𝑡 +𝑊𝐹𝑖𝑏𝑒𝑟𝑠 )

(1)

𝜇𝑇𝑢𝑟𝑘,𝑖 = (𝜎𝑖 ⁄𝜎𝜎10,95713 ) × (𝜁𝑇𝑢𝑟𝑘,𝑖 ⁄𝜁𝑇𝑢𝑟𝑘,1 ) , 𝑖 = 1, 2 ,3 … … … ,11

(2)

573

Earth and Space 2022

TSC: Materials Ratio

10.000

Wsalt Wb μi=σi/σ1 ×ζi/ζ1

Wcement Wwater Linear (Wsalt)

(%)

1.000 0

10

20

30

40

50

60

0.100

0.010

TSC Samples

Figure 1. Index of salty resistance vs material percentage, KTNSE (2019). Interpretation of TSC Samples In Figure 1 the materials used in the mix such as salt, water, cement, clay, bentonite and fibers are defined. It shows the proportion of each material in the mix in terms of weight of materials which it based on the ranking of low to high values of μTurk (salt resistance index). The curve obtained for μTurk shows an increasing trend, rising up from 0.15 to 4.08 for samples. While, figure ζTurk presents a similar concurrent ascending curve. Other curves for various materials an adverse μTurk curve behavior. In other words all material curves show a descending trend while ζTurk and μTurk show an increasing trend. This is a perfect mixture design. It means that the salt water merges into the mixture design forcing all material to go downwards. Figure 1 indicates that μTurk adapts the same as ζTurk behavior. According to Table 1, S970216P (i=3) contains values such as Wsalt:40%, ζTurk: 1.77, 𝜎(10days): 47kg/cm², μTurk:1.01 that compares to S980729P(i=11) with values WSalt: 67%, ζTurk: 4.78, 𝜎(10days): 58kg/cm² and 𝜇Turk:3.33, Turk(2018a). Comparison values are set as ΔWSalt: +27%, Δζ: +3.01%, Δ𝜎: +11% and Δ𝜇: +2.32 respectively which show the growth of resistance. If ΔμTurk is a positive value, the resistance of the mixture design will be modified, in other words TSC is capable of absorbing a higher percentage of salt, Equation 3. ∆𝜇 𝑇 𝑖𝑗 = 𝜇 𝑇 𝑗 − 𝜇 𝑇 𝑖 = (𝜎𝑗 𝜁𝑗 )/(𝜎1 𝜁1 ) − (𝜎𝑖 𝜁𝑖 )/(𝜎1 𝜁1 ) , 𝑖 = 1, 2 ,3, … … 𝑗 … ,49

(3)

TSC-MODULUS OF ELASTICITY The present study has shown that the mechanical properties of the TSC improved in as such that the Step Stiffness Method (SSM) was applied to analyze TSC using hydraulic jack output data figure EI (stiffness) and figure EA/L (axial stiffness). Samples were broken across the vertical axis in order to obtain a Modulus of Elasticity using the maximum tangency line of the

© ASCE

574

Earth and Space 2022

σ-ϵ curves. The Modulus of Elasticity ETURK increased by μTurk and sample curing period was up to 270 days as seen in Equations 4 to 7. The Modulus of Elasticity f(EZ) depends on the matrix ETurk [3×t] in which aging (t), axial stress (σZ), radial stress (σR) and Z-Z strain (ϵZ) are incorporated so that when the load suddenly returns to zero the Modulus of Elasticity is recalculated, Turk(2001). Axial stiffness is denoted by figure K in equation7. The behavior of figure K can be interpreted by -SSM. Figure 2 to Figure 3, and Table 2 to Table 5 explain the equations f (EZ) and K respectively. Figure (σZ) represents the exerted load in a Z-Z direction. Figures E and K contain the variables of salt percentage and sample curing period (t). 𝐸𝑍 = (∆𝜎⁄∆𝜖) = (𝜎𝑌 − 𝜎0 )⁄(𝜖𝑌 − 𝜖0 )

(4)

𝑓(𝐸𝑍 ) = 𝐸𝑇𝑢𝑟𝑘 (𝑡, 𝜎𝑧 , 𝜎𝑅 , 𝜖𝑧 ) = 𝑓[3 × 𝑡𝑖 ] , 𝑡𝑖 = 1, . . , 40, … , 270 𝑑𝑎𝑦𝑠 1𝑑𝑎𝑦 𝐸𝑇𝑢𝑟𝑘 (𝑡, 𝜎𝑧 , 𝜎𝑅 , 𝜖𝑧 ) = [ 𝜎𝑧1 𝜎𝑅1

2𝑑𝑎𝑦 𝜎𝑧2 𝜎𝑅2

𝑡𝑖 𝜎𝑧𝑖 𝜎𝑅𝑖

200𝑑𝑎𝑦 𝜎𝑧200 𝜎𝑅200

𝑡𝑛 𝜎𝑧𝑛 ] 𝜎𝑅𝑛

𝐾 = (𝜎𝑍 . 𝐴)/(𝜖𝑍 . 𝐿) = 𝐸𝐴/𝐿 25:2 σ(kg/cm²)

σ (kg/cm2)

60

σ(kg/cm²) 34: P12 σ

40

39: 20 σ 20

44RTA2:7-34 σ

0

45:34 σ 0

0.01

0.02

0.03

Figure 2. Sample RTA2-44-S980722P: 𝜇=3.19, ζ=4.08, KTNSE (2019).

Stress -Strain

43: 27 σ

σ (kg/cm2)

60

TA3: P5 σ(kg/cm²) 33: P5 σ(kg/cm²)

40

42: 27 σ 20

38: 13 σ 49: P30 σ

0 0

0.01

ε (%) 0.02

0.03

Figure 3. Sample TA3 (σR=1MPA) and Unloading 49S980729P30: 𝜇=3.33, ζ=4.78.

© ASCE

(6) (7)

Stress -Strain

ε (%)

(5)

575

Earth and Space 2022

Table 2. Figure E and Figure K of Tri-Axial and Axial: 𝜇=2.87, ζ=2.39, S980627P. No. Days E(kg/cm²) K (t/cm)

2 7 10,909 32,452

5 11 17,857 50,829

18 21 16,464 24,238

28 33 12,500 18,402

46 60 4,472 13,577

Table 3. Figure E and Figure K of Tri-Axial and Axial: 𝜇=2.39, ζ=3.90, S980720P. No. Days E(kg/cm²) K (t/cm)

23 1 300 702

26 4 2,727 5,797

32 9 6,667 20,423

35 14 7,738 23,254

47 36 10,145 34,230

Table 4. Figure E and Figure K of Tri-Axial and Axial: 𝜇=3.19, ζ=4.08 S980722P. No. Days E kg/cm2 K (t/cm)

25 2 3,690 10,753

TA2 7 8,065 25,117

34 12 11,780 26,126

39 20 9,400 20,317

44RTA2 34 6,494 21,776

45 34 4,032 19,386

51 37 5,208 15,063

Table 5. Figure E and Figure K of Tri-Axial and Axial: 𝜇=3.33, ζ=4.78, S980729P. No. Days E(kg/cm²) K (t/cm)

TA3 5 6,265 19,279

33 5 7,442 16,649

38 13 12,162 26,067

42RTA3 27 6,494 20,307

43 27 11,993 29,696

49 30 15,493 44,807

Comparison of Modulus of Elasticity in the Axial and Tri Axial Test Figures 2 and Figure 3 show the stress- strain curves as S980722P and S980729P through which the Modulus of Elasticity is obtained through the tangency of the σ-ϵ curve. It is common for stress-strain slopes to be extended over a wide angle which in itself is used as a measure for determining a modification in TSC strength behavior. A successful behavior of axial load curves. All curves have a tendency toward the maximum slope angle. In Figure 4, TA2-S980722P7 underwent a Tri-Axial test over a 7day curing period. Stress and strain were observed to have improved significantly through the application of pressure in the loading curve. The Modulus of Elasticity of TA2-S980722P7 had increased up to 8͵065 (kg/cm²) over the 7day curing period at 𝜎Z=71.5 (kg/cm²). Tri-Axial pressure was exerted to consolidate the sample utilizing a 1MPA oil hydraulic jack. It is of note that TA2-S980722P7 loading continued until the first crack appeared as seen in Figure 4. Axial load is exerted again to interpret behavior of the TA2S980722P7 after 34day=7day+27day curing period. The cracked sample 44-RTA2S980722P734 is represented by the rising resistance 𝜎Z=54.9 (kg/cm²) which indicates an improvement. Furthermore, as seen in Figure 5, TA3-S980729P5 which was put under pressure over a 5day curing process shows an increased Modulus of Elasticity up to 6͵265 (kg/cm²) in which 𝜎Z=66.5

© ASCE

576

Earth and Space 2022

(kg/cm²). TA3 was loaded until the first crack appeared as seen in Figures 3 and Figure 5. Axial load was then once more exerted to interpret the behavior of TA3-S980729P5 over a 27day=5day+22day curing period. The cracked sample TA3 was reloaded again. Sample 42RTA3S980729P5-27 represents the final resistance 𝜎Z=32.4(kg/cm²).

82

Figure 4. TA2: II and Retest 44-RTA2-II-S980722P7-34 after ultimate loading. 83

82

Figure 5. Retest 42-RTA3S980729P5-27 following immersion. The fact that TSC must be put under pressure in order for the values of E Z and K to increase during the pre-consolidation stage can be observed in Equations 5 and Equation 6. This stage is highly recommended as a means of modifying grout injections to prevent landslides and

© ASCE

Earth and Space 2022

coastline erosion. Cracks in TSC are rehabilitated through this means of self-repair. This was clearly seen in both TA2S980722P7 and TA3S980729P5 which after the initial Tri Axial pressure test over the seven day period, showed a fusing of material along the cracks after further pressure was applied over a longer period, Figures 4 Figure 5. TSC CRACK BEHAVIOR AND BROKEN PLANE STRESS Figure 6 shows broken samples upon which cracks are distributed over the exterior. Tensile stress or hook stress interact along the radial plane stress. Figure 6 demonstrates the effects of the radial stress and horizontal stress 𝜎R which is spread along the cylindrical skin. TSC deformation is similar to the destruction of a thin-walled cylinder. Strain 𝜖R enlarges into the plastic zone without being damaged. In Figure 6, skin cracks are detailed using samples 44RTA2S980722P734, 43S980729P27 and 64S980729P78 respectively, which are indicative of hook stress via the waste plastic fiber. These cracks expand from the point of failure that is from the external radius to the center. TSC often breaks under the axial in the σ - ϵ plane, Beer (2012).

Figure 6. Plastic fiber behavior at a ratio of 83% Salt Water in cracking spread. Modulus of Toughness (MT) in the Axial and Tri Axial Test TSC maximum resistance is obtained through the values obtained in the Tri Axial test. Samples are put under pressure to achieve greater strength. MT values are depended on the Tri Axial Test and μ Turk. Figure 7 demonstrate that MT behavior has a correlation relationship with the values of 𝜎max and 𝜖Z. In 2018, TSC was created using the ζTurk (0.50 to 1.50). In Table 6, MT increases with 𝜎max, 𝜖Z and the curing period. In Figure 1, waste plastic fibers (F) are added to increase TSC strength and ζTurk, Turk (2018b). The final laboratory results indicate an appropriate modulus of toughness which is increased X6 times when compared with non-fiber samples. Moreover, the axial resistance was reported as being σZ=147kg/cm2 after a 28day curing period.

© ASCE

577

578

Earth and Space 2022

Table 6. MT of S980729P, 𝜇=3.33, ζ=4.78, Beer (2012), and Turk (2018a, 2018b). No. 𝜖R % 𝜎max MT

TA3(5day) 4.995 66.5 2.15

33(5d) 0.754 34.8 0.17

38(13d) 0.723 49.6 0.22

42RTA3(27d) 0.846 32.4 0.33

𝜖𝑅

43(27day) 1.009 68.8 0.42

49(30day) 0.873 53.3 0.31

𝜖𝑅

𝑈𝑇𝑟𝑖𝐴𝑥𝑖𝑎𝑙 = ∫ 𝜎𝑅𝑢𝑝𝑡𝑢𝑟𝑒 . 𝑑𝜖𝑍 ; 𝑈𝐴𝑥𝑖𝑎𝑙𝑇𝑒𝑠𝑡 = ∫ 𝜎𝑅𝑢𝑝𝑡𝑢𝑟𝑒 . 𝑑𝜖𝑍 0

0

2

1.5

ϵRi/ϵR45

S980722P

1.730769231

σmax,i/σmax45

1.418650794

MTi/MT45

1.21196868

1

1 0.5 0 2day

7day

12day

20day

34day

34day

37day

25

TA2

34

39

44RTA2

45

51

Figure 7. MTTA2 /MT45 =1.73, σmax,TA2 /σmax,45 =1.41 and ϵR,TA2 /ϵR,45 =1.21 ratios. SSM GEOTECHNICAL ASPECTS The step stiffness method (SSM) interprets driving pile behavior at depth using the ASTM1143 static load test, Chellis (1951). In the loading stage (Figure 8) the curve develops a convex shape and in the unloading step, the curve output appears as a multi-linear concave. These linear slopes define the axial stiffness of zones, I, II and III. The Micro pile unloading curve is divided into three zones, Turk (2004) and Ulrich (2004). Zone I is defined through the axial stiffness of soil grouting (Table 7). The modulus of Elasticity is shown in Table 7 and Table 8 respectively and indicates the values obtained by SSM analysis in the unloading phase, Turk (2001, 2007 and 2008). Table 9 shows figure KZoneIII for tensile and compressive micro- pile load tests. Figure EZoneI further defines the modules of elasticity in zone I. Figure K ZoneII refers to soil axial stiffness which is shown in Table 9 and Figure 8. Equation 9 determines the ratio of ETSC/EZoneI ≥ 1/2 =50% and KTSC/KZoneII ≥ 3/5 =60%. It is believed that ratio E TSC/EZoneI = 26/47= 0.55 can determine the grouting requirements in salt layers, Equation 8. TSC Properties nearly parallel the values for micro pile grouting behavior in soils.

© ASCE

579

Earth and Space 2022

100

KQ05

ASTM 1143: Zone II (Soil Axial Stiffness)

80

KQ04 KQ09

Load (tons)

60

Zone III KQ15

40

TB1

zone II

20

TB2

Zone I

0

5 δ (mm) 10

0

15

20

25

KO

Figure 8. Driving Precast Concrete Pile, 35×35cm2×32m, Turk (2001, 2008). 𝐸𝑇𝑆𝐶 ⁄𝐸𝑍𝑜𝑛𝑒𝐼 = 26 ≅ 12 ; 𝐾𝑇𝑆𝐶 ⁄𝐾𝑍𝑜𝑛𝑒𝐼𝐼 = 44 ≅ 35 47 65

(8)

Table 7. Micro pile Load Test, KZoneIII (t/cm), EZoneI ×103 (kg/cm²), Turk (2004). i KZoneIII EZoneI

7A2 -35 47

49L12 T79 -31 -26 63 68

F194 -29 109

49L11 T80 +23 +12 74 118

T33 +23 124

T2 +22 157

F193 +23 178

Table 8. TSC Samples KTSC=EA/L (t/cm), E×103 (kg/cm²), KTNSE (2019). No. KTSC ETSC

X1:40C 63 26

12:12C 26 16

14:14C 31 15

TA1:13C 44 15

37RT32 27 9

40:32 38 13

30:14CF 16 8

Table 9. Precast Concrete Pile Load Test, KZoneII = KSoil (t/cm), Turk (2001). No. KQ9 KZoneIII 134 KZoneII 11

KQ5 163 16

TB2 118 24

KO 114 25

KQ15 137 28

TB1 118 30

KQ4 243 65

The ratio of ETSC/EZoneI in Equation 8, Table 7 and Table 8 indicate that injection in salt layers should be at values of at least 50%. That is to say, injection is a very viable option when dealing with pure salt layers. K TSC/KZoneII also indicates that the injection should be at maximum pressure to increase the radius of influence. INTERFACE GROUTING NEAR SHORELINES Saline water intrusion often occurs near river estuaries when the river outflow fails to prevents a backflow due to the head water potential (h) as seen in Figure 9. Equilibrium in the

© ASCE

580

Earth and Space 2022

interface plane produces Equation 9 in the normal boundary condition, USGS (2003). If TSC grouting values are added to the equilibrium of Equation 11 then the coefficient β will be the parameter of stability for interface grouting in Equation 12. Equation 12 and Equation 13 introduce the function (h/Z) versus β, γf and γs. Figure 9 to Figure 13 demonstrate the grouting values of f(h/Z) vs the normal condition where Z=40m changes into 30m, 20m and 10m. This in itself means that Figure h can decrease from 100cm to 7.8cm when β=0.15 is applied for Z=40m.

0.0 ≤ β.Z ≤ 0.15Z

Figure 9. Grouting operating in the interface at less than β= 0.15, USGS (2003). 𝛽=

𝐿𝑔𝑟𝑜𝑢𝑡𝑖𝑛𝑔 ⁄ , 𝐿𝑔𝑟𝑜𝑢𝑡𝑖𝑛𝑔 ≪ 𝛽𝑍, 𝑍

0%, 5%, 10%, 12%, 15% 𝐻𝑜𝑟𝑖𝑧𝑜𝑛𝑡𝑎𝑙

∑𝑧2=ℎ+𝑍 𝑓𝑋 = 0 𝑧1=0

⇒ 1

(9)

1

(ℎ + 𝑍). 𝛾𝑓𝑟𝑒𝑠ℎ × 1 × 𝑍 = 𝑍 2 𝛾𝑆𝑎𝑙𝑡𝑦 2 2 1

1

(10)

(ℎ + 𝑍). 𝛾𝑓 × 1 × 𝑍 = 𝑍 2 𝛾𝑆 [(1 − 𝛽)𝛽 + (1 − 𝛽)2 ] 2 2

(11)

1 1 2 (ℎ + 𝑍)⁄ 𝛾𝑆 2 𝑍 = ⁄𝛾𝑓 (𝛽 − 𝛽 + 2 + 2 𝛽 − β)

(12)

ℎ 𝛾𝑠 = (1 − 𝛽 2 ) − 1 𝑍 𝛾𝑓 ℎ

f (𝑍) h⁄ = 0.025 𝑍

𝛾

𝑇𝑢𝑟𝑘

= 𝛾𝑠 (1 − 𝛽 2 ) − 1 𝑓

𝑍=40𝑚

and β = 0.0 ⇒

ℎ = 1.0 𝑚

h⁄ = 0.01024 and β = 0.15 𝑍=40𝑚 ⇒ ℎ = 0.078 𝑚 𝑍 h⁄ = 0.01475 and β = 0.10 ℎ=20𝑚 ⇒ ℎ = 0.30 𝑚 𝑍

© ASCE

(13)

581

Earth and Space 2022

1.2 Z=40m, h=1m for β=0

f (h/Z)Turk

1

Water Table: h (m)

Z=40m, h=0.078m, for β=0.15

0.8

β=0

0.6

β=5 Z=20m, h=0.3m for β=0.10

0.4

β=10 β=12

0.2

β=15

0 40

30 20 interface water depth: Z (m)

10

Figure 10. Comparison of function f(h/Z) values with coefficient β, USGS (2003). Equation 9 to Equation 13 conclude that TSC grouting can decrease the Z depth to βZ.

β≤ 0.15

Figure 11. Zone of diffusion isolated by TSC injection, Lenntech (2021).

© ASCE

Earth and Space 2022

Figure 12. Salt Dome shoreline of GOTVAND reservoir will injected by TSC.

Figure 13. TSC isolation of saline water drains along KARUN River. SEM IMAGES Figure 14 and Figure 15 represent the SEM images of 46-S980627P60 with 0.27% plastic fiber (P), ζTurk=2.39 and a 60 day curing period. Pictures taken to study the micro behaviors of

© ASCE

582

Earth and Space 2022

TSC adhesiveness are presented as Figure 14 which shows waste plastic fiber details at a scale of 20μm and Zone Mag= 500 X, CENLAB (2019). It appears that the plastic fibers in the TSC amalgamate with the salt crystals thus creating an entity similar to a salt mummy, the literature believe that this process might be similar to the embalming procedure used in ancient societies, Waltari (1945) and Saltmen (1993). Figure 15 shows details of salt and fiber adhesion (crystallization). The fibers merge into salt concrete in as such that the materials cannot be distinguished. Figure 15 shows a 10μm scale plastic fiber encased in salt and completely merged into it. Figure 16 also demonstrates GRP plastic fiber from another angle of sample 55-980909P2, with 0.37% plastic fiber (P), ζTurk=5.11, 𝜇Turk=3.38 and a 2day curing period, CENLAB (2020). Figure 16 which shows GRP fiber details at a scale of 20μm and Zone Mag= 500 X. The roughness of the fiber surface is filled with salt concrete, and presents a feature as if salt has penetrated into the crushed fiber edges and created a uniform mixture. It is recommended that the plastic fiber used for TSC have a greater roughness and cutting edge to increase the adhesion between TSC and fibers (F). Plastic Fiber reduces friction during the mixing process. Hence the percentage of water contained is decreased, and more salt can be dissolved in the mixture.

Figure 14. Plastic fibers are surrounded by 46-S980627P60 materials of TSC, CENLAB (2019).

© ASCE

583

Earth and Space 2022

Figure 15. Mummify of plastic fibers, 46-P60, 𝜇=2.87, ζ=2.39, CENLAB (2019).

Figure 16. TSC covers Fibers in 55-980909P2 and Mummy-like process.

© ASCE

584

Earth and Space 2022

CONCLUSION Figure 1 is describes μTurk and ζTurk which are both interpretations of the TSC behavior. The salt ratio to other materials such as cement type (v, ii), bentonite and SCC increases. At the beginning of the current study, ζTurk was approximately at a ratio of 0.80; however this ratio increases up to 4.1, 4.80 and 5.65 in 2019 and 5.67, 6.13 to 6.69 in 2 020. These values are very significant and indicate the success of the TSC mixture design. In fact, more than 83% (to 87%) of the ingredients of TSC come from salt water whatsmore with the increase in salinity, the compressive strength of the TSC also increases, Figure 2, Figure 3, and Figure 6. The modulus of elasticity behavior is obtained through Figure 2 to Figure 3. Figure E is calculated as per Table 1 to Table 5 and Table 8. The modulus of elasticity based on its salt ratio increases. Tri-axial test pressure shows that the higher compressibility of samples greatly improves the elasticity and other properties of the TSC. It is recommended in along coastlines high-pressure grout injection be performed in several stages from low to high (1atm to 5atm); thus, TSC can be modified in shallower depths through the grouting curtains. This method can increase grout injection efficiency. TSC has the ability to repair itself after an initial crack, and will regain its initial strength after a few days. Cracks are repaired by the TSC after sliding. This fact is illustrated by Figure 2, (44RTA2, 7-34), Figure 3, (42RTA3S980729P7-27), Figure 4, Figure 5, Figure 7, Table 4, (44RTA2) and Table 8. This property is very useful in shoreline stabilization, land sliding, beach erosion and coastal protection structures. Only TSC can isolate submerged layers near beaches, along coastlines, for salt ridges, GHACHSARAN Formations (SW of IRAN) and permeable industrial and agricultural drainage canals. The total cost of TSC injection near the shoreline is less than USD $3.5 per square meter, which is an attractive economic factor for the financing of protection of freshwater aquifers. SEM images were used to show the extent of adhesion in TSC at higher resolutions. The TSC samples are remarkably mixed and difficult to separate. Plastic fiber is encased by the salt which merges with it. Salt crystallization shown through SEM imaging is indicative of the uniform arrangement of the ions. Based on the study and behavior of the laboratory specimens as observed in the SEM images, it can be stated that TSC parallels an ancient method of mummification which applied in Sub-Sahara and parts of the Middle East, Zoroastrian Salt Columbarium, Waltari (1945) and Saltmen (1993). It is proposed that TSC research should be further advanced to improve the properties of materials in Iran in order to develop an optimal mixture for TSC. REFERENCES Beer, F. P. E., Johnston, R., Dewolf, Jr. T., Mazurek, D. F., (2012). “Mechanic of Materials.” The McGraw-Hill Companies, Inc., 6th edition. CENLAB (2019). “SEM-Scanning Electronic Microscopic Images, TSFC Samples.” Act of 2019. Pub L. No. 3CSRDFST.” SHAHID CHAMRAN University of Ahvaz, Central Laboratory, LEO 417371395665VP. 98/CEN0328 IR.SCU. CENLAB (2020). “SEM-Scanning Electronic Microscopic Images, TSFC Samples.” Act of 2020. Pub L. No. 3CSRDFST.” SHAHID CHAMRAN University of Ahvaz, Central Laboratory, LEO 165413990721VP. 99/CEN0323 IR.SCU. Chegini, V., (2011). “Fundamental Design of Coastline Structures.” Oceanography Center of Iran, 1st edition, Iran.

© ASCE

585

Earth and Space 2022

Chellis, D. R., (1951). “Pile foundation.” McGraw-Hill Companies, Inc., 2nd edition, Tokyo, Japan. KTNSE, Soil Concrete Laboratory (2016), “SE Water Supply, Act of 2016. Pub L. No. 951027.” KWPA. 95/0119-K9631 IR.SE. KTNSE, Soil Concrete Laboratory (2017), “SE Water Supply, Act of 2016. Pub L. No. 961027.” KWPA. 96/0708 IR.SE. KTNSE, Soil Concrete Laboratory (2018), “SE Water Supply, Act of 2017. Pub L. No. 970217.” KWPA. 1200/126238-K9631 IR.SE. KTNSE, Soil Concrete Laboratory (2019), “SE Water Supply, Act of 2019. Pub L. No. 980714.” KWPA. 5200/626 IR.SE. Lenntech (2020). “Seawater intrusions in groundwater” < https://www.lenntech.com/ groundwater/seawater-intrusions.htm> Saltmen (1993). “Iran National Museum”. < https://en.wikipedia.org/wiki/Saltmen> Thompson, P.; Penning-Rowsell, E. C.; Parker, D. J; and Hill, M. I., (1987). “Interim Guidelines for Economic of Coast Protection and Sea Defense Schemes”, Flood Hazard Res. Center, Middlesex Polytechnic. Turk, A., (2001). “Driving Force by Kpile Method and Pile Loading Test, Concrete Piles”, CSCE, Canada. Turk, A. and Zaemeri, A. A., (2004). “Micro pile behaviors study using compressive (or tensile) pile load test and step stiffness method”, CSCE, Canada. Turk, A., Salehi, F., Kolahchi, A., Ghanavatizadeh, S., (2007). “Elasticity Module of Karkheh Earth Dam Study & Behavior through Inclinometers & SSM”, Proc., Int. ISEC04, Melbourne, Australia. Turk, A., Ghanavatizadeh, S., Zaamari, A. A., Kolahchi, A., (2008). “Regeneration of Missed Record Data, Vertical Axes of Karkheh Earth Dam Using Cell Pressure, Mathematical Aspects and SSM”, ASCE Aerospace Division, USA. Turk, A., Mombeni, B., Bahmaei, D., Khodabakhshi, H. R., Behdarvandi, A. M., and Ghanavatizadeh, S., (2017). “Coastal Protection of GOTVAND Reservoir DAM, EC Perfection of Salty Domes.” Proc., Int. ISEC-09, Valencia, Spain, 10.14455/ISEC.res.2017.61. Turk, A., (2018). “Stabilization of Pure Salty Formations of the GOTVAND Dam Lake and another Salty Drought Desert Regions through Invention of Turk Salty Mortar”, ASCE Aerospace Division, USA. Turk, A., Mombeni, B., Afrasiabi, S., and Ghanavatizadeh, S., (2018). “Turk Salty Mortar through Addition of Plastic and Cotton Waste Materials to Stabilize Problematic Coastlines.” ISEC Press, Australia, 10.14455/ISEC.res.2018.135. USGS (2003). “Salt water intrusion”

Ulrich, S., (2004). “Geotechnical Engineering Handbook.” Ernest & Son, A Wiley Company, Germany, Volume 3. Waltari, M., (1945) “Sinuhe Egyptiläinen”. WSOY Press, Finland.

© ASCE

586

587

Earth and Space 2022

Risk-Based Structural Optimization Framework for Connected Structural System Subjected to Extreme Events William Hughes1 and Wei Zhang2 1

Dept. of Civil and Environmental Engineering, Univ. of Connecticut, Storrs, CT. Email: [email protected] 2 Dept. of Civil and Environmental Engineering, Univ. of Connecticut, Storrs, CT. Email: [email protected] ABSTRACT Connected structural systems, such as overhead power distribution systems, could fail to function normally under extreme events or environments when the typically anticipated loadings are exceeded. Consequently, in structural designs, the tradeoffs between costs and accepted risks under such extreme loading scenarios need consideration. To this end, structural optimization methods could be applied to minimize costs subject to constraints on the design feasibility. While classical structural optimization techniques are deterministic, risk-based approaches considering the probabilities of structural failure under certain contingencies have been less explored. In the present study, a structural optimization framework is developed and tested on a power distribution system subject to extreme storms. Uncertainties in the material properties, loadings, and construction are integrated for a risk-based assessment with constraints on the acceptable pole failure probability. Various combinations of extreme storm scenarios and acceptable failure probabilities are investigated to explore the risk-cost tradeoff and optimal design criteria. By employing the use of high-class, shorter poles, the optimal configurations are found to reduce the cost for a similar reliability level by about 50%, while, for a similar cost, the system could be designed to a much higher reliability level. Such a framework could be adopted for a variety of structure types and systems under various extreme environments, including future habitats on the Moon or Mars. 1. INTRODUCTION When stressed under extreme events or environments beyond their typically anticipated daily loadings, structural systems could fail to function as designed. In connected systems, failure of individual components could trigger the loss of functionality of large portions of the system through cascading effects. As some failure risk always exists for typical structure components or systems to avoid prohibitive costs, a cost-benefit tradeoff plays an important role in the structural design process. Otherwise, networks could become over-dimensioned, leading to avoidable and unnecessarily large economic losses. To this end, the design of structures and systems could be optimized to minimize costs subject to constraints on the design feasibility and performance under extreme events. To improve the network resilience, the optimization must consider both the system as a whole and each component individually to ensure the system does not contain any weak links. Classical structural optimization methods treat the material properties and loadings as deterministic, neglecting variations and imperfections. However, risk-based approaches, which can incorporate accepted failure probabilities under certain contingencies, have been less

© ASCE

588

Earth and Space 2022

explored despite the practical benefits. Presently, a risk-based structural optimization framework is developed for connected structural systems as outlined in Figure 1. First, the structural model is defined, including topology, material properties, and forces, which could be treated as stochastic based on their uncertainties or return period. Following, the constraints are defined, such as on the serviceability, manufacturability, or strength of the structures alongside any budgetary confinements. By defining an acceptable risk of exceeding the defined structural constraints, the risk assessment could be conducted through methods such as Monte Carlo simulation. With the objectives and constraints defined, the problem can be solved, typically through heuristic methods, such as genetic algorithms, provided the complexity of the objective and constraint functions. Finally, design improvements can be explored based on the optimization results. In particular, the tradeoffs of designing for more extreme events or accepting a higher or lower risk could be directly analyzed. Such a framework could be adopted for a variety of structure types and hazards in extreme natural environments.

Figure 1. Outline of the proposed framework Presently, the framework is tested on a case study of power distribution systems (PDSs) subject to extreme storms. The uncertainties in the material properties, loadings, and construction are integrated for a risk-based assessment with constraints on the accepted pole failure probability, extending existing research by allowing system reliability-based optimization. The optimization problem for the automated PDS structural design is developed as a flexible framework that is alterable to account for specific reliability levels and loading conditions. For demonstration, the framework is evaluated for an existing distribution line in the state of Connecticut. To explore the risk-cost tradeoff and optimal design criteria to improve PDS resilience, various combinations of extreme storm scenarios and acceptable failure probabilities are defined. The results provide new insights into the design of such connected structural systems optimal designs under various storm severities. The remainder of the paper is organized as follows. First, a brief background of the power distribution system and relevant literature is discussed in Section 2. Following, the methodology specific to the power distribution system is outlined, including the objective function, constraints, and structural model. In Section 3, the methodology is then conducted on a network in the state of Connecticut. Results are then presented in Section 4, accompanied by a discussion of the findings. Lastly, in Section 5, the work is summarized with concluding remarks and notes of the limitations and future directions.

© ASCE

Earth and Space 2022

2. OPTIMIZATION METHODOLOGY 2.1 Power Distribution System Background The power distribution system (PDS), comprised of overhead timber utility poles and conductors, serves as a critical component in the efficient electricity delivery relied upon by residents and businesses for public functioning, safety, and well-being. However, under severe weather events, such as tornadoes, downbursts, or hurricanes, structural failures of the distribution poles or lines can produce cascading area-wide power outages. In addition to the societal impacts of extended outages, storm-induced power outages have immense economic impacts. With the substantial utility company PDS investment, improved system hardening and planning contain the lucrative promise of reducing power delivery interruptions. To this end, reliability analysis has been applied to analyze the PDS risk to better inform proper design, often through the development of fragility curves (Braik et al. 2019; Salman and Li 2016). However, classic network design approaches typically employ the determination of pole class, height, and span length which satisfy the code-specified minimum design loads, without substantial consideration of various design alternatives and their associated tradeoffs or probabilistic system risk assessment. Nevertheless, to date, the risk-based structural design optimization considering the specific sizing and spacing of power distribution poles has not received significant attention. Much research effort on the optimization of the PDS and power transmission system (PTS) design has been applied toward the planning of the power system from the standpoint of optimizing the number and placement of the feeders, switches, conductor sets, or distributed generation (DG) units (Kishore and Singal 2014). Meanwhile, the optimal placement of dampers to reduce aeolian vibrations has also been explored (Rezaei and Sadeghi 2019). The structural size, shape, and topology optimization of the lattice transmission towers have received more focus (de Souza et al. 2016; Teegala and Singal 2016), but often focus only on the design of single towers rather than the entire system (Raghavendra 2012). The optimization of the PSD utility poles based on reliability and structural performance constraints has received limited attention. Cicconi et al. (2019, 2020) developed a design approach for the automated design of overhead lines and later applied a constraint-based approach based on the dimensions, pole locations, and cable pre-tension. However, the design variables were considered deterministic, ignoring the randomness and inevitable imperfections in the design. Palheres and Brito (2020) developed a stochastic optimization scheme, treating the wind speed as random. Nevertheless, the risk-based optimization incorporating uncertainties in the loading and imperfections in the construction and material properties has not been studied. Presently, the risk-based optimization problem for the automated structural design of the PDS is developed to cost-effectively improve system resilience. For a demonstrative case study, the framework is evaluated on a PDS in the state of Connecticut. 2.2 Formulation The problem is formulated as an integer-constrained nonlinear programming problem. The objective is the minimization of the monetary cost function C which is considered as the sum of the pole and conductor installation and materials costs:

© ASCE

589

590

Earth and Space 2022

𝑖=𝑛

𝑗=𝑚

(1)

𝐶(𝑫𝒑 , 𝒉𝒑 , 𝐷𝑐 ) = ∑ 𝑐𝑝,𝑖 + ∑ 𝑐𝑐 𝑙𝑗 𝑖=1

𝑗=1

where Dp and hp are the vectors of pole diameters and heights, respectively, and Dc is the conductor diameter. cp,i is the cost of the ith pole out of n total poles in the system, cc is the conductor cost per unit length, and lj is the length of the jth span out of m total. The diameters and heights are constrained to meet typical manufacturing standards, such as certain pole classes. The system reliability is conducted assuming the poles form a series: 𝑛 (2) 𝑃𝑓,𝑠𝑦𝑠 = 1 − ∏(1 − 𝑃𝑓.𝑖 ) 𝑖=1

where Pf,sys is the system pole failure probability and Pf,i is the failure probability of the ith pole. A structural failure is considered if the pole or conductor stresses 𝜎 exceed their resistances R. (3) 𝜎−𝑅 ≤0 More specifically, for a pole, the failure probability Pf of a given pole could be determined as: (4) 𝑃𝑓 = 𝑃(𝜎𝑝 − 𝑅𝑝 ≤ 0) where 𝜎𝑝 is the groundline stress and Rp is the pole resistance or fiber strength. Therefore, a constraint is applied to ensure the pole designs meet a threshold acceptable probability failure or reliability level rp, expressed as an allowed exceedance percentage deemed acceptable: (5) 𝑃𝑓,𝑠𝑦𝑠 − 𝑟𝑝 ≤ 0 As the conductor material properties are treated as deterministic given the negligible impact of considering their stochasticity (Ma et al. 2021), the design is controlled by the worst case: (6) max (𝝈𝒄 ) − 𝑅𝑐 ≤ 0 where 𝝈𝒄 is the vector of conductor stresses and Rc is the conductor resistance. An additional constraint is added to ensure the conductor sagged height does not exceed its above-ground clearance c restrictions: (7) 𝑠 − min(𝒉𝒄 − 𝜹𝒄 ) ≤ 0 where hc is the vector of design (pre-sag) conductor heights for all conductors, δc is the vector of maximum conductor deflections, and s is the minimum allowable code-specified aboveground clearance. Finally, the maximum span length is limited to 200 m (Cicconi et al. 2020): (8) max(𝒍) − 200 ≤ 0 where l is the vector of span lengths. Therefore, the complete problem is written as: (9) minimize 𝐶(𝑫𝒑 , 𝒉𝒑 , 𝐷𝑐 ) Subject to Equations (5)-(8) 2.3 Wind and Ice Forces The structural modeling of the pole-wire system with various uncertainties in the loadings and material properties has been commonly studied throughout literature (Ma et al. 2020; Salman and Li 2016). While the finite element method has been applied (He et al. 2017; Yuan et al. 2018), the optimization computational costs could become excessively prohibitive when

© ASCE

591

Earth and Space 2022

incorporating the stochasticity. Therefore, to expedite the optimization process, analytical equations based on structural mechanics are utilized. In the design of the overhead power distribution system, wind forces are typically the major consideration. The wind pressures on the poles and conductors are calculated as (Yuan et al. 2018): Pr = 0.613 V2kzGCf (10) where Pr is the wind pressure in N/m 2, V is the 3-s gust (m/s), G is the gust response factor, kz is the topographic factor, and Cf is the shape factor. The distributions of the coefficients are listed in Table 1. The wind speed could be treated as random such as by a Weibull distribution as in (Palhares and Brito 2020), or a specific load case be chosen for analysis as in the present study. Table 1. Pole Wind Load Coefficients (Yuan et al. 2018) Parameter Cf (pole) Cf (conductor) G (pole) G (conductor) kz (pole) kz (conductor)

Distribution Normal Normal Normal Normal Normal Normal

Mean 1 1 (no ice), 1.2 (with ice) 0.97 0.88 1 1.1

COV 0.12 0.12 0.11 0.11 0.06 0.06

During winter storms, accreted ice on conductors could have severe impacts, particularly when coupled with strong winds. The distributed ice on the cable is assumed cylindrical and calculated as (Ma et al. 2020): (11) 𝑤𝑖 = [(𝑟𝑖 + 𝑟𝑐 )2 − 𝑟𝑐 2 ]𝜋𝜌𝑔 where wi is the distributed ice load for a unit length of cable, ri is the ice radius, rc is the conductor radius, 𝜌 is the ice density (0.917 kg/m 3), and g is the gravitational acceleration (9.81 m/s2). The increased vertical load and projected area from the poles are considered negligible (Ma et al. 2020), and ice is therefore only modeled on the conductor. Further, the shape factor is affected by the presence of ice and is increased by 20% (Ma et al. 2020). 2.4 Conductor Modeling The conductor deflection could be modeled via the hyperbolic catenary equation. However, the parabolic sag curve serves as an appropriate approximation provided the span length is under 400 m (Abebe and Rao 2016; Hatibovic 2014). If the supporting poles on each end of the span have equal height, the greatest sag will occur at midspan and is calculated as (Cicconi et al. 2020): (12) 𝑤𝑙 2 𝛿= 8𝑇 where w is the self-weight per unit length and T is the horizontal component of the wire tension. In the case of uneven pole heights at each end of the span, with a height differential h, the maximum sag occurs at a point a along the horizontal length calculated as (Cicconi et al. 2020): 𝑙 𝑇ℎ (13) 𝑎= − 2 𝑤𝑙 and the sag at a is (Cicconi et al. 2020):

© ASCE

592

Earth and Space 2022

(14) 𝑤𝑎2 𝛿1 = 2𝑇 Figure 2 shows a drawing of an uneven distribution line considering the sag with the various terms defined.

Figure 2. Diagram of sagged distribution line The stress increase in the wire due to the ice and wind loading and temperature changes could be calculated from the state change equation (Cicconi et al. 2020; Keselman and Motlis 1998; Pinto 2012): (15) 𝑇𝑓 − 𝑇𝑖 𝑙 2 𝑐𝑜𝑠 2 𝜀 𝑤𝑓2 𝑤02 ( 2 − 2) − − 𝛼(𝜃𝑓 − 𝜃0 ) = 0 24 𝐸𝐴𝑐𝑜𝑠𝜀 𝑇𝑓 𝑇0 where ε is the inclination angle of the span, Tf and T0 represent the final and initial conductor tensions, respectively, E is the conductor elastic modulus, and A is the conductor crosssectional area. α is the thermal expansion coefficient, θf and θ0 are the final and initial temperatures, w0 is the specific load per unit length in the initial conditions (i.e., the conductor linear weight, wc), and wf is the specific load per unit length in the final loading condition, which can be expressed as the total linear action on the wire (Cicconi et al. 2020): (16) 𝑤𝑓 = √(𝑤𝑐 + 𝑤𝑖 )2 + 𝑤𝑤 2 where ww is the wind force per unit length of the conductor. Solving Equation (15) for Tf provides the final conductor tension considering the wind and ice loadings. Presently, the temperature effects are not considered. 2.5 Pole Failure As noted, the limit state is presently considered as pole flexural failure at the groundline (Yuan et al. 2018), where failures are most common (Salman et al. 2017), although the failure could occur higher on the pole, such as where holes are drilled for utility cables. The modulus of rupture is compared to the groundline bending stress, which is determined based on the established principles of bending theory, assuming fixed bottom connections. Each pole is assumed to carry its own tributary length of the conductors, half of each span length to either side, as defined in Figure 3. The worst-case scenario of transverse wind loading perpendicular to the conductors is considered. As the height is small enough to neglect the variation the vertical wind profile, the wind load on the poles and wires is treated as a uniformly applied (Yuan et al. 2018). The bending stresses

© ASCE

593

Earth and Space 2022

caused by the wind pressure on the poles and conductors create moments about the x-axis. Meanwhile, if the span lengths to the left and right are uneven, the downward weight of the conductors and accreted ice could produce a moment about the z-axis. Finally, the groundline stress σ could be expressed as: 𝑊𝑠 𝑴𝑥 𝑴𝑧 (17) 𝜎= + + 𝐴𝑝 𝑆 𝑆 where Ws is the self-weight of the poles, wires, and ice within the tributary area, Ap is the crosssectional area of the pole, Mx and Mz are the moments about the corresponding axes, and S is the section modulus of the pole, which for the circular cross-section is equal in both the x and y directions and could be expressed as: (18) 𝐷𝑝 3 𝑆= 𝜋 32

Figure 3. Definition of pole tributary area Uncertainties in the loading and material and geometric properties are accounted for, such as via Monte Carlo simulation, to obtain the failure probabilities under a given loading condition. The specific material properties could differ for various design options such as the wood and conductor types. 2.6 Genetic Algorithm For problems with complex objective functions or constraints, meta-heuristic methods such as the genetic algorithm (GA) could be adopted. The GA (Holland 1975) is a powerful stochastic optimization technique inspired by the concepts of Darwinian natural evolution. GAs work with sets of potential solutions called populations, providing a computationally efficient algorithm that can run in parallel and requires no gradient information in the search process. Due to their consequent flexibility in implementation and the broad range of applicability, GAs have been widely used in the optimization of the PDS or PTS (Carrano et al. 2006; Guo and Li 2011; Teegala and Singal 2016). The general GA process begins with the creation of an initial population matrix spanning the design space. Individuals within the population are assigned chromosomes, collections of genes each storing the values of the design variables. The fitness of each individual is then evaluated, and parents are selected for reproduction with fitter parents favored (Beasley et al. 1993). During the reproduction phase, the next generation of children is created by recombining the parents’ chromosomes through the processes of mutation and crossover. In the crossover process, some parents are selected, and their chromosomes are cut randomly and swapped to create children with chromosomes unique from their parents (Beasley et al. 1993). In the mutation phase, following crossover, a small fraction of the children’s genes could be

© ASCE

Earth and Space 2022

altered slightly, creating more randomness to expand the search space. The reproduction process is then repeated with the newly created generation as the parents. Over successive generations, the mean and best fitness values improve toward the global optimum until th e specified convergence criteria are satisfied. 3. CASE STUDY For demonstration, an existing powerline located in southwestern Connecticut and displayed in Figure 4 is selected for optimization. The chosen line was identified as a critical segment of the larger system studied by Yuan et al. (2018). The line is about 300 meters (0.2 mi) in length and consists of 8 poles, with existing SYP poles of mostly class 2-3, typically around 40 feet tall, and roughly 30 years in age. The existing installation cost is estimated at $24,324. In the coming years, aging lines such as this may need replacement. The network is first simplified to two dimensions by considering the poles to lie along a straight line. Presently, the pole placements are considered at a ten-meter resolution to lower the computational costs and search space. While a more optimal design could therefore be improved further by using a smaller pole spacing resolution, such solutions could become unreasonably exact considering the difficulties in modeling the high-detailed terrain and related constraints. With the provided 10-meter resolution, a quasi-optimal solution is generated, providing a starting point allowing flexibility in the engineering judgment to adjust the placements slightly to account for design feasibility with respect to the natural topography. Elevation data were gathered from the 2016 Capitol Region Council of Government (2016) digital elevation model (DEM) as shown in Figure 4. The terrain is generally smooth and free of sudden changes with an average slope of 7% over the first 2 00 m and then flat for the remaining 100 m. Therefore, it is assumed the slope between any two poles in any feasible design can be estimated as constant in the clearance calculation. The allowable clearance over roads and sidewalks is taken as 5 meters (Cicconi et al. 2020; EverSource 2019).

