Advances in Mechanical Engineering: Select Proceedings of ICAME 2020 [1st ed.] 9789811536380, 9789811536397

Citation preview

Lecture Notes in Mechanical Engineering

Vilas R. Kalamkar Katarina Monkova   Editors

Advances in Mechanical Engineering Select Proceedings of ICAME 2020

Lecture Notes in Mechanical Engineering Series Editors Fakher Chaari, National School of Engineers, University of Sfax, Sfax, Tunisia Mohamed Haddar, National School of Engineers of Sfax (ENIS), Sfax, Tunisia Young W. Kwon, Department of Manufacturing Engineering and Aerospace Engineering, Graduate School of Engineering and Applied Science, Monterey, CA, USA Francesco Gherardini, Dipartimento Di Ingegneria, Edificio 25, Università Di Modena E Reggio Emilia, Modena, Modena, Italy Vitalii Ivanov, Department of Manufacturing Engineering Machine and Tools, Sumy State University, Sumy, Ukraine

Lecture Notes in Mechanical Engineering (LNME) publishes the latest developments in Mechanical Engineering - quickly, informally and with high quality. Original research reported in proceedings and post-proceedings represents the core of LNME. Volumes published in LNME embrace all aspects, subfields and new challenges of mechanical engineering. Topics in the series include: • • • • • • • • • • • • • • • • •

Engineering Design Machinery and Machine Elements Mechanical Structures and Stress Analysis Automotive Engineering Engine Technology Aerospace Technology and Astronautics Nanotechnology and Microengineering Control, Robotics, Mechatronics MEMS Theoretical and Applied Mechanics Dynamical Systems, Control Fluid Mechanics Engineering Thermodynamics, Heat and Mass Transfer Manufacturing Precision Engineering, Instrumentation, Measurement Materials Engineering Tribology and Surface Technology

To submit a proposal or request further information, please contact the Springer Editor of your country: China: Dr. Mengchu Huang at [email protected] India: Priya Vyas at [email protected] Rest of Asia, Australia, New Zealand: Swati Meherishi at [email protected] All other countries: Dr. Leontina Di Cecco at [email protected] To submit a proposal for a monograph, please check our Springer Tracts in Mechanical Engineering at http://www.springer.com/series/11693 or contact [email protected] Indexed by SCOPUS. The books of the series are submitted for indexing to Web of Science.

More information about this series at http://www.springer.com/series/11236

Vilas R. Kalamkar Katarina Monkova •

Editors

Advances in Mechanical Engineering Select Proceedings of ICAME 2020

123

Editors Vilas R. Kalamkar Department of Mechanical Engineering Visvesvaraya National Institute of Technology Nagpur, India

Katarina Monkova Faculty of Manufacturing Technologies Technical University of Kosice Presov, Slovakia

ISSN 2195-4356 ISSN 2195-4364 (electronic) Lecture Notes in Mechanical Engineering ISBN 978-981-15-3638-0 ISBN 978-981-15-3639-7 (eBook) https://doi.org/10.1007/978-981-15-3639-7 © Springer Nature Singapore Pte Ltd. 2021 This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, expressed or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. This Springer imprint is published by the registered company Springer Nature Singapore Pte Ltd. The registered company address is: 152 Beach Road, #21-01/04 Gateway East, Singapore 189721, Singapore

Preface

It is our pleasure and honour to bring you these findings of research and innovation from the International Conference on Advances in Mechanical Engineering (ICAME 2020) held on 10 and 11 January 2020 organized by the Department of Mechanical Engineering, Visvesvaraya National Institute of Technology, Nagpur, India. This conference was the beginning of a year-long Diamond Jubilee year celebrations of VNIT Nagpur’s foundation day. ICAME 2020 provided an international forum where researchers, academicians and scientists from interdisciplinary fields presented their synergistic solutions to frontier issues of mechanical engineering. We received around 200 research manuscripts from various domains like thermal engineering, CFD, machine design, sustainability, IoT, robotics, manufacturing engineering, biomechanics, machine learning, machine vision, optimization, industrial engineering and many other allied domains. During 12 technical sessions spread over two days, the conference witnessed the presentations by participants from different NITs, IITs and universities in India as well as abroad. Out of 200 plus received papers, only 101 manuscripts are accepted for inclusion in this proceedings. The keynote talks, technical sessions and panel discussions of the conference were focused on holistic contributions of mechanical engineering concerning the society in general and industry in particular. We are highly grateful to the authors for their contributions and all the expert reviewers for their valuable advice. We take this opportunity to thank the members of the organizing committees for their unwavering commitment. We are indebted to TEQIP-III, MSME-DI Nagpur, DST-SERB New Delhi, MOIL Nagpur, SBI VRCE Branch Nagpur and many other industries, establishments and agencies in India for their generous sponsorship and support for the conference.

v

vi

Preface

We extend our heartfelt gratitude to Springer Nature for its professional assistance and particularly Mr. Akash Chakraborty and Ms. Rini Christy who supported this publication. Nagpur, India Presov, Slovakia

Prof. Vilas R. Kalamkar Prof. Katarina Monkova

Contents

Dual Quaternion-Based Kinematic Modelling of Serial Manipulators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mohsin Dalvi, Shital S. Chiddarwar, Saumya Ranjan Sahoo, and M. R. Rahul Performance Analysis of Corrugated Inclined Basin Solar Distillation System Coupled with Parabolic Trough Collector . . . . . . . . . . . . . . . . . Sandeep Joshi, Shubham Tagde, Aboli Pingle, Nikhil Bhave, and Tushar Sathe Mechanical Design of Omnidirectional Spherical Wall Traversing Robot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Yogesh Phalak, Rajeshree Deotalu, Onkar, and Sapan Agrawal Fabrication and Performance Analysis of a Device to Transform Vibration Energy on an Automobile . . . . . . . . . . . . . . . . . . . . . . . . . . . Dheeraj H. Bonde, Nitin K. Panche, Hrishikesh S. Meshram, Vrushabh W. Dhongade, Atul V. Dharmik, Jayesh D. Parate, Mangesh G. Pardhi, and Vinit S. Gupta Robust Backstepping Controller for an Omniwheeled Mobile Robot with Uncertainties and External Disturbances . . . . . . . . . . . . . . . . . . . . Zeeshan Ul Islam, Saumya Ranjan Sahoo, Mohammad Saad, Uddesh Tople, and Amrapali Khandare

1

9

17

25

35

Micro-mechanical Analyses of Particle Reinforced ex situ Bulk Metallic Glass Matrix Composites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . S. Gouripriya and Parag Tandaiya

43

Life Estimation of Circumferentially Notch Round Bars Using J Integral . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Richa Agrawal, Rashmi Uddanwadiker, and Pramod M. Padole

51

vii

viii

Contents

Placement of Heated Blocks Under Forced Convection for Enhanced Heat Transfer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Shankar Durgam, Shakkottai Venkateshan, Thirumalachari Sundararajan, Milankumar Nandgaonkar, Pravin D. Sawarkar, and Aaryan Durgam Analysis of Track Vibration for Metro Rail . . . . . . . . . . . . . . . . . . . . . . Chaitanya V. Bhore, Atul B. Andhare, Pramod M. Padole, and Mayur D. Korde Localization of a Four-Wheeled Omnidirectional Mobile Robot Using Sensor Data: A Kalman Filter Approach . . . . . . . . . . . . . . . . . . . Saumya Ranjan Sahoo, Shital S. Chiddarwar, Mohsin Dalvi, and M. R. Rahul

59

67

75

Capacitated Vehicle Routing Problem with Interval Type-2 Fuzzy Demands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V. P. Singh and Kirti Sharma

83

Kinematic, Dynamic and Stiffness Analysis of an Asymmetric 2PRP-PPR Planar Parallel Manipulator . . . . . . . . . . . . . . . . . . . . . . . . . Deep Singh, Rutupurna Choudhury, and Yogesh Singh

91

CFD Analysis for Heat Transfer Enhancement of Microchannels Heat Sink Using Nanofluid Flow in Case of Electronics Device . . . . . . . . . . . Sushant Suresh Bhuvad, Arvind Kumar Patel, and S. P. S. Rajput

99

Burr Registration and Trajectory Planning of 3D Workpiece Using Computer Vision . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107 M. R. Rahul, Rohini Y. Bhute, Shital S. Chiddarwar, Mohsin Dalvi, and Saumya Ranjan Sahoo In-situ Microwave-Assisted Casting of ASTM B23 Tin-Based Babbitt Alloy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115 Sameer S. Gajmal and Dadarao N. Raut Optimization of Heat Transfer Behavior of Industrial Refrigerants Through Different Cross-Section Microchannels . . . . . . . . . . . . . . . . . . 127 Gourab Chakraborty, Shubhankar Sarkar, and Arunabha Chanda Evaluation of Two-Body Abrasive Wear Using FIS and ANN . . . . . . . . 139 Mehar Amit Kumar Computational Analysis of Dual Expander Aerospike Nozzle . . . . . . . . 151 Aswith R. Shenoy, T. S. Sreekumar, Pranav Menon, and Gerogi Alex A Study on Performance and Emission Characteristics of Diesel Engine for Lower Blends of Karanja Biodiesel . . . . . . . . . . . . . . . . . . . 159 V. R. Patil, S. S. Sane, and S. S. Thipse

Contents

ix

Experimental Comparison Between Friction Stir Welding and Underwater Friction Stir Welding on Al6061 Alloys . . . . . . . . . . . . 169 Hiten J. Mistry, Piyush S. Jain, and J. Vaghela Tinej Wear Particle Analysis Using Fractal Techniques . . . . . . . . . . . . . . . . . 179 Puja P. More and M. D. Jaybhaye Strategies for Low Engine Speed Torque Enhancement of Natural Gas Engine Used for Commercial Vehicles: Observations with Compression Ratio . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189 Pritesh J. Suple, Chandrakant R. Sonawane, S. S. Thipse, J. P. Mohite, and N. B. Chougule In-house Fabrication and Calibration of Silver Thin Film Gauge . . . . . 197 Akash Jadhav and Ravi K. Peetala Study of Shock Wave Boundary Layer Interaction in Hypersonic Flows Using Various Turbulence Models . . . . . . . . . . . . . . . . . . . . . . . . 205 Aniruddha Kane and Ravi K. Peetala Study of Effect on Engine Performance Using 15% HCNG Blend Versus CNG Using a Simulation Approach . . . . . . . . . . . . . . . . . . . . . . 213 K. P. Kavathekar, S. S. Thipse, S. D. Rairikar, S. B. Sonawane, P. S. Sutar, and D. Bandyopadhyay Behaviour of NiTi Based Smart Actuator for the Development of Planar Parallel Micro-Motion Stage . . . . . . . . . . . . . . . . . . . . . . . . . . 221 Deep Singh, Yogesh Singh, and Manidipto Mukherjee Multi-objective Optimization of Inconel 718 Using Combined Approach of Taguchi—Grey Relational Analysis . . . . . . . . . . . . . . . . . . 229 Manav Sheth, Kunj Gajjar, Aryan Jain, Vrund Shah, Het Patel, Rakesh Chaudhari, and Jay Vora The Effect of State Variables on Nucleation of Earthquake Using the Rate and State Friction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237 Nitish Sinha, Arun K. Singh, and Avinash D. Vasudeo Finite Element Analysis of Type I and Type II Fracture with PFN Implant—A Comparative Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243 Sandeep Rathor, Jayamalya Jena, Rashmi Uddanwadikar, and Ashutosh Apte Postural Evaluation of Construction Labourers Engaged in Excavation Work Using Newly Developed NERPA Method and Its Validation Through REBA and WERA Methods . . . . . . . . . . . . 253 Manoj T. Gajbhiye, Debamalya Banerjee, and Saurav Nandi

x

Contents

Influence of Stress Bar Length on the Response of a Stress Wave Force Balance Using Finite Element Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . 263 Sushmita Deka, Ramesh Babu Pallekonda, and Maneswar Rahang Relative Power Variation in Frequency Sub-bands of the EEG Signal During Painful Stimuli . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 271 Sameer Raj Singh and Ashish B. Deoghare A Study on the Effect of GTAW Input Current on Surface Distortion of Thin CRNO Electrical Steel Sheets . . . . . . . . . . . . . . . . . . . . . . . . . . 279 Bhushan Y. Dharmik and Nitin K. Lautre Heat Transfer and Pressure Drop Inside Duct with Different Surface Profiles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 289 P. P. Shirpurkar, V. M. Sonde, P. T. Date, and T. R. Badule A Hybrid Process Monitoring Strategy for Steel Making Shop . . . . . . . 299 Ashish Kumar, Anupam Das, and Swarnambuj Suman Analysis of Electrolyte Flow in IEG During Electrochemical Grinding of MMC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 307 Nisha Gupta, Avanish Kumar Dubey, and Dhruv Kant Rahi Static Structural Analysis of Roll Cage of an All-Terrain Vehicle . . . . . 317 Sushant Satputaley, Karan Ksheersagar, Bijay Sankhari, Rahul Kavishwar, and Kshitij Waghdhare Design of Motorcycle Handlebar for Reduction of Vibrations Using Tuned Mass Damper . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 329 Sumit S. Khune and Amit R. Bhende Development of a Diagnostive Tool for Prediction of Severity of Coronary Artery Disease . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 337 Pooja Jhunjhunwala, Pramod M. Padole, and S. B. Thombre Design, Modelling and Optimization of Artificial Limb for Lower-Extremity Amputees Based on CATIA . . . . . . . . . . . . . . . . . 345 Smit V. Motghare An Experimental Study on Surface Roughness in Slicing Tungsten Carbide with Abrasive Water Jet Machining . . . . . . . . . . . . . . . . . . . . . 353 Ranjan Singh, Virendra Singh, and T. V. K. Gupta Energy Absorption Characteristics of Single and Double-Walled Square Tubes Subjected to Axial Crushing . . . . . . . . . . . . . . . . . . . . . . 361 Sanjay S. Toshniwal and Raghu V. Prakash Field Data Analysis Using Work Measurement Techniques in a Packaging Industry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 371 Chinmay M. Salkar, Gaurao J. Tapare, Mayank A. Murkute, Chetan R. Zingre, Hansraj A. Mohod, and Vinit S. Gupta

Contents

xi

Comparative Study of Nanofinishing of Si (100) Using DDMAF and Allied Processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 377 Kheelraj Pandey, Ajendra Kumar Singh, and Gaurav Raj Pandey Optimization of Thickness of Hollow Punch–Die for Proposed Solar-Assisted Leaf Plate and Cup Making Machine . . . . . . . . . . . . . . . 385 Abhay Nilawar, Pravin Potdukhe, and Deepak V. Bhope Development of Briquette Cum Pellet Making Machine . . . . . . . . . . . . . 391 Yeshwant M. Sonkhaskar, Gajanan R. Nikhade, Saket Dharmik, Utkarsh Deshmukh, and Pramod Dhote Towards the Development of Low-Cost Vacuum Setup for Customized Implant Manufacturing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 399 Sanjay Randiwe, Dheeraj Bhiogade, and Abhaykumar M. Kuthe Simulation Study on Effect of Variable Curvature on the Modal Properties of Curved Cantilever Beams . . . . . . . . . . . . . . . . . . . . . . . . . 407 Aqleem Siddiqui, Girish Dalvi, Akshay Patil, and Surabhi Chavan Variation in the Properties of Spot Weldments of Cold Rolled Mild Steel Welded with Filler Metal by Annealing Treatment . . . . . . . . . . . . 415 Sushil T. Ambadkar and Deepak V. Bhope Comparison of Metro Track Vibration with Federal Transit Administration Limits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 425 Chaitanya V. Bhore, Atul B. Andhare, and Pramod M. Padole Effect of Moisture Content and Fiber Orientation on the Mechanical Behavior of GFRP Composites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 433 Alok Behera, M. M. Thawre, Atul Ballal, Prathamesh Babrekar, Pratik Vaidya, Satya Vijetha, and Tushar Sawant Experimental Investigation and Simulation of Modified Evaporative Cooling System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 441 Manju Lata and Dileep Kumar Gupta Effects of Different Vegetable Oils and Additives in Gearbox Operation and its Condition Monitoring . . . . . . . . . . . . . . . . . . . . . . . . 449 Anupkumar Dube and M. D. Jaybhaye Study and Analysis of Various Parameters of Bio-mechanization Plant . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 457 Deepak Patil, Rahul Barjibhe, Lakhan Meghani, Omkar Nanaware, Tejas More, and Aditya Pujari Robust Sliding Mode Controller (RSMC) for an Omniwheeled Mobile Robot with Uncertainties and External Perturbations . . . . . . . . . . . . . . 465 Mohammad Saad, Uddesh Tople, Amrapali Khandare, and Zeeshan Ul Islam

xii

Contents

The CFD Analysis of Convection Heat Transfer with Magnetic Field in the 2D Domain Using OpenFOAM . . . . . . . . . . . . . . . . . . . . . . . . . . 473 Ranjit J. Singh and Trushar B. Gohil Design of a Remote Racking Module for Racking Operation . . . . . . . . . 481 Alex Sherjy Syriac and M. R. Rahul A Coupled Heat Transfer and Artificial Neural Network Based Model for Accelerated Direct Cooling of Steel Plate . . . . . . . . . . . . . . . . . . . . . 487 Sagar Dave, Sirshendu Chattopadhyay, and Deepak Gupta Effect of Air Distribution on Cooling of Photovoltaic Panel and Its Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 495 Someshwar S. Bhakre and Pravin D. Sawarkar Numerical Investigations of Photovoltaic Phase Change Materials System with Different Inclination Angles . . . . . . . . . . . . . . . . . . . . . . . . 503 Tushar Sathe, A. S. Dhoble, Sandeep Joshi, C. Mangrulkar, and V. G. Choudhari Edge Feature Based Classification of Breast Thermogram for Abnormality Detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 511 Shawli Bardhan and Sukanta Roga Analytical Approach to Develop a Robust Mechanism for On-Orbit Gimballing of Satellite Antenna . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 519 V. Sri Pavan RaviChand, Anoop Kumar Srivastava, Abhishek Kumar, H. N. Suresha Kumar, and K. A. Keshavamurthy Impact of Rock Abrasivity on TBM Cutter-Discs During Tunnelling in Various Rock Formations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 527 N. N. Sirdesai, A. Aravind, and S. Panchal Tool Condition Prediction Using Acoustic Signal Processing and Learning-Based Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 535 Pranjali S. Deole and Priya M. Khandekar Finite Element Simulation of Ballistic Response of Metallic Sandwich Structures with Aluminium Foam Core . . . . . . . . . . . . . . . . . . . . . . . . . 543 Nikhil Khaire, Vivek Bhure, and Gaurav Tiwari Crushing Behavior of Thick Circular High Strength Aluminum Tube Against Quasi-static Axial Loading . . . . . . . . . . . . . . . . . . . . . . . . . . . . 551 Vivek Patel, Sanket Suresh Kalantre, Gaurav Tiwari, and Ravikumar Dumpala Estimation of Burr Dimensions Using Image Processing for Robotic Deburring . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 559 Rohini Y. Bhute and M. R. Rahul

Contents

xiii

Study of Structural and Mechanical Behaviour of Severe Plastically Deformed Al–Mg(AA 5052) Alloy Processed by Constrained Groove Pressing Technique . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 567 Jaya Prasad Vanam, Vinay Anurag Potnuri, and Sree Vidya Sravya Nallam Shear Rate Dependent Frictional Behavior of the Granular Layer . . . . 577 Pawan Kumar Soni and Arun K Singh Mathematical Overview on Omnidirectional Spherical Wall Traversing Robot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 583 Yogesh Phalak, Rajeshree Deotalu, Onkar, and Sapan Agrawal Finite Element Analysis of Ballistic Impact on Monolithic and Multi-layered Target Plate with and Without Air Gap . . . . . . . . . . 591 Rohit Kumar, Manoj Kumar, and Pramod Kumar Additive Manufacturing Process Selection Using MCDM . . . . . . . . . . . . 601 Vishwas Dohale, Milind Akarte, Shivangni Gupta, and Virendra Verma Evaluation and Improvement of Makespan Time of Flexible Job Shop Problem Using Various Dispatching Rules—A Case Study . . . . . . . . . . 611 Mohan Bihari and P. V. Kane The Impact of Building Orientation on Microhardness and Surface Roughness of Direct Metal Laser Sintered Inconel Alloy . . . . . . . . . . . . 619 Ajay Kumar Maurya and Amit Kumar Investigation on Elevated Temperature Tribological Performance of Alloy 718 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 629 S. Anand Kumar, Ravikumar Dumpala, K. Uday Venkat Kiran, and R. Gnanamoorthy Digital Twin for Shell and Tube Heat Exchanger in Industry 4.0 . . . . . 637 Himanshu Singh, Utkarsh Mishra, Prateek Saxena, Ganesh Shetiya, and Y. M. Puri Model-Based Synchronized Control of a Robotic Dual-Arm Manipulator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 645 Akshay Katpatal, Ajinkya Parwekar, and Alok Kumar Jha Prioritizing the Travelling Criteria for Customer-Centric Business Model of Public Transport System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 655 Prasad Lanjewar and Dhananjay A. Jolhe Effect of Friction Stir Processing on the Sliding Wear Characteristics of AZ91 Mg Alloy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 663 Hemendra Patle, K. Uday Venkat Kiran, B. Ratna Sunil, and Ravikumar Dumpala

xiv

Contents

Effect of Varying In-Plane Loads and Cutout Size on Buckling Behavior of Laminated Panels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 671 K. S. Subash Chandra, K. Venkata Rao, and T. Rajanna Effect of Machining Parameters on Surface Roughness and Tool Flank Wear in Turning of Haynes 25 Alloy . . . . . . . . . . . . . . . . . . . . . . . . . . . 679 Atul B. Andhare, K. Kannathasan, and Manoj Funde Performance Appraisal of Cryogenically Treated Tool in Dry, MQL and Cryogenic Machining of Inconel 718 . . . . . . . . . . . . . . . . . . . . . . . . 687 Yogesh V. Deshpande, Atul B. Andhare, and Pramod M. Padole Finite Element Simulation for Turning of Haynes 25 Super Alloy . . . . . 695 Atul B. Andhare, K. Kannathsan, and Manoj Funde Optimization of Machining Parameters for Turning of Haynes 25 Cobalt-Based Superalloy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 703 Atul B. Andhare, K. Kannathasan, and Manoj Funde A Compact Hinge Mechanism for Radial Rib Antenna . . . . . . . . . . . . . 711 Rahul Ghatak, Milind Undale, Mariya Ratlami, Prakher Singhal, G. Ravi Teja, N. S. Murali, and K. A. Keshavamurthy Suntracker on Rocker-Bogie Mechanism . . . . . . . . . . . . . . . . . . . . . . . . 719 Shruti Murarka, Aditya Wadichar, Shravar Tanawde, Abhijit Rehpade, Dhruv Agrawal, Mohammad Saad, and Sharan Bajjuri Market Basket Analysis: Case Study of a Supermarket . . . . . . . . . . . . . 727 Anup R. Pillai and Dhananjay A. Jolhe Experimental Investigation of Effect of Nanoparticle Concentration on Thermo-physical Properties of Nanofluids . . . . . . . . . . . . . . . . . . . . . 735 Prashant Maheshwary, C. C. Handa, K. R. Nemade, and N. N. Gyanchandani A Framework for Robot Programming via Imitation . . . . . . . . . . . . . . . 743 Abhishek Jha, Shital S. Chiddarwar, and Sanjay G. Sakharwade Optimizing EDM Parameters for Machining Cu102 and Finding Regression Equation of MRR and Surface Finish . . . . . . . . . . . . . . . . . 751 Amit Motwani, Y. M. Puri, and Gangadhar Navnage Multidisciplinary Solution Avenues in Mechanical Engineering . . . . . . . 759 Pradnya Gharpure OLSAC: Open-Source Library for Swarm Algorithms and Communication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 769 Harshad Zade, Mayuresh Bhoyar, Mayuresh Sarode, Neha Marne, Unmesh Patil, Ajinkya Kamat, and Vedant Ranade

Contents

xv

Development of Crank–Connecting Rod Attachment for Electric Discharge Machining of Curved Holes . . . . . . . . . . . . . . . . . . . . . . . . . . 777 Diwakar Makireddi, Y. M. Puri, and V. D. Ghuge Effect of Crack Angle on Stress Shielding in Bone and Orthopedic Fixing Plate Implant: Design and Simulation . . . . . . . . . . . . . . . . . . . . . 785 Ratna Raju Lam, V. V. Kondaiah, Y. Naidubabu, Ravikumar Dumpala, and B. Ratna Sunil IoT-Based Ambiance Monitoring System . . . . . . . . . . . . . . . . . . . . . . . . 793 Hritwik Singh Parihar, Rajesh Nagula, Mayank Bumb, Danish Gada, Sharan Bajjuri, Rishesh Agarwal, and Simran Chauhan Hand Gesture Control of Computer Features . . . . . . . . . . . . . . . . . . . . 799 Rishabh Runwal, Shivraj Dhonde, Jatin Pardhi, Suraj Kumar, Aadesh Varude, Mayuresh Sarode, Mayuresh Bhoyar, Simran Chauhan, and Neha Marne Industry 4.0 Applications in Agriculture: Cyber-Physical Agricultural Systems (CPASs) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 807 Rohit Sharma, Shreyanshu Parhi, and Anjali Shishodia Person Following Mobile Robot Using Multiplexed Detection and Tracking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 815 Khush Agrawal and Rohit Lal Sliding Wear Characteristics of Silver Particles Incorporated Electroless Nickel Phosphorus Composite Coatings . . . . . . . . . . . . . . . . 823 Bijoy Ramakrishnan, K. Uday Venkat Kiran, B. Ratna Sunil, and Ravikumar Dumpala Investigations on Engine Emission Using Biodiesel with Different Compression Ratios . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 831 Mohd Zeeshan and Sanjay K. Sharma Investigation of Thermal Desalination System Using Heat Recovery . . . 839 Rajan K. Petkar, Chandrakant R. Sonawane, and Hitesh N. Panchal

About the Editors

Dr. Vilas R. Kalamkar is a professor and head of mechanical engineering department of Visvesvaraya National Institute of Technology, Nagpur, India. He received his PhD in aerodynamics from the Indian Institute of Technology Bombay. His areas of research are CFD and turbo machinery. He has more than 20 years of teaching experience. Apart from several research articles in journals and conference proceedings, he has also filed 2 patents. He has undertaken 4 research projects and many consultancy projects. He was invited to deliver keynote lectures at various international conferences in places like Korea, Malaysia and Dubai. He was conferred with the best teacher award, research fellowships and best paper award for his contribution in teaching and research. Dr. Katarina Monkova is a full professor of Faculty of Manufacturing Technologies with the seat in Presov, Slovakia. She is a scientific researcher at technical university with pedagogical activities like computer-aided technical devices design, analysis and simulation, cellular material. She has done PhD in manufacturing technologies and metallurgy. She has more than 300 publications with 20 invited presentations at international conferences. She is also a leader of 7 national scientific-research projects and 1 international project and member of more than 20 national projects with educational and scientific grants. She is a recipient of various awards like Rector’s award and Dean’s Award.

xvii

Dual Quaternion-Based Kinematic Modelling of Serial Manipulators Mohsin Dalvi , Shital S. Chiddarwar , Saumya Ranjan Sahoo , and M. R. Rahul

Abstract In this paper, a dual quaternion-based methodology for computing the forward and inverse kinematic models for a serial manipulator is presented. A dual quaternion-based forward kinematics model is developed for the Kuka LBR IIWA 7 R800 cobot. An inverse kinematics model is developed that uses dual quaternion differential kinematics and includes Jacobian transpose and damped least squares methods for determining Jacobian pseudo-inverse. Implementation of these methods on a given trajectory shows that, compared to damped least squares, the Jacobian transpose method is faster, but is less immune to singularity and gives more jerky motions. Keywords Cobot · Dual quaternions · Differential motion operator · Jacobian transpose · Damped least squares

1 Introduction During robot programming, the orientation and position of robot end-effector must change as smoothly as a human hand. Quaternions, initially developed as a generalization of complex numbers for three dimensions, are robust than the popularly used Euler angles and rotation matrices for representing orientations. They are compact, computationally efficient, immune to gimbal lock and mathematical singularities, and also provide natural orientation interpolation [1]. However, using vectors and quaternions for representing simultaneous translations and rotations leads to inconsistencies [2]. A dual quaternion (DQ) uses dual numbers to unify rotations and translations into a single state instead of defining separate vectors for them [3]. DQs have been used for kinematic modelling and pose control of serial manipulators [4, 5]. The M. Dalvi (B) · S. S. Chiddarwar · S. R. Sahoo · M. R. Rahul Department of Mechanical Engineering, Visvesvaraya National Institute of Technology, Nagpur, Maharashtra 440010, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_1

1

2

M. Dalvi et al.

available literature on modelling with quaternions and DQs shows that there are ambiguities in representation (Hamiltonian and NASA-JPL), handedness (right and left) and reference frames (global and body) [6]. Differential kinematics is a widely used approach for inverse kinematics (IK) modelling. DQ and Plucker coordinates have been used to derive the geometrical Jacobian for serial manipulators [7]. DQs and quaternion exponential maps have been used to solve IK problems by considering joint limits [8]. Unlike the DQ differential operator, dealing with Jacobian matrix is well discussed in the literature. The flow of the paper is as follows: Sect. 2 introduces dual quaternions, which is used in Sect. 3 to develop the forward and inverse kinematic models for a serial manipulator. The approach is applied for Kuka LBR IIWA 7 R800 cobot in Sect. 4. The results are discussed in Sect. 5, followed by conclusion in Sect. 6.

2 Mathematical Preliminaries In this work, quaternions use Hamiltonian in right-handed coordi  representation nates. A quaternion is represented as p = p0 , p = p0 + p1 i + p2 j + p3 k, where p ∈ H, p0 ∈ R is a scalar and p = ( p1 , p2 , p3 ) ∈ R3 is a vector. The orthogonal unit vectors i, j , k satisfy the quaternion property i 2 = j 2 = k2 = i j k = −1. This property is used to define quaternion algebra. Addition and multiplication  + q , p + q and p ⊗ q = p0 q0 − p · q, p0 q + q0 p + are given by p + q = p 0 0  p × q , respectively. Multiplication is non-commutative, but follows distributive and associative properties. The multiplicative inverseis given by p −1 = p ∗ /  p2 , where p ∗ is the conjugate, defined as p ∗ = p0 , − p , and  p is the norm, given   √ p02 + p · p, 0 . The identity quaternion is 1 = [1, 0]. by  p = p ⊗ p ∗ = A dual number has the form a = ar + ad where a ∈ D, ar ∈ R is the real part, ad ∈ R is the dual part. The dual unit , satisfying  = 0,  2 = 0 is used to define dual number algebra. Higher-order terms get removed during Taylor series expansion  (a). This gives of dual function about real part to give f (a + b) =√f (a) + b f√ relations such as cos (a + b) = cos a − b sin a, and a + b = a −  2√b a . A dual quaternion is written as p = pr + pd , where pr , pd ∈ H and p ∈ DH. Addition and multiplication are given by p + q = ( pr + qr ) +  ( pd + qd ) and p ⊗ q = pr ⊗ qr + ( pr ⊗ qd + pd ⊗ qr ), respectively. A DQ has three conjugates,     namely dual conjugate p¯ = pr − pd , quaternion conjugate p ∗ = pr∗ + pd∗ ,   and dual quaternion conjugate p¯ ∗ = pr∗ − pd∗ . The identity DQ is 1 = [1, 0] + [0, 0]. The inverse p −1 = pr−1 − pr−1 ⊗ pd ⊗ pr−1 exists if pr =0. The DQ norm is         obtained from  p  = p ⊗ p ∗ =  pr 2 +  2 pr 0 pd0 + pr · pd , 0 . If  pr  = 1 and pr 0 pd0 + pr · pd = 0, then p is a unit DQ.

Dual Quaternion-Based Kinematic Modelling of Serial Manipulators

3

3 Kinematic Modelling of Serial Manipulator   The unit quaternion rˆ = cos θ2 , uˆ sin θ2 represents rotation by angle θ about unit

ˆ The quaternion t = [0, t] depicts translation t = tx , t y , tx . Then, the comvector u. posite transformation of rotation r followed by translation t is given by the unit DQ



p = t ⊗ r = 1 +  21 t ⊗ rˆ + 0 = rˆ + 2 t ⊗ rˆ     ˆ cos θ2 t + sin θ2 (t × u) ˆ = cos θ2 , sin θ2 uˆ +  21 −sin θ2 (t · u),

(1)

For a given unit DQ p = pr + pd , quaternions rˆ and t are obtained using rˆ = pr and t = 2 pd ⊗ pr∗ = 2 pd ⊗ rˆ ∗ . For rotational quaternion rˆ = [r0 , r], rotation parameters θ and uˆ are obtained as θ = 2 cos−1 (r0 ) and uˆ = r/ sin θ2 . When a vector v 0 is subjected to transformation p, the new vector v 1 is obtained from v1 = [1, 0] + [0, v 1 ] = p ⊗ ([1, 0] + [0, v 0 ]) ⊗ p¯ ∗ .

3.1 Forward Kinematics (FK) Modelling A link i of a serial manipulator as seen in Fig. 1 has frames {i − 1} and {i} attached to joints i and i + 1 present at its ends by following DH notations [9]. In order to coincide frame {i − 1} with frame {i}, an intermediate frame i  is defined at intersection of Zi−1 and  X i axes. Then, using Eq. 1, screw transforms q iZ = cos θ2i , 0, 0, sin θ2i +  − d2i sin θ2i , 0, 0, d2i cos θ2i about Z i−1 axis and     q iX = cos α2i , sin α2i , 0, 0 +  − a2i sin α2i , a2i cos α2i , 0, 0 about X i axes are carried out. Let Cθ = cos θ2i , Sθ = sin θ2i , Cα = cos α2i , Sα = sin α2i , A = a2i and D = d2i . Then, the DQ that maps frame {i} to frame {i − 1} is given by i−1

q i = q iZ ⊗ q iX = [Cθ Cα , Cθ Sα , Sθ Sα , Sθ Cα ] +  [−DSθ Cα − ACθ Sα , −DSθ Sα + ACθ Cα , DCθ Sα + ASθ Cα , DCθ Cα − ASθ Sα , ]

(2)

For n-degree serial manipulator, transformation DQ is 0 q n = 0 q 1 ⊗ · · · ⊗ n−1 q n .

3.2 Inverse Kinematics (IK) Modelling Differential kinematics involves mapping differential change in joint parameters θ˙ to differential change in pose q˙ by means of Jacobian matrix J as q˙ = Jθ˙ . For non-square Jacobian, pseudo-inverse is determined by using JT q˙ = JT Jθ˙ to

−1 T

−1 J q. ˙ One way to deal with JT J is to approximate it to obtain θ˙ = JT J

4

M. Dalvi et al.

Fig. 1 Screw motions for frame transformation

α=

 q, ˙ JJT q˙ JJT q, ˙ JJT q˙

 [10], where ·, · is a dot product operator. This is called the Jaco-

bian transpose (J T ) method. To further reduce the chances of JT J losing rank, the damped least squares method (DLS) is used [10]. Here, a damping constant δ ≈ 0.001

−1 T J q. ˙ The ith in the diagonal elements modifies the expression to θ˙ = JT J − δ 2 I ∂ 0 column of Jacobian J is ( q n ) where n = number of joints. ∂θi For given pose q = qr + qd , linear velocity v ∈ R3 and angular velocity w ∈ R3 , the DQ differential operator is q˙ = 21 ξ ⊗ q where ξ = [0, ω] +  [0, v + t × ω] is twist in world frame [11] and [0, t] = 2qd ⊗ qr∗ . Then, the DQ differential operator becomes q˙ = 21 ([0, ω] +  [0, v + t × ω]) ⊗ (qr + qd ). For initial pose q 0 of a serial manipulator, let corresponding joint parameters be θ 0 . Now, for some J (θ k ) and q˙ k , when θ˙ k is obtained from IK model, then θ k is updated to θ k+1 = θ k + θ˙ k , and updated pose from FK model becomes X F (θ k+1 ). However, due to various reasons such as linearization, this does not match the next pose q k+1 = q k ⊗ q˙ k . The pose error, given by (q k ) E = X F (θ k+1 )∗ ⊗ q k+1 , is fed back to the next differential pose q˙ k+1 .

Dual Quaternion-Based Kinematic Modelling of Serial Manipulators

5

4 Application to Kuka LBR IIWA 7 R800 Cobot The classical DH convention [9] is used to assign coordinate frames to the 7-dof Kuka LBR IIWA 7 R800 cobot (or collaborative robot) and derive the robot architecture as seen in Fig. 2. A library for dual quaternions is developed in Python 3.6 for implementing the algorithms. The tests are performed on a Windows 8 PC with Intel i5-4570 3.2 GHz processor, 8 GB DDR3 RAM and GeForce 625 graphics card. The DQ outputs of forward kinematics model for a series of joint inputs are tallied with results from

Joint i ai (m) 1 0 2 0 3 0 4 0 5 0 6 0 7 0 (a)

(b)

αi (deg) di (m) θi (deg) −π/2 0.34 θ1 π/2 0 θ2 −π/2 0.4 θ3 π/2 0 θ4 −π/2 0.4 θ5 π/2 0.4 θ6 0 0 θ7 (c)

Fig. 2 Cobot under study (a) with frame assignment (b) and DH parameters [9] (c) Fig. 3 Closed trajectory showing orientation frames

6

M. Dalvi et al.

RoboAnalyzer software. The two IK models are applied to the trajectory shown in Fig. 3. The red, green and blue arrows correspond, respectively, to the X , Y and Z axes of the end-effector coordinate frame.

5 Results and Discussion The DQ FK model results matched perfectly with those obtained from the homogeneous transform matrix (HTM) FK model and RoboAnalyzer software. The DQ FK model required 224 additions and 312 multiplications as opposed to 234 additions and 356 multiplications needed in HTM. It is seen that using DQs over HTMs saved computational time, the savings being up to 50% in some cases. Applying the J T and DLS methods on the trajectory shows that J T gives solutions faster, but also gives more jerky motions. For J T , mean solution time ranges from 16 to 33 ms, whereas for DLS, the same ranges from 27 to 66 ms. It is also observed that the DLS method outperforms the Jacobian transpose method when not near singularity. The graphs of two DQ components in Fig. 4 show that fluctuations in DLS method are less compared to those in J T method. It is seen that the trajectory loop did not get closed in either method, which means pose error is not eliminated even when feedback is used. This makes a case for exploring more feedback controllers.

6 Conclusion In this work, dual quaternions (DQs) were used to develop the forward and inverse kinematics models for a serial manipulator. The DQ differential operator and two methods of solving inverse differential kinematics with DQs were discussed. The approach was implemented on the Kuka LBR IIWA 7 R800 cobot, and error during simulation of tracing a loop trajectory was studied. Though DQs seem unintuitive

Fig. 4 Comparison of deviations in trajectory traced with J T and DLS IK algorithms taking first and sixth DQ elements as example

Dual Quaternion-Based Kinematic Modelling of Serial Manipulators

7

to humans, they lend themselves well to working with elaborate trajectories. Future works include velocity and acceleration control, as well as DQ implementation of faster IK algorithms such as FABRIK.

References 1. Shoemake K (1985) Animating rotation with quaternion curves. In: ACM SIGGRAPH computer graphics, vol 19. ACM, pp 245–254 2. Banerjee P, Zetu D (2001) Virtual manufacturing. Wiley 3. Jia YB (2013) Dual quaternions. Technical report, Iowa State University 4. Daniilidis K (1999) Hand-eye calibration using dual quaternions. Int J Robot Res 18(3):286– 298 5. Pham HL, Perdereau V, Adorno BV, Fraisse P (2010) Position and orientation control of robot manipulators using dual quaternion feedback. In: 2010 IEEE/RSJ international conference on intelligent robots and systems, pp 658–663 6. Sommer H, Gilitschenski I, Bloesch M, Weiss S, Siegwart R, Nieto JI (2018) Why and how to avoid the flipped quaternion multiplication. CoRR 7. Özgür E, Mezouar Y (2016) Kinematic modeling and control of a robot arm using unit dual quaternions. Robot Auton Syst 77:66–73 8. Kenwright B (2013) Inverse kinematics with dual-quaternions, exponential-maps, and joint limits. Int J Adv Intell Syst 6(1, 2):53–65 9. Siciliano B, Sciavicco L, Villani L, Oriolo G (2009) Robotics: modelling. In: Planning and control. Springer, London 10. Aristidou A, Lasenby J (2009) Inverse kinematics: a review of existing techniques and introduction of a new fast iterative solver. Technical report, University of Cambridge 11. Wang X, Han D, Yu C, Zheng Z (2012) The geometric structure of unit dual quaternion with application in kinematic control. J Math Anal Appl 389(2):1352–1364

Performance Analysis of Corrugated Inclined Basin Solar Distillation System Coupled with Parabolic Trough Collector Sandeep Joshi, Shubham Tagde, Aboli Pingle, Nikhil Bhave, and Tushar Sathe

Abstract Several designs of solar distillation system have been built over the past century. However, the development of an economical system with high productivity is a major challenge. Various researchers worked to improve the productivity of solar distillation system by improving the rate of evaporation and/or the rate of vapor condensation. In the current work, the evaporation rate is enhanced by basin modification and using a secondary heating medium. An inclined corrugated basin solar still is designed and fabricated and coupled with a parabolic trough collector. Experimental study was carried at Nagpur (21.14° N, 79.0882° E) during the months of April and May, and results indicate 13.58% increase in the thermal efficiency. Further, CFD analysis is carried out by using RNG (k − ε) turbulence model in ANSYS Fluent. The CFD results were found to be in good agreement with the experimental results, thus validating the CFD model to carry out any modifications in the future. Keywords Inclined basin solar still · CFD analysis · Corrugated basin · Parabolic trough collector

1 Introduction Water is one of the basic human needs. Water treatment amounts for about 8% of global energy consumption [1]. The use of solar still for water treatment and distillation is one of the most popular renewable energy solutions. However, low productivity of solar stills has been a matter of concern. Several researches have suggested various methods to enhance the productivity of the solar still with the use of fins, multi-basin, energy storage materials [2], wick [3, 4] reflectors [5–7], and coating the absorber surface with different films [8]. Also, S. Joshi · S. Tagde · A. Pingle (B) · N. Bhave Shri Ramdeobaba College of Engineering and Management, Nagpur, India e-mail: [email protected] T. Sathe Visvesvaraya National Institute of Technology, Nagpur, India © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_2

9

10

S. Joshi et al.

increasing the absorber area is suggested to help improve the performance of solar still which can be achieved by providing corrugations of V-shaped [9], wave-shaped [10], covered absorber plate with copper chips [8], etc. Some investigators have also used the solar collectors like flat plate collector [11, 12], evacuated tube collector [13], tracked parabolic trough collector [14], etc., as secondary heating source. The present work focuses on the improvement of rate of evaporation. An inclined corrugated basin solar still coupled with a parabolic trough collector is designed and fabricated. The performance of the system is simulated using CFD technique and validated using experimental analysis. The experiments are performed at Nagpur (21.14° N, 79.08° E) in the month of April and May. The details of simulation and experimental studies are discussed in the subsequent text.