Figure 4. (top) Existing distribution system overhead view, (bottom) simplified system with elevation view (not to scale)

© ASCE

594

595

Earth and Space 2022

3.1 Pole Structural Model The poles are assumed Southern Yellow Pine (SYP) following the work of Yuan et al. (2018). Fiberglass (FG) crossarms (9.2x11.7x244 cm) are adopted. Based on available data regarding the pricing, copper conductors are assumed. The material properties are presented in Table 2 as taken from (Yuan et al. 2018) considering the uncertainties in the material properties and strength. To comply with standard manufacturing sizes, the pole heights are allowed to range between 10.67 and 13.72 meters (35 – 45 feet) in increments of 1.52 meters (5 feet), while the pole classes range between 1 and 5 as listed in Table 3. The corresponding pole diameter is taken as uniformly distributed within the pole class bounds. The listed costs are the sum of the material costs as estimated from ATS (2019) plus the installation costs assumed as $2,500 (Salman and Li 2016). The unit wire costs are presented in Table 4 and adopted from Cicconi et al. (2020) assuming copper wires due to the data available at present. Table 2. Material Properties (Yuan et al. 2018) Parameter Fiber strength (kPA) SYP density (kg/m3) SYP elastic modulus (GPa) Pole diameter Elastic modulus conductor (GPa)

Distribution Lognormal

Mean 55,158

COV 0.15

Reference (Shafieezadeh et al. 2014)

Normal

500

0.04

(Doyle and Markwardt 1966)

Lognormal

11.59

0.04

(Doyle and Markwardt 1966)

Uniform Deterministic

Varies with class 121

-----

(Ukshini and Dirckx 2020)

Table 3. Available pole sizes and costs Class Price, 10.67 m ($USD2021) Price, 12.19 m ($USD2021) Price, 13.72 m ($USD2021)

1 3,050 3,153 3,282

2 2,979 3,084 3,170

3 2,930 3,039 3,110

4 2,868 2,991 3,039

5 2,816 2,938 2,998

Table 4. Available wire sizes (Cicconi et al. 2020) Cross-sectional area (mm2) 10 16 25 35 50

Diameter (mm) 4.1 5.1 6.3 7.5 9

Breaking load (kN) 4.02 6.37 9.72 13.77 19.88

Unit cost ($/m) 1.01 1.61 2.45 3.49 5.02

The presently selected demonstrative load scenarios are theoretical storms with gusts of 45 m/s, 50 m/s, and 55 m/s, corresponding to return periods of about 70, 100, and 175 years, respectively. Additionally, the wind speed could be considered with concurrent ice accretion of 5 mm (about 10-year return period). Consequently, six loading cases are defined in Table 5. As noted, the typical pole age in the system is about 30 years. The pole degradation over time, such as due to environmental factors and fungi, has been studied, and various methodologies have been developed and applied to model the time-dependent strength loss (Ryan et al. 2016; Salman and Li 2016). Based on NESC (2007) guidelines, the pole strength is assumed as 67% of the new

© ASCE

596

Earth and Space 2022

pole strength, which is estimated to occur between 30 and 70 years based on the differing strength decay models (Shafieezadeh et al. 2014; Wang et al. 2008). The initial design specifications of the system used aluminum conductor steel reinforced (ACSR) conductors (Yuan et al. 2018). However, in the present framework, due to the data availability, copper conductors are employed instead. For demonstration, therefore, the baseline system-level pole failure probability for the original system, Pf0, is calculated considering the smallest and largest wire sizes. In both cases, both the conductor and wire constraints are fully satisfied. Therefore, in comparison of the costs and failure probabilities from the optimal configuration to the original, the smallest wire size is assumed as the baseline. Table 5. Design loading scenarios Scenario

Gust (m/s)

Ice thickness (mm)

Pf0

1 2 3 4 5 6

45 50 55 45 50 55

0 0 0 5 5 5

0.04 0.19 0.50 0.84 0.98 0.99

By varying the allowable pole probability of failure rp and weather severity scenario, the economic tradeoffs of designing to various reliability levels are observable. The GA was employed using the Matlab function from the optimization toolbox (Mathworks 2018). A population size of 500 was selected, a crossover fraction of 0.6 was employed, and a bidirectional migration (Potts et al. 1994) with a migration rate of 0.4 and an interval of five generations was additionally utilized. A Gaussian mutation function was selected. 4. RESULTS 4.1 Sample optimal solution In all studied cases, the optimal configuration includes 35 ft poles, as the shorter poles will have less projected area and moment arm, while the clearance constraints remain satisfied . An example of the optimal configuration for Scenario 4 (45 m/s gust, 5 mm ice accretion) is presented in Figure 5. In both cases, the poles are primarily class 1, indicating that it is typically more cost-effective to use shorter, higher-class poles. When tightening the acceptable risk constraint from 20% to 1%, two additional poles are necessary to reduce the span lengths, resulting in increased cost, and some pole classes are changed.

Figure 5. Optimal configuration (Scenario 4) for (a) rp = 20% and (b) rp = 1%

© ASCE

Earth and Space 2022

4.2 Risk-cost tradeoff Figure 6 shows the Pareto fronts for design cost against system failure probability. For more extreme events, a large increase is observed in cost between an accepted 1% failure probability of 1% (rp ≤ 0.01) and 5 (rp ≤ 0.05), whereas for less extreme events, the difference between 1% and 5% rp is less substantial. In all cases, the difference between a 10% and 20 % acceptable failure is also seen as minimal, although it has a more noticeable effect in more severe cases. Therefore, cost tradeoff becomes most substantial when designing for an extreme event with a high-reliability level. Varying rp between 1 to 20% shows minimal impact on the system cost for the scenarios with no ice, as typically, for a given wind speed, the same number of poles are necessitated for all acceptable risks from 1 to 20%, with only minimal downgrades to the pole classes possible by increasing the acceptable failure risk. Figure 7 shows the price of installation to achieve reliability of rp ≤ 1% as a function of the design wind speed. The effect of accreted ice is also observed, as designing the system considering 5 mm of ice and simultaneous gusts requires about twice the investment as a system designed to the same Pf value without ice considered.

Figure 6. Optimal cost against reliability level for various storm scenarios

Figure 7. Effect of wind speed on optimal cost for failure probability rp ≤ 1%

© ASCE

597

Earth and Space 2022

The potential benefits of the optimized design are observed. For instance, Pf0 under Scenario 1 is around 4% (Table 6), while the optimized design for rp ≤ 5% is around $12,500, a roughly 48% reduction in installation costs. Similarly, under Scenario 2, Pf0 is around 19%, while the optimal design for rp ≤ 20% is also about $12,500. Under Scenario 4, the existing failure probability is 84%, while for a similar cost, the system can be designed to have a failure between 1-5%. Therefore, the optimal design of the system could cut the installation costs by roughly half to reach the same reliability standard, while also offering design options that have a much higher reliability level for a comparable price. However, the design optimization considers only wind and ice, while other loadings, such as impacts from falling trees, could be a critical design concern in some regions. In such cases, taller poles could be desired to reduce the probability of interaction with the surrounding vegetation. Further, the current framework also does not presently consider the repair and replacement costs. In some cases, it could be more cost-effective from a repair standpoint to use weaker conductors which would snap under tree impacts instead of dragging poles down. Such decisions are beyond the scope of the current study, although the framework could incorporate such concerns such as by increasing the minimum pole height near trees. 4.3 Constraints Figure 8 plots the average span length, minimum clearance, maximum conductor stress ratio, and conductor size for the various loads and reliability levels. In general, the average span length increases with increasing acceptable failure risk rp, reinforcing the use of fewer, higher-class poles is preferred over the use of a greater number of lower-class poles. For scenarios with no ice, the smallest conductor sizes are the most optimal as, in the absence of the additional stresses from the ice, the conductors are not overstressed. However, when considering ice, the wire size plays an important role and may not optimally be the smallest size. Between 1-5% rp for Scenarios 5 and 6, the conductor size must be increased to reduce the wire stresses, allowing for longer span lengths. Meanwhile, for Scenario 4 between rp of 5-10%, the conductor size can be decreased as the probability of failure constraint is relaxed. As the span length and conductor size increase, the above-ground clearance decreases, falling to 5.5 meters under the least severe load case, Scenario 1 with rp of 20%, as this case allows the longest span length.

Figure 8. Effect of accepted failure probability rp on (a) average span length, (b) conductor clearance, and (c) conductor stress, (d) conductor size

© ASCE

598

Earth and Space 2022

5. CONCLUSIONS In the present study, a risk-based optimization framework for connected structural systems was developed to minimize the construction and installation costs subject to design feasibility and reliability constraints using a genetic algorithm. A case study was conducted on the overhead power distribution system considering the uncertainties in the material properties and loadings. Pole sizes and heights and conductor diameters were constrained to standard manufacturing sizes, and the conductor sag and stresses constituted additional constraints. Under various storm scenarios, different acceptable probabilities of pole structural failure were tested to evaluate several resilience scenarios and the cost tradeoffs for designing a more robust system. The results indicate the use of shorter, higher-class poles provide the most cost-effective and reliable solutions, with around a 50% decrease in cost found for similar reliability levels compared to an existing system design. When the system is designed for extreme events, the pole and conductor stresses are the limiting active constraints. Meanwhile, for less severe storms, where the span length could be increased, the conductor sag becomes a more critical consideration. For scenarios with ice, smaller conductors produce more optimal solutions, while when considering ice, the increased stresses could require thicker conductors. The proposed methodology provides a flexible framework with potential applications to various connected systems and future use in other areas such as human habitats in extraterrestrial environments. Various additional constraints, such as obstacles at certain locations, could be employed. In the future, aerodynamic analysis of the conductor responses and the optimization of damper locations could be incorporated. Nevertheless, the framework still has room for substantial improvements. The pole-wire system was idealized as two-dimensional, although realistically, designs are required to navigate various obstacles, and designs in populated areas would typically follow roadways. Therefore, in the future, the framework could be expanded to include three-dimensional obstacles with a higher pole placement resolution. Further, the optimal design configurations do not consider possible failure modes such as tree impacts, which could necessitate stronger wires or taller poles. Ultimately, the framework provides an initial methodology for reliability-based optimization, which has not been well-studied despite its benefits over classical deterministic optimization methodologies. REFERENCES Abebe, Y. M., and Rao, P. M. 2016. “Overhead transmission line sag, tension and length calculation using affine arithmetic.” 2015 IEEE Power, Commun. Inf. Technol. Conf. PCITC 2015 - Proc., IEEE, 211–216. ATS (American Timber and Steel). 2019. “Products and Pricing.” Beasley, D., Bull, D. R., and Martin, R. R. 1993. “An overview of genetic algorithms: Part 1, fundamentals.” Univ. Comput., 2(15), 1–16. Braik, A. M., Salman, A. M., and Li, Y. 2019. “Risk-Based Reliability and Cost Analysis of Utility Poles Subjected to Tornado Hazard.” J. Aerosp. Eng., 32(4). Capitol Region Council of Governmnent (CRCOG). 2016. “Connecticut Statewide Lidar DEM 2016.” Carrano, E. G., Soares, L. A. E., Takahashi, R. H. C., Saldanha, R. R., and Neto, O. M. 2006. “Electric distribution network multiobjective design using a problem-specific genetic algorithm.” IEEE Trans. Power Deliv., 21(2), 995–1005.

© ASCE

599

Earth and Space 2022

Cicconi, P., Manieri, S., Bergantino, N., Raffaeli, R., and Germani, M. 2019. “A Design Approach for Overhead Lines Considering Configurations and Simulations.” Proc. CAD’19, 237–242. Cicconi, P., Manieri, S., Nardelli, M., Bergantino, N., Raffaeli, R., and Germani, M. 2020. “A constraint-based approach for optimizing the design of overhead lines.” Int. J. Interact. Des. Manuf., 14(3), 1121–1139. Doyle, D. V, and Markwardt, L. J. 1966. “Properties of southern pine in relation to strength grading of dimension lumber.” Dep. Agric. For. Serv. For. Prod. Lab. Madison, WI. EverSource. 2019. “Requirements for Electric Service Connection.” Guo, H. Y., and Li, Z. L. 2011. “Structural topology optimization of high-voltage transmission tower with discrete variables.” Struct. Multidiscip. Optim., 43(6), 851–861. Hatibovic, A. 2014. “Derivation of equations for conductor and sag curves of an overhead line based on a given catenary constant.” Period. Polytech. Electr. Eng. Comput. Sci., 58(1), 23–27. He, Y., Huang, Y., Cai, J., and Chen, Z. 2017. “A Numerical Analysis on the WindResistance Strengthening with Guy Wires for Distribution Lines.” Adv. Eng., 100, 402– 405. Holland, J. 1975. Adaptation in Natural and Artificial Systems: An introductory analysis with applications to biology, control, and artificial intelligence. The University of Michigan. Keselman, L. M., and Motlis, Y. 1998. “Application of the ruling span concept for overhead lines in mountainous terrain.” IEEE Trans. Power Deliv., 13(4), 1385–1390. Kishore, T. S., and Singal, S. K. 2014. “Optimal economic planning of power transmission lines: A review.” Renew. Sustain. Energy Rev., 39(2014), 949–974. Ma, X., Zhang, W., Bagtzoglou, A., and Zhu, J. 2021. “Local System Modeling Method for Resilience Assessment of Overhead Power Distribution System under Strong Winds.” ASCE-ASME J. Risk Uncertain. Eng. Syst. Part A Civ. Eng., 7(1), 1–15. Ma, Y., Dai, Q., and Pang, W. 2020. “Reliability Assessment of Electrical Grids Subjected to Wind Hazards and Ice Accretion with Concurrent Wind.” J. Struct. Eng., 146(7), 04020134. Mathworks Inc. 2018. “Matlab R2018b Optimization Toolbox.” NESC. 2007. “National Electrical Safety Code.” Inst. Electr. Electron. Eng. Palhares, P. H. da S., and Brito, L. da C. 2020. “Stochastic optimization method for mechanical design of overhead distribution power lines.” Eng. Optim., 52(2), 235–251. Pinto, R. E. 2012. “State Change Equation: Calculation Formula.” 2012 Work. Eng. Appl., IEEE. Potts, J. C., Giddens, T. D., and Yadav, S. B. 1994. “The Development and Evaluation of an Improved Genetic Algorithm Based on Migration and Artificial Selection.” IEEE Trans. Syst. Man Cybern., 24(1), 73–86. Raghavendra, T. 2012. “Computer Aided Analysis and Structural Optimization of Transmission Line Tower.” Int. J. Adv. Eng. Technol., 3(3), 44–50. Rezaei, A., and Sadeghi, M. H. 2019. “Analysis of aeolian vibrations of transmission line conductors and extraction of damper optimal placement with a comprehensive methodology.” Int. J. Eng. Trans. B Appl., 32(2), 184–191. Ryan, P. C., Stewart, M. G., Spencer, N., and Li, Y. 2016. “Probabilistic analysis of climate change impacts on timber power pole networks.” Int. J. Electr. Power Energy Syst., 78(2016), 513–523.

© ASCE

600

Earth and Space 2022

Salman, A. M., and Li, Y. 2016. “Age-dependent fragility and life-cycle cost analysis of wood and steel power distribution poles subjected to hurricanes.” Struct. Infrastruct. Eng., 12(8), 890–903. Salman, A. M., Li, Y., and Bastidas-Arteaga, E. 2017. “Maintenance optimization for power distribution systems subjected to hurricane hazard, timber decay and climate change.” Reliab. Eng. Syst. Saf., 168(2017), 136–149. Shafieezadeh, A., Onyewuchi, U. P., Begovic, M. M., and Desroches, R. 2014. “Agedependent fragility models of utility wood poles in power distribution networks against extreme wind hazards.” IEEE Trans. Power Deliv., 29(1), 131–139. de Souza, R. R., Fadel Miguel, L. F., Lopez, R. H., Miguel, L. F. F., and Torii, A. J. 2016. “A procedure for the size, shape and topology optimization of transmission line tower structures.” Eng. Struct., 111(2016), 162–184. Teegala, S. K., and Singal, S. K. 2016. “Optimal costing of overhead power transmission lines using genetic algorithms.” Int. J. Electr. Power Energy Syst., 83(2016), 298–308. Ukshini, E., and Dirckx, J. J. J. 2020. “Longitudinal and transversal elasticity of natural and artificial materials for musical instrument reeds.” Materials (Basel), 13(20), 1–13. Wang, C., Leicester, R. H., and Nguyen, M. 2008. “Probabilistic procedure for design of untreated timber poles in-ground under attack of decay fungi.” Reliab. Eng. Syst. Saf., 93(2008), 476–481. Yuan, H., Zhang, W., Zhu, J., and Bagtzoglou, A. C. 2018. “Resilience assessment of overhead power distribution system under strong winds for hardening prioritization.” ASCE-ASME J. Risk Uncertain. Eng. Syst. Part A Civ. Eng., 4(4), 04018037.

© ASCE

601

602

Earth and Space 2022

Viability of Construction Material within an Extraterrestrial Environment Linda E. Kuster1 and Justin D. Delorit2, Ph.D., P.E. 1

M.S. Candidate, Dept. of General Engineering Management, Air Force Institute of Technology, Wright-Patterson Air Force Base, OH. Email: [email protected] 2 Assistant Professor, Dept. of General Engineering Management, Air Force Institute of Technology, Wright-Patterson Air Force Base, OH. Email: [email protected] ABSTRACT Civil and military interest in space-based operations has increased dramatically in recent years due to advancements in space travel and discoveries, like water on Mars. Discussions have centered on the establishment of permanent infrastructure in extraterrestrial environments. Selecting appropriate construction materials and techniques requires that extraterrestrial civil engineers and understand candidate materials to determine their dynamic life-cycle costs. Robust and detailed systematic reviews have previously been completed on the research in this field however, these analyses do not provide intercomparisons of proposed materials or are not within the last decade. The rapid rate of new research in this area demands systematic reviews be performed at short intervals. To address these issues, this work produces a systematized review of literature that covers four construction materials: aluminum, biopolymer soil composite (BSC), sulfur concrete, and regolith. Here, several factors unique to space-based operations, including radiation resistance, high-impact strike resistance, thermal insulation/vulnerability, recyclability, material life cycle cost, and constructability are evaluated through a literature review to compare the feasibility of each material. The results of this analysis intend to inform both researchers and practitioners as they consider the vast distances materials must be transported, and the harsh and uncertain environmental conditions to which materials will be exposed. INTRODUCTION The 2020 National Space Policy Directive (NSPD) sets the goal that the United States establish a sustained presence on the moon by 2028. In line with the NSPD, the European Space Agency (ESA), the National Aeronautics and Space Administration (NASA), and their international peers have announced the desire to establish lunar and Martian habitats as early as 2040 (Naser 2019b). Civil engineers must interpret and understand what the goal “sustained presence” entails, and what methods and materials may be suitable to meet the challenge from an infrastructure perspective. For the past 17 years, the International Space Station (ISS) provided a habitable space environment for members of the global community (Naser 2019b). While the ISS provides a vehicle for permanent presence, it acts as an orbital body. To have a sustained presence on the moon, the same habitable standards must be achieved through engineers and planners. They must address the environmental challenges presented by space’s extreme habitat. Terrestrial environmental conditions ranging from snow and wind loads, seismic activity, and other environmental factors, are accounted for during civil design. Engineers will likely use the same design approach and account for similar unique conditions on the moon, Mars, and

© ASCE

Earth and Space 2022

other extraterrestrial bodies. Within the unfamiliar, diverse, and extreme environments found on extraterrestrial bodies, many unforeseen conditions will arise, complicating the design process. At this time, there is almost no design or construction precedent for lunar, Martian, or extraterrestrial habitats (Naser and Chen 2021). Despite the strides made by Naser comparing construction materials of like composition and Rabagliati et al. briefly discussing comparisons of different materials in an extraterrestrial environment, breaks in research exist in this field of study that this paper will address. As such, two critical gaps are filled by this research. First, it serves as a detailed analysis within the last decade, providing an intercomparison of proposed extraterrestrial construction materials through a systematic review. Recent publications have been limited to the evaluation of a single material, or the utility of materials in the presence of one extraterrestrial challenge. For the purposes of this research, an intercomparison of the sustainability and durability of four different viable extraterrestrial construction materials: aluminum, Biopolymer Soil Composite (BSC), regolith, and sulfur concrete, is presented comparing each’s ability to withstand radiation, high impacts, extreme thermal conditions, material life cycle cost, constructability, and recyclability. Second, this work identifies and amalgamates the abundance of research published in the field, providing engineers and planners an up-to-date synthesis, using the PRISMA framework, and articles from the 2021 American Society of Civil Engineers Earth and Space Conference proceedings (not yet entirely indexed such that they are captured by the PRISMA search). In total, the results of this research provide a systematic review of recent literature, cross comparing different categories of materials given environmental constraints in space. This allows practitioners and researchers a basis to expand on in the advancement of surface construction on extraterrestrial bodies. BACKGROUND: MATERIALS AND SPACE-BASED CONSTRUCTION Among the various materials proposed for extraterrestrial construction, an emphasis has been placed on In-Situ Resource Utilization (ISRU) (Rabagliati et al. 2021). ISRU categories highlighted in this research for intercomparison are concrete-like substances that may or may not require hydration; metals and metal alloys; and regolith. Aluminum is the material analyzed from the metal category. BSC is examined as the concrete substance that requires hydration while sulfur concrete serves as the concrete substance that does not require hydration. Regolith is assessed individually. These four materials are evaluated against one another, analyzing their durability and performance based on radiation-resistance (2.1), thermal durability (2.2), and high impact strike resistance (2.3). After analyzing the materials based on environmental factors, logistical considerations of constructability, material life-cycle cost, and recyclability are evaluated based on the procured literature for this paper. Radiation. Radiation exposure in an extraterrestrial environment causes short- and longterm effects on the human body to include cell death, cell mutation, and with prolonged exposure, loss of life (Akisheva and Gourinat 2021; Beemer and Worrells 2017). These hazardous effects can be further intensified by mission duration. With missions to Mars currently projected to last approximately 3 years, radiation represents the main health risk for deep-space missions (Giraudo et al. 2018). Additionally, electronics are also susceptible to radiation but can be built to withstand radiation in an extraterrestrial environment (Toutanji et al. 2012). Examining Earth’s neighboring celestial bodies (Table 1), the Martian atmosphere provides minimal shielding (Beemer and Worrells 2017) and the moon provides no shielding due to its lack of atmosphere (Naser 2019b). For these reasons, the impact of exposure to radiation is a major

© ASCE

603

604

Earth and Space 2022

factor to consider in long-term inhabitation of space. While it is apparent that other exoplanets have atmospheric presence, for the purposes of a large body of research, the moon is utilized to test extreme case scenarios as not everything is known about what humankind is to encounter as they seek colonizing extraterrestrial bodies (Naser 2019b). Table 1 – Differences in notable environmental influences on the Earth, Moon, and Mars. Data from (Katzer et al. 2021; Naser 2019b) Environmental Risk Radiation Atmospheric Protection Thermal limits

Earth 1–2.4 mSv Substantial, 101.3 kPa −89.2 °C to 56.9 °C

Moon 380 mSv None, 3 x 10-3 kPa −170 °C to 113 °C

Mars 100 mSv Minimal, 0.6 kPa −125 °C to 20 °C

Thermal Insulation/Vulnerability. Construction materials must be capable of withstanding extreme and rapidly changing temperatures. Temperatures on the moon and Mars can reach extreme ranges not seen in a terrestrial environment (Table 1). These intense ranges in temperature are due to lack of atmosphere and could impose acute and long-term thermal stresses on materials. Such stresses could lead to thermal fatigue resulting in a decrease in structural integrity in surface infrastructure (Akisheva and Gourinat 2021). High Impact Strike. The lack of atmosphere around the moon and Mars make them highly susceptible to micrometeroid and meteoroid strikes (Naser 2019b). Micrometeroids and meteoroids could hit the moon at speeds averaging 22.5 km s-1 with a wide uncertainty in mass (1 kg to over 4,500 kg). Without an atmosphere surrounding the moon to aid in the disintegration of these projectiles, any lunar structures within their vicinity would be highly susceptible to impact (Kalapodis et al. 2020). The same is true for Mars, but to a lesser extent, given its slightly more formidable atmosphere. While rare cases of meteoroid strikes are a possibility, most impacts will be from micro-meteorites. These strikes may not directly threaten the structural integrity of a facility however, vital communications equipment and sensors will be more susceptible to small impacts (Toutanji et al. 2012). METHODOLOGY A review using the Preferred Reporting Items for Systematic Review and Meta-Analyses (PRISMA) approach was accomplished to identify, screen, and retain applicable publications for this paper (Fig. 1). The PRISMA review process is a 27-item checklist that was created within the medical community to develop repeatable and transparent clinical practice guidelines. This practice provides clarity of reporting and validates the need for continued research in a certain area. Additionally, it reduces the probability of wasted efforts devoted to excessive reviews addressing the same issues. The benefits of the PRISMA review process was recognized outside of the medical community and has seamlessly transitioned to many fields of scientific literature (Moher et al. 2009). To begin this process for this body of research, an initial set of Boolean logic-based search criteria was used to search Scopus. In addition to this search, 2021 ASCE Earth and Space Conference Proceedings were accounted for. The total search yielded 654 publications from Scopus and 183 papers from the ASCE conference proceedings. ASCE Earth and Space Conference proceedings were included in this literature review as the conference interest specifically aligns with the research of this paper.

© ASCE

605

Earth and Space 2022

Using a pre-screening process followed by three levels of review, 57 articles were retained to construct this research paper. The remainder of this section explains the logic, filtering, and review processes.

Figure 1 – Execution of systematic exclusion and inclusion levels of review. To establish the initial paper retention criteria, several configurations of closely related Boolean test sets were used to search Scopus; the first thirty articles produced by each test set were reviewed for topical relevance to determine a ratio of articles applicable to the area of interest. All searches within Scopus adhered to this format, which targeted research focused on combinations of parent terms that addressed, where (environment), how long (longevity), and what (construction). The parent Boolean logic format that yielded the highest ratio was: Environment AND Longevity AND Construction This basic search also yielded many erroneous papers, and the search was too broad to capture papers that allowed an intercomparison of materials. To rectify this, several OR statements were included within the logic to refine the three previously mentioned parent search subjects. The final search term set was:

© ASCE

Earth and Space 2022

(Lunar OR Moon OR Extraterrestrial OR Martian OR Mars) AND (Permanent OR Endur* OR Temporary) AND (Construct* OR Build* OR Built OR Struct*) This initial search resulted in 654 eligible articles from Scopus. This search string was chosen because twenty of the first thirty results pertained to the areas of interest for this research. The pre-review filtered publications from Scopus to include just papers within the last ten years (2012 to 2022) and only articles within the subject areas of Engineering, Earth and Planetary Sciences, Physics and Astronomy, Environmental Science, Materials Science, Energy, Agricultural and Biological Sciences. After filtering irrelevant focuses, 258 Scopus-located publications remained. In addition to the Scopus results, conference proceedings from the American Society of Civil Engineers (ASCE) Earth and Space Conference were included for review. However, only proceedings from the 2021 conference were included because it is unlikely that they are fully captured in Scopus. The proceedings set from the conference contained Earth and Space 2021: Materials, Structures, Dynamics, and Control in Extreme Environments (E&S21: MSDC) and Earth and Space 2021: Space Exploration, Utilization, Engineering, and Construction in Extreme Environments (E&S21: SUEC). This search yielded 183 additional papers. After compiling publications from both Scopus (258) and the ASCE Earth and Space Conference (183), it was discovered that 11 articles were repeated. The repeated articles were removed leaving a total of 430 articles for the first level of review. This accounts for the removal of three repeat titles within Scopus and eight articles that were found in both Scopus and the ASCE Conference proceedings. The first level of review was a title review. For the title review process within Scopus, any titles that included key words were excluded. Themes of those key words included anything pertaining to moving bodies, biological studies, improvements in robotic technology, space law, space policies, and lava tube utilization as those were prevalent and irrelevant to this body of research. The title review reduced the number of Scopus-located titles to 143. The first level review process for the conference proceedings differed from the first level of review for Scopus-located papers. For the conference proceedings, key words were verified to be within a title. Inclusion criteria included key words pertaining to the construction of permanent or stationary infrastructure utilizing the materials of interest. Then, the title was reviewed to ensure no exclusion key words were present such as lunar landing, lava tube, propellant, ice, or spring. Of the total number of titles from the conference proceedings, 71 were retained for the next level of review. In total, 214 (Scopus + Earth and Space) titles remained after the second level of review. The second level of review was the abstract review. Articles that contained anything relating to dynamics, gravity analysis, ice drilling, plant studies, evolution of planets, construction management work-flow process, existence of extraterrestrial life, and inflatable facility design that did not include mention of the space environment the facility would be designed to withstand, were excluded. After the abstract review, 104 total titles remained. The third level of review was the full text review. Articles that did not contain information relating to aluminum, BSC, regolith, or sulfur concrete in the context of extraterrestrial construction recyclability, material cost, or constructability withstanding radiation, thermal stresses or high impact strikes were not included in the paper. After the full text review process, 53 articles remained pertinent to the topics of interest. While reviewing publications, it was recognized that some materials had more coverage than others. For this reason, publications mentioned in the references of the 53 retained articles

© ASCE

606

607

Earth and Space 2022

following the full text review were also included for research purposes as needed. In total, four articles were used to better the understanding of materials with minimal published research. Three of these articles related to BSC and one related to regolith. In total, 57 full texts were included in this research. Figure 2 illustrates the year of publication that made it through each level of review.

Figure 2 –Publication abundance (2012-2022), n = 430. Note: The majority of relevant titles emanated in 2021 were from the ASCE Earth and Space Conference. RESULTS Results of the systematic review are illustrated in Table 2 delineating the number of papers focused on the criteria of interest and materials being compared. It is apparent that there is an abundance of research on regolith regarding extraterrestrial construction with focus on constructability and radiation. Aluminum has the next most publicized research followed by sulfur concrete and BSC. The relatively low volume of literature on BSC in all the categories of interest could be attributed to the fact that most of the publications for this material come from only one location, Stanford, CA. Results of these papers are summarized in sections 4.1-4.4. Table 2 – Papers highlighting the environmental effects on Aluminum (Al), BSC, regolith (reg), and Sulfur Concrete (SC)

Radiation

Al

1. Naser & Chen 2021

1. Naser & Chen 2021

2. Akisheva & Gourinat 2021

2. Anderson & Thangavelautham 2021

3. Giraudo et al. 2018 (+1)

© ASCE

High Impact Strikes

Thermal Insulation/ Vulnerable

Recyclable

Material LCC*

1. Baasch et al. 2021 Naser 2019b

2. Rabagliati et al. 2021

Constructable 1. Naser 2019b

Naser 2019b

2. Naser & Chen 2021

3. Sgambati et al. 2018

3. Baasch et al. 2021

(+2)

(+4)

608

Earth and Space 2022

1. Allende et al. 2020 2. Allende et al. 2019

BSC

1. Allende et al. 2020 2. Roedel et al. 2015b

Roedel et al. 2019

1. Naser & Chen 2021 2. Spedding et al. 2020 3. Alkhateb et al. 2021 (+11)

Reg

1. Naser 2019b 2. Farries et al. 2021 3. Guo et al. 2021 (+20)

1. Naser 2019b 2. Farries et al. 2021 3. Dyke et al. 2021 (+6)

1. Naser 2019b 2. Farries et al. 2021 3. Akisheva & Gourinat 2021 (+5)

Sgambati et al. 2018

Baasch et al. 2021

1. Patel et al. 2021 SC

Toutanji et al. 2012

2. Toutanji et al. 2012

1. Dyke et al. 2021 2. Farries et al. 2021 3. Thangavelu & Adhikari 2017 (+ 21) 1. Naser 2019b

Sgambati et al. 2018

Troemner et al. 2021

2. Troemner et al. 2021 3. Naser & Chen 2021 (+5)

*Life Cycle Costs Aluminum. Aluminum has proven reliable in a space environment which is mostly attributed to its radiation shielding properties and durability (Naser and Chen 2021). It is able to deflect high-energy primary particles but secondary emissions, often neutrons, are able to penetrate an aluminum layer. For this reason, aluminum is often paired with polyethylene, as is seen in the ISS, to provide adequate protection against radiation (Akisheva and Gourinat 2021). Furthermore, in an experiment run by Giraudo et al, when directly compared to moon based regolith, aluminum ranked lower in regards to radiation resistance (Giraudo et al. 2018). Additionally, aluminum’s ability to withstand high impact strikes and vibrations due to its high material strength, has further made it an attractive material for use in space (Naser and Chen 2021). Moreover, aluminum’s strength and ductility increase under low temperatures although it is a poor thermal insulator (Naser and Chen 2021). As is seen in a terrestrial environment, aluminum can be recycled with less energy than is used in production (Rabagliati et al. 2021). While aluminum is the third most abundant metal found on the moon and Mars (Naser 2019b), extracting metal from regolith in large quantities will be challenging in a lunar environment (Naser and Chen 2021). With more than 8,800 tons of space debris in Earth’s orbit, aluminum could be harvested from the space debris and repurposed for extraterrestrial construction while methods are being advanced to extract metal from regolith (Baasch et al. 2021). Furthermore, utilizing space debris from orbit, would eliminate the cost to transport aluminum from Earth and provide an opportunity to recycle space trash. Any harvested aluminum from space debris would need to be cast into structural elements for erecting. Approaches for constructing permanent structures with aluminum in space are being conceptualized. The Institute of Space Systems and Orbit Recycling has experimented with casting aluminum using regolith in place of green sands which could be done with the harvested

© ASCE

Earth and Space 2022

aluminum already present in space. Two issues with this type of casting are mold degrading and binders of the mold burning, evaporating, or bonding with the material mold. This limits the amount of times the mold could be used (Baasch et al. 2021). More research and development are needed to adapt the mold for reuse using regolith. Following the casting of the aluminum, a crane would be needed to move the casted components. Teleoperated construction equipment has been tested for possible operation in an extraterrestrial environment (Narumi et al. 2018). Finally, to make a facility capable of withstanding impact strikes, honeycomb sandwiching has been proposed. Studies show that honeycombing performs better under a hypervelocity impact than a monolithic aluminum shield, with double honeycombing performing even better (Gunasekara and Jablonski 2021). Biopolymer Soil Composite (BSC). BSC is a term used to describe a soil that utilizes organic biopolymers as a binder. Research on this material has mostly been performed by NASA and Stanford using JSC-1A Lunar regolith, bovine blood proteins, and deionized water (Alkhateb et al. 2021). Tests demonstrated BSC was capable of reaching compressive strengths comparable to ordinary Portland Concrete Cement (PCC) when not compromised by water submersion, making this material attractive for space use (Roedel et al. 2015b). To test BSC for high impact strike resistance, Allende et al. performed a hypervelocity impact experiment followed by CTH simulations (a shock physics code developed by Sandia National Laboratories) with aluminum projectiles ranging from 0.1-25 mm traveling at 5-70 km s-1. Based on a 1 mm aluminum projectile at 5 km s-1, the CTH simulation resulted in a 0.3 mm deep crater. This demonstrated that craters in the 3D-printed BSC were smaller in volume and diameter than those formed in quartzite and sandstone based on the strength scaling relationships (Allende et al. 2020). Furthermore, the BSC’s components are relatively recyclable, as 95% of BSC can be crushed, dissolved, and centrifuged for reuse (Roedel et al. 2015b). For construction, BSC has exhibited an ability to retain its shape, making it a viable option for extrusion-based 3D-print technology (Alkhateb et al. 2021). A few drawbacks to 3D printers include high maintenance costs, time-intensiveness, energy-intensiveness, and complex equipment, not ideal for an initial settlement (Thangavelu and Adhikari 2017). The compositional makeup of BSC is ISRU, with the exception of bovine blood. Since this component cannot be found in an extraterrestrial environment, consideration would need to be given to the cost to transport the required amount of bovine blood to an extraterrestrial site. The estimated cost of transporting 1 kg from the Earth to the Moon (the closest celestial body) is between $5,000-$20,000 (Naser 2019a). The discovery of water in space opens up the opportunity for the hydration required for this material to come from space assets (Roedel et al. 2015a). Regolith. Regolith is formed through meteorite and micrometeorite strikes (Naser 2019b). It has characteristics such as high density and cohesion that make it suitable to be considered for construction (Dyke et al. 2021). It has been suggested for use as a shield against lunar radiation and an inhibitor to the effects of high impact strikes due to its density (Naser 2019b). The use of 3-5 meters of loose regolith has been proposed as a suitable amount of material to protect against radiation and micrometeorite strikes (Grandl 2017; Naser 2019b; Srivastava et al. 2016). Additionally, regolith has shown poor thermal conductivity properties making it a good thermal insulator in an extraterrestrial environment (Naser 2019b). To acquire regolith, mining for the material is necessary. A piece of equipment that is already designed that would be suitable for extracting regolith is called the stone crusher. To utilize this

© ASCE

609

Earth and Space 2022

technology, the stone crusher would need to be modified and automated with little to no human interaction due to the austere conditions in space (Thangavelu and Adhikari 2017). Two construction methods for utilizing regolith are sintering or melting. This is compatible with extraterrestrial construction as it only uses solar energy as input (Farries et al. 2021). Similar to BSC, regolith has been considered for 3D printing. A proposed technique for printing lunar regolith is utilizing selective laser sintered (SLS) technology, which involves placing powdered composites of regolith for each layer then melting the powder with a laser (Anderson and Thangavelautham 2021). Possible limitations include warping, which can be minimized by reducing the amount of energy input around the perimeter of the scan (Farries et al. 2021). Additionally, SLS needs direct access to the sunlight, its power source (Anderson and Thangavelautham 2021), and the equipment requires shielding and continuous maintenance to clean the mirrors and lenses (Naser 2019b). Microwave sintering is another sintering method however, it lacks efficiency and requires knowing the appropriate frequency based on the material type (Naser 2019b). Radiant furnace sintering thermally treats the regolith particles without fully melting the material to bind the particles into a solid mass. Casting regolith involves robust equipment with few moving parts to heat the material to a liquid state. It can then be placed in a mold and possibly buried to passively control the rate of cooling (Farries et al. 2021). After 53 experiments conducted by Farries et al., they concluded that cast regolith proved to be the most structurally sound as it is fully dense, very strong, and abrasion resistant (Farries et al. 2021). A characteristic of a sintered and cast regolith is brittleness, making structures susceptible to fractures. This is highly undesirable in an extraterrestrial environment as thermal cycling can be present (Farries et al. 2021; Thangavelu and Adhikari 2017). Sulfur Concrete. Another resource that is found on the moon and Mars is sulfur (Naser 2019b). Sulfur concrete can vary in its composition anywhere from 10-35% sulfur and 65-90% aggregate (Naser 2019a; Toutanji et al. 2012). This water-free concrete was previously used in terrestrial construction but due to highly volatile, noxious, and toxic characteristics, it is not as favorable as PCC (Patel et al. 2021). However, extraterrestrial habitats will likely already have air filtering features in place, therefore, it could be a viable option when it comes to construction in an extraterrestrial habitat. To test high impact strikes and radiation, Toutanji et al. conducted a hypervelocity impact test on a 5 cm3 sample of sulfur concrete struck by a 1 mm aluminum projectile at 5.85 km s -1. It was discovered that a 12.8 mm average diameter and 3.1 mm central depth crater was formed. For the radiation portion of the experiment, it was concluded that sulfur concrete would need to be combined with polyethylene to provide the same amount of radiation shielding as a wall made of only JSC-1 simulant, which is representative of regolith (Toutanji et al. 2012). Processing sulfur is not complex, and due to its melting point and viscosity, it can be used as a non-hydraulic binder. Sulfur is available on the moon and Mars and experiments indicate that 12–30% of the total sulfur in lunar regolith can be extracted at 750 °C, while 85-95% can be extracted at temperatures close to 1100 °C. Based on these results, it is estimated that thermal processing of 1000 m3 of mildly crushed regolith could yield one ton of sulfur. Additionally, sulfur can be extracted from regolith through oxidation (Naser 2019b). After successful extraction, it is heated to its point of liquification and injected into regolith to cool and harden. A drawback of sulfur concrete it separates from the particles of the regolith due to sulfur shrinkage when it cools. This creates voids in the material, which reduces the strength (Naser 2019a) contributing to its relatively short life span when used as a terrestrial construction

© ASCE

610

611

Earth and Space 2022

material (Troemner et al. 2021). Furthermore, solidified sulfur masses are soft and perform poorly during thermal cycling (Patel et al. 2021). Sulfur concrete is a good candidate for 3D printing since the material hardens quickly allowing remelting during the 3D printing process to increase layer interface strength (Troemner et al. 2021). The 3D printing technology suitable for processing sulfur concrete is still being developed (Troemner et al. 2021). 3D printed materials can be recycled but the process can be challenging likely proving far more difficult in an extraterrestrial environment (Sgambati et al. 2018). DISCUSSION Extraterrestrial environments present unique challenges to construction that must be considered during these early conceptualization and planning stages. To minimize the cost of transport and ensure accessibility to construction materials, ISRU resources are the most practical solution for extraterrestrial construction. Previously published literature regarding ISRU application include Naser’s 2019 systematic review which outlines and compares like materials. Metals were compared to metals. Regolith simulants were compared to each other. Concrete and concrete-like products were compared to each other and so forth (Naser 2019b). Rabagliati et al 2021 focused on the establishment of a twenty-member crew for long term habitation in a space outpost. The authors did not account for likely issues related to temperature extremes and fluctuations; only briefly touch on construction material comparisons; and the review itself was not systematized and is therefore not reproducible. Furthermore, the foundation of material comparisons for Rabagliati et al. stemmed from Happel’s Indigenous Materials for Lunar Construction published in 1993 (Rabagliati et al. 2021). With the rapid advancements within the last decade, it is important to review the most current methodologies, and materials proposed for extraterrestrial construction. Based on the results of this research, intercomparison analyses were made with respect to materials of interest referencing published literature within the last decade and ranking materials accordingly (Table 3). Due to the limited information regarding maintenance of the materials of interest, material life cycle cost was not ranked. Table 3 – Material rankings based on criteria analyzed Rank

Radiation

1

Regolith Sulfur Concrete

2

BSC

Thermal Insulation/ Vulnerability Regolith

Regolith

High Impact Strikes

Regolith

Material Life Cycle Cost -

Aluminum

Aluminum

-

Sulfur Concrete

Recyclability

Constructability Regolith

3

Aluminum

Aluminum

Sulfur Concrete

BSC

-

Aluminum

4

-

Sulfur Concrete

-

Sulfur Concrete

-

BSC

N/A

BSC

BSC

-

The most suitable material for radiation resistance is regolith, followed by sulfur concrete. The order of aluminum and sulfur depends on whether aluminum from space debris is used or if ISRU harvested material is prioritized. Regolith has proven its resistance to radiation penetration with the use of 3-5 meters of loose regolith. Aluminum and sulfur concrete performed at a

© ASCE

Earth and Space 2022

suitable level for space use after the inclusion of polyethylene which is not an in-situ resource. More research is need on BSC’s radiation resistivity before conclusive ranking can be made. For high impact strike resistance, BSC is most suitable followed by regolith and double honeycombed aluminum. The order of regolith and aluminum would again depend on the emphasis of ISRU. Sulfur concrete would be the least suitable for handling high impact strikes, consistent with characteristics of a softer material. Aluminum becomes more ductile and higher in strength as it gets colder, which is conducive to the cold environments in space (Table 1). When analyzing BSC and sulfur concrete, both materials were evaluated using hypervelocity experimentation with 1 mm aluminum projectiles. The results showed that the crater in BSC was 0.3 mm in depth, compared to the 3.1 mm depth for sulfur concrete. It is to be noted that the projectile for the BSC experiment was only traveling 5 km s -1 while the projectile for the sulfur concrete was traveling at 5.85 km s -1. Regolith’s dense characteristics make it a suitable material to dissipate high impact strike effects. The findings for thermal insulation and vulnerability revealed loose regolith is top ranked, followed by aluminum, and then sulfur concrete. The structural integrity of aluminum strengthens in colder climates. Unfortunately, as a good conductor, the thermal extremes that an aluminum structure would experience in space would not provide adequate insulation, without being paired with another material. Again, due to the dense nature of regolith, 3-5 meters of loose regolith provides thermal insulation in an extraterrestrial environment. Sintered and cast regolith do not perform well in a space environment due to brittleness and continuous thermal cycling. Sulfur concrete becomes weaker when it cools because sulfur particles separate from the aggregate. More research is need on BSC and sulfur concrete before rankings can be made. For recyclability, loose regolith is best, then aluminum, BSC, and lastly sulfur concrete. Aluminum can be recycled the same way that it is on Earth. In fact, aluminum that is al ready in space can be harvested, recycled, and cast for structural use. Depending on how regolith is used, loose regolith could be transported as needed to provide shielding for facilities. While the specifics of 3D-printed regolith recycling is not discussed, Sgambati et al. mentions that recycling 3D-printed material can be challenging. This applies to sulfur concrete as well as BSC, although, BSC specifically states that 95% of the components used for the material can be recycled through centrifugation. Based on the availability of a centrifuge, BSC would be easier to recycle than sulfur concrete. Material life-cycle cost includes the amount of time, money, and effort it would take to acquire, maintain, and dispose of a material. While little research has explored maintenance of these materials, sulfur concrete is likely to require maintenance due to the cracks that would occur from thermal cycling. Additionally, it is assumed that since there is 8,800 tons of aluminum still orbiting in space, it likely has a long-life cycle. To better understand this aspect of these materials, more research is needed. The research currently outlined in Table 2, in the Material Life Cycle Cost column, accounts mainly for recyclability, which has been described in detail as its own category due to its importance in the austere conditions of space. Based on the literature reviewed, and the qualitative analyses of performance criteria evaluated by this study, regolith would be the most feasible material used for extraterrestrial construction, followed by either sulfur concrete or aluminum (Table 3). Still, there are many factors to consider such as the acquisition of the material, the tools required to erect a facility, and the amount of manpower involved. To use aluminum debris, harvesting and processing would need to be accounted for. These two endeavors could prove to be time intensive and costly despite the benefits of clearing space debris. Aluminum could also be extracted from regolith

© ASCE

612

Earth and Space 2022

with some difficulty. With use of the teleoperated crane technology, minimum manpower is required. Sulfur concrete requires harvesting sulfur from regolith and transport of a 3D printer. The biggest factors contributing to the ranking of aluminum and sulfur concrete is the emphasis on ISRU and the ability to harvest aluminum. Regolith can be mined using a stone crusher but would require transport to the site. This affects sulfur concrete and BSC as well since they both consist of a meaningful percentage of regolith. Casting regolith requires equipment to be transported from Earth. Since there are few moving parts to maintain on the casting equipment, it is likely the maintenance would be more manageable than 3D-printed technology. Unfortunately, there was no mention of the manpower required to cast regolith therefore, SLS is most suitable for use on an extraterrestrial construction site. For BSC, a 3D laser printer must be transported from Earth followed by continual transportation of bovine blood. Additionally, water is required to hydrate the material. Competition for water could prove inimical to BSC in an extraterrestrial environment since water will likely be a precious resource even with the presence of ice caps. A limitation in this field of study is the availability of lunar and Martian regolith. Many of the findings evaluated use regolith simulants. It is apparent that the composition of lunar and Martian regolith is not the same, thus many of the unique properties exhibited by the respective lunar and Martian regolith could be missing. Despite the findings presented in this research, this study is limited in several ways. Construction opportunities such as lava tubes, underground facilities, and inflatable structures were not analyzed. Additionally, legal ramifications of building on an extraterrestrial environment were not considered. Furthermore, seismic, gravity, terraforming (the process of modifying the atmospheres of other celestial bodies), and dust effects were not considered. In addition, hydrocrater formation was not discussed. This could affect the emphasis on high impact strike resistance. Finally, for acquisition of material, no consideration was given to asteroid harvesting. CONCLUSION This work systematically analyzes published material research on aluminum, Biopolymer Soil Composite (BSC), regolith, and sulfur concrete for extraterrestrial construction. Each material was analyzed on its ability to withstand radiation, high impact resistivity, extreme thermal conditions, material life cycle cost, constructability, and recyclability. While each material presented in this body of research is feasible for use in an extraterrestrial environment, the extent of trade-offs between earth-procured materials and in-situ materials requires further investigation. Transportation logistics of earth-procured materials, such as cost and time, makes ISRU appealing for extraterrestrial construction. ISRU presents its own challenges such as reliability of unmanned equipment, mining and harvesting, and transportation of materials to the construction site. Furthermore, lack of research on maintenance and material duration in a space environment made assessment of materials based on life-cycle analysis difficult to accomplish leaving a gap in research that requires further development. Future work should focus on experiments that allow for intercomparison of materials given the same reference of measure. Additionally, there is an abundance of research on greenhouses and human habitations in space. While this technology will be required for extraterrestrial habitation, based on the state of science with respect to the quest to colonize space, consideration should be given to what type of structure to build as the test facility for analysis. An uninhabited communications facility might be suitable. This would allow engineers to better understand the

© ASCE

613

Earth and Space 2022

impacts that the extraterrestrial environment will have on a surface structure. Finally, consideration should be given to the possible negative effects that construction in space could have on the space environment. Despite the need for additional research, the rate of research production remains high, and there should be consistent efforts to provide intercomparisons of materials and extraterrestrial constraints, like the ones produced here, such that the space community has the best possible information as they work toward achieving the goals of the NSPD and colonization of space. REFERENCES Akisheva, Y., and Gourinat, Y. (2021). “Utilisation of Moon Regolith for Radiation Protection and Thermal Insulation in Permanent Lunar Habitats.” Applied Sciences, Multidisciplinary Digital Publishing Institute, 11(9), 3853. Alkhateb, H., Almashaqbeh, H., Edmunson, J., Fiske, M., Najjar, Y., and Stoddard, D. (2021). “Optimizing Magnesium Phosphate Binders with Boric Acid for Additive Construction Applications.” Earth and Space 2021, American Society of Civil Engineers, Virtual Conference, 13–28. Allende, M. I., Davis, B. A., Miller, J. E., Christiansen, E. L., Lepech, M. D., and Loftus, D. J. (2020). “Hypervelocity Impact Performance of Biopolymer-Bound Soil Composites for Space Construction.” Journal of Aerospace Engineering, 33(2), 04020001. Anderson, S. D., and Thangavelautham, J. (2021). “Solar-Powered Additive Manufacturing in Extraterrestrial Environments.” Earth and Space 2021, American Society of Civil Engineers, Virtual Conference, 732–744. Baasch, J., Windisch, L., Koch, F., Linke, S., Stoll, E., and Schilde, C. (2021). “Regolith as substitute mold material for aluminum casting on the Moon.” Acta Astronautica, 182, 1–12. Beemer, H. D., and Worrells, D. S. (2017). “Conducting Rock Mass Rating for Tunnel Construction on Mars.” Acta Astronautica, 139, 176–180. Dyke, S. J., Marais, K., Bilionis, I., Werfel, J., and Malla, R. (2021). “Strategies for the design and operation of resilient extraterrestrial habitats.” Sensors and Smart Structures Technologies for Civil, Mechanical, and Aerospace Systems 2021, D. Zonta, H. Huang, and Z. Su, eds., SPIE, Online Only, United States, 2. Farries, K. W., Visintin, P., Smith, S. T., and van Eyk, P. (2021). “Sintered or melted regolith for lunar construction: State-of-the-art review and future research directions.” Construction and Building Materials, 296, 123627. Giraudo, M., Schuy, C., Weber, U., Rovituso, M., Santin, G., Norbury, J. W., Tracino, E., Menicucci, A., Bocchini, L., Lobascio, C., Durante, M., and Tessa, C. L. (2018). “Accelerator-Based Tests of Shielding Effectiveness of Different Materials and Multilayers Using High-Energy Light and Heavy Ions.” Radiation Research, 190(5), 526. Grandl, W. (2017). “Human life in the Solar System.” REACH, 5, 9–21. Gunasekara, D., and Jablonski, A. M. (2021). “Technical Aspects of Micrometeoroid Impact on Lunar Systems/Structures.” Earth and Space 2021, American Society of Civil Engineers, Virtual Conference, 894–907. Kalapodis, N., Kampas, G., and Ktenidou, O.-J. (2020). “A review towards the design of extraterrestrial structures: From regolith to human outposts.” Acta Astronautica, 175, 540–569.