2 Working Principle The schematic diagram of the modified system of inclined corrugated basin solar still coupled with parabolic trough collector is shown in Fig. 1. In the modified system, the corrugated basin receives heat from two sources, viz. incident solar radiations and the parabolic trough collector. The parabolic trough collector acts as the secondary heating source which heats water, and the heat of the water is then passed through the corrugated channel in the basin. This water is recirculated in the heating water circuit of parabolic trough collector. The raw water to be distilled is circulated over the basin via header pipe. The header pipe is drilled at specific interval in such a way that the raw water falls drop by drop over the basin and thus get heated and evaporates. The evaporated water is condensed as it encounters the top glass, and thus, the distilled condensate is collected. The corrugated basin is modified by providing well-designed V-shaped corrugations and orienting it in inclined position to increase the surface area of the absorber plate.

Fig. 1 Schematic diagram of the inclined corrugated basin solar still coupled with parabolic trough collector

Performance Analysis of Corrugated Inclined Basin Solar …

11

Fig. 2 Experimental system of conventional and corrugated inclined basin still coupled with parabolic trough collector

3 Experimental Analysis The experimentation setup consists of conventional horizontal basin, corrugated basin, evacuated tube collector, and parabolic trough collector as shown in Fig. 2. Adequate piping arrangement has been done to circulate the heated water from the parabolic trough collector to the basin of solar still. During experiments, the basin temperature of both the solar stills, glass temperature, temperature of inlet water and outlet water from the parabolic trough collector were recorded throughout the period. The productivity of both the solar still was recorded by measuring the distillate output time to time.

4 Computational Analysis Geometry of corrugated inclined basin solar still was created in CATIA simulation software. The geometry file was then imported in ANSYS Fluent. The rectangular meshing was generated by AUTOMESH. Average value of the solar irradiance at the given location and average estimated temperature values were given as input. Glass front, and back wall temperatures were constant. Heat flux, opaque thermal, and semitransparent conditions were implemented on; front, side, and back walls of corrugated absorber plate and glass, respectively. The k − ε turbulence model is the most common model used in CFD to simulate mean flow characteristics for turbulent flow conditions. However, RNG (k − ε) model is selected for analysis as it gives a general description of turbulence by two transport equations. The turbulent kinetic energy Eq. (1) is same as in case of k − ε turbulence model; however, there is an additional term in its ε Eq. (3) which is significant in improving the accuracy for rapidly strained flows.

12

S. Joshi et al.

Equation of turbulent kinetic energy (k)     ∂(ρk) ∂ ρkx j ∂ μt ∂k + 2μt E i j E i j − ρε = + ∂t ∂x j ∂ x j σk ∂ x j

(1)

Equation of rate of dissipation of turbulent kinetic energy (ε) in (k − ε) model   ∂(ρε) ∂(ρεu i ) ε ε ∂ μt ∂k ε2 = − C1ε 2Ciε 2μt E i j E i j − C2ε ρ + ∂t ∂x j ∂ x j σk ∂ x j k k k

(2)

Equation of rate of dissipation of turbulent kinetic energy (ε) in RNG (k − ε) model   ∂(ρε) ∂(ρεu i ) ε ε ∂ μt ∂k ε2 − C1ε 2Ciε 2μt E i j E i j − C2ε ρ − Rε + = ∂t ∂x j ∂ x j σk ∂ x j k k k (3)

4.1 CFD Analysis The temperature contours of the inclined corrugated solar still and glass cover are shown in Fig. 3a, b, respectively. It is observed that the mean temperature of the inclined corrugated basin is 335 K.

Fig. 3 a Temperature contour for corrugated inclined basin solar still, b temperature contour for glass cover

Performance Analysis of Corrugated Inclined Basin Solar …

13

Fig. 4 Average distillate output of conventional and corrugated solar still at a given radiation intensity and time of the day

It is observed that the mean temperature of the glass cover is around 302 K. Based on the CFD analysis, the productivity of the modified solar still is calculated to be equal to 279 ml/h.

5 Result and Discussion 5.1 Experimentation Results Figure 4 shows the variation of average distillate output of conventional and corrugated solar still along with variation in solar radiation intensity with respect to the time of the day. It observed that corrugated inclined solar cell provided higher productivity than the conventional system irrespective of the time of the day. It can be observed that the highest output for conventional and corrugated solar still was found to be 204 ml/h and 232 ml/h, respectively, at 1300 h. The variation of average instantaneous experimental and theoretical efficiencies of conventional and corrugated basin solar still with respect to time is shown in Fig. 5. The efficiency of corrugated solar still was found to be higher, i.e., 32.9% as compared to that of conventional solar still which was just 23.62%.

14

S. Joshi et al.

Fig. 5 Average instantaneous theoretical and experimental efficiency of conventional and inclined corrugated solar still

6 Conclusion The present work focuses on the productivity improvement of the solar still by improving its rate of evaporation. An inclined corrugated basin solar still is designed. The inclined basin is modified as a corrugated channel, and it is coupled with a parabolic trough collector. The heat from the parabolic trough collector is transferred to the modified basin using suitable flow arrangement. Thus, the basin surface receives heat from direct solar radiations as well as from the coupled parabolic trough collector. The performance of the system is simulated using CFD technique. The simulation results show the improved performance of the modified system over conventional one. The CFD results are validated by the experimental analysis. The modified solar distillation system is fabricated. The experiments are performed at Nagpur (21.14° N, 79.08° E) in the month of April and May. The current work, solar still basin modification, and secondary heating source are used to enhance the overall productivity of solar still. Based on the experimental study, it was observed that the productivity of the conventional solar still was found to be 204 ml/h and that by modified solar still was 232 ml/h thus giving a clear indication of productivity improvement. Also, the efficiency of corrugated solar still was found to be higher, i.e., 32.9% as compared to that of conventional solar still which was just 23.62%. The CFD simulation shows good agreement with the experimental results, thus validating the CFD model to carry out further modification in the system.

Performance Analysis of Corrugated Inclined Basin Solar …

15

References 1. Pugsley A et al (2018) Solar distillation potential around the world (Chap. 2). In: Renewable energy powered desalination handbook, Butterworth-Heinemann, Oxford, pp 47–90. https:// doi.org/10.1016/C2017-0-02851-3 2. Panchal H et al (2017) Various methods applied to solar still for enhancement of distillate output. Desalination 415:76–89 3. Fathy M et al (2018) Experimental study on the effect of coupling parabolic trough collector with double slope solar still on its performance. Sol Energy 163:54–61 4. Aybar HS (2005) An experimental study on an inclined solar water distillation system. Desalination 180:285–289 5. Tanaka H et al (2006) Experimental study of a basin type solar still with internal and external reflectors. Desalination 197:205–216 6. Khalifa AJN et al (2013) Effect of inclination of the external reflector of simple solar still in winter. Desalination 264:129–133 7. Omara ZM et al (2013) Enhancing the stepped solar still performance using internal reflectors. Desalination 314:67–72 8. Joshi S et al (2019) Productivity improvement of single using basin inclined solar still using enamel coating and copper chips. Desal Water Treat 156:161–167 9. Shalaby SM et al (2016) An experimental investigation of corrugated absorber solar still with wick and reflectors. Desalination 398:247–255 10. Omara ZM et al (2013) Experimental investigation of corrugated absorber solar still with wick and reflectors. Desalination 318:25–33 11. Dwivedi VK et al (2010) Experimental validation of thermal model of a double slope active solar still under natural convection mode. Desalination 250:49–55 12. Taghvaei H et al (2014) A thorough investigation of the effects of water depth on the performance of active solar stills. Desalination 347:77–85 13. Singh RV et al (2013) Performance of a solar still integrated with evacuated tube collector in natural mode. Desalination 318:25–33 14. Hassan H et al (2018) Experimental work on the effect of saline water medium on the performance of solar still with tracked parabolic trough collector (TPTC). Renew. Energy 135:136–147

Mechanical Design of Omnidirectional Spherical Wall Traversing Robot Yogesh Phalak, Rajeshree Deotalu, Onkar, and Sapan Agrawal

Abstract Since the past few decades, many mechanisms were developed for wallclimbing robots. Robotic systems having omnidirectional surface traversing ability independent of its inclination require complex morphology transformation for a floor to wall transition as well as perfect adhesion on the vertical surface. This paper depicts the development of the mechanical design of the robot producing upthrust due to coaxial antiparallel propeller mechanism to grip the vertical surfaces. To provide the omnidirectional capability, 2-DoF gimbal mechanism is used. Bi-propeller coaxial thruster situated inside the gimbal mechanism is controlled resembling its localization on the wall and also it helps the robot to adhere to the wall. The novel design of this robot is newly aimed and its motion is mathematically modeled. Keywords Omnidirectional · Spherical robot · Mechanical design · Mathematical model

1 Introduction Robots having the ability to climb the vertical surfaces are categorized into wallclimbing robots. Wall climbers have been developed across the globe for surveillance, inspections, flaw detection, cleaning, and maintenance, etc. The proposed mechanism has a 2-DOF bi-propeller plane which allows getting any direction of thrust on to the wall, which helps it to move freely and remain attached to the surface.

Research work of this project is funded and supported by IvLabs VNIT, Nagpur, India. Y. Phalak (B) · R. Deotalu · Onkar Visvesvaraya National Institute of Technology, Nagpur, India e-mail: [email protected] S. Agrawal Worcester Polytechnic Institute, Worcester, MA, USA e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_3

17

18

Y. Phalak et al.

Producing perfect adhesion with the vertical surfaces has been a considerable quandary in wall-climbing robots. Since the past decade, mechanisms based on spiney gripper [1], magnetic [2], elastomer adhesive [3, 4], electromagnetic/electrostatic forces [5], grasping, wheel-driven [6, 7], and suction-vacuum [8, 9] are introduced in the wall climber robots. But their applications are highly limited to a specific purpose. The robots based on electromagnetic forces can climb only on ferromagnetic surfaces [5]. Robots using a vacuum suction mechanism fail to develop sufficient adhesion on dusty and rough walls [2–4, 10–12]. Another grasping-based bio-inspired robot require the design of a complex mechanism and controlled gait for accurate foot placement [10–12]. Hence, a propulsion-based gimbal mechanism is used where required normal reaction for frictional force is provided by the thrust [13]. The choice of two antiparallel propellers instead of one nullifies angular moment of the total system. During motion on the wall, the required magnitude and the direction of the thrust can be calculated. Such magnitude and direction can be adjusted by varying angular velocity and gimbal angles vice versa. The outer spherical shape allows horizontal to vertical surface alteration without the change in morphology.

2 Mechanical Design During floor to wall transition of the wall-climbing robot complex morphology transformation or deformation of the system is required. The choice of the external spherical shape of the robot provides super-symmetry and considerably solves the requirement of morphology change problem (Figs. 1 and 2).

2.1 Bi-propeller Mechanism The basic driving force of the proposed bot is derived from the propeller thruster. The thruster is consists of the coaxial 1045(10 × 4.5”) ABS propellers (1CW + 1CCW 1 pair Black) and a couple of A2212 2200KV BLDC motors. On rotating, CW and CCW propeller pair in an antiparallel direction results in the addition of individual propeller thrust and nullifies angular moment of the system. A2212 2200KV BLDC motors are placed adjacent to each other rather than placing coaxially, which reduces the distance between propellers and hence further dimensions of the model by the reduced size of the gimbal. Coaxial propeller axes are achieved by the gear mechanism with gear ratio 1:1.

Mechanical Design of Omnidirectional Spherical Wall …

19

2.2 2-DOF Gimbal Mechanism 2-DOF gimbal mechanism is used to control the direction of the thrust produced by the propeller mechanism. Two couple of V0150 servo motors are placed in an arrangement of perpendicular axes of rotation. One pair of motors is placed on the ring and it revolves propeller mechanism by an angle θin . Another pair turns the ring consisting of motors and propellers by an angle θout (Fig. 3).

2.3 Outer Shell and Gravitational Stabilization of the Base Plate Gimbal mechanism is gripped in a supporting frame which has a base plate at the bottom. Weights involving battery and the controllers are placed on the base plates to lower the center of gravity of the system. This arrangement automatically steadies the base plate and always keeps it horizontal with respect to ground. Due to gravitational stabilization, horizontal position can be taken as a reference which further simplifies the mathematics (Figs. 4 and 5). An outer shell is kept concentric with the gimbal ring with the support of ball bearings. Bearings provide free rotation of the spherical shell over a surface in contact. Radial perforations are provided to the outer shell for the free circulation of air required for the thrust. But residual surface from perforation causes obstruction to the thrust of propellers. This problem can be solved by choosing the strong material of the shell and increasing amount of perforation.

Fig. 1 Floor to wall transition mechanism comparison of a a multi-linked track robot [12] and b OsWalT

20

Fig. 2 The bi-propeller mechanism

Fig. 3 2-DOF gimbal mechanism

Fig. 4 a Perforated spherical shell. b Final model with spherical shell

Y. Phalak et al.

Mechanical Design of Omnidirectional Spherical Wall …

21

Fig. 5 a Support frame and base plate, b, c, d gravitational stabilization of base plate on horizontal, inclined, and vertical surfaces, respectively

Fig. 6 a CAD model of OsWalT. b Hardware of the first prototype of OsWalT

3 Hardware Design of First Prototype According to the requirement analysis, the hardware of the first prototype of OsWalT is designed to support the functional requirements of the gimbal and thruster. The RC unit is used to manual control of the gimbal and angular speed of the thruster. The RC unit comprises FS-TH9X 2.4 GHz 9CH Upgrade Transmitter with FS-IA10B Receiver. 1045(10 × 4.5”) ABS propellers (1CW + 1CCW-1 pair Black) and a couple of A2212 2200KV BLDC motors are used in the bi-propeller mechanism. Four V0150 servo motors are used to control gimbal mechanism. BLDC motors are controlled by a pair of SimonK 30 A ESCs. Orange 8000 mAh 3 S 30C/60C Lithium polymer battery Pack (LiPo) is used as main power source to the system. Components with a 5 V requirement such as receiver and servo motors are fed by 5 V voltage regulator IC 7805. Chassis and other framework are 3D printed using the PLA+ material (Fig. 6).

4 Results Experiments on the locomotion of fabricated prototype of the proposed OSWalT flat and inclined terrain are performed. The results of the experiment are stated below.

22

Y. Phalak et al.

Fig. 7 Time-lapse images of zero inclination test

Fig. 8 Time-lapse images of inclination test

4.1 Basic Traversing Test on the Ground (Zero Inclination) As a first step, the floor test was performed by manually controlling the prototype as shown in the figure. The angles and the required upthrust were adjusted remotely by the RC module. We found that the bot could traverse require trajectory in zero inclined surfaces. As shown in Fig. 12, the obtained results that are by changing the thrust and angles of the servo, we can achieve the required direction and can move the bot likewise (Fig. 7).

4.2 Inclination Test To perform the inclination test, the bot was placed on a wooden platform. The inclination angle of the platform was increased from 0◦ until the bot slips from its stable position. The servo angles and the upthrust were adjusted manually in order to maintain bot in position. The critical angle is found to be 45◦ . Figure 13 states the experimental setup and time-lapse images of motion tests on different inclinations (Fig. 8).

Mechanical Design of Omnidirectional Spherical Wall …

23

5 Conclusions This paper presented a novel design of a wall-climbing robot. This mechanism serves great improvement in previously proposed models due to the most simplified surface to surface transition mechanism. The spherical shape of the robot provides symmetry to the robot which makes it robust. Propulsion-based driving principal makes bot independent of surface parameters and allows it to work on any terrain. Acknowledgements The research work of this project is supported and funded by IvLabs, Robotics lab of Visvesvaraya National Institute of Technology, Nagpur, India.

References 1. Ma B, Liu R, Zhang R Design of wall climbing robots with transition capability 2. Murphy MP, Sitti M Waalbot: an agile small-scale wall-climbing robot utilizing dry elastomer adhesives 3. Hari P (2008) BIGALLO wall climbing robot. https://prabhuhari.wordpress.com/tag/wallclimbing 4. Sano S et al (2017) Development of wall climbing robot using passive joint and vacuum pad on rough surface. In: 2017 international symposium on micro-nanomechatronics and human science (MHS). Nagoya, pp 1–3 5. Cai J, He K, Fang H, Chen H, Hu S, Zhou W (2017) The design of permanent-magnetic wheeled wall-climbing robot. In: 2017 IEEE international conference on information and automation (ICIA). Macau, pp 604–608 6. Liu G, Liu Y, Wang X, Wu X, Mei T (2016) Design and experiment of a bioinspired wallclimbing robot using spiny grippers. In: 2016 IEEE international conference on mechatronics and automation. Harbin, pp 665–670 7. Schmidt D, Hillenbrand C, Berns K (2011) Omnidirectional locomotion and traction control of the wheel-driven, wall-climbing robot. Cromsci Robot 29:991–1003 8. Singh A, Paigwar A, Manchukanti ST, Saroya M,Maurya M, Chiddarwar S (2017) Design and implementation of Omni-directional spherical modular snake robot (OSMOS). In: 2017 IEEE international conference on mechatronics (ICM). Churchill, VIC, pp 79–84 9. Tanaka Y, Nozaki K, Ioi K (2017) Motion control of a wall climbing robot with coaxial propeller thruster. In: 2017 2nd IEEE international conference on intelligent transportation engineering (ICITE), pp 360–364 10. Chen R, Liu R, Chen J, Zhang J (2013) A gecko inspired wall-climbing robot based on electrostatic adhesion mechanism. In: International conference on robotics and biomimetics, pp 396–401 11. Bin M, Rong L, Rong Z, Tian L, Nashunbuhe (2007) Design of wall climbing robots with transition capability. In: IEEE international conference on robotics and biomimetics, pp 15–18 December. Sanya 12. Lee G, KimH, Seo K, Kim J, Seok Kim H (2015) Robotics and autonomous systems multitrack: a multi-linked track robot with suction adhesion for climbing and transition, vol 72. https://doi. org/10.1016/j.robot.2015.05.011 13. Murphy MP, Sitti M (2007) Waalbot: an agile small-scale wall-climbing robot utilizing dry elastomer adhesives. IEEE/ASME Trans Mechatron 12(3)

Fabrication and Performance Analysis of a Device to Transform Vibration Energy on an Automobile Dheeraj H. Bonde , Nitin K. Panche , Hrishikesh S. Meshram , Vrushabh W. Dhongade , Atul V. Dharmik , Jayesh D. Parate , Mangesh G. Pardhi , and Vinit S. Gupta Abstract In this article, a device design is described to convert vibration energy exerted on a vehicle during motion on different road conditions. This is achieved by converting the vibrational energy to electrical energy by utilizing Faraday’s law. Selection of mounting point is done by using first principle modelling approach for a two-wheeler vehicle as a spring-mass-damper system. Finally, the device is tested at different speeds and road surface conditions. It is concluded that the voltage induced is directly dependent on frequency and amplitude of vibrations. Keywords Vibration energy · Faraday’s law · Spring-mass-damper model · Amplitude and frequency of vibration

1 Introduction Vibration is nothing but an oscillatory motion in any body from its mean position [1]. Mechanical vibration can be clearly observed in every structure such as a vehicle, machine and any other solid construction when subjected to cyclic loads. Vibration generates when unbalanced forces are applied on any equilibrant or non-equilibrant body. Noise and vibrations are most important parameters that are analysed in any automobile design [2]. It is an un-utilized energy of the vehicle subjected to different road and driving conditions [3–5], that reduces comfort level in ride and also increases fuel consumption [6]. These vibrations not only affect a vehicle component as well as influence chassis design but also show hazardous effect on human body parts, i.e. hand joints, nervous system, etc. [7]. Longitudinal and transverse vibration in vehicle generates internally due to the combustion and power transmission system [8]. However, non-uniform road surfaces are again one of the major factors that result in production of random vibration in a vehicle [9, 10]. D. H. Bonde · N. K. Panche · H. S. Meshram · V. W. Dhongade · A. V. Dharmik · J. D. Parate · M. G. Pardhi · V. S. Gupta (B) Department of Mechanical Engineering, S. B. Jain Institute of Technology, Management and Research, Nagpur, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_4

25

26

D. H. Bonde et al.

Energy harvesting devices are created, nowadays, that intend to collect tiny amount of energies and convert it into useful electrical energy [11]. Such devices are playing a major role in increasing the efficiency of a vehicle and reducing the losses of energy in motor-operated hybrid vehicles [12, 13]. Most of the energy harvesting devices are working mainly on the concepts of Faraday’s law and piezoelectric effect [14, 15]. In this work, an attempt is made to study the harvesting of vibration energy and for utilizing it by converting it to other usable forms of energies. For this purpose, a fabrication method is developed using Faraday’s concept. This device is finally tested under experimental road conditions to evaluate the performance. Section 2 focusses on fabrication of device and selection of mounting point, Sect. 3 gives the test results on actual device and finally important conclusions are presented in Sect. 4.

2 Fabrication of Device and Selection of Mounting Point 2.1 Fabrication of Device The fabrication of the device to convert vibration energy to electric energy is carried out using Faraday’s law. It states that whenever there is a change in magnetic flux across a current-carrying conductor, an electromotive force is produced. The device is made to be mounted on a front wheel of a two wheel vehicle parallel to the suspension system.

2.1.1

Mechanical Assembly

The fabrication of mechanical assembly consists of two clamps which are made up of steel and are used to attach the device on the fork of suspension. Attaching link of L shape made of stainless steel is welded from both ends where one end is weld to the circumference of a clamp and another end is weld to the end face of cylinder as shown in Fig. 1. The main function of attaching link is to fix the cylinder parallel to the fork of suspension from the bottom. The piston having one L shape end is weld on the circumference of the clamp and another end is kept free to move inside the cylinder of diameter of 5 cm and length 20 cm which is again made up of stainless steel.

2.1.2

Electrical Assembly

The electrical assembly consists of insulated copper wire of 36 gauge which is wound on a thick cylinder of plastic in turns of 2400 turns having inner diameter 2.5 cm, outer diameter 3.4 cm and length 18 cm as shown in Fig. 1. The important component in

Fabrication and Performance Analysis of a Device …

27

Fig. 1 The combined assembly of a mechanical system and an electrical system of the fabricated device

electrical assembly is a cylindrical magnet made of neodymium material of 2.54 cm diameter and 2.54 cm length. This magnet is two in number; one is attached to the piston while other is fixed inside the cylinder so as to create a repulsive magnetic field. In this assembly, magnet is moving while the windings of insulated copper wire are kept stationary. As seen in Fig. 1, the copper windings cylinder is fixed inside the stainless cylinder with the two end terminal of coil extended and left out. The piston is attached with the help of clamp below the head of two wheel vehicle in such a way that makes to move it between the coils inside the cylinder. To take the output readings, multi-meter is attached at the two end terminal of coil.

2.2 Selection of Mounting Point Mathematical modelling of a suspension of two-wheeler is carried out to study the vertical motion of a vehicle body due to the road surface [16]. The modelling is carried out for both front wheel and back wheel of a two-wheeled vehicle. Important elements in a passive suspension system consist of sprung mass, un-sprung mass, spring and a damper [16]. Un-sprung mass is the mass of the air tight tire with alloy wheel and sprung mass is the mass of the vehicle.

2.2.1

For Front Wheel

The front wheel suspension system of a two-wheeler (test device) consists of two telescopic forks that act as dampers c1 and c2 between sprung mass m1 and un-sprung mass m2 as shown in Fig. 2a:

28

D. H. Bonde et al.

Fig. 2 a Schematic diagram of front wheel shock absorber of a two-wheeled vehicle, b schematic diagram of mass-spring-damper system of back wheel shock absorber of a two-wheeler

In Fig. 2a, c1 and c2 represent coefficient of damping, k 1 represents stiffness of tire, m1 and m2 are un-sprung and sprung mass, respectively, x 1 , x 2 represent vertical displacement of sprung mass and un-sprung mass, respectively, and y(t) represents road excitation which is assumed to be sinusoidal [1]. Also, c1 and c2 include the damping effect of entire telescopic fork system considering also the effect of internal coiled springs. By applying Newton’s second law of motion to mass-spring-damper arrangement as shown in Fig. 2a, equation of motion can be represented by: For sprung mass m1, m 1 x¨1 + (c1 + c2 )(x˙1 − x˙2 ) = 0

(1)

For un-sprung mass m2, m 2 x¨2 + (c1 + c2 ) (x˙2 − x˙1 ) + k1 (x2 − y) = 0

(2)

Putting the values of x˙1 = x3 and x˙2 = x4 we get the following equations, m 1 x˙3 + (c1 + c2 )(x3 − x4 ) = 0

(3)

m 2 x˙4 + (c1 + c2 )(x4 − x3 ) + k1 (x2 − y) = 0

(4)

x˙3 = −

1 [(c1 + c2 )(x3 − x4 )] m1

(5)

Fabrication and Performance Analysis of a Device …

x˙4 = −

1 [(c1 + c2 )(x4 − x3 ) + k1 (x2 − y)] m2

29

(6)

Equations 5 and 6 give the acceleration of the chassis and wheel, respectively. By using these equations, theoretical values of acceleration, velocity and displacement can be calculated.

2.2.2

For Back Wheel

The back wheel of a two-wheeler (test device) consists of two external spring and two dampers attached between sprung mass and un-sprung as shown in Fig. 2b. In Fig. 2b, m1 represents sprung mass, m2 represents the un-sprung mass of vehicle, c1 and c2 represent the damping coefficient of the suspension, k 1 and k 2 represent the stiffness of suspension spring while k 3 represents the stiffness of tire, x 1 and x 2 represent vertical displacement of sprung mass and un-sprung mass, respectively, and y(t) represents road excitation which is assumed to be sinusoidal. By applying Newton’s second law, equation of motion for Fig. 2b can be represented by, For sprung mass: m 1 x¨1 + (k1 + k2 )(x1 − x2 ) + (c1 + c2 )(x˙1 − x˙2 ) = 0

(7)

For un-sprung mass: m 2 x¨2 + k3 (x2 − y) + (k1 + k2 )(x2 − x1 ) + (c1 + c2 )(x˙2 − x˙1 ) = 0

(8)

Putting the values of x˙1 = x3 and x˙2 = x4 in the above equations we get, m 1 x˙3 + (k1 + k2 )(x1 − x2 ) + (c1 + c2 )(x3 − x4 ) = 0 m 2 x˙4 + k3 (x2 − y) + (k1 + k2 )(x2 − x1 ) + (c1 + c2 )(x4 − x3 ) = 0 x˙3 = − x˙4 = −

1 [(k1 + k2 )(x1 − x2 ) + (c1 + c2 )(x3 − x4 )] m1

1 [(c1 + c2 )(x4 − x3 ) + (k1 + k2 )(x2 − x1 ) + k3 (x2 − y)] m2

(9) (10) (11) (12)

The above equation describes the mathematical model of mass-spring-damper system used in both the suspension for front as well as back wheel. By solving these equations, displacement, velocity and acceleration are calculated. Equations 11 and 12 are showcasing the acceleration of chassis and wheel, respectively. For solving the equations, the unknown variables are required. Table 1 gives all the values of mechanical parameters related with front wheel while Table 2 showcases

30 Table 1 Mechanical parameters of front wheel of the vehicle [values are taken from vehicle’s manual of Hero Honda Achiever]

Table 2 Various mechanical parameters of back wheel of the vehicle [values are taken from vehicle’s manual of Hero Honda Achiever]

D. H. Bonde et al. Parameters

Value

Sprung mass, m1

90 kg

Un-sprung mass, m2

14 kg

Shock absorber damping coefficient c1, c2

6000 Ns/m2

Tire stiffness k 1

105,428 N/m

Parameters

Value

Sprung mass, m1

90 kg

Un-sprung mass, m2

16 kg

Shock absorber damping coefficient c1, c2

6000 Ns/m2

Spring stiffness, k 1, k 2

70,000 N/m

Tire stiffness, k 3

131,286 N/m

the values for the back wheel. All the mechanical parameters are found and calculated for the two-wheeled vehicle (Hero Achiever). Above equations are solved using MATLAB for the time range of 0–10 s in the steps of 0.1 s. Figure 3 depicts the displacement versus time plot of both front and back wheel. The displacements are plotted providing a unit step input to the suspension system. The graph clearly shows that the systems represented by the given set of equations have maximum displacement in the front wheel as compared to that of back wheel. Also, as the suspension system is designed to maximize driver’s comfort, it is evident that the vertical displacement of back wheel will always be less than the front wheel.

Fig. 3 Plot of displacement generated in both the wheels (x 2 ) of two-wheeler drive with respect to time

Fabrication and Performance Analysis of a Device …

31

Considering the above result, the device is designed to be mounted on front wheel parallel to the suspension system.

3 Testing of the Device 3.1 Methodology The fabricated device is attached on the selected mounting point of vehicle. The device output is evaluated for stationary condition. In stationary condition, pumping of front suspension is carried out by applying external manual load at certain fixed amplitude and frequencies. The device is also inspected by running the vehicle on road surface at different velocities.

3.2 Testing in Stationary Condition In stationary condition, the device is tested for varying amplitude and frequency. The terminal voltage is recorded for different amplitude. The variation in output voltage for different amplitude of vibrations is shown in Fig. 4a. The graph plotted, i.e. of voltage against amplitude, shows that the induced emf increases with amplitude almost linearly. The graph of voltage against amplitude is showing the trend as linear law and is represented by equation mentioned below: V = 392.7A − 466.6

(13)

Fig. 4 a Representation of terminal voltage against amplitude of vibration on the mounted device, b representation of voltage from the vibration harvester for different frequency of vibration in the front suspension of vehicle

32

D. H. Bonde et al.

where V is the terminal voltage of device and A is the amplitude of vibration produced in the wheel. Figure 4b gives the idea of variation in terminal voltage for the device at different frequency collected from the road surface at a constant amplitude of 3.5 cm. Voltage is showing the linear trend against the frequency of vibration and it is followed by equation given below: V = 212.3F + 307.6

(14)

Here, F is the frequency of vibration from road surface and V is the terminal voltage across device.

3.3 Testing in Running Condition Figure 5 depicts the voltage variation caused due to different speed conditions. The data was collected for constant speed of 20, 40, 60 kmph. The graph shows that there is increment in the induced emf when vehicle is accelerated at different constant speeds. The graph is showing the linear trend represented by the equation: V = 1.038S + 6.375

(15)

Here, S is the speed of the vehicle which is constant and V is the terminal voltage of the device. Fig. 5 Representation of voltage at different constant speeds during motion of vehicle

Fabrication and Performance Analysis of a Device …

33

4 Conclusion The fabrication and performance analysis of a device for converting vibration to electrical energy is discussed. The results clearly indicate that generation of power within the small space is possible. The work also concludes that maximum vibration is obtained on the front wheel of two-wheeler and the magnet should move freely and travel maximum distance within the coil to obtain efficient output. Also the output voltage is directly dependent on frequency and amplitude of vibrations. However, the device can be further improved in order to obtain a higher output. This device can also be developed in a way to replace the ordinary suspension system of vehicle.

References 1. Thomson W (2018) Theory of vibration with applications. CRC Press, Boca Raton 2. Rao MD (2003) Recent applications of viscoelastic damping for noise control in automobiles and commercial airplanes. J Sound Vib 262:457–474 3. Ibrahim IM, Crolla DA, Barton DC (1996) Effect of frame flexibility on the ride vibration of trucks. Comput Struct 58:709–713 4. Lee TK, Byoung SK (2008) Vibration analysis of automobile tire due to bump impact. Appl Acoust 69:473–478 5. Wei CF, Taghavifar H (2017) A novel approach to energy harvesting from vehicle suspension system: half-vehicle model. Energy 134:279–288 6. Abdelkareem MAA, Lin X, Ali MKA, Elagouz A, Mi J, Guo S, Liu Y, Zuo L (2018) Vibration energy harvesting in automotive suspension system: a detailed review. Appl Energy 229:672– 699 7. Mou R, Li H, Yiqiang J, Yong Z, Yongqiao W (2015) Study of automobile suspension system vibration characteristics based on the adaptive control method. Int J Acoust Vibr 20:101–106 8. Dawange SV, Kadlag VL (2015) A review paper on vibration analysis of DI engine. Int J Sci Res 4:759–761 9. Ahirrao NS, Bhosle SP, Nehete DV (2018) Dynamics and vibration measurements in engines. Proc Manuf 20:434–439 10. Burdzik R (2013) Identification of vibration transfer to car-body from road roughness by driving car. Vibroeng Proc 1:27–30 11. Mydeen A, Kelwin I, Venkatesh M, Suthesh C (2017) Hybrid power harvester using engine source. Int J Smart Sens Intell Syst 10:198–211 12. Wang W, Junyi C, Zhang N, Lin J, Liao WH (2017) Magnetic-spring based energy harvesting from human motions: design, modeling and experiments. Energy Convers Manag 132:189–197 13. El-Hami M, Glynne-Jones P, White NM, Hill M, Beeby S, James E, Brown AD, Ross JN (2001) Design and fabrication of a new vibration-based electromechanical power generator. Sens Actuators, A 92:335–342 14. Glynne-Jones P, Tudor MJ, Beeby S, White NM (2004) An electromagnetic, vibration-powered generator for intelligent sensor systems. Sens Actuators, A 110:344–349 15. Siddique AR, Mahmud S, Heyst BV (2015) A comprehensive review on vibration based micro power generators using electromagnetic and piezoelectric transducer mechanisms. Energy Convers Manag 106:728–747 16. Tan WH, Cheah JX, Lam CK, Lim EA, Chuah HG, Khor CY (2017) Vibration analysis on compact car shock absorber. J Phys Conf Ser 908:012025

Robust Backstepping Controller for an Omniwheeled Mobile Robot with Uncertainties and External Disturbances Zeeshan Ul Islam , Saumya Ranjan Sahoo , Mohammad Saad , Uddesh Tople , and Amrapali Khandare

Abstract The application and implementation of portable mobile robots are increasing in industrial and warehouse environments. A robot that can efficiently maneuver in confined spaces is perfectly suited for such a task. One such variety is omniwheeled mobile robots. In the paper, we present the design of a robust control policy for the accurate trajectory tracking of four-wheeled mecanum mobile robot in the presence of external uncertainties and disturbances. In the beginning, the kinematical and the dynamical modeling of the mobile robot are performed. Afterwards, a robust control policy is designed and implemented. The efficacy of the proposed controller was tested on a reference trajectory. Simulation results prove that the proposed controller tracks trajectory with greater accuracy than other standard controllers. Keywords Control methods · Mobile robotics · Robust controller · Trajectory tracing · Backstepping controller

1 Introduction Mobile robot platforms with rollers on the wheel’s periphery constitute a part in the class of omnidirectional mobile robots. The rollers on the wheels leverage an edge to these platforms as they can perform translatory and rotatory motions independently Z. U. Islam (B) · S. R. Sahoo · M. Saad · U. Tople · A. Khandare Visvesvaraya National Institute of Technology, Nagpur, Maharashtra 440010, India e-mail: [email protected] S. R. Sahoo e-mail: [email protected] M. Saad e-mail: [email protected] U. Tople e-mail: [email protected] A. Khandare e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_5

35

36

Z. U. Islam et al.

and synchronously. These robots can steer and maneuver in very confined spaces and hence are well suited for warehouses and factories and workshop floors. One kind of omnidirectional mobile robot is four-wheeled mecanum mobile robot (FWMMR). The wheels of these platforms have rollers aligned at an angle of 45◦ with the axis of the wheel, thus helping it slide in the lateral direction. The kinematic and dynamic modeling of the FWMMR is challenging because of four independent motors controlling the motion. Kinematic [1] and dynamic [2] modeling of mecanum-wheeled mobile robots have been done before [3]; however, the majority of them were torque-based approaches. Here, the modeling is done by considering voltage input to motors. In real world, uncertainty and disturbance will creep in, so its effects have been added to the system. All these factors are added in dynamic equation derived by Newton–Euler approach [4]. The system is having nonlinear dynamics, hence the linear controllers like PID, cannot achieve accurate trajectory tracking. In the situation of bounded but unknown uncertainties, backstepping controller, which is a nonlinear controller, can effectively trace the trajectory. However, it has one disadvantage that it is not robust enough to track the trajectory to a good precision in presence of higher disturbances. This disadvantage is taken care of by adding a switching control law, which effectively and quickly converges the system to the required trajectory. Hence, the proposed controller is a robust backstepping (RBS) controller. In the paper, the above methodology is proposed which is verified against tuned PID controller. The main difference between this work and the previous works [5] done in this domain are (1) The dynamical equation is adjusted for disturbances. (2) The backstepping controller with and without robustness is compared to PID controller in situations having varying uncertainty. (3) Control input voltage approach helps include motor dynamics into consideration. (4) Controller is designed with Lyapunov stability guaranteeing asymptotic stability. The paper has been organized as, Sect. 2 discusses the kinematic modeling of the FWMMR. The dynamic equation is derived in Sect. 3. The proposed controller design is explained in Sect. 4. Simulation results and comparison of backstepping controller with and without robustness with PID is done in Sect. 5. The paper ends in Sect. 6 with the conclusion.

2 Kinematic Modeling The top view of the FWMMR is shown in Fig. 1. For the system analysis, a total of six frames are considered. Owi X wi Ywi are the frames assigned to the wheels, where i represent the wheel number. The frame of motion of robot is Om X m Ym Z m and the world frame is On X n Yn Z n . The angle of roller is represented by γi for the i th wheel, which is +45◦ for first and third wheels and −45◦ for the wheels second and fourth. The position vector is represented as Pm = [xm ym φm ]T and it shows the position of the robot in the Om X m Ym Z m frame, where the positions of the mobile robot is given as xm and ym and φm is the rotational orientation of the mobile robot with the X m

Robust Backstepping Controller for an Omniwheeled Mobile Robot …

37

Fig. 1 Schematical top view of mobile robot

axis. The value of velocity vector P˙ m given as [x˙m y˙m φ˙ m ] in terms of the wheel’s velocity, i.e., θ˙i and wheel’s radius R is. ⎡

⎤ ⎡ −1 x˙m ⎣ y˙m ⎦ = R ⎣ 1 4 1 φ˙ m

d1 +d2

1 1

−1 1

−1 −1 d1 +d2 d1 +d2

⎡ ⎤ θ˙1 1 ⎢θ˙2 1 ⎦⎢ ⎣θ˙3 1 d1 +d2 θ˙4

⎤ ⎥ ⎥ ⎦

(1)

Taking the transformation matrix to relate the frames Om X m Ym Z m and On X n Yn Z n , the velocity vector P˙n = [x˙n y˙n φ˙ n ] in the world frame is given as ⎡ ⎤ ⎤ ⎤ θ˙ ⎡ −C(φ) − S(φ) C(φ) − S(φ) −C(φ) − S(φ) C(φ) − S(φ) ⎢ ˙1 ⎥ x˙n ⎣ y˙n ⎦ = R ⎣−S(φ) + C(φ) S(φ) + C(φ) −S(φ) + C(φ) S(φ) + C(φ)⎦ ⎢θ2 ⎥ ⎣θ˙3 ⎦ 4 1 −1 −1 1 φ˙ n d1 +d2 d1 +d2 d1 +d2 d1 +d2 θ˙4 (2) where C(φ) = cos(φ) and S(φ) = sin(φ). ⎡

3 Dynamic Modeling The free-body diagram of all the forces acting on the system is shown in Fig. 2. The driving forces due to actuator of the i th wheel in X- and Y-directions acting on the system is given as Fxi and Fyi , respectively. The net torque on the entire platform by all the external forces is represented by τ . The external force acting on the system is shown by Fe at an angle of δ with Ym at a distance of b from the front wheel’s axis. Applying the Newton’s second law of translation and rotation, i.e., F = ma , and τ = I α in the robot frame Om X m Ym Z m and using rotation transformation matrix,

38

Z. U. Islam et al.

Fig. 2 Free-body diagram of the FWMMR

we have P˙n = Rmn (φ)P˙m and Fn = Rmn (φ)Fm . Upon solving and rearranging [6], we have ˙ 3X 1 + g(x)3X 4 u(t)4X 1 + h(Fe , φ, δ)3X 1 + ξ(t)3X 1 x(t) ¨ 3X 1 = f (x) where

⎡

(3) ⎤

1 (β S(2φ) y˙ − β S(2φ) y˙ + 2β x˙ + 2C 2 (φ)(β − β ) x˙ + 4K x˙ x n y n y n x y n 1 n ⎢ 2M 1 (−β S(2φ) x˙ + β S(2φ) x˙ + 2β y˙ + 2C 2 (φ)(β − β ) y˙ + 4K y˙ ⎥ f (x) ˙ = ⎣ 2M x n y n x n x y n 1 n ⎦ 1 2 ˙ 2In (−2(d1 + d2 ) K 1 − βz )φ

⎡

⎤ − M1 (Sφ + Cφ) M1 (−Sφ + Cφ) − M1 (Sφ + Cφ) M1 (−Sφ + Cφ) K2 ⎣ 1 (−Sφ + Cφ) M1 (Sφ + Cφ) M1 (−Sφ + Cφ) M1 (Sφ + Cφ) ⎦ g(x) = √ M 2 1 1 (d1 + d2 ) − I1n (d1 + d2 ) − I1n (d1 + d2 ) (d1 + d2 ) In ⎡ I1n ⎤ (2Fe S(φ)S(δ) + 2Fe C(φ)C(δ)) 2M 1 (−2Fe S(φ)C(δ) + 2Fe C(φ)S(δ)) ⎦ ξ(t) = [ξx ξ y ξφ ]T h(Fe , φ, δ) = ⎣ 2M 1 (d1 Fe C(δ) − d2 Fe S(δ) + bFe S(δ)) 2In

where C(φ) = cos(φ), S(φ) = sin(φ), C(δ) = cos(δ), and S(δ) = sin(δ) and K 1 and K 2 are motor coefficients in the driving force by motor which is given as Fdr = K 1 R θ˙ + K 2 u [7] and ξ(t) is the uncertainty in environment.