© ASCE

614

Earth and Space 2022

Katzer, J., Suchocki, C., Błaszczak-Bąk, W., Zarzycki, P. K., and Damięcka-Suchocka, M. (2021). “Reliability and effectiveness of laser scanners in future construction efforts on the Moon and Mars.” Automation in Construction, 132, 103979. Moher, D., Liberati, A., Tetzlaff, J., Altman, D. G., and and the PRISMA Group. (2009). “Reprint—Preferred Reporting Items for Systematic Reviews and Meta-Analyses: The PRISMA Statement.” Physical Therapy, 89(9), 873–880. Narumi, T., Aoki, S., Yokoshima, T., Uyama, N., Fukushima, S., Tabuchi, G., Kanamori, H., and Wakabayashi, S. (2018). “Preliminary System Design for Teleoperating Construction in Extreme Environments.” Taipei, Taiwan. Naser, M. Z. (2019a). “Space-native construction materials for earth-independent and sustainable infrastructure.” Acta Astronautica, 155, 264–273. Naser, M. Z. (2019b). “Extraterrestrial construction materials.” Progress in Materials Science, 105, 100577. Naser, M. Z., and Chen, Q. (2021). “Extraterrestrial Construction in Lunar and Martian Environments.” Earth and Space 2021, American Society of Civil Engineers, Virtual Conference, 1200–1207. Patel, H., Suermann, P., Shumaker, N., Deitrick, S., Bullard, J. W., and Ewing, R. C. (2021). “Applying Terrestrial Geo-Material Science Methodology to Lunar ISRU Construction.” Earth and Space 2021, American Society of Civil Engineers, Virtual Conference, 171–180. Rabagliati, L., Devecchi, M., Lovagnini, A., Pino, P., and Thirion, G. (2021). “Regolith Mining in Shackleton Crater on the Moon: Propellant, Building Materials and Vital Resources Production for a Long Duration Manned Mission.” International Journal of Astronautics and Aeronautical Engineering, 6(1). Roedel, H., Lepech, M. D., and Loftus, D. J. (2015a). “Protein-Regolith Composites for Space Construction.” Earth and Space 2014, American Society of Civil Engineers, St. Louis, Missouri, 291–300. Roedel, H., Plata, I. R., Lepech, M., and Loftus, D. (2015b). “Sustainability Assessment of Protein–Soil Composite Materials for Limited Resource Environments.” Journal of Renewable Materials, 3(3), 183–194. Sgambati, A., Berg, M., Rossi, F., Daurskikh, A., Imhof, B., Davenport, R., Weiss, P., Peer, M., Gobert, T., and Makaya, A. (2018). “URBAN: conceiving a lunar base using 3D printing technologies.” 69th International Astronautical Congress (IAC). Srivastava, V., Lim, S., and Anand, M. (2016). “Microwave processing of lunar soil for supporting longer-term surface exploration on the Moon.” Space Policy, 37, 92–96. Thangavelu, M., and Adhikari, P. (2017). “MPIT: Minimally Processed ISRU Technology Structures for Rapid Extraterrestrial Settlement Infrastructure Development.” AIAA SPACE and Astronautics Forum and Exposition, American Institute of Aeronautics and Astronautics, Orlando, FL. Toutanji, H. A., Evans, S., and Grugel, R. N. (2012). “Performance of lunar sulfur concrete in lunar environments.” Construction and Building Materials, 29, 444–448. Troemner, M., Ramyar, E., Marrero, R., Mendu, K., and Cusatis, G. (2021). “Marscrete: A Martian Concrete for Additive Construction Applications Utilizing In Situ Resources.” Earth and Space 2021, American Society of Civil Engineers, Virtual Conference, 801–807.

© ASCE

615

Earth and Space 2022

Detection of Corrosion-induced Damage in Bolted Steel Structure Using Piezoceramic Transducers Wen-I Liao1 and Chien-Kuo Chiu2, Rei-Ching Huang3 and Meng-Xun Lu4 1

Department of Civil Engineering, National Taipei University of Technology, Taipei, Taiwan; PH (886) 2-27712171; FAX (886) 2-27814518; email:[email protected] 2 Department of Civil and Construction Engineering, National Taiwan University of Science and Technology, Taipei, Taiwan; PH (886) 2-27376580; FAX (886) 227376606; email: [email protected] 3 Department of Civil Engineering, National Taipei University of Technology, Taipei, Taiwan; email:[email protected] 4 Department of Civil Engineering, National Taipei University of Technology, Taipei, Taiwan; email:[email protected]

ABSTRACT Bolted steel structures may be subjected to many adverse effects, such as corrosion, vibration, shock, and seismic excitation, which reduce the strength of the bolted connection and thereby weaken the whole structure. Therefore, monitoring the health of bolted connections is very important in ensuring the safety of structures. In this paper, a two-story bolted steel frame with braces was used in the detection of damage to a bolted connection. The piezoelectric-based active sensing method was utilized to evaluate the damage condition of the connection. Various degrees of corrosion and bolt loosening in various positions were simulated during the test to verify the effectiveness of the proposed method. Both the impact waves and the sinusoidal excitation waves produced by impact test and a piezoceramic actuator respectively, provided active sensing stress waves. The continuous wavelet transform was adopted to process the received piezoelectric signal and to evaluate the damage index. The energy-based damage index was used to predict the severity of damage to a bolted joint, allowing slightly corroded bolts to be accurately detected. This work reveals that the health status of the bolted connection is influenced by the severity of corrosion of the bolt, and the presented diagnostic method is effective in monitoring the severity of corrosion and the degree of bolt loosening of a bolted connection. INTRODUCTION Monitoring the performance, structural integrity, and safety of civil infrastructure under environmental loading is an important task of the civil engineering community. Bolted steel structures may be subject OR to many adverse processes, such as corrosion, vibration, shock, and seismic excitation, which reduce the strength of bolted connections and thereby weaken the whole structure. Therefore, monitoring the health of bolted connections is very important in ensuring the safety of structures. Detecting a loose bolted joint has been a topic of interest for many years. Several

© ASCE

616

Earth and Space 2022

techniques for so doing have been proposed and studied; they include the ultrasonic technique, electromagnetic resonance and the impedance technique. A number of significant and relevant investigations have been published. In recent years, piezoelectric materials have been successfully used in monitoring the health of various structures owing to their active sensing, low cost, rapid response, availability in various shapes, and ease of use. Piezoceramic materials are some of the most used piezoelectric materials for sensing purposes. Numerous studies have focused on the detection of bolt loosening using an active sensing system that is based on surface-bonded PZT patches. Wang et al. (2017) investigated the relationship between incident wave energy and applied torque in bolted joints. They showed that the wave energy that propagated through the connection increased with the applied torque. Feng et al. (2017) proposed a time reversal method for monitoring the health status of bolted joints. Wang and Song (2019, 2020) monitored the multi-bolt connection looseness using a novel vibroacoustic method. Wang et al. (2018) proposed a method to quantify EMI based methods through the integration of fractal contact theory for quantitative monitoring of bolted looseness. The relationship between the electrical impedance of a piezoceramic patch installed on the joint and the mechanical impedance of the bolted joint can be determined based on this method. A new percussion-based nondestructive approach to determine the health condition of bolted joints with the help of machine learning was presented by Kong et al. (2018). This proposed approach has great potential to be implemented with intimately weaving robotics and machine learning to produce a cyber-physical system that can automatically inspect and determine the health of a structure. Huo et al. (2017a) used the electromechanical impedance (EMI) method to develop a smart washer for studying applied pre-stress in bolted connections. Their results showed that the impedance signal decreased as the torque that is applied to the bolt decreased, so their proposed smart washer can be used to monitor the loosening of bolts. Yao et al. (2016) used the PZT sensor to detect damage to a steel structure that was caused by spot welding using both time-domain and frequency-domain techniques. Yin et al. (2016) designed a new type of smart washer to monitor the looseness of bolts. Experimental studies have established that the looseness of a bolt joint can be precisely detected using their device. Huo et al. (2017b) proposed a new analysis model, based on fractal contact theory, for monitoring the looseness of bolts and they applied the finite element method to determine the relationship between the actual contact area between the bolt and the plate and the peak amplitude of the focus signal. A PZT-based steel bar corrosion monitoring approach was proposed by Huo et al. (2018) utilizing the time reverse technology to monitor the corrosion of the steel bars with a high signal to noise ratio. Xiao et al. (2020) conducted an experimental study to compare the deterioration of mechanical properties of HPS specimens subjected to electrochemical corrosion and neutral salt spray accelerated corrosion. Yuan et al. (2019) proposed a novel percussion-based bolt looseness monitoring approach using intrinsic multiscale entropy analysis and back propagation neural network for practical applications of bolt looseness monitoring. Additional advanced studies on the structural health

© ASCE

617

618

Earth and Space 2022

monitoring of RC or other structures also can be found in the literatures (Kong et al., 2016; Huo et al., 2019; Song et al., 2022; Kong et al., 2013; Lu et al., 2021). .

In this study, simple bolted connection specimens are firstly pre-tested to investigate the relationship between the looseness of an un-corroded bolt and its damage index, and then the relationship between the severity of corrosion and the damage index of corroded bolts. A piezoelectric active sensing system is used to detect the damage condition of the bolted connection. Then, a large-scale two-story bolted frame structure with diagonal braces is used in a health monitoring experiment. Various conditions of bolt looseness or bolt corrosion are simulated and investigated. During the test, a sinusoidal wave is generated using a PZT actuator and an impact pulse is generated using a hammer as excitation waves. A continuous wavelet transform is carried out to process the received signal, and an energy-based damage index is used to quantify the damage status of the bolted connection. According to the experimental results, the damage index that is used herein is effective for not only monitoring the looseness of a bolt connection but also accurately diagnosing even the very slight corrosion of bolts. Therefore, the piezoelectric-based diagnostic method is effectively for monitoring the severity of bolt corrosion or degree of bolt looseness, and should be widely used in the detection of damage to bolted connections in the near future to structural safety. SETUP FOR PRE-TESTING BOLTED CONNECTION The specimens that were used in the pre-testing of bolted connections were two steel plates, each of size 144mm×50mm×2mm, that are bolted together by an M10 bolt. Three types of specimen were used in the test. The first type was a non-corrosive bolt specimen to which were applied torques of 0, 5, 8, 11, 14, and 17 Nm. The control factor of this type of specimen was the applied torque to find the relationship between bolt looseness and detected stress wave. Figure 1 presents the test setup. A piezoelectric patch was installed on the left-hand side of the specimen and used as an actuator that generated an excitation stress wave, and another was installed on the right-hand side as a receiver for detecting transmitted signals. Table 1 provides the piezoelectric strain constant coefficient for the direct piezoelectric effect in the direction normal to the PZT patch (APC International Ltd., 2012) and the piezoelectric voltage constants of the PZT patches that were used as sensors and actuators that were obtained from the catalog. An NI (National Instrument, Inc.) system was used to generate the excitation source signal for the actuator and to record signals from the sensors. Single-frequency sinusoidal signals with a duration of 10 seconds and frequencies of 1 kHz, 5 kHz, and 10 kHz were used as excitation sources for the PZT actuator. The sensing unit was low-cost and supported active sensing. Active sensing allows active excitation by the desired waveform so that distributed sensors can detect the responses. The sensor signals could be analyzed to determine many important properties of the connections, such as the severity of corrosion and loosening status.

© ASCE

619

Earth and Space 2022

Table 1. The piezoelectric constants* of the PZT patches. function

piezoelectric strain constant

piezoelectric voltage constants

d 33

g 33

430x10-12 m/V 650x10-12 m/V

25x10-3 V-m/N 14x10-3 V-m/N

Sensors Actuator *catalogue value

Figure 1. Configuration and photo of the test setup. Un-corroded specimen

Corroded specimen (0.53% w.t. loss)

Corroded specimen (0.96% w.t. loss)

Corroded specimen (3.61% w.t. loss)

Figure 2. Photographs of corroded specimens of type 2 with various weight losses. The second type of specimen was corroded after the application of a torque of 10Nm. The steel plate was coated with red lead paint to protect it against corrosion, causing only the bolt to corrode during the relevant process. The corrosion weight losses of the specimens were 0%, 0.09%, 0.18%, 0.39%, 0.53%, 0.96%, 1.68%, 2.2%, 3.25% and 3.61%. Figure 2 shows the photographs of these corroded specimens with various weight losses. The purpose of this test was to determine values of the damage index following different degrees of corrosion. In the third type of test specimen, bolts were corroded to weight losses of 0%, 4.0%, 9.5%, 12.2%, 15.4%, 27.5%, and 35.2%, and then used to fasten two steel plates to each other. A torque of 10 Nm was then applied

© ASCE

620

Earth and Space 2022

and health monitoring carried out to obtain the corresponding values of the damage index. Figure 3 displays the corroded bolts of type three at various corroded hours.

Figure 3. Photograph of corroded bolts of type 3 at various corroded hours.

DAMAGE INDEX USED IN HEALTH MONITORING The operating principle of the sensing system is that one PZT patch is used as an actuator to generate excitation waves while other distributed patches are used as sensors to detect the transmission responses. Looseness or corrosion in the bolted connection relieves stress relief in the wave propagation, and the energy of the propagation waves is attenuated accordingly. The reduction of the transmitted energy correlates with the degree of damage. The transmission energy of the sensor is given by tf

E = ∑ u 2 (ti )

(1)

ti =t0

where t0 and t f are respectively the starting time and end time of recording, and u(ti) is the sensor voltage at time ti . After the received signal is processed by continuous wavelet transform (CWT) filtering, the filtered signal is integrated to obtain the transmission energy. Using Eq.(1), the energy Eh ,k is calculated when connections are tightened or healthy at the k-th excitation frequency (k=1, 2, and 3 for 1 kHz, 5 kHz, and 10 kHz), and the energy that correspond to the damage state of connections i is denoted as Ei ,k . The square root of the sum of the squared (SRSS) values of the damage index is defined as 3

I=

∑ (E k =1

h,k

− Ei ,k ) 2 (2)

3

∑E k =1

2 h,k

The proposed damage index simply quantifies the loss of transmitted energy that is caused by the looseness or corrosion of the bolt. When the damage index is close to zero, the connection is considered to be healthy; otherwise, it is loose or damaged by corrosion. A large value of the index corresponds to more severe damage.

© ASCE

621

Earth and Space 2022

RESULTS OF PRE-TEST OF BOLTED CONNECTION The magnitude of the torque that is applied to the un-corroded bolt specimens of the first type is controlled to simulate the looseness of a bolt connection. The proposed diagnostic method and damage index are then used to evaluate the degree of looseness of the connection. Figure 4 shows the experimentally determined relationship between the applied torque and the damage index. The figure reveals that the damage index gradually increases as the applied torque decreases, but ultimately reaches a plateau at a torque of 11 Nm. because the bolt is loosened to a certain degree, the steel plates are separated by a tiny gap, changing the transmitted stress wave in the connection from a combination of compression and tension waves to a compression wave only. The experimental results show that the proposed method and damage index correctly and effectively diagnose the looseness of the bolted connection. Figure 5 presents the test results concerning the relation between the weight loss of a specimen of the second type and its damage index. The bolt in the specimen is initially tightened with a torque of 10Nm and then corroded; this situation is similar to that of a bolted connection in the natural environment. The corroded parts are the bolt, the bolt hole, and the parts of the steel plate that is in contact with the bolt. The other parts of the specimen are pre-coated with red lead paint to prevent corrosion. Figure 5 indicates that the damage index reaches unity at a very low corrosion rate, indicating that even slight corrosion of the parts of the bolt and steel plate that are in contact with each other can block the transmission of stress waves and that even very slight corrosion of a bolt connection can be detected by the active sensing method.

Damage index

0.4 0.3 0.2 0.1 0 0

3

6

9

12

15

18

Torque (N-m)

Figure 4. Relationship between damage index and applied torque for specimen of type I. Specimens of the third type have an initially corroded bolt , which is then fastened with a torque of 10Nm. Figure 6 shows the test results concerning the relatioship between the weight loss of the bolt and its damage index. It shows that the damage index increased with the weight loss up to the maximum value of around 0.2, which is not close to unity. One possible reason for this finding is that the parts of the steel plates that were in contact with the bolt were not corroded so the transmission of the

© ASCE

622

Earth and Space 2022

waves was not completely blocked. The difference between the values of the damage index in Figure 5 and Figure 6 corresponds to the corrosion of the contact area between bolt and plate.

Figure 5. Relationship between damage index and weight loss of specimens of type II.

Damage index

0.25 0.2 0.15 0.1 0.05 0 0

10

20

30

40

Rate of weight loss %

Figure 6. Relationship between damage index and weight loss for specimens of type III. MONITORING HEALTH OF BRACED STEEL FRAME WITH BOLTED CONNECTIONS The tested steel frame structure was a two-story diagonal braced frame with a height of 1000mm per floor and a span of 1280mm. Each column was a hollow cylinder with a diameter of 140mm; each beam was 65mm×65mm×10mm L-shaped steel, and each diagonal brace was 1295mm×65mm×10mm L-shaped steel. M16 bolts were used in the connections. Figure 7 displays a side view and photograph of the test frame. To prevent interference with the received signal by the excitation wave that is transmitted to the ground and reflected back to the structure, rubber layers with a thickness of 10 mm were installed under the bottom of each steel column to isolate the transmitted waves. As shown in Figure 8, the bolts were numbered from B1 to B8, and the sensors were numbered from SA1 to SA7. The actuator was marked “A”, and the impact points in the impact test were marked in the figure by asterisks.

© ASCE

623

Earth and Space 2022

Figure 7. Photograph of bolted frame.

Figure 8. Side view of frame, showing sensors and bolts. The detection of damage to the bolted connection frame was divided into two parts. The first part was the detection of bolt loosening, and the second part was the detection of the corrosion of the bolts. In the first part, bolts B1~B8 in the connections were tightened with a torque of 25Nm, and the received piezoelectric signal in this state was treated as the signal without damage. Bolts B1 to B8 were loosened one by one. The piezoelectric signals that were obtained with different numbers of loose bolts were processed and the corresponding values of the damage index calculated. In the second part, all eight bolts from B1 to B8 were corroded to weight losses of 0.0%, 0.93%, 1.8%, and 3.98%, respectively. After the corrosion process was completed, the bolts were tightened using a 25Nm torque for subsequent damage diagnosis. Single-frequency (1kHz, 5kHz, and 10kHz) sine waves were generated by the function generator and impact pulses that were generated using a hammer to provide excitation waves. The peak-to-peak voltage of each sine wave was 20 V. The hammer that was used in the impact test weighed 1kgf and fell from a fixed height of 45 cm. Since the impact forces could not be made exactly equal, an additional sensor was set up next to the position of impact and the stress wave energy

© ASCE

624

Earth and Space 2022

that was received by it was used as the denominator in normalizing the energy of energies that were received by sensors SA1-SA6.

1

SA1 SA2 SA3 SA4 SA6

0.2

Damage index

Damage index

Figure 9 presents the values of the SRSS damage index for different numbers of bolts that had been loosened under the excitation waves from the PZT actuator. SA5 malfunctioned during the test, so only the results of the other sensors are displayed. Figure 10 plots the diagnostic results concerning the stress waves that were generated by the impact pulses. Figures 9 and 10 show that the damage index gradually increased with the number of loosened bolts. Therefore, the loosening of bolted connections can be monitored using the damage index.

0.1

SA1 SA2 SA3 SA4 SA6

0.8 0.6 0.4 0.2

0

0

0

1

2

3

4

5

6

7

0

8

1

2

3

4

5

6

7

8

Number of bolts loosened

Number of bolts loosened

Figure 9. Relationship between damage index and number of bolts loosened (under sine wave excitation).

Figure 10. Relationship between damage index and number of bolts loosened (impact test).

Figures 11 and 12 plot the relationships between the damage index and degree of bolt corrosion under excitation waves that were generated by the PZT actuator and the impact of the hammer, respectively. The damage index under single-frequency sine wave excitation reaches a constant value at very slight corrosion, beyond which it is independent of the degree of corrosion, and the values of the damage index of all of the sensors are close to each other. However, in the impact test, the values of the damage index of the sensors differed markedly. The experimental results confirm that the severity of corrosion of bolted connections can be detected using the active sensing method.

Damage index

0.3

0.2 SA1 SA2 SA3 SA4 SA6

0.1

0 0

1

2

3

4

Average rate of weight loss %

Figure 11. Relationship between damage index and weight loss of bolt (under sine wave excitation).

© ASCE

625

Earth and Space 2022

Damage index

1

SA1 SA2 SA3 SA4 SA6

0.8 0.6 0.4 0.2 0 0

1

2

3

4

Average rate of weight loss %

Figure 12. Relationship between damage index and weight loss of bolt (impact test). CONCLUSION A bolted steel structure suffers from many unfavorable phenomena including corrosion, vibration, shock, and seismic excitation, resulting in a loss of strength of its bolted connections, which damages the entire structure. Therefore, monitoring the health of bolted connections is very important to ensuring structural safety. In this work a large-scale two-story bolted frame structure with a diagonal brace was used in a health monitoring experiment. Various conditions of bolt looseness and bolt corrosion are simulated and investigated. A piezoelectric active sensing system is used to detect damage bolted connections. During the test, sinusoidal waves that are generated by a PZT actuator and impact pulses that are generated by a hammer are used excitation waves. A continuous wavelet transform is carried out to process the received signal of sensors, and an energy-based damage index is used to quantify the damage status of the bolted connections. According to the experimental results, the damage index that is used in this study can be used not only to monitor the looseness of bolt connections but also to diagnose accurately even the very slight corrosion of bolts. This work demonstrates that the piezoelectric-based diagnostic method is effectively for monitoring the degree of bolt corrosion or the degree of bolt looseness in bolted steel structure. Therefore, this technique should be widely applied to the detection of damage to bolted connections in the near future to improve structural safety and our understanding thereof. However, the problem to separate the damages from bolt looseness or from the bolt corrosion is of importance when the two kinds of damages were coupled together. The method to decouple the sensing signal should be conducted in further research. ACKNOWLEDGMENTS The authors would like to thank the Ministry of Science and Technology of the Republic of China, Taiwan, for financially supporting this research under contract MOST 109-2625-M-027-001-MY2. The facilities for research provided by National Center for Research on Earthquake Engineering are also highly appreciated.

© ASCE

Earth and Space 2022

REFERENCES APC

International Ltd. (2002). “Piezoelectric Ceramics: Principles and Applications.” APC International Ltd., Pennsylvania, USA. Feng, Q., Kong, Q., Jiang, J., Liang, Y., and Song, G. (2017). “Detection of Interfacial Debonding in a Rubber-Steel-Layered Structure Using Active Sensing Enabled by Embedded Piezoceramic Transducers.” Sensors 17(9), 2001. Huo, L.S., Chen, D., Liang, Y., Li, H., Feng, X., and Song, G. (2017a). “Impedance based bolt pre-load monitoring using piezoceramic smart washer.” Smart Materials & Structures, 26(5), 057004. Huo, L., Cheng, H., Kong, Q. and Chen, X., (2019). “Bond-slip monitoring of concrete structures using smart sensors—A review.” Sensors, 19(5), 1231. Huo, L., Li, C., Jiang, T. and Li, H.N. (2018). “Feasibility study of steel bar corrosion monitoring using a piezoceramic transducer enabled time reversal method.” Applied Sciences, 8(11), 2304. Huo, L. S., Wang, F. R., Li, H. N., and Song, G. (2017b). “A fractal contact theory based model for bolted connection looseness monitoring using piezoceramic transducers.” Smart Materials & Structures, 26, (10), 9. Kong, Q., Hou, S., Ji, Q., Mo, Y.L. and Song, G. (2013). “Very early age concrete hydration characterization monitoring using piezoceramic based smart aggregates.” Smart Materials and Structures, 22(8), 085025. Kong, Q., Robert, R.H., Silva, P. and Mo, Y.L. (2016). “Cyclic crack monitoring of a reinforced concrete column under simulated pseudo-dynamic loading using piezoceramic-based smart aggregates.” Applied sciences, 6(11), 341. Kong, Q., Zhu, J., S.C.M., Song, G. (2018). “Tapping and listening: A new approach to bolt looseness monitoring.” Smart Materials and Structures, 27(7), 07LT02. Lu, G., Wang, Q., Song, H., Liu, Z. and Wang, T. (2021). “Actuating Performance Analysis of a New Smart Aggregate Using Piezoceramic Stack.” Applied Sciences, 11(20), 9599. Soh, C.K., Tseng., K., Bhalla., S., Gupta., A. (2000). “Performance of smart piezoceramic patches in health monitoring of RC bridge.” Smart materials & structures, 9, (4), 533. Song, H., Xiang, M., Lu, G. and Wang, T. (2022). “Singular spectrum analysis and fuzzy entropy-based damage detection on a thin aluminum plate by using PZTs.” Smart Materials and Structures, 31, 035015. Wang, B., Huo, L., Chen, D., Li, W. and Song, G. (2017). “Impedance-Based PreStress Monitoring of Rock Bolts Using a Piezoceramic-Based Smart Washer—A Feasibility Study.” Sensors, 17(2). Wang, T., Song, G., Wang, Z., and Li, Y. (2013). “Proof-of-concept study of monitoring bolt connection status using a piezoelectric based active sensing method.” Smart Materials and Structures 22(8):087001. Wang, F., Huo, L., S.C.M., Song, G. (2018). “A novel fractal contactelectromechanical impedance model for quantitative monitoring of bolted joint looseness.” IEEE Access, 6, 40212-40220.

© ASCE

626

Earth and Space 2022

Wang, F., and Song, G. (2019). “Bolt early looseness monitoring using modified vibro-acoustic modulation by time-reversal.” Mechanical Systems and Signal Processing, 130, 349-360. Wang, F. and Song, G. (2020). “Monitoring of multi-bolt connection looseness using a novel vibro-acoustic method.” Nonlinear Dynamics, 100(1), 243-254. Xiao, L., Peng, J., Zhang, J., Ma, Y. and Cai, C.S. (2020). “Comparative assessment of mechanical properties of HPS between electrochemical corrosion and spray corrosion.” Construction and Building Materials, 237, 117735. Yao, P., Kong, Q., Xu, K., Jiang, T., Huo, L. and Song, G. (2016). “Structural health monitoring of multi-spot welded joints using a lead zirconate titanate based active sensing approach.” Smart Materials and Structures, 25, (1), 015031. Yin, H. Y., Wang, T., Yang, D., Liu, S. P., Shao, J. H., and Li, Y. R. (2016). “A Smart Washer for Bolt Looseness Monitoring Based on Piezoelectric Active Sensing Method.” Applied Sciences-Basel, 6, (11), 10. Yuan, R., Lv, Y., Kong, Q., Song, G. (2019). “Percussion-based bolt looseness monitoring using intrinsic multiscale entropy analysis and BP neural network.” Smart Materials and Structures, 28(12), 125001.

© ASCE

627

628

Earth and Space 2022

Seismic and Resilient Property Analysis of SMA-Based Replaceable BRBs Q. Y. Pan1; S. Yan2; and X. Su3 1

School of Civil Engineering, Shenyang Jianzhu Univ., Shenyang Liaoning, China. Email: [email protected] 2 School of Civil Engineering, Shenyang Jianzhu Univ., Shenyang Liaoning, China. Email: [email protected] 3 School of Civil Engineering, Shenyang Jianzhu Univ., Shenyang Liaoning, China. Email: [email protected] ABSTRACT A novel type of buckling restrained brace (BRB) utilized the super-elasticity of shape memory alloys (SMAs) is developed, and a new assembly of restraint parts is proposed to solve the problem of poor restorability and non-replacement of traditional BRBs. The novel BRB components are simplified and modeled by using the ABAQUS platform to study the seismic and resilient properties. The axial carrying capacity and hysteresis energy consumption capacity of both the novel BRB components and the traditional ones are analyzed and compared, respectively. The numerical results show that the new assembly of restraint parts of the novel BRB component can improve the hysteresis energy-consuming capacity about 10% under the same condition as the core energy-consuming components. The SMA-based novel BRB component as the core energy-consuming one behaves typically as the “flag-shaped” hysteresis curves, and the SMA-based and replaceable BRB components have good seismic and resilient performances. The influence of gap measurement between the restraint component and the energy-consuming component on the performance of BRB is also analyzed. INTRODUCTION Traditional braced frames are generally used to solve the problem of pure frame structure with unfavorable earthquake resistance such as the small lateral stiffness and excessive inter-story drifts under lateral loading. However, the traditional braced frames still have a problem that it is prone to compressive buckling under lateral loading. (1971) proposed BRBs in the 1970s to solve the problem. The frame structure with BRBs greatly improves the ability of resisting horizontal earthquakes without buckling before the whole section yielding. The traditional BRBs are generally composed of two parts: core elements and constraint ones. The core elements generally choose low-yield-point steel to bear axial load. However, although the traditional BRBs can improve the structural rigidity and seismic capacity, they still have the problem of large residual deformation after strong earthquakes, affecting the maintenance and continued use of the structure in the future. Therefore, a new type of BRB with performances of good recoverability and replaceability as well as high-efficiency energy consumption has become the focus of research and development, and some corresponding research results have been achieved. The self-centering BRBs (SC-BRBs), which generally consist of an energy-consuming system (ECS) and a self-centering system (SCS), are currently focused. The SCS is mainly based on prestressed tendons / cables, high-strength springs / disc springs, SMA wires / cables, etc. The ECS is mainly based on friction damping devices, low yield-point steel, etc. By combining

© ASCE

Earth and Space 2022

the two systems with different structures, a variety of SC-BRBs can be developed. Self-centering dampers can also be designed through this path. The boundary between self-centering dampers and SC-BRBs is gradually blurred. SMA as a novel functional material has found increasing applications in seismic absorbing or isolation fields. These studies and applications have guiding significance for the design of SCBRB. Song et al (2006) researched that SMA materials can be applicated for passive, active and semi-active controls of civil structures. Hong et al (2020) utilized superelastic SMA wires to confine concrete cylinders to enhance their axial compressive behavior. The results verified that superelastic SMA wires can increase the axial loading capacity and enhance deformation performance of concrete columns. De Domenico et al (2020) presents a base isolation layout that combines low-friction curved surface sliders (CSSs) with SMA gap dampers (SMAGDs). The SMAGDs introduce additional stiffening and energy dissipation to the isolation system only when the displacement of the CSS exceeds a certain threshold or gap displacement. Chen et al (2013) proposed an energy-consuming enhanced SMA damper to make the SMA wire produce greater axial displacement. The damper uses the principle of leverage to amplify the tensile deformation of the SMA wire. The research results show that the hysteresis loops of the damper are plump. Zhou et al (2018) present a cable-stayed bridges using SMA SC-damper. The piston structure with three limit points ensures that one of the two SMA wires of the damper is always in tension. The experimental results demonstrated that the SC-damper reduced tower accelerations and relative displacements. Enferadi et al (2019) used SMA dampers to control the jacket platform oscillations and the numerical analysis method is used to prove the effectiveness of vibration control. The short SMA bars in the damper compressed by piston and the short lengths of SMA bars made the buckling effects can be ignored. Viscoelastic materials or rubber in the damper can increase the damper’s dissipation capability. The numerical results shown that the damper can work as semiactive devices and control the jacket platform’s vibration. Sheikhi et al (2021) investigated the behavior of a natural rubber bearing system with U-shaped dampers by numerical analysis. Scholars have made achievements in the direction of SC-BRBs. Li et al (2008) proposed a damper based on superelastic SMA wires. The experimental research shows that the damper has good energy-consuming and self-centering ability under cyclic loading. Zhu et al (2008) introduced a self-centering friction damping brace using SMA wires as the SCS and a friction damping washer as the ECS. The brace can be adjusted by preloading the SMA wires and the performance of the brace can be changed by the friction compressive strength. Shi et al (2021) studied mechanical properties of SMA cables and added an SCS based on prestressed SMA cables to the traditional BRB. The performance of the brace was studied through theoretical analysis and numerical simulation. Christopoulos et al (2008) first introduced an SCS of a dual tube with left and right movable endplates, which realized that the self-centering elements in the system were always in a state of further tension no matter whether the brace was under tension or compression. Miller et al (2012) used a pre-tensioned SMA rod as an SCS anchored on the movable endplate. The dual tube of the SCS is welded in the core transition section of the traditional steel BRB to realize the parallel connection of the SCS and BRB. The BRB provides initial axial bearing capacity, energy dissipation capacity and deformability. The SCS adds additional self-centering capacity and initial stiffness to the brace system. Qiu et al (2020) proposed a new type of SC-BRB with two SMA bars as an SCS. The BRB allowing one SMA bar is subjected to tension and the other SMA bar concurrently in compression. That make up for the asymmetrical hysteretic behavior of individual SMA bar. Xu et al (2018) proposed a SCBRB that uses pre-compressed disc springs as an SCS and a friction device as an ECS. When

© ASCE

629

630

Earth and Space 2022

the friction device is not working, the static friction provides the axial bearing capacity, and after the friction device starts working, the friction device provides the energy consuming capacity. The pre-compressed disc spring is in a compressed state under tension and compression, providing self-reset capability for the brace. Qian et al (2016) experimentally investigates shaking table tests of a new superelastic SMA friction damper (SSMAFD). The damper with pre-tensioned SMA wires as the SCS and friction devices as the ECS. The experimental results demonstrated that the SSMAFD was effective in suppressing the dynamic response of the building structure subjected to strong earthquakes by dissipating a large portion of the energy. SC-BRBs have both good self-centering capability and hysteretic energy dissipation capability. However, friction dampers and low-yield-point steels often used as ECSs are difficult detecting and replacing after earthquakes, being unconducive to the rapid recovery of structural functions. Therefore, a novel type of BRB utilizing the SMAs is developed in this paper to solve abovementioned problems. The main contribution is to provide greater self-centering forces and better hysteretic energy consumption than the traditional one by using SMA steel angle as the SCS and the ECS of the novel BRB, improving the structural resilient capacity by using the replaceable core component. The ABAQUS platform is used to model the replaceable BRB. The axial bearing capacity, hysteresis energy-consuming capacity, and failure modes are analyzed. The seismic and resilient properties of the new functionally recoverable BRB components are studied, focusing on the restraint effect of the constrained parts on the core SMA components. A parameter analysis is also performed on the gap measurement between the constrained component and the core component to explore the influence of machining error on the performance of the novel BRB. NOVEL BRB COMPONENTS AND ITS NONLINEARITY Components of the Novel BRBs The novel BRB is composed of inner-restraint plunger component, inner-restraint sleeve components, angle SMA components, outer-restraint components, high-strength springs, and highstrength bolts. The angle SMA-based replaceable BRB and the specific assembly are shown in Figure 1 and Figure 2, respectively. A retractable inner-restraint part is formed by two inner-restraint sleeves, one inner-restraint plunger, and two high-strength springs. Two inner-restraint sleeves are installed on the inner-restraint plunger, and sleeves and plunger are connected by high-strength springs. Four SMA angles are fixed on the inner-restraint part by high-strength bolts. Then, two U-shaped outer-restraint components are fixed outside the SMA angles by high-strength bolts. The specific working principle and design method have been described in the literature (2021).

Figure 1. Angle SMA-based replaceable BRB.

© ASCE

Figure 2. Assembly of the novel BRB.

631

Earth and Space 2022

Nonlinearity of the Novel BRB There are three kinds of nonlinearity issues in the novel BRB under loading, namely, material nonlinearity, geometric nonlinearity, and contact nonlinearity. Material nonlinearity is related to the characteristics of the material itself, and the effect on component performance can be obtained through experiments and model verification. The material nonlinearity of the novel BRB mainly comes from the deformation of superelastic SMA angles and steel components. The nonlinear deformation of SMA angles can be described by the SMA constitutive model which is based on the phenomenological thermodynamic constitutive model in this paper, and the various parameters refer to the SMA bar test data of Dong (2018). The constitutive model of steel mainly includes bilinear kinematic hardening model and combined hardening model. The bilinear kinematic hardening model can be used to reduce the amount of calculation when the material does not have large strain or does not bear cyclic load. However, the bilinear kinematic hardening model defaults that the yield stress surface of the material is fixed and does not expand with the increase of plastic strain during the plastic deformation process, which is inconsistent with the actual situation (2012). The combined hardening model (combination of kinematic hardening model and cyclic hardening model) used in this paper can solve this problem. The model can consider the deformation under the influence of multiple back stresses, which is more aligned with the actual performance of BRBs. Due to the complexity of the novel BRB structure, the influence of the geometric nonlinearity caused by both buckling of SMA angles and the nonlinear contact caused by the contact between the steel angles and the restrained component after buckling on the performance of novel BRBs are difficult to obtain through theoretical calculations. Therefore, the ABAQUS platform is used for numerical simulation analysis of the above-mentioned issues. ESTABLISHMENT OF NOVEL BRB FINITE ELEMENT MODEL Model Verification of Traditional BRBs Selection of test components. A traditional BRB model verification was conducted to better simulate the buckling of BRB core components and verify the restraint ability of the novel BRB. The traditional BRB test result (Guo et al, (2016) was selected for model verification. The traditional BRB with Q345 steel casting cross-section inner-core as the ECS and steel outersleeve filled with mortar as the restraint component was selected. The basic mechanical property of testing steel is shown in Table 1. Table 1. Basic Mechanical Property of Testing Steel. Component

Materials

Inner-core Outer-sleeve

Q345 steel Q235 steel

Yield stress / MPa 350 244

Ultimate stress / MPa 575 385

elasticity modulus / GPa 199.6 204

Yield ratio 1.64 1.58

Yield strain /% 0.175 0.120

Elongation /% 24.5 32.0

Modeling method of model verification. The dimensions of the tested traditional BRB is shown in Figure 3. The model includes a Q345 steel energy-consuming inner-core, a Q235 steel outer-sleeve, and mortar filling. When modeling, the large deformation caused by axial quasi-static loading and the linear contact between core components and mortar filling are considered. The

© ASCE

632

Earth and Space 2022

measurement of the core part is the same as that of the test part, and the 1 mm unbonded layer is simplified to 1 mm gap. After carried out the BUCKLE analysis on the core components and extracted the buckling displacement file, the displacement file is introduced into the model as the initial defect for simulation. The loading amplitude is consistent with the test loading amplitude. Material constitutive laws of model verification. The constitutive law of the Q345 steel adopts the combined hardening model, and the specific parameters are shown in Table 2. Table 2. Q345 Steel Combined Hardening Model Parameters. Parameters

fy / MPa

𝐶1 / MPa

𝛾1

𝐶2 / MPa

𝛾2

Value

230

5600

25

10700

1000

𝑄∞ / MPa 100

𝑏 5

The filling mortar adopts concrete damage plastic constitutive model, the specific parameters are shown in Table 3. Table 3. Concrete Damage Plastic Constitutive Model Parameters. Parameters

Young’s modulus / GPa

Poisson’s ratio

Dilation angle /°

Eccentricity

𝑓b0 / 𝑓c0

K

Viscosity parameter

Value

44.56

0.2

30

0.1

1.16

0.6667

0.0005

Load / kN

Result of model verification. The comparison of the simulation curve with the test one is shown in Figure 4. The figure shows that the simulation curve is in good agreement with the test one. The secant stiffness and the bearing capacity are very close, validating that the simulation method can be used for BRB simulation.

600 400 200 0 -200 -400 -600 -800

test simulation

-20-15-10 -5 0 5 10 15 20 Displacement / mm

Figure 3. Dimensions of test traditional BRB (Guo et al, 2016).

© ASCE

Figure 4. Comparison of simulation curve with the test one.

633

Earth and Space 2022

Modeling Method of Novel BRB and Comparison Components The replaceable SMA angles in the core of the novel BRB serve as SCS and ECS at the same time, providing the brace function with self-centering force and hysteretic energy-consuming capacity. High-strength springs provide axial stiffness and make up for the deficiency of self-centering force during phase transformation of SMA angles. Based on the previous theoretical structural parameter analysis of the novel BRB (Yan et al, (2021), the mechanical properties and restraint effect of SMA angles in the core energy-consuming section of the new BRB are mainly studied. A solid model with C3D8R element type were adopted to model the novel BRB. To simplify the model, high-strength springs and bolts are neglected, and tie-rod is adopted to connect SMA angles and inner-restrained sleeve. The total length of the core energy-consuming components is 900 mm, and the core section consists of four 500 mm long steel angles with section measurement of L 30 mm × 30 mm × 3 mm. The thickness of constraint material is 6 mm. The lengths of the outer-restrained and the inner-restrained are 700 mm and 600 mm, respectively. The modeling and mesh division of each member are shown in Figure 5. A gap of 1 mm between the restraint member and the core member is set.