4 Controller Design The FWMMR system is a multiple input multiple output (MIMO) system, and here, the purpose of the controller is to track the trajectory accurately and with least deviation in presence of external uncertainty and disturbances by optimizing and controlling the various inputs. In backstepping control design [8], a system is first broken down into simpler subsystems and inputs of each subsystem is taken as virtual

Robust Backstepping Controller for an Omniwheeled Mobile Robot …

39

input. Finally, all the virtual control laws are combined by stepping back to generate the final control law, and the stability is guaranteed by Lyapunov function. Let the position vector be defined by x1 = [xn yn φn ]T and velocity vector by x2 = [x˙n y˙n φ˙ n ]T and control inputs as u(t) = [u 1 u 2 u 3 u 4 ]T . The two subsystems are given as (4) x˙1 = x2 x˙2 = f (x) + g(x)u(t)

(5)

here, x2 is taken as the virtual control input. The errors are defined as e1 = [x1 − x1d ] = [xn − xnd yn − ynd φn − φdn ]T and e2 = [x2 − x2d ] = [x˙n − x˙nd y˙n − y˙nd φ˙ n − φ˙ dn ]T . Here, x1d and x2d represent desirable position and velocity, respectively. Taking the Lyapunov function as V (e1 , e2 ) = 21 [e1 ][e1 ]T + 21 [e2 ][e2 ]T and the derivative is V˙ (e1 , e2 ) = [e˙1 ][e1 ]T + [e˙2 ][e2 ]T . To have, V˙ (e1 , e2 ) ≤ 0, we take [e˙1 ] = −K a [e1 ] and [e˙2 ] = −K b [e2 ] where K a and K b are 3 × 3 positive definite matrix. Hence, we have [x˙1 ] = [x˙1d ] − [K a ][e1 ] and [e˙2 ] = [x˙2 ] − [x¨1d ] − [K a ]2 [e1 ]. Finally the backstepping controller equation is given as in Eq. (6) and its switching function is given as in Eq. (7) u(t)bs = g −1 (x)[K a2 [e1 ] + [x¨1d ] − f (x) − h(Fe , φ, δ) − K b [e2 ]]

(6)

u(t)sw = g −1 (x)[K c Sign(e˙1 + λe1 )]

(7)

where, Sign(x) = 1, ifx ≥ 0 and Sign(x) = −1, ifx < 0. Here, ξ(t)max ≤ |K c | and λ is 3 × 3 matrix. The robust backstepping controller is finally given as u(t) = g −1 (x)[K a2 [e1 ] + [x¨1d ] − f (x) − h(Fe , φ, δ) − K b [e2 ]] + g −1 (x)[K c Sign(e˙1 + λe1 )]

(8)

5 Simulation Results To verify the efficacy of the designed controller, we perform a reference nonlinear trajectory tracking with external uncertainties and disturbances. The values of the physical dimensions and parameters of the robot [7] are taken as M = 6 kg, In = 0.0945 kgm2 , d1 = 11 cm d2 = 18 cm b = 10 cm βx = β y = βz = 0.02 K 1 = 0.087 N/V K 2 = −11.4 kg/s. Results of error as deviation from the trajectory are quantified in the form of integral square error (ISE), integral average error(IAE), and integral time average error (ITAE). The equation of trajectory ∀t in (0s, 30s) xd =

3 sin(t) 9 sin(t) cos(t) yd = φd = 0 2 1 + sin (t) 1 + sin2 (t)

(9)

40

Z. U. Islam et al.

Fig. 3 Trajectory tracing plot

and the uncertainty is given as ξx (t) = 15 sin(t) ξ y (t) = 17 sin(t) and ξφ (t) = 13 sin(t) and Fe = 10 N for 4 ≤ t ≤ 6 and δ = 1rad for 4 ≤ t ≤ 6 The parameters of the RBS controllers are obtained after over 200 iterations of parameter estimation toolbox in MATLAB R2016a. ⎡ ⎡ ⎤ ⎤ ⎡ ⎤ 64.06 0 0 128.08 0 0 5.34 0 0 K a = ⎣ 0 70.29 0 ⎦ K b = ⎣ 0 140.04 0 ⎦ λ = ⎣ 0 17.14 0 ⎦ 0 0 60.04 0 0 140.63 0 0 2.6

and K c is a diagonal 3 × 3 matrix with each value = 20 for aggressive convergence. The initial position and orientation of the FWMMR is [xn yn φn ] = [3 0 0]T . Here, the PID controller is also tuned by the parameter estimation toolbox of MATLAB R2016a. The controller gain values obtained for PID are the optimum suited for the situation. From Fig. 3, it can be observed that the RBS controller tracks the trajectory with accuracy greater than PID. PID undergoes a lot of chatter in the last few seconds. However, RBS follows the reference path smoothly. Figure 4 shows the plot of error or deviation from trajectory, RBS is able to track the trajectory with maximum error of 0.12 m, when the external forces applied were of 10 units. The error will asymptotically converge to zero as given by Lyapunov stability. Figure 5, shows the fluctuation of voltage supplied to actuators, RBS follows a step curve with few fluctuations but PID undergoes fluctuation for the entire interval. In practice, this will damage the actuator. Hence, for practical purposes, RBS is a better choice. The errors of RBS, PID, and backstepping are compared in Table 1. In the table, it can be clearly seen that RBS has small error magnitudes of each kind with respect to other standard linear and nonlinear controllers.

Robust Backstepping Controller for an Omniwheeled Mobile Robot …

41

Fig. 4 Plot of error versus time

Fig. 5 Plot of control input versus time Table 1 Tracking performance comparison Controller ISE (m) RBS PID BS

0.1790 0.6227 612.384

IAE (m)

ITAE (m)

2.2546 3.9551 116.373

34.6501 61.1205 1616.351

6 Conclusion In this paper, a robust backstepping controller for FWMMR is developed. The robot platform is given external disturbances and nonlinear uncertainties. Using the backstepping technique of integrating all the virtual inputs of the subsystem, the robust backstepping control law is developed, and stability is verified by Lyapunov method. Lastly, simulations were performed to check the efficacy of the proposed controller on a reference path. The proposed controller outperforms PID in terms of trajectory tracking, error magnitude, and practical usability.

42

Z. U. Islam et al.

References 1. Taheri H, Bing Q, Nurallah G (2015) Kinematic model of a four Mecanum wheeled mobile robot. Int J Comput Appl (0975-8887) 113(3) 2. Williams II et al (2002) Dynamic model with slip for wheeled omni-directional robots. IEEE Trans Robot Autom 3. Huang L, Lim YS, Li D (2004) Teoh CEL. Design and analysis of a four-wheel omnidirectional mobile robot. In: Proceedings of second international conference on autonomous robots and agents, December 2004. Palmerston North, New Zealand, pp 425–428 4. Mittal RK, Nagrath IJ (2015) Robotics and control, Tata Macgraw Hill Publication 5. Hamid RK (2014) Adaptive tracking with external force disturbance rejection for uncertain robotic systems. Int J Control Autom Syst 12(1):169–176 6. Alakshendra V, Chiddarwar S (2016) Adaptive robust control of Mecanum wheeled mobile robot with uncertainties. In: Non linear dynamics, Springer 7. Viet TD, Doan PT, Hung N, Kim HK, Kim SB (2012) Tracking control of a three-wheeled omnidirectional mobile manipulator system with disturbance and friction. J Mech Sci Technol 26(7):2197–2211 8. Behera L, Kar I (2010) Intelligent systems and control: principles and applications, 2nd ed. Oxford University Press

Micro-mechanical Analyses of Particle Reinforced ex situ Bulk Metallic Glass Matrix Composites S. Gouripriya

and Parag Tandaiya

Abstract Three-dimensional finite element simulations of mechanics of deformation and failure of ductile particle reinforced ex situ bulk metallic glass matrix composites (BMGCs) are carried out in the present work. Reinforcements in the form of spherical particles are introduced in a BMG matrix. Appropriate constitutive models are chosen for the ductile reinforcements and the BMG matrix, which are representative of the broad class of respective materials. Finite element simulations are carried out for uniaxial tension and compression loading of ex situ BMGCs having particle volume fraction of 50% using a unit cell method and a multi-particle full model. The underlying deformation and failure mechanism of particle reinforced ex situ BMGCs are elucidated. The results from multi-particle analyses and unit cell analysis are compared with experimental data and are found to be in good agreement with each other, thus, validating the unit cell modeling approach which is further applied for several parametric studies. Keywords Bulk metallic glass composites · Nonlinear finite element analysis · Unit cell method

1 Introduction Ex situ bulk metallic glass matrix composites (BMGCs) are broadly classified as particle or fiber-reinforced BMGCs. Choi Yim and co-workers [1] have conducted experiments on uniaxial compression and tension behavior of niobium (Nb) particle reinforced Vitreloy 106 BMGCs, having particle volume fraction of 50%. The Nb particle reinforced BMGCs show a plastic strain of up to 13% under uniaxial compression and 2.21% under uniaxial tension. In the present work, finite element (FE) analyses of a three-dimensional (3D) multi-particle model and a unit cell model of a particle reinforced ex situ BMGC having Nb particle volume fraction of 50% is S. Gouripriya · P. Tandaiya (B) Department of Mechanical Engineering, Indian Institute of Technology Bombay, Mumbai 400076, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_6

43

44

S. Gouripriya and P. Tandaiya

carried out and it shows a very good agreement with the experiments of Choi Yim et al. The unit cell modeling approach, which is validated by the experiments and the multi-particle simulations, can further be used to study the behavior of several types of ex situ BMGCs depending on the reinforcement shape, volume fraction and hardening behavior.

2 Three-Dimensional Models 2.1 3D Multi-particle Model A 3D FE multi-particle model is described in this section. A set of particles are assumed to be dispersed randomly in the amorphous matrix. The multi-particle model is developed to understand realistic formation of shear bands under uniaxial compression and tension. Figure 1a, b shows the finite element mesh of the matrix and particles, respectively, of the multi-particle model having 30 spherical particles dispersed randomly in a prismatic box with square cross section having an aspect ratio of 0.5:0.5:1. The particle volume fraction is 50%. The element type used in the FE analysis is ABAQUS C3D10M and the mesh consists of 18,245 elements and 26,616 nodes. The boundary conditions are applied to the base nodes such that they

Fig. 1 Finite element mesh of the multi-particle simulation model having particle reinforcement volume fraction f = 50% showing, a the matrix material with nomenclature of the node sets and b the spherical particles

Micro-mechanical Analyses of Particle Reinforced …

45

are restricted in the Y direction. One of the corner nodes at the base is fixed in X, Y and Z directions to prevent rigid body motion of the model. Uniaxial compression and tension loads are applied by displacing the ‘top’ nodes in the Y direction at a quasi-static strain rate of 4 × 10−4 s−1 , corresponding to the experiments of Choi-Yim [1].

2.1.1

3D Unit Cell Model

The 3D FE simulations using unit cell method for ex situ BMGCs are developed to study the uniaxial tension and compression response of ex situ BMGCs. A triply periodic array of spherical particles is assumed to be spaced uniformly in the amorphous matrix. Owing to this assumption, a single unit cell surrounding a reinforcement is sufficient for capturing the response of the entire composite. The geometry of the simulation model developed for spherical particle BMGC having reinforcement volume fraction of 50%, is shown in Fig. 2. The model represents 1/4th of the entire unit cell (exploiting symmetry in the geometry and isotropy in the material properties) of the composite. Node sets named ‘right’, ‘top’ and ‘front’ are created at respective sides of the unit cell as shown in Fig. 2. A set of nodes named as ‘left’, ‘base’ and ‘back’ (which are opposite to the former three node sets) are given symmetry boundary conditions in the X, Y and Z directions, respectively. ‘Right’ and ‘front’ nodes are tied to the ‘tie’ node such that X displacement of ‘right’ and ‘tie’ nodes are the same; similarly, Z displacement of ‘front’ and ‘tie’ nodes are the same. These boundary conditions ensure that the unit cell boundaries remain flat and free of shear traction throughout the analysis. This ensures compatibility between adjacent unit Fig. 2 Geometry of the simulation model of unit cell

46

S. Gouripriya and P. Tandaiya

cells. Uniaxial compressive and tensile loads are applied to the model by displacing the ‘top nodes’ in the Y direction. In both the cases of multi-particle and unit cell analyses, the material of the particles is Nb and that for the matrix is Vitreloy 106. The matrix materials follow the Anand-Su [2] constitutive model. The constitutive equations have been reformulated and implemented using an implicit numerical integration scheme by Tandaiya et al. [3]. In order to model the behavior of Nb, an empirical formula suggested in the work of Zamiri [4] using Hollowman equation σ = K εn is used. The values of parameters in this equation are obtained by fitting the equation to the experimental stress–strain curve of Nb [4]. The material properties for Vitreloy 106 BMG and Nb are taken from literature [1].

3 Results and Discussion Figure 3a, b show the nominal stress versus strain curves of Nb particle reinforced Vitreloy 106 ex situ BMGC, having particle volume fraction 50% under uniaxial compression and tension, respectively. These figures show the simulation results obtained from the multi-particle and the unit cell analysis. A fairly good match is obtained between both the simulation curves and between the experiments in literature [1]. The salient points of the stress–strain curve are marked as A, B and C. Point A corresponds to the initial yielding of the composite. Thus, the material exhibits an elastic behavior till point A. The slope of the stress–strain curves of multi-particle model and the experiments matches exactly until point A. This match with unit cell

Fig. 3 Nominal stress–strain curves of Nb particle reinforced Vitreloy 106 BMGC having f = 50% under uniaxial, a compression and b tension. A comparison between the multi-particle and unit cell analyses is made along with the experiments of Choi-Yim et al. [1]

Micro-mechanical Analyses of Particle Reinforced …

47

model is comparable with the rest of the two curves even though an exact match is not obtained. Further loading imparts nonlinearity to both the curves obtained from multi-particle and the unit cell analysis up to point B. A shift in the stress–strain curve is observed at point B. The unit cell curves deviate from the multi-particle curves beyond elastic limit. This phenomenon is bound to occur since unit cell considers a uniform distribution of particles. Beyond point B, the multi-particle curve shows strain hardening effect, whereas the unit cell curve tends to flatten similar to experiment curves. Though the simulation curves slightly under-predict the nature of the stress–strain curves, it can be noted that a qualitative match is obtained between the three curves. At point C, the unit cell and multi-particle simulation models show a representative failure phenomenon. p Figure 4a, b show the contour plots of ln λ1 in the matrix for unit cell and multiparticle simulation models at point A under uniaxial compression. The contours show that the plastic deformation begins to nucleate at the diametrical center of the sphere where the particle contacts the matrix. As the loading increases, plastic deformation spreads toward the corners of the sphere. At point B, the deformation spreads toward the thin ligaments in the unit cell as seen from Fig. 4c. Similar effects are observed in the multi-particle model where there are criss-cross shear band patterns forming within the matrix as shown in Fig. 4d. On loading beyond B, plastic deformation begins to localize and intensify within the shear bands. At C, the maximum principal logarithmic plastic strain reaches 10% within the shear bands. Figure 4e, f represents this point for unit cell and multi-particle simulation models, respectively. This coincides with an overall strain of about 20% estimated as the failure point in experiments. Thus, under uniaxial compression, for Nb particle reinforced Vitreloy 106, the failure point is represented by the formation of shear bands accumulating a maximum principal logarithmic plastic strain of 10%. Figure 5 shows the evolution of contour plots of maximum principal logarithmic p plastic strain ln λ1 in the BMG matrix for the unit cell and multi-particle models under uniaxial tension. The evolution of plastic strain at points A, B and C is similar as seen in the case of compression. Up to point A, the matrix deforms elastically and it is seen that matrix exhibits elastic deformation as shown in Fig. 5a, b, respectively. Beyond A, as loading increases, plastic deformation nucleates at the center of the sphere at the particle interface. At B, plastic deformation spreads around the spherical interface as seen in the unit cell model (Fig. 5c) and it begins to take shape of crisscross shear bands in the multi-particle model (Fig. 5d). With further loading, the plastic deformation begins to localize as shear bands. The analysis is stopped at the experimental failure strain denoted by point C in the stress–strain curve (Fig. 5b). It is observed that the shear bands form a criss-cross pattern and plastic strain has localized within these shear bands at this failure point. The maximum strain in the model is about 10%, though the plastic strain within the dominant shear band is lower than the maximum value. The failure criterion that holds true under uniaxial compression as mentioned previously does not apply to the case of uniaxial tension. Thus, for a qualitative study presented in the later sections, it is observed that the strain to failure under tension is over predicted.

48

S. Gouripriya and P. Tandaiya

p

Fig. 4 Contour plots of maximum principal logarithmic plastic strain ln λ1 in the matrix at points A, B and C marked on Fig. 3a for a, c, e unit cell, and b, d, f multi-particle model, under uniaxial compression

4 Conclusion The presence of particles in BMG matrix has shown to improve compressive strain to failure in experiments. By adopting a multi-particle model and comparing with the unit cell model and experiments, the present work elucidates the mechanics and mechanisms of deformation and failure of particle reinforced ex situ BMGCs at every stage of their loading process. While monolithic BMGs are known to be brittle, particle reinforced ex situ BMGCs can exhibit enhanced ductility suggesting that these ex situ BMGCs can be effectively deployed in load bearing structural applications.

Micro-mechanical Analyses of Particle Reinforced …

49

p

Fig. 5 Contour plots of maximum principal logarithmic plastic strain ln λ1 in the matrix at points A, B and C marked on Fig. 3b for a, c, e unit cell, and b, d, f multi-particle model, under uniaxial tension

References 1. Choi-Yim H, Conner RD, Szuecs F, Johnson WL (2002) Processing, microstructure and properties of ductile metal particulate reinforced Zr57 Nb5 Al10 Cu15.4 Ni12.6 bulk metallic glass composites. Acta Mater 50(10):2737–2745 2. Anand L, Su C (2005) A theory for amorphous viscoplastic materials undergoing finite deformations, with application to metallic glasses. J Mech Phys Solids 53(6):1362–1396

50

S. Gouripriya and P. Tandaiya

3. Tandaiya P, Ramamurty U, Narasimhan R (2011) On numerical implementation of an isotropic elastic-viscoplastic constitutive model for bulk metallic glasses. Modell Simul Mater Sci Eng 19(1):015002 4. Zamiri A, Pourboghrat F (2007) Characterization and development of an evolutionary yield function for the superconducting niobium sheet. Int J Solids Struct 44:8627–8647

Life Estimation of Circumferentially Notch Round Bars Using J Integral Richa Agrawal , Rashmi Uddanwadiker, and Pramod M. Padole

Abstract Nuclear reactor’s structural components are subjected to high-temperature gradients at the time of shutdowns and start ups. These temperature gradients lead to loading conditions leading to low-cycle fatigue. Additionally, the presence of flaws, defects, and welds results in areas of stress concentration in the components. Therefore, the procedure of estimating the life of such components should consider the effects of stress concentration and temperature. In the present work, the lowcycle fatigue (LCF) life of specimens with circumferential notch is estimated when subjected to strain-controlled loading condition. Notched specimens mimic the multiaxial stress conditions which are results of defects present in the component and LCF conditions are at high temperature mimic the loading conditions due to temperature gradients. In order to study the effect of notches on the life of specimen, LCF were first conducted on plain or smooth specimens, i.e., specimens with notch and then on notched specimens at the same loading condition. The specimens were made of 316 LN austenitic stainless steel and the tests were done in strain control mode at room temperature and at 873 K. LCF loading conditions are conditions when stresses and strains are beyond the elastic limit, and hence, the elasto-plastic fracture mechanics approach was applied for life estimation using principles of fracture mechanics. In the present investigation for fatigue life, the J integral for the geometry is calculated to find out the root stress strain magnitude. The fatigue life is then estimated using the local strain-life method. The predicted life when compared with the experimental results was found to be within a factor of 1.2. Keywords Fatigue · Life · Notch · Fracture

R. Agrawal (B) Pillai College of Engineering, Navi Mumbai 410206, India e-mail: [email protected] R. Uddanwadiker · P. M. Padole Visvesvaraya National Institute of Technology, South Ambazari Rd., Ambazari, Nagpur, Maharashtra 440010, India © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_7

51

52

R. Agrawal et al.

1 Introduction Fatigue leads to almost ninety percent failure of structural components. These components are designed safe for the static load conditions but fail much below the design stress when subjected to cyclic loading conditions. Therefore, life estimation for fatigue becomes necessary to avoid failure during working condition. There are two types of fatigue failure. Fatigue life cycles more than 105 are termed as highcycle fatigue failure and less than 105 as failure due to low-cycle fatigue (LCF). Under LCF failure, the life cycles are less due to high stress and strain conditions. 316 LN is the most favored material [1] for structural component in liquid metal cooled fast breeder reactor (LMFBR) due to its better mechanical properties at high temperature and resistance to corrosive environment. LCF lives of smooth specimens and notched specimens have been predicted at room temperature and at 823 K. Notches are incorporated to study multi-axial state of stress effect present in components due to existence of flaws, welds, and geometrical discontinuities. In the present study, fatigue life is first determined experimentally and then estimated through finite element and fracture mechanics approach.

2 Strain-Life Behavior Basquin [2] stated that stress life data can be modeled using a power relationship, which is a straight line on a log-log plot. This examination is equivalent to elastic material behavior in the strain-life approach. Coffin [3] and Manson [4] separately established that plastic strain-life data can be modeled using a power relationship. The strain-life equation could be formed by adding the elastic and plastic components and is given by Eq. 1. εt =

σ f  b  c 2N f + εf 2N f E

where εt σ f Nf b E εf c

total cyclic strain amplitude regression intercept known as fatigue strength coefficient number of cycles to failure regression slope known as fatigue strength exponent modulus of elasticity regression intercept known as fatigue ductility coefficient regression slope known as fatigue ductility exponent

(1)

Life Estimation of Circumferentially Notch Round Bars …

53

3 Methods of Life Prediction 3.1 Experimentation Local strain-life method is applied on the assumption that the life of the component spent during crack nucleation and growth could be estimated with the help of laboratory specimens subjected to similar cyclic conditions at the site of crack initiation. Fatigue life of a component subject to cyclic loading can be determined using this concept once the relation between the fatigue life and localized strain is known. This strain-life relation is represented as a curve of fatigue life versus strain and is obtained by performing strain-controlled fatigue tests on plain and polished specimens of the material under study. The material under stress concentration due to notches present in it may be under plastic deformation even if the component behaves elastically during cyclic loads, and therefore, strain-controlled axial fatigue tests are recommended in these situations. Low-cycle fatigue tests are conducted on plain and circumferentially notched specimens of 316 LN SS. These tests are performed at room temperature as shown in Fig. 1 and at 823 K and the specimens are subjected to loading conditions resulting in different strain amplitudes. The plain specimens were fabricated as per ASTM standard E606. The notched specimens, as shown in Fig. 2, were of same dimensions as that of plain specimens but with a circumferential circular notch at the center of the gauge length. Fig. 1 Experimental setup for RT tests

54

R. Agrawal et al.

Fig. 2 Notched specimen geometry (all dimension in mm)

1

NODAL SOLUTION STEP=1 SUB =7 TIME=1 EPTOEQV (AVG) DMX =.150266 SMN =.948E-04 SMX =.022947

JAN 24 2018 23:02:03

MN

MX

Y Z .948E-04

.002634

.005173

X

.007712

.010251

.01279

.015329

.017868

.020408

.022947

Fig. 3 Total strain plot at RT for notched geometry with R2.5 mm

The strain-life parameters for the material under study obtained by conducting LCF test on plain specimens are tabulated in Table 1. LCF life can now be estimated by substituting these values in Eq. 1 for the corresponding value of strain. For notched specimens, the strain value will be the maximum strain at the stress concentrated area. Table 1 Strain-life parameters for 316 LN Temperature

Cyclic strength coefficient

Fatigue strength coefficient

Fatigue strength exponent

Fatigue ductility coefficient

Fatigue ductility exponent

E (MPa)

σ f (MPa)

b

εf

c

RT

2×

105

2636

−0.22

0.6501

−0.541

823 K

1.55 × 105

1359

−0.161

0.0845

−0.394

Life Estimation of Circumferentially Notch Round Bars … Table 2 LCF life of notched specimens at room temperature

Table 3 LCF life of notched specimens at 823 K

55

% Strain amplitude

Experimental

FEA

EPFM

±0.4

605

705

576

±0.6

281

318

309

±0.8

137

190

194

% Strain amplitude

Experimental

FEA

EPFM

±0.4

550

184

778

±0.6

218

47

222

±0.8

98

21

123

3.2 Finite Element Analysis The most flexible and commonly used method for estimation of the local stress and strain magnitudes for life prediction is the finite element method (FEM). In LCF condition, as the material is subjected to plastic strain, nonlinear analysis is performed to study LCF behavior of the material. In the present study, rate independent kinematic and isotropic combined hardening nonlinear plastic model is used to calculate the stress and strain values at the notch root using ANSYS. The model was proposed by Chaboche [5, 6] and it uses an associated flow rule with the von Mises yield criterion. The constants incorporated in the model are derived by curve fitting experimental plastic strain and true stress. The bilinear stress versus strain curve describes the bilinear isotropic hardening model used in the finite element analysis. Elastic modulus of the material defines the initial slope of the curve. The user-specified tangent modulus (E T ) defines the slope of the line showing the stress versus total strain beyond the user-specified yield stress when plastic strain develops in the material. Tables 2 and 3 show the life predicted from the maximum total strain obtained by the FEA (as shown in Fig. 3) of the notched specimens at room temperature and at 823 K, respectively. The values show that the predicted life by finite element approach is approximately equal to that obtained by experimentation at room temperature but varies significantly at 823 K.

3.3 Elastic-Plastic Fracture Mechanics LCF life of notched specimens is also estimated by fracture mechanics approach. The notch can be considered as a pre-defined crack. As LCF condition is subjected a high stress and strain condition, elastic-plastic fracture mechanics approach (EPFM) and nonlinear elastic fracture mechanics (LEFM) are applied to predict the notch root stress and strain values. The values thus obtained are then used to estimate the LCF by Eq. 1.

56

R. Agrawal et al.

The stress intensity factor (K I ) is obtained by using Eq. 2, where D = 10 mm and d = 7.07 mm for the geometry similar to that under consideration [7]. The load P is obtained as per the loading condition corresponding to the specific strain amplitude.     D P − 1.27 K I = 3/2 1.72 D d

(2)

Once K I is known, the J integral is obtained using the relation between J and K I and given by Eq. 3 [7]. J=

K I2 E

(3)

Rice [8] proposed a relation to obtain the notch tip strain value at condition when the yield strain is less than the applied strain and was differently given for large and small scale yielding. These relations are represented by Eqs. 4 and 5.   3 J large scale yielding εy + 4 σ y rt    3 σ∞ 2 a ≈ εy 1 + π small scale yielding 4 σy rt

εmax ≈ εmax

(4)

(5)

where εy σy rt a σ∞

yield strain, yield stress, root radius, crack length, and nominal stress

The LCF life is predicted on the basis of the obtained notch root strain and the strain-life equation and is shown in Tables 2 and 3 for the notched specimen with radius 2.5 mm at room temperature and 823 K, respectively. Small scale yielding specified by Eq. 5 is considered for room temperature life estimation and large scale yielding for life estimation at 823 K.

4 Conclusion The various life estimation techniques explained in the present paper were applied for life prediction of circumferentially notched specimens of 316 LN SS specimens. Average ratio of the life predicted by FEA to the experimental life and that by EPFM and experimental life is calculated to judge the level of error in life estimation methods. The life estimated by FE analysis is within a factor of 1.23 at room temperature,

Life Estimation of Circumferentially Notch Round Bars …

57

whereas it underestimated the LCF life at high temperature. This is due to the limitation of completely incorporating the material behavioral changes occurring at high temperature in FE analysis. On the other hand, the life estimation by EPFM approach is within a factor of 1.14 at room temperature and at 823 K. Thus, EPFM approach can be fairly applied for life estimation of notched specimens once J integral is obtained for the geometry under consideration. The variation in the life estimation by EPFM may be due to the use of the stress intensity factor equation available for V notch on round specimen, whereas the specimen used is a round specimen with circular notch.

References 1. Roy SC, Goyal S, Sandhya R, Ray SK (2012) Low cycle fatigue life prediction of 316L(N) stainless steel based on cyclic elasto-plastic response. Nucl Eng Des 253:219–225 2. Coffin LF, Tavernelli JF (1959) The cyclic straining and fatigue of metals. Trans Am Inst Min Metall Pet Eng 215:794–806 3. Manson SS (1966) Thermal stress and low-cycle fatigue. McGraw-Hill, New York, Gr, pp 125–192 4. Basquin OH (1910) The exponential law of endurance tests. Am Soc Test Mater Proc 10:625–630 5. Chaboche JL (1989) Constitutive equations for cyclic plasticity and cyclic viscoplasticity. Int J Plast 5(3):247 6. Chaboche JL (1991) On some modifications of kinematic hardening to improve the description of ratcheting effects. Int J Plast 7(7):661 7. Dieter G (1988) Mechanical metallurgy. McGraw-Hill, London 8. Rice JR, Rosengren GF (1968) Plane strain deformation near a crack tip in a power-law hardening material. J Mech Phys Solid 16:1–12

Placement of Heated Blocks Under Forced Convection for Enhanced Heat Transfer Shankar Durgam , Shakkottai Venkateshan , Thirumalachari Sundararajan , Milankumar Nandgaonkar , Pravin D. Sawarkar , and Aaryan Durgam

Abstract This paper presents numerical and experimental investigation on optimal placement of discrete heated blocks under forced convection. Flowing air velocities of 0.6–1.4 m/s have been used for cooling of heated blocks in a vertical channel. Threedimensional laminar developing flows over-heated blocks, representing integrated circuit components for electronic cooling, have been studied using conjugate heat transfer. Experiments are conducted for FR4 and bakelite as substrate board materials having thermal conductivities of 0.3 and 1.4 W/m K to study the fluid flow and heat transfer characteristics with effects of substrate thermal conductivity. Finite elementbased software is used to solve the coupling between heat transfer in solids and fluid region. The air cooling of substrate boards mounted with heated blocks is modeled and simulated to present heat transport in combination with the fluid flow resulting from the forced air circulation at velocities 0.6–1.4 m/s at constant heat flux values of 1500, 2000, and 2500 W/m2 . The optimal configuration that gives maximum heat dissipation is identified. Experiments indicate a deviation of under 5% with simulations. Keywords Forced convection · Optimal distribution · Vertical channel · Discrete heated blocks · Electronic thermal control

S. Durgam (B) · M. Nandgaonkar College of Engineering, Pune 411005, Maharashtra, India e-mail: [email protected] S. Venkateshan Indian Institute of Information Technology D and M, Chennai 600127, India T. Sundararajan Indian Institute of Technology Madras, Chennai 600036, India P. D. Sawarkar Visvesvaraya National Institute of Technology, Nagpur, Maharashtra 440010, India A. Durgam Indian Institute of Technology, Kharagpur 721302, India © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_8

59

60

S. Durgam et al.

1 Introduction Recent advancement in integrated circuit technology due to microminiaturization and increased chip-level power densities, cooling of electronics becomes challenging. Consistently good in quality or performance of electronic component and its increased life demands efficient cooling methods for heat dissipation from substrate board. Optimal placement of the heat sources and its size plays a vital role to meet the effective cooling scenario. Hajmohammadi et al. [1] presented studies on laminar forced convection cooling of heated blocks and found the optimal configuration to minimize the hot spot temperature of the plate. Juncu [2] studied numerically time dependant conjugate heat transfer from a flat plate and found the influence of aspect ratio and physical property ratio was significant effect on Nusselt number. Gavara and Kanna [3] numerically studied natural cooling of flush-mounted heaters in horizontal channel and showed that high substrate conductivity contributes toward higher heat transfer. Nardini et al. [4] performed natural convection heat transfer in a square channel and found increase in Nusselt number and velocity values. Radhakrishnan et al. [5] showed the results of heat transfer and fluid flow characteristics in enclosure. The authors found that the response surface method was effective in optimizing for a target temperature. Madadi and Balaji [6] studied optimal distribution of heated blocks in a ventilated cavity using ANN and GA. Hotta et al. [7, 8] investigated mixed convection cooling experiments and presented optimal distribution of discrete heated blocks. Muftuoglu and Bilgen [9] determined optimized positions of heat sources in an open square channel under natural convection cooling and found increased Nusselt number and volume flow rate. Da Silva et al. [10] studied natural convection cooling to find optimal placement and heater size in a vertical channel. Hotta et al. [11] conducted experiments under mixed convection and found enhanced heat transfer in optimal placement of heaters. Kumar and Rao [12] studied conjugate mixed convection numerically and found the effect of buoyancy on temperature. Ghorab [13] numerically studied forced convection cooling of heaters in a porous channel and found enhanced heat transfer. Review of above literature shows that many studies are available for electronic cooling but the effect of thermal conductivity along with the optimal placement of heated blocks has not been studied in detail. Therefore, the present work focuses on determining optimal configuration of heated blocks under forced convection and to find the effect of substrate board thermal conductivity.

2 Configurations and Substrate Boards In the present study, 300 combinations obtained by permutation following procedure as given in [12, 14] were simulated for flow velocity of 1 m/s, heat flux value of 1500 W/m2 , and thermal conductivity of 1.4 W/m K. From the simulation results, five representative configurations were selected such that they satisfy the entire range of the combinations. Experiments were conducted with three heat flux values of

Placement of Heated Blocks Under Forced Convection …

FR4_Config 110

61

Bakelite_Config 115

Fig. 1 Substrate boards mounted with heat sources in different configurations

Table 1 Thermophysical properties of substrates at T ∞ = 30 °C and 1 atm

Material

K, W/m K

ρ, kg/m3

Cp, J/(kg K)

FR4

0.3

1900

1370

Bakelite

1.4

1300

1465

1500, 2000, and 2500 W/m2 , three velocities of 0.6, 1, 1.4 m/s and two substrate boards of thermal conductivities of 0.3 and 1.4 W/m K. The five representative configurations are 110, 115, 116, 145, and 189, respectively. Heat sources in the bottom row are assigned numbers 1–5, in the middle row 6–10, and in the top row 11–15 associated with location numbers on the substrate board from 11 to 15 in the first lower row, 21–25 in the middle row, and 31–35 in the top row, respectively. The substrate boards used are glass/epoxy FR4 and bakelite of thickness 5 mm. The fifteen aluminum heat sources are arranged on these substrate boards using five representative configurations. The photographs of FR4 and bakelite substrate boards mounted with aluminum heat sources arranged in configuration 110 and 115 are shown in Fig. 1. Thermophysical properties of FR4 and bakelite are given in Table 1.

3 Experimental Set Up Steady state experiments are conducted under forced convection cooling of fifteen rectangular heated blocks of non-identical sizes, arranged in five different configurations referred to above in a vertical channel. The characteristic length (Lh ) of the heated blocks varies from 10 to 42 mm. The experiments are conducted for a wide range of Reynolds numbers, thermal conductivities, and heat inputs. The temperature

62

S. Durgam et al.

Fig. 2 Photograph of vertical channel with test section

distribution on the substrate board and in each heat source is obtained by conducting experiments. The vertical channel test facility fabricated in-house is shown in Fig. 2. The total number of 90 experiments are conducted (5 configurations × 3 velocities × 3 heat fluxes × 2 substrate conductivities). The 90 cases studied experimentally are simulated for the same ambient temperatures that exist in each case for the purpose of comparison and validation. The detailed procedure of experimentation is given in [13].

4 Numerical Results The pre-processor generated extremely fine mesh is used for the simulation. In the present study, free tetrahedral extremely fine mesh was used for all simulations. Complete mesh consists of 160,404 domain elements, 25,866 boundary elements, and 1564 edge elements, and the number of degrees of freedom solved for 117,898. The Mesh size slightly greater or less than the mesh size used gives variation of temperature of less than 1 °C. The multi-physics used is laminar forced conjugate heat transfer (heat transfer in solids and the fluid) to solve the governing equations [15]. The detailed procedure of grid sensitivity study, error analysis is followed as in reference [12]. The famous Navier–Stokes equations for mass, momentum, and energy for three-dimensional, steady flow in the fluid domain, the solid region energy

Placement of Heated Blocks Under Forced Convection …

63

Fig. 3 Substrate board temperature of FR4 and bakelite in configuration 145

equation, and the body force in vertical direction is given in [14–16] have been solved using simulation software. Boundary conditions are: x = 0, T = T ∞, x = L, p = p∞. Lateral boundaries are adiabatic. The top, bottom boundaries and the edges of substrate and the vertical channel are assumed to be adiabatic. The numerical results of the temperature plots show that the maximum excess temperatures of configurations 110, 115, 116, 145, and 189 are 28.4, 24.4, 27.6, 22.7, and 30.8 °C, respectively. The lowest value of excess temperature, in this scenario, is found in configuration 145. Figure 3 shows the substrate temperature of FR4 board is higher than that for bakelite. The substrate temperature values are the lowest in configuration 145 compared to all other four configurations. To compare the effects of substrate thermal conductivity on maximum excess temperature, the configuration 145 is considered as it is found to be the optimal one. The maximum substrate temperature observed in FR4 is about 68 °C and that in case of bakelite is about 66 °C. The plums of heat can be observed in the direction of airflow.

5 Comparison of Numerical and Experimental Results Figure 4 shows the numerical and experimental results of forced convection cooling of heated blocks on FR4 ad bakelite substrate board using flow velocity of 0.6 m/s and heat flux value of 1500 W/m2 for different configurations.

6 Conclusions Steady state forced convection experimental and numerical results for fifteen rectangular protruding discrete heat sources showed that maximum excess temperature in

64

S. Durgam et al.

Fig. 4 Comparison of experimental and simulation results

configuration number 145 is lowest in all cases considered in the present study. The maximum excess temperature found to be lowest at higher value of inlet air velocity considering bakelite substrate board. High heat fluxes show higher maximum excess temperatures. From this study, it is found that position or location of heated blocks has a significant effect in heat dissipation. The configuration number 145 was found to be the optimal configuration. Results show that excess temperatures in case of bakelite are lower compared with FR4. From the results, it is concluded that maximum temperature of any configuration is a function of Reynolds number, substrate thermal conductivity, heat flux, position, and size of heated blocks.

References 1. Hajmohammadi MR, Shirani E, Salimpour M, Campo A (2012) Constructal placement of unequal heat sources on a plate cooled by laminar forced convection. Int J Thermal Sci 60(4):13– 22 2. Juncu G (2008) Unsteady conjugate forced convection heat/mass transfer from a finite flat plate. Int J Thermal Sci 47:972–984 3. Gavara M, Kanna PR (2013) Three-dimensional study of natural convection in a horizontal channel with discrete heaters on one of its vertical walls. Heat Transf Eng 35(14–15):1235–1245 4. Nardini G, Paroncini M, Carvaro F (2013) Effect of heat transfer on natural convection in a square cavity with two source pairs. Heat Transf Eng 35(9):875–886 5. Radhakrishnan T, Balaji C, Venkateshan S (2010) Optimization of multiple heaters in a vented enclosure—a combined numerical and experimental study. Int J Thermal Sci 49:721–732 6. Madadi R, Balaji C (2008) Optimization of the location of multiple discrete heat sources in a ventilated cavity using artificial neural networks and micro genetic algorithm. Int. J. Heat Mass Transfer. 51:2299–2312 7. Hotta T, Venkateshan S (2015) Optimal distribution of discrete heat sources under natural convection using ANN-GA based technique. Heat Transf Eng 36:200–211 8. Hotta T, Balaji C, Venkateshan S (2015) Experimental driven Ann-GA based technique for optimal distribution of discrete heat sources under mixed convection. Exp Heat Transf 28(3):298–315 9. Muftuoglu A, Bilgen E (2008) Natural convection in an open square cavity with discrete heaters at their optimal positions. Int J Thermal Sci 47:369–377

Placement of Heated Blocks Under Forced Convection …

65

10. da Silva A, Lorenzini G, Bejan A (2005) Distribution of heat sources in a vertical open channels with natural convection. Int J Heat Mass Transf 48:1462–1469 11. Hotta T, Balaji C, Venkateshan S (2014) Optimal distribution of discrete heat sources under mixed convection—a heuristic approach. J Heat Transf 136:104503 12. Kumar G, Rao C (2011) Interaction of surface radiation with conjugate mixed convection from a vertical plate with multiple non-identical discrete heat sources. Chem Eng Comm 198:692–710 13. Ghorab M (2014) Forced convection analysis of discrete heated porous convergent channel. Heat Transf Eng 36(9):829–846 14. Durgam S, Venkateshan S, Sundararajan T (2017) Experimental and numerical investigations on optimal distribution of heat source array under natural and forced convection in a horizontal channel. Int J Thermal Sci 115:125–138 15. Durgam S, Venkateshan S, Sundararajan T (2018) A novel concept of discrete heat source array with dummy components cooled by forced convection in a vertical channel. Appl Therm Eng 129:979–994 16. Patankar S (1980) Numerical heat transfer and fluid flow. McGraw-Hill, New York

Analysis of Track Vibration for Metro Rail Chaitanya V. Bhore , Atul B. Andhare , Pramod M. Padole , and Mayur D. Korde

Abstract Buildings located near surface Metro and surface trains are subjected to mechanical vibrations induced by their motion. High-precision instruments in laboratories nowadays are very much sensitive to external vibrations which affect their performance. It is thus imperative to measure amount of vibration and noise transmitted from the railway track to ground. In this paper, frequency content and amplitude of vibration levels are measured on the metro rail track. Track vibrations were measured at a point on track for three different positions of metro from that point, while it was in motion. Amplitude and frequency content were found when (a) train was some distance away from the point of observation, (b) train was passing above the point and (c) train had left the point by some distance. This experiment was carried out near Khapri metro station, Nagpur, Maharashtra, India. These experimental results are studied and discussed in this paper. Further, these results can be compared with limits given by Federal Transit Administration (FTA) to ensure human comfort and safety of nearby structures. Keywords Metro rail · Accelerometer · Track · Vibration

1 Introduction Due to industrialisation, their is exponential increase in population in Nagpur city, therefore, their is a need to create new transport facilities. The best transport facility nowadays is metro rail. Metro rail has proved its worth in many major cities of India such as Delhi, Bangalore and Chennai by providing ease of travelling, better conC. V. Bhore (B) · A. B. Andhare · P. M. Padole · M. D. Korde Department of Mechanical Engineering, Visvesvaraya National Institute of Technology, South Ambazari Rd., Nagpur, Maharashtra 440010, India e-mail: [email protected] URL: http://vnit.ac.in/ A. B. Andhare e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_9

67

68

C. V. Bhore et al.

nectivity, time saving and indirectly reducing traffic on roads. Therefore, metro rail is being constructed in Nagpur, which is the third largest city in Maharashtra, India, with population of around 2.5 million. Nagpur metro design consists of 1435 mm gauge and maximum permissible speed of 80 km/h. Maximum acceleration and maximum speed of train according to design are 1 m/s2 and 90 km/h respectively. Train composition consists of three-car train (DMC-TC-DMC). DMC and TC stand for driving motor car and trailer car, respectively. Due to ease of access and real estates affordable prices, construction of buildings and other facilities are often pushed in the vicinity of subway and surface trains. Surface trains and subways induced vibrations in the nearby structures and buildings hamper the operation of equipment which are highly sensitive to vibration. Vibration induced to the ground by motion of metro acts as line source of vibration and lies in the range of 10–400 Hz [1]. Hassan [1] studied and showed that vibrations produced by subway and metro trains propagate through the ground in different ways. There are three different ways in which the vibration propagates, (1) surface waves, (2) compressive wave and (3) shear wave. Adam and Estroff [2] studied and showed that vibration transmitted to the structure through foundation and the waves propagated close to soil surface are Rayleigh waves. Anderson [3] measured vibration in two different buildings near the vicinity of railway track which were subjected to vibration at the foundation induced by motion of railway. He found that vibration that is perceptible and can propagate through ground lies in 5–50 Hz frequency domain. Sanayei et al. [4, 5] studied vibration propagated through ground in different areas of Boston, USA, which was induced by train and subway motion. Transmission of vibration in four-storey building through the bottom floor was also reported. Such studies help designers to design and eliminate vibrations in areas in the building which are highly sensitive to vibration. This study aims to measure and analyse vibration on metro rail track. These vibration from track propagate through ground and hence act as base excitation to nearby buildings and overhead structures. Vibrations induced by train on track are measured and analysed. These findings provide basis for base excitation, which is necessary in designing of vibration mitigation models and hence minimising the induced vibration.

2 Measurement Set-Up The measurements were carried out with the help of OROS data acquisition and vibration analyser. It consists of DAQ, accelerometer and a software. It was carried out on track near Khapri metro station, Nagpur, Maharashtra, India. This portion of metro rail is on ground which makes it easy to measure track vibration. Figure 1 shows schematic diagram of arrangement of various components used for measurement of track vibration. Accelerometer was placed on track to measure the vertical vibration of the track when the train is moving. The accelerometer

Analysis of Track Vibration for Metro Rail

69

Fig. 1 Schematic diagram of measurement set-up

is in contact with the metro track with the help of strong magnet attached to its base. Position of the accelerometer was such that its axis was perpendicular to the track surface which helps in getting vibration in vertical direction. Figure 1 shows the schematic of the measurement set-up and Fig. 2 shows the actual measurement set-up. As seen in Fig. 1 accelerometer is connected to DAQ system. For frequency analysis, Fast Fourier Transform (FFT) was used. The measurements were taken for 5 s interval when the train was in motion. Data were acquired in three different conditions: when the train is away and approaching the measurement point, train, when the train is passing over the measurement point, and when the train has crossed the measurement point. Figure 2 shows actual measurement set-up for track vibration. The sampling frequency was 52.1 kHz and number of data points measured were 256,000.

3 Results and Discussion Results were obtained through OROS software in terms of acceleration (m/s2 ) versus time (s) graph. Waveform shows track vibration amplitude with time. As the observed/measured data are in time domain, it is necessary to convert it to frequency domain for ease of analysis. This conversion to frequency domain helps us to pinpoint the maximum amplitude for its respective frequency, which is done using Fast Fourier Transform (FFT) algorithm. Time domain and frequency domain signals for three different positions of train with respect to measurement point are discussed below.

70

C. V. Bhore et al.

Fig. 2 Measurement set-up for measuring vibration of the track

Fig. 3 Spectrum of signal when the train is away and approaching the measurement point

3.1 When Train Is Away and Approaching Measurement Point Figure 3 shows spectrum of signal when the train is away and approaching the measurement point. When the train is away from the point of observation maximum amplitude of acceleration is 525.0 (mm/s2 ) at a frequency of 1600 Hz. Other significant amplitudes corresponding to their frequencies are shown in Table 1.

Analysis of Track Vibration for Metro Rail

71

Table 1 Frequency and amplitude of vibration when train is away and approaching the measurement point Frequency (Hz) Acceleration (mm/s2 ) Frequency (Hz) Acceleration (mm/s2 ) 100 1050 1200

299.4 198.3 183.3

1600 1800 2000

525.0 229.4 160.2

Fig. 4 Spectrum of signal when the train is passing over the measurement point Table 2 Frequency and amplitude of vibration when train is passing over the measurement point Frequency Acceleration Frequency Acceleration Frequency Acceleration (Hz) (mm/s2 ) (Hz) (mm/s2 ) (Hz) (mm/s2 ) 200 1150 1600

663 1137 1094

2000 4050 4800

1860 861 516

5700 7250 8100

499 480 2117

3.2 When Train Is Passing over the Measurement Point Figure 4 shows the spectrum of signal obtained when the train is passing over the measurement point. When the train is passing over the observation point for period of 5 s, maximum amplitude is 2117 (mm/s2 ) at a frequency of 8100 Hz. Other significant amplitudes corresponding to their frequencies are shown in Table 2.

72

C. V. Bhore et al.

Fig. 5 Spectrum of signal when the train has crossed the measurement point Table 3 Frequency and amplitude of vibration when train has crossed the measurement point Frequency Acceleration Frequency Acceleration Frequency Acceleration (Hz) (mm/s2 ) (Hz) (mm/s2 ) (Hz) (mm/s2 ) 1750 2000 4050 5750

131.4 181.6 64.5 122.6

6800 7250 7650 7950

121.6 235.1 363.5 390.5

8700 11,050 11,450

67.3 185.3 120.5

3.3 When Train Has Crossed the Measurement Point Figure 5 shows the spectrum of signal obtained when the train has crossed the measurement point. When the train has crossed the measurement point, for period of 5 s, maximum amplitude is 390.5 (mm/s2 ) at a frequency of 7950 Hz. Other significant amplitudes corresponding to their frequencies are shown in Table 3.

4 Conclusion Design of nearby buildings and overhead structures requires vibrational analysis as the vibration induced through train motion transmitted through ground to their base. As the metro rail in Nagpur is yet to start in its full capacity, therefore, its speed in the

Analysis of Track Vibration for Metro Rail

73

trial period lies in the range of 25–30 km/h. Vibration induced due to train motion to the ground in the adjacent to its track was measured and studied. Maximum acceleration amplitudes corresponding to its frequencies were found out. This study helps in providing base excitation for vibration mitigation system to design safe and disturbance free structures. Further, these results may be compared with limits provided by Federal Transit Administration (FTA) for human comfort and safety.

References 1. Hassan OAB (2006) Train-induced ground-borne vibration and noise in buildings. MultiScience 2. Adam M, Von Estorff O (2005) Reduction of train-induced building vibrations by using open and filled trenches. Comput Struct 83(1):11–24 3. Anderson DC (1994) Engineering prediction of railway vibration transmitted in buildings. Environ Eng 7(1) 4. Sanayei M, Maurya P, Moore JA (2013) Measurement of building foundation and ground-borne vibrations due to surface trains and subways. Eng Struct 53:102–111 5. Ngamkhanong C, Kaewunruen S (2018) The effect of ground borne vibrations from high speed train on overhead line equipment (OHLE) structure considering soil-structure interaction. Sci Total Environ 627:934–941

Localization of a Four-Wheeled Omnidirectional Mobile Robot Using Sensor Data: A Kalman Filter Approach Saumya Ranjan Sahoo , Shital S. Chiddarwar , Mohsin Dalvi , and M. R. Rahul

Abstract In this paper, a model-based method to estimate the position of an omnidirectional mobile is proposed using a Kalman filter. The state variables of a fourwheeled omnidirectional mobile robot are estimated and are used for a feedback loop for trajectory control of the robot. Due to noise in the sensor measurement, the signal cannot directly be used for feedback closed-loop control. The system control without filter mode is not favourable as it produces significant tracking error and leads to high fluctuation in control input. The proposed method is demonstrated using numerical simulations on an omnidirectional mobile platform. Results are presented for different measurement sets. Effects of measurement noise level, filter parameters and modelling error (process noise covariance) are also presented, and it is observed that the position parameter estimation is robust with respect to measurement noise. Keywords Mecanum-wheeled mobile robot · Kalman filter · Flatness-based control · State estimation · Trajectory tracking

1 Introduction Mecanum-wheeled mobile robots are widely used in many applications for its special wheel structure allows for lateral movement. The rotation of the rollers around the mecanum wheel will result in slippage of the robot, and it is difficult to model [1]. Accurate localization is very crucial for navigation of autonomous mobile robot in any environment. The gap between two adjacent rollers causes the periodic vibration of the moving platform. Due to slippage, vibration and noise in the sensor, an accumulation of error can exist using vision-based tracking system. The problem of estimation is widely solved by Kalman filter method which was invented in 1960 by Kalman [2].

S. R. Sahoo (B) · S. S. Chiddarwar · M. Dalvi · M. R. Rahul Visvesvaraya National Institute of Technology, South Ambazari Rd., Ambazari, Nagpur, Maharashtra 440010, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_10

75

76

S. R. Sahoo et al.

Kalman filter reduces the error in least square sense from the measurement from different sensors. Among many applications, the Kalman filter is the most essential part of a robotic vision system. It uses the vision measurements that contain noise and uncertainty by the visual sensors over time and produce value that is close to the true measurement value of the robot (self-localization) [3, 4] or the target (object detection) [4]. Object tracking is used to detect the motion of an object by accruing input from a series of images. The vision system is used to track 3D object [5], people [6] and some planner contour [7] by vision sensor. In the present work, a Kalman filter is developed for a four-wheeled omnidirectional mecanum-wheeled mobile robot. The proposed algorithm is tested for the different level of sensor noise data. The effect of sensor noise on trajectory tracking and control command is illustrated.

2 Dynamics of the Robot In this part, referring to [8], the dynamic model of the four-wheeled mecanum robot is introduced. The robot consists of four independently actuated mecanum wheel over a robotic platform. Figure 1 shows a four-wheeled mecanum robot. The dynamic model of the robot is given as (Eqs. 1–4). τ1w (cos θ − sin θ ) + τ2w (cos θ + sin θ ) + τ3w (cos θ − sin θ ) +τ4w (cos θ − sin θ ) − x˙cq (2Rβcx cos2 θ + 2Rβcy cos2 θ ) − y˙cq (2Rβcx cos θ sin θ − 2Rβcy cos θ sin θ ) 2m R

x¨cq =

(1)

τ1w (cos θ + sin θ ) + τ2 (− cos θ + sin θ ) + τ3w (cos θ + sin θ ) +τ4w − (cos θ + sin θ ) − y˙cq (2Rβcy cos2 θ + 2Rβcx cos2 θ ) y¨cq =

Fig. 1 Four-wheeled mecanum robot

−x˙cq (2Rβcx cos θ sin θ − 2Rβcy cos θ sin θ ) 2m R

(2)

Localization of a Four-Wheeled Omnidirectional Mobile …

θ¨ =

(τ1w − τ2w − τ3w + τ4w ) βcv θ˙ (a + b) − (2I R) I

V = La

di dw + Ra i + kb ω; Jm = T − bm ω − Tl dt dt

77

(3) (4)

where m: mass of the robot, I: inertia of the robot, βcx , βcy , βcω : viscous friction along longitudinal, lateral and yaw direction.

3 Flatness-Based Controller The dynamic model of the omnidirectional mobile robot is differentially flat, and a set of flat output is defined [9]. Based on these flat outputs, the differential flatness controller is defined as ⎤⎡ ⎤ ⎡ ⎤ ⎡ d⎤ ⎡ y¨1 k p1 0 0 e y1 y¨1 ⎣ y¨2 ⎦ = ⎣ y¨ d ⎦ + ⎣ 0 k p2 0 ⎦⎣ e y2 ⎦ 2 y¨3 y¨3d e y3 0 0 k p3 ⎡ ⎤ ⎡ ⎤⎡  ⎤⎡ ⎤ ki1 0 0 kd1 0 0 e˙ y1  e y1 dt + ⎣ 0 ki2 0 ⎦⎣ e y2 dt ⎦ + ⎣ 0 kd2 0 ⎦⎣ e˙ y2 ⎦ (5)  e˙ y3 0 0 ki3 e y3 dt 0 0 kd3 ⎤ ⎡ e y1   where k p 3×3 , [ki ]3×3 , [kd ]3×3 are the controller gain parameters and ⎣ e y2 ⎦ = e y3 ⎤ ⎡ d y1 − y1     ⎣ y d − y2 ⎦ is the error vector. y1 y2 y3 T = x y θ T are the flat output of the 2 y3d − y3 differentially flat system. The proposed differential flatness controller is used for trajectory tracking of the robot.

4 Kalman Filter for Improved Sensing and Control In real systems, trajectory tracking to obtain the desired variables for the feedback loop relies on sensors data. Sensors are prone to noise due to their internal structure and inherent environmental conditions, and their data cannot be used directly for measurement. The Kalman filter is a multi-input multi-output filter for optimal realtime state estimation in systems with noisy measurements. As the filter is recursive, it does not require all the previous sensor data to be maintained, thus simplifying its hardware implementation. The Kalman filter can estimate current and upcoming

78

S. R. Sahoo et al.

states even when accurate models and sensor data are unavailable. In this study, we employed the Kalman filter to timely estimate the state for feedback control by linear difference equations of the omnidirectional mobile robot.

4.1 Kinematic Model for Kinect-Based Vision System The vision system based on the Kinect sensor (Microsoft Corp., Redmond, WA, USA) can determine the position of the robot from its centre of gravity using camera data. The system equation with process noise and sensor noise can be modelled as x˙ = Ax + Bu + w, y˙ = C x + Du + v where w and v are the process and measurement noise following zero-mean Gaussian distributions, respectively. Based on this model, a linear stochastic system is defined for the implementation of the Kalman filter as xˆt = At xˆt−1 + Bt u t + wt , yt = Ct xt + Dt u t + vt ,

(6)

wk ≈ N (0, Q k ) and vk ≈ N (0, Rk ).

(7)

where

At is an n × n matrix that describes the state evolution from t − 1 to t in the absence of control input and noise, Bt is an n × l matrix that describes the effect of control ut on the state from t − 1 to t, Ct is an n × n matrix that relates state x t to observation yt , and Dt is an n × l matrix that describes the effect of control ut on the measurement from t − 1 to t. Process noise wt is assumed to be independent and Gaussian distributed with covariance Q k , and measurement noise vt is assumed to be independent and Gaussian distributed with covariance Rk . For the omnidirectional mobile robot considered in this study, the system matrices are defined as ⎡ ⎡ T 2 ⎤ ⎤ 0 0 1 0 0 T 0 0 2 ⎢ 0 1 0 0 T 0 ⎥ ⎢ 0 T 2 0 ⎥ ⎢ ⎢ ⎥ ⎥ 2 2 ⎥ ⎢ ⎢ ⎥ ⎢ 0 0 1 0 0 T ⎥ ⎢ 0 0 T2 ⎥ At = ⎢ ⎥, Bt = ⎢ ⎥, ⎢0 0 0 1 0 0 ⎥ ⎢ T 0 0 ⎥ ⎢ ⎢ ⎥ ⎥ ⎣0 0 0 0 1 0 ⎦ ⎣ 0 T 0 ⎦ 000 0 0 1 0 0 T

Localization of a Four-Wheeled Omnidirectional Mobile …

⎡

100 ⎢0 1 0 ⎢ ⎢ ⎢0 0 0 Ct = ⎢ ⎢0 0 0 ⎢ ⎣0 0 0 000 ⎡ 2 σq x 0 0 ⎢ 0 σ2 0 qy ⎢ ⎢ 0 0 σ2 ⎢ qt wt = ⎢ ⎢ 0 0 0 ⎢ ⎣ 0 0 0 0 0 0

0 0 0 0 0 0

0 0 0 0 0 0

0 0 0 σq2x˙ 0 0

⎡ ⎤ ⎤ 0 0000 ⎢0 0 0 0⎥ 0⎥ ⎢ ⎥ ⎥ ⎢ ⎥ ⎥ 0⎥ ⎢0 0 0 0⎥ ⎥, Dt = ⎢ ⎥ ⎢0 0 0 0⎥ 0⎥ ⎢ ⎥ ⎥ ⎣0 0 0 0⎦ 0⎦ 0 0000 ⎤ ⎡ 2 0 0 σr x 0 ⎢ 0 σ2 0 0 ⎥ ⎥ ⎢ ry ⎢ 0 0 ⎥ ⎥ ⎢ 0 0 ⎥, v = ⎢ 0 0 ⎥ t ⎢ 0 0 ⎥ ⎢ ⎣ 0 0 σq2y˙ 0 ⎦ 2 0 0 0 σq t˙

79

0 0 σr2t 0 0 0

0 0 0 σr2x˙ 0 0

0 0 0 0 σr2y˙ 0

⎤ 0 0 ⎥ ⎥ ⎥ 0 ⎥ ⎥ 0 ⎥ ⎥ 0 ⎦ σr2t˙

Equation (7) provides the information of the predicted state in the next time step provided that the current system state is available. Hence, this model can be used in the prediction step of the Kalman filter.

4.2 Kalman Filter Algorithm for Localization The Kalman filter consists of the three processes illustrated in Fig. 2. The filter includes a prediction or prior estimation of the state, the use of measurement data and calculation of Kalman gain and the estimation update or posterior estimation. Estimation is the first step of the algorithm. It provides the evolution of the system state matrix (xˆt− ) and the process covariance matrix (Pt− ) over time if the state (xˆt−1 ) at

Fig. 2 Kalman filter algorithm

80

S. R. Sahoo et al.

the previous time step is known. This is the prior estimation (xˆt−1 ) of the state. Then, the vision system measurements are considered, and the Kalman gain calculated. Specifically, the measurement data are collected from the sensor (yt ) and, along with the measurement covariance matrix, the Kalman gain (kt ) is calculated. Finally, the state is updated, and the process covariance matrix calculated. The updated value of the system state (xˆt ) and the process covariance matrix (Pt ) are calculated based on the Kalman gain (K t ). The algorithm implementing the proposed Kalman filter for the omnidirectional mobile robot is detailed as follows: 1. 2. 3. 4. 5. 6. 7. 8. 9. 10.

Algorithm Kalman Filter (xˆ t−1 , u t , Pt−1 , yt ): Prediction: xˆt− = A xˆ t−1 + B u t + wt : Expected state x Pt− = A Pt−1 A T + Q k : Expected covariance P Correction: − K t = H PP−t HH+R = (H Pt− H + R)−1 Pt− H : Calculation of Kalman gain t xˆt = xˆt− + K t (yt − H xˆt− ): Correction of state x Pt = (1 − K t )Pt− : Correction of covariance P Return (xˆt Pt ): Return corrected x and P Go to step 3

5 Simulation Results The proposed dynamic model, control law and Kalman filter described in Sects. 2– 4 were implemented for simulation using MATLAB Simulink (Mathworks, Inc., Natick, MA, USA). To test the effectiveness of the proposed method, we added various levels of noise to the feedback signals during simulations.

5.1 Tracking Results on Circular Trajectory Case 1: wk ≈ N (0, 0.001) and vk ≈ N (0, 0.01) The Kalman filter was applied to mitigate this noise, and the resulting state estimation with and without the estimator for the x- and y-axis robot positions is shown in Fig. 3a, b. Figure 3c shows the controller tracking performance regarding the sensor data and state estimation through the Kalman filter. The controller accurately produced the input command for correct measurements. Case 2: wk ≈ N (0, 0.001) and vk ≈ N (0, 1). The Kalman filter algorithm is also tested under high level of noise in the measurement. From the state estimation (Fig. 4a, b) and trajectory tracking Fig. 4c, the Kalman filter is able to track the desired trajectory under high level of noise.

Localization of a Four-Wheeled Omnidirectional Mobile …

(a) x-axis robot position

(b) y-axis robot position

81

(c) Tracked Trajectory

Fig. 3 Estimated states during circular trajectory under low noise level

(a) x-axis robot position

(b) y-axis robot position

(c) Tracked Trajectory

Fig. 4 Estimated states during circular trajectory under high noise level

6 Conclusion We use the Kalman filter to fuse position information from a vision sensor and encoders and embed this estimation into a flatness-based controller. This control method was implemented for trajectory tracking along prescribed trajectories on an omnidirectional robot. We observed that the substantial amount of noise in the measured signals coming from the Kinect sensor causes high fluctuation of the control input. Such input may result in catastrophic damage to the actuators. However, state estimation using the Kalman can efficiently mitigate noise. This proposed method can be used for simultaneous localization and mapping of the robot.

References 1. Pin FG, Killough SM (1994) A new family of omnidirectional and holonomic wheeled platforms for mobile robots. IEEE Trans Robot Autom 10:480–489 2. Kalman RE (1960) A new approach to linear filtering and prediction problems. J Basic Eng 82:35–45

82

S. R. Sahoo et al.

3. Bishop G, Welch G (2001) An introduction to the Kalman filter. Proc of SIGGRAPH, Course. 8:59 4. Correa J, Soto A (2010) Active visual perception for mobile robot localization. J Intell Rob Syst 58:339–354 5. Tesli´c L, Škrjanc I, Klanˇcar G (2011) EKF-based localization of a wheeled mobile robot in structured environments. J Intell Rob Syst 62:187–203 6. Weinstein AJ, Moore KL (2010) Pose estimation of Ackerman steering vehicles for outdoors autonomous navigation 7. Rae A, Basir O (2009) Reducing multipath effects in vehicle localization by fusing GPS with machine vision 8. Sahoo SR, Chiddarwar SS, Alakshendra V (2017) Intuitive dynamic modeling and flatness-based nonlinear control of a mobile robot. Simulation 0037549717741192 9. Sahoo SR, Chiddarwar SS (2019) Flatness-based control scheme for hardware-in-the-loop simulations of omnidirectional mobile robot. Simulation 0037549719859064

Capacitated Vehicle Routing Problem with Interval Type-2 Fuzzy Demands V. P. Singh

and Kirti Sharma

Abstract Capacitated vehicle routing problem is extension of vehicle routing Problem where the purpose is to determine a set of routes to fulfil all the transportation requests with a given fleet of vehicles in such a way that the travelling cost comes out to be a minimum. In this paper, the demands of the customers will be very well known in advance, but they are known imprecisely. To deal with such kind of impreciseness, interval type-2 fuzzy number (IT2FN) can be used. Therefore, a capacitated vehicle routing problem with interval type-2 fuzzy number has been modelled. A procedure to solve the model has been introduced using Clark and Wright algorithm. Keywords Vehicle routing problem · Interval type-2 fuzzy numbers · Clark and Wright algorithm

1 Introduction The very first work regarding the vehicle routing problem was done in 1959, approximately 60 years before, by Dantzig and Ramser [3] who introduced the vehicle routing problem, then called as truck dispatching problem as a real-world application concerning the delivery of gasoline to gas stations. In the paper published in 1959 by Dantzig and Ramser, the authors proposed the first mathematical programming formulation and algorithmic approach for the vehicle routing problem. In 1964, Clark and Wright proposed [1] an effective greedy heuristic for the approximate solution of the vehicle routing problem. If the transportation tasks are to be performed at the vertices of the network, the corresponding

Supported by Organization VNIT Nagpur. V. P. Singh (B) · K. Sharma Visvesvaraya National Institute of Technology, South Ambazari Rd., Nagpur, Maharashtra 440010, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_11

83

84

V. P. Singh and K. Sharma

problem is known as node routing problem [7], and if the tasks are to be performed on street segments, then the corresponding routing problem is known as arc routing problem [2]. The imprecision may occur in various forms, e.g. imprecise demands of customers. This paper deals with the imprecise demands of the customers. The capacity of the vehicle is crisply defined, but the demands of the customers are defined in the form of an interval type-2 triangular fuzzy number. Inspired by [5, 6], we have considered CVRP with IT2FN. The vehicle routing problem is a NP-hard problem, and introduction of uncertainty gives a new variation of the problem. Since VRP itself is a NP-hard problem, it is not surprising that the very first numerical methods for solving VRPSD consisted of heuristics. These relatively simple techniques based on savings were later complemented by more efficient algorithms such as the Tabu search. This paper is structured in the following manner: in Sect. 2 of this paper, basic preliminaries of fuzzy set theory have been reviewed. In Sect. 3, a mathematical model of CVRP with IT2FN has been presented. Sections 4 and 5 deal with the algorithm of the proposed model and the procedure of the method used. In Sect. 6, an example has been presented with seven customers and a depot node where the demands of the customers are given by interval type-2 triangular fuzzy number. The last section comprises the concluding remarks.

2 Preliminaries and Concepts Definition 1 Uncertainty in the primary membership of a type-2 fuzzy set A˜˜ consists of a bounded region called the footprint of uncertainty, that is, FOU = ∪x∈X Ix [4]. The footprint of uncertainty of A˜˜ is bounded below by a lower membership function (LMF), usually denoted by μ A˜ , and is bounded above by an upper membership function (UMF), usually denoted by μ A˜ [4]. Definition 2 The interval type-2 fuzzy set is called the interval type-2 fuzzy number [4] when its UMF and LMF are type-1 fuzzy numbers. ˜˜ = Definition 3 The core of the interval type-2 fuzzy number A˜˜ = [A, A] is Cor e( A) [min Cor e(A), max Cor e(A)] where φ = Cor e(A) ⊆ Cor e(A) when φ is an empty ˜˜ set and μ A˜ and μ A˜ are the lower and upper membership functions of A, respectively [4]. ˜˜ is defined on the interval [a, c], and Definition 4 Interval type-2 fuzzy number, A, its lower membership function takes the value equal to h ∈ [0, 1] at b and h ∈ [0, 1] at b, respectively, where a ≤ a ≤ b = b ≤ c ≤ c.

Capacitated Vehicle Routing Problem with Interval …

85

Fig. 1 A type-2 triangular fuzzy number

˜˜ is notated as A˜˜ = (A, A) = [4] Thus, interval type-2 triangular fuzzy number, A, ((a, b, c), (a, b, c)). ˜˜ is A graphical interpretation of the interval type-2 triangular fuzzy number, A, shown in Fig. 1.

3 Mathematical Model Before formulating the mathematical model, we state some important assumptions which will be followed throughout the paper. It has been assumed that the goods to be delivered are not divisible in nature and the demands of the customers are given in units and not in weights. The paper assumes that the demands of the customers are independent; i.e. the demand of a customer is not influenced by other customers’ demand. The fleet K = {1, 2, 3, . . . , |K |} is assumed to be homogeneous; i.e. all the |K | vehicles available at the depot have the same carrying capacity Q > 0, and all of them operate at identical costs. While moving from vertex i to vertex j, the travel cost incurred is denoted by ci j . It has also been assumed that the demand of any customer is greater than the vehicle capacity since it does not make any sense to accept a customer who cannot be serviced successfully on an average. It has also been assumed that the cost matrix is symmetric; i.e. the cost of travelling between any two vertices is always independent of direction of traversal. Overall, a capacitated vehicle routing problem is uniquely defined by a complete weighted graph G = (V, E, ci j , qi ) together with the size |K | of the vehicle fleet and vehicle capacity Q. A route (or tour) is a sequence r = (i 0 , i 1 , i 2 . . . , i s , i s+1 ) with i 0 = i s+1 = 0 in which the setS = {i 1 , i 2 . . . , i s } ⊆ N of the customers is visited. The route r has the cost c(r ) = sp=0 ci p i p+1 .

86

V. P. Singh and K. Sharma

 A route r is said to be feasible if it satisfies the capacity constraint, i.e. q(S) := i∈S qi ≤ Q, and no customer is visited more than once. In such cases, one says that S ⊆ N is a feasible cluster. A solution to a CVRP consists of |K | feasible routes, one for each vehicle. The routes r1 , r2 , . . . rk and the corresponding clusters S1 , S2 , . . . Sk provide a feasible solution to a CVRP if all the routes are feasible and cluster forms a partition of N . Thus, the CVRP consists of two interdependent tasks 1. The partitioning of the customer set N into feasible clusters S1 , S2 , . . . S|K | . 2. The routing of each vehicle k ∈ K through {0} ∪ SK . The objective 2 requires the solution of the travelling salesman problem over {0} ∪ Sk . Both the objectives are interrelated, since the cost of cluster depends on routing and the routing needs clusters as an input. Let P be the set of the feasible CVRP routes. Each route p ∈ P is of the form  p p = (i 0 , i 1 , i 2 . . . , i p , i p+1 ), i 0 = i p+1 = 0, so that we can assign the cost c p = j=0 ci j i j+1 

Minimize

cpλp

(1)

∀i ∈ N

(2)

p∈P

such that



αi p λ p = 1,

p∈P



λ p ≥ ϑ(N )

(3)

p∈P

λ p ∈ {0, 1}

∀ p in P

(4)

In the given mathematical model, αi p is a binary parameter whose value is 1 if the route p visits the node i and 0 otherwise. λ p is a binary decision variable whose value is 1 if the route p is visited and 0 otherwise. In the algorithm, different symbols are used with different descriptions, e.g. 1. 2. 3. 4. 5. 6.

C[n][n] D[n] crdemand[i] S[n-1][n-1] Ci Q

Cost matrix. Demand of customer as IT2TFN. Defuzzy demand of customer i. Half savings matrix. Cost of route i Capacity of vehicle.

Capacitated Vehicle Routing Problem with Interval …

87

4 Algorithm

Input: C[n][n], d[n], Output: C.

Q.

1: Start. 2: Input C[n][n] and D[n] 3: i ← 1 4: for i ≤ n do 5: crdemand[i] ← defuzzy value of demand using PGMIR expression. 6: end for 7: i ← 1 8: j ← i + 1 9: for i ≤ n do 10: for j ≤ n do 11: s[i][j] ← c[0][i]+c[0][j]-c[i][j] 12: end for 13: end for 14: i ← 1 15: for i ≤ n do 16: ri ← (0 → i → 0) 17: end for 18: m[k] ← Sorted(s[i][j]). %Sorted(s[i][j]) here means sorted in descending order. 19: k ← 0 20: c ← 0 %Choose ri and r j corresponding to first entry of m[k]. 21: while m = φ do 22: if qri + qr j ≤ Q and ri = r j and (i = ri [1] or j = r j [1]) then 23: qri ← qri + qr j 24: ri ← (0 → ri → r j → 0) 25: m[k] ← Sorted(s[i][j]-m[k]) 26: c ← c + cri + cr j 27: end if 28: end while 29: Return c.

5 Example To illustrate the given model, let us suppose an instance where there are seven different customers having their demands in the form of an interval type-2 triangular fuzzy numbers. The network corresponding to the customers and distributor is given in Fig. 2. The vertex node 0 represents the depot node. The adjacency matrix storing the cost incurred in travelling any two vertices is given by C. The demand data of all the customers is given in Table 1. According to the algorithm, the first task we need to do is to defuzzify the demands using parametric graded mean integration representation method; i.e. if A˜˜ = (A, A) = ((a, b, c), (a, b, c)) is an interval type-2 triangular fuzzy number, then applying PGMIR on it gives P A˜˜ =

1 1 (a + c + a + c) + (b + b) 12 3

88

V. P. Singh and K. Sharma

∞ 4 8 3 9 6 5 11

4 ∞ 9 6 10 3 4 6

8 9 ∞ 8 4 8 12 5

3 6 8 ∞ 9 8 7 3

9 10 4 9 ∞ 11 12 2

6 3 8 8 11 ∞ 8 15

5 4 12 7 12 8 ∞ 12

11 6 5 3 2 15 12 ∞

Fig. 2 Cost matrix and network Table 1 Demands of customers

Customer

Demands

1 2 3 4 5 6 7

((2, 3, 4), (1, 3, 5)) ((4, 6, 7), (3, 6, 10)) ((10, 12, 14), (9, 12, 15)) ((8, 9, 10), (7, 9, 12)) ((9, 12, 15), (8, 12, 16)) ((5, 6, 7), (4, 6, 8)) ((3, 6, 9), (2, 6, 10))

Consider the demand of distributor to be 0 units and the capacity of vehicle to be 25 units. Now, we construct the half savings matrix by using the formula given in the algorithm, i.e. s[i][ j] = c[0][i] + c[0][ j] − c[i][ j] The next highest demand corresponds to the second row and seventh column. In case, we join vertex 7 and vertex 2, the new cycle formed will be 0-4-7-2-0. But this will happen only if all the conditions given in Step 5 are satisfied. Since crdemand[4]+crdemand[7]+crdemand[2]=21 < 25 which means that the capacity constraint is still satisfied. Here also, both 7 and 2 belong to two different cycles and they are, respectively, the last and first customers. So, the new route is 0-4-7-2-0. Continuing in the same way, there exist three different routes, namely 0-6-1-50, 0-4-7-2-0 and 0-3-0 travelling, which never violate the constraints and fulfil the requirement of travelling salesperson of visiting every node exactly once. The cost of traversal of these routes comes out to be 48 units.

Capacitated Vehicle Routing Problem with Interval …

89

6 Conclusion In this paper, we have attempted to solve capacitated vehicle routing problem when the demands of the customers are more imprecise. In this paper, the demands of the customers are given by interval type-2 triangular fuzzy number. This paper dealt with defuzzifying the customers’ demand by using PGMIR and then finding the feasible routes with the help of Clark and Wright algorithm. In this paper, it has been observed that the cost comes out to be a crisp number.

References 1. Clarke G, Wright JW (1964) Scheduling of vehicles from a central depot to a number of delivery points. Oper Res 12(4):568–581 2. Corberán Á, Laporte G (2013) Arc routing: problems, methods, and applications. SIAM 3. Dantzig GB, Ramser JH (1959) The truck dispatching problem. Manage Sci 6(1):80–91 4. Javanmard M, Nehi HM (2019) A solving method for fuzzy linear programming problem with interval type-2 fuzzy numbers. Int J Fuzzy Syst 21(3):882–891 5. Singh VP, Chakraborty D (2017) Solving bi-level programming problem with fuzzy random variable coefficients. J Intell Fuzzy Syst 32(1):521–528 6. Singh V, Chakraborty D (2015) A dynamic programming algorithm for solving bi-objective fuzzy knapsack problem. In: Mathematics and computing. Springer, Berlin, pp 289–306 7. Toth P, Vigo D (2002) The vehicle routing problem. SIAM

Kinematic, Dynamic and Stiffness Analysis of an Asymmetric 2PRP-PPR Planar Parallel Manipulator Deep Singh , Rutupurna Choudhury , and Yogesh Singh

Abstract This paper presents an assessment of the comparison of three-degreeof-freedom 2PRP-PPR planar parallel robotic manipulator (x, y, θz motion stage) with other standard planar parallel manipulators such as 3PPR U-base, 3PPR delta ()-base and 3RRR in respect of static structural stiffness, optimal kinematic design and dynamic performances. Adams/View, a multibody dynamics software, has been utilized to analyze the kinematic and dynamic performance of the motion stage. Analysis of static stiffness has been performed and compared by the joint space Jacobian method along with the matrix structural analysis method. Also, static stiffness was verified through NASTRAN, a standard finite element software. The findings of numerical simulation conclude that the 3PPR U-base configuration and the proposed 2PRP-PPR manipulator possess a number of favorable optimum design characteristics such as good isotropy, better manipulability, better dynamic performances (power, energy), higher stiffness and singularity-free workspace as compared to other manipulators. Note: P and R refer to prismatic and revolute, respectively. Keywords Planar parallel manipulator · Kinematic analysis · Isotropy · Manipulability · Structural stiffness

1 Introduction 1.1 Planar Parallel Manipulators Many scientists have extensively researched parallel configured manipulators that address their benefits over serial manipulators such as higher stiffness, higher payload capacity, higher precision and accuracy etc. [1]. On the contrary, the significant drawbacks related to parallel manipulators are complicated forward kinematic analysis, decreased workspace and control difficulties (due to coupled dynamics). Countless D. Singh (B) · R. Choudhury · Y. Singh Department of Mechanical Engineering, NIT Silchar, Assam, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_12

91

92

D. Singh et al.

applications such as micro-machining, fabrication, material handling and processing, planar parallel manipulators (PPMs) demands proper analysis for efficient design [2]. Analysis of singularity in manipulator’s workspace is significant as it impacts the controller efficiency and optimal design aspects [1]. Stiffness is directly linked to the manipulator’s rigidity and accuracy. It is a function of mechanism geometry, end-effector pose and function of internal forces which may apply considerable instability and low position accuracy problems [3]. Structural stiffness impacts the manipulator’s speed, feed and capacity [4].

1.2 Contribution in the Present Paper This paper proposes a 2PRP-PPR manipulator and addresses the comparison of kinematic optimal design aspects, dynamic parameters, singularity-free workspace and stiffness analysis with respect to 3RRR (revolute-revolute-revolute), 3PPR (prismatic-prismatic-revolute) -base, 3PPR U-base PPMs. The proposed 2PRPPPR PPM consists of one leg with PPR configuration along x-direction and two other legs with PRP configuration along y-direction. For the estimation and comparison of performance indices, Adams/View and MATLAB were used. The second part is associated with the static stiffness analysis and its comparison. This paper is aimed at developing a method to formulate stiffness model of the PPMs. To validate and verify, a finite element analysis (FEA) software MSC-NASTRAN is used. These methods can further be adopted for the kinematic, dynamic and stiffness analysis of similar PPMs.

2 Jacobian of Planar Parallel Manipulators The joint space velocities are mapped to the task space velocities by the Jacobian matrix (J) of the manipulator as shown in Table 1 for the above-mentioned PPMs (Fig. 1).

3 Performance Analysis The numerical simulations were carried out using the Jacobian matrices to obtain the performance index and workspace for the mentioned PPMs. Horizontal (x-direction) and vertical (y-direction) span range is bounded to 1 m by 0.866 m (equilateral triangle base in plane) for 3RRR and 3PPR delta shape base PPM and 1 m by 1 m (square) for 3PPR U shape base and proposed PPM. The active prismatic actuators (si ) displace within the interval [0, 1] meters. The rotation of the revolute joints (θi ) of 3RRR PPM limits within the interval [0°, 180°]. The shape of the mobile platform

2PRP-PPR [7] ⎡ 1 0 ⎢ 1 − tan θ J =⎢ z ⎣ − tan θz 1

and J = Jq−1 Jx

⎥ sin(α + θz ) ⎥ ⎦ sin(α − θz )

0

⎤ −x cos2 θz (s−x) cos2 θz

⎥ ⎥ ⎦

where a and h are side length and height of an equilateral triangle based mobile platform respectively, α is the internal

T angle which is 60° in case of 3PPR PPM and x y θz represents the task space vector

3PPR U-base [6] ⎤ ⎡ 2h cos θ 10 z ⎢ h 3 ⎥ a ⎥ J =⎢ ⎣ 0 1 − 3 sin θz + 2 cos θz ⎦ 0 1 − h3 sin θz − a2 cos θz

where Di j = f (x, y, θz ) Q i j = f (x, y, θz )

cos α sin α

a 2 a 2

h cos θz

⎤

3PPR-base [6] ⎡ 1 0 ⎢ J =⎢ − cos α sin α ⎣

3RRR [5] ⎤ ⎡ D11 0 0 ⎥ ⎢ ⎥ Jq = ⎢ ⎣ 0 D22 0 ⎦ 0 0 D33 ⎤ ⎡ Q Q Q ⎢ 11 12 13 ⎥ ⎥ Jx = ⎢ ⎣ Q 21 Q 22 Q 23 ⎦ Q 31 Q 32 Q 33

Table 1 Jacobian of planar parallel manipulators

Kinematic, Dynamic and Stiffness Analysis of an Asymmetric … 93

94

D. Singh et al.

Fig. 1 Planar parallel manipulators frame diagram: (a) 3RRR, (b) 3PPR -base, (c) 3PPR U-base, (d) proposed 2PRP-PPR

Fig. 2 Workspace of different manipulators

has been considered as an equilateral triangle with 160 mm side length. It is assumed that the masses of actuators and links are negligible. For the above-mentioned PPMs, the mass of the mobile platform is assumed to be 10 kg. Common task space vector assigned for the workspace is: x = 10–825 mm, y = 10–800 mm, θz = 30°.

3.1 Workspace The analysis of singularity-free workspaces is based on the condition |J | = 0. Figure 2 depicts that the 2PRP-PPR and 3PPR U-base PPMs obtained larger workspace as compared to the other manipulators due to the singularity-free behavior.

3.2 Performance Index ◦

Due to the presence of singularity at θz > ±45 in case of the proposed 2PRPPPR PPM, a range of safe orientation angle (−45° to +45°) has been selected for comparison. For all the mentioned PPMs, performance parameters such as isotropy and manipulability were analyzed in Adams/View and compared. Figure 3(a) and 3(b) depicts ◦ that manipulability is lower at ±45 and higher at 0° orientation angle for all the

Kinematic, Dynamic and Stiffness Analysis of an Asymmetric …

95

Fig. 3 Kinematic and dynamic performance, (a) manipulability, (b) isotropy, (c) power consumption, (d) energy requirement

PPMs. Isotropy also shows similar behavior. 3PPR U-base and the proposed 2PRPPPR PPM possess better performance indices which mean higher manipulability, isotropy and lower resistivity as compared to other selected manipulators. Energy and power consumption were researched, analyzed and compared revealing data on the actuator choice. The simulation was performed by Adams/View software. The basic assumptions considered for the analysis are listed below: 1. The mobile platform is rigid. 2. The manipulator components follow Hooke’s law. 3. The manipulator was considered to be present in a gravity-less space. The following common singular-free end-effector/task space circular trajectory is given to the manipulator: x = 500 + 50 × sin(0.12t); y = 500 + 50 × cos(0.12t); θz = 30◦

(1)

where t refers to simulation time. Figure 3(c) and 3(d) depicts that the power and energy requirement are less for 3PPR U-base PPM and the proposed 2PRP-PPR PPM in comparison to others. Hence, the proposed 2PRP-PPR PPM and 3PPR U-base PPM possess superior performance as a motion-based platform.

4 Stiffness Analysis This paper relates the end-effector torques and forces to the corresponding angular and linear joint displacement vectors by utilizing the joint space Jacobian stiffness matrix in addition to matrix structural analysis (MSA) matrix. The basic assumptions taken are listed below: 1. The mobile platform is rigid. 2. Weights of the manipulator components are negligible. 3. The joints are non-frictional and experience negligible deformation.

96

D. Singh et al.

4.1 Stiffness of the PPMs Structure The PPMs are divided into nine numbers of space beam elements with each node having forces and moments in specific directions as depicted in Fig. 4.

4.2 Simulation Results on Stiffness In this section, the proposed PPM is compared with other considered manipulators. The manipulator’s physical and geometrical parameters are listed in Table 2. The task space variables (x, y, θz ) for the manipulators 3RRR and 3PPR  base are taken as (500, 577, 0°–45°), and for the manipulators, 3PPR U-base and 2PRPPPR are taken as (500, 500, 0°–45°). The horizontal span is bounded as 1000 mm, and the orientation varies from 0° to 45° for singularity-free workspace. To demonstrate, the stiffness analysis of the system has been performed using  MATLAB and NASTRAN (FEA) software and compared. Five sets Fx , Fy , Mθz of end-effector force (in N) and bending moment (in N-mm) taken for the analysis are (10, 50, 50), (20, 60, 60), (30, 70, 70), (40, 80, 80) and (50, 90, 90). The end-effector

Fig. 4 FEA model of the manipulators

Table 2 Parameters of various manipulators Parameters

RRR

3PPR -base

3PPR U-base

2PRP-PPR

Link length, li (in mm)

300 (i = 1 to 6)

800 (i = 1 to 3)

800 (i = 1 to 3)

1000 (i = 1 to 2)

Link cross-sectional area, Ac (in mm2 )

2500

2500

2500

2500

Side length (in mm) of end-effector (equilateral triangle)

160

160

160

160

Moment of Inertia, Iz (in mm4 )

520 × 103

520 × 103

520 × 103

520 × 103

Range of si (in mm) for i = 1 to 3

–

50–950

50–950

50–950

Elasticity of steel, E (in N/mm2 )

2.1 × 103

2.1 × 103

2.1 × 103

2.1 × 103

Kinematic, Dynamic and Stiffness Analysis of an Asymmetric …

97

deflection analysis has been performed  in three directions on the plane, (1) x-axis deflection (δx ), (2) y-axis deflection δ y , (3) angular deformation (δθz ). As the simulation of end-effector deformation for multiple cases is timeconsuming, the first set of forces and moments (10 N, 50 N, 50 N-mm) has been considered for simulation in FEA software (NASTRAN) with orientation angle θz = 30° for all the manipulators. Comparison result concludes a few points as mentioned below: 1. The estimated analytical findings match well with those of NASTRAN software which verified the model of theoretical stiffness. 2. In the case of 3-PPR U-base and proposed manipulator, the end-effector deflections (δx , δ y and δθz ) are lower. 3. Stiffness of the manipulators will be less in case of higher orientation angle. 4. Stiffness in y-direction will be lower for all the manipulators as compared to other directions. But, the stiffness in case of 3PPR U-base and proposed manipulator will be higher in all planar direction. This paper presents the NASTRAN model in Fig. 5 for a specific orientation angle 30° as remeshing costs a lot of computation time and is very tedious. The comparison result only for the proposed PPM is listed in Table 3 due to page constraint. NASTRAN results are close to analytical. This study showed that the end-effector deflection is lower for 2PRP-PPR and 3PPR U-base PPM as compared to others. As deflection and stiffness are directly related, it can be concluded that 2PRPPPR and 3PPR U-base PPMs will possess higher stiffness as compared to other manipulators.