Figure 5. Modeling of new BRB. Two comparison models are established. One keeps the structural form of the novel BRB, but replacing the SMA with the Q235 steel angles. The other is the traditional BRB with cross section Q235 steel as the core ECS and steel sleeve filled with concrete as the restrained component. The variable parameter pairs of three BRB components are shown in Table 4. Table 4. Comparison of Variables and Parameters of BRBs Modeling. Model Novel BRB (SMA) Novel BRB (Q235) Traditional BRB (Q235)

Material of the ECS SMA Q235 Q235

Length of the ECS / mm 500 500 900

Loading displacement / mm 45 20 20

Sectional area / mm2 684 684 684

Material Constitutive Model of the ECS The super-elastic constitutive model of ABAQUS platform is adopted for SMA angles, and the model parameters are shown in Table 5. The combined hardening model is adopted for Q235 steel, and the model parameters are shown in Table 6.

© ASCE

634

Earth and Space 2022

Table 5. Constitutive Model Parameters for SMA Angles. Parameters

𝐸𝐴 /GPa

𝐸𝑀 /GPa

Value

78

78

′

𝜎𝑀𝑠 /MPa 340

′

𝜎𝑀𝑓 /MPa 700

′

𝜎𝐴𝑠 /MPa 300

′

𝜎𝐴𝑓 /MPa

𝜀𝐿

120

0.063

Table 6. Combined Hardening Model Parameters for Q235 Steel. Parameters

fy/MPa

𝐶1 /MPa

𝛾1

𝐶2 /MPa

𝛾2

𝑄∞ /MPa

𝑏

Value

157

4600

25

9800

1000

100

5

Influence of Gap Measurements on Seismic Performance of Components The gap measurements between SMA angles and restrained components are changed by altering the measurements of restrained components to explore the influence of the gap measure on the performance of components. The quasi-static analysis of new BRB components with gap measurements of 1 mm, 2 mm, and 3 mm is performed respectively. The hysteresis curves and skeleton curves of the novel BRB are compared for understanding the reasons of performance change. SIMULATION RESULT AND DISCUSSION Analysis of Axial Compression Performance

Load / kN

Results of axial compression simulation. The axial compression performance differences between the restrained or unrestrained SMA-based ECS and two comparison models are compared from the curves in Figure 6. When the ECS of SMA angles are unconstrained and the compressive load reaches 180 kN, the compressive bearing capacity suddenly drops and soon loses its bearing capacity. On the contrary, when the ECS are restrained, the novel BRB shows good compressive bearing capacity. Compared with the BRB with Q235 steel as the ECS, the novel BRB with SMA angles as the ECS has higher axial compressive bearing capacity under axial loading. Both the traditional BRB and the novel BRB with the same sectional area and materials (Q235 steel) of ECS show a decrease in compressive bearing capacity when the load reaches 180 kN, and they soon enter the yield state. However, the difference is that the decrease in compressive bearing capacity of novel BRB is not obvious.

700 600 500 400 300 200 100 0 -100

SMAcore novel BRB Q235core novel BRB Q235core traditional BRB SMAcore without brace

0

10

20

30

40

Displacement / mm

Figure 6. Comparison of axial compression performance of BRBs. © ASCE

Figure 7. Stress nephogram of two comparative BRB models.

635

Earth and Space 2022

Discussion of axial compression simulation. Benefiting from the material mechanical properties advantage of superelastic SMA, the SMA angle-based novel BRB has higher axial compressive bearing capacity than the others. Comparing the stress nephogram of two comparative BRB models with Q235 steel as the ECS in Figure 7, the loading end of the traditional BRB is buckling at the initial stage of compression, which leads to the rapid decline of the compressive bearing capacity. Traditional BRBs need to reserve large gaps at both ends to ensure the normal operation of the ECS, which leads to the loading end buckling at the initial stage of loading. Compared with the curve in Figure 6, it can be found that after the loading end contacts with the restraint member, the buckling trend is restrained, and the bearing capacity is gradually increased. With the same material and sectional area, the ECS of the novel BRB has a large polar moment of inertia, and the whole section is well restrained by the restraint members, so it is not easy to have low order buckling and the reduction of compressive bearing capacity is not significantly. Analysis of Hysteretic Performance Results of quasi-static simulation. A quasi-static analysis is carried out on the novel BRB and two comparative models, and the quasi-static loading system is shown in Figure 8. The simulation results are shown in Figure 9.

60 40 20 0 -20 -40 -60

Load / kN

Displacement / mm

Q235core traditional BRB Q235core novel BRB SMAcore novel BRB

0.0 0.1 0.2 0.3 0.4 0.5 0.6

800 600 400 200 0 -200 -400 -600 -800 -60 -40 -20

0

20

40

60

Displacement / mm

Time of analysis step / s Figure 8. Quasi-static loading system.

Figure 9. Quasi-static simulation results of different BRBs.

Three BRB models all show full hysteretic curves under quasi-static loading. Comparing the three curves, the novel BRB with SMA angles as the ECS behaves a “flag-shaped” hysteretic curves and has larger axial displacement and higher self-centering force than the BRB with Q235 steel as the ECS. Like the results of axial compression simulation, both with Q235 steel as the ECS and with the same sectional area, the hysteretic energy dissipation capacity of the novel BRB is better, and the bearing capacity is improved by about 10 %. Discussion of quasi-static loading simulation. Benefiting from the good mechanical properties of SMA and the restraint components providing enough restraints for the ECS, the

© ASCE

636

Earth and Space 2022

SMA-based novel BRB has better self-centering ability, greater axial displacement, and smaller residual deformation than the traditional BRB. However, the energy dissipation capacity under small displacement is weaker than that of traditional BRB. Like the results of axial compression simulation, the buckling at the loading end has a negative impact on the hysteretic energy dissipation capacity of traditional BRB. Analysis of the Influence of Gap Measurement on Novel BRBs

1mm gap 800 2mm gap 600 3mm gap 400 200 0 -200 -400 -600 Fig.10b -800 -1000 -60 -40 -20 0 20 40 60 Displacement / mm (a)

Load / kN

Load / kN

Comparing the hysteretic curves of SMA angle-based novel BRBs with different gap measurements (Figure 10(a)), the hysteretic curves have similar shapes when the gap measurements are 1 mm and 2 mm, respectively. However, the hysteresis loop of the novel BRB in the compression section is reduced significantly, and the stiffness is changing constantly. Then, the compression buckling sections in the 5 th, 8th, 11th and 14th lap hysteretic curves with a gap measurement of 3 mm were extracted for analysis (Figure 10(b)). In the 5th lap curve, the decline of the compression bearing capacity of the model first appeared when the displacement reached 5 mm, and the bearing capacity started to rise again in a short time, showing a buckling occurs at the moment. The 3mm gap BRB buckling section of different laps The 5th lap -180 The 8th lap -200 The 11th lap -220 The 14th lap -240 -260 -280 -300 -320 -340 -20 -16 -12 -8 -4

Displacement / mm (b)

Figure 10. Analysis of hysteretic curves of SMA-based novel BRBs with different gap measurements. (a) Hysteretic curves. (b) Buckling in detail. It shows that the restraint component prevents the ECS from continuing buckling out of the plane. With the increase of the number of laps (maximum displacement), the axial displacement corresponding to the first decline of peak load in each lap is increasing, and the bearing capacity tends to be stable gradually. Based on the analysis and combined with the stress nephogram, the initial buckling is low order type with the rapid decreasing of bearing capacity. With the increase of axial displacement, the ECS gradually enters a higher order buckling mode. With the increasing of lap number, the ECS enters the buckling state earlier and produces higher order buckling modes faster than before. Therefore, with the increase of the number of loading cycles, the buckling section of the hysteretic curve becomes stable.

© ASCE

637

Earth and Space 2022

800 600 400 Fig. 11b 200 0 -200 -400 -600 -800 -1000 -60 -40 -20 0

-200 -250

1mm gap 2mm gap 3mm gap

Load / kN

Load / kN

The skeleton curves of novel BRBs with different gap measurements are compared in Figure 11(a). Like hysteretic curves, the skeleton curves are almost the same when the gap measurements are 1 mm and 2 mm, respectively. The strength declines significantly when the gap measurement is 3 mm and the displacement of the ECS reaches 4 mm. According to the analysis of loading lap change in Figure 11(b), the strength of BRB also gradually increases with the increase of maximum displacement. When the peak displacement reaches 30 mm, the load is close to the strength when the gap measurement is 2 mm, which shows that the generation of higher order buckling modes in the later stage can improve the bearing capacity of members.

The 5th lap The 8th lap

-300 The 11th lap -350

1mm gap 2mm gap 3mm gap

-400 -450

20 40 60

-30 -25 -20 -15 -10 -5

Displacement / mm

0

Displacement / mm

(a)

(b)

400 300 200 100 0 -100 -200 -300 -400

1mm gap 3mm gap

2mm gap

Load / kN

Load / kN

Figure 11. Analysis of skeleton curves of SMA based novel BRBs with different gap measurements. (a) Skeleton curves. (b) Strength decline in detail.

-20 -10

0

Fig. 12b 10 20

Displacement / mm (a)

The 5th lap The 11th lap 0 -50 -100 -150 -200 -250 -300 -350 -20 -10

The 8th lap The 14th lap

0

10

20

Displacement / mm (b)

Figure 12. Analysis of hysteretic curves of Q235 steel based novel BRBs with different gap measurements. (a) Hysteretic curves. (b) Strength decline in detail. The hysteretic curves of novel BRB with Q235 steel as the ECS, which is like that with SMA as the ECS, is shown in Figure 12. The hysteretic curve fluctuates in the compression section and the hysteretic loop area of the hysteretic curve is obviously reduced when the gap © ASCE

638

Earth and Space 2022

1mm gap 3mm gap

400 300 200 100 0 -100 -200 -300 -400

2mm gap

Load / kN

Load / kN

measurement is 3 mm. With the increase of the number of loading cycles, the hysteretic curves of compression buckling sections are extracted and compared. Like the SMA based new BRB, the buckling section of the hysteretic curve becomes stable with the increase of loading cycles. From the skeleton curves shown in Figure 13, the traditional BRB bearing capacity of compression section degrades obviously. However, compared with the novel BRB with SMA as the ECS, due to the limitation of material properties, after high order buckling, the performance of Q235 steel angles fails to reach normal performance gradually with the increase of the number of cycles.

Fig. 13b -20

-10

0

10

Displacement / mm (a)

20

-160 The 5th lap -180 The 8th lap -200 -220 -240 1mm gap -260 2mm gap -280 3mm gap -300 -10 -8 -6 -4 -2 0 Displacement / mm (b)

Figure 13. Analysis of skeleton curves of Q235 steel based novel BRBs with different gap measurements. (a) Skeleton curves. (b) Strength decline in detail. Through the analysis above, the allowable manufacturing error of the gap measurement between the ECS and the restrained members of novel BRB cannot exceed 3 mm, otherwise it will decrease the mechanical property of the novel BRB. CONCLUSIONS After the finite element simulation, the SMA-based replaceable BRB shows better selfcentering ability, greater axial displacement, and smaller residual deformation than the traditional BRB. The novel BRB, which restrains the whole ECS inside restraint members, can fully play the function of the mechanical properties of the ECS materials. Mechanical properties of the novel BRB is not affected, when the gap measurement between the ECS and the restrained member of novel BRB less than 3 mm. ACKNOWLEDGMENTS This work was partially funded by National Key R&D Program of China with grant No. 2017YFC1503106 and Science & Technology Project of China Energy Engineering Group Planning and Engineering Co., Ltd with grant No. GSKJ2-T02-2020.

© ASCE

Earth and Space 2022

REFERENCES Chen, Y., Lv, X. L. and Jiang, H. (2013). “Design and hysteretic energy analysis of new enhanced energy dissipation SMA damper.” Journal of Central South University (Science and Technology), 2527-2536. (in Chinese) Christopoulos, C., Tremblay, R., Kim H. J. and Lacerte, M. (2008). “Self-centering energy dissipative bracing system for the seismic resistance of structures: Development and validation.” J. Struct. Eng., 134(1), 96-107. De Domenico, D., Gandelli, E. and Quaglini, V. (2020). “Adaptive isolation system combining low-friction sliding pendulum bearings and SMA-based gap dampers.” Engineering Structures, 212, 110536. Dong, J. Z. (2018). “Experimental research on the self-centering device based on large diameter SMA bar.” Structural Engineers, 34(4), 101-108. (in Chinese) Enferadi, M. H., Ghasemi, M. R. and Shabakhty, N. (2019). “Wave-induced vibration control of offshore jacket platforms through SMA dampers.” Applied Ocean Research, 90, 101848. Guo, L. and Wu, J. (2016). “Experimental study on the seismic performance of casting cruciform-shape buckling-restrained brace.” Progress in Steel Building Structures, 18(3), 10-17. (in Chinese) Hong, C., Qian, H. and Song, G. (2020). “Uniaxial compressive behavior of concrete columns confined with superelastic shape memory alloy wires.” Materials, 13(5), 1227. Li, H. N., Qian, H., Song, G. and Gao, D. W. (2008). “A type of shape memory alloy damper: design, experiment and numerical simulation.” Journal of Vibration Engineering, 21(2), 179-184. (in Chinese) Miller, D. J., Fahnestock, L. A. and Eatherton, M. R. (2012). “Development and experimental validation of a nickel–titanium shape memory alloy self-centering bucklingrestrained brace.” Eng. Struct., 40, 288-298. Qian, H., Li, H. and Song, G. (2016). “Experimental investigations of building structure with a superelastic shape memory alloy friction damper subject to seismic loads.” Smart Materials and Structures, 25(12), 125026. Qiu, C., Fang, C., Liang, D., Du, X. and Yam, M. C. (2020). “Behavior and application of self-centering dampers equipped with buckling-restrained SMA bars.” Smart Materials and Structures, 29(3), 035009. Sheikhi, J., Fathi, M., Rahnavard, R. and Napolitano, R. (2021). “Numerical analysis of natural rubber bearing equipped with steel and shape memory alloys dampers.” Structures, 32, 1839-1855 Shi, Y. F., Qian, H., Kang, L. P., Li, Z. G. and Xia, L. K. (2021). “Cyclic behavior of superelastic SMA cable and its application in an innovative self-centering BRB.” Smart Mater. Struct., 30, 095019 (25pp). Song, G., Ma, N. and Li, H.-N. (2006). “Applications of shape memory alloys in civil structures.” Engineering structures, 28(9), 1266-1274. Xu, L. H., Xie, X. S. and Li, Z. X. (2018). “A self-centering brace with superior energy dissipation capability: development and experimental study.” Smart Mater. Struct., 27, 095017 (17pp).

© ASCE

639

Earth and Space 2022

Yan, S., Pan, Q. Y. and Su, X. X. (2021). “Design method of SMA property-based replaceable BRBs.” Journal of Shenyang Jianzhu University (Natural Science) 37(3), 402-410. (in Chinese) Yang, C. M., Li, H. N., Toshio, M. and Hao, X. Y. (2012). “Finite element simulation and hysteretic-performance analysis of buckling-restrained braces.” Journal of Disaster Prevention and Mitigation Engineering, 32(2), 145-151. (in Chinese) Yoshino, T. and Karino, Y. (1971). “Experimental study on shear wall with braces: Part 2. Summaries of technical papers of annual meeting.” Structural Engineering Fascicle, Architectural Institute of Japan, 11, 403-404. Zhu, S. H. and Zhang, Y. F. (2008). “Seismic analysis of concentrically braced frame systems with self-centering friction damping braces.” J. Struct. Eng., 134(1), 121-131.

© ASCE

640

641

Earth and Space 2022

Seismic Evaluation of Lava Tubes Subjected to Moonquakes Hamed Seifamiri1, Pooneh Maghoul1, Richard Boudreault2, Najib Bouaanani1, and Roberto de Moraes3 1

Department of Civil, Geological and Mining Engineering, Polytechnique Montreal, Montreal, Quebec, H3C 3A7 2 Canadian Space Mining Corporation, Toronto, Ontario, M4G 1Z4 3 AECOM, Burnaby, British Columbia, Canada Email: [email protected]; [email protected]; [email protected]; [email protected]; [email protected] ABSTRACT In this paper, the seismic response of lava tubes was estimated. Several dimensionless numerical analyses were carried out for various dimensionless geometries. The effects of key influencing factors such as the depth of the lava tubes, the gravitational effect of the Moon, and the frequency content of the incident motion on the dynamic response of lava tubes were studied. The findings can be useful in geotechnical moonquake engineering and lunar seismology and provide some practical insights into the design of future moonquakeresilient human habitats on the Moon. INTRODUCTION Underground caves such as lava tubes are suitable candidates for permanent human bases on the Moon since they provide instant protection from harsh environments. According to GRAIL, SELENE, and other space missions, the diameter, and length of these tunnels could be several kilometers (Haruyama et al., 2010; Haruyama et al., 2009; Robinson et al., 2012; Robinson et al., 2010). The search for lava tubes and underground cavities on the Moon has been one of the core activities of lunar missions. In such lunar investigations, the presence of semi-collapsed lava tubes and intact lava tubes, which are substantially larger than the lava tubes found on Earth, was unfolded (Daga et al., 2009; Sauro et al., 2020). Several studies on the stability of lunar lava tubes of various dimensions and geometry have been conducted in the past (Blair et al., 2017; Modiriasari, Boener, et al., 2019; Modiriasari, Theinat, et al., 2019; Oberbeck, 1969; Theinat et al., 2018). However, the lunar lava tubes are not thoroughly studied against seismic waves. The Moon experiences a variety of seismic events. Based on the signal's waveform, frequency content, amplitude,

© ASCE

Earth and Space 2022

and duration of reverberation, seismic occurrences on the Moon were categorized into four groups: deep moonquakes, shallow moonquakes, meteorite impacts, and thermal moonquakes. The first seismic waves on the Moon were recorded during the APOLLO missions, and hence seismic waves on the Moon have received attention (Cooper & Kovach, 1975; Duennebier & Sutton, 1974; Horvath et al., 1980). It is also well-known that the lunar regolith at shallow layers has a very low damping ratio, likely due to the absence of liquid (Dainty et al., 1974). Also, seismic waves recorded on the Moon's surface had a high frequency and a long reverberation period (Dainty & Toksöz, 1981; Garcia et al., 2019). Furthermore, reported waves were smaller in magnitude than earthquakes. The most powerful moonquakes are shallow ones that can eventually cause significant damages and regolith mass instabilities. Various studies have been conducted to explore the geometrical and geotechnical effects on the seismic response of underground spaces on Earth. These studies have revealed that the parameters impacting the seismic response of the underground spaces depend on the material properties, geometric and topographic configurations, and the seismic excitation wave (Luco & De Barros, 1994; Smerzini et al., 2009). The one-dimensional seismic site effect on the Moon has been recently studied by Amini et al., )2021). Also, the effect of indirect seismicity of meteor impacts on the lava tubes is studied by Modiriasari et al., )2019). However, to the best of our knowledge, no studies have been performed to parametrically evaluate the seismic response of lunar lava tubes and the seismic response. In this paper, parametric studies are presented to investigate the dynamic response of underground lava tubes subjected to vertical SV waves at different depth ratios (defined as the ratio of the radius of the lava tube to the vertical distance of the lava tube's crest from the surface). To this end, we have investigated the effect of the lava tube's depth ratio on the ground surface motion and the dynamic stress concentration factor (DSCF) of the lava tube wall. Besides, the effects of incident wave frequency on the DSCF of lunar lava tubes have been studied. The main goal is to compare the responses of lunar lava tubes to those on Earth to better understand their behavior. Furthermore, the findings can be used as a basis for future work to study influencing parameters in sustainable lunar construction. PROBLEM STATEMENT Here, we consider an idealized 2D plane-strain lava tube with a circular cross-section buried in an isotropic, dry/linear-elastic, and homogeneous half-space medium, as illustrated in Figure 1. The lava tubes are subjected to a narrowband single pulse Ricker type SV wavelet.

© ASCE

642

643

Earth and Space 2022

Figure 1. Schematic view and the material mechanical properties of an embedded lava tube in half-space impinged by a plane wave The radius of the tubes is indicated by r, whereas the burial depth of the lava tube is d. The material properties are given by shear wave velocity (𝛽), Poisson's ratio (ν), and relative density (𝜌). By separating site effects into two main categories, in particular topographic and geotechnical effects, coupled and separate effects have been studied using numerical results. A hybrid FEM/BEM numerical code, called SiteQUAKE (https://commerce.eduzone.ca/boutique/produits-et-outils-en-sciences-et-genie/sitequakea-seismic-site-effect-software/), is used to simulate and analyze the wave propagation through the domains .The SiteQUAKE numerical code was developed to perform static/dynamic analysis over 2D dry/saturated/unsaturated porous media (Maghoul & Gatmiri, 2017; Maghoul et al., 2011a, 2011b). As designed and validated by developers, FE and BE parts are compatible and work separately or in a coupled way for more advanced analyses (Gatmiri et al., 2009). As we note, the solution of wave equations in multidimensional configurations needs to satisfy the infinite radiation condition, which BEM fulfills these requirements in the far-field. In this framework, we have modeled the lava tube by FEM and substratum and far-field by BEM, which is interconnected with finite elements peripheral nodal points. As stated, the incident wave is a Ricker-type plane seismic wavelet, propagating vertically from the bottom of the model (Figure 2). 2

u(t) = A0 (a2 − 0.5)e−a a = (t − t s )⁄t p

where u(t) denotes the horizontal imposed displacement at the base of the models, while the other component of the incident motion is taken to be zero. A0, ts, and tp denote the amplitude of the incident motion, time-shift parameter, and predominant period of the incident SV wave, respectively.

© ASCE

644

Earth and Space 2022

Normalized Amplitude

0.5 0 -0.5 -1 0

0.5

1

1.5 time

2

2.5

3

Normalized Fourier Transform

1 0.5 0 0

1

2 frequency

3

4

Figure 2. Normalized incident Ricker-type wavelets properties. (top) time domain, and (bottom) frequency domain. We employed the mechanical properties of the Moon's surface layers, retrieved using a newly developed surface wave inversion algorithm (Liu & Maghoul, 2021). This method characterizes the dispersion image of the lunar subsurface using experimental seismic data from the Moon's surface. The inversion then uses the spectral element method and the trust region optimization method to determine the material's properties (Liu et al., 2020). To better understand the effect of gravity on the seismic response on the Moon, we employed the same mechanical properties for lava tubes on both Earth and the Moon. As such, the proposed Shear modulus, Poisson's ratio, and density for lunar lava tubes agree well with the properties of Earth's basaltic lava caves. Therefore, the damping ratio of the material is taken to be zero. The non-dimensional frequency (η = fr/β) is defined as the ratio of the lava tube radius to the wavelength of the incident motion. Except for Figure 7, the dimensionless frequency has been assumed to be one in all sections. RESULTS AND DISCUSSION Effect of depth ratio of lava tubes on the surface displacement response Here we discuss the effect of presence and the burial depth of a lunar lava tube in a site on the ground surface motion during a moonquake. As such, Figure 3 presents the seismic response of the Moon surface at the top of the lava tube (point A in Figure 1). Results depict that the burial depth is an important affecting factor on the ground's surface. The seismic waves hitting the lava tubes can easily be trapped between the lava tube's upper section and the ground surface. Consequently, the reflected waves can easily induce higher amplifications above the lava tube on the ground surface. It can be seen that the time-

© ASCE

645

Earth and Space 2022

domain response of point A follows a similar pattern; however, shallow lava tubes have a greater impact on the ground surface, which is important in terms of the stability of the lava tube's top soil layer. 0.8 r/d=0.2 r/d=0.4 r/d=0.6 r/d=0.8

Amplitude (m)

0.6 0.4 0.2 0 -0.2 -0.4 -0.6 -0.8 0

2

4

6 Time (sec)

8

10

12

Figure 3. Effect of lava tube burial depth on the surface seismic response for point A. Effect of depth ratio on the DSCF of lunar lava tubes Severe earthquakes may cause significant damages to underground spaces on Earth, including caves, tunnels, and lava tubes (Šebela, 2010; Wang et al., 2001; Yashiro et al., 2007). Similarly, lunar lava tubes, a potential location for future habitat on the Moon, may be prone to seismic-induced hazards. So, it is important to better understanding seismicinduced dynamic stresses in such tubes on the Moon. This is of great importance for designing lining systems of lava tubes. In order to comprehend the seismic energy distribution around embedded lava tubes, it is worthwhile to compare the dynamic stress concentration factor (DSCF) along the interior section of lava tubes. We define the DSCF, according to Pao et al. (1973), as the ratio of the hoop stress, to the maximum amplitude of the incident wave stress in the half-space. σθ σθ DSCF = | | = | | σ0 μk β in which σθ and σ0 are, respectively, hoop stress of the lava tube and stress amplitude of the incident wave. μ and k β denote the shear modulus and wavenumber of the plane wave in the half-space. To get a clearer image of the depth ratio effect on the seismic response of the lava tubes, the DSCF is plotted in Figure 4. It depicts the comparison of DSCF along the interior section of the tubes embedded in half-space. As can be seen, the DSCF amplitude is

© ASCE

Earth and Space 2022

significantly reduced by increasing the depth ratio, with the greatest DSCF occurring in the shallowest tubes. In other words, decreasing the burial depth of lava tubes from a depth ratio of 0.8 to 0.2 amplifies the induced DSCF by 60% at the upper portion of the lava tubes (30, 150 degrees), while it amplifies by 40% at the lower portion (220, 320 degrees). This unbalanced amplification shifts the maximum DSCF location from the bottom to the top of the tubes. As a result, the shallow lava tube roofs may be subjected to more violent induced stresses during in-plane shear seismic waves propagation.

Figure 4. Comparison of the DSCF along the interior ring of the lava tubes on the Moon for different depth ratios. Comparison of the results for the lunar and terrestrial lava tubes with similar mechanical and geometrical properties In Figure 5, time-domain responses for different points of the embedded lava tubes on the Moon and Earth are presented for various depth ratios. As can be seen, the terrestrial lava tubes display a more pronounced displacement at points B and C when compared to the lunar lava tubes. This clarifies the Earth's gravitational amplifying effects on the seismic response of the lava tube. When the depth ratio of the lava tubes is considered, shallow ones, particularly those with a depth ratio of 0.2 or less, exhibit similar reactions in both lunar and earth environments. This could be because these lava tubes have a lower gravitational overburden difference in both cases. Obviously, the greater the depth ratio, the greater the difference in gravitational overburden between the Moon and the Earth at the same depth. This is clearly due to the different gravitational pulls of these bodies. On the other hand, it can be seen that the maximum amplitude occurs at the top of the lava tubes with depth ratios of 0.2 and 0.4, whereas it occurs at the bottom of the lava tubes with depth ratios of 0.6 and 0.8. It is understandable that as the depth increases, the amplitude of the displacement response decreases; however, this decrease occurs at a steeper rate at the top of the lava tubes. As a result, the maximum displacement occurs at

© ASCE

646

Earth and Space 2022

the bottom of deep lava tubes and the top of shallow lava tubes. However, the highest amplitude occurs at the lowest depth ratio for all points.

Figure 5. Time-domain displacement on the roof (point B), side (point C), and the bottom of the circular lava tube (point D) for different values of depth ratio. Solid and dashed lines respectively represent the results for lava tubes on Earth and the Moon. As outlined above, the gravitational conditions play an important role in determining the seismic displacement response of lava tubes. Of course such a critical factor, should be examined more expeditiously for the DSCF of the lava tubes and its stability. Therefore, assuming a depth ratio equal to 0.2, we extracted the DSCF for the lunar and terrestrial lava tubes (Figure 6). Note that the DSCF curves for the lunar and terrestrial lava tubes have a different general pattern with a different magnitude which implies the lava tubes face various wave scattering interference patterns.

Figure 6. Comparison of the DSCF along the interior wall of lava tubes on the Moon and 1g earth conditions.

© ASCE

647

Earth and Space 2022

In terrestrial lava tubes, we observe an additional amplification of the wall hoop stresses, which is because of the higher specific gravity of the materials. The specific weight of the materials is representative of lateral soil pressure and accordingly an indicator of confining pressure on the underground spaces. In sum, given identical mechanical properties with similar excitation sources, the higher gravity on Earth would result in a higher induced stress amplitude around the embedded cavities compared to lunar lava tubes. Effect of the dimensionless frequency on the DSCF of the lunar lava tubes In this section, the effect of dimensionless frequency on the response of the lunar lava tubes is examined. The depth ratio is assumed to equal 0.2 for all the simulations. We note that the dimensionless frequency determines the sensitivity of incoming incident waves to the relative size of the lava tube. Figure 7 depicts the DSCF of the lava tubes concerning dimensionless frequency. It can be seen that a very long wavelength (η=0.25) results in a lower value of circumferential stresses on the interior wall of the lava tubes. This is because very long wavelengths cannot see the structure properly. The DSCF, on the other hand, has the highest values when the case is η=1, 0.5, which is close to the high-frequency resonant frequency band. At higher frequency (η=2), because the wavelength is smaller than the size of the lava tube, each point on the surface of the lava tube acts as a point source of wave scattering and disperses the wave energy in all directions, resulting in a lower DSCF, according to Huygens' Principle.

Figure 7. DSCF spatial variation of lunar lava tubes for different dimensionless frequencies. CONCLUSION In this paper, we have conducted a parametric study via a coupled FEM/BEM numerical technique to investigate the seismic response of an underground lava tube in terms of timedomain displacement response and the DSCF of the lava tube's wall. First, we evaluated

© ASCE

648

Earth and Space 2022

the effect of the presence of lava tubes at different burial depths on the surface seismic response using calculated displacement seismograms. The calculated seismograms show that the lava tubes at various depths can alter the surface wavefield, and shallow embedded tubes amplify the surface displacements. Furthermore, the calculated DSCF of the lunar lava tubes and those on Earth show that the higher gravity situation generally amplifies the dynamic hoop stress of interior walls. We also examined the seismic time domain displacement response of the various points on the lava tubes' interior wall, and compared the results with the corresponding lava tubes on Earth. We then examined the input seismic wave frequency content on the DSCF of the lunar lava tubes. At the heart of these parametric studies, the salient points of the dynamic response of lava tubes can be highlighted as follows: •

•

• • • •

Our findings suggest that the shallowest lava tubes strongly influence surface motion and lava tube displacements. As a result, shallow lava tubes will experience more intense responses. It was shown that as the depth of lava tubes increases, the shape of the DSCF spatial variation changes in such a way that the location of the maximum DSCF shifts toward the bottom of the inner wall of the tubes. The maximum displacement occurs at the bottom of the deep lava tubes and the top of the shallow lava tube. The calculated DSCF of the lunar lava tubes and those on Earth show that the higher gravity generally amplifies the dynamic hoop stress of interior walls. As the depth of lava tubes increases, the role of gravity becomes more perceptible in the seismic response of such tubes. Except for a very long wavelength, the DSCF of the lava tubes generally decreases as the dimensionless frequency increases and oscillates more frequently.

The findings of this study will hopefully be used as a preliminary evaluation of the seismic response of lunar lava tubes and the moonquake-resilient design of underground habitats on the Moon. REFERENCES Amini, D., Liu, H., & Maghoul, P. (2021). Seismic Site Effect Investigation for Future Moonquake-Resistant Structures by Considering Geometrical and Geotechnical Characteristics of Lunar Bases. In Earth and Space 2021 (pp. 724-731). Blair, D. M., Chappaz, L., Sood, R., Milbury, C., Bobet, A., Melosh, H. J., Howell, K. C., & Freed, A. M. (2017). The structural stability of lunar lava tubes. Icarus, 282, 4755. Cooper, M., & Kovach, R. (1975). Energy, frequency, and distance of moonquakes at the Apollo 17 site. Lunar and Planetary Science Conference Proceedings,

© ASCE

649

Earth and Space 2022

Daga, A. W., Allen, C., Battler, M., Burke, J., Crawford, I., Léveillé, R., Simon, S., & Tan, L. (2009). Lunar and Martian lava tube exploration as part of an overall scientific survey. Annual Meeting of the Lunar Exploration Analysis Group, Dainty, A. M., & Toksöz, M. N. (1981). Seismic codas on the Earth and the Moon: A comparison. Physics of the Earth and Planetary Interiors, 26(4), 250-260. Dainty, A. M., Toksöz, M. N., Anderson, K. R., Pines, P. J., Nakamura, Y., & Latham, G. (1974). Seismic scattering and shallow structure of the Moon in Oceanus Procellarum. The Moon, 9(1), 11-29. Duennebier, F., & Sutton, G. H. (1974). Meteoroid impacts were recorded by the short‐ period component of Apollo 14 Lunar Passive Seismic Station. Journal of Geophysical Research, 79(29), 4365-4374. Garcia, R. F., Khan, A., Drilleau, M., Margerin, L., Kawamura, T., Sun, D., Wieczorek, M. A., Rivoldini, A., Nunn, C., & Weber, R. C. (2019). Lunar seismology: An update on interior structure models. Space science reviews, 215(8), 1-47. Gatmiri, B., Maghoul, P., & Arson, C. (2009). Site-specific spectral response of seismic movement due to geometrical and geotechnical characteristics of sites. Soil Dynamics and Earthquake Engineering, 29(1), 51-70. Haruyama, J., Hara, S., Hioki, K., Morota, T., Yokota, Y., Shirao, M., Hiesinger, H., van der Bogert, C., Miyamoto, H., & Iwasaki, A. (2010). New discoveries of lunar holes in Mare Tranquillitatis and Mare Ingenii. 41st Annual Lunar and Planetary Science Conference, Haruyama, J., Hioki, K., Shirao, M., Morota, T., Hiesinger, H., van der Bogert, C. H., Miyamoto, H., Iwasaki, A., Yokota, Y., & Ohtake, M. (2009). Possible lunar lava tube skylight observed by SELENE cameras. Geophysical Research Letters, 36(21). Horvath, P., Latham, G. V., Nakamura, Y., & Dorman, H. J. (1980). Lunar near‐surface shear wave velocities at the Apollo landing sites as inferred from spectral amplitude ratios. Journal of Geophysical Research: Solid Earth, 85(B11), 6572-6578. Liu, H., & Maghoul, P. (2021). Apollo Seismic Data Interpretation Using an Elastodynamic Space-Time Spectral Element Technique and Dispersion Image Inversion Method. In Earth and Space 2021 (pp. 99-107). Liu, H., Maghoul, P., Shalaby, A., Bahari, A., & Moradi, F. (2020). Integrated approach for the MASW dispersion analysis using the spectral element technique and trustregion reflective method. Computers and Geotechnics, 125, 103689. Luco, J., & De Barros, F. (1994). Dynamic displacements and stresses in the vicinity of a cylindrical cavity embedded in a half‐space. Earthquake engineering & structural dynamics, 23(3), 321-340. Maghoul, P., & Gatmiri, B. (2017). Theory of a time-domain boundary element development for the dynamic analysis of coupled multiphase porous media. Journal of Multiscale Modelling, 8(03n04), 1750007.

© ASCE

650

Earth and Space 2022

Maghoul, P., Gatmiri, B., & Duhamel, D. (2011a). Boundary integral formulation and twodimensional fundamental solutions for dynamic behavior analysis of unsaturated soils. Soil Dynamics and Earthquake Engineering, 31(11), 1480-1495. Maghoul, P., Gatmiri, B., & Duhamel, D. (2011b). Wave propagation in unsaturated poroelastic media: Boundary integral formulation and three-dimensional fundamental solution. Computer Modeling in Engineering and Sciences, 78(1), 5176. Modiriasari, A., Boener, A., Theinat, A., Bobet, A., Melosh, H., Dyke, S., Ramirez, J., Maghareh, A., & Gomez, D. (2019). Effect of Induced Seismicity of Indirect Meteorite Impacts on the Stability of Lunar Lava Tubes. Lunar and Planetary Science Conference, Modiriasari, A., Theinat, A., Melosh, H., & Bobet, A. (2019). Stability Analysis of Lunar Lava Tubes for Permanent Extraterrestrial Habitation. 53rd US Rock Mechanics/Geomechanics Symposium, Oberbeck, V. R. (1969). On the origin of sinuous lunar rilles. Modern Geology, 1, 75. Pao, Y.-H., Mow, C.-C., & Achenbach, J. (1973). Diffraction of elastic waves and dynamic stress concentrations. Robinson, M., Ashley, J., Boyd, A., Wagner, R., Speyerer, E., Hawke, B. R., Hiesinger, H., & Van Der Bogert, C. (2012). Confirmation of sublunarean voids and thin layering in mare deposits. Planetary and Space Science, 69(1), 18-27. Robinson, M., Brylow, S., Tschimmel, M., Humm, D., Lawrence, S., Thomas, P., Denevi, B., Bowman-Cisneros, E., Zerr, J., & Ravine, M. (2010). Lunar reconnaissance orbiter camera (LROC) instrument overview. Space science reviews, 150(1-4), 81124. Sauro, F., Pozzobon, R., Massironi, M., De Berardinis, P., Santagata, T., & De Waele, J. (2020). Lava tubes on Earth, Moon and Mars: A review on their size and morphology revealed by comparative planetology. Earth-Science Reviews, 103288. Šebela, S. (2010). Effects of earthquakes in Postojna cave system. Acta carsologica, 39(3). Smerzini, C., Aviles, J., Paolucci, R., & Sánchez‐Sesma, F. (2009). Effect of underground cavities on surface earthquake ground motion under SH wave propagation. Earthquake engineering & structural dynamics, 38(12), 1441-1460. Theinat, A., Modiriasari, A., Bobet, A., Melosh, J., Dyke, S., Ramirez, J., Maghareh, A., & Gomez, D. (2018). Geometry and structural stability of lunar lava tubes. 2018 AIAA SPACE and Astronautics Forum and Exposition, Wang, W., Wang, T., Su, J., Lin, C., Seng, C., & Huang, T. (2001). Assessment of damage in mountain tunnels due to the Taiwan Chi-Chi earthquake. Tunnelling and underground space technology, 16(3), 133-150. Yashiro, K., Kojima, Y., & Shimizu, M. (2007). Historical earthquake damage to tunnels in Japan and case studies of railway tunnels in the 2004 Niigataken-Chuetsu earthquake. Quarterly Report of RTRI, 48(3), 136-141.

© ASCE

651

Earth and Space 2022

Examination of Smart Sandbags for Semi-Permanent Structures on the Lunar Surface Yinan Xu1; Jiawei Qiu2; Virupakshan Vilvanathan3; Athip Thirupathi Raj4; and Jekan Thangavelautham5 1

Space and Terrestrial Robotic Exploration (SpaceTREx) Laboratory, Dept. of Aerospace and Mechanical Engineering, Univ. of Arizona, Tucson, AZ 2 Space and Terrestrial Robotic Exploration (SpaceTREx) Laboratory, Dept. of Aerospace and Mechanical Engineering, Univ. of Arizona, Tucson, AZ 3 Space and Terrestrial Robotic Exploration (SpaceTREx) Laboratory, Dept. of Aerospace and Mechanical Engineering, Univ. of Arizona, Tucson, AZ 4 Space and Terrestrial Robotic Exploration (SpaceTREx) Laboratory, Dept. of Aerospace and Mechanical Engineering, Univ. of Arizona, Tucson, AZ 5 Associate Professor, Dept. of Aerospace and Mechanical Engineering, Univ. of Arizona, Tucson, AZ. Email: [email protected] ABSTRACT Development of the lunar surface will likely be a pivotal step in the emerging space economy. In recent years it has been confirmed that water is present in the polar craters of the Moon. The Moon is also rich in iron, titanium, and silicon; which may be mined with the appropriate lunar facilities. Motivation to erect a lunar base aligns with the NASA Artemis objectives for a human return to the Moon. In order to achieve Artemis objectives, bases must be erected in a manner that allows for flexibility and mobility. In this way, semi-permanent structures are an ideal mode for facilities to exist on the lunar surface. The optimal method to achieve these structures would be to utilize existing sandbag technologies to innovate lunar-appropriate bases. In the early stages of prospecting and open pit mining, there will likely need to be mobile pilot bases setup that need to perform in-depth evaluation and attempt pilot scale mining at different sites before permanent structures can be installed. We propose simple, multifunctional building blocks such as “smart sandbags” for constructing rapid, low-cost semi-permanent structures. Smart sandbags are filled with lunar-regolith and embedded with structure and impact sensors, and adopt a designed 3D customizable shape. The sandbags will be made of carbon fiber fabrics and integrated with silicone to combat the abrasiveness of lunar sand. Options for rigidization of the sandbags will also be explored with methods such as UV-cured resin or hydrogel. Innovatively designed features of our sandbag structures include the ability to be quickly assembled and disassembled, utilization of in situ resources, and effectiveness under a variable number of ground conditions. The semi-permanent structures are expected to provide shielding from collisions, radiation, maximize surface traction, and provide human habitat. Our studies show that the feasibility of the sandbag structure for use under different lunar surface conditions. Further in-depth investigations will need to be performed to quantify the potential improvement offered by sandbag structures over conventional brick laying and additive manufacturing. 1. INTRODUCTION NASA has set forth an ambitious program to return humans to the Moon through the Artemis program. It is expected the return of humans will be a step towards development of semi-

© ASCE

652

Earth and Space 2022

permanent or permanent base development on the Moon. The surface of the Moon is a harsh, inhospitable environment that undergoes extreme temperature swings between lunar day and night (ranging from -173 oC to 130 oC), faces solar and cosmic radiation that is hazardous to human health, in addition, the surface is often bombarded with micro-meteorites that can pierce through thin fabric and structures (Heiken et al., 1991). Furthermore, much like Earth, the terrain, local rock composition and natural weathering conditions all have an impact on the overall surface conditions. Site-preparation and construction methods that can be adapted to varying site conditions and enable base mobility would be ideal as it reduces constraint on overall mission planning and minimizes variability in end-structure. Our focus will be to produce robust solutions that minimize energy use, time and cost of excavation, site preparation activity and construction in support of base operations and pilotscale surface resource mining activity (Thangavelautham et al., 2008) (Figure 1). In the early stages of prospecting and open pit mining, it is likely there will need to be mobile pilot bases setup that need to perform in-depth evaluation and attempt pilot scale mining at different sites before truly permanent structures can be installed.

Fig. 1. (Top) Artist rendering of a lunar base with shielding. (bottom) Sandbags assembled into Super Adobe structures and homes invented by N. Khalili (bottom right), Katauskas (1998) and Husain (2000). The team will be exploring the use of simple, multifunctional building blocks such as “smart sandbags.” The demonstration and success of this technology can also have major applications in terrestrial environments, where there is a need for rapid, low-cost construction of semipermanent structures for use in post-disaster/emergency situations (Figure 1, Bottom) (Katauskas, 1998, Husain, 2000 and Khalili, 2008). Our work on “smart sandbags” is derived from our 8-year work on inflatables for spacecraft in collaboration with NASA JPL (Babuscia et al., 2020 and Chandra et al., 2018). The bladder fabric for the “smart sandbag” needs to be thoroughly researched to handle the crushed glass properties of lunar sand and candidates include Vectran, Kevlar, Technora, Twaron and Zylon among others. The “smart sandbag” instead of being filled with a gas, would be filled with lunar-regolith and would be embedded with structural and impact sensors, and adopt a designated 3D customizable shape (borrowing from advancements in soft robotics). The

© ASCE

653

Earth and Space 2022

embedded electronic modules make these sandbags smart. Depending on the scheme of electronics, they may act as a distributive processing network for the lunar base; they may communicate among each other and act as the lunar base information infrastructure; they may monitor different parameters of the structure such as temperature and radiation levels, and even the health of the sandbags; they may detect the proximity of another sandbag to facilitate the construction process; they may be embedded with LED matrices as signages for human and robotic vision; they may act as beacons for rovers to travel along a roadway. The specific electronic schemes for different used mentioned are to be explored and are beyond the scope of this paper. The sandbags could be powered in one of several ways including use of primary batteries, RF power beaming, micro-fuel cells (Thangavelautham et al., 2017). Lunar regolith much like Earth sand can be used to build reinforcement structures such as berms, sufficient 0.5 m or greater thickness offers shielding from solar and cosmic radiation. In addition, the sand forms a shock absorption system from micro-meteor impacts, collisions and on-site accidents. The proposed sandbag can be made hardened like a brick by having gravel and sand, vacuum sealed or having soft properties much like a ‘bumper shield’ with loose sand. The sandbag would be sealed shut using a solar-sintering process (Anderson and Thangavelautham 2021 and Kayser, 2011). These smart sandbags could then be fused together through solar-sintering to form pillars, walls, light-posts, radiators, parabolic communication antennas, ceilings and unibody hemispheres such as the Super Adobe structure (Khalili, 2008). Our hypothesis is that sandbagging structures can be quickly assembled and disassembled, utilize in-situ resources such as solar-sintering efficiently and are effective even under a variable number of ground conditions. Sandbag structures can also be easily patched up or repaired. The sandbags could be used to form a blanket like shield to be placed over rovers to provide shielding from collisions, radiation shielding and ballast to maximize surface traction. Our overall goals are to enable scalable teams of robots to autonomously perform end to end site-preparation and construction in support of base operations. This is after all a dull, dirty, dangerous, and dexterous task for human astronauts. Our earlier work has shown the promise of multirobot teams performing excavation and site preparation activity. These multirobot teams can autonomously interpret 3D excavation blueprints and learn to collect and dump regolith in designated ‘dumping areas.’ In this paper, we envision the collected regolith would be put to good use by filling of smart sandbags. Using these smart sandbags we intend to build base shelters. The highest priority lunar base facilities include storage and repair shelters for a team of lunar infrastructure robots. These teams of robots will be active during the lunar day performing all kinds of site preparation and base construction activity. However, during the night the temperatures will plummet to below -150 oC and the rovers will need to be sheltered within envision Super Adobe structures that would be semi buried to 1-2 m below the surface. At depth below 0.5 m, the lunar surface temperatures reach a steady -25 oC and hence this is satisfactory storage temperature for the rovers over the lunar night. The sandbag structures also have other uses on base include reinforcing berms around landing pads, reinforcing base human facilities and to reinforce the base/anchor large steerable communication antennas, solar panels and radiators. In the following section, we present related work followed by a description of the modular smart sandbag architecture, structure, thermal and radiation analysis, simulation experiments in assembly followed by results and discussion. Finally, we offer some preliminary conclusions in Section 6.