Fig. 5 FEA result of PPMs, (a) 3RRR, (b) 3PPR -base PPM, (c) 3PPR U-base, (d) 2PRP-PPR

98

D. Singh et al.

Table 3 Comparative result of 2PRP-PPR end-effector for the first set of loading condition (10 N, 50 N, 50 N-mm) with 30° constant orientation angle Deflections

Analytical

FEA software (NASTRAN)

Difference

δx (m)

0.004173e−06

0.005733e−06

0.00155e−06

δ y (m)

0.0463e−03

0.0523e−03

6.0e−06

δθz (rad)

0.000514e−03

0.000763e−03

2.49e−07

5 Conclusions The following are a few significant points observed from the present study: a. The proposed 2PRP-PPR PPM possesses less singularity in the workspace due to the partially coupled kinematic relationship. b. 2PRP-PPR and 3PPR U-base PPMs exhibits favorable outcomes of optimal design and performance index as it possesses lesser singularity workspace. c. Due to structural simplicity and avoidance of interference of links, 3PPR Ubase and the proposed 2PRP-PPR PPM require minimum power and energy to perform the operation/trajectory as compared to other PPMs. d. The comparison showed that the stiffness of 3PPR U-base and 2PRP-PPR PPM is higher as investigated and validated through NASTRAN. Acknowledgements Authors acknowledge TEQIP-III under National Institute of Technology, Silchar, for financial support.

References 1. Gosselin C, Angeles J (1988) The optimum kinematic design of a planar three-degree-of-freedom parallel manipulator. ASME J Mech Trans Autom 110:35–41 2. Alvarado JG, Campos JHT (2019) A parallel manipulator with planar configurable platform and three end-effectors. Math Prob Eng 2019 3. Bolboli J, Khosravi MA, Abdollahi F (2019) Stiffness feasible workspace of cable-driven parallel robots with application to optimal design of a planar cable robot. Robot Auton Syst 114:19–28 4. Gosselin C (1990) Stiffness mapping for parallel manipulators. IEEE Trans Robot Autom 6:377– 382 5. Wu J, Wang J, You Z (2011) A comparison study on the dynamics of planar 3-DOF 4-RRR, 3-RRR and 2-RRR parallel manipulators. Robot Comput Integr Manuf 27:150–156 6. Choi KB (2003) Kinematic analysis and optimal design of 3-PPR planar parallel manipulator. KSME Int J 17:528–537 7. Vinoth V, Singh Y, Santhakumar M (2014) Indirect disturbance compensation control of a planar parallel (2-PRP and1-PPR) robotic manipulator. Robot Comput Integr Manuf 30:556–564

CFD Analysis for Heat Transfer Enhancement of Microchannels Heat Sink Using Nanofluid Flow in Case of Electronics Device Sushant Suresh Bhuvad, Arvind Kumar Patel, and S. P. S. Rajput

Abstract A CFD analysis is performed using nanofluids flow on a microchannels rectangular heat sink under uniform heat flux condition for forced convection cooling of electronic devices. In the present investigation, eight varying concentrations of Aluminum Oxide, Titanium Dioxide, Copper Oxide, Silicon Dioxide and Zinc Oxide nanoparticles, and EG20 (mixture of ethylene glycol 20% wt. and water), water as base fluids are considered. By considering the single-phase model, numerical computation is performed using ANSYS Fluent software. To examine the validity, results are compared with previous experimental and numerical research data. Further, different heat transfer parameters are presented and analyzed. From this analysis, it was noted that with the addition of nanoparticles there are sharp decrements in local thermal resistance and increment in local heat transfer coefficient compared to base fluid. There is a large improvement in heat transfer parameters is noticed in the case of CuO nanoparticles having a concentration of 1 and 4% in water base fluid. Keywords Nanofluid · Microchannel heat sink · ANSYS fluent

1 Introduction Today, also the rate of heat dissipation from IC chips is main concerned even simple liquid cooling is insufficient, hence there is increased in interest of nanofluids flow. Azizi et al. [1] studied that for concentrations of 0.05, 0.1, and 0.3 mass percentages of CuO nanoparticle in water; the Nusselt numbers increases by 17%, 19%, and 23%, respectively, in comparison to water. Dongsheng et al. [2] used Al2 O3 -Water nanofluid for their experiment in the case of a circular tube and found out that the local heat transfer coefficient increases up to 47% for 1.6% concentration of Al2 O3 nanoparticles.

S. S. Bhuvad · A. K. Patel (B) · S. P. S. Rajput Department of Mechanical Engineering, Maulana Azad National Institute of Technology, Bhopal 462003, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_13

99

100

S. S. Bhuvad et al.

From the literature review, it is observed that there may be further research required in case of some points like comparison of heat transfer enhancements using various concentrations of different nanoparticles in base fluids. Therefore, in the present analysis, this particular aspect will be analyzed numerically. Furthermore, commercially available CFD software ANSYS Fluent 13.0 is used for simulation purpose.

2 Physical Model and Mathematical Formulation 2.1 Heat Sink Model and Thermophysical Properties Figure 1 is the illustration of heat sink and single microchannel domain. Further information on the model other than dimension can be found from Toh et al. [7]. In this study, three different cases of heat sink dimensions are considered as discussed in Table 1. Here case 0 and case 1 dimensions are considered only for validation purpose and Case 3 dimensions are used for further numerical simulation. The thermophysical properties of base fluid and nanoparticles are considered from the previous study paper. Here particularly Xiang-Qi et al. [3], Khalil et al. [4], Batchelor et al. [5], and Nilesh Purohit et al. [6] equations are considered to find out properties of nanofluid. To find out local thermal resistance and local heat transfer coefficient following equations are used, respectively, Tmax(z) − Tin q q hz = Twall − Tb

R=

Fig. 1 a Diagram of sink, b view of microchannel

(1) (2)

2

1.4

1.4

0

1

2

2

2

1.5

Wt

56

56

64

Wc

44

44

36

Ws

533

533

489

Ht

420

320

280

Hc

4.7

4.7

1.277

Q˙ a

90

181

34.6

Flux q

200

200

150

No. of channels

L t total length of microchannels (cm), W t width of heat sink (cm), W c width of microchannels (µm), W s width of single heat sink (µm), H t height of sink (µm), H c height of microchannels (µm), Qa total flow rate through heat sink (cm3 /s), q heat flux (W/cm2 )

Lt

Cases

Table 1 Dimension for three different sets of microchannels

CFD Analysis for Heat Transfer Enhancement of Microchannels … 101

102

S. S. Bhuvad et al.

2.2 Governing Equations and Numerical Method Here, the Cartesian tensor system is used for more detail information of governing equation is found from Toh et al. [7] research article. Furthermore, numerical analysis is performed on ANSYS FLUENT 13.0, Solver is specified at simple pressure based on absolute velocity formation. For the simulation design solution, Standard ViscousLaminar model is opted with initial iteration value of 0.3 is taken as a relaxation factor. Further second-order upwind spatial discretization is considered for momentum and for other species. The residuals energy equation is set at 1 × 10−4 , for momentum and the other component equations it is 1 × 10−6 and for continuity equations 1 × 10−3 .

3 Results and Discussions 3.1 Validation of Model For validation of models, the results from the simulation are compared with previous data from literature article [7, 8] as shown in Fig. 2a, b. The simulation results are in good agreement with experimental data; there is little deviation seen because of the certain assumption which is made during the study.

Fig. 2 Correlation between experimental and present study for local thermal resistance parameters along with fluid flow: a for case-0, b case-1

CFD Analysis for Heat Transfer Enhancement of Microchannels …

103

Fig. 3 Effects of nanoparticles varying concentration in water at z = 0.9 cm plane: a on local thermal resistance, b on local heat transfer coefficient [∅b % = percentage concentration of nanoparticles]

3.2 Analysis of Varying Concentration of Nanoparticles in Deionized Water Base Nanofluids From Fig. 3a, it is evident that for volume fraction of 0.005; thermal resistances is decrease up to 5.62 percentages for different nanoparticles except for SiO2 . The decrement is sharp as even addition of a 0.5% Al2 O3 increases thermal conductivity of base fluid by 70.17 percentage. Since heat capacity (ρcp ) of CuO is higher, decrement in thermal resistance is more in case of CuO-Water nanofluid. For other cases of nanoparticle, thermal resistance remains constant, increases or decreases slowly as their volume concentration increase. Figure 3b shows that for 0.5% Al2 O3 in water there is 47.35% of increment in heat transfer coefficient and for 4% of Al2 O3 , it is 61.53% compared to water.

3.3 Analysis of Varying Concentration of Nanoparticles in EG20 (Where 20 Weight Percentages of Ethylene Glycol Mixed with Water) Figure 4 is the illustration of the variation in local thermal resistance for varying concentrations of nanoparticles in EG20. It is observed that for 4% concentration of CuO in EG20; thermal resistance is decreased by 36.52% and for the same concentration of Al2 O3 , it is 36.13% when compared with ethylene glycol. After analyzing the above two sections, a concrete conclusion can be made that CuO and Al2 O3 nanoparticles are the best choices. Hence for further study, these particular two nanoparticles are selected.

104

S. S. Bhuvad et al.

Fig. 4 Effects of nanoparticles varying concentration in EG20: on local thermal resistance at z = 0.9 cm plane

3.4 Comparison Between Water/EG20-Based Nanofluid with Varying Concertation of Al2 O3 and CuO Nanoparticles From Fig. 5a it is demonstrated that water-based nanofluid is more efficient than EG20-based nanofluid. As 0.5% concentration of CuO in water has thermal resistance 7% lower than 10% CuO in EG20. It is observed because EG20 fluid has lower conductivity than water. From Fig. 5b, it is observed that when a comparison is

Fig. 5 Comparison between water- and EG20-based fluid for different concentration of Al2 O3 and CuO nanoparticles, at Z = 0.9 cm plane: a on local thermal resistance, b on local heat transfer coefficients

CFD Analysis for Heat Transfer Enhancement of Microchannels …

105

made between the local convective heat transfer coefficient, EG20 is more effective than water-based except for lower concentration below 1% of nanoparticles, it is almost the same for both cases. Furthermore, in both cases, convective heat transfer coefficients increases linearly as a concentration of nanoparticles are rises.

4 Conclusions CFD analysis is performed using ANSYS FLUENT 13.0 software and the following observations are made from the above study: 1. Even lower concentration of oxide nanoparticles in water and EG20 base fluids; decreases local thermal resistance and increases local convection heat transfer coefficients effectively. 2. For ∅b = 0.5%, in case of CuO-water nanofluid, the local thermal resistance decreases by 5.62% when compared to water, and for CuO-EG20 it is decreased by 37.21% compared to ethylene glycol. 3. In spite of higher local convective heat transfer coefficients in EG20 compared to water, still it is less effective for cooling of devices because of higher viscosity for the same Reynolds number. 4. For this specific application in the case of electronics device, Water is the best choice as base fluid compared to EG20. The use of EG20 as a base fluid in nanofluid flow is preferable for higher working temperature condition as it has higher boiling point. 5. From this simulation study, after comparing all nanoparticles with given two base fluids. A concrete observation can be made that CuO-water nanofluid is more effective for cooling of electronic devices. As with 4% concentration of CuO in water, local convective heat transfer coefficient increases up to 61.53% other than that it has less cost, more stability and power required by the pump is also less. Acknowledgements This research was supported by the mechanical engineering department at Maulana Azad National Institute of Technology, Bhopal.

References 1. Azizi Z, Alamdari A, Malayeri MR (2015) Convective heat transfer of Cu–water nanofluid in a cylindrical microchannel heat sink. Energy Convers Manag 101:515–524. https://doi.org/10. 1016/j.enconman.2015.05.073 2. Wen D, Ding Y (2004) Experimental investigation into convective heat transfer of nanofluids at the entrance region under laminar flow conditions. Int J Heat Mass Transf 47:5181–5188. https://doi.org/10.1016/j.ijheatmasstransfer.2004.07.012

106

S. S. Bhuvad et al.

3. Wang X-Q, Mujumdar AS (2018) A review on nanofluids—part I: theoretical and numerical investigation. Braz J Chem Eng 25:613–630. https://doi.org/10.1590/S010466322008000400001 4. Khanafer K, Vafai K (2011) A critical synthesis of thermophysical characteristics of nanofluids. Int J Heat Mass Transf 54:4410–4428. https://doi.org/10.1016/j.ijheatmasstransfer.2011.04.048 5. Batchelor GK (1977) The effect of Brownian motion on the bulk stress in a suspension of spherical particles. J Fluid Mech 83:97–117. https://doi.org/10.1017/S0022112077001062 6. Purohit N, Purohit VA, Purohit K (2016) Assessment of nanofluids for laminar convective heat transfer: a numerical study. Eng Sci Tech Int J 19:574–586. https://doi.org/10.1016/j.jestch. 2015.08.010 7. Toh KC, Chen XY, Chai JC (2002) Numerical computation of fluid flow and heat transfer in Microchannels. Int J Heat Mass Transf 45:5133–5141. https://doi.org/10.1016/S00179310(02)00223-5 8. Tuckerman DB, Pease RFW (1981) High—performance heat sinking for VLSI. IEEE Electr Dev J 2:126–129. https://doi.org/10.1109/edl.1981.25367

Burr Registration and Trajectory Planning of 3D Workpiece Using Computer Vision M. R. Rahul , Rohini Y. Bhute, Shital S. Chiddarwar , Mohsin Dalvi , and Saumya Ranjan Sahoo

Abstract This paper proposes an effective way to identify the burr using image processing technique. The proposed method uses a simple mirror setup to capture 3D details of the burr. A set of image processing algorithm is used to calculate the dimensions of the burr. The burr dimensions are verified using the coordinate measuring machine. These details are then used to generate the trajectory for robot deburring. The deburring path generated is verified using ABB RobotStudio simulation software. The experimental result shows that the proposed approach provides an efficient method for robotic deburring. Keywords Burr registration · Image processing · Trajectory generation · Computer vision

1 Introduction Automating deburring process is highly challenging since the burr geometry is irregular and unpredictable. Though it is a critical finishing process, very few efforts are reported in the literature. One such attempt is to extract burr parameter from the original workpiece using parametric modeling is made by Ryuh et al. [1]. However, this method was time-consuming. To overcome this, Lee et al. [2] developed a programming technique by automatically scanning the workpiece to generate a trajectory for robot-assisted grinding. However, for every product, the scanning has to done separately. To overcome the above drawbacks, Jinno et al. [3] and Princely et al. [4] developed an automatic programming method using computer vision and force feedback. However, estimation of burr dimensions from the image data remains a challenge. In view of this, Idaku et al. [5] proposed an advance coded structured light M. R. Rahul (B) · S. S. Chiddarwar · M. Dalvi · S. R. Sahoo Visvesvaraya National Institute of Technology, South Ambazari Rd., Ambazari, Nagpur, Maharashtra 440010, India e-mail: [email protected] R. Y. Bhute Dassault Systems, Pune 411057, India © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_14

107

108

M. R. Rahul et al.

projection method for the acquisition of three-dimensional images at a high frame rate. However, this method requires a sophisticated camera setup which makes it time-consuming and impractical to implement in an industrial scenario. To overcome this, Nakao et al. [6] used a CCD camera and a simple mirror setup. However, the burr measurement was done manually. These techniques have some limitations like (1) Parametric modeling and other methods are possible only for an industry having no change in their product over a long period, (2) The image acquisition processes for burr detection are sufficient, however, cannot be fully autonomous. Therefore, these methods are not suitable for automating the process of deburring for workpiece with different profiles. In this paper, a teachless robotic deburring system for 3D workpiece using image processing is introduced. The outline of the paper is as follows: in Sect. 2, the methodology used for image processing and trajectory generation for the robot is explained. In Sect. 3, simulation and the experimental procedure are discussed. In Sect. 4, the results of the image processing are discussed. Finally, Sect. 5 discusses the conclusion drawn from the study.

2 Methodology In order to get 3D view of the burr, a setup of four mirrors placed at 45-degree angle is used (Fig. 1). Image acquisition is done using a CCD camera placed at a fixed distance above the workpiece. The mirror setup transforms the 3D view into multiple 2D projected images. These images are binarized for estimating the height and thickness of burr by using height and thickness algorithm developed by authors [7]. These estimated values are further used to generate the robot trajectory necessary for burr removal. This trajectory is simulated using ABB’s RobotStudio software for verification. On successful simulation, it is given to the robot controller for performing a deburring operation.

Fig. 1 a Binary Image of four sections captured in mirror. b Example of Left section image with size of 10 by 10

Burr Registration and Trajectory Planning of 3D Workpiece …

109

2.1 Mirror Setup The proposed image acquisition setup consists of four mirrors kept at 45-degree angle. A CCD camera is placed at a fixed height above the mirror to capture the image. This setup enables us to capture 3D burr and convert them into 2D images. The image consists of five sections and are named Lt (Left), Rt (Right), Up (Upper), Lo (Lower), and C (Central). The Up, Lo, Lt and Rt section of the whole image provide back, front, and both side views of the burr image, respectively. These sections are used to identify and estimate the burr dimensions.

2.2 Estimation of the Burr Dimensions The burr dimensions, i.e., the thickness and height are estimated from the image obtained through the CCD camera. The height data of the burr is obtained from the binarized Up, Lo, Lt, and Rt sections of the image (Fig. 1a). It can be observed that the presence of burr shows high intensity, i.e., 1 and 0 when the burr is absent. A unit matrix scanning method is employed to find the height of the burr. For example, suppose the Left section image is of size 10 by 10 as shown in Fig. 1B, each row is scanned for the presence of “1”. If “1” is found in a location, the counter value increases by “1” and is saved in a separate table. Then the search variable moves to the next row and repeats the same procedure till the last row. After this, the maximum value in the table is determined. This value gives us the height of the Left section. The same method is used to find the height of Right section. However, in the case of Up and Lo section, the search variable is moved column-wise as the height of the section is directing from top to bottom and vice versa.

2.3 Trajectory Generation Trajectory is generated using thickness and height of the burr. The fundamental inverse kinematic equations are used to calculate roll, pitch, yaw of the ABB IRB 120 robot. A data extraction algorithm is used to extract the data points for trajectory planning. In this algorithm, first the inner edge of the burr part is detected by using “Prewitt edge detection algorithm”, then number of passes is calculated by considering diameter of the tool. Since multiple edges are found during inner edge detection as shown in Fig. 2a, some additional steps like closed curve fitting, smoothening and backtracking coordinates elimination are used to assure accurate trajectory planning. Closed curve fitting method is used to eliminate the multiple edges of the burr parts. As the data points extracted from image are random, it was necessary to cluster data points based on the features. Gaussian mixture model based on clustering is used for curve fitting. Since this model is suitable only for open-loop data, the image is

110

M. R. Rahul et al.

Fig. 2 a Multiple inner edges of the burr part. b Sectioning of inner edge image to four equal quadrants for open-loop data

sectioned into four as shown in Fig. 2b to make it open-loop data. This data is then clustered by using K-mean clustering technique. A probabilistic Gaussian mixture model is used to identify centers of various clusters and a curve is fitted to acquire correct geometry as shown in Fig. 3a. After curve fitting, all quadrants are re-joined to get a single geometry as shown in Fig. 3b. The single edge obtained by using closed curve fitting algorithm has too much waviness, which is used directly, results in the jerky motion of the manipulator. Hence, it is necessary to make the curve smooth. For smoothening, an iterative discrete cosine transform algorithm is used. By minimizing the generalized cross-validation score, smoother curve is generated (Fig. 4a, b). The data of inner edge has some backtracking points as seen in Fig. 4b. These points are to be removed as they will lead to unnecessary points during data extraction for trajectory planning. A backtracking algorithm using the directional derivative is developed to eliminate these points. The algorithm calculates the direction vector of

Fig. 3 a Curve fitting done by using GMR model for all four quadrants. b Data obtained after curve fitting and re-joining all quadrants

Burr Registration and Trajectory Planning of 3D Workpiece …

111

50

50

50

45

45

45

40

40

40

35

35

35

30

30

30

25

25

25

20

20

20

15

15

15

10

10

10

5

5

0

0

5

10

15

20

25

30

35

40

45

50

5 0

0 0

5

10

15

20

25

30

35

40

45

50

0

5

10

15

20

25

30

35

40

45

50

Fig. 4 a Edge before smoothening. b Edge after smoothening. c Result of backtracking coordinate elimination algorithm

each point and saves it in a separate variable. For a particular location, if the direction vector is in reverse to earlier point, then that point is eliminated. The result of backtracking algorithm is shown in Fig. 4c.

3 Simulation and Experiment To verify the accuracy of the generated trajectory points, simulation using ABB RobotStudio software is done before actual deburring. The generated coordinate for the deburring trajectory is imported into RobotStudio software. The simulation shows that the robot tracked the burr trajectory accurately. To access the feasibility of the proposed method, deburring on a workpiece is conducted. The experimental setup developed for this work consists of a CCD camera, a workpiece for deburring on a temporary fixture and ABB IRB 120 six degrees of freedom industrial robot with a dummy tool attached. The RAPID code generated by the ABB RobotStudio software is transferred to the ABB IRB 120 robot controller. The trajectory generated had jerky motion in the early stage of execution. It is corrected by applying additional conditioning to the path like smoothening and backtracking coordinate elimination which is discussed in the previous section. The outcome of the experiment clearly shows that the robot tracked the burr path perfectly.

4 Results To validate the algorithm, burr height of multiple specimens is estimated using algorithm and CMM. It is observed that the average value of burr height is 4.95 mm from the algorithm and 5.3 mm from CMM resulting into average error of 0.72 mm. The coordinates generated by closed curve fitting had waviness and direction points. To filter this, smoothening function and backtracking point elimination algorithm is used. It is observed that the robot trajectory became smooth resulting in a Jerk’s free

112

M. R. Rahul et al.

Fig. 5 Maximum linear acceleration with and without Jerks

Fig. 6 a Power consumption without filtering. b Power consumption with filtering

motion of robotic arm (Fig. 5). It is also noticed that the total power consumption is reduced from 35 W to 16 W after filtering which is about 50% reduction (Fig. 6).

Burr Registration and Trajectory Planning of 3D Workpiece …

113

5 Conclusion In this paper, an attempt is made to detect the burr dimension using a simple mirror setup. The approach provides an effective way of generating a deburring trajectory without explicitly teaching the robot. The simulation and experiment show that this technique can be used for deburring any workpiece. Furthermore smoothening and backtracking point elimination algorithm helps in effective trajectory planning and Jerk free motion of the robotic arm.

References 1. Ryuh BS, Pennock GR (2006) Robot automation systems for deburring. In: Industrial robotics: programming, simulation and applications, p 15 2. Li CJ, Han GJS, Kyung JH, Park CH (2011) Study on teaching path reconstruction algorithm based direct teaching and playback method. In: Eighth international conference on ubiquitous robots and ambient intelligence, Incheon 3. Jinno M et al (1999) Teaching-less robot system for finishing workpieces of various shapes, Kyongju, South Korea, South Korea, 17–21 Oct 1999 4. Leo S, Princely F (2014) Vision assisted robotic deburring of edge burrs in cast parts. 12th Glob Congr Manuf Manag GCMM 97:1906 5. Idaku I, A coded structured light projection method for high-frame-rate 3D image acquisition. In: Ventzas D (ed) Advanced image acquisition, processing techniques and applications I, 14 March 2012 6. Yohichi Nakao YW (2011) Drilling burr profile measurement method using image processing technique. Trans Jpn Soc Mech Eng 7. Bhute RY (2017) Burr detection and trajectory planning using image processing for robotic deburring. Dissertation, Master of Technology, VNIT, Nagpur

In-situ Microwave-Assisted Casting of ASTM B23 Tin-Based Babbitt Alloy Sameer S. Gajmal

and Dadarao N. Raut

Abstract Metal casting is one of the primary manufacturing processes used in industries. It is also one of the economical routes of producing components varying from simple to complex shapes. But conventional casting methods possess some drawbacks in terms of higher energy consumption, longer melting times, air pollution and higher defects. Microwave-Assisted Casting is one of the latest manufacturing technique which can overcome these drawbacks owing to its several advantages like time-saving, clean and environmentally friendly. This paper reports on in situ microwave-assisted die casting process of bush with 35 mm OD, 25 mm ID and 40 mm Length, using microwave furnace at 2.45 GHz and 1.4 kW and also conventional die casting process by using muffle furnace of 3.5 kW of ASTM B23 Tin-based Babbitt Alloy. It was found that microwave-assisted die casting process reports time-saving of 07 min and power-saving of 1.1 kW h as compared with the conventional die casting process. Further SEM images of microwave bush casting showed uniform grain distribution as compared to SEM images of conventional diecasted bush. It was found that microwave-assisted die-cast bush is having 1.4 times higher hardness as compared with that of conventional die-cast bush. Keywords Die casting · Microwave · Microwave-assisted casting · Babbitt metal

1 Introduction Casting is one of the most preferred manufacturing techniques used in the industries. It is also one of the efficient techniques for producing the components. However, there are some drawbacks in the conventional casting process like environmental pollution, higher energy requirement, longer melting times, and a large amount of defects. To overcome these drawbacks, new techniques are being developed and many more in the process of development. Hence industries are now aiming at utilizing new techniques that are clean, energy-efficient and economical [1]. S. S. Gajmal (B) · D. N. Raut Department of Production Engineering, Veermata Jijabai Technological Institute, Mumbai, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_15

115

116

S. S. Gajmal and D. N. Raut

In microwave processing, the raw materials are exposed to the microwaves. Depending upon the method of exposure, microwave processing can be divided into two types—direct and indirect. In the direct method, the raw materials are directly exposed to the microwaves and in the indirect method, susceptor materials are used for processing the raw materials. Susceptor avoids the direct contact of the microwaves and the raw material. This method is also known as hybrid method [2, 3]. Utilization of microwaves for metal casting is one of the recent applications of the microwaves. Microwave-Assisted Casting process seems to be capable of overcoming the drawbacks of conventional casting processes. Microwaves are electromagnetic rays contain a mixture of electric and magnetic field with the frequency range of 0.3–300 GHz. Microwaves with 2.45 GHz frequency is generally used for processing of materials [4, 5]. Advantages like time-saving, cost-effectiveness, environment-friendly make microwave processing suitable for industrial applications [6, 7]. Mishra et al. [8] have studied the application of microwave in producing the casting of Al 7039. Microwave energy at 2.45 GHz frequency and 1400 W power were used for producing casting. Total three different solidification conditions, i.e. closed cavity cooling, open cavity and water-cooled cavity cooling were used for producing the casting and comparison of the castings was done. This paper reports on in situ microwave-assisted die casting at 2.45 GHz and 1.4 kW of ASTM B23 Babbitt bulk alloy and conventional die casting process using muffle furnace.

2 Experimental Setup In situ casting process of ASTM B23 Babbitt bulk alloy was done first by conventional die casting method by using muffle furnace and then by microwave-assisted die casting using microwave furnace. The component chosen was shaft bush used in one the special purpose cutting machine for performing teeth cutting operation of files. The bush size is as follows: Outer Diameter [OD] = 35 mm, Inner Diameter [ID] = 25 mm and Length = 40 mm (Fig. 1). A silica crucible of 250 ml capacity, which absorbs microwaves rapidly, was used for the melting of the charge in both cases [9]. Table 1 shows the chemical composition of ASTM B23 Babbitt bulk alloy.

In-situ Microwave-Assisted Casting of ASTM B23 …

117

Fig. 1 Bush

2.1 Method-01: Conventional Die Casting of Bush by Using Muffle Furnace Muffle Furnace as shown in Fig. 2 of make Shital Scientific Industries, Mumbai Model: SSI-48 (Sr. No. 32672008) was used. The maximum temperature which was possible to attain was 1200 °C. The heating chamber size: 12 × 6 × 6 . Maximum power rating of the muffle furnace was 3.5 kW with voltage of 240 V. The raw Babbitt metal as shown in Fig. 3 was used as charge and was subjected to a temperature of 360 °C by using silica crucible. After reaching the pouring temperature of 360 °C the molten Babbitt was poured in the mold cavity and was allowed to cool in the open air. After solidification, the bush casting was removed from the mold cavity. Figure 4 shows the as cast in situ Babbitt Alloy casting. Finally, the cast bush was machined by using a turning operation followed by boring operation to achieve the final size of the bush Outer Diameter [OD] = 35 mm, Inner Diameter [ID] = 25 mm and Length = 40 mm. Figure 5 shows the finished Babbitt bush as per the required dimensions.

2.2 Method-02: Microwave-Assisted Die Casting of Bush by Using Microwave Furnace Microwave Furnace as shown in Fig. 6 of make Phoenix, Mumbai Model: SSI-48 was used. The maximum temperature which can be attained is 1200 °C. The heating chamber size: 21 × 11.6 × 7.6 cm. Maximum power rating of the muffle furnace was 1.4 kW with voltage of 240 V. The raw Babbitt metal as shown in Fig. 3 was used as charge and was subjected to temperature of 360 °C by using silica crucible. The charge was subjected to microwave exposure for about 18 min at 1.4 kW and pouring temperature of 360 °C was achieved. After reaching the pouring temperature of 360 °C the molten Babbitt was poured in the mold cavity and was allowed to cool in open air. After solidification, the bush casting was removed from the mold cavity. Figure 7 shows the as cast in situ

Sn

89.35

Content

% Value

7.12

Sb

3.14

Cu 0.24

Pb

Table 1 Chemical composition of ASTM B23 Babbitt bulk alloy 0.023

Bi 0.002

Zn 0.005

Al

0.032

As

0.003

Cd

0.008

Fe

0.005

Ni

118 S. S. Gajmal and D. N. Raut

In-situ Microwave-Assisted Casting of ASTM B23 …

119

Fig. 2 Muffle furnace

Fig. 3 Babbitt raw material

Babbitt Alloy casting. Finally, the cast bush was machined by using turning operation followed by boring operation to achieve the final size of the bush Outer Diameter [OD] = 35 mm, Inner Diameter [ID] = 25 mm and Length = 40 mm. Figure 8 shows the finished Babbitt bush as per required dimensions.

120

S. S. Gajmal and D. N. Raut

Fig. 4 Babbitt bush casting using muffle furnace

Fig. 5 Bush casting after machining

3 Results and Discussion 3.1 Power and Time Consumed In both cases, the time required for melting the charge was noted. Also by using energy meter the actual power consumed during conventional and microwave melting was noted [10]. The in-situ bush casting obtained by using muffle furnace was found to have surface defects like porosity, blowholes, whereas the in situ bush casting obtained by using microwave furnace was found to defects free. The results obtained are compiled in Table 2.

In-situ Microwave-Assisted Casting of ASTM B23 …

121

Fig. 6 Microwave furnace

Fig. 7 Babbitt bush casting using microwave furnace

3.2 Microstructure Microstructure analysis was done by light microscopy and then by using a scanning electron microscope [11]. Figure 9 shows the microstructure of conventional and microwave-assisted die-casted bush done by light microscopy. It is clear that Tin-based Babbitts contain three phases α, β and η. The basic αphase is a solid solution of antimony (Sb) and copper (Cu) in tin (Sn). The β-phase is Sn–Sb and η-phase is Cu–Sn. The properties of tin-based Babbitt like hardness and wear resistance depend upon the amount and morphological size of the hard phases present in the alloy, especially Sn–Sb, i.e. β-phase and Cu–Sn, i.e. η-phase. When these hard phases are well defined in the microstructure, the wear resistance of the alloy gets improved (Fig. 10). Further microstructure analysis was also carried out by using scanning electron microscope (Oxford Instruments, X-MaxN , 50, GeminiSEM 300) available at

122

S. S. Gajmal and D. N. Raut

Fig. 8 Bush after machining

Table 2 Results of in situ bush casting Parameters

Conventional Casting

Microwave-Assisted Casting

Remark

Melting temperature

350 °C

350 °C

–

Pouring temperature

360 °C

360 °C

–

Time required for melting in minutes

25 min

18 min

Saving of 07 min

Power rating of furnace

3.5 kW

1.4 kW

–

Power consumed in kW h

1.45 kW h

0.35 kW h

Saving of 1.1 kW h

a. Microstructure of Conventional Die Casted Babbitt at 50 x

b. Microstructure of Microwave Assisted Die Casted Babbitt at 50 x

Fig. 9 a Microstructure of conventional Die Casted Babbitt at 50x, b microstructure of microwave assisted Die Casted Babbitt at 50x

In-situ Microwave-Assisted Casting of ASTM B23 …

a. Conventional Die Casted Babbitt at 100x

b. Microwave Assisted Die Casted Babbitt at 100x

c. Conventional Die Casted Babbitt at 500x

d. Microwave Assisted Die Casted Babbitt at 500x

e. Conventional Die Casted Babbitt at 2000x

f. Microwave Assisted Die Casted Babbitt at 2000x

g. Conventional Die Casted Babbitt at 5000x

h. Microwave Assisted Die Casted Babbitt at 5000x

123

Fig. 10 Microstructure of conventional and microwave-assisted die-casted Babbitt at 100×, 500×, 2000×, 5000×, 10,000×, 250,000×

124

S. S. Gajmal and D. N. Raut

i. Conventional Die Casted Babbitt at 10k x

j. Microwave Assisted Die Casted Babbitt at 10k x

k. Conventional Die Casted Babbitt at 25k x

l. Microwave Assisted Die Casted Babbitt at 25k x

Fig. 10 (continued)

Microstructural Mechanics and Micro-Forming Laboratory (MMMF), IIT-Bombay, Maharashtra, India. SEM images reveal that microstructure of microwave-assisted die-casted bush contains uniform distribution of grains as compared to that of conventional diecasted bush. The hard phases viz. β-phase of Sn–Sb and η-phase of Cu–Sn is present in the matrix of tin, i.e. basic α-phase.

3.3 Hardness Testing Hardness test was carried out on the conventional and microwave-assisted die-casted bush samples (polished) by using Hardness Testing Machine model-MRB-250, Sr. No. 2001/059 (Manufactured by; Meeta Test Instruments Pvt. Ltd., Miraj, Maharashtra) Rockwell hardness number was measured by using 1/16 ball indentor, scale-B with load of 100 kgf (Table 3). It is clear from the above readings that hardness of the microwave-assisted cast bush is almost 1.4 times higher than the conventional die-cast bush. This means there is an improvement in the hardness of the microwave-assisted casting.

In-situ Microwave-Assisted Casting of ASTM B23 …

125

Table 3 Hardness testing results Rockwell hardness numbers

Average Rockwell hardness number

Conventional die-cast bush

Microwave-assisted die-casted bush

12

19

13

20

16

19

17

21

13

18

14.20

19.40

4 Conclusion Microwave-assisted casting technique seems to be more advantageous than the conventional die casting method. • Microstructure of microwave-assisted die-casted bush contains uniform distribution of grains as compared to that of conventional die-casted bush. Also the hard phases viz. β-phase of Sn–Sb and η-phase of Cu–Sn present in the matrix of tin, i.e. basic α-phase and are well defined, because of which there is enhancement of the mechanical properties like hardness, wear resistance. • Hardness of the microwave-assisted cast bush is almost 1.4 times higher than conventional die-cast bush. • The power-saving is more than three times in the microwave-assisted casting compared with conventional die casting method. • Time and as further manufacturing cost reduction is also high in the microwave casting process. From this study, it can be concluded that microwave-assisted casting technique is a new technique which is more advantages than conventional die casting method in terms of power-saving and time-saving. Acknowledgements The authors are grateful for the support and encouragement from the Gharda Institute of Technology, Lavel, Taluka Khed, District Ratnagiri, Maharashtra. The authors would also like to acknowledge Nya. Tatyasaheb Aahalye Arts, Ved. S. R. Sapre Commerce and Vid. Dadasaheb Pitre Science College, Devrukh, District Ratnagiri, Maharashtra for providing microwave furnace facility to carry out this research work (Ref. No.: Sr. 216/2018-19 Dated: 30-07-2018).

References 1. Mishra RR, Sharma AK (2015) A new in situ casting technique using microwave energy at 2.45 GHz. In: Proceedings of the India international science festival—young scientist meet, Dec 2015

126

S. S. Gajmal and D. N. Raut

2. Mishra RR, Sharma AK (2016) A review of research trends in microwave processing of metalbased materials and opportunities in microwave metal casting. Crit Rev Solid State Mater Sci 41(3):217–255 3. Agrawal D (2006) Microwave sintering, brazing and melting of metallic materials. Sohn Int Symp Adv Process Metals Mater 4:183–192 4. Mishra RR, Sharma AK (2017) Structure-property correlation in Al–Zn–Mg alloy cast developed through in-situ microwave casting. Mater Sci. Eng A 688:532–544 5. Kashimura K, Sato M, Hotta M, Agrawal DK, Nagata K, Hayashi M, Mitani T, Shinohara N (2012) Iron production from Fe3 O4 and graphite by applying 915 MHz microwaves. Mater Sci Eng A 556:977–979 6. Mishra RR, Sharma AK (2018) Experimental investigation on in-situ microwave casting of copper. In: IOP conference series: materials science and engineering, vol 346 7. Rathee Y, Aggarwal V, Sehgal A (2016) Microwave assisted casting for fabrication of micro components. Indian J Sci Tech 9(36), Sept 2016 8. Mishra RR, Sharma AK (2012) Effect of solidification environment on microstructure and indentation hardness of Al–Zn–Mg alloy casts developed using microwave heating. Int J Metal Cast 12:370 9. Chandrasekaran S, Basak T, Ramanathan S (2011) Experimental and theoretical investigation on Microwave melting of metals. J Mater Process Tech 211(3):482–487 10. Singh S, Gupta D, Jain V (2016) Novel microwave composite casting process: theory, feasibility and characterization. Mater Des 111(5):51–59 11. Mishra RR, Sharma AK (2016) Microwave–material interaction phenomena: heating mechanisms, challenges and opportunities in material processing. Compos Part A: Appl Sci Manuf 81:78–97

Optimization of Heat Transfer Behavior of Industrial Refrigerants Through Different Cross-Section Microchannels Gourab Chakraborty , Shubhankar Sarkar , and Arunabha Chanda

Abstract Microchannel heat exchanger, nowadays, has become an area of interest in all kinds of industries. It requires the removal of high heat flux by the means of cooling optimization. In this numerical investigation, different cross-section geometries have been simulated, as it is used in different industries. The depletion of the ozone layer is now raising concern for the environmentalists. According to the Montreal Protocol, CFCs were eliminated by January 1996 giving chance for fair use of HFCs by the end of 2020. Our investigation approaches for selecting suitable refrigerants for the sake of environmental betterment. Two different fluids, namely R-410A and R22 have been selected along with water as a refrigerant. This analysis also covers their heat transfer characteristics and statistical optimization technique like response surface methodology for cost-cutting in computational methods. This investigation objectifies a high applicability of commercial coolants in the cooling industries. This investigation further opens different scope for study microfluidics boiling behavior as well as the heat transfer phenomenon through microchannel. Keywords Microchannel · CFD simulation · Industrial refrigerants · Response surface methodology

G. Chakraborty Department of Mechanical Engineering, SDET-BGI, Kolkata, India e-mail: [email protected] S. Sarkar (B) Aeronautical Society of India, New Delhi 110092, India e-mail: [email protected] A. Chanda Mechanical Engineering Department, Jadavpur University, Kolkata 700032, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_16

127

128

G. Chakraborty et al.

1 Introduction In the performance point of view, microchannel heat exchanger is way better than conventional heat exchanger, because of its higher heat transfer performance, low pressure drops, reduced size and above of all low coolant requirements which makes it perfect to lower the operational cost. Removal of high heat flux still keeps the cooling industries in Dire Straits. In the area such as Aerospace Industries, macro-electronic devices and nanorefrigerant applications there are limited cooling techniques. For the past few decades, it has been seen for cooling in microfluidic devices fluids pollute the environment such a way that in recent trends fluids that have lower pollution factors are introduced. In this numerical investigation, global warming potential have also been considered as a prime focus for selecting suitable materials. Tucker and Pease [1] first investigated the microchannel and introduce the concept of electrical cooling. Zhou et al. [2] experimentally presented the Evaluation of Correlations of Flow Boiling Heat Transfer of R-22 in Horizontal Channels. Shakarah et al. [3] theoretically investigated the Boiling Two-Phase Flow in Microchannels where they found the boiling categorized flow rules in microchannels. Duryodhan et al. [4] experimentally simulated the Boiling flow through diverging microchannel and studied the effect of heat flux and mass flux on pressure drop and heat transfer coefficient with boiling in the diverging microchannel. Shamirzaev et al. [5] experimentally present the flow boiling heat transfer for water and refrigerants in microchannel heat exchangers. Ekhlas et al. [6] investigated Flow Boiling Heat Transfer of R134a in Multi Microchannels. Lee and Mudawwar [7] had investigated A Mechanistic Critical Heat Flux Model For Subcooled Flow Boiling based on local Bulk Flow Conditions. Kandlikar [8] experimentally investigated A General Correlation for Saturated Two-Phase Boiling Heat Transfer Inside Horizontal and Vertical Tubes. Kew and Cronwell [9] have investigated the Correlations for the Prediction of Boiling Heat Transfer in Small Diameter Channels. In their study, they have wanted to set up the comparison of heat transfer coefficient between narrow existing flow boiling heat transfer and nucleate boiling type. Hsieh and Lin [10] predicted Saturation flow boiling heat transfer and pressure drop of refrigerant R-410A in a vertical plate heat exchanger. In their prediction, they used the ozone-friendly refrigerant R-410A which is the mixture of half percent of R-32 and half percent of R-125.

2 Theoretical Background 2.1 Governing Equation Navier–Stokes equations are among the primary governing equations of CFD. It is based on the conservation law of physical properties of fluid. The perception of conservational law is the exchange of properties, for instance, mass, energy, and momentum in multiphase flows, in an object is decided with the aid of the input

Optimization of Heat Transfer Behavior of Industrial …

129

parameters and output parameters. ρV (∇ · V ) = −∇ P + μ∇ 2

(1)

Further, in the flow field where the flow’s spatial properties are considered, The substantial derivative applies to any flow-field variable, for instance, DT /Dt, Dp/Dt, Du/Dt, etc., where T and p are the temperature and static pressure, respectively.  DT ∂ T − ∂T ∂T ∂T ∂T ≡ + → v ·∇ T ≡ +u +ν +w Dt ∂t ∂t ∂x ∂y ∂z where

∂T ∂t

(2)

→  local derivative and − v · ∇ T is convective derivative.