© ASCE

654

Earth and Space 2022

2. RELATED WORK A promising technique for autonomous robotic construction mimics eusocial insects such as ants, termites and bees. Termites have evolved exceptional engineering talent building skyscraper-like nests called cathedral mounds equipped with temperature-regulating internal heating and cooling shafts, in addition to facilities to farm food. Insects are particularly noted for their ability to build pathways and ramps utilizing amorphous construction methods. Borrowing from these principles, a laboratory robot is used to process foam by heating and melting into shape to produce bridges, ramps and other complex structures (Napp and Nagpal, 2012). Halbach et al. (2013) have simulated robot teams performing excavations on Mars. These robots are used to construct a human base and use in-situ materials and processes for construction. Humans are involved in the high level planning, designation of key facilities, making strategic decisions and for troubleshooting. The planning and construction of a lunar human habitat will need teams of infrastructure robots cooperating to achieve a common goal. Swarm robotics is well placed to accomplish such tasks as it involves multiple autonomous robots that work in one or more redundant teams towards achieving a common goal. Some of these swarm robots have been designed to mimic identified key strategies used by eusocial insects to accomplish collective organization. These strategies include use of self-organization, templates and stigmergy. Bonabeau et al. (1997) showed self-organization is how macroscopic order is generated from interactions among distinct lower-level units within the group using only local information. Templates are environmental features such as spatiotemporal gradients perceptible to individual units within the swarm (Bonabeau et al., 1999). In robotics, templates have implemented using static light fields to facilitate construction of linear (Stewart and Russell, 2003), circular walls (Wawerla et al., 2002) and annulus structures (Wilson et al., 2004). Stewart and Russell (2004) study showed time varying templates consisting of moving a light field gradient have been used to produce even more intricate structures. Grasse (1959) study showed stigmergy is a low-bandwidth, indirect communication scheme channeled via the surrounding environment. Stigmergy has been the mainstay of swarm construction, including box pushing (Mataric et al., 1995), heap formation by Beckers et al. (1994), tiling pattern formation (Thangavelautham et al., 2003) and ‘blind’ bulldozing (Parker et al., 2003). However, often, conventional swarm control methods face two roadblocks. Firstly, many rely on deterministic “if-then” rules developed by humans or on stochastic behaviors. Without sufficient domain knowledge, it is challenging to manually design controllers that facilitates self-organization, as there exists few if any theory to model and predict/control the macroscopic system behaviors that emerge from local actions. Designing controllers that “just works” can descend into an arduous task of trial-and-error learning. Chantemargue et al. (1996) showed swarm systems also face another challenge called antagonism. This occurs when multiple individuals performing the same task unexpectedly obstruct each other, degrading the overall workings of the system or worse, resulting in gridlock. Such an outcome becomes an inevitable challenge with multiple decentralized robots operating in close quarters. Overall, such an outcome limits transferability of the solution to size of the task area and number of robots. Site preparation tasks had initially been envisioned to be done using astronauts. However, the danger of the lunar surface makes this a high-risk effort. Instead, thanks to the promise of collective robotics as discussed above, what is more compelling is the use of swarm robotics to perform excavation, site-preparation and base construction ready for human astronauts to live

© ASCE

655

Earth and Space 2022

in. There are two compelling pathways, using tele-robotics and autonomous robotics. We envision the use of both pathways, with autonomous robotics being the default mode and telerobotics stepping in to handle unexpected situations (Thangavelautham et al., 2009). Our ongoing work in autonomous robot team show that we can provide robots a 3D-blue print and they can autonomously excavate that structure. The robot controllers are developed using Artificial Neural Tissue and can be applied on a variety of excavation devices and number of vehicles (Thangavelautham et al., 2008 - 2020). Our research and other identified bucket wheels as the most effective excavation device in low gravity (Skonieczny and Wettergreen, 2016 and Thangavelautham et al., 2009). We find ANT exceeds the performance of other neural network methods including Deep Learning and finds optimal number of robots to perform excavation (Thangavelautham et al., 2017) (Figure 2, 3).

Fig. 2. (Left) Using ANT, we can determine the optimal number of robots needed for an excavation task. In comparison human devised controller drop in performance with increased number of robots. (Right) (a)-(d) We have successfully tested ANT evolved multirobot controllers performing excavation in the laboratory, controlled field conditions, in high-fidelity Digital Space Simulation and Mauna Kea lunar analogue site (d) Our work in this area began with laboratory demonstration and culminated with demos at a lunar analog site (Figure 2). That work can be further advanced with the current proposal. The excavated material would be filled into “smart sandbags” and form the building blocks for preparatory structures and for base construction. Other alternatives include additive manufacturing of the lunar base structure. Additive manufacturing process is energy and resource intensive (Tang et al., 2003), but still more efficient than conventional brick and mortar or steel, and glass (Thangavelautham et al., 2019). Our proposed approach of solar sintering would be to seal and mate smart sandbags (Anderson and Thangavelautham, 2021), (Kayser, 2011). This would take considerably less resources, although the sandbags will need to be brought in at first from Earth. Other methods require special forms of concrete (Khsohnevis and Bekey, 2019) or composite paste, by AI SpaceFactory, winner of the NASA habitat challenge (Reid, 2020) and needs water

© ASCE

656

Earth and Space 2022

which will be hard to find and extract on the Moon (at least at initial stages) and complex material processing (Anderson and Thangavelautham, 2021), (Kayser, 2011).

Fig. 3. Three front-loaders excavated out a 9 m × 10.5 m × 2 m landing pad and a 5 m × 4 m ramp reaching a depth of 2 m over a 4-hour period. 3. STRUCTURAL ANALYSIS There is a need for multi-functional construction material to serve as the building blocks of an emerging lunar base. As discussed earlier, the first lunar bases will likely need to have qualities of a hardened permanent base but will also need to be mobile to move from one location to another in search of viable local resources. Therefore, there is a need for semi-permanent shelters and support for critical structures, facilities, and equipment on the base. These structures and facilities need to be well protected from the lunar surface environment between day and night. Our efforts to date have shown emerging success with using teams of robots to excavate according to a 3D blueprint. The regolith collected from excavation could be put to good use by filling up sandbag modules. Use of sandbags stands as an alternative to current methods of 3D printing and construction of more traditional brick like building components. According to our research, large scale 3D printing of structures on Earth requires forms of cement/concrete paste that requires water. Brick also starts off as wet paste that is put through a mold and dried. In both cases water in large supply is essential. On the Moon, surface missions will be faced with water scarcity. Contemplating tapping into the large water supply in the PSR regions for early missions is unrealistic, thus early building structures will need to be dry and not be reliant on water.

Fig. 4. Super Adobe structure concept using sack shaped sandbags (left) and dimension parameters (right).

© ASCE

657

658

Earth and Space 2022

The advantage of sandbags is that it can take on loose, dry material that could vary from site to site. It is the bag fabric in combination with some mounting or gripping device that operates as the interface between sandbag modules. The sandbags can be used to build impressive structures including Super Adobes (see Figure 4). The pragmatic steps in building a conventional Super Adobe dome will illustrate these parameters, which can be later generalized shown below (see Figure 5).

Fig. 5. Sandbag cross-section dimension (top left) model of the geometry of sandbag (right) example of the result of Equivalent Stress (bottom). Next, the relationship between sack shaped sandbags dimensions (see Figure 5) and SuperAdobe structure dimensions (see Figure 4) is estimated by the below equations, for storage 2 rovers: 𝐻2 = 6𝑊 (1) 𝐻1 = 16𝑊 (2) 𝑅1 = 9(𝐿1 + 2𝐿2) (3) For storage 5 rovers: 𝐻2 = 8𝑊 (4) 𝐻1 = 21𝑊 (5) 𝑅1 = 18(𝐿1 + 2𝐿2) (6) From the above equations and Super Adobe structure dimensions, the total volume can be estimated: 1 2 𝑅1 (3 𝐻1 + 3 𝐻2) ∗ 𝜋 ∗ ( 2 )2 (7) Table 1. Parameters of the sandbag structure Notation 𝑊 𝐿1 𝐿2 𝐻1 𝐻2 𝑅1

ᶱ

© ASCE

Description Width of the sandbag Length of the sandbag External length of sandbag (Independent from 𝐿1 and 𝑊) Total height of SuperAdobe Height of side wall Diameter of the SuperAdobe base Incline angle of the roof

659

Earth and Space 2022

Numerical Results and Discussions A Super Adobe structure’s geometry consists of using sack-shaped sandbags, side by side. These sandbags form a ring shape. These rings have a decreasing radius as they ascend, and the decrease of each ring’s radius with respect to the previous. A very critical fact to consider in a Supe Adobe’s geometry, is the amount of support (contact area) between two consecutive rings, being the minimum located between the two top rings. To avoid the situation illustrated above, a coherent set of values for independent parameters should be defined before construction. ANSYS 19.2, the software has been used for numerical simulations. A total number of 53007 nodes and 28772 elements are generated. The physical properties of the sandbag structure are the same as for sand cast magnesium, that is, young’s modulus = 40 GPa, tensile (compressive) yield strength = 21 MPa, Poisson’s ratio = 0.35, and density = 1738𝑘𝑔/𝑚3 .

Fig. 6. Model of a Super Adobe structure (left) example of the result of Equivalent Stress. Parametric study in ANSYS and MATLAB

Fig. 7. ANSYS and MATLAB integration API

© ASCE

660

Earth and Space 2022

The effort to design efficient and safe structures has led to the growing use of material simulation which aims to provide the most exact description possible of the real behavior of structures. With these issues, special attention is paid to Super Adobe structures made from lunar regolith. The process of the identification of the parameters of the material model (see Table 2) from the ANSYS workbench environment via the repeated numerical calculation (see Figure 7) of the exciting models. An advantage of this procedure was the simpli city of calling an external script for the evaluation of different parameters. The programmed batch carried out the assembly of the geometry of the sandbag (length, width, and external radius), generated mash as geometry changed, and called the script which enabled the calculation of the objective function value. Then sandbag geometry study could be performed. Next, the material properties study was conducted by 5 sets of advanced materials to generate a set of data to estimate the result. Table 2. Material Properties Bag Material

Kevlar Mylar Teflon Carbon Fiber Sand Cast Magnesium

Young’s Modulus (Gpa)

Thermal Conductivity (W/mK)

Density 𝑘𝑔 ( 3)

76 5.24 6.89 183 40

5.00 0.189 0.250 21-180 159

1380 1380 2200 1750 1738

𝑚

The aim of the parametric study was to carry out the identification of the parameters of the Super Adobe model in the ANSYS Workbench. As the task was conducted, only the young’s modulus, tensile strength, and materials density can be considered as realistically identified parameter values. As you can see (see Figure 8), the density has a large impact on the maximum equivalent stress after greater than 4 × 103 𝑘𝑔/𝑚3. Tensile yield strength should be considered below 4 × 108 𝑃𝑎. The above study shows that even though ANSYS does not offer its own calculator enabling the evaluation of the objective function, the appropriate results can be provided for consideration.

Fig. 8. Parameter Chart material density vs Maximum equivalent stress (top) Tensile yield strength vs Maximum equivalent stress (bottom)

© ASCE

661

Earth and Space 2022

4. THERMAL ANALYSIS Without an atmosphere, the temperature change within a lunar day can be very dramatic. The temperature rises to 374 K during the lunar day and drops to 120 K at lunar night at the equatorial latitude. The wide temperature variation may pose a significant challenge to the rovers operating longterm on the lunar surface. Beyond being a physical shelter, the Super Adobe can also provide a space with a smaller temperature range compared to complete exposure. In lieu of employing finite element thermal analysis in commercial software, we elected to establish a thermal circuit model of a Super Adobe system that approximates the temperature range of apparatus stored within. This method is more readily applicable to the iterative algorithm and faster compared to a complete finite element analysis, which can be used for verification rather than being used in the algorithm loop. Thermal Circuit Model

Fig. 9. Thermal model of the Super Adobe: the physical representation (left) and the thermal circuit representation (right). 𝑻𝟏 is the temperature of the external adobe surface; 𝑻𝟐 is the temperature of the internal adobe surface; 𝑻𝟑 is the temperature of the regolith enclosed within the adobe structure; 𝑻𝟒 is the temperature of the sample rover stored within the adobe; 𝑻𝟓 is the temperature of the regolith in the vicinity. 𝑻𝟏 and 𝑻𝟓 are known, 𝑻𝟏 and 𝑻𝟐 are intermediary target variables and 𝑻𝟒 is the target variable. There are different bodies and temperature nodes and heat exchanges that we are interested in. We have chosen those that are most important and simplified the model to the thermal circuit in Figure 9. They are a combination of conduction and radiation, as there is negligible atmosphere at the lunar surface. The heat fluxes are listed in Table 3. Table 3. Physical meanings of the important heat fluxes in the thermal circuit. Heat Flux [W] Q1 Q2 Q3 Q4

© ASCE

Physical Meaning Conduction external to internal wall Conduction between external ground to internal wall Conduction between external ground to internal ground Radiation between internal ground to internal wall

Heat Flux [W] Q5 Q6 Q7

Physical Meaning Radiation between internal wall to rover Conduction between internal ground to rover Radiation between internal ground to rover

Earth and Space 2022

Using the Kirchoff’s law of junctions, we get the heat flux relationships as well as the expression of individual heat fluxes. We get three equations for the three unknowns 𝑇2 , 𝑇3 , and 𝑇4 . 𝑄1 + 𝑄2 − 𝑄4 − 𝑄5 = 0 𝑄3 − 𝑄4 − 𝑄6 = 0 { (8) 𝑄5 − 𝑄7 + 𝑄6 = 0 𝑄2 − 𝑄3 = 0 This allows us to analyze each heat flux separately and obtain a system of equations between the temperatures. We use 𝑄𝑖 = 𝑘𝑖 𝐴𝑖 Δ𝑇/𝐿 for conduction, where 𝑘 is the thermal conductivity, 𝐴 is the contact area, Δ𝑇 is the temperature differential, and 𝐿 is the length of the object; and 𝑄𝑖 = ∑𝑗 ϵσ𝐹𝑖𝑗 𝐴𝑖 (𝑇𝑖4 − 𝑇𝑗4 ) for radiation, where 𝜖 is the emissivity, 𝐹𝑖𝑗 is the view factor, and 𝜎 is the Stefan-Boltzmann constant. 𝑄1 = 𝑘12 𝐴dome (𝑇1 − 𝑇2 )/𝑡sand 𝑄2 = 𝑘25 𝐴ring (𝑇5 − 𝑇2 )/ℎdome 𝑄3 = 𝑆𝑘35 (𝑇3 − 𝑇5 ) 𝑄4 = 𝜎[𝜖bag 𝐹23 𝐴dome 𝑇24 − 𝜖ground 𝐹32 (𝐴ground − 𝐴rover )𝑇34 ] (9) 𝑄5 = 𝜎[𝜖bag 𝐹24 𝐴dome 𝑇24 − 𝜖rover 𝐹42 𝐴rover 𝑇44 ] 𝑄6 = 𝑘34 𝐴rover (𝑇4 − 𝑇3 )/ℎrover 𝑄7 = 𝜎[𝜖ground 𝐹34 𝐴ground 𝑇34 − 𝜖rover 𝐹43 𝐴rover 𝑇44 ] { We may simply substitute these equations into the equations obtained from the Kirchoff’s law of junction to get the explicit relationships of the unknown temperatures.

Solving the Equations We see that it contains different physical parameters along with known and unknown temperatures. And now we will try to obtain all the necessary parameters to find the unknown temperatures 𝑇2 , 𝑇3 , and 𝑇4 . We find 𝑇1 and 𝑇5 from solar power ϕsun and surface absorptivity αbag and αsand to temperature of the external wall $𝑇1 $ and the external ground 𝑇5 with surface emissivity ϵbag and ϵsand . This means 𝑇1 and 𝑇5 is known. We will use the maximum ϕsun to find the temperature extremes, where the moon is closest to the Sun.

© ASCE

662

663

Earth and Space 2022

{

ϕsun αbag = ϵbag σ𝑇14

(10) ϕsun αsand = ϵsand σ𝑇54 For conduction between the external wall to the inner wall, we first treat the sandbag as a 3part system illustrated in Figure 9, two parts of sandbag material in contact with each other as well as the one part of lunar regolith. We find the equivalent thermal conductivity via thei r thermal resistance. 𝑡 𝑡 𝑅bag = 𝑘 bag𝐴 , 𝑅sand = 𝑘 sand𝐴 (11) bag

𝑅bag 𝑅sand

𝑅eq = 𝑅bag + 𝑅 𝑡

bag +𝑅sand

sand 𝑘eq = 𝐴𝑅 = 𝑡sand (𝑡 eq

sand

1

= 𝐴 (𝑡

𝑡bag 𝑡sand

bag 𝑘sand +𝑡sand 𝑘bag

+𝑘 ) bag

𝑡

+ 𝑘bag )

(12)

= 𝑘12

(13)

bag 𝑘sand +𝑡sand 𝑘bag −1 𝑡bag 𝑡sand 𝑡bag

bag

For the conduction between the external ground to the inner wall, the conduction contact area is a ring between radii of inner and outer wall of the adobe at ground level. We approximate the length that heat travels is the height of the adobe ℎdome . Thus 𝑘12 = 𝑘25 = 𝑘eq , 𝑘35 = 𝑘sand , 𝑘34 = 𝑘rover . We assume the Super Adobe to be a hemisphere structures for simplicity. We recognize that we must separate the lunar regolith directly beneath the area enclosed by the adobe wall with the regolith in the immediate surrounding. The reason being that the regolith not shaded by the adobe will receive heating from the Sun, and this is not the case for those shaded by the adobe. This is observed in the simulation reported by the Mottaghi (2015) that the temperature profile of the regolith within the shaded area is different from that of that in the vicinity. In our case, we treat the regolith underneath the adobe to be a separate object modeled by a cylinder buried in the lunar regolith. It shares its thermal properties with the lunar regolith, but with a depth significantly larger than its cross-section diameter. This makes 𝑆, the shape factor 2π𝐿/ ln(4𝐿/𝐷), 𝐿 being the depth and 𝐷 being the diameter. For the view factors in the radiation models, we approximate that the system to be a domed structure with a flat ground and a disk resting at the center on the ground. To improve the accuracy of our model, we included both the radiation path and the conductive path for heat transfers between the ground and the rover. The internal wall surface would exchange radiation with both the ground and the top of the rover, and the ground would exchange radiation with both the wall surface and the bottom of the rover. We approximate the rover being a flat plate in this case. Thus, we find the view factors using identities: 𝐹22 + 𝐹23 + 𝐹24 = 1 𝐴4 𝐹43 = 𝐴3 𝐹34 { 𝐹32 + 𝐹34 = 1 𝑎𝑛𝑑 {𝐴2 𝐹23 = 𝐴3 𝐹32 (14) 𝐹42 + 𝐹43 = 1 𝐴2 𝐹24 = 𝐴4 𝐹42 𝐹43 =

√(𝑑4 +𝑑3 )2 /𝑙2 +4−√(𝑑3 −𝑑4 )2 /𝑙2 +4 2𝑑3 /𝑙

(15)

with 𝑙 being the distance between 3 (ground) and 4 (rover). With 𝐹43 calculated from distances, we can find all the view factors 𝐹𝑖𝑗 in the temperature equations. Now all the necessary parameters are determined. For the thermal analysis, we noticed that there are some parameters assumed that can also be modified as variables to be investigated. For example, we may investigate different Super Adobe shapes: cylinders standing upright, lying half-cylinders, cones, cuboids, and tetrahedrons. This will change the adobe dimensions and areas of thermal radiation. The algorithm may reveal which shape is the best in thermal isolation and which is the worst and compare it to their structural properties. This report only considered maximum solar heating power. In future

© ASCE

Earth and Space 2022

iterations, the latitude of the adobe may also be considered. This allows tailoring the sandbag design and dimensions to the location of the structure, as different latitudes will receive different solar heating power. Adobes being heated by only half of its external surfaces may not require as much thermal isolation ability compared with those that receive solar heating on its entire external surface. Additionally, we only have one rover housed in the Super Adobe. Perhaps the data would change if we included several rovers parked in different patterns. The rovers can also be represented more accurately. The rovers contain different components with different thermal requirements and properties. Having them modelled more accurately may be more informative to our algorithm. As an additional path, if a finite-element thermal analysis software that can be readily incorporated into the iterative algorithm, it can be employed to run in parallel with our thermal model analysis as verification. 5. RADIATION ANALYSIS One of the biggest concerns of developing a habitable human base on the lunar surface is the danger of space radiation. The space around the moon contains many different types of ionizing radiation. There are large fluxes of low-energy solar particles, smaller fluxes of high-energy galactic cosmic rays and the occasional intense particle fluxes emitted by solar flares. The lunar surface receives around 30 rem/yr of radiation compared to 0.36 rem/yr on the earth’s surface (Churchill 1997). This radiation can cause long-term damage in unprotected crew members as well in case of short-term events like solar flares, the radiation dosage can even lead to fatality. The structure built out of “smart sandbags” should be capable of reducing radiation levels to acceptable limits and the sandbags are designed to be functional under the radiation levels on the lunar surface. Summary of the three major types of radiation in the lunar environment by Heiken (1991) showed the three major categories of radiation in the lunar environment: the solar wind, solar flare associated particles also known as Solar cosmic rays (SCR) and Galactic cosmic rays (GCR). These particles interact with the lunar surface differently based on their size and energy and this results in varying penetration depths from micrometers to meters. Solar wind particles have low energy (keV) and are shielded by less than a micrometer of regolith. Solar event particles aren’t significantly mitigated until they traverse through at least ~50-100 cm of regolith. (Nealy, 1988). Heavy nuclei GCR particles are stopped by ~10 cm of lunar regolith while all other particles are stopped by 1000g/cm 3 of regolith (Heiken, 1991). The average density of lunar regolith is known to be around 2 g/cm 3 which implies at least 5 meters of lunar regolith. However, since the permissible radiation levels is higher than zero, less than 5 meters of lunar regolith shielding should be sufficient. Therefore, based on the radiation penetration depths of the different types of ionizing radiation on the lunar surface, at least 1 meter of lunar regolith should provide the bare minimum radiation protection for the crew inside (Silberberg, 1988). 6. CONCLUSION In the paper, we have identified the need for multi-functional building blocks in base preparation and construction on the lunar surface. Thanks to advances in robotic excavation, we propose to use the collected regolith to fill sandbags that would be assembled into base structures. Here we are evaluating the feasibility of the sandbag concept for use on the lunar

© ASCE

664

Earth and Space 2022

surface. High-priority facilities include shelters for a team of autonomous infrastructure rovers that will be involved in cleaning, site preparation, and excavation tasks. The structure would need to have at least 1 meter of regolith shielding to effectively shield the base from harmful space radiation and protect the crew from radiation sickness. We have also established a thermal circuit model of the Super Adobe to evaluate its thermal isolation ability. The iterative algorithm is able to call the system of equation derived from our thermal circuit model and determine the temperature of a simplified rover when then sun is shining directly on the Super Adobe. 7. REFERENCE Abu El Samid, N., J. Thangavelautham, G.M.T. D’Eleuterio (2008). “Infrastructure Robotics: A Technology Enabler for Lunar In-Situ Resource Utilization, Habitat Construction and Maintenance,” in Proceedings of the International Astronautic Conference, Scotland. Anderson, S., J. Thangavelautham (2021). “Solar-Powered Additive Manufacturing in Extraterrestrial Environments,” Proceedings of the Earth and Space Conference. Bradley, D., D. Seward (1998). The development, control and operation of an autonomous robotic excavator. Journal of Intelligent and Robotic Systems, 21:73–97. Beckers, R., O. E. Holland, J. L. Deneubourg (1994). From local actions to global tasks: Stigmergy and collective robots. In Fourth Int. Workshop on the Syntheses and Simulation of Living Systems, pages 181–189. MIT Press. Bonabeau, E., G. Theraulaz, J. Deneubourg, S. Aron, S. Camazine (1997). Self-organization in social insects. In Trends in Ecology and Evolution, volume 12, pages 188–193. Bonabeau, E., M. Dorigo, G. Theraulaz (1999). Swarm Intelligence: From Natural to Artificial Systems. Oxford Univ. Press, New York. Babuscia, A., T. Choi, K. Cheung, J. Thangavelautham, M. Ravichandran, A. Chandra (2015). “Inflatable antenna for CubeSat: Extension of the previously developed s -band design to the X-band,” AIAA Space, 2015. Bame S. J., W. C. Feldman, J. T. Gosling, D. T. Young, R. D. Zwickl (1983). What magnetospheric workers should know about solar wind composition. In Energetic Ion Composition in the Earth’s Magnetosphere (R. G. Johnson, ed.), pp. 73–98. Terra, Tokyo. Babuscia, A., J. Sauder, J. Thangavelautham, E. Choi, A. Chandra (2016). “Inflatable Antenna for CubeSats: Development of the X-band Prototype, Proceedings of the IEEE Aerospace Conference. Babuscia, A., J. Sauder, A. Chandra, J. Thangavelautham (2017). “Inflatable Antenna for CubeSats: A New Spherical Design for Increased X-band Gain,” Proceedings of the IEEE Aerospace Conference. Babuscia, A., J. Sauder, A. Chandra, J. Thangavelautham (2019). “Inflatable Antenna for CubeSat,” Chapter 6, CubeSat Antenna Design, Chahat, N., Editor, Wiley/IEEE, pp. 197231, 2020. Bergman, T. L., A. Lavine, F. P. Incropera. Fundamentals of Heat and Mass Transfer. John Wiley & Sons, Inc. Chantemargue, F., T. Dagaeff, M. Schumacher, B. Hirsbrunner (1996). Implicit cooperation and antagonism in multi-agent systems.

© ASCE

665

Earth and Space 2022

Chandra, A., J. Thangavelautham, A. Babuscia (2018). “Composite Inflatable Antennas for Small-Satellite and Backup Communication,” 24th Ka-band and Broadband Communications Conference, Niagara Fall, Canada. Churchill, Susanne (1997). Fundamentals of Space Life Sciences (Malabar, Florida: Krieger Publishing Company): 13-16. Earon, E., J. Thangavelautham, T. Liu, H. Armstrong, G.M.T. D’Eleuterio, D. Boucher, M. Viel, J. Richard (2009). “A Multiagent Methodology for Lunar Robotic Mission Risk Mitigation,” in Proceedings of AIAA Space Conference. Finkenzeller, Klaus (2010). RFID handbook: fundamentals and applications in contactless smart cards, radio frequency identification and near-field communication. John wiley and sons, Mendell, Wendell W. Lunar bases and space activities of the 21st century. Lunar and Planetary Institute, 1985. Feldman W. C., J. R. Ashbridge, S. J. Bame, J. T. Gosling (1977). Plasma and magnetic fields from the Sun. In The Solar Output and its Variation (O. R White, ed.), pp. 351– 382. Colorado Assoc. Univ., Boulder. Grasse, P. (1959). La reconstruction du nid les coordinations interindividuelles; la theorie de stigmergie. In Insectes Sociaux, volume 35, pages 41–84. Halbach, E., V. Zhmud, A. Halme (2013). Simulation of robotic regolith mining for base construction on mars. In Proceedings of the 12th Symposium on Advanced Space Technologies in Automation and Robotics (ASTRA), pages 1–7. Heiken, G., D. Vaniman, B. M. French (1991). “Lunar sourcebook: a user’s guide to the Moon,” Cambridge (England), Cambridge University Press. Husain, Y. (2000). “Space-Friendly Architecture: Meet Nader Khalili.” Space.com. Hassan, Mohamed A., A. Elzawawi (2015). “Wireless Power Transfer through Inductive Coupling.” Recent Advances in Circuits pp. 115-118. Khalili, N. (2008). “Emergency Sandbag Shelter and EcoVillage Manual,” Cal -Earth Press. Katauskas, T. (1998). “Dirt-Cheap Houses from Elemental Materials.” Architecture Week. Kayser, M. (2011). “Solar Sinter.” Master’s Thesis, Royal College of Art, London. Kurs, A., A. Karalis, R. Moffatt, J. D. Joannopoulos, P. Fisher, and M. Soljacic (2007). “Wireless power transfer via strongly coupled magnetic resonances,” Science, vol. 317, no. 5834, pp. 83–86. Khsohnevis, B., G. Bekey (2002). “Automated Construction Using Contour Crafting: Applications on Earth and Beyond,” International Symposium on Automation and Robotics in Construction, 19th (ISARC). Proceedings. National Institute of Standards and Technology. Lu, X., P. Wang, D. Niyato, D. I. Kim, Z. Han (2016). “Wireless Charging Technologies: Fundamentals, Standards, and Network Applications,” in IEEE Communications Surveys & Tutorials, vol. 18, no. 2, pp. 1413-1452. Mataric, M. J., M. Nilsson, K. T. Simsarian (1995). Cooperative multi-robot box-pushing. In IEEE/RSJ IROS, pages 556–561. McGuire R. E., T. T. von Rosenvinge, F. B. McDonald (1986) The composition of solar energetic particles. Astrophys. J., 301, 938–961. Mottaghi, S., H. Benaroya (2015). “Design of a Lunar Surface Structure. I: Design Configuration and Thermal Analysis.” Journal of Aerospace Engineering 28, no. 1 : 04014052. https://doi.org/10.1061/(asce)as.1943-5525.0000382.

© ASCE

666

Earth and Space 2022

Napp, N., R. Nagpal (2012). Distributed amorphous ramp construction in unstructured environments. In Proceedings of the Distributed Autonomous Robotic Systems (DARS) 12., pages 1–15. Nealy, John E. et al. (1988). Solar-Flare Shielding With Regolith at Lunar-Base Site, NASA Technical Paper 2869 (Hampton, Virginia: Langley Research Center). Parker, C. A., H. Zhang, R. C. Kube (2003). Blind bulldozing: Multiple robot nest construction. In IEEE/RSJ Int. Conference on Intelligent Robots and Systems, pages 2010–2015. [35] Prawiro, Satrio Yudo, and Muhammad Ary Murti (2018). “Wireless power transfer solution for smart charger with RF energy harvesting in public area.” 2018 IEEE 4th World Forum on Internet of Things (WF-IoT). IEEE. Reid, R. (2020). “Space Place”, The Civil Engineering Magazine, American Society of Civil Engineers. Skonieczny, K., D. Wettergreen (2016). Advantages of continuous excavation in lightweight planetary robotic operations. 35:1121–1139. Stewart, R., A. Russell (2003). Emergent structures built by a minimalist autonomous robot using a swarm inspired template mechanism. In The First Australian Conference on ALife (ACAL2003), pages 216–230. Stewart, R., A. Russell (2004). Building a loose wall structure with a robotic swarm using a Spatio-temporal varying template. In IEEE/RSJ Int. Conference on Intelligent Robots and Systems. pages 712–716. Sonawane, N., J. Thangavelautham (2017). “Precision Pointing of Antennas and Science Instruments in Space using Arrays of Shape Memory Alloy based Linear Actuators,” Proceedings of the 40th AAS Guidance and Control Conference, 2017. Simpson, J. A. (1983) Elemental and isotopic composition of the galactic cosmic rays. Annu. Rev. Nucl. Part. Sci., 33, 323–381. Silberberg, R. et al. (1985). “Radiation Transport of Cosmic Ray Nuclei in Lunar Material and Radiation Doses,” Lunar Bases and Space Activities of the 21st Century, NASA Symposium Publication (Houston, Texas: Lunar and Planetary Institute). Tang, Y., J. Fuh, H. Loh, Y. Wong, L. Lu (2003). “Direct laser sintering of a silica sand,” Materials and Design, vol. 24, no. 8, pp. 623–629 Thangavelautham, J., T. Barfoot, G. M. T. D’Eleuterio (2003). Coevolving communication and cooperation for lattice formation tasks (updated). In Advances In Artificial Life: Proc. of the 7th European Conference on ALife (ECAL), pages 857–864. Thangavelautham, J., K. Law, T. Fu, N. Abu El Samid, A. D. S. Smith, G. M. T. D’Eleuterio (2017).” Autonomous Multirobot Excavation for Lunar Applications,” Robotica, pp. 1-33. Thangavelautham, J. (2008). “A Regulatory Theory of Cortical Organization and Its Application to Robotics.” PhD Thesis, University of Toronto Thangavelautham, J., N. Abu El Samid, P. Grouchy, E. Earon, T. Fu, N. Nagrani, G. M. T. D’Eleuterio (2009). “Evolving Multirobot Excavation Controllers and Choice of Platforms Using Artificial Neural Tissue Controllers,” Proceedings of the IEEE Symposium on Computational Intelligence for Robotics and Automation, Injeon, South Korea, pp. 1-10.

© ASCE

667

Earth and Space 2022

Thangavelautham, J., D. Strawser, S. Dubowsky (2017). “Long-Life Micro Fuel Cell Power Supply for Mobile Field Sensor Network Modules,” International Journal of Hydrogen Energy, Vol 42, No. 31, pp. 1-22 Thangavelautham, J., A. Chandra, E. Jensen (2019). “Autonomous Multirobot Technologies for Mars Mining Base Construction and Operation,” 70th International Astronautic Congress. Thangavelautham, J. (2020). “Autonomous Robot Teams for Lunar Mining Base Construction and Operation,” IEEE Aerospace Conference 2020, Big Sky, MT Thangavelautham, J. (2020). “Autonomous Robot Swarms for Off-World Construction and Resource Mining,” 2020 AIAA SciTech Forum, Orlando, FL Thangavelautham, J., Y. Xu (2021). “Modelling Excavation, Site-Preparation and Construction of a Lunar Mining Base Using Robot Swarms,” Earth and Space Conference, Seattle, WA Wawerla, J., G. Sukhatme, M. Matari (2002). Collective construction with multiple robots. In IEEE/RSJ Int.l Conference on Intelligent Robots and Systems, pages 2696–2701. Wilson, M., C. Melhuish, A. B. Sendova-Franks, S. Scholes (2004). Algorithms for building annular structures with minimalist robots inspired by brood sorting in ant colonies. In Autonomous Robots, volume 17, pages 115– 136. Xu, D., R. Su, J. Yeh, M. Momayez (2019). “Mapping Soil Layers Using Electrical Resistivity Tomography and Validation: Sandbox Experiments”, J. Hydr, Vol. 575, pp. 523-536. [42] R. Williams, J. Jadwick, “Handbook of Lunar Materials,” NASA RP -1057, 1980. Xu, Y., J. Thangavelautham, J. (2020). “Co-Evolution of MultiRobot Controllers and Task Cues for Off-World Open Pit Mining,” iSAIRIS Conference

© ASCE

668

669

Earth and Space 2022

Automatic Reading Method for Pointer Meter Based on Computer Vision Weijin Xu1; Weihua Zhang2; Liang Xing3; Hongjun Lu4; Dongyou Li5; and Yang Du6 1

State Grid Changchun Power Supply Company, Email: [email protected] 2 State Grid Changchun Power Supply Company, Email: [email protected] 3 State Grid Changchun Power Supply Company, Email: [email protected] 4 State Grid Changchun Power Supply Company, Email: [email protected] 5 State Grid Changchun Power Supply Company, Email: [email protected] 6 State Grid Changchun Power Supply Company, Email: [email protected]

Chaoyang District. Chaoyang District. Chaoyang District. Chaoyang District. Chaoyang District. Chaoyang District.

ABSTRACT The pointer meter is an essential component of industrial facilities such as substations, oil fields, buildings, and construction. The readings represented by its pointer record various environmental parameters and equipment operating parameters of industrial facilities. The traditional manual meter reading method is time-consuming and requires professionals to operate it, which has a high demand on labor costs. Therefore, we propose a computer vision-based method for the reading of pointer meter. Through image local feature matching, the algorithm realizes the autonomous correction of image in-plane rotation, and effectively reduces the recognition error caused by image in-plane rotation. Moreover, this method can realize the function of automatic reading of the pointer meter without additional training. The results show that the readings obtained by the proposed method are close to those obtained by manual meter reading and meet the requirements of industrial use. INTRODUCTION With the continuous improvement of social productivity, industrial production processes are moving towards intelligence, meanwhile production informatization and digitalization are becoming the trend of development. Instruments are an important component of industrial facilities such as substations, oil fields, construction and building, an important safety guarantee to monitor the production process, and an important source of data acquisition in production informatization. Traditional pointer-type meters, with their excellent reliability, low cost, and resistance to electromagnetic interference, are still widely used in various industrial sites. Unlike digital meters, which have convenient data acquisition and recording methods, most pointer meters do not have data interfaces and cannot collect data independently, relying mainly on manual meter reading methods and requiring high professionalism of meter readers. This is not only time-consuming and labor-intensive, but also has the risk of introducing human errors. With the progress of technology, especially the development of computer vision technology, the use of computers instead of manual meter reading is gradually becoming a reliable means.

© ASCE

Earth and Space 2022

At present, most readout algorithms are based on pictures obtained from fixed meters and fixed cameras, or meter dials without large angle tilt (Li et al. 2014; Li 2014; Yang et al. 2015; Yue and Zhang 2014). In many cases, the algorithms can only achieve good reading accuracy after complex manual processing, which is unrealistic for many industrial sites. In addition, to ensure the accuracy of the readings, the background of the meter dial is often required not to be too messy, and the meter must be correctly placed in the photo without large rotations. Moreover, some meters do not require real-time monitoring, but only periodic recording by inspectors, which would also cause a waste of cost if a fixed camera is installed. If fixed cameras are not used for photography, images captured by other means are often affected by the angle and position of the shooting, and the resulting image rotation creates high requirements for the reading algorithm. Especially for dial rotation, in some places, the meter is in a special position, or the meter is installed with an initial rotation, the meter in the captured image will also have a rotation angle. This will greatly affect the automatic reading of the pointer meter. To solve the above problems, some researchers have proposed feasible methods. Kang et al. (2021) manually crop out the ring area where the scale lines are located, extract the scales with edge detection, then judge the starting and ending positions according to the maximum included angle between the scale lines. Sowah et al. (2021) use high-level feature extraction to identify the scale value of the instrument, conduct binary regression between the scale value and its corresponding angle value, then get the reading according to the pointer angle. This method relies on a large number of feature extraction operations, and the program runs for a long time. Zhou et al. (2020) extract the meter scale lines with the connected region eccentricity, and determined the meter rotation center according to the centrality of the scale lines, which can effectively avoid the center positioning error caused by lens distortion. Bao et al. (2019) eliminate the perspective effect of the acquired meter image by using the inverse perspective mapping, improving measurement precision. In addition, some researchers use deep learning method for pointer meter reading (Fang et al. 2019; Xiao 2021). Zheng et al. (2020) and Meng et al. (2020) use the deep learning method to identify the key points such as the beginning and end position of the scale and the pointer in the dial, and use the angle method to read, which has good robustness to different meters. Alexeev et al. (2020) propose a method combine AMR detector to identify the dial and local detection to identify symbols, generate a grid of the key points for the symbols on the center of meter surface, which has a good recognition effect on the meter images in different environments. but did not pay attention to pointer detection. Meanwhile they do not describe the methods for the pointer’s position determination. Dong et al. (2021) use the deep learning algorithm (VDN) to predict the finger vector to identify the meter pointer, which has good robustness in different environments. Zhang et al. (2021) present a target-key point detection-based recognition method for water meter instruments. The method combines the latest target detection technology (YOLOv4-Tiny) and key point detection technology (an improved RFB-Net). Li et al. (2020) use the depth learning method to identify the meter scale value text, determine the measuring center based on the text coordinates, transform the arc scale area into a linear scale area through polar coordinate transformation, then use the distance method to read.

© ASCE

670

Earth and Space 2022

However, deep learning requires a large number of data set training to ensure the robustness of the model. If the meter is changed, it is likely to need retraining to achieve satisfactory results. As a result, the data had to be collected again, which is labor-intensive work. Therefore, considering the meter rotation, and in order to avoid training, this paper proposes an automatic reading method for pointer meter reading based on local feature matching is proposed. Through image local feature matching, the algorithm realizes the autonomous correction of image in-plane rotation, and effectively reduces the recognition error caused by image in-plane rotation. Moreover, this method can realize the function of automatic reading of the pointer meter without additional training. The results show that the readings obtained by the proposed method are close to those obtained by manual meter reading and meet the requirements of industrial use. METHODOLOGY The pipeline of automatic reading method for pointer meter proposed in this paper is shown in Figure 1. Firstly, the meter image is preprocessed. The methods used include Mean Shift Filtering (Fukunaga and Hostetler 1975), Gray Transferring and adaptive Threshold. Secondly, after obtaining the binary image, the Probabilistic Circle Hough Transform (Ballard 1987) is used to detect the circular meter dial and remove the redundant background outside the meter. Thirdly, for the image containing only the dial area, the Contour Search method is used to identify and locate the meter scales, and then fit the center position of the circle. Fourthly using Line Hough Transform, the position of the pointer can be found. Combined with the position of the center of the circle, the pointer line segment can be obtained. Then, SIFT feature extraction and K-D Trees feature matching is used to match the characters with obvious features on the meter to obtain the reference point, which is the key-point to correct the rotation of the meter. Finally, the dial reading can be calculated out combining the information acquired above.

Figure 1. The pipeline of automatic reading method for pointer meter The principle of the method used will be briefly introduced below.

© ASCE

671

Earth and Space 2022

Mean Shift Filter. Mean Shift Filter algorithm is a general clustering algorithm. Its basic principle is that for a given number of samples, select any one of them, delimit a circular area with the sample as the center point, obtain the centroid of the sample in the circular area, which is the point with the maximum density, and then continue the above iterative process with the point as the center until the final convergence. This feature of Mean Shift Filter algorithm can be used to realize color image segmentation. Adaptive Threshold. Adaptive threshold is a binarization method. General binarization algorithms, such as Otsu algorithm and Maximum Entropy algorithm, belong to global threshold algorithm. For images with uneven illumination, this kind of algorithms has poor recognition accuracy. The adaptive threshold method solves this problem better. Its idea is not to calculate the threshold of the global image, but to calculate the local threshold according to the brightness distribution of different regions of the image. Therefore, different thresholds can be calculated adaptively for different regions of the image. Therefore, it is called adaptive threshold method, that is, local threshold method. The local threshold can be determined by calculating the mean, median and Gaussian weighted average (Gaussian filtering) of a neighborhood (local). Finally, binarization is carried out according to the comparison relationship between each pixel and the local threshold. Hough Transform. Hough Transform was first proposed by Paul Hough in 1962 and then popularized by Richard Duda and Peter Hart in 1972. It is one of the basic methods to detect geometry from images in the field of image processing. The classical Hough transform is used to detect lines in the image. Later, the Hough transform can be extended to recognize objects of arbitrary shape, such as circles and ellipses. Hough Transform transforms the problem of detecting arbitrary shape into the problem of statistical peak value by using the transformation between two coordinate spaces. The basic idea of Circle Hough Transformation is that every non-zero pixel on the image may be a point on a potential circle. Like Line Hough Transformation, it also generates a cumulative coordinate plane by voting and sets a cumulative weight to locate the circle. In the Cartesian coordinate system, the equation of circle is：, where is the center of the circle, and is its radius. And it can also be represented as or. Therefore, in the three-dimensional coordinate system composed of, a point can uniquely determine a circle. In the Cartesian coordinate system, all circles passing through a certain point mapped to the coordinate system is a three-dimensional curve. All circles passing through all non-zero pixels in coordinate system constitute many three-dimensional curves in coordinate system. In the coordinate system, the circular equations of all points on the same circle are the same, and they map to the same point in the coordinate system. Therefore, in the coordinate system, the intersection of all curves is a circle. SIFT feature extraction. Scale Invariant Feature Transformation (SIFT) is a description used in the field of image processing. This description has scale invariance and can detect key points in the image, which

© ASCE

672

Earth and Space 2022

is a local feature descriptor. It has the many advantages. It has good stability and invariance, can adapt to the changes of rotation, scale scaling and brightness, and can be free from the interference of angle change, affine transformation, and noise to a certain extent. And it is fast and can produce many features. It mainly includes three steps. The first step is to extract key points. Key points are some very prominent points that will not disappear due to lighting, scale, rotation and other factors, such as corner points, edge points, bright spots in dark areas and dark spots in bright areas. This step is to search for image positions in all scale spaces. Gaussian differential function is used to identify potential points of interest with scale and rotation invariance. The second step is to locate keys and orient features. At each candidate location, the location and scale are determined by a fine fitting model. The selection of key points depends on their stability. Then, one or more directions are assigned to each key position based on the local gradient direction of the image. All subsequent operations on image data are transformed relative to the direction, scale and position of key points, so as to provide invariance to these transformations. The last step is by comparing the feature vectors of each key point, several pairs of matching feature points are found, and the corresponding relationship between scenes is established. K-D Trees feature matching. Feature matching operators can be roughly divided into two categories. One is the linear scanning method, which compares the distance between the points in the data set and the query points one by one. The disadvantage is obvious, that is, it does not use any structural information contained in the data set itself, and the search efficiency is low. The second type is to establish data index, and then quickly match. Because the actual data generally presents a cluster shape, the speed of retrieval can be greatly accelerated by designing an effective index structure. Index tree belongs to the second category, and its basic idea is to divide the search space hierarchically. According to whether the divided space has aliasing, it can be divided into clipping and overlapping. K-D Trees is the representative of the former, which has no overlapping space; The latter division space overlaps with each other, which is represented by R-tree. K-D Trees is a binary tree structure. Each node of KD tree records the characteristic coordinates, the segmentation axis, the pointer to the left branch and the pointer to the right branch. K-D Trees algorithm can be divided into two parts. The first part is about establishing the data structure of K-D Trees. The other part is how to find the nearest neighbor on the established K-D Trees. Through binary search, we can quickly find the nearest approximate point, that is, leaf node, along the search path. The leaf node found is not necessarily the nearest one. The nearest one must be closer to the query point and should be located in the circle centered on the query point and passing through the leaf node. CASE STUDY The technical route adopted in this paper is shown in Figure 1. The intelligent pointer meter reading algorithm used in this paper mainly includes the following parts: image preprocessing, dial recognition and positioning, scale lines identification and circle center positioning, meter rotation correction, reading recognition. The specific steps of each part will be described in the next five sections.