2.2 Boundary Condition To solve the problem, boundary condition should be well-defined. No slip boundary condition is applied in this investigation where the inlet has some velocity and pressure where outlet is computed by the simulation. Through the microchannel constant heat flux is applied. PISO method with Second-order upwind has been selected for perfect simulation as multiphase flow is considered (Fig. 1). u = v = w = 0 at the surface (when a viscous flow is considered)

(3)

− → − V ·→ n = 0 at the surface (when an inviscid flow is considered)

(4)

2.3 Properties of Fluids R-410A is an HFC refrigerant, which is blended using same amount of R32 and R125 as an equal amount of mixture. It is nonflammable refrigerant and nontoxic. Pressure generated by R-410A is 50% lower than R-22 although the operating pressure for 410A is slightly higher. R-410A has high vapor heat capacity which makes R-410A better coolant than R-22. As per European standards for using new air conditioning, Fig. 1 Type of boundary condition in microchannel geometries

130

G. Chakraborty et al.

R-22 was discontinued in 2010 because its effect on ozone layer depletion. This numerical investigation will contribute selecting material for industrial cooling. On the other hand, R-410A met the new standard for air conditioning application, with no contribution to ozone layer depletion. As per the performance point of view, it has been seen that R-410A can absorb and as well as release more heat than R-22. Solubility of R-410A is more than R-22.

3 Model Specification Considering various industrial application five different geometries have been selected. For keeping the investigation for unique aspect, hydraulic diameter has kept the same in entire investigation (0.1 µm). Constant Heat flux has been applied in the base surface of the microchannel geometries and refrigerants are passing through the microchannel. Fine mesh has been created using ANSYS software. For computing results without compromising accuracy and less computational cost mesh convergence test has been performed. It has been seen that all velocity profiles are identical to each other (Figs. 2, 3, 4, 5, and 6).

Fig. 2 Circular geometry

Optimization of Heat Transfer Behavior of Industrial …

131

Fig. 3 Rectangular geometry

Fig. 4 Semicircular geometry

4 Result and Discussion 4.1 Effect of Inlet Temperature in Heat Flux Profile Further, this investigation shows the influence of geometry and input parameters. It is pretty evident that input parameters have an important factor than the material properties. In this investigation, three different fluids passed through the microchannel. After the implementation of all boundary condition, it has been seen that the heat is gradually decreasing from the surface to center in the temperature profile for

132

G. Chakraborty et al.

Fig. 5 Trapezoidal geometry

Fig. 6 Triangular geometry

all microchannel. Where it has been found from inlet to outlet the temperature is increasing with respect to position although general conditions have been applied in the flow field. Heat flux profile abbreviation is written as H.F.P (Figs. 7, 8, 9, 10 and 11). Heat flux profiles are very important for microchannel fluid flow. Figure shows the temperature profile enabling the coolants to understand the most crucial parts for cooling application. As per safety concern temperature profiles also ensure the maximum heat profile zones to be treated priorities. In case in any anomaly standard protocols are to be considered for dodging any kind of accidents. In this investigation, it has been noted regarding inlet temperature plays a maximum role. Inlet

Optimization of Heat Transfer Behavior of Industrial … Fig. 7 Triangular H.F.P

Fig. 8 Trapezoidal H.F.P

Fig. 9 Rectangular H.F.P

133

134

G. Chakraborty et al.

Fig. 10 Semi-circular H.F. P

Fig. 11 Circular H.F.P

temperature is to replace with coolant temperature for cooling application, thus these two different fluids are considered for comparison in order to find the more effective fluid. Investigation required all possible heat flux values for comparison are observed in different temperature and the values are plotted (Fig. 12). Investigation clearly shows that with almost the same configuration and external condition for a microchannel coolant R-22 produces more heat flux than R-410A for same geometries. Investigation furthermore extended to performance point of view, simulation results are noted for heat transfer values which is a major deciding parameter for selecting suitable coolant with respective for geometries. ANSYS software is used for the simulation and values are tabled (Table 1). Obtained values are pretty evident to prove the performance of R-410A over R-22. Still, for more accurate and precised result statistical analysis has been also observed.

Optimization of Heat Transfer Behavior of Industrial …

Inlet temp. vs heat flux

150

Heat Flux in w/cm2

Fig. 12 Inlet temperature versus heat flux values of R-22 and R-410A

135

100

50

R-22 R-410A

0 0

20

40

60

80

Inlet temperature in oc

Table 1 Heat flux values for different shape with respect to fluid Shape

R22 (W/cm2 )

R410a (W/cm2 )

Water (W/cm2 )

Circular

0.000825136

0.0001353165

0.001175561

Semi-circular

0.00037193

0.0002737194

0.003420678

Triangular

0.0000505

0.0004560873

0.00114894527

Rectangular

0.0009160594

0.00215477709

0.00040108186

Trapezoidal

0.0005887209

0.003740303

0.00070111689

4.2 Response Surface Methodology RSM is a statistical technique for modeling numerical experiments for generating error response which enables reduce cost in experimental simulation. For a numerical investigation, incomplete convergence of iterative methods and round-off errors are often the challenging situation. In order to overcome this, response surface methodology is used. Statistically it has been found that R-410A is more effective than R-22 and Water. Heat transfer through circular microchannel is more effective than other geometries (Fig. 13).

5 Conclusion Analysis has been carried out for different microchannels of various configurations, in order to find optimum configuration of microchannels. It can be concluded that the results in this study were compared with the results from literature reviews and it indicated that the numerical results are in fine agreement with the reviews. Result shows R-410A is a more suitable material for selection. Geometrically circular microchan-

136

G. Chakraborty et al.

Fig. 13 Response surface of microchannel

0.004 0.0035 0.003 0.0025 R22 0.002

R410a

0.0015

Water

0.001 0.0005 0 Circular

Semi-circular Triangular

Rectangular Trapezoidal

Fig. 14 Effect of heat transfer

nel is preferred over the other microchannels because of its heat transfer potentiality. This work also contributes by concluding the preference of R-410A over R-22 for selecting suitable refrigerant, keeping in mind global warming potential concerning environmental betterment (Fig. 14).

Optimization of Heat Transfer Behavior of Industrial …

137

References 1. Tuckerman DB, Pease FW (1981) High-performance heat sinking for VLSI. IEEE Electron Device Lett EDL-2 126–129 2. Zhou Z, Fang X, Li D (2013) Evaluation of correlations of flow boiling heat transfer of R22 in horizontal channels. Hindawi Publishing Corporation 3. Shkarah AJ, Sulaiman MYB, Hj MRB (2013) Boiling two phase flow in microchannels: a review. Indian J Sci Technol 6(11):5514–5521 4. Duryodhan VS, Singh SG, Agrawal A (2013) Boiling flow through diverging microchannel. Sadhana 38(6):1067–1082 5. Shamirzaev AS, Mordovskoy AS, Kuznetsov VV (2018) An experimental investigation of flow boiling heat transfer for water and refrigerants in microchannel heat exchangers. AIP Publishing 6. Fayyadh EM, Mahmoud MM, Karayiannis TG (2017) Flow boiling heat transfer of R134A in multi micro channels. Int J Heat Mass Trans 422–436. (Elsevier Publication) 7. Lee CH, Mudawwar I (1988) A mechanistic critical heat flux model for subcooled flow boiling based on local bulk flow conditions. Elisevier Publication 14:711–728 8. Kandlikar SG (1990) A general correlation for saturated two-phase boiling heat transfer inside horizontal & vertical tubes. ASME 112/219:219–228 9. Kew PA, Cronwell K (1997) Correlations for the prediction of boiling heat transfer in small diameter channels. Appl Thermal Eng 17:705–715. (Elsevier Science Ltd. Publications) 10. Hsieh YY, Lin TF (2002) Saturation flow boiling heat transfer & pressure drop of refrigerant R-410A in a vertical plate heat exchanger. Int J Heat Mass Transfer 45:1033–1044. (Elsevier Science Ltd)

Evaluation of Two-Body Abrasive Wear Using FIS and ANN Mehar Amit Kumar

Abstract In the most recent three decades, many embed materials have been made of metals, compounds, earthenware production, polymers, and so forth. Most metals and earthenware production are a lot stiffer than bone tissue which can bring about mechanical bungle between the embed and the nearby bone tissue. Notwithstanding other biocompatibility issues, metals are excessively firm while pottery is excessively fragile and polymers are excessively adaptable and feeble to meet the mechanical quality. Thus, composites of polymers and inorganic materials may offer the ideal properties for embed materials. Polymers are well known because of their low thickness, great mechanical quality, and simple formability. At the point, when the composite is utilized as embed material, its development causes scraped spot at the joint. Henceforth, a study on scraped spot wear of composites is fundamental before utilizing it as embed material. In this work, two-body grating test has been completed on HAp-HDPE and HAp-UHMWPE bio-composites to ponder the impact of different test parameters on scraped spot wear. Fluffy derivation framework (FIS) and Artificial neural systems (ANNs) are utilized to foresee the wear qualities of composites. It has been seen that HAp-HDPE composite gives palatable outcomes contrasted with HAp-UHMWPE composite as far as grating wear test. The analysis results recommend that HAp-HDPE bio-composite has the potential for use as an elective material for burden-bearing orthopedic applications. Keywords Hydroxyapatite (HAp) · Fuzzy inference system (FIS) · Artificial neural network (ANN) · High-density poly ethylene (HDPE) · Ultra-high molecular weight polyethylene (UHMWPE)

M. A. Kumar (B) Raghu Engineering College, Visakhapatnam, Andhra Pradesh, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_17

139

140

M. A. Kumar

1 Introduction Hydroxyapatite (HAp)-strengthened polymer bio-composites offer a powerful framework to design engineered bone substitutes with custom-made mechanical, organic, and careful capacities [1]. This biomaterial is practically equivalent to bone having mechanical properties to impersonate those of bone tissues. Bone is the essential mineralized tissue in mammalian bodies whose principle capacity is ‘loadconveying’ [2]. The real constituents of bone tissues are minerals, natural material, and water; the natural material is for the most part collagen and the mineral is for the most part HAp (Ca10 (PO4 )6 (OH)2 ) which records for 69% of the heaviness of the bone [3]. HAp fortified polymer composites have been created as of late as simple materials for bone substitution, cement bone concretes and degradable inward obsession gadgets [4–7]. Among a few applicant polymers, engineered biocompatible polymers including Ultra-High Molecular Weight Poly Ethylene (UHMWPE), Poly Ether Ketone (PEEK) and High Density Poly Ethylene (HDPE), have been effectively strengthened with bioactive HAp for substitution or recuperating of bone [8–10].The motivation behind making such composites is to fortify the polymer and improve the bone holding properties of the material since it has been discovered that including HAp into a polymer network may transform an at first non-bioactive polymer into a bone holding composite and may all the while improve the mechanical properties. Standard wear tests are utilized for relative material positioning of a particular test parameter as stipulated in the test strategy. For increasingly reasonable estimations of material disintegration in modern applications it is important to direct wear testing under conditions recreating the accurate wear process [11]. The investigation of the procedures of wear is a piece of the control of tribology. The perplexing idea of wear has postponed its examinations and brought about disconnected investigations towards explicit wear systems or procedures [12].

2 Experimental Details To do the trial work, HAp is blended with high thickness polyethylene (HDPE) and ultra-high atomic weight polyethylene (UHMWPE) polymers with various volume rates. HAp powder is set up by blending Calcium Hydroxide (Ca(OH)2 ) powder, Ortho Phosphoric (H3PO4 ) corrosive, and Ammonia(NH3) arrangement. The technique is known as a wet concoction precipitation course [13–15].

2.1 Full Factorial Design of Wear Testing The chosen control parameters and their qualities at various levels are recorded in Table 1 underneath.

Evaluation of Two-Body Abrasive Wear Using FIS and ANN

141

Table 1 Control parameters and their levels Control parameters

Symbol

Level 1

Unit 2

3

HAp volume percentage

A

20

30

40

Load applied

B

500

1000

1500

% MPa

Sliding distance

C

100

200

300

Cycle

The experimental layout using full factorial design of experiment along with S/N ratio for HAp-HDPE and HAp-UHMWPE bio-composite are shown in Table 2. Table 2 S/N ratio for HAp-HDPE and HAp-UHMWPE bio-composite Factor

Factor

Factor

Abrasion rate

S/N ratio

Abrasion rate

S/N ratio

A

B

C

(HAp-HDPE)

(HAp-HDPE)

(HAp-UHMWPE)

(HAp-UHMWPE)

1

1

1

0.0043

47.33

0.0006

64.436

1

1

2

0.0049

46.196

0.0011

59.172

1

1

3

0.0092

40.724

0.0063

44.013

1

2

1

0.0088

41.11

0.0005

66.02

1

2

2

0.0092

40.724

0.0009

60.915

1

2

3

0.0119

38.489

0.0007

63.098

1

3

1

0.0099

40.087

0.0012

58.416

1

3

2

0.015

36.478

0.0013

57.721

1

3

3

0.0097

40.264

0.0009

60.915

2

1

1

0.0059

44.582

0.0631

23.999

2

1

2

0.0052

45.679

0.0041

47.744

2

1

3

0.0106

39.493

0.0094

40.537

2

2

1

0.0078

42.158

0.0059

44.582

2

2

2

0.0133

37.522

0.0011

59.172

2

2

3

0.0134

37.457

0.0015

56.478

2

3

1

0.0095

40.445

0.0041

47.744

2

3

2

0.0129

37.788

0.0014

57.077

2

3

3

0.0131

37.654

0.0056

45.036

3

1

1

0.0075

42.498

0.0249

32.076

3

1

2

0.0092

40.724

0.0209

33.597

3

1

3

0.0106

39.493

0.0093

40.63

3

2

1

0.0097

40.264

0.0039

48.178

3

2

2

0.0161

35.863

0.0125

38.061

3

2

3

0.019

34.424

0.0043

47.33

3

3

1

0.0195

34.199

0.0133

37.523

3

3

2

0.0202

33.892

0.0087

41.209

3

3

3

0.024

32.395

0.0035

49.118

142

M. A. Kumar

Fig. 1 Block diagram of fuzzy inference system of Fuzzy Mamdani inference system

3 Fuzzy Inference System (FIS) Square outline of Fuzzy Inference arrangement of Fuzzy Mamdani derivation framework appeared in Fig. 1.

4 Artificial Neural Network Fundamental structure of Artificial Neural Network appeared in Fig. 2.

Fig. 2 Basic structure of artificial neural network

Evaluation of Two-Body Abrasive Wear Using FIS and ANN

143

ANNs might be characterized as structures contained thickly interconnected versatile straightforward preparing components (called counterfeit neurons or hubs) that are fit for performing greatly parallel calculations for information handling and learning portrayal [16]. The information sign proliferates through the system a forward way, on a layer-by-layer premise [17]. Presently set a detailing to figure the blunder between the current and anticipating information we use Back-Engendering Algorithm [18]. The engaging quality of ANNs originates from the amazing data preparing attributes of the natural framework, for example, nonlinearity, high parallelism, power, shortcoming and disappointment resistance, learning, the capacity to deal with uncertain and fluffy data, and their ability to sum up [19]. Different layer feed-forward counterfeit neural system (ANN) is utilized to foresee the particular wear rate and frictional coefficient [20]. Counterfeit neural system which was connected to foresee explicit wear rate glass-filled PTFE composites through a well-prepared fake neural system which is done in MATLAB R2009a programming [21]. Displaying of grating wear opposition by methods for fake neural systems of Al-SiCp composites created by virus squeezing technique were acquired utilizing a back-proliferation neural system that utilizations slope drop learning calculation [22, 23].

5 Results and Discussion The experiment has been performed in the experimental layout gives the following results.

5.1 ANOVA Table for S/N Ratio of HAp/HDPE Bio-Composite The Analysis of Variance (ANOVA) Table for S/N ratio of HAp-HDPE composite is shown in Table 3. Table 3 ANOVA Table for S/N ratio of HAp-HDPE bio-composite Source

DF value

Seq SS

Adj SS

Adj MS

F value

P value

A

2

86.459

86.459

43.230

15.73

0.000

B

2

169.692

169.692

84.846

30.86

0.000

C

2

58.095

58.095

29.047

10.57

0.001

–

–

–

–

Error

20

54.979

Total

26

369.225

54.979 –

2.749 –

144

M. A. Kumar

Table 4 ANOVA Table for S/N ratio of HAp-UHMWPE bio-composite Source

Seq SS

Adj SS

Adj MS

F value

P value

A

DF value 2

1610.72

1610.72

805.36

17.09

0.000

B

2

558.40

558.40

279.20

5.92

0.010

C

2

60.95

60.95

30.48

0.65

Error

20

942.48

942.48

47.12

Total

26

3172.56

3172.56

–

0.534

–

–

–

–

Data Means % HAp

Load Applied

42

Mean

40 38 1

2 No Of Cycles

3

1

2

3

1

2

3

42 40 38

Fig. 3 Main effects plot (data means) for S/N ratio of HAp-HDPE bio-composite

The probability (P) value for each factor is less than 0.05 which means that all the factors are significant given by ANOVA Table 4. The Main Effects Plot (data means) for S/N Ratio is also given in Fig. 3 which shows the most optimum value given by the Main Effects Plot is 3, 3, 3.

5.2 ANOVA Table for S/N Ratio of HAp/UHMWPE Composite Analysis of Variance (ANOVA) Table for S/N ratio of HAp-UHMWPE biocomposite is shown in Table 4. Here the probability (P) value of A, B, and C factors are 0.000, 0.010 and 0.534, respectively shown above. Here we have seen that the value of A and B factor is less than 0.05, it means that it is significant while the value of factor C is greater than

Evaluation of Two-Body Abrasive Wear Using FIS and ANN

145

Data Means Load Applied

% HAp

60 55 50

Mean

45 40 1

2

3

1

2

3

No Of Cycles

60 55 50 45 40 1

2

3

Fig. 4 Main effect plot (data means) for S/N ratio of HAp-UHMWPE bio-composite

0.05 means it is not significant. The Main Effects Plot (Data means) for S/N ratio is shown in Fig. 4. Here, the most optimum value given from Fig. 4 is 3, 1, 1.

5.3 Fuzzy Inference System In a FIS, a standard base is developed to control the yield variable. A fluffy n guideline is a basic IF-THEN rule with a condition and an end. The assessments of the fluffy principles and the mix of the after-effects of the individual standards are performed utilizing fluffy set tasks and are finished by taking triangular Fuzzy sets. The tasks on fluffy sets are not quite the same as the activities on non-fluffy sets. The fluffy set guidelines for 27 trials which we have taken for HDPE and UHMWPE framework for getting wanted yield. Principle watcher for HDPE and UHMWPE are shown in Figs. 5 and 6. The test information for HAp-HDPE and HAp-UHMWPE bio-composites through the Fuzzy rationale. Table 2 demonstrated every one of the 27 trials with S/N proportion in which least estimation of S/N proportion is 32.395 and the most extreme worth is 47.33 for HAp-HDPE bio-composite and least estimation of S/N proportion is 23.999 and the greatest worth is 66.02 for HAp-UHMWPE bio-composite. Here, the prescient worth is determined by composing the directions in MINITAB 15. At that point, by exploratory worth and prescient worth, we determined the mean total rate blunder by taking the normal or mean of all examinations by condition (1).

146

M. A. Kumar

Fig. 5 Rule viewer for HDPE

Mean absolute % error =

 experimental value − predicted value predicted value total number of experiments

× 100 (1)

Percentage error calculated for HAp-HDPE and HAp-UHMWPE composites are 0.5037 and 0.9063, respectively.

5.4 Artificial Neural Networks By ascertaining, we can get the mean supreme rate blunder by condition (1) for preparing is about around 0.01 and 0.085 for HAp-HDPE and HAp-UHMWPE biocomposites through ANN model individually. So also, we can get the mean total rate blunder for testing information. By ascertaining we can get the mean total rate blunder by condition (1) for testing is about around 2.115 and 10.579 for HAp-HDPE and HAp-UHMWPE bio-composites through ANN model individually.

Evaluation of Two-Body Abrasive Wear Using FIS and ANN

147

Fig. 6 Rule viewer for UHMWPE

5.5 Microstructural Analysis Figure 7 shows the SEM micrographs of abrasive wear of UHMWPE at 30 vol.% of HAp at a magnification of 1000X. HAp not successfully compounded as compare to 30 vol.% of HAp-HDPE polymer matrix. Fig. 7 SEM micrographs of abrasive wear of UHMWPE at 30 vol.% of HAp

148

M. A. Kumar

Fig. 8 SEM micrograph of 40 vol.% HAp/HDPE

Figure 8 shows composites with 40 vol.% of HAp were successfully compounded and SEM images revealed good dispersion of HAp in both the polymer matrix.

6 Conclusion In this investigation, Fuzzy Mamdani and ANN methodologies were utilized to anticipate the abrasiveness in the two bodies wear trial of HAp-HDPE and HAp-UHMWPE composites. It recommends that the Fuzzy Mamdani model is best for the foreseeing information for the trial. The created models were assessed as far as their forecast capacity. Fake neural systems (ANN) have potential applications in mechanized discovery and finding of machine conditions. A large number of the ANNs for machine condition checking utilized the pre-handled recurrence space highlights of the deliberate vibration signals. The proposed ANN and FIS models would thus be able to be utilized successfully to foresee the two-body wear test. For HAp-HDPE proposed level of 40% HAp, connected burden 1500 MPA and number of cycles (sliding separation) 300 are the better recommended for the structure for ideal outcome and less wear between the bodies. For HAp-UHMWPE proposed level of 30% HAp, connected burden 500 MPA and number of cycles (sliding separation) 100 are the better recommended for the piece for ideal outcome and less wear between the bodies. By the test performed and the outcomes recommend that the proposed composite is HAp-HDPE for the implantation as fake bone in the human body instead of HAp-UHMWPE and further reason.

Evaluation of Two-Body Abrasive Wear Using FIS and ANN

149

References 1. Roeder RK, Converse GL, Kane RJ, Yue W (2008) Hydroxyapatite-reinforced polymer biocomposites for synthetic bone substitutes. JOM 60(3):38–45 2. Jones MH, Scott D (eds) (1983) Industrial tribology: the practical aspects of friction, lubrication, and wear. Elsevier Scientific Publishing Company, New York 3. Garnett J, Dieppe P (1990) The effects of serum and human albumin on calcium hydroxyapatite crystal growth. J Biochem 266(3):863–868 4. Coombes AGA, Meikle MC (1994) Resorbable synthetic polymers as replacements for bone graft. J Clin Mater 17(1):35–67 5. Juang HY, Hon MH (1994) Fabrication and mechanical properties of hydroxyapatite-alumina composites. J Mater Sci Eng Part C 2(1–2):77–81 6. Sendemir A, Altintas S (1997) Production of HAp reinforced polymer composites for biomedical applications. In: Proceedings of the 2nd international conference biomedical engineering days, 5, pp. 114–117 7. Kannan S, Balamurugan A, Rajeswari S (2001) Bio-composites: a review of literature. Trends Biomater Artif Organs 14(2):30–36 8. Chattopadhyay R (2001) Surface wear—analysis, treatment, and prevention. ASMInternational, OH, USA. ISBN 0-87170-702-0 9. Zhang Z, Friedrich K, Velten K (2002) Prediction on tribological properties of short fibrecomposites using artificial neural networks. Elsevier Wear 252:668–675 10. Pompe W, Worch H, Epple M, Friess W, Gelinsky M, Greil P, Hempel U, Scharnweber D, Schulte K (2003) Functionally graded materials for biomedical applications. J Mater Sci Eng Part A 362(1–2):40–60 11. Sousa RA, Reis RL, Cunha AM, Bevis MJ (2003) Coupling of HDPE/hydroxyapatite composites by silane-based methodologies. J Mater Sci Mater Med 14(8):475–487 12. Pramanik S, Agarwal AK, Rai KN (2005) Development of High strength Hydroxyapatite for hard tissue replacement. J Trends Biomater Artif Organs 19(1):46–51 13. Viswanath B, Ravishankar N, Nayar S, Sinha A (2005) Synthesis, Sintering and Micro structural characterization of nano crystalline Hydroxyapatite composites. Mater Res Soc Symp Proc 845(11–12):1–6 14. Lohfeld S, Barron V, Mc Hugh PE (2005) Bio-models of bone: a review. J Ann Biomed Eng 33(10):1295–1311 15. Jagur GJ (2006) Polymers for tissue engineering, medical devices, and regenerative medicine, Concise general review of recent studies. J Polym Adv Technol 17(6):395–418 16. De Xie Z, Qian D Huang, Abdi F (2006) Crack growth strategy in composites under static loading, Structures, structural dynamics, and materials conference. Am Inst Aeronaut Astronaut 5:1–8 17. Chowdhury AC, Kulkarni A Basak, Roy SK (2007) Wear characteristic and biocompatibility of some hydroxyapatite-collagen composite acetabular cups. J Wear 262(11–12):1387–1398 18. Kothamasu R, Haung SH (2007) Adaptive Mamdani fuzzy model for condition-based maintenance. Fuzzy Sets Syst 158:2715–2733 19. Tin-Oo MM, Gopalakrishnan V, Samsuddin AR, Al Salihi KA, Shamsuria O (2007) Antibacterial property of locally produced hydroxyapatite. Archives Orofacial Sci 2(11):41–44 20. Neuendorf RE, Saiz E, Tomsia AP, Ritchie RO (2008) Adhesion between biodegradable polymers and hydroxyapatite. J Acta Biomater 4(5):1288–1296 21. Huang AH, Farrell MJ, Kim M, Mauck RL (2010) Long-Term dynamic loading improves the mechanical properties of chondrogenicmesenchymal stem cell-Laden Hydrogels. J Euro Cells Mater 19(215):72–85 22. Eslami H, Solati-Hashjin M, Tahriri M, Bakhshi F (2010) Synthesis and characterization of nano crystalline HAp obtained by the wet chemical technique. J Mater Sci 28(1):5–13 23. Ren Q, Baron L, Balazinski M (2011) Type-2 fuzzy modelling for acoustic emission signal in precision manufacturing. Model Simul Eng 696947:1–12

Computational Analysis of Dual Expander Aerospike Nozzle Aswith R. Shenoy, T. S. Sreekumar, Pranav Menon, and Gerogi Alex

Abstract To increase the performance of current space launch capability new rocket designs are required. One concept that would undertake this need is an aerospike nozzle. The aerospike nozzle has proved its better performance compared to conventional bell nozzles. The major advantage of the aerospike nozzle is its altitude compensating ability since the expansion of the jet is not bounded by a wall. In the case of dual expander aerospike nozzle (DEAN) it uses separate expander cycles for both fuel and oxidizer with an aerospike nozzle. The present study includes the design of the combustion chamber and nozzle of the DEAN and its numerical analysis for producing a thrust of 2000 KN and a specific impulse of 420 s. The given input conditions are exit Mach number of 3.9, combustion chamber pressure and temperature of 180 bar and 3818 K and use kerosene as fuel and liquid oxygen as oxidizer with a mixture ratio of 2.65. The validation of the design is done with ANSYS FLUENT and a C++ program is prepared for the development of geometry of the combustion chamber and nozzle of the DEAN. The designed nozzle produces a specific impulse of 472 s. Keywords Aerospike nozzle · Dual expander aerospike nozzle · Numerical analysis · Specific impulse

1 Introduction A launch vehicle is designed to send a spacecraft designed for specific purpose into the required orbit. Due to simplicity, manufacturability, mission effectiveness bell nozzle rocket engine became the choice of engine and it is dominating today. The major disadvantage of bell nozzle is that it does not work optimally at all altitudes [1].

A. R. Shenoy (B) · T. S. Sreekumar · P. Menon · G. Alex Albertian Institute of Science and Technology, AISAT, Kochi, Kerala 682022, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_18

151

152

A. R. Shenoy et al.

Here comes the importance of aerospike nozzle. It is a type of altitude compensating nozzle that maintains its aerodynamic efficiency over a wide range of altitudes. The aerospike nozzle engine has small engine size and weight compared to a bell nozzle [2].

1.1 Aerospike Nozzle An aerospike nozzle has an optimal operation at different altitudes. Optimal operation means the exit pressure of the nozzle becomes equal to the ambient pressure. In the case of a bell nozzle, exhaust pressure is fixed by nozzle geometry. While in the case of aerospike nozzle, since it has no outer boundary, the working gas can expand to match the current atmospheric pressure at all altitudes. In the case of an aerospike nozzle, combustion occurs in a closed chamber similar to a bell nozzle. The hot gases coming out of the combustion chamber are accelerated in an internal expansion section to supersonic speed. The flow in this section is similar to gases moving between two walls which are coming to a point. When the flow achieves low supersonic speed one wall ends, while other wall continues down by forming spike contour [5].

1.2 Dual Expander Aerospike Nozzle The dual expander aerospike nozzle utilizes both fuel and oxidizer. It uses a separate expander cycle for both fuel and oxidizer. The schematic diagram of the DEAN is shown in Fig. 1. It uses separate turbomachinery for both fuel and oxidizer and separate heating channels. The expander cycle is powered by using heat from the walls of combustion chamber and nozzle. The heat obtained from the combustion chamber is used to produce power in both expander cycle and hence in turn to power both fuel and oxidizer pump [4]. The DEAN employs the oxidizer turbine and fuel turbine to power oxidizer pump and fuel pump, respectively. Since both the fuel and oxidizer remain separated until injection into the combustion chamber no interpropellant seals are required. In Fig. 1, red dotted line represents the fuel flow and the blue dotted line represents the oxidizer flow. The propellant first flows from the fuel tank to the pump. The pump delivers the oxidizer to cooling channels in the combustion chamber and fuel to cooling channels in the combustion chamber. By using the heat energy obtained from the cooling channel it drives the turbine. From here oxidizer and fuel are injected into the combustion chamber using injectors. After the combustion, the combusted product expands against the aerospike nozzle to produce the desired thrust.

Computational Analysis of Dual Expander Aerospike Nozzle

153

Fig. 1 The schematic of DEAN [3]

2 Methodology The given inputs are fuel and oxidizer, the oxidizer to fuel ratio, exit Mach number (M e ), combustion chamber pressure (Pc ), stagnation temperature of combustion chamber (T0 ),specific impulse (I sp ), thrust (F). The kerosene and oxygen are used as fuel and oxidizer with a mixture ratio of 2.65. The given input conditions are given in Table 1.

2.1 Combustion Chamber Design

Isp =

F mg ˙ 0

(1)

From the above equation, we can find the value of m˙ Table 1 The given input condition Pc

Tc

Me

Thrust

Specific impulse

180 bar

3818 K

3.9

2000 kN

420 s

154

A. R. Shenoy et al.

 ∗

C =

  γ +1 1 γ + 1 γ −2 RT0 γ 2 M (2)

With the value of characteristic velocity(C * ), we can find the value of At [3]. At =

mC ˙ ∗ Pc

(3)

We can find the value of the combustion chamber area (Ac ) from the below equation. Value of M ranges from 0.1 to 0.5. Ac 1 = At M



2 γ +1

 γ +1  γ − 1 2 γ −1 1+ M 2

(4)

Now input the value of L * according to the fuel and oxidizer used to find the volume of combustion chamber. L∗ =

Vc At

(5)

An optimized dimensions of combustion chamber is selected by considering minimum length required for complete combustion.

2.2 Nozzle Design Angelino’s approximate method is used for designing the contour of aerospike nozzle. Consider Fig. 2. A sonic flow is present along throat AB. The flow is assumed to expand by a centered wave originating at point A. For the desired contour of aerospike nozzle, we have to consider the streamline passing through point B. A characteristic Fig. 2 Aerospike nozzle [5]

Computational Analysis of Dual Expander Aerospike Nozzle

155

line is a straight constant flow property line. It is inclined to the sonic line AB by an angle A. Where a =μ−v (6) In Eq. (6), μ is the Mach angle and ν is the Prandtl–Meyer expansion angle. μ(M) = sin−1  v(M) =

(γ + 1) (γ − 1)

0.5

tan−1



1 M

  0.5 (γ − 1)  3 M −1 − tan−1 M 2 − 10.5 (γ + 1)

(7) (8)

In Eqs. (7) and (8), M represents the Mach number. The distance ‘l’ from the point A at which characteristic line crosses the nozzle boundary is fount by the continuity equation. A 1 l = lt sin μ At

(9)

where A is the area normal to velocity vector and polar equation is given by Eq. (10). a(M) = μ(M) − v(M) (10) In which M is the Mach number and it varies from 1 to desired exit Mach number M e.

2.3 Thrust Calculation The thrust is calculated by using momentum Eq. (11).     τ = p2 A2 + ρ2 A2 c22 − p1 A1 + ρ1 A1 c12

(11)

Putting the relation ρc2 = γ pM 2 , the Eq. (11) becomes     τ = p2 A2 1 + γ M22 − P1 A1 1 + γ M12 where subscripts 2 represents outlet section and 1 represents inlet section.

(12)

156

A. R. Shenoy et al.

2.4 Expander Cycle Design The input variables required for the design of expander cycle are chamber pressure (Pc ), chamber temperature (T c ), mass flow rate of fuel and oxidizer, pump efficiency, turbine efficiency, pressure ratio of turbine and pump. The pressure and temperature at which fuel and oxidizer must store can be calculated using this data by using turbine efficiency equation and pump efficiency equations.

3 Result and Discussion By using Angelino’s approximation the contour of the aerospike nozzle designed is shown in Fig. 3. The numerical analysis of the designed aerospike contour is done using ANSYS FLUENT. Pressure-based solver was used for the simulation at the start as it has higher accuracy for high Mach number and also the ease of convergence than density based solver. In terms of Turbulence modeling, SST k-ω 2-equation model was used because of its accuracy in case of high speed and especially for separated flows. The temperature and Mach number contour obtained at sea level are shown in Fig. 4. The total thrust was found to be 2800 Kn at sea level. Then, using Eq. (1) corresponding specific impulse was calculated as 472 s. For the designed aerospike nozzle numerical analysis has been done at varying altitudes. The specific impulse was calculated at different altitudes and it is plotted against altitude is shown in Fig. 5. The pressure and temperature at which kerosene and oxygen stored and details of nozzle are shown in Table 2.

Fig. 3 Aerospike nozzle contour for exit Mach no. 3.9

Computational Analysis of Dual Expander Aerospike Nozzle

157

Fig. 4 Temperature contour at sea level (Left), Mach number contour at sea level (Right)

Fig. 5 Thrust versus altitude

Table 2 Details of fuel storage

Throat area

0.0133 m2

Nozzle length

2.5 m

Lox temperature

700 K

Lox pressure

120 bar

LH2 temperature

600 K

LH2 pressure

68 bar

4 Conclusion The geometry of the Dual Expander Aerospike Nozzle is prepared for the given conditions. The DEAN nozzle is providing superior performance compared to a bell nozzle. A theoretical study on the various design aspects of aerospike nozzle has been conducted for cryogenic applications with given performance parameters. Thrust variation for different altitudes has been studied and found that specific impulse is

158

A. R. Shenoy et al.

increasing continuously with an increase in altitude. It is shown in Fig. 5. Based on the result obtained the DEAN would meet and exceed the thrust and specific impulse. The specific impulse obtained is 472 s. The increase in specific impulse obtained by comparing with bell nozzle is 32 s. A C++ program is prepared for the preparation of aerospike nozzle geometry, combustion chamber, and expander cycle.

References 1. Hammond W (2001) Design methodologies for space transportation systems. AIAA 2. Humble RW (1995) Space propulsion analysis and design. McGraw-Hill Inc 3. Martin DF (2008) Computational design of upper stage chamber, Aerospike, & cooling jacket of dual-expander rocket engine. MS Thesis, Air Force Institute of Technology 4. Simmons J, Branam R (2011) Parametric study of dual-expander aerospike nozzle upper stage rocket engine. AIAA J Spacecraft Rockets 48(2):355–367 5. Angelino G (1964) Approximate method for plug nozzle design. AIAA 2(10)

A Study on Performance and Emission Characteristics of Diesel Engine for Lower Blends of Karanja Biodiesel V. R. Patil, S. S. Sane, and S. S. Thipse

Abstract Increased number of vehicles are causing higher consumption of petroleum fuels leading to depletion of conventional fuel reserves. Hence, there is a need for alternative fuel, which will fulfil the demand. Biodiesel is one of the environment-friendly, renewable alternative biofuel which can be obtained from vegetable oils. Blends of various biofuels were used by researchers but there are limitations on the percentage of vegetable oil in diesel as emission norms are becoming stringent. Recently, the Government of India has announced that country is going to implement BS VI emission norms by 2020 and higher blends, i.e. B20 and above B20 are not satisfying these norms. Presently, very limited information is available on the use of lower biodiesel blend as a fuel in diesel engines. In this work, efforts are made to study the effect of lower blends of Karanja biodiesel as a fuel in a diesel engine to evaluate its performance and emission characteristics. The engine performance test was conducted on single cylinder four stroke diesel engine with lower blends of Karanja biodiesel (B5, B7, B10), B20 and diesel fuel to find brake thermal efficiency and BSFC. HC, CO, CO2 and NOx emissions also have been monitored. Results showed the brake thermal efficiency of all lower blends (B5, B7, B10) is high compared to diesel fuel at full load of the engine. BSFC of all lower blends (B5, B7, B10) is low at part load and almost the same as that of diesel fuel for maximum load. HC and CO emissions were less compared to diesel fuel. Also, a significant reduction of NOx was observed for B5, B7, B10. Keywords Biodiesel · Higher blends · Lower blends · Diesel engine · Emission

V. R. Patil (B) Department of Mechanical Engineering, AISSMS College of Engineering, Pune, India e-mail: [email protected] S. S. Sane Department of Mechanical Engineering, Pune Vidyarthi Griha’s College of Engineering and Technology, Pune, India S. S. Thipse Power Train Engineering, Automotive Research Association of India, Pune, India © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_19

159

160

V. R. Patil et al.

1 Introduction The consumption of conventional fuel is increasing much faster as compared to their natural production. The available fuel resources are insufficient to fulfil the demand in the future and are not renewable. Many efforts are made to find alternative fuels that will overcome this scarcity. Many researchers are working on biodiesel and proven that it is one of the best substitute renewable fuels having properties closer to diesel fuel. Various biodiesel blends and their feasibility in using an engine without doing any modification has been checked. Edible as well as nonedible oils after transesterification, their methyl ester or ethyl ester part is added in diesel fuel to form biodiesel blend. Transesterification is a process in which a glyceride reacts with an alcohol (methanol or ethanol) in the presence of a catalyst forming fatty alkyl esters and glycerin. Various blends are used in different proportions to find the performance and emissions of diesel engines. Agarwal and Rajamanoharan investigated the performance and emission characteristics of CI engine fuelled with different blends of Karanja oil [1]. Deepakkumar et al. used different blends of Mahua, linseed, Ricebran in CI engine and found a B20 blend gives close performance to diesel fuel [2]. The effect of engine compression ratio and fuel injection pressure was studied by Jindal and Nandwana using B100 Jatropha oil [3]. Rehman and Ghadage also studied the performance engine using Mahua oil (B100) [4]. Most of the researchers observed that B20 blend is closer to diesel fuel as far as performance and emission are considered. But as blend proportion is increased from B20 to B100 several problems arise such as filter clogging, carbon deposition due to the higher viscosity of blends. Thus, few researchers preheated biodiesel blend to reduce viscosity and studied their performance on engine. Hanbey Hazar tested performance and emission characteristics of diesel engine for preheated raw rapeseed oil at 100 °C [5]. Nagaraja used various blends of Palm oil with 90 °C temperature and checked performance parameters of the engine [6]. If higher blends are used after preheating performance is nearer to diesel fuel, but the emission is not satisfying the norms as well as separate arrangement is required for preheating. Nowadays, lower blends of biodiesel are becoming more widespread because emission norms are becoming more stringent in nearing future and higher blends are not satisfying the norms. Also, very limited information is available regarding the performance and emission of lower blends of biodiesel. Ertan Alptekin prepared six different oils and their lower blends (B2, B5, B10) and studied the effect of increase in the concentration of biodiesel on viscosity and density [7]. Murari Mohan Roy evaluated the effect of lower blends (B5, B10) of canola oil on the emission [8]. Mohammad Hafizil, Mat Yasin used a lower blend (B5) of Palm oil and found a significant reduction in NOx and reduction in CO, HC emissions [9].

A Study on Performance and Emission Characteristics of Diesel …

161

Table 1 Properties of biodiesel Sr. No.

1

2

3

4

5

6

Property

Density (kg/m3 )

Calorific value (kJ/kg)

Viscosity (mm2 /sec)

Cetane no

Cloud point (°C)

Pour point (°C)

Diesel

830

42,500

2.7

49

−4

−9

B5

832

42,360

2.85

49.1

−3

−7

B7

836

42,250

2.94

49.39

−2.5

−6

B10

839

42,010

3.15

49.44

−2

−5

B20

841

41,980

3.32

49.49

1

−2.5

B100

876

38,500

5.2

50.7

7

3

Ref. std. ASTM 6751

D1448

D6751

D445

D613

D2500

D2500

Instrument

Hydrometer

Bomb-calorimeter

Red wood-viscometer

Derived cetane number

Pour/cloud point apparatus

Pour/cloud point apparatus

2 Materials and Methods 2.1 Biodiesel Preparation Karanja oil from the seeds was extracted using a screw expeller. Oil was preheated to remove water contents at about 100 °C for 10 min in the cylindrical stainless steel tank. The Potassium methoxide is mixed with Karanja oil extract. After 8 h, glycerin and methyl ester separates out. The suspended impurities like unconverted methanol, unconverted catalyst were then removed by water washing method as methanol and catalyst mixes easily with water and having more density than methyl ester. The moisture was removed by heating the mixture at about 100–110 °C. Magnetic stirring method was used for blending. The various properties of biodiesel blends are listed in Table 1 which shows that density, calorific value, viscosity and cetane number of B5, B7, B10 are closer to diesel fuel. The reduction in calorific value was observed with the increase in proportion of the blend.

2.2 Experimental Setup and Procedure The technical specifications of the diesel engine (Fig. 1) used for experimentation are given in Table 2. The engine speed used was 1500 RPM with a compression ratio of 18. Eddy current dynamometer was used for load variation. The tests were carried out first for diesel fuel and then with different blends of Karanja oil biodiesel (B5, B7, B10, B20). The CO, HC and NOx emissions were measured with the help of Saj make HG 540-5 channel gas analyzer. CO is measured by NDIR absorption principle. In this method, CO gas is passed through the sample cell and air is passed

162

V. R. Patil et al.

Fig. 1 Experimental Set up of diesel engine test rig Table 2 Specifications of diesel engine set up

Sr. No.

Descriptions

Specifications

1

Make

Kirloskar TV1 Engine

2

Power

3.5 kw

3

Compression ration

18

4

Speed

1500 RPM

5

No. of cylinder

1

6

No. of stroke

4

7

Cylinder diameter (Bore)

87.5 mm

8

Stroke length

110 mm

6

Orifice diameter

20 mm

7

Dynamometer arm length

185 mm

A Study on Performance and Emission Characteristics of Diesel …

163

through the reference cell. In sample gas cell nondispersive infrared rays are absorbed by CO content. In reference, cell air is passed and there is no absorption of rays. This variation is detected in the detector cell and with the help of photovoltaic cell and electric signal it is measured. Zirconia sensor is used to measure oxygen in exhaust gas which is based on solid-state electrochemical fuel cell. Its two electrodes provide output voltage corresponding to oxygen in the exhaust gas. HC is measured by FID method. Mass air flow sensor was used to record air flow rate. Various engine parameters like brake thermal efficiency, brake specific fuel consumption and emissions like CO, HC and NOx were measured and the results were plotted. Enginesoft Software was used to capture the results.

3 Results and Discussion It is important to emphasize that the lower blend fuels and the diesel fuel that was used as reference were evaluated under the same conditions. Density, calorific value and viscosity of B5 and B7 found to be closer to that of diesel, whereas for B10 blend these properties are slightly higher than diesel. The increase in density as well as viscosity of the blends (B5, B7, B10, B20…) with increase in proportion of biodiesel was observed. It was found that the viscosity of B5, B7 and B10 is comparable with diesel. For B20 and above, i.e. higher blends may create filter clogging problem and poor atomisation. The calorific value decreases with an increase in blend proportion. A cetane number of lower blends are also close to diesel.

3.1 Performance Parameters 3.1.1

Brake Thermal Efficiency

The variation of brake thermal efficiency of lower blends of biodiesel (B5, B7, B10), B20 and diesel against engine load is shown in Fig. 2. Brake thermal efficiency is almost same as diesel at part load for B5 and B7 blend. For maximum load brake, thermal efficiency of (B5, B7, B10) is high compared to diesel. Increase in brake thermal efficiency for B5, B7, B10 was found to be 5%, 12% and 8% than diesel as there is more oxygen chemically bonded with oxygen in biofuel, which is an additional source of oxygen other than oxygen present in the intake air which helps in better combustion.

3.1.2

Brake Specific Fuel Consumption

Figure 3 shows the variation of Break Specific Fuel Consumption (BSFC) of all blends and diesel against engine load. The BSFC was found to be decreasing with

164

V. R. Patil et al. 30

Fig. 2 Variation of Brake Thermal Efficiency (BTE) of lower Karanja oil blends with load BTE in (%)

25 20

B00%

15

B5%

10

B7% B10%

5 0

B20% 0

5

10

15

Load in (kg)

3.5

B SFC in (Kg/Kw-hr)

Fig. 3 Variation of Break Specific Fuel Consumption (BSFC) of lower Karanja oil blends with load

3 2.5

B00%

2

B5%

1.5

B7%

1

B10%

0.5 0

B20% 0

5

10

15

Load in (Kg)

an increase in the engine load. It was noted that BSFC was lower than diesel fuel for part load and is nearer to diesel fuel for lower blends at maximum load as biofuel blend combustion efficiency is slightly better than diesel.

3.2 Emission Parameter The exhaust gas emission was measured using Saj make HG-540 5-Channel Gas Analyzer with the specifications listed in Table 3. It measures CO2 , CO, HC concentrations in the exhaust gas by non-diffractive infrared radiation (NDIR). O2 and NOx measurement by electrochemical method.

3.2.1

Hydrocarbon

Figure 4 shows the variation of HC emission of lower blends of biodiesel, B20 and diesel against engine load. Results show that the HC emission for B5, B7 and B10

A Study on Performance and Emission Characteristics of Diesel …

165

50

Fig. 4 Variation of Hydrocarbon (HC) of lower Karanja oil blends with engine load HC in (%)

40

B00%

30

B5%

20

B7%

10 0

B10% B20% 0

5

10

15

Load in (kg)

blends is low compared to diesel fuel for maximum load condition. HC emission for B20 was also lower than diesel at maximum load. Reduction of HC for B5, B7 and B10 was 60%, 50% and 20% as compared to diesel. Reduction of HC for B20 was also 10% than diesel. The cetane number of ester-based fuel is higher than diesel, it reduces the delay period and results in better combustion.

3.2.2

Carbon Monoxide

The carbon monoxide (CO) in exhaust gas is a result of incomplete combustion of fuel. Figure 5 shows the variation of CO emission for lower blends of biodiesel, B20 and diesel against engine load. CO emissions are low for B5, B7 and B10 as compare to B20 and diesel. Reduction of CO for B7 and B5 was 8% and 15% than diesel fuel as it is oxygenated biofuel thereby combustion of blends is better than diesel.

0.14

Fig. 5 Carbon monoxide variation of lower Karanja oil blends with engine load

0.12

CO in (%)

0.1

B00%

0.08

B5%

0.06

B7%

0.04

B10%

0.02 0

0

10

Load in (kg)

20

B20%

166

V. R. Patil et al.

Nox in (ppm)

2000 1500

B00% B5%

1000

B7%

500 0

B10% 0

5

10

15

B20%

Load in (kg)

Fig. 6 Variation of NOx of lower Karanja oil blends with load

3.2.3

Oxides of Nitrogen

Figure 6 shows the variation of NOx emission of lower blends of biodiesel, B20 and diesel against engine load. A higher reduction of NOx was observed for all lower blends compared to diesel fuel at part load as well as full load conditions. These lower NOx emissions might be due to lower temperatures in the combustion chamber. Reduction of NOx emission for B5, B7, B10 was almost 15–17%. Trend explains as the concentration of blend increases NOx percentage may increase. Formation of NOx does not only depend on temperature of combustion but also on content of nitrogen in the fuel itself. As nitrogen content in biodiesel is less, its emissions are less. Also, inconsistencies appear among studies as engine type, engine technology and fuel feedstock too.

4 Conclusion An experimental investigation has been conducted in this work to find the performance and emissions of lower blends of Karanja biodiesel (B5, B7, B10) and B20. Results are compared with diesel. Test results showed that the brake thermal efficiency of all lower blends is high compared to diesel at maximum load. Brake specific fuel consumption of all lower blends is nearer to diesel fuel. Considerable reduction in hydrocarbon emissions was observed. CO emission for B5, B7 and B10 blends is low compared to diesel fuel. There was a considerable reduction in NOx for B5, B7, B10 as compared to diesel fuel as well as a B20 blend. Thus, considering future stringent norms of emission and need for alternative fuel it will be better to use the lower blends of biodiesel for diesel engines.

A Study on Performance and Emission Characteristics of Diesel …

167

References 1. Agarwal AK, Rajamanoharan K (2009) Experimental investigations of performance and emissions of Karanja oil and its blends in a single cylinder agricultural diesel engine. Appl Energy 86:106–112 2. Agarwal D, Kumar L, Agarwal AK (2008) Performance evaluation of a vegetable oil fuelled compression ignition engine. Renewable Energy 33(6):1147–1156 3. Jindal S, Nandwana BP (2010) Experimental investigation of the effect of compression ratio and injection pressure in a direct injection diesel engine running on Jatropha methyl ester. Appl Thermal Eng 30(5):442–448 4. Raheman H, Ghadge (2008) Performance of diesel engine with biodiesel at varying compression ratio and ignition timing. Fuel 87(12):2659–2666 5. Hazar H, Aydin H (2010) Performance and emission evaluation of a CI engine fueled with preheated raw rapeseed oil (RRO)–diesel blends. Applied Energy 87(3):786–790 6. Nagaraja S, Sooryaprakash K, Sudhakaran R (2015) Investigate the effect of compression ratio over the performance and emission characteristics of variable compression ratio engine fuelled with preheated palm oil—diesel blends. Procedia Earth Planet Sci 11:393–401 7. Alptekin E, Canakci M (2008) Determination of the density and the viscosities of biodiesel– diesel fuel blends. Renewable Energy 33(12):2623–2630 8. Roy MM, Wang W, Alawi M (2014) Performance and emissions of a diesel engine fueled by biodiesel–diesel, biodiesel–diesel-additive and kerosene–biodiesel blends. Energy Convers Manage 84:164–173 9. Yasin MM, Paruka P, Mamat R, Yusop AF, Najafi G, Alias A (2015) Effect of low proportion palm biodiesel blend on performance, combustion and emission characteristics of a diesel engine. In: International conference on Energy procedia, Applied energy-ICAE 2015

Experimental Comparison Between Friction Stir Welding and Underwater Friction Stir Welding on Al6061 Alloys Hiten J. Mistry, Piyush S. Jain, and J. Vaghela Tinej

Abstract This study compared friction stir welding undertaken parameters such as tool rotation speed, transverse speed, and tool shoulder diameter on weld joint and compared ultimate tensile strength weld joint by underwater friction stir welding and normal friction stir welding using vertical machining center. For statistical analysis, the Taguchi L9 method was adopted. Ultimate tensile strength is tested as a response which shows that the ultimate tensile strength achieved by the underwater friction stir welding was higher than the normal friction stir welding. Regression analysis and analysis of variance were used as a statistical analysis method. From the analysis, it was observed that high tool rotation speed with minimum tool transverse speed gives maximum tensile strength for underwater friction stir welding and that of normal friction stir welding. Keywords Friction stir welding · Ultimate tensile strength · Underwater friction stir welding · Taguchi L9

1 Introduction Friction stir welding (FSW) was carried out in four stages; they are plunging stage, the dwelling stage, the welding stage, and the tool exit or retracting stage [1]. FSW process leads to the softening of the grains due to dissolution and coarsening of the grains which indirectly leads to a decrease in the mechanical properties of welded joints; for resolving these issues, underwater friction stir welding (UFSW) method is used which improves the cooling rate and lower pick temperature [2]. In comparison with other conventional methods of welding, only FSW is the method that consumes a considerable amount of energy. There is neither usage of flux nor cover gas, thus making the process eco-friendly [3]. The process of FSW H. J. Mistry (B) · P. S. Jain Department of Mechanical Engineering, S N P I T&R C, Umrakh, India e-mail: [email protected] J. Vaghela Tinej Department of Production Engineering, S N P I T&R C, Umrakh, India © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_20

169

170

H. J. Mistry et al.

is simply carried out by plunging a rotating FSW tool into the interface of two sheets which are rigidly clamped, till the shoulder riches out to the surface of the material being welded [4]. The movement of the tool transverses along the weld line as well as it transverses along the joint line which causes the material movement from advancing side to retracting side, and tool shoulder consolidates the material at the backside of the pin leading to a solid state [5]. The surrounding environment underwater, as well as an increase in travel speed, helps to increase the precipitation volume fraction, and by these, it reduces the average grain and precipitation size of the weld nugget [6].

2 Experimental Procedure To perform the welding Al6061 was selected as work material 150 × 90 × 6 mm. H13 tool steel was selected for this welding material for the tool. There was a 6 mm diameter of tool pin, length of the pin was 5.5 mm, and the tool shoulder diameter was 22, 24, and 26 mm. The friction stir welding was completed with and without the use of water as a working medium and compared the result of them in the term ultimate tensile strength (UTS). The experiment setup is shown in Fig. 1. The experimental work was completed at Vertical Milling Center PX20. UTS of the welded specimen will be tested according to ASME Section IX [7]. Effect of response characteristics ANOVA was used to find the significant variables. The main effect plots and their interaction plots were plotted using a statistical analysis method for each parameter at all levels which were used to examine the parametric effects on the response’s characteristics. All the results were analyzed using “Minitab 18”—statistical software (Table 1). Fig. 1 Underwater UWFSW

Experimental Comparison Between Friction Stir Welding …

171

Table 1 Process parameters Parameters

Levels

Tool rotation speed (rpm) (TRS)

Low

Medium

High

2250

2500

2750

Tool transverse speed (mm/min) (TTS)

10

15

20

Tool shoulder diameter (mm) (TSD)

22

24

26

3 Result and Discussion The study aims to identify the effect of friction stir welding parameters on ultimate tensile strength that was conducted using the Taguchi L9 method. Main effect plots and interaction plots were plotted using ANOVA for each parameter at all levels. The main effect plots were used to examine the parametric effects on the responses characteristics. All the results were analyzed using “Minitab 18”—statistical software.

4 Ultimate Tensile Strength Testing UTS of the welded specimen will be tested according to ASME Section IX. The Result of the tested welded component is as shown in Table 2. Table 2 UTS response of FSW and UWFSW Tool rotation speed (rpm)

Tool shoulder diameter (mm)

Tool transverse speed (mm/min)

UTS (N/mm2 ) FSW

UWFSW

1

2250

22

10

140

84

2

2250

24

15

154

80

3

2250

26

20

109

58

4

2500

24

10

139

41

5

2500

26

15

0

34

6

2500

22

20

50

93

7

2750

26

10

64

97

8

2750

22

15

72

0

9

2750

24

20

51

37

Run No.

172

H. J. Mistry et al.

5 Analysis of Variance for Underwater Friction Stir Welding for UTS Figure 2 shows that the behavior of UTS is low rotation speed gives maximum strength that of high rotation speed. Maximum UTS is achieved from 24 mm shoulder diameter and shows UTS decreases as tool shoulder diameter increases. For lower tool transverse speed, maximum UTS was achieved. The optimum parameters were tool rotation speed 2250 rpm, tool transverse speed 10 mm/min, and tool shoulder diameter 24 mm for UWFSW. Table 3. shows that the affecting parameters on UTS were investigated using ANOVA. Main Effects Plot for UTS Data Means

TRS

140

TSD

TTS

130 120

Mean

110 100 90 80 70 60 50 2250

2500

22

2750

24

26

10

15

Fig. 2 Main effect plot for underwater friction stir welding

Table 3 Result of ANOVA for underwater FSW Source

DOF

Adj SS

Adj MS

P-Value

TRS

2

10,273

5136

0.021

TTS

2

3515

1757

0.044

TSD

2

4876

2437

0.036

1404

Error

2

2802

Total

8

21,472

S = 37.4715

R-Sq = 86.92%

R-Sq (adj) = 67.69%

20

Experimental Comparison Between Friction Stir Welding …

173

6 Statistical Analysis of Friction Stir Welding At lower tool rotation speed, UTS was maximum, and at higher tool rotation speed, UTS was minimum. The maximum UTS was achieved at lower tool transverse speed. For tool shoulder diameter, the UTS was maximum achieved at higher tool shoulder diameter (Fig. 3). The optimum process parameter for the friction stir welding without the use of water was tool rotation speed 2250 rpm, tool transverse speed 10 mm/min, and tool shoulder diameter 26 mm. Table 4 shows that the affecting parameters on UTS were investigated using ANOVA.

Fig. 3 Main effect plot for UTS for friction stir welding without use of water

Table 4 Result of ANOVA for FSW without use of water Source

DOF

Adj SS

Adj MS

TRS

2

1312.9

656.44

0.044

TTS

2

2032.9

1016.44

0.034

TSD

2

162.9

81.44

0.049

Error

2

5066.9

2533.44

Total

8

S = 50.33

P

8575.6 R-Sq = 73.919%

R-Sq (adj) = 66.47%

174

H. J. Mistry et al.

Table 5 Regression analysis ANOVA for underwater friction stir welding Source

DF

Adj SS

Adj MS

F

P

Regression

3

12,044

4015

2.13

0.039

TRS

1

7776

7776

4.12

0.028

TTS

1

2948

2948

1.56

0.043

TSD

1

1320

1320

0.70

0.046

Error

5

9428

1886

Total

8

21,472

S = 43.4232

R-sq = 76.09%

R-sq (adj) = 49.75%

7 Statistical Equation for Underwater Friction Stir Welding Regression equation for underwater friction stir welding is (Table 5) UTS = 691 − 0.1441 TRS − 4.43 TTS − 7.42 TSD

(1)

8 Regression Analysis for Friction Stir Welding Without Use of Water Regression equation for friction stir welding is (Table 6) UTS = 198 − 0.0587 TRS − 1.13 TTS + 1.00 TSD

(2)

Table 6 Regression analysis ANOVA for friction stir welding Source

DF

Adj SS

Regression

3

1507.33

TRS

1

1290.67

TTS

1

192.67

TSD

1

24.00

Error

5

7068.22

1413.64

Total

8

S = 37.5985

Adj MS

F-Value

P-Value

502.44

0.36

0.041

1290.67

0.91

0.038

192.67

0.14

0.046

24.00

0.02

0.048

8575.56 R-sq = 77.58%

R-sq (adj) = 69.75%

Experimental Comparison Between Friction Stir Welding …

175

Fig. 4 UTS comparison chart

9 Comparison of Both Parameters with Respect to Ultimate Tensile Strength To identify the best properties achieved by the process with or without the use of water, comparison of both processes must be required. For better understanding or better comparison of both processes, a graphical method was used. The bar chart was generated for both processes. The chart is as shown in Fig. 4. From the chart, it is cleared that the UTS achieved by the UWFSW was higher than the UTS achieved by the FSW. The maximum UTS achieved by the use of water was 154 N/mm2 and by the friction stir welding without the use of water was 97 N/mm2 .

10 Conclusions This research work presents an experimental comparison between friction Stir welding and underwater friction stir welding on Al6061 alloys. The ANOVA was performed for the analysis of the work, and also, regression was done; the following conclusions were achieved: 1. UTS decreases as tool rotation speed increased, and UTS increases as tool rotation speed decreased. 2. At lower tool transverse speed, UTS has achieved maximum, and tool transverse speed increases, and UTS of the underwater friction stir welding decreases. 3. Tool shoulder diameter increases UTS, and in UWFSW, UTS increases up to a certain level after that level and then decreases. 4. In FSW, at lower tool rotation speed, UTS increases, and as tool rotation speed increases, UTS decreases.

176

H. J. Mistry et al.

5. At the lower tool transverse speed, UTS was maximum; as tool transverse speed decreases, UTS also decreases up to a certain level; further increment of tool transverse speed results in increment of UTS. 6. In friction stir welding, as the tool shoulder diameter increases, UTS also decreases up to a certain level; after that level, UTS increases. 7. From various experiments performed, the maximum UTS was achieved at 2250 rpm tool rotation speed, 10 mm/min tool transverse speed, and 24 mm tool shoulder diameter for the underwater friction stir welding, and for the FSW, without use of water, maximum UTS was achieved at 2250 tool rotation speed, 10 mm/min tool transverse speed, and 26 mm tool shoulder diameter. 8. From the results and analysis, it is clear that the UTS achieved by underwater friction stir welding is higher than the underwater friction stir welding. 9. According to ANOVA, most significant parameters are tool rotation speed and tool transverse speed which affect the UTS of underwater FSW and FSW, respectively.

References 1. Mishra RS, Ma ZY (2005) Friction stir welding and processes. Mater Sci Eng, R 50(2005):1–78. https://doi.org/10.4236/wjet.2018.62025 2. Kumar K, Kailash SV (2008) The role of friction stir welding tool on material flow and weld formation. Mater Sci Eng A 485:367–374. https://doi.org/10.1016/j.msea.2007.08.013 3. Wahid MA, Siddiquee AN (2018) Review on underwater friction stir welding: a variant of friction stir welding with great potential of improving joint properties. Trans Nonferrous Met Soc China 28:193 − 219. https://doi.org/10.1016/s1003-6326(18)64653-9 4. Clark TN (2005) An analysis of microstructure and corrosion resistance in underwater friction stir welded 304 l stainless steel. Brigham Young University, Provo. https://doi.org/10.1007/ s11771-012-12-2 5. Lingaraju D, Salavaravu L (2016) A review on underwater friction stir welding modified with normal friction stir welding setup. IJARSE 5(10) 6. Mofid MA, Abdollah-Zadeh A, Ghaini FM, Gür CH (2012) Submerged friction-stir welding (SFSW) underwater and under liquid nitrogen: an improved method to join Al alloys to Mg alloys. Miner Met Mater Soc ASM Int. https://doi.org/10.1007/s11661-012-1314-2 7. Liu HJ, Zhang HJ, Yu L (2014) Homogeneity of mechanical properties of underwater friction stir welded 2219-T6 aluminum alloy. J Mater Eng Perform 20(8). https://doi.org/10.1007/ s11665-010-9787-x 8. Rui-Dong F, Zeng-Qiang S, Rui-Cheng S, Ying L, Hui-jie L, Lei L (2011) Improvement of weld temperature distribution and mechanical properties of 7050 aluminum alloy butt joints by submerged friction stir welding. Mater Design 32:4825–4831. https://doi.org/10.1016/j. matdes.2011.06.021

Experimental Comparison Between Friction Stir Welding …

177

9. Vivek Kumar P, Prakash KB, Soundrapandian E (2017) Investigation on underwater friction stir welding of Al 6061 and AZ31. ICSTM-17. ISBN: 978-93-86171-78-8 10. Liu HJ, Zhang HJ, Yu L (2011) Effect of welding speed on microstructures and mechanical properties of underwater friction stir welded 2219 aluminum alloy. Mater Design 32:1548– 1553. https://doi.org/10.1016/j.matdes.2019.108315

Wear Particle Analysis Using Fractal Techniques Puja P. More

and M. D. Jaybhaye

Abstract Wear particle characterization plays a important role in condition monitoring of machine as most of the breakdown occurs due to wear particle saturation in the lubricating oil. Traditional methods for wear debris analysis depend on human expertise to conclude the results, which are subjective in nature, time consuming, and costly. The objective of this paper is to categorize different techniques of fractal analysis to study the wear particle morphology and calculate fractal dimension of wear particles. Fractal analysis is used to give information about different features of wear particles like fractal dimension, shape, size, color, boundary representation, and surface/texture analysis. This data can be used to detect the fault and decide prognostic maintenance period. Keywords Wear particle · Fractal analysis · Fractal dimension · Condition monitoring · ImageJ

1 Introduction Wear Particle Examination is one of the most important factors to monitor machine condition. Kumar [1] discussed wear particle morphology like shape, size, color, and texture which can be used to analyze the machine condition. Information obtained from wear particle analysis can be used to detect abnormal wear which helps to detect mode and mechanism of wear and wear severity. It can also help to give insight into the condition of machine and decide the maintenance period [2]. Wear debris is classified as rubbing, laminar, cutting, chunk (surface fatigue), sliding. There are few techniques used for wear particle analysis like ferrography, spectroscopic oil analysis, atomic absorption spectroscopy, _atomic emission spectroscopy, etc. Kumar and Kumar [3] studied that data collected from wear particle analysis is further studied under optical microscope or scanning electron microscope to check P. P. More (B) · M. D. Jaybhaye Department of Production Engineering and Industrial Management, College of Engineering Pune, Pune, Maharashtra 411005, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_21

179

180 Table 1 Wear particle size and shape descriptor [5]

P. P. More and M. D. Jaybhaye Method

Attribute

Descriptor

Fourier analysis

Contour and edge detail

1st, 2nd, 3rd, …, Harmonics

Form factor

Contour

Aspect-ration, roundness factor

Fractal analysis

Profile and edge detail

Structure and texture

Analysis of curvature

Edge detail

Standard deviation, skewness

Analysis of size

Size

Weibull parameter

the behavior of different wear particles. The images captured through this study acts as an input to fractal analysis. The images captured through this study act as an input to fractal analysis. Fractal analysis is used to compute fractal dimension which provides characterization of wear particle like graphic analysis, shape classification, and texture segmentation (profile and edge detection). The generalized formula to calculate fractal dimension is [4]: D =1−m

(1)

where D: Fractal dimension, m: slope of log(perimeter length) versus log(step-length), There are different wear particle size and shape descriptor discussed by Raadnui [5], which are mentioned in Table 1. Amongst all these methods, most commonly used method is fractal analysis as it gives overall wear particle morphology which can be used to detect failure mechanism and maintenance period [6]. Ghosh et al. [7] calculated fractal dimension of wear particle using image vision system and applied fractal mathematics to check gearbox condition. After analysis, it is concluded that the condition of gearbox was severe and researches suggested to change the gearbox oil. Lope and Betrouni [8] discussed the techniques used to calculate the fractal dimension for fractal and multi-fractal surfaces and their applications in the field of medical science. Debnath [9] discussed the brief history of fractals, fractal geometries, and calculation of fractal dimension. He has discussed few applications of fractal in the field of fracture mechanism and turbulence. Kang et al. [10] applied fractal analysis for ongoing process monitoring of tool wear with the help of fractal characteristics of machined surfaces. Researchers compared the variation in fractal dimension value and surface roughness value with change in cutting conditions. Tool wear showed an increase in fractal dimension

Wear Particle Analysis Using Fractal Techniques

181

value along with increasing surface roughness. Shah and Hirani [11] carried out wear debris analysis on spur gearbox to study the occurrence of type of wear particle. The analysis showed the presence of pitting and fatigue wear. Increase in wear particle and acid number showed the degradation of lubricant additives. Kirk et al. [12] applied fractal descriptors to wear particle boundaries. They concluded that fractal techniques can be used in machine condition monitoring to detect the change in wear process as fractal dimension is a function of wear progression involved. Stachowiak et al. [13] carried out oil analysis using ferrography for unused lubricating oil, which showed the presence of silicon-based mineral particle, aluminumbased mineral particle, and flat iron-rich metallic particle. Researchers applied boundary fractal analysis to calculate the texture and structural fractal dimension of these particles. So et al. [14] proposed a method named novel box counting method to estimate the fractal dimension of arbitrary-sized image. This method helped to overcome the problems related to conventional box counting method like problem with large window size is number of data points are not enough. Li et al. [15] presented a new box counting approach to improve accuracy of fractal dimension. Researchers proposed an algorithm that can suggest minimum number of boxes to cover the required image for all selected scale. Gonzata et al. [16] explained visual screen box counting method to determine fractal dimension. Hong-tao and Shi-rong [17] applied fishbone graph fractal method to depict boundaries of wear debris. Along with this, fractal analysis is applied to study the flow of river known as fractal geography of rivers, to calculate area of islands, mountains, lakes known as fractal nature of the earth, to calculate the perimeter of coastline known as fractal nature of coastline, to construct cables and bridges, to study the complexity of maps, to quantify, describe and diagnose cancer known as fractal medicine.

2 Fractal Analysis Fractal geometry of wear particle is used to represent condition of wear for any tribological system [18]. Large value of fractal dimension indicates more complex surface [19]. Procedure to carry out the fractal analysis is mentioned in Fig. 1. Fractal is a uneven geometric object which can be divided into subparts, each one of which is a reduced-size replica of the whole. Fractals are self-similar but do not depend on scale. Fractal structures are used to describe any complex shape in the world. Multifractal is a set of tangled fractals. Self-similarity of multifractals depends on scale. There are different algorithms used to determine fractal dimension of any image. Few of the algorithms are listed in flowchart shown in Fig. 2. Some of the techniques to calculate fractal dimension are discussed in this paper as follows:

182

P. P. More and M. D. Jaybhaye Collection of Lubricating Oil Samples

Separation of Wear Particle from Oil using Ferrography

Analytical Ferrography (Qualitative Analysis)

Direct Reading Ferrography (Quantitative Analysis)

Study of Wear Particles using Bi-chromatic Microscope Image Processing

Selecting Fractal Technique Calculation of Fractal Dimension

Fig. 1 Flowchart for wear particle analysis using fractal dimension

Fractal Dimension Computing Methods

Scale Invariance Limited by Pixel Sizes

1.Box Counting Method 1.1 Differential Box Counting Method 2.Variation Method 3.Hurst Orientation Method

Scale Invariant Method based on a Partition Iterated Function System Fractal Dimension by Partition Iterated Function System

Fractional Brownian Motion Model

Area Measurement Method

1.Power Spectrum 2.Maximum Likelihood Estimation 3.Grey Scale Variation

1.Isarithm Method 2.blanket Method

Fig. 2 Flowchart for selection of fractal analysis techniques

2.1 Box Counting Method (BC) This method is used for connected as well as non-connected forms of images. Fractal dimension D is calculated by covering entire three-dimensional surfaces with grid of

Wear Particle Analysis Using Fractal Techniques

183

squares. For this, sufficient amount of date is required, approximately 20 data points with self-similarity up to second or third order of magnitude. Log(N (β)) = a − D log(β)

(2)

N (β)

Mq (β) =



q

Pi

(3)

j=1

Where N(β): Number of objects whose linear dimension exceeds β M q (β): Moment formula used to calculate high-order dimensions (capacity, information, and correlation dimensions) D: Fractal dimension.

2.2 Differential Box Counting Method (DBC) This method works on the idea of self-similarity. Fractal dimension for this method for boundary set can be defined as      1 log D = lim log(Nr ) r →0 r

(4)

n r (i, j) = l − k + 1

(5)

Nr =



n r (i, j)

(6)

ij

where r is scale, nr (i, j) is number of boxes covered (i, j)th block, k and l are minimum and maximum gray level in (i, j)th block. Fractal dimension can be calculated from the least square linear fit of log(N r ) versus log(1/r). This method has limitation of over-counting or under-counting the box height, number of boxes, and image intensity [15].

3 Calculation of Fractal Dimension of Wear Particles HP EP 90 gear oil was collected after 500 h of running gear box. Quantitative and qualitative oil analyses were carried out on collected sample. Quantitative analysis using DR ferrography showed the presence of large- and small-sized wear particles. Concentration of large-sized particle (DL ) was 156.4 and concentration of smallsized particle (Ds ) was 123.2. As the concentration of wear particles was high in

184 Fig. 3 Steps to calculate fractal dimension using ImageJ

P. P. More and M. D. Jaybhaye

Selection of Required Image in ImageJ Software

Conversion of RGB Image to 8-bit Image Conversion of 8-bit image to Binary Image Fractal Box Count Tool Selecting Box sizes

Obtaining Fractal Dimension

given sample, qualitative analysis was carried out using analytical ferrography to prepare ferrogram slide to study morphology of wear particles using bichromatic microscope. Images captured from bichromatic microscope were used to calculate fractal dimension using Fractal Box Count Tool from ImageJ software. Initially, the required image whose fractal dimension needs to be calculated is selected in ImageJ tool [20]. After that the image is converted into 8-bit binary image. Fractal Box Count Tool is selected from analyze tool bar, which is used to select the box sizes for calculation part of fractal dimension. Change in box sizes affects the value of fractal dimension. Following steps are used to calculate fractal dimension of any image using ImageJ software, Fig. 3.

3.1 Fractal Dimension of Normal Rubbing Wear Particle The image was captured at 10× magnification factor as shown in Fig. 4a. Original image was converted to 8-bit binary image as shown in Fig. 4b. Then, box size is selected as 2, 3, 4, 6, 8, 12, 16, 32, 64. Table 2 shows the sizes of boxes and number of boxes occupied by wear particle for respected number of boxes. The notation 2, 3, 4, … n shows box sizes and D indicates the value of fractal dimension. Graph 1 shows the values of log(box sizes) plotted against log(count), where box size is the size of individual boxes which used to measure any object and count refers to the number of boxes occupied by given image for respective box size. Slope of the line in Graph 1 gives the value of fractal dimension. Fractal dimension value for given image of normal rubbing wear particle is 1.886.

Wear Particle Analysis Using Fractal Techniques

185

Fig. 4 a Original image of normal rubbing wear particle, b 8-Bit binary image of normal rubbing wear particle

Table 2 Size of boxes and corresponding number of boxes Box size

2

3

4

6

8

12

16

32

64

D

Number of boxes

892,060

413,400

241,421

113,384

66,372

31,105

18,164

4821

1270

1.886

Graph 1 Log (box size) versus log (count)

3.2 Fractal Dimension of Red Oxide Particle The image was captured at 50× magnification factor as shown in Fig. 5a. Original image was converted to 8-bit binary image as shown in Fig. 5b. Then, box size is selected as 2, 3, 4, 6, 8, 12, 16, 32, 64. As the original image of red oxide is blur, boundaries are not clearly separated after converting the image to 8-bit binary. Red oxides are formed due moisture content in lubricating oil. They appear as orange– reddish in color.

186

P. P. More and M. D. Jaybhaye

Fig. 5 a Original image of red oxide particle, b 8-bit binary image of red oxide particle

Table 3 shows the values of number of boxes corresponding to box size for red oxide. Graph 2 shows the log of box size versus log of count for red oxide particle. Fractal dimension of red oxide is 1.852. Table 3 Size of boxes and corresponding number of boxes Box size

2

3

4

6

8

12

16

32

64

D

Number of boxes

46,175

21,028

12,019

5559

3214

1528

892

250

77

1.852

Graph 2 Log (box size) versus log (count)

Wear Particle Analysis Using Fractal Techniques

187

4 Conclusions Fractal geometry is a powerful tool used to characterize and segment the wear particles in many mechanical systems. Fractal analysis is used to illustrate shape, texture/surface complexity of wear, and particle boundary. Few of the techniques used to determine fractal dimension are discussed in this paper. Based on literature survey, it is suggested that box counting algorithm is extensively used to calculate fractal dimension due to its easy understanding and simplicity. ImageJ software is used to estimate fractal dimension of these wear particle using Fractal Box Count Tool. Fractal dimension is calculated for two different wear particles that are normal rubbing wear particle and red oxide. As these particles are neither one-dimensional nor \two-dimensional, therefore, fractal dimension value for both particles is between 1 and 2. By calculating fractal dimension for large data of different types of wear particles, they can be distinguished using the fractal dimension ranges.

References 1. Kumar M (2013) Advancement and current status of wear debris analysis for machine condition monitoring: a review. Ind Lubr Tribol 65:3–11 2. Stachowiak GW (1998) Numerical characterization of wear particles morphology and angularity of particles and surfaces. Tribol Int 31:139–157 3. Kumar C Kumar M (2016) Wear debris analysis using ferrography. Int J Recent Trends Eng Res 2(8):398–404 4. Podsiadlo P, Podsiadlo GW (2000) Scale-invariant analysis of wear particle surface morphology: II. Fractal Dimension Wear 242:180–188 5. Raadnui S (2005) Wear particle analysis—utilization of quantitative computer image analysis: a review. Tribol Int 38(10):871–878 6. Kirk TB, Panzera D, Anamalay RV, Xu ZL (1995) Computer image analysis of wear debris for machine condition monitoring and fault diagnosis. Wear 181:717–722 7. Ghosh S, Sarkar B (2005) Wear characterization by fractal mathematics for quality improvement of machine. J Q Maintenance Eng 11(4):318–332 8. Lopes R, Betrouni N (2009) Fractal and multifractal analysis: a review. Med Image Anal 13:634–649 9. Debnath L (2006) A brief historical introduction to fractals and fractal geometry. Int J Math Educ Sci Technol 37:29–50 10. Kang MC, Kim JS, Kim KH (2005) Fractal dimension analysis of machined surface depending on coated tool wear. Surf Coat Technol 193(1–3):259–265 11. Shah H, Hirani H (2014) Online condition monitoring of spur gears. Int J Condition Monit 4:1–8 12. Kirk TB, Stachowiak GW, Batchelor AW (1991) Fractal Parameters and computer image analysis applied to wear particles isolated by ferrography. Wear 145:347–365 13. Stachowiak GW, Kirk TB, Stachowiak GB (1991) Ferrography and fractal analysis of contamination particles in unused lubricating oils. Tribol Int 6:329–334 14. So GB, So HR, Jin GG (2017) Enhancement of the box-counting algorithm for fractal dimension estimation. Pattern Recognit Lett 98:53–58 15. Li J, Du Q, Sun C (2009) An improved box-counting method for image fractal dimension estimation. Pattern Recogn 42(11):2460–2469

188

P. P. More and M. D. Jaybhaye

16. Gonzato G, Mulargia F, Marzocchi W (1998) Practical application of fractal analysis: problems and solutions. The Charlesworth Group 132:275–282 17. Hong-tao L, Shi-rong G (2009) Fishbone graph fractal description to UHMWPE wear debris boundary. Tribol Int 42(11–12):1624–1628 18. Shirong G, Guoan C, Xiaoyun Z (2001) Fractal characterization of wear particle accumulation in the wear process. Wear 251(1–12):1227–1233 19. Klinkenberg B (1994) A review of methods used to determine the fractal dimension of linear features. Math Geol 26(1):23–46 20. Karperien A (2004) FracLac advanced user manual. Charles Sturt University, Australia

Strategies for Low Engine Speed Torque Enhancement of Natural Gas Engine Used for Commercial Vehicles: Observations with Compression Ratio Pritesh J. Suple , Chandrakant R. Sonawane , S. S. Thipse , J. P. Mohite , and N. B. Chougule Abstract Since many years diesel engines are powering commercial vehicles. Of late, governments are promoting the use of natural gas (NG) as a fuel for such vehicles to reduce pollution. Thereafter, natural gas engines have witnessed faster development, especially for use in commercial vehicles. City bus is probably the most common NG commercial vehicle, dedicated to ferry passengers across the city. Some places impose a safety speed limit on such vehicles considering local traffic conditions. Thus, a typical scenario faced by such vehicles includes low drive speeds, high loads, frequent halts for passenger pick up and drop, signals, etc. Such vehicles thus need high torque at low engine speeds to manage these daily occuring conditions. Aim of this paper is to enhance torque at low-engine speed zone. Number of options such as the use of a turbocharger, direct injection of fuel, variable valve actuation, programmable waste-gates, etc. can help to realize higher engine outputs. The intent here is to study the effect of compression ratio and understand the extent of change in torque in engine low-engine speed region. Current study consists of modelling a reference commercial vehicle engine of six cylinders. A virtual model is built and its ability to represent actual engine performance from testbed is verified. Further, such model undergoes iterations of change in compression ratio and different parameters are studied for their relation with torque. Keywords Compressed natural gas engine · Low-speed torque · Compression ratio

P. J. Suple · C. R. Sonawane (B) Department of Mechanical Engineering, Symbiosis Institute of Technology, Symbiosis International Deemed University, Pune, India e-mail: [email protected] S. S. Thipse Automotive Research Association of India, Pune, India J. P. Mohite · N. B. Chougule Tata Motors, Pune, India © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_22

189

190 Table 1 Specifications of base engine

P. J. Suple et al. Bore × Stroke (mm)

97 × 128

Number of cylinder

6 Cylinder, Inline

Firing order

1st-5th-3rd-6th-2nd-4th

Rated speed

2500 RPM

Swept volume

5.7 L

Aspiration

Natural

Emissions

BS-IV compliant

Fuel

Natural Gas, spark-ignited

1 Introduction Urban dwellings are home to all categories of vehicles. Based on the consumption pattern, it has been observed that personal vehicles demand comparatively more energy consumption and are more energy demanding than public transport. Use of natural gas is an economical alternative which has less polluting, have ease of control, delivers better combustion, etc. [1–3]. Compressed Natural Gas engines for commercial vehicles are redesigned from their diesel counterparts. Although this approach is relatively easy and quick, it imposes a number of limitations on performance as not all parts are designed taking natural gas into consideration [4– 7]. A basic thermodynamic principle mentions that compression ratio plays a very important part in deciding output of an engine. Compression ratio, although a function of clearance volume, needs consideration of many parameters such as design limits, ability to withstand high temperatures, engine knocking, etc. [8–12].

1.1 Engine Modifications for Low-Speed Torque Enhancements A compressed natural gas, naturally aspirated, spark-ignited engine consisting of six cylinders is taken for reference. Specifications are as follows in Table 1. Performance of reference engine at full load is depicted graphically in Fig. 1.

2 Simulation and Engine Testing Initially, a software model of reference engine was built and compared against performance data from actual engine testbed performance test. In Fig. 2, close match is seen between in-cylinder pressure trace from actual measurement as well as a simulated one. Other parameters of engine performance significance are highlighted in Table 2. There was a reasonable agreement in different parameters, the model is

Strategies for Low Engine Speed Torque Enhancement of Natural …

191

400

300

300

200

Torque Power kW

100 0 700

1200

1700

200

SFC Air_Flow

2200

100

2700

0

SFC gm/kWhr, Air flow kg/hr

Torque Nm, Power kW

NA Engine Performance 400

Engine speed RPM Fig. 1 Graphical representation of base engine performance

Cylinder pressure

Pressure bar

60 50

Meas

40

Sim

30 20 10 0

-40

-20

0

20

40

60

80

100

120

Crank angle degrees Fig. 2 Comparing simulated and actual engine combustion pressure

Table 2 Test data and simulation output comparison Parameter

Test data

Sim Data

% diff

2500 RPM @ Max Power Torque

357.11

Power

93.44

356.76 93.401

Test data

Sim data

% diff

1500 RPM @ Max Torque 0.10

392.96

389.26

0.04

61.69

61.11

0.94

216.03

2.65 1.87

Air flow (kG/hr)

355.65

348.29

2.07

Fuel flow (kG/hr)

22.09

21.53

2.54

13.38

13.13

7.91

7.89

0.25

8.71

8.43

3.21

33.05

2.51

BMEP Brake efficiency

30.5

30.65

– 0.49

221.9

0.94

33.9

said to be calibrated and ready for iterating different setups such as valve timings, compression ratios, etc. Compression Ratio Study: It signifies the extent to which the fluid is compressed after entering the cylinder. Study here investigates the effect of compression ratio

192

P. J. Suple et al.

from 10.5:1 to 18.5:1, in an attempt to understand the gain in low-speed torque if any and the short-comings that must be compromised. Point to be noted that, we have not using variable compression ratio, rather we are testing the engine at fixed compression ratio. Hence during specified test a particular compression ratio (which is a fixed value) is maintained and by changing stepwise compression ratio for each specified steps. From Fig. 3, it is seen that torque enhances by about 6.5% as compression ratio is raised from 10.5:1 to 18.5:1 at an engine speed of 750 RPM. Similar pattern is observed at other engine speeds (Figs. 3, 4, 5 and 6). Torque

420

Torque (Nm)

400 380 360 340 10.5 15.5

320 300 700

11.5 16.5

12.5 17.5

1200

13.5 18.5

1700

2200

14.5

2700

Engine speed RPM Fig. 3 Comparision of Torque produced at various engine speed for various values of fixed compression ratio

BMEP and FMEP

9

3.25

8

7

2.25 10.5 13.5 16.5

11.5 14.5 17.5

12.5 15.5 18.5

1.75

FMEP (bar)

BMEP (bar)

2.75

1.25 6 700

1200

1700

2200

0.75 2700

Engine Speed RPM Fig. 4 Comparision of BMEP and FMEP produced at various engine speed for various values of fixed compression ratio

Strategies for Low Engine Speed Torque Enhancement of Natural …

Exhaust Energy

51

Exhaust energy (%)

193

49 47 45 43 41 39 37 35 700

10.5 15.5

11.5 16.5

1200

12.5 17.5

13.5 18.5

1700

14.5

2200

2700

Engine Speed RPM Fig. 5 Comparision of Exhaust energy (%) expelled at various engine speed for various values of fixed compression ratio

Exhaust Temp & Brake Efficiency

40

800 600

11.5 16.5

12.5 17.5

13.5 18.5

400

36 34

200 0 700

14.5

Brake eff (%)

Exhaust temp (°C)

38 10.5 15.5

32

1200

1700

2200

30 2700

Engine speed RPM Fig. 6 Comparision of Exhaust Temeprature and Brake Efficiency (%) at various engine speed for various values of fixed compression ratio

3 Conclusions It has been observed that the use of higher compression ratio engine helped to enhance the low-speed torque by about 6.5%. In all, it can be inferred that this enhancement to low engine speed torque is due to higher compression ratio originates principally from better utilization of fuel and enhanced thermodynamic expansion process. Practical constraints permitted engine testing with 12.5 CR only. Engine performance as compared to simulation can be seen as matching in Fig. 7. Hence findings from other simulations can be taken into consideration. They can be summarized in Table 3. It is to be noted that, at higher compression ratio engine may knock especially at higher engine speeds. Thus, one cannot increase compression ratio beyond a limit.