© ASCE

673

674

Earth and Space 2022

Image preprocessing. Due to the uncertainty of the working environment, shooting equipment, shooting time and place, the original collected meter images will contain many interference factors, which will affect the accuracy of algorithm recognition. In order to improve the applicability of the algorithm, a set of autonomous preprocessing program for the collected image is designed in this paper. 1) Optimize the image size. If the image size is too large, the running time of the algorithm will increase. If the size of image is too small, it will affect the recognition accuracy. After the experiment of multiple images, this paper uniformly sets the width of the input image to 800 pixels. 2) The image is filtered by Mean Shift. The image also contains pixels that interfere with the reading, such as meter logo, image background, etc. The Mean Shift filtering algorithm can be used for smooth filtering at the color level. It can smooth the color details with similar color distribution, erode the color area with small area, and improve the overall smoothness of the image. 3) Gray transferring. Image graying can reduce the storage space occupied by the image and be used as the input image for Circle Hough Transform. Figure 2 shows the changes of the image during the preprocessing.

Figure 2. Image preprocessing Meter dial identification and positioning. For the circular pointer meter, the circular dial is directly extracted and positioned. On this basis, a mask can also be established according to the meter position to filter useless image background pixels. 1) Circle Hough Transform. Input the gray image after Mean Shift filtering, the possible circles can be detected by Circle Hough Transform. one of the circles with the greatest probability is automatically selected as the meter circle. After testing for multiple images and adjusting the input parameters, the dial detection has high reliability. 2) Create a mask. All the detected pixels outside the circle are taken as 1, that is, a white mask is established. In this way, the background is removed from the image, and only the information in the meter dial will be retained in the image. The process and results of meter recognition are shown in Figure 3.

© ASCE

675

Earth and Space 2022

Figure 3. Meter identification and circle positioning Scale lines identification and positioning. The meter dial detected by Circle Hough Transform is a perfect circle. Due to the distortion of the photo, the actual center of the meter may not set on the circle center detected. In order to solve this problem, this paper identifies the scale lines in the image, and calculates the position of the pointer rotation center according to the intersection set of the scales’ extension line. 1) Preprocess the meter image. The Mean Shift filter is no longer applicable due to the need to accurately obtain the marks details. Therefore, taking the extracted meter as the input, Gaussian filtering is used here to smooth the image as much as possible and retain enough details. The obtained image is grayed as the input of binarization. 2) Adaptive threshold binarization. The binary image can not only extract the morphological skeleton features of the meters in the image, but also effectively improve the accuracy of the edge detection algorithm and Line Hough Transform. Figure 4 shows the preprocessing process.

Figure 4. preprocessing process for meter dial 3) Scale lines contour fitting. The generated binary image is searched for contour. Through contour search, all not black parts (scale lines, pointer, and interference point) can be found. According to the characteristics of scale lines, they are filtered according to the following aspects: a) Distance. The center point of the scale line is near the radius area.

© ASCE

676

Earth and Space 2022

b) Aspect ratio. The scale line is a slender area, and the aspect ratio is rectangular, reaching more than 1:4. c) Area. Through the above filtering, the area statistics of the selected contour is carried out. The scale line area now accounts for the majority. Take the area near the statistical median for secondary filtering. 4) Circle center positioning. Refine the obtained scale lines and calculate their extension lines. It is obvious that the tick lines will intersect at the center of the dial. The extension lines set is randomly divided into two sets, and the intersection of the two lines sets is calculated to obtain the node set. The average coordinate of the intersection nodes near the circle center is taken as the requested center of the dial. The process of scale lines recognition and center positioning is shown in Figure 5.

Binary image

Scale lines Identification

Extension lines of scales

Node set

Figure 5. The process of scale lines recognition and circle center positioning Pointer recognition. After further processing the binary image and removing the scale lines and miscellaneous points, we can get an image containing only pointers and discs. The image has good morphological features and Line Hough Transform can be directly used to identify the start and end point of the pointer. It is easy to get that the point further from the center of the circle is the required pointer vertex coordinate. The process of pointer recognition and vertex positioning is shown in Figure 6.

Figure 6. The process of pointer recognition and vertex positioning

© ASCE

677

Earth and Space 2022

Meter rotation correction. If we can find the position of the 0 scale, we can get the line between the 0-scale position and the center of the circle, and then through the included angle between the line and the line segment of the pointer, we can get the reading of the meter. This is hard to do practically due to the meter may rotate in the image, where the position of the 0 scale is not on the centerline of the meter. To solve this problem, this paper uses the image local feature matching method based on K-D Tree to correct the rotation. However, this paper does not directly match the position of 0 scale, because there are multiple ‘0’s on the meter, so ‘0’ cannot be used as a unique matching feature. The specific solution process is as follows: 1) K-D Trees feature matching. First of all, we need to find the unique feature on the meter that is easy to be recognized by the computer. After many attempts, this paper comes to the conclusion that the scale indication ‘50’ is a better choice. Different from other numbers in the meter, ‘50’ only appears directly below the panel, and the extension lines of two ‘50’ lines pass through the center of the meter, which can effectively avoid matching error. The result of matching is shown in Figure 7, which shows good reliability. 2) ‘50’ scale coordinate positioning. As can be seen from Figure 7, multiple key points can be matched for the local region. In this paper, the coordinates of the matched point set in the image are averaged, and the coordinates are set as ‘50’ scale display coordinates. The in-plane rotation correction is carried out according to the coordinate and the central coordinate of the meter.

Figure 7. The result of matching Reading identification. Through the above steps, the absolute coordinate values of meter center 𝑂𝐶 , pointer vertex 𝑃𝐼 and ‘50’ scale indication 𝑍50 in the image have been obtained. Next, the meter reading is obtained by the angle method according to these three coordinates.

© ASCE

678

Earth and Space 2022

As shown in Figure 8, from the coordinate values, the polar coordinate values, which is established clockwise with 𝑂𝐶 as the origin, of vectors 𝑂𝐶 𝑍50 and 𝑂𝐶 𝑃𝐼 can be obtained respectively. Therefore, the included angles between 𝑂𝐶 𝑍50 and X-axis, and 𝑂𝐶 𝑃𝐼 and X-axis are respectively 𝛼 and 𝛽. So the included angle between 𝑂𝐶 𝑍50 and 𝑂𝐶 𝑃𝐼 is 𝛾. And the included 100 angle between pointer and zero scale 𝜃. Therefore, the meter reading 𝑥 = 𝜃 × 2𝜋 .

Figure 8. Schematic diagram of reading RESULT DISPLAY In order to verify the accuracy of the method, 10 images are used to be recognized by the 𝑇𝑒𝑠𝑡−𝑅𝑒𝑎𝑙 algorithm. The error is calculated by 𝐸𝑟𝑟𝑜𝑟 = 𝑅𝑒𝑎𝑙 %. The verification results are shown in Figure 9 and Table 1. It can be seen from Figure 9 and Table 1 that the method proposed in this paper is very accurate in the identification of pointer and 50 scale, meanwhile it can well recognize the reading of meter. Table 1. Verification results Image number 1 2 3 4 5 6 7 8 9 10

© ASCE

Manual reading 81.0 47.2 65.3 81.0 4.5 4.5 39.5 58.0 25.0 96.0

Test reading 80.97 47.44 65.61 80.72 4.69 4.63 39.99 58.04 24.55 95.76

𝑬𝒓𝒓𝒐𝒓 -0.04% 0.51% 0.47% -0.35% 4.20% 2.83% 1.24% 0.06% -1.80% -0.25%

679

Earth and Space 2022

Figure 9. Verification results CONCLUSION For the pointer-type circular meter, an autonomous pointer reading recognition algorithm is proposed in this paper. The algorithm innovatively adopts K-D Trees feature matching technology to effectively solve the problem of image rotation correction. Using this algorithm, the meter dial in the image can be accurately located and extracted, and the meter pointer can be accurately recognized. Through the identified pointer vertex coordinate, meter center coordinate and corrected feature point coordinate, the meter reading can be obtained accurately and quickly. Based on the real meter images, the recognition algorithm proposed in this paper is verified by experiments, and shows good reliability. In conclusion, the recognition algorithm proposed in this paper meets the requirements of industrial use and can be well applied to the actual industrial production site. REFERENCES Alexeev, A., Kukharev, G., Matveev, Y., and Matveev, A. (2020). “A Highly Efficient Neural Network Solution for Automated Detection of Pointer Meters with Different Analog Scales Operating in Different Conditions.” Mathematics, 8(7), 1104. Ballard, D. H. (1987). “Generalizing the Hough Transform to Detect Arbitrary Shapes.” Readings in Computer Vision, M. A. Fischler and O. Firschein, eds., Morgan Kaufmann, San Francisco (CA), 714–725. Bao, Tan, Liu, and Miao. (2019). “Computer Vision Measurement of Pointer Meter Readings Based on Inverse Perspective Mapping.” Applied Sciences, 9(18), 3729. Dong, Z., Gao, Y., Yan, Y., and Chen, F. (2021). “Vector Detection Network: An Application Study on Robots Reading Analog Meters in the Wild.” IEEE Transactions on Artificial Intelligence, 2(5), 394–403.

© ASCE

Earth and Space 2022

Fang, Y., Dai, Y., He, G., and Qi, D. (2019). “A Mask RCNN Based Automatic Reading Method for Pointer Meter.” 2019 Chinese Control Conference (CCC), 8466–8471. Fukunaga, K., and Hostetler, L. (1975). “The estimation of the gradient of a density function, with applications in pattern recognition.” IEEE Transactions on Information Theory, 21(1), 32–40. Kang, L., Li, H., Zheng, H., and Li, X. (2021). “Research on Automatic Recognition Algorithm of Pointer Meter Based on Machine Vision.” 2021 6th International Conference on Intelligent Computing and Signal Processing (ICSP). IEEE, 2021, 265268. Li, Q., Fang, Y., He, Y., Yang, F., and Li, Q. (2014). “Automatic reading system based on automatic alignment control for pointer meter.” IECON 2014 - 40th Annual Conference of the IEEE Industrial Electronics Society, 3414–3418. Li, X. (2014). “Key Technology Research of Point Meter Detection Based on Computer Vision.” Guangdong University of Technology. Li, Z., Zhou, Y., Sheng, Q., Chen, K., and Huang, J. (2020). “A High-Robust Automatic Reading Algorithm of Pointer Meters Based on Text Detection.” Sensors, 20(20), 5946. Meng, X., Cai, F., Wang, J., Lv, C., Liu, H., Liu, H., and Shuai, M. (2020). “Research on Reading Recognition Method of Pointer Meters Based on Deep Learning Combined with Rotating Virtual Pointer.” 2020 5th International Conference on Information Science, Computer Technology and Transportation (ISCTT). IEEE, 2020, 115-118. Sowah, R. R., Ofoli, A. R., Mensah-Ananoo, E., Mills, G. A., and Koumadi, K. M. M. (2021). “An Intelligent Instrument Reader: Using Computer Vision and Machine Learning to Automate Meter Reading.” IEEE Industry Applications Magazine, IEEE Industry Applications Magazine, 27(4), 45–56. Xiao, N. (2021). “Meter Reading System Based on Computer Vision and Deep Learning.” Beijing University of Posts and Telecommunications. Yang, Z., Yuan, Z., Qiao, Y., Dai, Y., and Hu, F. (2015). “Research of Intelligent Recognition Method of Pointer Instrument Based on Image Processing.” Computer Measurement & Control, 23(05), 1717–1720. Yue, X., and Zhang, J. (2014). “Automobile Meter Pointer Recognition Algorithm Based on Image Processing Technology.” Mechanical Engineer, (12), 161–163. Zhang, Q., Bao, X., Wu, B., Tu, X., Jin, Y., Luo, Y., and Zhang, N. (2021). “Water meter pointer reading recognition method based on target-key point detection.” Flow Measurement and Instrumentation, 81, 102012. Zheng, Y., Chen, M., Peng, J., and Qi, D. (2020). “An automatic reading method for pointer meter based on one-stage detector.” 2020 International Conference on Image, Video Processing and Artificial Intelligence. Vol. 11584. International Society for Optics and Photonics. Zhuo, H., Bai, F., Xu, Y. (2020). “Machine vision detection of pointer features in images of analog meter displays.” Metrology and Measurement Systems, 589-599.

© ASCE

680

Earth and Space 2022

Identification Method for Displacement of Substation Structure Based on Machine Vision Weijin Xu1; Weihua Zhang2; Liang Xing3; Hongjun Lu4; Dongyou Li5; and Yang Du6 1

State Grid Changchun Power Supply Company, Chaoyang District. Email: [email protected] 2 State Grid Changchun Power Supply Company, Chaoyang District. Email: [email protected] 3 State Grid Changchun Power Supply Company, Chaoyang District. Email: [email protected] 4 State Grid Changchun Power Supply Company, Chaoyang District. Email: [email protected] 5 State Grid Changchun Power Supply Company, Chaoyang District. Email: [email protected] 6 State Grid Changchun Power Supply Company, Chaoyang District. Email: [email protected] ABSTRACT The building structure and equipment of substations are exposed and are subject to environmental factors, which can produce deformation. Once the equipment of the substation fails due to excessive displacement, it will seriously endanger the substation safety and affect the power supply. The traditional manual inspection method of a substation is challenging to observe the slight deformation of the substation. Thus, a displacement recognition method of substation is proposed based on vision. The state-of-the-art object recognition technology was applied to build a calibration object recognition model. In recognition of structural displacement, the calibrator is fixed on the substation structure. The camera recognizes the pixel change of the calibrator, and then the structural displacement is calculated. Finally, the proposed method is validated. Results show that the application of the mobile terminal-based method proposed in this paper is considerably close to the actual displacement, which meets the requirements for industrial use. INTRODUCTION Substation is the place where the voltage is changed and the voltage and current of electric energy are transformed, concentrated and then distributed in the power system. In order to ensure the quality of electric energy and the safety of equipment, voltage adjustment, power flow control and protection of transmission and distribution lines and main electrical equipment are also needed in the substation (Gao & Zhang, 2006). Power flow refers to the direction and distribution of voltage, current and power in nodes and branches of a power system. The main facilities of substation include distribution equipment, power transformer, control equipment, automatic protection device, communication facilities and compensation device, etc. In normal circumstances, the service time of the substation structures can be as long as several decades. However, the building structure and equipment of substations are exposed all year round and are subject to wind, rain, snow, and other environmental factors in the actual situation, which leads that steel material of the transmission tower no longer meets the specified mechanical properties

© ASCE

681

682

Earth and Space 2022

and can easily produce deformation, even collapse (McDonald, 2003; Thompson & Wall, 1981). Once the equipment of the substation fails due to excessive displacement, it will seriously endanger the safety of the substation and may affect the power supply. Although the location selection of substation construction has been changed to indoor or underground and the original open outdoor equipment has been gradually developed to fully enclosed gas combination appliances and semi-enclosed gas combination appliances (Carrano et al., 2005; Qin et al., 2019), there are still a large number of existing open-air substations. The traditional manual inspection method of a substation is challenging to observe the slight deformation of the substation, and the manual inspection greatly reduces the efficiency and accuracy of detection. With the development of machine vision, more and more scholars have applied it to engineering field, and made some achievements(Lei, Ren, Wang, Huo, & Song, 2020; Lei, Wang, Xu, & Song, 2018; Luo & Feng, 2018; Spencer, Hoskere, & Narazaki, 2019; Xiao-wei & Chuan-zhi, 2019). Therefore, an intelligent substation damage detection method is needed to improve the detection efficiency and accuracy. Based on this, a substation displacement monitoring method based on machine vision target detection technology is proposed. In this method, marker points are set on the substation structure, and then the displacement of the structure can be obtained by the displacement transformation of marker points. METHODOLOGY The target recognition method based on machine vision (Li, Yin, & Huang, 2010; Wang, Zhang, Dou, & Sugisaka, 2014) mainly includes three steps: region of interest extraction, target recognition and coordination transformation. Region of interest extraction In order to improve the accuracy of recognition, the region of interest containing the targets is extracted first based on the grey values of images (Kuang et al., 2016). The image is colored, and made up of red, green and blue (GRB) light in different proportions. A set of RGB is the smallest display unit. Normally, RGB has 256 levels of brightness, ranging from 2 to 256. When RGB is different, the brightness is equal to the sum of the three. Figure 1 shows the RGB values corresponding to different pixels and colors. 255

255

255

0

0

0

100

100

100

120

172

180

50

100

150

Figure 1 the RGB values corresponding to different pixels and colors OpenCV provides a real-time optimized Computer Vision library, tools, and hardware (Brahmbhatt, 2013). In opencv, ‘cv2. InRange’ function is used to binarize the image and obtain the mask, in which lower and upper are the two limits of the threshold. The calculation formula is as follows: lower [ x, y, z ] (1)

upper [ X , Y , Z ]

© ASCE

(2)

683

Earth and Space 2022

255 x B(i ) X ; y G (i ) Y ; z R(i ) Z (3) 0 others Then, ‘cv2.findContours’ function is used to get the outermost contour of the target, and then fill the interior of the contour with black. After a series of image processing, the image containing the region of interest is finally output, shown in Figure 2. dst (i )

(a) the original image

(b) Binarized image

(c) Binary non-compute image

(d) Filled image

(e) Binary non-compute image

(f) Superimposed image

Figure 2 The process of obtaining an image of the region of interest Target recognition The main purpose of target recognition is to obtain the coordinates of the center point of the target circle. In the region of interests, the center method is used to calculate the coordina tes of the center of the circle, and the formula is as follows. 0 dst ( x, y ) 0 f ( x, y ) (4) 1 dst ( x, y ) 255

( x0 , y0 )

f ( xi , yi ) xi

,

f ( xi , yi ) yi

(5) f ( xi , yi ) f ( xi , yi ) Where, (𝑥0 , 𝑦0 ) represent the coordinates of the center of mass in pixels. (𝑥𝑖 , 𝑦𝑖 ) is the coordinates of the ith pixel. Coordinate transformation In order to obtain the actual displacement of the structure in the image, it is necessary to establish the relationship between the pixel coordinates and the actual coordinates (Ahn, 2011; Choulakian, 2003). The scale factor is obtained based on the size of the known reference target and its corresponding pixel size. The target circle radius used in this paper is 10 mm and 15 mm, respectively.

© ASCE

684

Earth and Space 2022

EXPERIMENTS SETUP To verify the validity of the proposed method, experiments were designed and performed. The displacement monitoring method based on machine vision mainly consists of two parts: machine vision module and marking module. The machine vision module consists of an industrial camera, telephoto lens, raspberry PI, display screen, keyboard and mouse. The marking module is a black plate marked with two white circles. Data is transferred between the industrial camera and the raspberry PI via a data link. Raspberry PI is an ARM-based microcomputer motherboard. It has the advantages of small size, light weight and low price. Raspberry PI performs image processing and data calculation through Python algorithm, and outputs structural displacement changes. Moreover, a sensor board is fixed on the raspberry, which can measure acceleration, inclination and other parameter information. Linear variable differential transformer (LVDT) is used as a comparison and stepper motor is applied to load or unload. The parameters of devices are listed in Table 1, and devices are shown in Figure 3. Table 1 The parameters of devices in experiments Devices Industrial camera Telephoto lens Raspberry PI

Brand Ruishikeni JAR Weixuedianzi

(a) Industrial camera

(b) Telephoto lens

Model RS-500C 420-800mm/f 8.3-16 PI4B 4G

(c) Raspberry PI

(d) The sensor board

Figure 3 Hardware devices used in experiments RESULT AND DISCUSSION Step tests To verify the validity of the proposed method, step tests were designed and performed. The loading step was 0.1 mm. The total displacement was 1mm after 10 loading times, and then unloading in the same step until the displacement was 0 mm. The results of the comparison are displayed in Figure 4. From Figure 4, the displacement curves measured by machine vision and LVDT have good consistency and stability, which indicates that the proposed method based on machine vision is effective and accurate. Moreover, some data in the green box is chosen and their average of every step is calculated. The results are listed in Table 2 and the unit is mm. MaVi represents the results measured by the proposed method based on machine vision and LVDT is the results measured by LVDT.

© ASCE

685

Earth and Space 2022

Displacement (mm)

LVDT Machine vision

Time (s)

(a) The curves of displacement

Difference (mm)

Difference

Time (s)

(b) The change of difference

Figure 4 The displacement curves and the errors with the step of 0.1 mm Table 2 The average and errors in green box in displacement tests MaVi LVDT Difference Error (%)

0.591 0.605 0.014 2.369

0.694 0.710 0.016 2.305

0.797 0.809 0.012 1.506

0.901 0.913 0.012 1.332

1.002 1.011 0.009 0.898

0.901 0.914 0.013 1.443

0.799 0.814 0.015 1.877

0.697 0.714 0.017 2.439

0.592 0.607 0.015 2.534

The errors are calculated to distinguish the difference between the proposed method and LVDT. With the increase of the movement, the average value is corresponding to increase. Comparison shows the difference changes within 0.017, and the error ranges from 0.898% to 2.534%. The difference is stable, which may be caused by systematic errors. Long-term tests Moreover, to verify the stability of the proposed method, long term tests were performed. The marker was pasted on the structure. A series of data is obtained, and the results in seven days are drawn in Figure 5. The temperature changes from 5 ℃ to 18 ℃. From Figure 5, the results of long-term monitoring are satisfactory. Although the monitoring data is changed in one day, the fluctuation is small, only 0.2 mm. Besides, the fluctuation is periodic, which also shows that the proposed method is feasible and can effectively monitor the change of small displacement in structures.

© ASCE

686

Earth and Space 2022

Displacement (mm)

sunny day with the wind

rainy day with the wind

Time (s)

Figure 5 Displacement results of long-term monitoring CONCLUSIONS The building structure and equipment of substations are exposed all year round and are subject to wind, rain, snow, and other environmental factors, which can easily produce deformation. Once the equipment of the substation fails due to excessive displacement, it will seriously endanger the safety of the substation and may affect the power supply. The traditional manual inspection method of a substation is challenging to observe the slight deformation of the substation, so we propose a displacement recognition method of substation structures based on vision, which uses computer vision technology to identify the deformation of substation structures. Different tests were designed and carried out. Results show that the proposed method is effective and can monitor the change of small displacement in structures. The proposed method based on machine vision will serve as a promising measuring alternative in structural health monitoring of substation structures. REFERENCES Ahn, B. S. J. E. J. o. O. R. (2011). Compatible weighting method with rank order centroid: Maximum entropy ordered weighted averaging approach, 212(3), 552-559. Brahmbhatt, S. (2013). Practical OpenCV: Apress. Carrano, E., Takahashi, R., Cardoso, E., Saldanha, R., Neto, O. J. I. P.-G., Transmission, & Distribution. (2005). Optimal substation location and energy distribution network design using a hybrid GA-BFGS algorithm, 152(6), 919-926. Choulakian, V. (2003). The optimality of the centroid method. Psychometrika, 68(3), 473475. Gao, X., & Zhang, P.-c. (2006). Main features and key technologies of digital substation. Power System Technology, 30(23), 67-71, 87. Kuang, H., Chen, L., Gu, F., Chen, J., Chan, L., & Yan, H. J. I. I. s. (2016). Combining region-of-interest extraction and image enhancement for nighttime vehicle detection, 31(3), 57-65.

© ASCE

Earth and Space 2022

Lei, B., Ren, Y., Wang, N., Huo, L., & Song, G. (2020). Design of a new low-cost unmanned aerial vehicle and vision-based concrete crack inspection method. Structural Health Monitoring-an International Journal, 19(6), 1871-1883. doi:10.1177/1475921719898862 Lei, B., Wang, N., Xu, P., & Song, G. (2018). New Crack Detection Method for Bridge Inspection Using UAV Incorporating Image Processing. Journal of Aerospace Engineering, 31(5). doi:10.1061/(asce)as.1943-5525.0000879 Li, X., Yin, M., & Huang, S. (2010). Design of Target Recognition System Based on Machine Vision. Paper presented at the 2010 International Conference on Computational Intelligence and Software Engineering. Luo, L., & Feng, M. Q. (2018). Edge-Enhanced Matching for Gradient-Based Computer Vision Displacement Measurement. Computer-Aided Civil and Infrastructure Engineering, 33(12), 1019-1040. doi:10.1111/mice.12415 McDonald, J. D. (2003). Substation automation. IED integration and availability of information. IEEE Power & Energy Magazine, 1(2), 22-31. doi:10.1109/mpae.2003.1192023 Qin, C., Li, B., Shi, B., Qin, T., Xiao, J., & Xin, Y. J. M. (2019). Location of substation in similar candidates using comprehensive evaluation method based on DHGF, 146, 152-158. Spencer, B. F., Jr., Hoskere, V., & Narazaki, Y. (2019). Advances in Computer Vision -Based Civil Infrastructure Inspection and Monitoring. Engineering, 5(2), 199-222. doi: 10.1016/j.eng.2018.11.030 Thompson, G. L., & Wall, D. L. (1981). A branch and bound model for choosing optimal substation locations. IEEE Transactions on Power Apparatus and Systems, 100(5), 26832688. doi:10.1109/tpas.1981.316784 Wang, J., Zhang, X., Dou, H., & Sugisaka, M. J. J. R. N. A. L. (2014). Study on the Target Recognition and Location Technology of Industrial Sorting Robot Based on Machine Vision, 1(2), 108-110. Xiao-wei, Y., & Chuan-zhi, D. (2019). Review of computer vision-based structural displacement monitoring. China Journal of Highway Transport, 32(11), 21.

© ASCE

687

688

Earth and Space 2022

Applying Architectural Design and Construction Principles to Lunar and Martian Construction Erin Brayley|Werkema1 and Patrick Suermann, Ph.D, P.E., F.ASCE2 1

Graduate Student, Dept. of Construction Science, Texas A&M Univ. Email: [email protected] 2 Associate Professor and Dept. Head, Texas A&M Univ., College Station, TX. Email: [email protected] ABSTRACT Before and since Vitruvius Pollio’s recognition as the first “architect,” the basic principles of architectural design and construction have been debated and reworded with each influential individual defining specific principles of design pertaining to their own era of design. Despite all the debate, at their core, these principles outline the various considerations required to create a logical, inspired, and beautiful design for a space, a building, or even a city. The advancement of technology has challenged some of these primary principles and added additional complexity to their traditionally understood definitions. Similarly, construction practices have advanced due to the development of technologies such as computer assisted drafting and the more recent advent in the last two to three decades of building information modeling. No longer are people carving large blocks of stone by hand in order to form a wall. However, the progression of the advancement of construction practices has not evolved at the same rate as technological advancement. This disconnect creates exponential challenges when considering construction methodologies, architectural design, and materiality features in extraterrestrial construction. Understanding the advantages and limitations of the known techniques and principles provides a launch point for future development specifically designed for extraterrestrial construction requirements. Beyond the traditional design principles, sustainability and materiality must move to the forefront for the project's consideration due to the expense and challenge of moving materials, products, and construction teams to a lunar or Martian environment. This means traditional principles and practices for architecture and construction need to be developed and adapted to meet the unique requirements of the challenges associated with an extraterrestrial project site. Use of native, local materials takes on a new meaning when considering the types of resources available in a Martian or lunar environment. Site selection and orientation is also limited by topography, soil (regolith) conditions, and the availability of adjacent spaces for landing pads or longer distance “road” networks. Finally, simplicity in constructability becomes of paramount importance when considering the autonomous and robotic techniques likely to be required when preparing for extraterrestrial construction. INTRODUCTION Before written, recorded history, visual representations depicting humanity’s fascination with and understanding of space and the astrological bodies dancing in the heavens can be found scattered around the world in ancient cave paintings and architectural monuments (Prehistoric cave art reveals ancient use of complex astronomy, 2018). Prehistoric people used the planets and stars to keep track of seasons and dates, aid in migrations, and, on occasion, predict

© ASCE

Earth and Space 2022

catastrophe. During the renaissance, astronomers such as Brahe, Copernicus, and Kepler, (Whose Revolution? Copernicus, Brahe & Kepler, 2021) discovered that space was another physical place, not merely a dome covering the earth; and that the planets and stars were other heavenly bodies that moved according to specific physical laws. In the 1860s, a man named Edward Everett Hale wrote a book called “The Brick Moon” which told the story of a satellite that unexpectedly entered orbit, with people on board (Mann, 2012). This became one of the first ideations of a space station. In the early 1900s, George Melies produced the movie “A Trip to the Moon,” describing one of the first expeditions to another astronomical body (A Trip to the Moon (1902) - IMDb, n.d.). In the early 20th century, the first and second world wars produced technological developments, specifically in regards to rocket science. On October 4th, 1957 (Sputnik launched, 2021), The Soviet Union launched the first satellite, Sputnik, into space. This launch and successful orbital insertion initiated the competition that came to be known as the Space Race. During the second half of the 20th century, the Space Race fueled some of the most impressive feats of human innovation and triumph to date. After several successful unmanned satellite launches, the challenge evolved to creating vehicles that were robust enough to maintain human life while outside of the protective atmosphere of Earth. In 1961, the first human, Yuri Gagarin, was launched into space aboard the Soviet Union’s spacecraft, Vostok, and returned safely to Earth (Redd, 2018), proving the possibility of creating an artificial environment for life off of the home planet. Since this moment, the movement towards creating and maintaining an artificial environment has expanded. The International Space Station (ISS), often credited with being one of the most complex international scientific and engineering projects in history, is the largest structure humans have ever constructed in space. Five different space agencies, the National Aeronautics Space Administration (NASA), Roscosmos, the Canadian Space Agency (CSA), the Japan Aerospace eXploration Agency (JAXA) and the European Space Agency (ESA) (NASA, 2011), representing fifteen countries have spent approximately $100 billion (Minkel, 2010) constructing this technological marvel. Celebrating its 20th year of continual human occupation, the International Space Station’s life expectancy has already been extended nearly 10 years beyond initial predictions (Fecht, 2015). So, while the ISS continues to remain operational, there is an escalating need to study design and construction methodologies to develop the future of extraterrestrial habitation. There are numerous considerations that need to be taken into account while pursuing the development of space habitation technologies. Technical, scientific challenges revolving around payload capacity, materiality, microgravity, atmospheric containment and structural resilience will need to be resolved as long term extraterrestrial habitats develop. However, the technological needs for a project of this scope, should not disregard the long history of sustainable architectural and construction principles and their applications for extraterrestrial conditions. BACKGROUND Sustainable Principles Traditionally, there are six (6) principles that are referenced and used to guide sustainable design and construction practices: optimizing site potential, minimizing the consumption of nonrenewable resources, encouraging the use of environmentally friendly products and materials, protecting and conserving water, enhancing indoor environmental air quality, and finally

© ASCE

689

690

Earth and Space 2022

optimizing operational and maintenance practices (General Services Administration, 2021). Terrestrially, optimizing the potential of a site has to do with choosing the right site, while taking the location, landscape, orientation and reuse of existing buildings, transportation infrastructure and energy systems into consideration. In the case of a lunar or martian habitat or facility development, the criteria of site selection has slightly different implications. While the Moon and Mars do have unique and varied topographies (see Figure 1 and 2) that should be taken into account, the purpose of the facility will have the most significant impact on the location selection based on what natural features or resources will need to be adjacent to the facility for study.

Figure 1. Unified Geologic Map of The Moon (Fortezzo, 2020)

Figure 2. Geologic Map of Mars (Tanaka, 2014)

© ASCE

Earth and Space 2022

In terrestrial construction, sustainable buildings aim to minimize non-renewable energy consumption by focusing on various ways to limit the overall required energy load, improve equipment and system efficiency and promote renewable energy sources such as on site solar production. Sustainable buildings also aim to utilize environmentally friendly materials that are intelligently developed, responsibly produced and integrated into the construction to minimize overall environmental impact. With growing concerns over water scarcity, even in the developed world (UNICEF, n.d.), using water efficiently is increasingly important in sustainable construction. This includes, not only minimizing potable water usage where possible, but also implementing techniques such as rainwater harvesting or onsite grey-water treatment. People spend approximately 90 percent of their lives indoors (Jafari, 2015) and in an extraterrestrial environment, people will spend 100 percent of their lives indoors, either inside the facility, a mobile unit such as a rover, or their spacesuits, which are essentially miniaturized mobile habitats. The prevalence of sick building syndrome (SBS) in a terrestrial environment, where air can be exchanged with the exterior and there are opportunities for individuals to spend time outside of a created habitat, is already estimated to impact 30% of office buildings in the United States (United States EPA, 1991). Factors such as inadequate ventilation, chemical or biological contaminants, and a lack of natural light within a space are most frequently attributed to the symptoms associated with SBS. Finally, working to optimize a building's operation and maintenance requirements improves building performance, reduces energy and resource costs, and helps to prevent critical system failures. The remainder of this research paper will briefly discuss the historical development of sustainable construction practices and a select few of the numerous rating systems that exist today to quantify the environmental performance of buildings. After discussing this history, an in depth analysis of the Leadership in Energy and Environmental Design (LEED) Building Design & Construction: New Construction scorecard will be conducted. The various attributes will be discussed, focusing on their applicability to the development and construction of an extraterrestrial habitat. History of Sustainable Construction In the early 1990s, the term global warming was just beginning to emerge in the public consciousness (Kibert, 2016). The global population was growing from only 5.2 billion people in 1990 (Kibert, 2016) to over 7.9 billion in 2021 (Current World Population, n.d.). The United Nations projects this growth will continue rapidly and the global population will exceed 10 billion by 2057 (Current World Population, n.d.). Nearly double the population of a mere 67 years later. This type of population boom has never before been experienced in human history (see Figure 3). In addition to this booming population growth, global temperatures have been on the rise. Although the Earth has experienced hotter temperatures than today over it’s 4.54 billion year history (Scott, 2014), the current average temperatures are 75% hotter (Scott, 2014) than th ey have been over the past 11,000 years (see Figure 4). This 1.18℃ (Global surface temperature, 2021) rise in temperature over the past century from some of the coldest in recent history to some of the hottest (see Figure 5) raises concerns for the scientific community. Similarly, global CO2 atmospheric concentrations, from human activities, have also significantly increased during this same time period (see Figure 6).

© ASCE

691

692

Earth and Space 2022

Figure 3. Population, 1000 to 2021 (Our World in Data, n.d.)

Figure 4. Global temperature anomalies over the past 11,300 years compared to historic average (1961-1990). “The purple line shows the annual anomaly, and the light blue band shows the statistical uncertainty (one standard deviation). The gray line shows temperature from a separate analysis spanning the past 1,500 years.” (Scott, 2014)

© ASCE

693

Earth and Space 2022

Figure 5. Global Land-Ocean Temperature Index (Global surface temperature, 2021)

Figure 6. Global CO2 atmospheric concentration (Our World in Data, 2018) As climate awareness and the idea of ‘sustainable building’ grew in prominence, several organizations developed systems to quantify the various criteria that had been determined to positively impact sustainability. In the United Kingdom, the Building Research Establishment Environmental Assessment Method (BREAM) was launched in 1989. This rating system became the first successful

© ASCE

Earth and Space 2022

effort in standardizing assessments of ‘green construction’. To date, over 1 million buildings have been registered under BREAM (Kibert, 2016). Japan launched the Comprehensive Assessment System for Building Environmental Efficiency in 2004 (Kibert, 2016). Australia followed up, two years later, with Green Star and in 2009, the German Green Building Council and the German Government launched Deutsche Gesellschaft für Nachhaltiges Bauen (DGBN). The United States Green Building Council (USGBC) publicly launched the Leadership in Energy and Environmental Design (LEED) rating system in 2009 (U.S. Green Building Council, n.d.). This green building movement is often considered to be one of the most successful environmental movements in America. Since its introduction, the LEED rating system has gone through several revisions and has celebrated several significant milestones in terms of project registration. In April 2004, 100 projects had been certified by the USGBC (U.S. Green Building Council, n.d.). LEED v2009 introduced weighted credits that underlines the increasing rigor and objective scientific intentions in assigned credit values. By 2010, USGBC certified the 5,000th LEED project (U.S. Green Building Council, n.d.). In 2015, the newest version of LEED was introduced (U.S. Green Building Council, n.d.). Version 4, and subsequently version 4.1, released, providing improvements to the rating system. These improvements include increased credit flexibility, a smart grid approach, and a large emphasis on the materials and resources that are utilized to build the project. APPLYING LEED ANALYSIS TO EXTRATERRESTRIAL CONSTRUCTION The LEED scorecard is broken up into nine sections: Location and Transportation, Sustainable Sites, Water Efficiency, Energy and Atmosphere, Materials and Resources, Indoor Environmental Quality, Integrative Process, Innovation and Regional Priority (U.S. Green Building Council, 2019). Each of these sections is further disaggregated into specific requirements and opportunities to demonstrate environmental sustainable practices in the built environment. The remainder of this research will analyze the credits within each category, excluding the Integrative Process, Innovation and Regional Priority categories, that could potentially be applied to the construction of an extraterrestrial habitat. Location and Transportation The Location and Transportation credits focus on, obviously, the location of the project, and its access to various types of transportation. In the Building Design and Construction rating system, the first credit listed under Location and Transportation is the LEED for Neighborhood Development. This credit is unique because earning points under this credit precludes the project from attaining any other credits in this category. LEED for Neighborhood Development is an entirely separate rating system within LEED. For the purposes of this study, only the applicable Neighborhood Development Credits will be analyzed because the intention behind LEED for Neighborhood Development aligns most accurately with the development considerations that should be taken into account for the creation of a lunar or martian habitat. Sustainable Sites In LEED v4, the Sustainable Sites credit category was developed to acknowledge that buildings and the landscapes on which they are built are deeply intertwined. Protecting and preserving the natural environment throughout every step of the construction process earns credits in this section. On the lunar surface, the natural environment is completely alien to what

© ASCE

694

695

Earth and Space 2022

is experienced terrestrially. Because of this, the Site Assessment credit becomes even more critical to the success of a sustainable lunar development project. This credit requires project teams to gather an in-depth understanding of the site conditions in terms of topography, hydrology, climate, soils, and the potential for human use and health effects. The other cred it that is applicable to a lunar or martian environment is the Construction Activity Pollution Prevention Prerequisite. While the soil erosion and waterway sedimentation aspects of this credit may not be completely applicable to an off-world project, the portion that deals with controlling airborne dust becomes critical. “The lunar regolith has grain-size characteristics similar to a silty sand, with mean grain sizes mostly between 45 to 100 µm. Many of the grains are sharp and glassy; analogies have been made to fine-grained slag or terrestrial volcanic ash. This fine material has very low electrical conductivity and dielectric losses, permitting accumulation of electrostatic charge under ultraviolet (UV) irradiation (Heiken, 2005).” Water Efficiency NASA’s Stratospheric Observatory for Infrared Astronomy (SOFIA) confirmed the presence of water (H2O) on the sunlit lunar surface (NASA, 2020). While this discovery opened up possibilities and questions regarding deep space exploration, the fact remains that the lunar surface is 100 times drier than the Sahara Desert (NASA, 2020), which means that water, as a resource, will be limited. LEED’s water efficiency credits, especially Indoor Water Use Reduction and Optimize Process Water Use both have applications for sustainable development of an extraterrestrial environment. The International Space Station (ISS) has already demonstrated the validity of a water recycling process (see Figure 7). Implementation of this technology at a large scale will be necessary to support, not only the individuals living in a lunar habitat, but any process water that may be required for equipment and facilities to function.

Figure 7. Water Recycle Process on the ISS (Sullivan, 2019)

© ASCE

Earth and Space 2022

Energy and Atmosphere The Energy and Atmosphere credit category is focusing on minimizing the use of fossil fuels and to promote better building energy performance. In an environment, such as the lunar surface, energy efficiency is paramount to success. Because this project will not tie into an existing grid, it is important that any and all energy needs are produced locally and used as efficiently as possible. Fundamental and Enhanced Commissioning are used on sustainable projects to support the design, construction, and operation of the project to meet the energy, water, indoor environment and resiliency requirements of the project (U.S. Green Building Council, 2019). Minimizing and Optimization of Energy Performance sets a minimum level of efficiency for the building and it’s required systems and establishes systems and processes that will meet and exceed those limitations, reducing the overall building energy consumption. Finally, to help meet the energy requirements of the facility, LEED rewards projects that install onsite renewable energy generation, typically in the form of solar panels. As this is an environment without existing infrastructure, all energy required for the project must be produced locally. Materials and Resources The Materials and Resources category for LEED credits presents a unique challenge for a lunar project. Because of the large expense, both financially and environmentally, associated with transporting materials from Earth to the Moon, material selection and use must be thoughtfully considered prior to execution. This aspect of lunar construction would be tracked in the Building Life-Cycle Impact Reduction credit, which requires projects to demonstrate a 10% reduction from baseline in at least three of the following impact categories, without allowing an increase of more than 5% in any of the categories: global warming potential, depletion of the stratospheric ozone layer, acidification of land and water sources, eutrophication, formation of the tropospheric ozone or depletion of nonrenewable energy resources. The three Building Product Disclosure and Optimization credits reward projects that utilize certified, environmentally friendly products in the construction of the project. Healthcare facilities also offer additional credits for designing for flexibility, which could apply to an Innovation credit under the New Construction rating. Finally, reducing the waste during construction by diverting multiple materials streams from a landfill would also be applicable to an extraterrestrial project, as all waste materials would have to be either recycled or potentially incinerated for energy production. Indoor Environmental Quality The final credit category analyzed in this paper is Indoor Environmental Quality. As noted previously, the prevalence of SBS causes major concerns when a majority of time is spent in an indoor environment. These concerns are exacerbated in a lunar habitat scenario, because 100% of time will be spent within an “indoor” environment. Under the Indoor Air Quality Performance credit, LEED sets minimum thresholds for mechanical ventilation of indoor spaces based on ASHRAE Standard 62.1-2010 (U.S. Green Building Council, 2019). Because there is no air in the exterior environment, all air within the facility must be recycled and reused without compromising the quality of the interior atmosphere for occupants. Prohibiting the use of tobacco and smoking within the facility, in addition to meeting the requirements of another credit

© ASCE

696

Earth and Space 2022

in this category, helps to facilitate the recycle and reuse process. During construction, management of the air quality will also prove important in a lunar environment. Because the lunar regolith is so small, and often carries an electromagnetic charge (Heiken, 2005), managing regolith disruption and intrusion into the facility during the construction may actually prove more feasible than attempting to flush it out after construction is completed and the facility has been sealed. For the occupants of the facility, the LEED requirements for Low Emitting Materials will help to maintain the indoor air quality by preventing or limiting materials that produce offgassing of volatile organic compounds (VOCs). Occupant well being is largely influenced by Thermal Comfort, Interior Lighting and Acoustic Performance of the space they work and live in. Meeting the thresholds set forth in these LEED credits helps to provide an environment suitable for full time habitation. The Daylighting and Quality Views credits of this category are more directed at the mental health of the occupants. However, the requirements for these credits, and the limitations of the lunar environment may make them challenging to achieve. It is possible that a project team would be able to pursue these credits with alternative methodology reviewed and approved by the U.S. Green Building Council, but that is currently outside the scope of this review. CONCLUSION The history of measurable sustainable construction on Earth has been limited to the past several decades, when the problems of climate change or the concept of ‘green’ first arose within the public consciousness. However, the developments that have occurred in this sector of the AEC industry during that time have produced incredible results and improvements. Working to apply what is known to an unknown endeavor, such as a lunar or martian habitat, is critically important and organizations, such as NASA, are in the process of developing plans for long term space exploration and habitation. This brief review demonstrates the possibility of achieving a LEED Gold certified building (see Figure 8).

© ASCE

697

698

Earth and Space 2022

Figure 8. Hypothetical LEED Checklist for Lunar Habitat

© ASCE

Earth and Space 2022

REFERENCES Current World Population. (n.d.). Retrieved December, 2021, from https://www.worldometers.info/world-population/#ref-1 Fecht, S. (2015). How Many Years Does the Space Station Have Left? [online] Popular Science. Available at: https://www.popsci.com/how-many-years-does-space-station-haveleft/ [Accessed 6 Apr. 2021]. Fortezzo, C. M., Spudis, P. D., & Harrel, S. L. (2020, March 3). Unified Geologic Map of the Moon [Map]. In USGS Astrogeology Science Center. Retrieved December, 2021, from https://astrogeology.usgs.gov/search/map/Moon/Geology/Unified_Geologic_Map_of_the _Moon_GIS_v2 General Services Administration. (2021, March 17). Sustainable design. Retrieved December, 2021, from https://www.gsa.gov/real-estate/design-construction/designexcellence/sustainability/sustainable-design Global surface temperature. (2021, October 20). Retrieved December, 2021, from https://climate.nasa.gov/vital-signs/global-temperature/ Heiken, G., Vaniman, D., & French, B. M. (2005). Lunar sourcebook: A user's guide to the Moon. Houston: Cambridge University Press / Lunar and Planetary Institute. Retrieved December, 2021, from https://www.lpi.usra.edu/publications/books/lunar_sourcebook/pdf/Chapter03.pdf History.com Editors (2019). Sputnik launched. [online] HISTORY. Available at: https://www.history.com/this-day-in-history/sputnik-launched [Accessed 5 Apr. 2021]. IMDb. n.d. A Trip to the Moon (1902) - IMDb. [online] Available at: https://www.imdb.com/title/tt0000417/plotsummary [Accessed 3 May 2021]. Jafari, M. J., Khajevandi, A. A., Mousavi Najarkola, S. A., Yekaninejad, M. S., Pourhoseingholi, M. A., Omidi, L., & Kalantary, S. (2015). Association of Sick Building Syndrome with Indoor Air Parameters. Tanaffos, 14(1), 55–62. Kibert, C. J. (2016). Sustainable construction: Green building design and delivery. Hoboken, NY: Wiley. Retrieved December, 2021, from https://ebookcentral.proquest.com/lib/tamucs/detail.action?docID=4462528# Mann, A., 2012. Strange Forgotten Space Station Concepts That Never Flew. [online] Wired. Available at: https://www.wired.com/2012/01/space-station-concepts/ [Accessed 3 May 2021]. Minkel, J. (2010). Is the International Space Station Worth $100 Billion? [online] Space.com. Available at: https://www.space.com/9435-international-space-station-worth100-billion.html [Accessed 6 Apr. 2021]. NASA. (2011). International Cooperation. [online] Available at: https://www.nasa.gov/mission_pages/station/cooperation/index.html [Accessed 8 Apr. 2021]. NASA. (2020, October 26). NASA’s SOFIA Discovers Water on Sunlit Surface of Moon [Press release]. Retrieved December, 2021, from https://www.nasa.gov/pressrelease/nasa-s-sofia-discovers-water-on-sunlit-surface-of-moon Our World in Data (n.d.). Population, 1000 to 2021 [Graph]. Retrieved December 2021 from https://ourworldindata.org/grapher/population?time=1000..2021 Our World in Data (2018). Global CO2 Atmospheric Concentration [Graph]. Retrieved December 2021 from https://ourworldindata.org/atmospheric-concentrations?country=

© ASCE

699

Earth and Space 2022

Phys.org. 2018. Prehistoric cave art reveals ancient use of complex astronomy. [online] Available at: https://phys.org/news/2018-11-prehistoric-cave-art-reveals-ancient.html [Accessed 30 April 2021]. Redd, N. (2018). Yuri Gagarin: First Man in Space. [online] Space.com. Available at: https://space.com/16159-first-man-in-space.html [Accessed 5 Apr. 2021]. Scott, M. (2014, September 14). What’s the hottest earth has been “lately”? Retrieved December, 2021, from https://www.climate.gov/news-features/climateqa/what%E2%80%99s-hottest-earth-has-been-%E2%80%9Clately%E2%80%9D Sullivan, K. (2019, June 21). How the ISS recycles its air and water [Web log post]. Retrieved December, 2021, from https://www.popsci.com/how-iss-recycles-air-andwater/ Tanaka, K. L., Skinner, J. A., Jr., Dohm, J. M., Irwin, R. P., III, Kolb, E. J., Fortezzo, C. M., Platz, T., Michael, G. G., and Hare, T. M., 2014, Geologic map of Mars: U.S. Geological Survey Scientific Investigations Map 3292, scale 1:20,000,000, pamphlet 43 p., https://dx.doi.org/10.3133/sim3292. The Library of Congress. 2021. Whose Revolution? Copernicus, Brahe & Kepler. [online] Available at: https://www.loc.gov/collections/finding-our-place-in-the-cosmos-with-carlsagan/articles-and-essays/modeling-the-cosmos/whose-revolution-copernicus-brahe-andkepler [Accessed 30 April 2021]. U.S. Green Building Council. (n.d.). U.S. Green Building Council: LEED. Retrieved December, 2021, from https://www.usgbc.org/ U.S. Green Building Council. (2019). LEED v4 for Building Design and Construction. U.S. Green Building Council. Retrieved December, 2021, from https://www.usgbc.org/sites/default/files/LEED%20v4%20BDC_07.25.19_current.pdf UNICEF. (n.d.). Water scarcity. Retrieved December 15, 2021, from https://www.unicef.org/wash/water-scarcity United States EPA. (1991). Indoor Air Facts No. 4 (revised) Sick Building Syndrome (Publication). Environmental Protection Agency. Retrieved December, 2021, from https://www.epa.gov/sites/default/files/2014-08/documents/sick_building_factsheet.pdf

© ASCE

700

701

Earth and Space 2022

A Summary of Technical Requirements, Environmental Factors, and Loading for Lunar Infrastructure Nerma Caluk1 and Atorod Azizinamini2 1

Dept. of Civil and Environmental Engineering, Florida International Univ., Miami, FL. ORCID: https://orcid.org/0000-0002-7134-0593. Email: [email protected] 2 Dept. of Civil and Environmental Engineering, Florida International Univ., Miami, FL. ORCID: https://orcid.org/0000-0001-7627-6757. Email: [email protected] ABSTRACT Humankind has not set foot on the Moon in more than half a century. Since then, our best chance to get back to the Moon and stay, is now, with the Artemis program, led by National Aeronautics and Space Administration (NASA). To achieve a lunar presence for longer than 3 days, the astronauts will need a safe lunar habitat that will protect them from the harsh environment on the lunar surface. When designing for a structure that is quite literally out of this world, there are many different factors to be considered, which are not present on Earth. Humanity went through thousands of years of construction to develop their understanding of structural shape effectiveness, construction material, understanding of the stresses and forces that are present in the Earth environment, and summarized them into various building codes. However, most of that knowledge is not applicable for the lunar environment due to different factors and lack of atmosphere. Various conceptual designs were presented and published around the world, but a more comprehensive study needs to be conducted to account for all the technical requirements and extreme factors affecting the decision of the lunar habitat. This paper briefly summarizes the most important data on the technical requirements and environmental factors to be considered for a successful design and continues with development of a preliminary in-depth and systematic framework, through the Pugh Matrix, to determine an optimal design for a lunar structure. 23 different evaluation criteria were used. Even though numerous different lunar habitat concepts were published throughout last 50 years, they were summarized in eight different concepts. The full decision matrix is presented, with the 3D printed structures as baseline concepts, and Modular Block Structures as the most viable option. INTRODUCTION Humanity’s last visit to the Moon was half a century ago, and it only lasted for a brief period of only about 3 days. There were several attempted programs within NASA to send American astronauts back to the lunar surface since then, but due to many economic constraints, the programs were terminated. However, with the rise of the private space sector, development of commercial rockets, discoveries of significant amounts of water around the Lunar pole areas (Clark, 2009), and plans of the NASA’s Artemis program, our chances of putting humans back on the Moon have increased significantly. To make the trip back to the Moon worthwhile, the need for a long-term base is unavoidable. The lunar base will not only allow the astronauts a safe long-term habitation, but will also provide shelter for manufacturing, exploration, farming and science experiments. To design for a safe, sophisticated, economical, and sustainable habitat is a great challenge for the engineers due to the harsh lunar environment, lack of atmosphere and

© ASCE

Earth and Space 2022

cost of space transportation. Terrestrial construction has been in development for centuries, including understanding of various structural designs, construction material properties and how the stresses are developed in each structural member. The gathered information on terrestrial structural design and analysis has been summarized in various building codes and used in everyday civil engineering practices. Most of those codes and terrestrial assumptions are not applicable for the lunar construction due to different environmental factors, construction material and lack of atmosphere. Furthermore, new construction methods and processes will need to be developed, incorporating robotics and automation, minimizing the human assistance as much as possible. Various lunar habitat design concepts have been developed during and after the Apollo missions, in hopes to send humans to the Moon for a longer period of time and continuing to other planets of our Solar System (Cesaretti et al., 2014; Ganapathi et al., 1993; Jolly et al., 1994; Kaplicky & Nixon, 1985; Malla & Chaudhuri, 2006a; Mottaghi & Benaroya, 2015; Nowak et al., 1992). Some contain more data and analysis than other, but there is still a lack of a more comprehensive study that accounts for all NASA’s technical requirements, environmental and loading factors required for a functional lunar habitat. The absence of a comprehensive lunar base design is due to lack of research and experimental results, which indicate need for various assumptions. Numerous publications mention different ways of categorizing the lunar infrastructure, based on its transportation method, construction material or intended use. With the current technology developed and high-cost of Earth-to-Moon material transportation, scientists have focused on implementing In-Situ Resource Utilization (ISRU), where they considered lunar regolith as the main construction material for protecting a lunar infrastructure from extreme environmental conditions. However, there still exist a lot of assumptions and gaps in our knowledge for the technology that will keep humans safe on the lunar surface for months, with a specified design criteria for the habitats. Furthermore, due to its immense complexity, this type of infrastructure will require intensive communication between various disciplines, including architects, civil, mechanical, geotechnical, and aerospace engineers, material scientists, roboticists, biological scientists, etc. Prior to taking any design and analysis steps for the lunar infrastructure, operational guidelines, technical requirements, and environmental conditions need to be fully defined in order to make a decision on a feasible habitat geometry, its size, functionality and construct ion method. This paper intents to briefly summarize the technical requirements presented in NASA’s Artemis program, together with the environmental and loading factors that will be crucial in the development of any lunar infrastructure and provides a preliminary decision matrix for determining the most optimal lunar habitat design. TECHNICAL REQUIREMENTS Based on the Artemis Plan: NASA’s Lunar Exploration Program Overview (NASA, 2020b), the early stages of the mission include development of the Space Launch System (SLS) rocket and the Orio spacecraft that will bring humans back to the lunar surface by the end of 2028, together with the lunar orbiting station, known as the Gateway. The plan is to establish an Artemis Base Camp with human presence on the lunar South Pole region (area withing 6 o of latitude from the lunar south pole), with help of its international partners. One of the Base Camp elements defined is a lunar Foundation Surface Habitat (FSH), with implementation of in-situ resources and accommodating four crew members for duration of one to two months. The

© ASCE

702

Earth and Space 2022

construction and robotic capabilities were not mentioned in the reports, but we can assume that the goal of building a lunar habitat will be to minimize the human on-site construction as much as possible, either through sending an already-constructed lunar habitat directly from Earth, or with implementation of autonomous robotics and 3D printers to manufacture one on the lunar surface. Establishing a lunar habitat will provide for a long-term exploration of Earth’s nearest neighbor and development of new technologies that are considered as the steppingstone towards exploration of Mars. The exact landing location of Artemis III astronauts is not yet determined, and it depends on several factors. Some of those factors include the change in the sunlight throughout the year, to assure minimal temperature variations, continuous line-of-sight with the Earth for effective commutation, grading of the surface around the landing site, walking and rover mobility, and proximity to water ice resources. One of the potential sites for the Artemis Base Camp is near Shackleton Crater (NASA, 2020b). Furthermore, to better understand the regolith’s properties, NASA has planned to retrieve approximately 35kg of the lunar samples during the planned Artemis III Surface Operations Mission. Besides brining the local soil back to Earth, based on the Artemis III Science Definition Report Goal 6l (NASA, 2020a), astronauts will conduct investigation of lunar regolith as construction material, determining its performance and durability. With these experiments, we will understand the exact behavior of the construction material to be used for any lunar infrastructure, its behavior, including its advantages and disadvantages. LUNAR ENVIRONMENT AND PROTECTION With the basic technical requirement being determined, the next step is to analyze the lunar environment, with all its factors that could potentially affect the design of any lunar infrastructure. A brief understanding and data on lunar environmental factors will be presented here. Lunar environment differs from Earth’s environment in many aspects, including lower gravity, having a value of 1/6th of Earth’s, lack of atmospheric pressure, extreme temperature fluctuations, seismic activity that can last over an hour, lunar dust that sticks to everything and can be very abrasive, three different types of radiation, meteorite impacts and geophysical features (Figure 1). Lunar Gravity The gravitational acceleration on the lunar surface is approximately calculated to be 1.62m/s2, which represented a major difference in design criteria when comparing the lunar and earth environments. Due to the lower gravity, the structural components will possess lower weight, indicating possibilities of structures with longer spans. Engineers will have to reconsider the concept of dead loads and live loads within the lunar environment and might need to implement a mass-based criteria, rather than weight based. Higher ceilings will also be necessary, since operating and walking in low gravity will come with higher reach vertically upwards due to lower weight of the astronauts. Furthermore, structural response under dynamic excitation in lunar gravity is an important behavior that needs to be studied in depth, since gravity plays an important effect in stabilization of different systems (Kalapodis et al., 2019). The effect of dynamic response will need to be considered during different stages of construction, and of the global system after the construction has been complete and the structure is at full operation.

© ASCE

703

Earth and Space 2022

Figure 1. Lunar Environmental Factors to be considered for design of lunar infrastructure Lunar Seismic Activity Dynamic effects will mostly be present during the moonquakes, which can last up to an hour (Figure 2). Prior to the Apollo missions, the Moon was considered seismically nonactive. However, the seismometers placed during the Apollo landings, with their four sensors that included three long-period sensors in the X,Y and Z direction (LPX, LPY, LPZ) that recorded frequencies of ground motion below 2Hz, and one short -period sensor in Z direction (SPZ), which recorded frequencies up to 10Hz (Cooper & Kovach, 1975; Latham et al., 1969; Yosio Nakamura et al., 1975). During the period of 8 years, over 12,000 seismic events were recorded by four seismometers, while more were discovered later with re examining the data (Bulow et al., 2005). There are four types of moonquakes identifies by the Apollo seismometers (Y. Nakamura, 1980; Y. Nakamura et al., 1974; Oberst & Nakamura, 1992), including thermal moonquakes, moonquakes caused due to meteorite impacts, deep moonquakes, and shallow moonquakes. Figure 2 shows an example of typical records of seismic activity by each of the four sensors from the seismometers, presented by Nakamura (Y. Nakamura et al., 1974). Based on the Apollo 15 Preliminary Science Report, it was estimated that the annual seismic energy release on the Moon is between 10 11 and 1015 ergs, while the Earth’s annual seismic energy is 1025 (Y. Nakamura, 1980; NASA, 1972; Nunn et al., 2020). Even though the lunar seismic energy is significantly less than the Earth’s, seismic design, together with microgravity, should be taken into account for additional factor of safety, since humans will incorporate novel construction methods, innovative design and material properties. Potentially, lunar infrastructure will require a specific type of base isolation system to decouple the horizontal motions of the ground from the structural system, however extensive research and ground motion data is required to determine the base isolation requirements and limitations.

© ASCE

704

Earth and Space 2022

Figure 2. One of the recordings from the lunar seismometer from the deep moonquakes, shallow moonquake and meteoroid impact (Y. Nakamura et al., 1974). Extreme Thermal Variations The length of the full lunar day, starting from the beginning of the lunar daylight until the end of the lunar night, is around 29 days, 12 hours, and 44 minutes. During this time, the temperature on the lunar surface can range from 127°C during the lunar daylight to -153°C during the night-time (Williams et al., 2017). This change in temperatures in such a short amount of time will impose thermal stresses on the structural material, leading to thermal fatigue, crack formation and propagation, presenting a real threat to surface infrastructure. Furthermore, the location of the habitat will significantly affect the temperature changes. The goal of the Artemis program is to construct a lunar outpost on the South Pole, where the coldest temperatures are present. The lunar South Pole possesses different diurnal temperatures that are significantly less than on the equatorial sites, with sunlight striking at a low angle. Taking into consideration the location of the sun and the way it radiates, at one moment in time, one side of a lunar habitat will be completely in sunlight, while the other side of the base will be completely in the shade. This leads to different thermal stresses happening simultaneously, indicating the need for a material that can sustain such variable thermal stresses. Utilizing in-situ material will be the most feasible way of providing an effective protection from the extreme variations of lunar temperature. In order to determine an optimum thickness of a regolith shield, engineers need to take into account thermal conductivity of the lunar soil, specific heat, lunar albedo, emissivity of the regolith, among other material properties related to any additives used for the construction. Explaining the thermal data of the lunar regolith is out of scope of this paper. Radiation and Meteorites Lack of atmosphere implies no atmospheric protection from the radiation and meteorites. There are three different types of radiation the astronauts on the lunar surface need to be protected from, including Galactic Cosmic Rays (GCRs), Solar Winds and Solar Flares. Two types of radiation protection could be employed: inactive – use of local material (regolith), or active (plasma shielding, magnetic, electrostatic). Passive radiation protection might seem more

© ASCE

705

Earth and Space 2022

favorable than the active one, due to its abundance on the site and compatibility with other environmental impacts that the astronauts will need to be shieled from, but more thorough research needs to be conducted. Thickness of a potential regolith shield from radiation will be determined based on low-earth orbit (LEO) limits that are recommended for human skin, Blood Forming Organs (BFOs) and ocular lenses. NASA’s Standard 3001 Volume on Crew Health gives several statement on radiation limits (Lewis, 2015). The potential regolith shield can also be used to protect the base from micrometeorite and meteorite impacts. Based on several research studies, the flux of the meteorites is relatively low, while the mean velocity of those particles is determined to be around 15.3 km/s (Cremonese et al., 2013). Even though it would not be feasible to design a structure than can sustain a high meteoroid impact, a certain level of protection must be accommodated for crucial lunar operational system. Average thickness of the protective regolith shell Different values of the protective regolith shield have been mentioned in many different literatures and publications (Aulesa et al., 2000; Cesaretti et al., 2014; Malla et al., 1995) throughout the years. Combining the environmental factors mentioned above, with the in-depth understanding of load distribution and structural geometry, an optimum thickness can be estimated. In the Lunar Sourcebook, it was mentioned that the minimum regolith thickness for a lunar base against radiation is around 2 meters, but for the full protection from extreme flare events, thickness of 3.5 meters will be required (Heiken et al., 1991; Silberberg et al., 1985). However, for the case of thermal insulation, 30cm of regolith was recommended which would minimize the high surface temperature fluctuations (Heiken et al., 1991). Based on Aulesa, regolith thickness of 1.5m is required for sufficient thermal insulation, while 5.4m of regolith was recommended for protection from extreme solar flares, with the assumed density of unconsolidated regolith to be 1300 kg/m 3 (Aulesa, 2000). Presenting the idea of framemembrane lunar base structure, Malla et al have used the regolith shield thickness of 1.5m (Malla & Chaudhuri, 2006b). Cesaretti et al. presented their innovative idea of lunar base construction with the regolith shield depending on the meteorite flux (minimum thickness of 0.8m), radiation and thermal insulation (minimum thickness of 1.5m) (Cesaretti et al., 2014). The values between 1.5m and 3.5m for a lunar regolith shield thickness are the most common ones, but more indepth analyses are required, which should take into account of the maximum human presence on the lunar surface, temperature variations of the South Pole, sunlight angle and factor of safety the engineers will want to take for the meteorite flux probabilities. GEOTECHNICAL PROPERTIES The only direct data on the lunar geotechnical properties was gathered during the Apollo missions, over 50 years ago and this paper contains a summary of the data analyzed, which will be needed for the design of lunar infrastructure. We are only familiar with the first 60 cm of the regolith, defined as the unconsolidated and fragmented layer of rock material which resulted from continuous impact of meteorite and bombardment from the highly charged particles from the Sun and the Galactic Cosmic Rays (Heiken et al., 1991). Furthermore, the material that we brought back is only from the near side of the Moon, while NASA is planning to go to the highlands regions, located at the lunar South Pole. Some of the primary factors that affect the

© ASCE

706

707

Earth and Space 2022

physical and geotechnical properties of the lunar soil, while also directly affecting the foundation design, include: i. ii.

iii. iv.

Specific Gravity (G) of the soil particles, which ranges from 2.3 to >3.2 for lunar soil In-Situ Bulk Density for the top 15cm of the lunar soil is found to be 1.50 ± 0.05 g/cm3, while for the top 60cm of the lunar soil, the value is 1.66 ± 0.05 g/cm3 (Figure 3 and Table 1) (Heiken et al., 1991); Porosity, defined as ratio of the volume of void space between the particles and total volume, is estimated to be around 49 ± 2% at the surface (Table 2) (Heiken et al., 1991). Shear Strength of lunar soil was defined using Mohr-Coulomb equation with the recommended values of lunar soil cohesion and friction angle presented in Table 3 (Heiken et al., 1991).

Figure 3. In-situ bulk density of the lunar soil, as function of depth (Heiken et al., 1991) Table 1. Density of lunar regolith based on depth ranges (Heiken et al., 1991) Density range (cm) 0 – 15 0 – 30 30 – 60 0 – 60

Bulk density, ρ (g/cm3) 1.50 ± 0.05 1.58 ± 0.05 1.74 ± 0.05 1.66 ± 0.05

Relative density, DR (%) 65 ± 3 74 ± 3 92 ± 3 83 ± 3

Table 2. Estimated in-situ porosity of the lunar soil, with its corresponding void ratio as a function of depth (Heiken et al., 1991) Density range (cm) 0 – 15 0 – 30 30 – 60 0 – 60

© ASCE

Average porosity, n (%) 52 ± 2 49 ± 2 44 ± 2 46 ± 2

Average void ratio, e 1.07 ± 0.07 0.96 ± 0.07 0.78 ± 0.07 0.87 0.07

708

Earth and Space 2022

v.

vi. vii. viii.

Bearing Capacity (ultimate and allowable, static, and dynamic) was estimated based on limited experimental data. Figure 4 is from the Lunar Sourcebook (Heiken et al., 1991) and it presents the modulus of subgrade values (k) that might be sufficient for design of a structure. An assumption that soils are normally consolidated is made. Regolith’s modulus of subgrade reaction is calculated to be around 1,000 kN/m2/m Compressibility ranges from 0.3 (loose) to 0.05 (dense) If Factor of Safety is assumed to be 1.5, the calculations from Lunar Sourcebook show that a vertical cut might be possible up to 3m, and a slope of 600 could be maintained to a depth of 10m.

Table 3. Estimated values of cohesion & internal friction angle based on the depth intervals (Heiken et al., 1991) Density range (cm) 0 – 15 0 – 30 30 – 60 0 – 60

Cohesion (kPa) Mean Range 0.52 0.44 – 0.62 0.90 0.74 – 1.10 3.00 2.40 – 3.80 1.60 1.30 – 1.90

Internal friction angle (0) Mean Range 42 41 – 43 46 44 – 47 54 52 – 55 49 48 – 51

Figure 4. Estimations of modulus of subgrade reaction, from calculations of maximum allowable bearing capacity, with data generated using astronauts boot prints (Heiken et al., 1991) LOADING FACTORS The design of the lunar base structural system presents a great challenge for both the architects and civil engineers, since it comes with a large number of unknown parameters and implementation of robotic construction methods. As mentioned before, the goal is to implement autonomous robots for the complete construction process, with minimal human assistance. Unlike Earth, there exist no design codes for any lunar infrastructure. Additionally, any structure to be designed on the Moon will require different design criteria, in comparison with the terrestrial ones. Besides the environmental factors mentioned above, various loading factors and conditions need to be considered during the thought process and design of lunar infrastructure. Those loading types include dead loads, live and vibrational loads, impact loads, temperature and fatigue effects, internal air pressurization and durability, as shown in Figure 5.

© ASCE

Earth and Space 2022

Dead loads are defined as structural loads with constant magnitude over time. The main type of dead loads to be considered in the structural analysis of lunar infrastructure will be their self-weight, and any other attachments that will be permanently attached onto the structure. If a structure is to be built on the lunar surface with implementation of robotics and automation, the load path of the structure will need to be carefully monitored throughout the whole construction process, since that part of the process is considered most demanding and challenging when it comes to terrestrial construction. The live and vibrational loads are considered any machinery loads induced during the construction, together with occasional secondary vibrations due to the landing and takeoff of the rockets. In addition to the artificial vibrational loads, the natural ones caused by the moonquakes will need to be taken into consideration. When determining structural properties for any vibrational analysis, investigation of known boundary conditions which are dependent on the regolith-to- foundation properties and the regolith-to-surface interaction must be accounted for.

Figure 5. Lunar LoÇ ading Factors to be considered for design of lunar habitats Two types of impact loads can be can distinguished: artificial (due to landing, take-off, EVAs) and natural (meteorites and micrometeorites). To design a proper protection from these types of loads, engineers will have to look at how does the protective shield, potentially regolithbased, transmit stresses based on the impact and wave velocity, indicating a need for an in-depth understanding of the shield material properties. As mentioned before, the lunar environment possesses periodic and rapid t emperature gradients, while also being dependent on the position and angle of the sunlight, indicting a need for understanding the fatigue effects that will be presented in the structure that would either be brough from the Earth or built on the surface of the Moon. To effectively determine and analyze the fatigue effects and their limits, engineers will need to investigate the loading cycles on the infrastructure, estimating its design life and any options for maintenance and retrofit. Once all the technical requirements, environmental and loading factors have been set in place, the most feasible alternative should be proposed, and the design can be started. The design

© ASCE

709

Earth and Space 2022

procedure of a lunar infrastructure should be split into two steps. In the first step, t he conceptual/functional design is considered, which represents a preliminary design of a lunar habitat. In this step, the type of structure and its construction material will be determined, together with determining the rough dimensions that will satisfy the intended use. The second step is the design specification phase, during which selection and arrangement of structural components is done, while defining their exact dimensions. During this step the connections and foundation design should be provided, while also accounting for service loads. TRADE STUDY A thorough trade study is required to establish an effective evaluation methodology and determine the most feasible lunar habitat design that incorporates various design and importance factors. The process of trade study for lunar habitats is not an easy one (Benaroya & Ettouney, 1992; Bodiford et al., 2005; Drake & Richter, 1992; Higgins & Benaroya, 2020; Ruess et al., 2006), due to multiple complexities that come with the lunar environment, while also lacking knowledge on various topics and understanding capabilities of the current technology, indicating need for assumptions and high factors of safety. For this study, 45 different habitats concepts were identified, and grouped in 8 different categories: prefabricated structures (Adam et al., 2021; Grandl, 2007; Litaker et al., 2013), deployable structure (Dronadula & Benaroya, 2021; Gruber et al., 2007; Lak & Asefi, 2022), pneumatic structures (Langlais & Saulnier, 2000; Sadeh et al., 2000), cable structures (Ettouney et al., 1992; Zhang et al., 2021), mobile structures (RaisRohani, 2005), natural structures (lava tubes) (Horz et al., 1985; Theinat et al., 2020), modular blocks structures (Faierson et al., 2010; Romo et al., 2018; Zhou et al., 2019) and 3D printed structures.(Cesaretti et al., 2014; Fateri et al., 2019; Melodie et al., 2021) All known advantages and disadvantages of each design concept were elaborated, but that data is out of the scope of this paper. The evaluation method used in this study is the Pugh Matrix. This decision-making process is initiated by defining the evaluation criteria that will be related to all the proposed solutions. Implementing trade studies conducted several years ago (Bodiford et al., 2005; Drake & Richter, 1992), and adding other criteria which have been related with recent technology development and research findings, 23 of criteria were identified, that were grouped in 8 different sections: 1) Ease of Earth-to-Moon Transportation a. Total Weight b. Size Limitation c. Redeployabiliy 2) Ease of Lunar Construction a. Degree of Development Difficulty b. Erectability c. Ease of Soil Placement 3) In-Situ Resource Utilization (ISRU): a. ISRU Implementation for Construction 4) Operations/Growth/Reusability a. Connection of Modules/Components b. Variation in Infrastructure Use 5) Architectural/Psychological a. Psychological Effects on Humans © ASCE

710

Earth and Space 2022

b. Space Utilization Efficiency c. Internal Rearrangement Ease 6) Environmental Impact a. Dust Mitigation b. Temperature and Insulation Effects c. Radiation Resistance d. Meteorite Resistance 7) Structural Integrity a. Component Stability b. Failure Modes/Safety c. Differential Pressure Stability d. Fatigue and Durability Effects 8) Maintenance & Inspectability a. Ease of inspectability b. Ease of maintenance Then, the importance factor for each criterion is defined, taking in opinion of several engineers. The values of importance factors go from 1 (the least important) to 5 (the most important). Afterwards, the different alternatives were presented, with the baseline concept being the 3D printed structures. For each criterion, each alternative was rated with +1 (better), -1 (worse) or 0 (same), which was then multiplied with its corresponding importance factors and total weighted value was determined. The preliminary results of this trade study can be found in Table 4. Even though thorough analysis and understanding of existing concepts was done, there still exist a lot of assumptions present within this study, indicating need for further understanding and development of the decision-making process. CONCLUSION With the current development of the Artemis mission and growth of the private space sector, humanity is getting back on the right track to visit the Moon once again, and this time to stay. To provide the astronauts with a safe haven on the lunar surface, engineers will need to design a structure that will protect them from the harsh lunar environment, including vacuum conditions, radiation, seismic activity, extreme temperatures, and micrometeorites. To achieve this goal and minimize the space transportation cost, in-situ resource utilization needs to be implemented, together with automated robotic construction that will minimize any assistance from the astronauts. Even though many research papers have been published on the matter of lunar construction and design, with a variety of conceptual ideas and analyses, we are in a need of a comprehensive study that incorporates NASA’s technical requirements, operational guidelines and environmental factors applied onto a geometrically defined lunar habitat. However, factors that are determining the geometric design are dependent on each other. This means that determining the shape of any lunar infrastructure will be both dependent on the limitations of construction methods and environmental conditions. If we can design a lunar habitat that is able to sustain all the environmental factors, and keep the crew safe and healthy for months, but the construction methods are limited in their capabilities, the design must be iterated to accommodate for easy and autonomous construction. This paper is intended to present a summary of those factors, together with the preliminary comprehensive trade study of 8 different lunar habitat concepts and 23 evaluation criteria. Pugh © ASCE

711

Earth and Space 2022

Matrix was selected as the decision-making tool to identify the most acceptable lunar habitation option, with the knowledge that we have available so far. Structures made from modular section blocks were selected as the winning design, presenting superior applicability in other lunar infrastructure designs (berms, landing pads, wall, equipment shelters, etc.), while also having easier way of maintenance and inspectability. Further research will be conducted at FIU in designing and analyzing a lunar habitat comprised of section blocks, together with other lunar infrastructure designs. Table 4. Preliminary trade study of 8 different lunar habitat designs, incorporating Pugh Matrix as the decision-making method.

REFERENCES Adam, A., Stefan, A., Emily, A., Gautier, B. de F., Karl, B., Gianmarco, C., Monica, E., Alan, G., Adina, G., Damiana, I., Leonard, L. C. L., Giorgio, L., Shay, M., Joao, M., Aiden, O., Orin, O., Laura, P., Charlotte, P., Alexander, S., … Alberto, Z. (2021). Solutions for Construction of a Lunar Base: A Proposal to Use the SpaceX Starship as a Permanent Habitat. 72nd International Astronautical Congress (IAC). Aulesa, V. (2000). Architecture of Lunar Habitats. 4th International Conference on the Exploration and Utilisation of the Moon, 462(July), 289–292. https://ui.adsabs.harvard.edu/abs/2000ESASP.462..289A/abstract

© ASCE

712

Earth and Space 2022

Aulesa, V., Ruiz, F., & Casanova, I. (2000). Structural requirements for the construction of shelters on planetary surfaces. Space 2000, 403–409. https://doi.org/10.1061/ 40479(204)49 Benaroya, H., & Ettouney, M. (1992). Framework for Evaluation of Lunar Base Structural Concepts. Journal of Aerospace Engineering, 5(2), 187–198. https://doi.org/10.1061/ (asce)0893-1321(1992)5:2(187) Bodiford, M. P., Fiske, M. R., McGregory, W., Pope, R. D., Fiske, M. R., McGregory, W., & Pope, R. D. (2005). In Situ Resource-Based Lunar and Martian Habitat Structures Development at NASA/MSFC. Bulow, R. C., Johnson, C. L., & Shearer, P. M. (2005). New events discovered in the Apollo lunar seismic data. Journal of Geophysical Research: Planets, 110(E10), 1–22. https://doi.org/10.1029/2005JE002414 Cesaretti, G., Dini, E., De Kestelier, X., Colla, V., & Pambaguian, L. (2014). Building components for an outpost on the Lunar soil by means of a novel 3D printing technology. Acta Astronautica, 93, 430–450. https://doi.org/10.1016/j.actaastro.2013.07.034 Clark, R. N. (2009). Detection of adsorbed water and hydroxyl on the moon. Science, 326(5952), 562–564. https://doi.org/10.1126/science.1178105 Cooper, M. R., & Kovach, R. L. (1975). Energy, frequency, and distance of moonquakes at the Apollo 17 site. 6th Lunar Science Conference; 17-21 March 1975; Houston, TX, 3, 2863–2879. https://ui.adsabs.harvard.edu/abs/1975LPSC....6.2863C/abstract Cremonese, G., Borin, P., Lucchetti, A., Marzari, F., & Bruno, M. (2013). Micrometeoroids flux on the Moon. Astronomy and Astrophysics, 551, A27. https://doi.org/10.1051/00046361/201220541 Drake, R. M., & Richter, P. J. (1992). Concept Evaluation Methodology for Extraterrestrial Habitats. Journal of Aerospace Engineering, 5(3), 282–296. https://doi.org/10.1061/(asce)0893-1321(1992)5:3(282) Dronadula, R., & Benaroya, H. (2021). Hybrid lunar inflatable structure. Acta Astronautica, 179, 42–55. https://doi.org/10.1016/j.actaastro.2020.10.039 Ettouney, M., Benaroya, H., & Agassi, N. (1992). Cable Structures and Lunar Environment. Journal of Aerospace Engineering, 5(3), 297–310. https://doi.org/10.1061/(asce)08931321(1992)5:3(297) Faierson, E. J., Logan, K. V., Stewart, B. K., & Hunt, M. P. (2010). Demonstration of concept for fabrication of lunar physical assets utilizing lunar regolith simulant and a geothermite reaction. Acta Astronautica, 67(1–2), 38–45. https://doi.org/10.1016/ j.actaastro.2009.12.006 Fateri, M., Meurisse, A., Sperl, M., Urbina, D., Madakashira, H. K., Govindaraj, S., Gancet, J., Imhof, B., Hoheneder, W., Waclavicek, R., Preisinger, C., Podreka, E., Mohamed, M. P., & Weiss, P. (2019). Solar Sintering for Lunar Additive Manufacturing. Journal of Aerospace Engineering, 32(6), 04019101. https://doi.org/10.1061/(asce)as.19435525.0001093 Ganapathi, G. B., Ferrall, J., Seshan, P. K., Ferrall, J., & Seshan, P. K. (1993). Lunar base habitat designs: Characterizing the environment, and selecting habitat designs for future trade-offs. June. Grandl, W. (2007). Lunar Base 2015 Stage 1. Acta Astronautica, 60(4-7 SPEC. ISS.), 554– 560. https://doi.org/10.1016/j.actaastro.2006.09.031

© ASCE

713

Earth and Space 2022

Gruber, P., Häuplik, S., Imhof, B., Özdemir, K., Waclavicek, R., & Perino, M. A. (2007). Deployable structures for a human lunar base. Acta Astronautica, 61(1–6), 484–495. https://doi.org/10.1016/j.actaastro.2007.01.055 Heiken, G. H., Vaniman, D. T., & French, B. M. (1991). Lunar Sourcebook - A User’s Guide to the Moon. Cambridge University Press. Higgins, M., & Benaroya, H. (2020). Utilizing the Analytical Hierarchy Process to determine the optimal lunar habitat configuration. Acta Astronautica, 173, 145–154. https://doi.org/10.1016/j.actaastro.2020.04.012 Horz, F., (1985). Lava tubes - Potential shelters for habitats. Lbsa, 405–412. https://ui.adsabs.harvard.edu/abs/1985lbsa.conf..405H/abstract Jolly, S. D., Happel, J., & Sture, S. (1994). Design and Construction of Shielded Lunar Outpost. Journal of Aerospace Engineering, 7(4), 417–434. https://doi.org/10.1061/ (asce)0893-1321(1994)7:4(417) Kalapodis, N., Kampas, G., & Ktenidou, O. J. (2019). Revisiting the fundamental structural dynamic systems: the effect of low gravity. Archive of Applied Mechanics, 89(9), 1861– 1884. https://doi.org/10.1007/S00419-019-01548-7/TABLES/1 Kaplicky, J., & Nixon, D. (1985). A surface-assembled superstructure envelope system to support regolith mass-shielding for an initial-operational-capability lunar base. Lunar Bases and Space Activities of the 21st Century, 375. https://ui.adsabs.harvard.edu/ abs/1985lbsa.conf..375K/abstract Lak, A., & Asefi, M. (2022). A new deployable pantographic lunar habitat. Acta Astronautica, 192, 351–367. https://doi.org/10.1016/j.actaastro.2021.12.049 Langlais, D. M., & Saulnier, D. P. (2000). Reusable, presurized dome for lunar construction. Space 2000, 791–797. https://doi.org/10.1061/40479(204)95 Latham, G., Ewing, M., Press, F., & Sutton, G. (1969). The Apollo passive seismic experiment. Science, 165(3890), 241–250. https://doi.org/10.1126/ SCIENCE.165.3890.241/ASSET/45902147-1A03-4326-9ED7E9348E43D8B1/ASSETS/SCIENCE.165.3890.241.FP.PNG Lewis, R. (2015). NASA Standard 3001. http://www.nasa.gov/hhp/standards Litaker, H. L., Archer, R. D., Szabo, R., Twyford, E. S., Conlee, C. S., & Howard, R. L. (2013). Human habitation field study of the Habitat Demonstration Unit (HDU). Acta Astronautica, 90(2), 391–405. https://doi.org/10.1016/j.actaastro.2012.04.018 Malla, R. B., Adib-Jahromi, H. R., & Accorsi, M. L. (1995). Simplified Design Method for Braced Double-Skinned Structure in Lunar Application. Journal of Aerospace Engineering, 8(4), 189–195. https://doi.org/10.1061/(asce)0893-1321(1995)8:4(189) Malla, R. B., & Chaudhuri, D. (2006a). Analysis of a 3D Frame — Membrane Structure for Lunar Base. Earth and Space 2006 - Proceedings of the 10th Biennial International Conference on Engineering, Construction, and Operations in Challenging Environments, 2006, 1–8. https://doi.org/10.1061/40830(188)61 Malla, R. B., & Chaudhuri, D. (2006b). Analysis of a 3D frame - Membrane structure for lunar base. Earth and Space 2006 - Proceedings of the 10th Biennial International Conference on Engineering, Construction, and Operations in Challenging Environments, 2006, 61. https://doi.org/10.1061/40830(188)61 Melodie, Y., Morris, M., Pailes-Friedman, R., Elshanshoury, W., Esfandabadi, Gomez, D., Guzeev, A., Netti, V., & Rajkumar, A. (2021). Project Olympus: Off-World Additive

© ASCE

714

Earth and Space 2022

Construction for Lunar Surface Infrastructure. 50th International Conference on Environmental Systems, 19. Mottaghi, S., & Benaroya, H. (2015). Design of a Lunar Surface Structure. I: Design Configuration and Thermal Analysis. Journal of Aerospace Engineering, 28(1), 04014052. https://doi.org/10.1061/(asce)as.1943-5525.0000382 Nakamura, Y. (1980). Shallow moonquakes: how they compare with earthquakes. Geochimica et Cosmochimica Acta, Supplement, 14(3), 1847–1853. https://ui.adsabs.harvard.edu/abs/1980LPSC...11.1847N/abstract Nakamura, Y., Dorman, J., Duennebier, F., Ewing, M., Lammlein, D., Latham, G., Dorman, J., Duennebier, F., Ewing, M., Lammlein, D., & Latham, G. (1974). High-frequency lunar teleseismic events. Nakamura, Yosio, Dorman, J., Duennebier, F., Lammlein, D., & Latham, G. (1975). Shal low lunar structure determined from the passive seismic experiment. The Moon, 13(1–3), 57– 66. https://doi.org/10.1007/BF00567507 NASA. (1972). Apollo 15: Preliminary Science Report. NASA SP-289. Apollo 15: Preliminary Science Report, 289, 546. http://adsabs.harvard.edu/abs/1972NASSP.289...... NASA. (2020a). Artemis III Science Definition Report. https://doi.org/NASA/SP20205009602 NASA. (2020b). NASA’s Lunar Exploration Program Overview. In Nasa (Issue September). https://www.nasa.gov/sites/default/files/atoms/files/artemis_plan-20200921.pdf Nowak, P. S., Sadeh, W. Z., & Morroni, L. A. (1992). Geometric Modeling of Inflatable Structures for Lunar Base. Journal of Aerospace Engineering, 5(3), 311–322. https://doi.org/10.1061/(asce)0893-1321(1992)5:3(311) Nunn, C., Garcia, R. F., Nakamura, Y., Marusiak, A. G., Kawamura, T., Sun, D., Margerin, L., Weber, R., Drilleau, M., Wieczorek, M. A., Khan, A., Rivoldini, A., Lognonné, P., & Zhu, P. (2020). Lunar Seismology: A Data and Instrumentation Review. Space Science Reviews, 216(5), 1–39. https://doi.org/10.1007/S11214-020-00709-3/FIGURES/14 Oberst, J., & Nakamura, Y. (1992). A seismic risk for the lunar base. In NASA. Johnson Space Center, 231. https://ui.adsabs.harvard.edu/abs/1992lbsa.conf..231O/abstract Rais-Rohani, M. (2005). On structural design of a mobile lunar habitat with multi -layered environmental shielding. NASA Contractor Report, NASA Marshall Space Flight Center, April 2005, 1–5. http://scholar.google.com/scholar?hl=en&btnG=Search&q=intitle: On+structural+design+of+a+mobile+lunar+habitat+with+multilayered+environmental+shielding#0 Romo, R., Andersen, C., Defore, K., Zacny, K., Thangavelu, M., & Lippitt, T. (2018). Planetary lego: Designing a construction block from a regolith derived feedstock for in situ robotic manufacturing. Earth and Space 2018: Engineering for Extreme Environments - Proceedings of the 16th Biennial International Conference on Engineering, Science, Construction, and Operations in Challenging Environments , 289– 296. https://doi.org/10.1061/9780784481899.029 Ruess, F., Schaenzlin, J., & Benaroya, H. (2006). Structural Design of a Lunar Habitat. Journal of Aerospace Engineering, 19(3), 133–157. https://doi.org/10.1061/(ASCE)08931321(2006)19:3(133) Sadeh, E., Sadeh, Z. W., Criswell, M., Rice, E. E., & Abarbanel, J. (2000). Inflatable Habitats of Lunar Base Development. ESASP, 462, 301. https://ui.adsabs.harvard.edu/ abs/2000ESASP.462..301S/abstract

© ASCE

715

Earth and Space 2022

Silberberg, R., Tsao, C. H., Adams, J. H. J., & Letaw, J. R. (1985). Radiation transport of cosmic ray nuclei in lunar material and radiation doses. http://inis.iaea.org/Search/ search.aspx?orig_q=RN:17071190 Theinat, A. K., Modiriasari, A., Bobet, A., Melosh, H. J., Dyke, S. J., Ramirez, J., Maghareh, A., & Gomez, D. (2020). Lunar lava tubes: Morphology to structural stability. Icarus, 338. https://doi.org/10.1016/j.icarus.2019.113442 Williams, J.-P., Paige, D. A., Greenhagen, B. T., Sefton-Nash, E., Williams, J.-P., Paige, D. A., Greenhagen, B. T., & Sefton-Nash, E. (2017). The global surface temperatures of the Moon as measured by the Diviner Lunar Radiometer Experiment. Icar, 283, 300–325. https://doi.org/10.1016/J.ICARUS.2016.08.012 Zhang, D., Zhou, D., Zhang, G., Shao, G., & Li, L. (2021). 3D printing lunar architecture with a novel cable-driven printer. Acta Astronautica, 189, 671–678. https://doi.org/ 10.1016/j.actaastro.2021.09.034 Zhou, C., Chen, R., Xu, J., Ding, L., Luo, H., Fan, J., Chen, E. J., Cai, L., & Tang, B. (2019). In-situ construction method for lunar habitation: Chinese Super Mason. Automation in Construction, 104, 66–79. https://doi.org/10.1016/J.AUTCON.2019.03.024

© ASCE

716

Earth and Space 2022

717

Stone, Brick and Concrete Masonry on Mars Dr. Peter Carrato1, Keith Kennedy2, Cory Brugger3, Jennifer Heldman4, Darlene Lim5 1

Bechtel Fellow Emeritus, Fellow ASCE, Bechtel Global, Reston, VA, [email protected] Chief Architect, Bechtel Global, Reston, VA, [email protected] 3 Chief Technology Officer, HKS, [email protected] 4 NASA, Ames Research Center, [email protected] 5 NASA, Ames Research Center, [email protected] 2

ABSTRACT Sedimentary rock and clay mineral deposits are known to exist on Mars. Substantial amounts of water ice in the form of polar caps, glacial deposits, and subsurface ice are also a recognized natural resource. Dry stacked stone construction, stones placed without the use of mortar, using sedimentary rocks is the oldest form of durable building used by humans. Mud bricks, composed of clay and water, used by early civilizations in the deserts of Earth, created monumental and iconic structures, some of which have endured for more than a thousand years. Concrete a mixture of sand, gravel, and a binding agent can also be produced from Martian materials. Masonry construction, the use of stones, bricks, and concrete will create the first generations of truly Martian buildings, which will endure for centuries. Construction techniques associated with heavy masonry construction on Mars are described. INTRODUCTION The most fundamental need for survival is shelter. For humans this is followed closely by food, water and sanitation. On Mars, early exploration will probably utilize shelters imported from Earth. Long-term facilities, that are constructed as part of a large-scale urban master plan1, must take advantage of in situ resources. This will minimize the mass of imported building material and ensure economically viable development. For illustration purposes a masonry structure that would be part of an early construction phase for a permanent settlement will be considered. Figure 1 shows a concept of a Mars settlement 2 that will take many decade to build. This project will develop from the bottom up, with early installation of subsurface and surface structures. Figure 2 shows one possible detail of the access corridor from the launch pads to the inhabitted settlement. This access corridor would be well suited to the application of masonry construction. A portion of this structure might begin its useful life as a shelter used for robot storage and maintenance, and become a portion of a launch pad access corridor as the settlement grows.

© ASCE

718

Earth and Space 2022

Figure 1: Mars Settlement Concept (ref 2)

Figure 2: Launch Pad Access Corridor To get a rough order of magnitude estimate of the amount of masonry needed to construct one access corridor the overall dimensions are assumed to be; length 700 meters, width 25 meters, depth below grade 10 meters. The access ways are assumed to take up approximately 75% of the cross-sectional area of the corridor. A total volume of masonry needed for one access corridor is 43750 cubic meters. When planning for facilities that will form part of a permanent presence on Mars, masonry construction should consider projects that use tens of thousands of cubic meters of material. MASONRY ATTRIBUTE FOR MARTIAN STRUCTURES Masonry construction provides many inherent benefits for survival in the Martian environment that are not provided by other structural systems. For example, a masonry wall by virtue of its material properties, will provide both; a structural system to resist external loads such as wind and gravity, and a substantial amount of mass that insulates the building’s inhabitants from extreme variations in outside temperature and provides radiation shielding. A comparable structure constructed using metal framing such as steel or aluminum would require a secondary subsystem © ASCE

719

Earth and Space 2022

of thermal insulation and radiation shielding to match the same overall building performance. Non-structural attributes of masonry that are common to stone, brick or concrete construction include; • durability • impact resistance • thermal mass • acoustic damping • radiation shielding • fireproofing Not all aspects of masonry construction are beneficial for Marian construction. The most significant draw back to be considered is the relatively low tensile strength of masonry which leads to cracking. These structural systems will require some form of liner or sealer to provide a pressure tight structure. An epoxy coating is suitable for this purpose. Terrestrially, epoxies have been used for decades in critical sealing application including containment structures for Canadian nuclear power plants3. Masonry construction relies on in situ resource utilization (ISRU) for most of the material employed. The structural systems considered may require the use of water extracted on the surface, binders produce from local resource, and/or additives imported from off the planet. This is especially true of brick and concrete construction. Factors that will determine the most efficient masonry system to choose are presented in the following table. Table 1: Masonry Construction Considerations ATTRIBUTE Material Availability Energy Utilization Extraction Manufacture/process Installation

STONE

BRICK

CONCRETE Low Impact Moderate Impact High Impact

Construction Considerations Limited geometry Special tools Temporary support Sensitive to means/methods

Table 1 provides a high-level qualitative assessment of various attributes of masonry that will affect the selection of a specific structural system. The general categories considered include; material availability, energy utilization, and construction considerations. These attributes are discussed for each material type in the following sections. STONE Stacking stones is the oldest form of durable building construction on Earth. Terrestrial examples are readily found that have endured for thousands of years. The most recognizable of these are the great pyramids at Giza, which required highly advanced technology and a labor-intensive construction organization to quarry, transport and place the stones. A more rudimentary example are the stone walls found in Ireland shown in Figure 3.

© ASCE

720

Earth and Space 2022

Figure 3: Dry Stacked Stone Wall Aran Island. Dun Aonghasa 1000 BC (ref 4) The dry stacked (constructed without the use of mortar) walls of Aran Island could readily be replicated on Mars Material Availability: Sedimentary rocks suitable for building construction are available on Mars as seen in Figure 4. It is not uncommon to transport stones great distances for special structures. Forty-ton megaliths used to construct Stonehenge were moved nearly 30 kilometers using means and methods available in 3000 BC5. However, it is most advantageous for a project to have a quarriable deposit of sedimentary rocks close by. Because of the significant mass associated with all forms of masonry construction, the energy associated with transporting raw materials is a common consideration for these structural systems.

Figure 4: Natural Stone on Mars (ref 6, 7) Energy Utilization: The masonry material that requires the least amount of energy to extract is natural stone. For certain sedimentary rock deposits material extraction is either low energy quarrying or in some cases simply picking up the rocks off the ground. Most energy utilization will be in loading and hauling to the building site. For some deposits of stones, their natural shapes may be such that they may be placed without any energy needed for processing. The deposit shown on the right side of Figure 4 falls into this category. For large or irregularly shaped stone, such as those shown on the left side of Figure 4, some shaping of the stones using chisels may be required before placing. This simple process for shaping is relatively low in energy utilization.

© ASCE

Earth and Space 2022

721

Initial stone construction will be dry stacked, using no mortar for joining. This requires a minimal use of energy, essentially the potential energy needed to lift the stones into place. Construction Considerations: Dry stacked stone structures utilize two general categories of stones; as quarried (as found) material or carefully shaped or “dressed” (ashlar – dressed square or rectangular). The simplest form of stone construction uses as-found stones. This type of construction has been used for walls for centuries as shown in Figure 3. To create an enclosed structure, one with walls and a roof a construction technique called corbelling is used. Rows of stones are slightly off-set from the wall inward. As row upon row are built-up a simple vaulted roof is formed. This type of construction has been used since Neolithic times and is still used today in Italy’s Itria Valley of Apulia. In this region, a traditional house called a trullo is built of stone without mortar. Figure 5 shows cross-section of a trullo, note how the inner layer of stones are corbelled inward to form a dome roof. Trullo-like structures can readily serve as non-pressurized shelters for robot storage and maintenance, and to house mechanical and electrical equipment.

Figure 5: Trullo Construction, Foti et al 2017 (ref 8) Note that the structures shown in Figures 3 and 5 were built by simply stacking one stone on top of another. The stones used are essentially as extracted from the quarry. To facilitate robotic construction, the three-dimensional geometry of each stone should be scanned. Given a final outline of a wall or dome, mapping software could be developed to create an optimal construction placing sequence of individual quarried stones to their final in-place location and orientation. Such a software tool would greatly facilitate robotic construction. More complex structural forms may be produced by the use of temporary support structures (false-work) and dressed stones (stone that have been shaped to produce precise geometry). False-work maybe used to produce a variety of stone arches, vaults, and domes, that can be combined to create complex and monumental structures. Tools for dressing and quarrying include low technology devices such as chisels and stone saws. Figure 6 shows some examples of vaults and false-work.

© ASCE

Earth and Space 2022

722

Figure 6: False-work and Vaults (ref 9) Dry stacked stone structures benefit from readily available material, and requires little energy for extraction and construction. Construction using this structural system does not require sophisticated equipment and is not inhibited by prevailing environmental conditions i.e. no need for temporary enclosures. There are limited structural forms for dry stacked stones; vaults and domes, and open spans are expected to be less than those from more sophisticated masonry systems such as brick and concrete. MUD BRICKS Although mud bricks are a very low technology building material, they are more complex and more energy intensive to produce than natural stone. Many monumental and iconic structures have been constructed of locally produced bricks10. One of the most impressive facilities built of mud bricks is the Jameh Mosque of Yazd. This facility encloses an area of more than 10000 square meters comprising the main court and supporting facilites11.

Figure 7. Jameh Mosque at Yazd 1324 -1365 (ref 11) Another interesting brick structure, that is more than one thousand years old, is the Ctesiphon Arch, which is 37 meters tall and 26 meters across. It is worth noting that this arch was built without the aid of false-work. A similar structure might be used as a service building for space ships. The figure below shows a SpaceX Starship superimposed on a photograph of the arch.

© ASCE

723

Earth and Space 2022

Figure 8: Ctesiphon Arch, (Tāq Kasrā) Persia 540 AD Material Availability: Clay deposits suitable for manufacturing strong durable bricks may be found on Mars12. These deposits include clay minerals that are similar to those used for centuries in Africa and the Middle East13. Terrestrial mud brick usually includes some fibrous material such as that found in animal dung. On Mars chopped basaltic fibers will add to the mechanical properties of the locally manufactured bricks. Unlike natural stone, mud bricks are manufactured building components. They require the combination of two materials, clay and water, both of which are available on Mars. Consideration must be given to the proximity of clay deposits to a source of water when planning this type of construction. Energy Utilization: The energy required to support brick construction is greater than that needed for stacked stone methods. In addition to the need for two different materials, and their associated transportation requirements, bricks are manufactured components that require energy to produce. The energy required for processing the “mud” prior to brick construction includes that associated with: • proper proportioning the component material, either a mass or volumetric process • mixing the clay and water • placing and consolidating the mud into molds • curing the bricks, typically using elevated temperature • removing the bricks from their molds, this may be done before curing • stacking and storing the completed bricks prior to delivery to the job site This last bullet point should be emphasized. First raw material must be collected and delivered to a manufacturing facility, then the final bricks can be delivered to their point of application. It is also important to note that brick making must be done in a controlled environment such as a pressure-tight, temperature-controlled enclosure. Construction Considerations: One of the most advance robotic construction methods on Earth is placing brick and block masonry. FBR (fast brick robot) Ltd, an Australian company, manufactures the Hadrian X brick laying robot (shown in Figure 9) which can place as many as

© ASCE

Earth and Space 2022

724

240 blocks per hour14. Similar equipment modified for Martian environmental condition will be needed for efficient brick construction. The list of equipment needed for brick construction is much longer than that needed to build using stacked stones, including devices for proportioning, mixing, curing, as well as additional handling. Since this equipment must be imported the impact on landed mass to be delivered to Mars will influence the over all construction planning process.

Figure 9. Hadrian X Placing blocks for commercial building Mud bricks may be place using the same techniques as those for stacked stones. The uniformity of bricks when compared to as quarried stones allows for more precise placement and a generally cleaner look to the final structure. However, typical brick construction utilizes mortar to bind them together. In terrestrial structures a mixture of gypsum, sand was used by ancient Egyptians as mortar15. This type of binding agent would be challenged by the Martian environment. Issues with extreme cold temperature and low atmospheric pressure will probably make this form of mortar impractical unless building is done inside of temporary construction enclosure. Developing a Mars friendly masonry mortar would make this construction method more efficient. Although there are challenges associated with brick construction on Mars, this structural system allows the building designer more sophisticated and esthetically pleasing architectural forms than stacked stones. Numerous structural details such as pendentives, squinches and muqarnas are more readily created using bricks than stones16, 17. Mud brick construction benefits from readily available material, however if mortar is specified for this building system a locally produced mixture needs to be developed. Energy utilization with this structural system, over and above that needed for material extraction, is that associated with manufacturing the bricks and mortar. Construction using mud bricks would benefit from purposebuilt robots. and may be inhibited by prevailing environmental conditions depending on the properties of the mortar being used. Structural forms for brick expand on those used with stack stones, and open spans are expected to be greater. CONCRETE Production of concrete on Mars has been well studied18, 19. The NASA Centennial Challenges program sponsored a three-phase, four-year-long, competition focused on 3D printing of concrete habitats on Mars. This competition addressed materials used in concrete mixtures, means and

© ASCE

Earth and Space 2022

methods of construction, and architectural design. Figure 10 shows the winning entry from the last phase of the competition. This structure is a one-third scale prototype of the proposed habitat constructed using a basaltic concrete with a polymer binder. The American Concrete Institute defines concrete as a “mixture of hydraulic cement, aggregates, and water, with or without admixtures, fibers, or other cementitious materials21.” Extraterrestrial use of material that requires hydraulic cement and water is not necessarily practical or efficient. A modified definition of concrete is more appropriate to Martian application. Merriam-Webster defines concrete as “a mass formed by concretion or coalescence of separate particles of matter in one body22.” Building on Webster’s definition, for Mars concrete is Figure 10. AI Space Factory, considered to be a mixture of fine aggregate (sand) and coarse Basaltic Concrete, Reduced aggregate (gravel) cemented together using a binding agent. Scale Mars Habitat (ref 20) Material Availability: Concrete is primarily composed of the aggregate used to make up the mixture. By volume, aggregate will typically represent 75% or more of the material placed in the final structure. This makes concrete an attractive building material when compared to stacked stones or brick. Rather than relying on locating and developing sedimentary rock quarries, or deposits of suitable clay minerals, material suitable as concrete aggregate is ubiquitously available on the surface of Mars. Fine aggregate (sand) for use in concrete should have a particle sizes distribution primarily from 0.60 mm to 2.36 mm23. Coarse aggregate (gravel) includes particle sizes from 2.36 mm up to 100 mm. The range of sizes used in the design of the final mixture will depend on a number of parameters including flowability and workability of the fresh concrete, as well as heat generated during curing, configuration of structural elements, etc. Suitable deposits of coarse and fine aggregate may be found close to most construction sites. In some cases locally available regolith may need to be processed by crushing, sieving, sorting and blending to produce the needed material. In any case concrete aggregate is readily available on Mars. Although coarse and fine aggregate are natural resources that are relatively easy to extract, material that makes up the binder that cements the aggregate into concrete is a much greater challenge. A few examples of potential binders include Portland cement, geopolymers, molten sulfur, and high density polyethylene (HDPE). All these materials were investigated by teams that competed in the Mars 3D Printed Habitat Challenge24. Although other binding agents are possible, the material availability for these four will be discussed. Portland cement is the traditional binder used in terrestrial concrete. It requires the use of water to initiate the chemical reaction that causes the concrete to gel. The ability to produce this cement on Mars was demonstrated by the team from CTL (Construction Testing Laboratory) Group during Phase 2 of the Mars habitat challenge. Portland cement is made up of four main compounds: tricalcium silicate, dicalcium silicate, tricalcium aluminate, and a tetra-calcium

© ASCE

725

Earth and Space 2022

aluminoferrite. These materials must be appropriately proportions, blended and then clinkered (fused under high heat). The clinkers are then finely ground to produce the cement. Portland cement is often blended with supplementary cementitious materials (SCM) to either utilize industrial waste products and/or enhance the properties of concrete such as workability or durability. The two most common SCMs are fly ash or ground blast furnace slag. Both these materials can have significant cementitious properties. In fact some of the earlies terrestrial concrete relied on fly ash deposited by volcanic eruptions, called pozzolan, as their binding agent25. Production of Portland cement on Mars will require a significant effort to find and refined the appropriate ingredients. Use of volcanic deposits as a source of cementitious pozzolanic material is an intriguing possibility. Mars is home to many volcanoes and a geologic survey of their ejecta many identify valuable building material. Geopolymer concrete does not use pure water as part of the chemical process need to cause the mixture to gel. SCMs such as fly ash and ground blast furnace slag are combined with coarse and fine aggregate and then the mixture is “activated” by an alkali such as a solution of sodium hydroxide26. Use of molten sulfur as a binding agent in extraterrestrial concrete was first proposed during the Apollo era27. Mars is a sulfur-rich planet28. Much of the sulfur is bound up in sulfates and would need to be extracted to a purer elemental form prior to use in making concrete. The process of concrete making is then relatively simple, melt the sulfur and blend it with aggregate and allow the mixture to cool. Using HDPE as a binder is similar to using sulfur; melt it, mix it, cool it. However, HDPE is not a naturally occurring substance on Mars. It can be produced, like any long chain hydrocarbon, from materials found on the planet. For example, the atmosphere is primarily carbon dioxide and thus a ready source of carbon, and hydrogen is available from the hydrolysis of water. Energy Utilization: Of the locally produced construction material concrete, when compared to brick and stones, is the most energy intensive option. Making concrete on Mars requires quarrying and processing both coarse and fine aggregate. The real challenge for concrete manufacturing is producing the binder needed to cement the aggregate into concrete. Locating recoverable deposits of volcanic pozzolans or high-grade sulfur deposits appears to be the least energy intensive options of the four considered. This implies that the two most promising binders to consider are molten sulfur and geopolymers. Lacking ground-truth information on availability and recoverability of these materials it is impossible to make a reasonable estimate of the energy required for producing suitable quantities of concrete. Construction Considerations: Concrete is one of the most widely used building materials on Earth. There are many mature construction techniques that are optimized for a variety of different types and configurations of structures, and the available local sources of material and equipment. As discussed, concrete mix formulation is not well defined and thus the means and methods of construction cannot be forecast with much certainty. This makes it impractical to develop a list of functional requirements for concrete related equipment as was provided for brick construction. It

© ASCE

726

Earth and Space 2022

is expected that a final equipment list for concrete manufacturing, mixing, and placing will be significantly more complex than that for brick and mass of imported equipment will also be greater. Regardless of the complexity and energy demand required to building using concrete this material provides opportunities to enhance structural design that are not available to brick and stone. The two most salient advantages are adaptability to a wide range of final building geometries, and the potential for including reinforcing elements such as steel or basalt bars, or wires. Reinforced concrete construction techniques for use on Mars are yet to be developed. However, this technology will no doubt be employed as permanent facilities mature. The adaptability of concrete to many different structural forms has been demonstrated in NASA’s 3D-Printed Habitat Challenge. Figure 11 show an extremely complex geometry, note the tapering rib of the domed shape, that was placed without the use of false work or shoring to support the overhanging section. This example used a HDPE based basaltic aggregate concrete mixture.

Figure 11. 3D Habitat Final Structure by Branch Technology (ref 20) Concrete construction benefits from readily available coarse and fine aggregate. However suitable binders need to be developed and the availability of local material to produce them evaluated. Energy utilization with this structural system, over and above that needed for material extraction, is difficult to estimate until further mix designs and placing methods are more fully developed. Construction using concrete will require purpose-built robots and will be inhibited by prevailing environmental conditions. Temporary construction enclosures will be required. Structural forms associated with concrete are not nearly as restrictive as those for brick and stone. With the potential for including tensile reinforcing in concrete almost unlimited building configurations are possible.

CONCLUSION Masonry construction provides an energy efficient means of building durable, functional, and esthetically pleasing Martian facilities. Tens of thousands of cubic meters of material will be needed to deliver the projects required to support a permanent presence on the planet. The first phases of construction should consider stack stones as the material of choice. When compared to other masonry systems, brick and concrete, use of locally quarried stones will require: • Fewest and simplest pieces of equipment (robots) thus lower mass landed on Mars • The least amount of material processing thus lowest energy utilization • The simplest construction operations. Stone vaults and domes will provide facilities that last for centuries. © ASCE

727

Earth and Space 2022

REFERENCES 1. Hudgins, E., Pitfalls of Remote, Extreme Settlements: The Case for Urban Planning Practice for Future Space Colonies, ASCE Earth and Space 2021. 2. https://techcrunch.com/2017/09/28/everything-spacex-revealed-about-its-updated-plan-toreach-mars-by-2022/ 3. https://inis.iaea.org/collection/NCLCollectionStore/_Public/29/030/29030513.pdf 4. https://singersongblog.me/2017/04/15/the-stone-walls-of-aran-a-triumph-of-adaptability/ 5. https://en.wikipedia.org/wiki/Stonehenge#Function_and_construction 6. https://geology.com/stories/13/rocks-on-mars/ 7. https://www.nasa.gov/jpl/msl/pia19074 8. Foti, A., A. Fraddosio, N. Lepore, M. Piccioni, On the Mechanics of Corbelled Domes: New Analytical and Computational Approaches, Research on Engineering Structures and Materials, January 2017. 9. https://blog.stephens.edu/arh101glossary/ 10. Byron, Robert, The Road to Oxiana, John Lehmann, 1937. 11. Mahdavinejad and Javanroodi, Analysis the Physical Proportions of Main Courts in Azeri Style Mosques, Journal of Advanced Social Research Vol.2 No.6, August 2012, 269-279. 12. Bristow et al, Clay mineral diversity and abundance in sedimentary rocks of Gale crater, Mars, Science Advances, 2018;4: eaar3330 6 June 2018. 13. Serry et all, Egyptian Smectite-rich clays for lightweight and heavy clay products, Periodico di Mineralogia (2015), 84, 2, 351-371. 14. https://www.fbr.com.au/ 15. https://www.newworldencyclopedia.org/entry/mortar_(masonry) 16. Carrillo, A., The Sasanian Tradition in ʽAbbāsid Art: Squinch Fragmentation as The Structural Origin of the Muqarnas, IEM, Jan-Jun 2016/ISSN 1676-5818. 17. Sumini, V., M. M. Esfandabadi, J. Paradiso, and G. Trotti, Structural Insight of Persian Bathhouse Architecture for Designing Greenhouses on Mars, ASCE Earth and Space 2021. 18. Mueller et al, NASA Centennial Challenge: Three Dimensional (3D) Printed Habitat, Phase 3, IAC-19-E5.1.8, 70th International Astronautical Congress (IAC), Washington, D.C., 21-25 October 2019. 19. Carrato, P., A. Ellis, T. Kim, Additive Manufacturing for Martian Habitats, 40th IABSE Symposium, 19-21 September 2018, Nantes, France. 20. Carrato, P., personal photograph. 21. ACI CT-20, ACI Concrete Terminology, American Concrete Institute, May 2020. 22. https://www.merriam-webster.com/dictionary/concrete 23. ASTM C33/C33 M-18, Standard Specification for Concrete Aggregate, American Society for Testing and Materials, 2018. 24. Phase 2 Report, NASA Centennial Challenges Program 3D-Printed Habitat Challenge, Phase 2 Structural Member Competition, February 9, 2018. 25. https://www.britannica.com/technology/pozzolana 26. https://www.the-possible.com/geopolymer-concrete-carbon-neutral-alternative-to-cement/ 27. ACI SP-125, Lunar Concrete, American Concrete Institute, May 1991. 28. Franz, H.B., P. King, F. Gaillard, Volatiles in the Martian Crust, Chapter 6 – Sulfur on Mars from the Atmosphere to the Core, Elsevier, 2019.

© ASCE

728

Earth and Space 2022

Structural and Durability Properties of MgO-Al2SiO3 Concrete for ISRU Martian Construction Milap Dhakal1; Allan N. Scott2; Rajesh P. Dhakal3; Christopher Oze4; and Don Clucas5 1

Dept. of Civil and Natural Resource Engineering, Univ. of Canterbury, Christchurch, New Zealand. Email: [email protected] 2 Dept. of Civil and Natural Resource Engineering, Univ. of Canterbury, Christchurch, New Zealand. Email: [email protected] 3 Dept. of Civil and Natural Resource Engineering, Univ. of Canterbury, Christchurch, New Zealand. Email: [email protected] 4 Geology Dept., Occidental College, Los Angeles, CA. Email: [email protected] 5 Dept. of Mechanical Engineering, Univ. of Canterbury, Christchurch, New Zealand. Email: [email protected] ABSTRACT The development and properties of MgO-clay-based engineered M-S-H cement mortar and concrete with Martian regolith simulant filler is presented in this study. Engineering properties for the M-S-H cement include compressive strength, tensile strength, Poisson’s ratio, elastic modulus, bond strength, and incorporate durability indicators such as porosity, permeability, and resistivity. Additionally, the structural response of reinforced engineered M-S-H concrete columns under lateral loads is compared with the performance of columns cast with conventional portland cement (PC) concrete. These findings demonstrate that the properties of the MgO-clay cement concrete can be effectively manipulated through variations, either in the cement formulations or in the mix proportions, to suit the functional requirements of proposed Martian applications. Engineering properties and the structural performance of the reinforced concrete sections made with the engineered M-S-H concrete were similar to those demonstrated by the columns made with PC concrete, thereby, making it a suitable material for both Martian and Earth-based construction. INTRODUCTION The pursuit of a permanent human presence on Mars will likely follow a sequential infrastructure development and can be expected to be divided into multiple phases of colonization (Arnhof 2016). The early stage of Mars exploration will comprise a limited number of missions and personnel. These missions are expected to be characterized by dedicated cargo spaces for prefabricated habitats and life support systems. The missions primary objectives will be the investigation of performances of various ISRU techniques developed for the production of life support amenities and construction. However, at the later stages, the isolated complexes will likely be expanded and interconnected through structures constructed using ISRU. At this stage, fully operational regolith processing capabilities will be established on Mars to produce indigenous raw materials, thus rendering the Martian colony more sustainable and at least semiindependent from Earth-based resources. The ability to supply power, water and a viable enclosed atmosphere are fundamental to ongoing development.

© ASCE

729

Earth and Space 2022

Though Mars has been regarded as the most habitable planet in the solar system after Earth, the temporal variation, lower atmospheric pressure, and micro-gravity of Mars still presents a set of significant challenges for construction and colonization. Most of the efforts in the area of Martian construction have been concentrated on the identification and development of ISRUbased construction technology, which helps in gaining independence from the Earth-based construction material. The use of ISRU-based-construction will lead to reduced IMLEO and maximize overall chance of success of the mission. As such concrete has been proposed as a convenient construction material for establishing a permanent human colony on Mars. One benefit of this approach is that majority of the concrete volume is provided by inert fillers or aggregate, a function served by local regolith which requires minimum processing interventions. Additionally, concrete offers advantages such as ease of preparation, element formations, and adaptability for mechanized construction (Sanjayan and Nematollahi 2019). Based on the elemental composition of Martian minerals, cement, which binds the Martian regolith particles, such as Portland cement, sulfur, geopolymers, plastic polymers has been proposed to develop Martian concrete (Naser 2019). However, some of the proposed cements may present greater challenges for production in Martian conditions. For instance, Portland cement, which on Earth is manufactured from the widely available limestone deposits around the planet, may not be available in the required quantities of cement production on Mars (Ehlmann and Edwards 2014). Similarly, sulfur, which may be an abundant element in the Martian regolith and can demonstrate reasonable mechanical properties under normal conditions, seems to be vulnerable against alternating freezing and thawing conditions. In this regards, based on the regional availability of minerals containing MgO along with and SiO2-rich clay minerals, a magnesium-silicate-hydrate (M-S-H) cement has also been proposed for Martian concrete (Dhakal et al. 2021b; Scott and Oze 2018). The presence of large deposits of magnesium-rich minerals such as olivine, magnesite, and clay on the Martian surface makes magnesium-based cementitious binder system a potential option for ISRU-based Martian construction. Magnesium oxide (MgO) can be produced from olivine containing minerals, or in some local areas magnesite, that are distributed across much of the planet (Scott and Oze 2018). One potential extraction process, as suggested by Scott et al. 2021, makes use of a closed system in which recycled acid is used to digest olivine, followed by a process of silica and iron recovery and finally electrolysis to produce Mg(OH) 2. The Mg(OH)2 can easily be converted to MgO. The inputs to the process are water, olivine containing minerals and electricity. Aluminosilicate, derived from clay minerals (Rashad 2013) could provide another valuable material for Martian construction. The reactive MgO and aluminosilicate thus produced can be used together in a hydration reaction to produce two cementing phases, namely M-S-H and hydrotalcite. This binder formulation develops compressive strengths above 60 MPa (Shah and Scott 2021a). This particular MgO-cement formulation, due to the addition of hydrotalcite as a secondary binding phase, has been reported to outperform the MgO-SiO2 cement formulation in both mechanical and durability properties. In a previous study, using a locally sourced Martian regolith simulant and MgO-clay-based cementitious binder, the authors reported compressive strengths as high as 40 MPa (Dhakal et al. 2021b). The majority of past studies investigating the mechanical properties of the proposed M-S-H cement have been associated with the use mortars, with very little information available on the detailed properties of M-S-H in concrete systems. Here, we report recent findings on the multiple mechanical properties of the MgO-clay-based concrete, such as tensile strength and bond

© ASCE

730

731

Earth and Space 2022

strength, along with its durability properties. Additionally, we include the load-displacement behavior of reinforced MgO-clay cement concrete column under lateral loading and compare it with that of the PC concrete column. Overall, M-S-H cements provide a means for construction on Mars that can be compared to PC concrete properties, workability, and applications. MATERIALS AND METHODS Regolith simulant selected as concrete aggregates was composed of trachybasaltic rock dust (nominal size 6.5mm) and crushed stone (nominal size 18 mm) sourced from the East Otago region of New Zealand (i.e., the same stimulant used in the previous study re ported by the authors (Dhakal et al. 2021b)). The primary binder ingredients for the M-S-H cement were the reactive or light-burnt magnesia (MgO), obtained from Australia, and high purity kaolinite clay (Metakaolin, Mk) purchased from Scotts Chemicals, New Zealand. The chemical compositions of these ingredients are shown in Table 1. General Purpose (GP) cement, provided by Golden Bay Cement, New Zealand, was used for the PC concrete and hydrated magnesium aluminosilicate powder, Acti-gel®, for thixotropy and segregation prevention. Industrial grade limestone powder and magnesite powder were used as fillers for PC and MS-H concrete, respectively. Admixtures used during the concrete preparation were superplasticizer (SP), SIKA ® Viscocrete 5-555®, and air-entraining admixture (AEA), BASF ® Masterair® AE200. Binder Formulation/Mix Design Ensuring the development of sufficient strength at a workable consistency in MgO-Mk mortars and concrete requires a high dose of superplasticizer, which renders rheopectic consistencies to the mortar and concrete (Dhakal et al. 2021a). Therefore, self-compacting consistency was adopted for both the control and M-S-H concretes. A second M-S-H cementitious formulation, with 30% quartz powder replacement of the metakaolin, was used to study the effects of impure clay on the concrete properties. The target mix design for the concrete mixes used in the study is shown in Table 2. For convenience, the concrete with MgO-Mk cement was designated as MSH concrete, while that with MgO-Mk-Qz was named MSH-Qz. Table 1. Oxide composition (wt.%) of the cementing materials

© ASCE

MgO

Mk

PC

CaCO3

MgCO3

Actigel®

SiO2

4.55

51.75

21.54

1.24

4.54

55.2

Al2O3

0.24

44.9

3.81

0.4

0.79

12.2

Fe2O3

0.34

0.48

2.69

0.17

0.5

4.05

CaO

3.09

0.13

64.51

54.87

3.46

1.98

MgO

79.22

0.1

0.94

0.49

41.59

8.56

Na2O

6.0

MSH-f31-A0.05

2.6

rebar

>15.0

>5.8

Column Test ID PC-f27-A0.2

© ASCE

739

Earth and Space 2022

Durability

7-days

28-days

91-days

Porosity (%)

15 10 5 0

7-days

Resistivity (kW.cm)

20

PC

MSH

MSH-Qz

(a) Permeable porosity

28-days

91-days

100

Permeability Coeff. (k, 10-10 m/s)

Concrete durability is associated with the ingress of environmental agents into it, and it is a function of the nature of available capillary pores and their interconnectivity. While the permeable porosity quantifies the amounts of pores within the concrete, the other two indices, electrical resistivity and oxygen permeability index (OPI), provide a qualitative indication of the pore connectivity and tortuosity. Cement hydration continued to progress with time, and the reaction products resulted in a decrease in permeable porosity of the concrete, as shown in Figure 9a. Until 28-days, the porosities of the concretes MSH and MSH-Qz were not significantly different than that of the control concrete, despite the control concrete mix using higher mixing water. However, at 91 days, the porosities of MSH and MSH-Qz were lower than that of the control concrete. This observation provides the microstructural validation of the slower hydration of the MgO-Al2O3SiO2 cement system, which also explains the increase of the concrete strengths in the M-S-Htype concrete during this period. The electrical resistivity measurements made in this study were conducted on the concrete samples saturated with tap water and did not use NaCl. Therefore, the difference in the electrical resistivity observed in the PC and MSH concretes is partly due to the nature of the pore solutions of the cement systems in addition to the microstructure (Figure 9 b). However, with age, as was revealed by the porosity results, with progress in the cement hydration, the increased electrical resistance in all the concrete suggests increased pore tortuosity. The addition of Qz filler in the MgO-Mk cement tends to enhance the durability properties of the concrete. The observed permeability coefficients imply that the MgO-Mk-based concrete has better pore distribution and tortuosity, which provides superior durability properties to this concrete than control PC concrete.

2

7-days

28-days

91-days

1.5

10

1

0.5

1

PC

MSH

MSH-Qz

0

(b) Electrical resistivity

PC

MSH

MSH-Qz

(c) Oxygen permeability

Figure 9. Durability indices of concrete Concrete and mortars developed with a cementitious formulation based on MgO and Al2SiO3, are feasible to be produced on an industrial scale utilizing indigenous Martian minerals and can be expected to have a perform similar to conventional PC -based systems. The lower atmospheric pressure and reduced gravity on Mars however can be expected to influence the performance of the proposed concrete (Naser 2019). The near-vacuum atmosphere for instance, will induce higher porosity while causing accelerated moisture loss

© ASCE

Earth and Space 2022

due to rapid degasification of the mixing water from the cement matrix, leading to decreased concrete strength (Khoshnevis et al. 2017) for any cast concrete directly exposed to the atmosphere. Therefore, the production and application of the proposed concrete for Martian applications may be dependent on options to mitigate the effects of Martian conditions, such as curing membranes or precast elements in a climate controlled facility. Future studies are required to quantify the effects of such conditions on the performance of the concrete and to establish suitable construction techniques for the proposed concrete. The use of steel reinforcing in this investigation is meant for comparison purposes to demonstrate the effectiveness of the concrete under structural loading conditions. Alternative reinforcing such as basalt fiber sourced from Martian materials may be a more appropriate option for efficient ISRU construction. CONCLUSION Several conclusions can be derived from the results of the different mechanical and durability properties of the concrete derived using MgO-clay cement system. The shear-thickening flow of the proposed concrete with a high dose of superplasticizer is suitable for a self-compacting consistency. At its fresh state, it was possible to fill into molds of different shapes and sizes, including the ones with closely spaced reinforcement bars. The self-compacting consistency facilitated the ease of filling and finishing. The slow nature of hydration of the MgO-clay cement system resulted in gradual tensile strength gain. The rate of compressive strength gain in the proposed MgO-Mk concrete is similar to that observed in the PC concrete. The concrete-rebar bond performance of the proposed MgO-Mk concrete is similar to that in the PC concrete. The codal provisions used to ensure adequate rebar embedment in RC structures made with conventional concrete can be safely adopted for those made with the proposed concrete. The load-displacement behavior under lateral loading of the reinforced M-S-H concrete was comparable to that of the PC concrete columns. The tendency of the proposed concrete to undergo rapid deterioration will demand greater sectional area of the section. Durability indicators of the proposed concrete suggest its microstructure should perform, at least, like that of the PC concrete to resist the ingress of the environmental agencies within it. Material needed to create M-S-H concrete, as assessed here, can all be sourced from deposits accessible at the Martian surface. This provides one future pathway for construction on Mars working with a material similar to PC concrete. ACKNOWLEDGEMENT The authors acknowledge the financial support received from the Quake Core, NZ, and University of Canterbury Doctoral Scholarship along with the technical support during the planning and testing phase of the research provided by Prof. Timothy Sullivan, Mr. Tim Perigo, Mr. Mosese Fifita, Mr. Alan Thirlwell, and the Structural Engineering Laboratory team. © ASCE

740

Earth and Space 2022

REFERENCES Alexander, M., Y. Ballim, and J. M. Mackechnie. 2009. “Durability Index Testing Procedure Manual.” Res. Monogr. No. 4, 29. University of Cape Town. De Almeida Filho, F. M., M. K. El Debs, and A. L. H. C. El Debs. 2008. “Bond-slip behavior of self-compacting concrete and vibrated concrete using pull-out and beam tests.” Mater. Struct. Constr., 41 (6): 1073–1089. https://doi.org/10.1617/s11527-007-9307-0. ASTM C1437-07. 2009. “Standard Test Method for Flow of Hydraulic Cement Mortar.” ASTM Int. ASTM C1611/C1611M-18. 2018. “Standard Test Method for Slump Flow of SelfConsolidating Concrete.” ASTM Int. West Conshohocken, PA, USA: ASTM Internationals. https://doi.org/10.1520/C1611. ASTM C496/496M-17. 2017. “Standard Test Method for Splitting Tensile Strength of Cylindrical Concrete Specimens.” ASTM Int. West Conshohocken, PA, USA: ASTM Internationals. https://doi.org/10.1520/C0496_C0496M-17. ASTMC469/469M. 2014. “Standard Test Method for Static Modulus of Elasticity and Poisson’s Ratio of Concrete in Compression.” ASTM Int. West Conshohocken, PA, USA: ASTM Internationals. https://doi.org/10.1520/C0469_C0469M-14. Dhakal, M., A. N. Scott, V. Shah, R. P. Dhakal, and D. Clucas. 2021a. “Development of a MgO-metakaolin binder system.” Constr. Build. Mater., 284 (May 2021): 122736. https://doi.org/10.1016/j.conbuildmat.2021.122736. Dhakal, M., A. Scott, V. Shah, C. Oze, R. Dhakal, D. Clucas, M. W. Hughes, and R. P. Mueller. 2021b. “Magnesia-Metakaolin Regolith Mortar for Martian Construction.” Earth Sp. 2021, 808–817. Reston, VA: American Society of Civil Engineers. Ehlmann, B. L., and C. S. Edwards. 2014. “Mineralogy of the Martian Surface.” Annu. Rev. Earth Planet. Sci., 42 (1): 291–315. https://doi.org/10.1146/annurev-earth-060313055024. Khoshnevis, B., A. Carlson, and M. Thangavelu. 2017. ISRU-Based Robotic Construction Technology for Lunar and Martian Infrastructures. NASA Tech. Rep. Naser, M. Z. 2019. “Extraterrestrial construction materials.” Prog. Mater. Sci., 105 (June): 100577. Elsevier. https://doi.org/10.1016/j.pmatsci.2019.100577. NZS 3101. 2006. “Concrete structures standard.” Stand. New Zeal. New Zealand: Standards New Zealand. Park, R. 1989. “Evaluation of ductility of structures and structural assemblages from laboratory testing.” Bull. New Zeal. Soc. Earthq. Eng., 22 (3): 155–166. https://doi.org/https://doi.org/10.5459/bnzsee.22.3.155-166. Park, R., and T. Paulay. 1974. Reinforced Concrete Structures. Christchurch, New Zealand: John Wiley & Sons. RILEM-FIP-CEB. 1973. “Armatures de béton armé et de béton precontraint.” Matériaux Constr., 6 (5): 319–375. https://doi.org/10.1007/BF02473623. Sanjayan, J. G., and B. Nematollahi. 2019. 3D Concrete Printing for Construction Applications. 3D Concr. Print. Technol. Elsevier Inc. Scott, A. Oze, C. Shah, V. Yang, N. Shanks, B. Cheeseman, C. Marshall, A. Watson, M. 2021. Transformation of abundant magnesium silicate minerals for enhanced CO2 sequestration, Communications Earth and Environment, 2 (25).

© ASCE

741

Earth and Space 2022

Scott, A. N., and C. Oze. 2018. “Constructing Mars: Concrete and Energy Production from Serpentinization Products.” Earth Sp. Sci., 5 (8): 364–370. https://doi.org/10.1029/ 2017EA000353. Shah, V., and A. Scott. 2021a. “Use of kaolinite clays in development of a low carbon MgO clay binder system.” Cem. Concr. Res., 144 (March): 106422. Elsevier Ltd. https://doi.org/10.1016/j.cemconres.2021.106422. Shah, V., and A. Scott. 2021b. “Hydration and microstructural characteristics of MgO in the presence of metakaolin and silica fume.” Cem. Concr. Compos., 121 (April): 104068. Elsevier Ltd. https://doi.org/10.1016/j.cemconcomp.2021.104068. Soretz, S. 1972. “A comparison of beam tests and pull-out tests.” Matériaux Constr., 5 (4): 261–264. https://doi.org/10.1007/BF02474074.

© ASCE

742

743

Earth and Space 2022

Lunar Demandite—You Gotta Make This Using Nothing but That A. Ellery1 1

Dept. of Mechanical and Aerospace Engineering, Carleton Univ., Ottawa, ON. Email: [email protected] ABSTRACT We consider whether an entire spacecraft can be constructed from lunar resources by consideration of its materials availability. We first consider what functional materials would be required to construct all the major spacecraft subsystems. We then examined each lunar mineral to determine what metals, ceramics, and glasses can be extracted. We suggest that lunar resources—with certain provisos—can indeed supply all the functional materials required to construct a spacecraft. We suggest that this could constitute a first generation ISRU capability and that a subsequent generation to extract KREEP materials would offer enhanced performance but no significantly greater capacities. INTRODUCTION In-situ resource utilisation makes possible the construction of a lunar infrastructure (Steurer 1982). We have developed an inventory of functional materials available on the Moon and scoped their applications to determine the degree of self-sufficiency attainable without reliance on an Earth-based supply chain. If we take a long-term view, the decisions made today will impose legacy effects on future lunar settlers. We must eliminate the Earth-based supply chain as early as possible to prevent the foreseeable political strife of imbalanced opportunity for selfdetermination for which near-self-sufficiency is a prerequisite. The construction of a lunar infrastructure such as an evolving Moon Village (even perhaps Lunar University) will require a range of products – lunar base complexes, closed ecological life support systems including lunar agriculture, robotic and human-rated rovers, etc – with consumables support – energy, water, oxygen and other recyclable CNPKS, etc - and machines to implement such constructions – mining machines, unit (electro/thermo)chemical processors, manufacturing and assembly machines. Such a Moon Village must be as self-sufficient and self-maintainable as is feasible without reliance on the extensive Earth-based infrastructure that has evolved to serve Earth. THE DEMANDITE We begin with an hypothetical but typical robotic spacecraft comprised of all the major subsystems as a core requirement for construction from lunar resources on the observation that all space systems are variations on a spacecraft. This provides us with our demandite concept. These are the approximate breakdowns for an orbital space vehicle – if we are manufacturing spacecraft on the Moon, this is a reasonable starting point though a solar power satellite’s design would be dominated by its energy chain. Similarly, propulsion from the lunar surface includes launch – we ignore this aspect here other than to observe that launch to the Gateway requires an LH2/LOX mass fraction of close to 2.5 including tankage. If we are concerned with lunar bases (Ellery 2021), this table must be interpreted carefully – human-rated spacecraft include extensive

© ASCE

744

Earth and Space 2022

environmental control and life support systems which we exclude from our dry mass allocation. Terrestrial materials consumption is dominated by bulk civil engineering structures. We might expect lunar bases to utilise regolith-derived material for compressive structures but these are primarily conceived as outer shells for radiation /micrometeoroid protection and thermal insulation. It seems reasonable that inner structure would include metal frames and panels. Thermal control on spacecraft is primarily passive involving redistribution between short hot sunlight and cold eclipse cycles - on the Moon, the two-week sun/eclipse periods suggest that thermal control will be more challenging. A lunar base on the other hand is a static structure unlike a spacecraft so it requires no GNC/ACS or propulsion. Common features to both lunar bases and space vehicles that are expected to be approximately similar associated with avionics - the power system (though a lunar base has fewer restrictions on deployed size), wiring harness, communications and onboard computing systems. We have adopted a standard 25% for tensile structure for a spacecraft assuming that additional compressive structure required for a lunar base would constitute a significant additional mass. Thermal insulation (nighttime) and thermal radiators (daytime) have been allocated 3% each – for lunar bases, outer regolith protection offers thermal insulation during both day and night. Thermal straps re-distribute heat internally. The wiring harness – dominated by electrical power distribution - common to spacecraft and bases comprises electrical wiring and some electrical insulation (glass fibre) and allocated 8%. Given that avionics constitutes an increasing fraction of spacecraft and bases or both computing, control and communications, it is allocated 12%. Mechanical actuation and sensors are allocated 5% each because both spacecraft and lunar bases require orientable mechanisms and sensory networks – in the latter case, this includes rover/manipulator support. Most spacecraft payloads are based on imaging systems so we allocate 11% for optical systems – in the case of lunar bases, this is dominated by mirrors/lenses for thermal ISRU processing. Hard structures and high thermal tolerance material are also specialized requirements for ISRU processing allocated 3% and 4% respectively. Hence, ISRU thermal processing systems constitute 18% of the lunar base mass. Power generation and storage are allocated 20% for both spacecraft and lunar base. Table 1. Typical subsystem mass allocation to 100% spacecraft mass (adapted from Brown C (2002) Elements of Spacecraft Design, AIAA Education Series) Subsystem

Structure/Mechanisms Thermal Control Attitude Control Power System Wiring Harness Propulsion Communications CDHS Payload

Subsystem mass allocation of on-orbit dry mass 26-29% 3% 9-10% 19-21% 7-8% 13-15% 6-7% 6-7% 11%

We examine the Moon’s inventory, aware that the Moon is deficient in certain elements, most notably Cu, Na and halogens. The Moon exhibits a range of resources from volatiles to regolith to minerals from which useful ceramics, metals and possibly plastics may be extracted (Appendix). Although volatiles are relatively scarce, careful husbanding of carbon in small

© ASCE

745

Earth and Space 2022

quantities may provide a valuable resource. This defines gaps to realising the construction of this spacecraft from lunar resources but there are material substitutions that can be made or strategies implemented to minimise the use of deficient materials, e.g. recycling Earth-imported reagents so that they are not consumed. A key facet is in the exploitation of multi-functionality of materials. There are two related issues that must be relegated to future studies and are not addressed here: (i) the influence of minor natural “contaminants” in minerals which on reduction will yield contaminant metals in the metal product; (ii) the omission of deliberately added alloy constituents in product alloys due to their lack on the Moon. Most metal alloys comprise a majority metal with minor additives such as weldalite, an Al-Li alloy (90% of highland anorthosite rock - aluminium enrichment is up to 33% by mass. Anorthite is highly soluble to HCl acid leaching at low temperature yielding amorphous silica and alumina with high recovery efficiencies via AlCl 3.6H2O. It is a two-part process. HCl leaching of anorthite as HCl gas is carried out at 160oC followed by precipitation and crystallization of AlCl3.6H2O. Silica is removed during this stage by filtration. Removal of Cl and H2O is achieved by heating at 400 oC which recycles HCl. Roasting AlCl 3 at a maximum of 900oC for 90 minutes in oxygen yields alumina. Lunar orthoclase may similar be treated with HCl acid to yield silica and kaolinite. Kaolin – rock bearing kaolinite – is the basis of porcelain. On Earth, alumina is often extracted from kaolin using HCl leaching (Pak et al 2019) adding orthoclase as another source of silica and alumina. Both alumina and silica are refractory ceramics with wide utility especially in thermal control applications – silica for thermal insulation and alumina for reactor crucible lining. Lime (CaO) can be reduced to the metal Ca that is more than twice as electrically conductive as Cu – in the vacuum of the Moon, it will not oxidise. Ca is malleable and therefore offers a source of electrically conducting wiring. Alumina can also be reduced to aluminium metal. The Washington monument completed in 1885 was capped with aluminium as an expression of extravagance - aluminium was an extremely precious metal until the development of the Hall-Heroult electrolytic process in the late 1880s. The low melting point of Al and cryolite permitted the input of liquid solution of aluminum oxide in cryolite and liquid metal output. For the Moon, the FFC electrochemical process can reduce both aluminium and silicon independently to high metallurgical purities >99% (Ellery et al 2021). It is worth noting that aluminium extracted from bauxite requires 227-342 MJ/kg while steel manufactured from iron ore requires 40-75 MJ/kg (Raabe et al 2019). The FFC process requires considerably less energy for metal reduction including aluminium as it operates in the solid state. As well as its desirable qualities as a high specific strength structural material and electrical conductor,

© ASCE

Earth and Space 2022

Al powder may be burned in oxygen as a solid fuel or fed as a fluid fuel by a powder pump or auger but it has poorer performance than hydrogen/oxygen. Silicon has traditionally been earmarked for solar cell construction but several disadvantages of PV solar cells that render its manufacture and use marginal (Ellery 2021). It has other uses including as a as a metal alloy additive (such as silumin with 87% Al and 13% Si). Silicon carbide powder is a versatile material that may be pressure-pressed at >1400 bar and sintered at around 2000oC but carbon in lunar volatiles is relatively scarce. Silicon’s primary use is in its oxide form, silica as a source of fused silica glass. LUNAR ILMENITE Lunar ilmenite has often been touted as a source of oxygen through hydrogen (or less commonly carbon) reduction but here we adopt it as a source of iron and titanium: FeTiO3 + H2 → Fe + TiO2 + H2O. The water is electrolysed to yield oxygen and recycled hydrogen. Although chemical reduction requires temperatures of 900-1000oC, a temperature of 1600 oC yields molten iron (liquation) which can be tapped off. Iron offers the basis of many different functional alloys (steels) and as a constituent of permanent magnets as metal or ferrite. Steel structures are typically formed from I-beams which impart high stiffness relative to their crosssection area. Silicon steel (up to 3% Si) has dramatically increased electrical resistivity for electromagnet and motor cores. We have sourced the silicon additive but further additives are required for other iron alloys Kovar (53.5% Fe, 29% Ni, 17% Co, 0.3% Mn, 0.2% Si and 99% pure titanium (Ellery et al 2021). Titanium has very high tensile strength and toughness, superior strength-to-weight ratio than steel and aluminium, and excellent corrosion resistance at high temperature up to 400 oC. The most common alloy comprises 6% Al and 4% V (Ti6Al4V) and