194

P. J. Suple et al.

Simulation vs measured NA engine Torque (Nm)

100

300

80 200

60

100 0 700

Msr Torque Nm Msr Power kW

Sim Torque Nm Sim Power kW

Power (kW)

120

400

40 20

1200

1700

2200

2700

Engine speed RPM Fig. 7 Simulated and measured engine output for 12.5 compression ratio

Table 3 Summarized results from different compression ratios

Observations from different compression ratio Parameter

18.5 CR

10.5 CR

Gain

Torque (Nm)

383.45

341.16

12.40

Power (kW)

100.39

89.31

12.40

BSFC (gm/kWhr)

214.85

240.77

−10.77

IMEP (bar)

10.17

9.13

11.42

BMEP (bar)

8.49

7.55

12.40

FMEP (bar) A/Flow (kG/hr) Fuel (kg/hr)

1.68

1.58

6.76

348.92

347.93

0.29

21.57

21.51

0.30

Brake Eff (%)

33.76

30.13

12.06

Tot Exh Energy (%)

43.93

50.06

−12.25

777.77

875.29

−11.14

63.14

41.85

50.87

Exh Temp (°C) Pmax (bar)

Most common means to control knocking, i.e. retarding ignition, especially at high speeds and loads, is not recommended as it leads to loss of efficiency, higher exhaust temperatures, and increases emissions. Other ways to enhance torque at low engine speeds are to be explored if higher torque is expected.

References 1. Reddy BS (1995) Transportation, energy and environment: a case study of Bangalore. Econ Polit Wkly 30(3):161–170 2. Parker RS, Pettijohn CE (1997) The use of alternative fuels in the private trucking industry: is there a viable target market? J Mark Theory Pract 5(4):88–93. (Fall) https://doi.org/10.1080/ 10696679.1997.11501783

Strategies for Low Engine Speed Torque Enhancement of Natural …

195

3. Frost & Sullivan (2016) HD transit bus market—global analysis. In: Presentation in market engineering, NF63-18 4. Jacob J, Abhimanyu Y, Kumar AH, Singh S (2014) Strategic analysis of compressed natural gas (CNG) passenger cars market in Europe. In: Presentation in market engineering, MA30-18. Frost & Sullivan 5. Gharehghani A, Koochak M, Mirsalim M, Yusaf T (2013) Experimental investigation of thermal balance of a turbocharged SI engine operating on natural gas. Appl Thermal Eng 60:200–207. https://doi.org/10.1016/j.applthermaleng.2013.06.029 6. Thipse S, Dsouza A, Sonawane S, Rairikar S et al (2017) Development of multi cylinder turbocharged natural gas engine for heavy duty application. SAE Int J Engines 10(1). https:// doi.org/10.4271/2017-26-0065 7. Kalam MA, Masjuki HH (2011) An experimental investigation of high performance natural gas engine with direct injection. Energy 36:3563–3571. https://doi.org/10.1016/j.energy.2011. 03.066 8. Semin I, Ismail A, Bakar A (2009) Investigation of torque performance effect of development of sequential injection CNG engine. J Appl Sci 9(13):2416–2423 9. Yan B, Wang H, Zheng Z, Qin Y et al (2017) The effect of combustion chamber geometry on in-cylinder flow and combustion process in a stoichiometric operation natural gas engine with EGR. Appl Thermal Eng. https://doi.org/10.1016/j.applthermaleng.2017.09.067 10. Yadollahi B, Boroomand M (2013) The effect of combustion chamber geometry on injection and mixture preparation in a CNG direct injection SI engine. Fuel 107:52–62. https://doi.org/ 10.1016/j.fuel.2013.01.004 11. Wu C, Deng K, Wang Z (2015) The effect of combustion chamber shape on cylinder flow and lean combustion process in a large bore spark-ignition CNG engine. J Energy Institute 1–8. https://doi.org/10.1016/j.joei.2015.01.023 12. Zhao J, Ma F, Xiong X, Deng J et al (2013) Effects of compression ratio on the combustion and emission of a hydrogen enriched natural gas engine under different excess air ratio. Energy 59:658–665. https://doi.org/10.1016/j.energy.2013.07.033

In-house Fabrication and Calibration of Silver Thin Film Gauge Akash Jadhav

and Ravi K. Peetala

Abstract Measurement of transient surface temperature and surface heat flux got vast attention with recent technological advancement. Thin film gauge (TFG) is the most suitable instrumentation for such measurements. Present work aimed to fabricate and calibrate the silver TFG. In the present work, performance parameters of TFG were estimated through static calibration. Further TFG was subjected to constant heat flux using in-house fabricated calibration setup in conduction mode to assess the thermal performance and reliability of silver TFG. The silver TFG measured the transient surface temperature, which was discretized using cubic spline technique and heat flux was recovered by considering one-dimensional heat conduction modeling in semi-infinite solid. Results revealed that silver thin film gauge can be considered for heat flux measurement for short time scale (s and ms) applications. Keywords Transient surface temperature · Surface heat flux · Silver TFG · Cubic spline · Semi-infinite

1 Introduction The capability to capture the transient surface temperature and heat flux is a demanding task in the present engineering problems. This task is easy if the experimentation time scale is large. However, for a short duration, capturing the transient surface temperature and heat flux is very difficult. TFGs are the most worthy sensors for such short duration measurements. TFG is the resistance temperature detector (RTD) having a metallic thin film of sensing material (platinum/silver/nickel) on a thermally insulating substrate such as Macor, Quartz, and Pyrex. The working principle of TFG A. Jadhav (B) · R. K. Peetala Department of Mechanical Engineering, Visvesvaraya National Institute of Technology, Nagpur 440010, India e-mail: [email protected] R. K. Peetala e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_23

197

198

A. Jadhav and R. K. Peetala

is the same as RTD. The resistance of TFG varies linearly with temperature. The calibration parameters of TFG such as temperature coefficient of resistance (TCR) and sensitivity can be determined using the oil bath calibration technique [1]. When TFG is exposed to sudden heat load, the electrical resistance of TFG varies based on the type of heat load. This resistance response can be converted to temperature signals with the knowledge of TCR [2, 3]. Short duration surface heat flux can be predicted by inverse analysis using the transient surface temperature history. The first step in inverse heat prediction is the discretization of temperature signals. Various discretization techniques are available in the literature [4, 5]. Heat flux can be predicted analytically by considering TFG as a thin film mounted on the semi-infinite substrate. Details of fabrication technique, calibration and heat flux prediction are given in the following sections.

2 In-house Fabrication of Silver Thin Film Gauge In the present work, TFG was fabricated using silver as a sensing material on Quartz rod of 6 mm diameter and 10 mm depth. The choice of silver was made by considering good adhesion and low cost. Quartz was selected as a substrate due to its thermal stability over the wide range of temperature up to 1000 °C. The first step in TFG fabrication is substrate preparation. In this process, the Quartz rod was first polished with the commercially available metallurgical grade silicon carbide sandpaper of 400 grit size (Made-3 M). The polishing process was continued using 1000 and then 2000 grit size sandpaper until the visible polished surface was observed by naked eyes. This substrate was washed with ethanol and further treated in an ultrasonic cleaning machine at a high frequency of 30 kHz (Made-Telsonic). This frequency is sufficient to remove the adhered impurities from Quartz rod. This step is necessary to ensure the impurity-free surface. Surface impurities may reduce the adhesion of the silver thin film to the Quartz substrate [6]. Next step is the deposition of sensing film. In this process, silver metallo-organic paint (SPI Supplies, USA) was used for making the sensing thin film. This paint was first mixed with thinning agent n-butyl acetate (SPI Supplies, USA) to make the paint of required viscosity. The thermophysical properties of materials are given in Table 1. The sophisticated brush with fine hairs was used to deposit the silver thin film on Quartz substrate. The thickness of the film was kept as minimum as possible (below 10 μm). If the silver film is too thin, it may break during testing, and if it is too thick, the effective resistance of TFG will decrease. The decrement in resistance may cause the decrement in sensitivity, and it may affect the inverse heat flux prediction. The final step is an electrical connection. For this purpose, the silver films were painted on a round surface of the substrate. These connecting thin films were kept intentionally thick compared to sensing thin film. Thickening of connecting thin film will reduce its resistance compared to the sensing thin film. So that resistance of connecting thin film can be neglected. This painted substrate was annealed at 350 °C in microcontroller-based muffle furnace for 30 min to stabilize the thin film. The

In-house Fabrication and Calibration of Silver Thin Film Gauge

199

Fig. 1 Silver thin film gauge a schematic, b photographic view

Table 1 Thermophysical properties of materials [1, 4] Density ρ (kg/m3 )

Material

Thermal conductivity k (W/mK)

Specific heat c (J/kgK)

Thermal product √ ρck (J/m2 s0.5 K)

Quartz

1.40

670

2200

1436.52

Silver

429

233

10490

32381.30

tin-coated copper wire of 0.4 mm diameter was then soldered to connecting films with sufficient care. Tin-coated copper wire was used to ensure the better structural stability of the solder joint. This TFG is then covered with Teflon tape to provide thermal insulation for connecting leads on the non-sensing surface. The in-house fabricated silver TFG by this process is shown in Fig. 1. The effective resistance of this silver TFG was found to be 1.4590  at a temperature of 30 °C.

3 Estimation of Calibration Parameters The performance parameters of TFG were estimated using the oil bath calibration technique. This is also referred to as static calibration. The experimental setup fabricated for static calibration is shown in Fig. 2. The oil bath calibration setup consists of air beaker fixed inside a silicone oil beaker placed on a flat plate heater using clamps. The silicone was selected as a heating medium for air due to its chemical and thermal stability over the wide range of temperature. Both the silver TFG and calibrated glass thermometer were fixed at the same height using a clamp arrangement. Mismatch in height may affect the calibration results due to different temperatures sensed by TFG and thermometer. TFG was first activated by 10 mA current using a constant current source (CCS). To examine the resistance response of silver TFG, the temperature was gradually increased from room temperature of 30 °C to 65 °C. The resistance was

200

A. Jadhav and R. K. Peetala

Fig. 2 Static calibration of TFG

recorded using the data acquisition system (DAS) at an interval of 5 °C during both heating and cooling processes as shown in Fig. 3. The good linear behavior of TFG was observed in static calibration. However, some hysteresis was observed. This is due to the thermal expansion and contraction of sensing thin film during the heating and cooling processes. The coefficient of correlation (r) was calculated using the linear regression analysis, and it was found to be better than 0.90. This reflects the usefulness of this TFG for accurate temperature measurement. The sensitivity (S) of silver TFG was determined using the following correlation which is essentially the slope of the resistance-temperature response of TFG [1]. S =

R (R − R0 ) = T (T − T0 )

(1)

where R is resistance at arbitrary temperature T, and R0 is the initial resistance of TFG at room temperature T 0 of 30 °C. The average values of TCR and sensitivity were Fig. 3 Resistance response of TFG

In-house Fabrication and Calibration of Silver Thin Film Gauge

201

Fig. 4 Estimation of TCR

considered for further study for greater accuracy. The average sensitivity obtained for the present silver gauge is 0.001235 /K. To estimate the TCR, the graph was plotted between the resistance ratio R/R0 and temperature difference T as shown in Fig. 4. The numerical value of TCR was calculated by considering the first-order linear fit equation. This equation was compared with correlation (2), and the average TCR was found to be 0.0008514 K−1 . R = 1 + αT R0

(2)

4 Dynamic Calibration The aim of dynamic calibration is to access the thermal performance of silver TFG when exposed to a constant heat flux. A special experimental device was fabricated for this purpose as shown in Fig. 5. This device consists of a Teflon insulated copper rod (k = 391 W/mK) having a heater at one end, and another end is free to environmental conditions. The heat input to the copper rod can be controlled with the help of a dimmerstat. Five K types of calibrated thermocouples are attached along the axial direction of the rod to monitor the temperature along the length. During calibration, the axial temperature distribution was continuously monitored until the constant heat flux was achieved in the copper rod which is given by the correlation (3). qc = −k

dT dx

(3)

202

A. Jadhav and R. K. Peetala

Fig. 5 Dynamic calibration of TFG

Here, qc is the constant heat flux established in the copper rod (W/m2 ); k is a is a temperature gradient along thermal conductivity of copper rod (W/mK), and dT dx the length of copper rod (K/m). The constant heat flux of 1.5 kW/m2 was achieved in the copper rod. The silver TFG was first activated with a constant current of 10 mA using a CCS, and then, its gentle contact was made to the free surface of copper rod for 1 s. The established constant heat flux was sensed by TFG, and its resistance varied in a parabolic pattern. This transient resistance signal was then converted to a transient temperature signal using correlation (2) as shown in Fig. 6. This parabolic nature of the temperature signal confirms the constant heat flux condition. Heat flux was predicted from transient temperature signal using inverse analysis. In this analysis, heat transfer through TFG was modeled as heat transfer in semiinfinite solid. The following assumptions were considered in this analysis. (i) The temperature of the sensing element and substrate surface is the same due to the thinness of film. (ii) Heat conduction takes only in the axial direction, and lateral heat conduction is neglected. (iii) Thermophysical properties of sensing and substrate materials are constant within the operating temperature range. (iv) Substrate is infinite so that for test duration of 1 s, the temperature rise at infinity is zero. The general solution for this problem was obtained by Duhamel’s superposition integral given by the following correlation [4–7].  q=

ρs cs ks π

t √ 0

1 d{Ts (τ )} dτ dτ t −τ

(4)

In-house Fabrication and Calibration of Silver Thin Film Gauge

203

Fig. 6 Transient temperature response of TFG

where q is the heat flux; ρ s , cs, and k s denote density, specific heat, and thermal conductivity of substrate material, respectively; t and τ denote experimental time and scaled time, respectively, and T S (τ ) denotes the surface temperature of the substrate. To solve the above equation numerically, first temperature signal needs to be discretized. In the present work, the temperature signal was discretized using the third-order cubic spline technique. Heat flux recovered by this technique was 1.42 kW/m2 . The error in heat flux recovery was 5.33%.

5 Conclusions The silver thin film gauge was fabricated in-house successfully. TCR and sensitivity were obtained by static calibration. However, some fluctuations in resistance–temperature linear behavior and transient temperature data were observed. Further, the thermal performance of silver TFG was assessed through dynamic calibration. Heat flux was under-predicted by silver TFG within the range of 6%. Finally, it was concluded that silver TFG is advantageous due to the low cost of fabrication and better adhesion of thin film to the substrate compared to platinum gauge but at the cost of accuracy in the heat prediction. This TFG can be used for small time scale (s and ms) applications but may not be useful for ultra-short duration (μs) applications such as shock tubes and shock tunnels.

204

A. Jadhav and R. K. Peetala

References 1. Kinnear K, Lu F (1997) Design, calibration and testing of transient thin film heat transfer gauges. In: 20th AIAA advanced measurement and ground testing technology conference, p 2504 2. Kumar R, Sahoo N, Kulkarni V, Singh A (2011) Laser based calibration technique of thin film gauges for short duration transient measurements. J Therm Sci Eng Appl 3(4):044504 3. Smith DE, Bubb JV, Popp O, Diller TE, Hevey SJ (1999) A comparison of radiation versus convection calibration of thin-film heat flux gauges. ASME-Publications 364:79–84 4. Sahoo N, Peetala RK (2011) Transient surface heating rates from a nickel film sensor using inverse analysis. Int J Heat Mass Transf 54(5–6):1297–1302 5. Niranjan S, Peetala RK (2010) Transient temperature data analysis for a supersonic flight test. J Heat Transf 132(8):084503 6. Sarma S, Sahoo N, Unal A (2016) Thin-film gauges using carbon nanotubes as composite layers. J Eng Mater Technol 138(4):041014 7. Jadhav A, Peetala R, Kulkarni V, Multi-walled carbon nano-tubes for performance enhancement of thin film heat flux sensors. Heat Mass Transf

Study of Shock Wave Boundary Layer Interaction in Hypersonic Flows Using Various Turbulence Models Aniruddha Kane

and Ravi K. Peetala

Abstract Shock wave boundary layer interaction (SWBLI) is an important phenomenon in hypersonic vehicle designing. For accurate predictions of the thermal and pressure loads due to SWBLI turbulent flow analysis is essential. In the present study two dimensional ramp induced SWBLI in hypersonic flows is investigated by using SA and k–ω SST turbulence models. In the present study it has been found that the prediction of the separation size is nearer to experimental results by using SA model as compared to k–ω SST model. Keywords SWBLI · Turbulent flows · Hypersonic flow

1 Introduction Shock wave boundary layer interaction (SWBLI) and shock-shock interactions are the prime topics of research in supersonic and hypersonic flow analysis research [1]. SWBLI is commonly observed in various parts of the vehicle such as wing-body junctions, inlets, wedge like sections etc. These interactions enhance the heating and pressure loads on the vehicle body and can damage it [2]. Thus it is necessary to predict these loads accurately in order to utilize these predicted loads for designing suitable thermal protection system for the vehicle. For understanding SWBLI, analyzing flow over complete vehicle is not required but instead canonical geometries such as ramp, wedge, double wedge are frequently used in the studies [3]. Supersonic and hypersonic flow over flat plate, ramp has been analyzed experimentally by few researchers to study the wall pressure variation, condition of incipient separation and length of flow induced separation [4–6]. The response of turbulent boundary layer to A. Kane (B) · R. K. Peetala Department of Mechanical Engineering, Visvesvaraya National Institute of Technology, Nagpur 440010, India e-mail: [email protected] R. K. Peetala e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_24

205

206

A. Kane and R. K. Peetala

the shock wave has been studied by Smits and Muck [7]. The shock induced separation can also be studied by numerical modelling techniques and thus capabilities and limitations of various models must be evaluated. Knight and Degrez [8] and Knight et al. [9] reviewed capabilities of various RANS based Navier-Stokes codes to predict pressure field, skin friction, heat transfer, shock-induced separation extent for both laminar and turbulent shock boundary layer interactions. Hirsch [10] reviewed the numerical work done on SWBLI considering a variety of flow solvers, grids and turbulence models for, particularly impinging-reflecting shock case. It is observed from the literature, reattachment of the separated boundary layer is turbulent for stronger shocks which rises in hypersonic flow regime. Thus it is necessary to study the behavior of shock wave and turbulent boundary layer interaction occurring in hypersonic regime. It is also observed that the RANS based turbulence models such as Spalart-Allmaras one equation model and k–ω SST model are suitable in case of flows with separated boundary layer. Thus the objective of the present investigation is the study of shock induced boundary layer separation in hypersonic flow over a 2D ramp using various turbulence models.

2 Governing Equations and Numerical Strategy The integral cartesian form of compressible Navier-Stokes equations is used in order to capture the physics involved in the present investigation. The equations are given below.    ∂ W d V + [F − G] · d A = HdV (1) ∂t ⎧ ⎫ ⎧ ⎧ ⎫ ⎫ ρv 0 ρ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎨ ⎬ ⎨ ⎨ ⎬ ⎬ τxi ρvu + piˆ ρu , G = where W = ,F= ⎪ ⎪ ⎪ ⎪ τ yi ρv ⎪ ρvv + p jˆ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎩ ⎭ ⎩ ⎩ ⎭ ⎭ τi j vi j + q ρE ρvE + pv W is the vector of conservative variables. Vector F represents the convective fluxes while G indicates viscous fluxes.H represents the source terms which are absent in current case. ρ, v, p indicate density, velocity and pressure of the fluid respectively. τ is the viscous stress tensor and q is the heat flux. Total energy (E) and total enthalpy (H o ) are linked as E = H o – p/ρ while H o = h + |v|2 /2 where h is enthalpy of fluid. Present study has been carried out using an open source platform openFOAM. In openFOAM, a transient density based finite volume solver named ρ CentralFoam [11] is employed to solve compressible Navier-Stokes equations. Implicit approach is utilized for temporal discretization while second order, semi discrete, non-staggered upwind schemes of Kurganov and Tadmor [12] is employed for convective flux calculations. The flow of air is modeled by perfect gas law while Sutherland’s model is considered to calculate the viscosity and Prandtl number of 0.7 is considered.

Study of Shock Wave Boundary Layer Interaction in Hypersonic …

207

Turbulence models such as k–ω SST and Spalart-Allmaras models are employed for capturing the actual flow physics of shock induced separation because these models are found to performing well in case of flows with separated boundary layer [13].

3 Results and Discussion Flow over a single ramp model studied by Marini [14] has been adopted for the present investigation. Computational domain consists of a 0.05 m flat plate with a 15° ramp attached to it. Figure 1 illustrates computational domain and boundary conditions. Freestream conditions are mentioned in Table 1. At inlet freestream values of pressure, velocity and temperature are provided. Zero gradient condition is provided at the outlet while freestream condition is employed at the farfield boundary. The wall is considered as isothermal and no slip. Grid is generated with proper clustering near the corner of 2D ramp for capturing the SWBLI effects. Mesh independence study is carried out by considering grid sizes of 180 × 90, 240 × 120, 360 × 180, 420 × 240 cells. Minimum cell size is considered as 1 × 10−05 m with 1.12 as expansion ratio. It is found that 240 × 120 cells give mesh independent solution. The grid and clustered section of 240 × 120 mesh is shown in Figs. 2 and 3 respectively. The high-speed flow with M = 6 reaches the compression corner and there is a formation of an oblique shock. The shock interacts with the boundary layer and submits it to the adverse pressure gradient and causes flow reversal which ultimately

Fig. 1 2D Ramp model

Table 1 Freestream parameters

M∞

6

Ho

1.08 MJ/kg

T∞

131.7 K

P∞

199.4345 Pa

Re∞ (unit)

8 × 105 m−1

208

A. Kane and R. K. Peetala

Fig. 2 Mesh

Fig. 3 Clustered grid near the corner

produces a separation bubble which is seen in Fig. 4. The variation of coefficient of pressure and heat flux over the surface are shown in Figs. 5 and 6 respectively. The wall shear stress and ultimately skin friction coefficient (C f ) becomes negative where the flow separation starts and again attains positive value at reattachment point. The length of the separation bubble is the length between separation and reattachment points. The plot of skin friction coefficient is compared for both the turbulent models as presented in Fig. 7. Figure 5 depicts both turbulent models captures the physics of the shock induced separation qualitatively. But quantitatively Spalart-Allmaras (SA) model gives some closer results. Heat flux prediction of k–ω SST model is higher than SA model. k–ω SST model predicts smallest separated region. Figure 8 shows the length of separation bubble as captured by two models with respect to experimental value. Error reported while predicting the lengths of the bubble for SA and k–ω SST models are 26 and 77%. These large errors may be because of the higher value of turbulent intensity of 5% considered which otherwise is in the range of 0.5–1% for high speed flows. In the present investigation 5% of turbulent intensity

Fig. 4 Mach contour and enlarged view of the boundary layer separation

Study of Shock Wave Boundary Layer Interaction in Hypersonic …

209

Fig. 5 Coefficient of pressure variation

Fig. 6 Heat flux variation

is considered as per the literature [15]. But the inaccuracy occurring in the present results indicates that the value of turbulent intensity is application specific and for the current case lower value may provide appropriate results. Thus there is need to study the effect of turbulent intensity on the shock induced turbulent boundary layer separation.

210

A. Kane and R. K. Peetala

Fig. 7 Skin friction variation

Bubble Size (mm)

20 16 12 8 4 0

Experimental

Spalart Allmaras

K-omega SST

Fig. 8 Separation size for various techniques

4 Conclusion The investigation of shock induced separation is carried out successfully using two turbulence models. The two models capture the SWBLI effects qualitatively but the quantitative mismatch for size of separation is observed as compared to the experimental results. SA model predicts the separation length to be 13 mm and k–ω SST model predicts very small separation length of 4 mm against the experimental value of 17.75 mm. The error reported in the prediction of the lengths of the bubble for SA and k–ω SST models are 26 and 77%. These large errors may be because of the higher value of turbulent intensity of 5% considered which otherwise is in the range of 0.5–1% for high velocities. The study will be extended to investigate the effect of turbulence intensity on SWBLI.

Study of Shock Wave Boundary Layer Interaction in Hypersonic …

211

References 1. Bertin JJ, Cummings RM (2003) Fifty years of hypersonics: where we’ve been, where we’re going. Prog Aerosp Sci 39(6–7):511–536 2. Gaitonde DV (2015) Progress in shock wave/boundary layer interactions. Prog Aerosp Sci 72:80–99 3. Délery J, Dussauge JP (2009) Some physical aspects of shock wave/boundary layer interactions. Shock Waves 19(6):453–468 4. Gadd GE, Holder DW, Regan JD (1954) An experimental investigation of the interaction between shock waves and boundary layers. Proc Royal Soc A Maths Phys Eng Sci 226(11650):227–253 5. Greber I, Hakkinen RJ, Trilling L (1958) Laminar boundary layer oblique shock wave interaction on flat and curved plates. J Appl Mathe Phys (ZAMP) 9:312. https://doi.org/10.1007/ BF02424755 6. Needham DA, Stollery JL (1966) Boundary layer separation in hypersonic flow. AIAA 4th Aerosp Sci Meeting 66:66–455 7. Smits AJ, Muck KC (1987) Experimental study of three shock wave/turbulent boundary layer interactions. J Fluid Mech 182(1) 8. Knight D, Degrez G (1998) Shock wave boundary layer interactions in high mach number flows a critical survey of current numerical prediction capabilities. Agard Advisory. Report, 2 9. Knight D, Yan H, Panaras AG, Zheltovodov A (2003) Advances in CFD prediction of shock wave turbulent boundary layer interactions. Prog Aerosp Sci 39(3):121–184 10. Hirsch C (2010) Lessons Learned from the first AIAA-SWBLI Workshop CFD Simulations of Two Test Cases. AIAA Conference pp 1–14 11. Greenshields GJ, Weller HG, Gasparini L, Reese JM (2010) Implementation of semi-discrete, non-staggered central schemes in a colocated, polyhedral, finite volume framework, for highspeed viscous flows. Int J Numeri Meth Fluids 63(1):1–21 12. Kurganov, Tadmor E (2000) New high-resolution central schemes for nonlinear conservation laws and convection-diffusion equations. J Comput Phys 160(1):241–282 13. Babinsky H, Harvey JK (2011) Shock wave–boundary-layer interactions. Cambridge University Press 14. Marini M (1998) Effects of flow and geometry parameters on shock-wave boundary-layer interaction phenomena. AIAA J 98(1570):319–329 15. John B, Senthilkumar P (2018) Alterations of cowl lip for the improvement of supersonic-intake performance. J Appl Fluid Mech 11(1)

Study of Effect on Engine Performance Using 15% HCNG Blend Versus CNG Using a Simulation Approach K. P. Kavathekar , S. S. Thipse , S. D. Rairikar , S. B. Sonawane , P. S. Sutar , and D. Bandyopadhyay

Abstract The present work deals with modeling an existing CNG engine using a 1D simulation tool (GT-SUITE), performing baseline validation of the model, generating simulation results of HCNG (15% hydrogen by volume), study of simulated results to bring out various trends and studying the effects of varying fuel–air equivalence ratio on optimum spark advance, peak in-cylinder pressure and peak in-cylinder temperature through simulation on validated engine model. This work is intended as a preliminary step to bring about a host of potential suggestions, which will help to improve the current power performance of the engine while using 15 HCNG as the fuel. The primary objective is to develop 1D engine model using the actual engine data as input and to generate simulation results. Actual experiments will be carried out to generate experimental results. Simulation model will be validated for accuracy within 10%. The validated engine model is now used for the purpose of simulation to study the effects of varying fuel–air equivalence ratio on optimum spark advance, peak in-cylinder pressure, and power performance for selected HCNG blend. Keywords HCNG · 1D simulation

1 Introduction Over the preceding decade, natural gas has turned out to be a demanding alternative fuel for the growth of sustainable transportation. Heavy commercial vehicles plying on natural gas represent a budding technology, while natural gas heavy-duty transport vehicles are widely popular. As a fuel source for large scale automotive applications, natural gas provides rewards in automotive technology because of its socioeconomic benefits in comparison with fossil-fueled engines. K. P. Kavathekar (B) Symbiosis International University, Pune, India e-mail: [email protected]; [email protected] K. P. Kavathekar · S. S. Thipse · S. D. Rairikar · S. B. Sonawane · P. S. Sutar · D. Bandyopadhyay Powertrain Engineering Lab, Automotive Research Association of India (ARAI), Pune, India © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_25

213

214

K. P. Kavathekar et al.

There is huge potential of hydrogenated fuel to increase combustion efficiency and minimize emissions in natural gas engines. Hydrogen blends with natural gas evenly as both are in a gaseous state at STP. Hydrogen has a greater stoichiometric laminar flame speed in air than natural gas.

2 Engine Model Development A four-cylinder naturally aspirated CNG engine is modelled in GT-POWER using all the relevant templates and is used for baseline validation with test data. The validated model is then used to simulate the results when HCNG is used as the fuel instead of CNG. This simulation data is further validated through experimentation. The validated model is then further used to study the effects of various engine parameters like spark advance, fuel–air equivalence ratio, fuel injection timing, and compression ratio on the engine performance. The final 1D Engine model is shown in Fig. 1. In this model, CNG was defined using the standard species like methane vapor, ethane vapor, propane vapor, n-butane vapor, and nitrogen vapor available in the GT library in typical proportions by volume [1]. In this model, since different blends of HCNG will be used in the simulation, the user can define any fuel by using the ‘FluidMixture’ template by either choosing from the standard fuels available in the GT library or by a combination of two or more species available in the GT library.

Fig. 1 Complete engine 1D simulation model

Study of Effect on Engine Performance Using 15% HCNG Blend …

215

7 6

2

5 1

4

8

Servo motor

3

Signal Conditioning

Exhaust

1. Cat-con 3. Exhaust 5. HCNG Injectors 7. Pressure Regulator

2. Intake 4. Muffler 6. Gas Flow Meter 8. HCNG Cascade

Fig. 2 Schematic of test setup

3 Experimental Setup The CNG engine test setup was used to generate test data with HCNG blends as the fuel. It can be seen from the experimental setup that, the blend (Hydrogen and CNG) was supplied to the engine intake manifold [2]. The storage pressure of HCNG was 200 bars. The experimental setup schematics is as shown in Fig. 2. A four-cylinder NA CNG-fueled engine was used for the test. Full throttle performance (FTP) tests were carried out for 15% HCNG fuel throughout the engine speed range. This FTP data was then used to validate the CNG 1D model with CNG test data.

4 Baseline Model Validation After validation of base CNG engine, and before using the HCNG model, it was first necessary to optimize the spark advance for the given engine specifications and each at each selected engine speed. Brake torque was selected as the dependent variable to find the optimum spark advance at each selected engine speed. Optimum spark advance at any given engine speed is that spark timing which yields maximum brake torque [3].

216

K. P. Kavathekar et al.

4.1 Prediction Using Selected 15% HCNG Blend as Fuel After completion of the baseline validation of the model with CNG fuel, the ‘fuel object’ attribute in fuel injector object of 1D model was altered for prediction of HCNG blend (15% by volume).

4.2 Simulation Prediction for 15 HCNG Blend with Following Parameters

Effect of varying fuel–air equivalence ratio (φ) on engine performance for 15 HCNG blend in comparison to CNG Figure 3 shows that if we change the fuel–air equivalence ratio from 1 (stoichiometric) to 0.9 (lean-burn) without making any other modifications to the existing system, there is a drop in the brake torque (10.2% for CNG and 9.98% for 15 HCNG at rated torque speed) values across the entire speed range. The slight difference in the value of the drop for CNG and 15 HCNG suggests that HCNG blends have inherently better ‘lean-burn’ characteristics over CNG, which can be exploited by suitable modifications to the existing system to achieve optimum lean-burn performance using 15 HCNG as the fuel. Figure 4 shows that if we change the fuel–air equivalence ratio from 1 (stoichiometric) to 0.9 (lean-burn) without making any other modifications to the existing system, there is a drop in the brake power values across the entire speed range. In Fig. 5, the BSFC drops (improves) for both CNG (by 0.81% at rated torque speed) and 15 HCNG (by 0.85% at rated torque speed) as the φ is changed from 1 to 0.9 (lean) for almost the entire speed range (except near the rated speed where slightly increases). This may be due to the fact that the engine model is tuned to a stoichiometric operation actual engine and not to a lean-burn actual engine. Hence,

Brake torque in N-m ------>

Simulation prediction of effect of varying fuel-air equivalence ratio on brake torque for 15% HCNG and CNG 190 170

CNG (equivalence ratio of 1) CNG (equivalence ratio of 0.9 i.e. Lean) 15HCNG (equivalence ratio of 1) 15HCNG (equivalence ratio of 0.9 i.e. Lean)

150 130 800

1200

1600

2000

2400

2800

Engine speed in RPM ----->

Fig. 3 Effect of fuel–air equivalence (φ) ratio on brake torque for 15% HCNG and CNG

3200

Brake power in kW ----->

Study of Effect on Engine Performance Using 15% HCNG Blend …

217

Simulation prediction of effect of varying fuel-air equivalence ratio on brake power for 15% HCNG and CNG

55 45 35

CNG (equivalence ratio of 1) CNG (equivalence ratio of 0.9 i.e. Lean) 15HCNG (equivalence ratio of 1) 15HCNG (equivalence ratio of 0.9 i.e. Lean)

25 15 1000

1500

2000

2500

3000

Engine speed in RPM -----> Fig. 4 Effect of varying φ on brake power for 15 HCNG and CNG

Simulation prediction of effect of varying fuel-air equivalence ratio on BSFC for 15% HCNG and CNG BSFC in g/kW hr ----->

250 240 230

CNG (equivalence ratio of 1) CNG (equivalence ratio of 0.9 i.e. Lean) 15HCNG (equivalence ratio of 1) 15HCNG (equivalence ratio of 0.9 i.e. Lean)

220 210 200 1000

1500

2000

2500

3000

Engine speed in RPM -----> Fig. 5 Effect of varying φ on BSFC for 15 HCNG and CNG

the subtle changes, which go into the process of converting a stoichiometric operation actual engine into a lean-burn engine, are not incorporated in this engine model. In Fig. 6, the maximum in-cylinder pressure falls (by 3.63% for CNG and 5.78% for 15 HCNG at the rated speed) with change of fuel–air equivalence ratio from 1 to 0.9. Figure 7 shows that the optimum spark advance must be advanced slightly (for both CNG and 15 HCNG fuels) across the entire speed range when the fuel–air equivalence ratio is changed from 1 to 0.9. This can be attributed to the fact that the flame speed is slightly lower under lean conditions as compared to stoichiometric condition for both the fuels thereby increasing the combustion duration [4, 5]

K. P. Kavathekar et al.

Maximum in-cylinder pressure in bar ---->

218

Simulation prediction of effect of varying fuel-air equivalence ratio on maximum in-cylinder pressure for 15% HCNG and CNG 60

55

CNG (equivalence ratio of 1) CNG (equivalence ratio of 0.9 i.e. Lean) 15HCNG (equivalence ratio of 1) 15HCNG (equivalence ratio of 0.9 i.e. Lean)

50

45 1000

1500

2000

2500

3000

Engine speed in RPM -----> Fig. 6 Effect of varying φ on maximum in-cylinder pressure for 15 HCNG and CNG

Optimum spark advance (BTDC) -->

Simulation prediction of effect of varying fuel-air equivalence ratio on optimum spark advance for 15% HCNG and CNG 1000 0 -5 -10

1500

2000

2500

3000

CNG (equivalence ratio of 1) CNG (equivalence ratio of 0.9 i.e. Lean) 15HCNG (equivalence ratio of 1) 15HCNG (equivalence ratio of 0.9 i.e. Lean)

-15 -20 -25 -30 -35

Engine speed in RPM ----->

Fig. 7 Effect of varying φ on optimum spark advance for 15 HCNG and CNG

5 Conclusions 1. The principal outcome of this study was achieved by obtaining a validated engine model in selected 1D simulation tool (GT-SUITE) for CNG as well as selected HCNG blend. 2. The predicted results showed that there was an encouraging improvement in brake-specific fuel consumption (BSFC) with a small but definite improvement in brake power and in-cylinder pressure while using 15 HCNG as fuel as compared to CNG-fueled operation. 3. For a given fuel, the optimum spark advance needs to be slightly advanced for both, rich and lean mixtures as compared to stoichiometric operation.

Study of Effect on Engine Performance Using 15% HCNG Blend …

219

4. The baseline engine model (CNG as fuel) was validated experimentally with the simulation results following the same trend as the experimental results and lie within an accuracy of 6.8% (maximum) across the entire speed range. Acknowledgements Author would like to thank to The Automotive Research Association of India (ARAI), Pune, and Symbiosis International University, Pune, for their support and inspiration to work up on this exclusive area. Author also thanks Mr. Suyash Gawade for his support in simulation work.

References 1. GT Suite user guide, flow theory, GEM-3D and Engine Performance user manuals, Version v7.4 2. Kahraman N et al (2009) Investigation of combustion characteristics and emissions in a sparkignition engine fuelled with natural gas-hydrogen blends. Int J Hydrogen Energy 34:1026–1034 3. Das LM et al (2005) Experimental evaluation of a hydrogen added natural gas (HANG) operated SI engine, SAE Paper No. 2005-26-029 4. Ma F et al (2012) Effect of compression ratio and spark timing on the power performance and combustion characteristics of an HCNG engine. Int J Hydrogen Energy 37:18486–18491 5. Zhao J et al (2013) Effects of compression ratio on the combustion and emissions of a hydrogen enriched natural gas engine under different excess air ratio. Energy 59:658–665

Behaviour of NiTi Based Smart Actuator for the Development of Planar Parallel Micro-Motion Stage Deep Singh , Yogesh Singh , and Manidipto Mukherjee

Abstract Recently, the application of smart materials such as shape memory alloys (SMAs) as actuators is gaining huge importance. SMA-based actuators are light in weight and provide higher work/mass. SMA undergoes simple actuation process such as Joule heating. Nitinol (NiTi) SMA can restore larger strains as compared to others. It can serve as active prismatic joint to provide linear motion in various robotic manipulators. The study correlates the deflection of (a) single NiTi spring and (b) series-connected NiTi springs with input parameters (time and current) to understand its behavioural complexity. The study revealed that the rate of NiTi spring contraction is dependent on time and current. To predict the actuation motion, several regression models were developed. This study defined the feasible current range for the actuation of NiTi spring based on contraction rate and precision. The contraction rate for single NiTi spring differs from the series connection of two NiTi springs which results a new set of polynomial regression model. The developed mathematical models can help control the smart actuation-based planar parallel robotic manipulators. Keywords Shape memory alloy · Smart actuator · Nitinol · Mathematical model

1 Introduction Currently, the need of smart materials is emerging in every field due to its amazing behaviour to change itself on application of external parameters such as temperature, stress, electric and magnetic fields [1]. Shape memory alloys (SMAs) are smart materials which can restore its memorized shape when heated till the austenite region from martensite region due to change in crystal orientation from monoclinic to the body-centred cubic structure. SMAs are being widely used in many applications D. Singh (B) · Y. Singh Department of Mechanical Engineering, NIT Silchar, Assam, India e-mail: [email protected] M. Mukherjee Advanced Manufacturing Centre, CSIR-CMERI, Durgapur, West Bengal, India © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_26

221

222

D. Singh et al.

such as biomedical, commercial and aerospace industries with one of its important application as actuators [2]. SMA actuators are very compact with higher power/mass ratio, silent operation, low voltage activation and most importantly biocompatibility [2, 3]. In the reference [4], the authors studied the workspace of the three-legged Ushaped base 3PRP planar robotic manipulator experimentally by implementing nitinol as an actuator. The workspace was identified based on full contraction of SMA actuator, which incurs difficulty in control of end-effector. The desired end-effector position can be achieved by the displacement control of actuator. It can be achieved by correlating the contraction rate of SMA with respect to current and time. However, one of the main drawbacks of SMAs is the hysteresis which makes complications in controlling the displacement because of its continuous nonlinear change in length, particularly, when the micro-motion is in consideration. The displacement behaviour of SMA springs under Joule heating is merely studied. Therefore, in the present study, the correlation of the contraction of NiTi springs with current and time as input variables was determined to understand its behavioural complexity. Also, the series connection behaviour of the SMA spring was analysed.

2 Experimental Procedure 2.1 Material Selection and Specification The selected actuation material is the readily available NiTi SMA spring of 0.75 mm diameter with 19 helix windings. The general composition of NiTi is 50 wt% Ni and 50 wt% Ti. NiTi SMA spring has the ability to provide a maximum strain of up to 8% which is highest as compared to other SMA materials [5].

2.2 DC Power Supply In order to provide variable DC current to the NiTi SMA spring, a variable DC power supply has been developed and the schematic circuit diagram is shown in Fig. 1(a). The developed variable DC power supply, used in the study, is shown in Fig. 1(b). The necessity of the variable current is to understand the behaviour of the contraction (deflection) of springs at various current values to predict the displacement necessary for the manipulator. The accuracy of the DC variable power supply is 10 mA.

Behaviour of NiTi Based Smart Actuator for the Development …

223

Fig. 1 Variable DC power supply (a) circuit diagram and (b) prototype

2.3 SMA Springs in Series Connection In order to understand the behaviour of the SMA springs in series, two NiTi springs were taken. One end of first spring was fixed at the origin (0, 0), and one end of second spring was fixed at location B (l, 0) where l = 129 mm. The other end of the two springs was connected together at point A (r, 0) with the help of a pin made of ABS plastic to restrict the flow of current from one spring to other as depicted in Fig. 2. Initially, the second spring is contracted fully by supplying electrical current and then allowed to cool down to room temperature. Then, constant current is supplied to the first spring, and the position of A is noted at an interval of 10 s. Similarly, different values of constant current were supplied to the spring with the help of variable DC power supply (as discussed in Sect. 2.2), and the position of A was determined.

Fig. 2 NiTi SMA spring connected in series

224

D. Singh et al.

Table 1 Category of current based on contraction speed Category

Actuation and precision

Current range (I mA)

Contraction rate (Vc mm/s)

I

Rapid actuation

1080 < I

0.59 < Vc

II

Fast actuation and low precision

930 < I < 1080

0.23 < Vc 0.59

III

Slow actuation and high precision

700 < I < 930

0.10 < Vc 0.23

IV

Infeasible region

I < 700

Vc < 0.10

3 Results and Discussion 3.1 Determination of Current–Time Domain for Single-Spring Connection In this study, an effort has been made to control the displacement of the SMA springbased linear actuator, so that it can effectively control the end-effector motion in micron level. The end-effector motion is directly controlled by SMA actuator. In the experiment, it was observed that the SMA actuator contracted by applying an electric current from its initial elongated length (L 0 ) of 100 mm. In this study, a particular current value was supplied to the spring, and the contraction time was noted for a standard contraction length (L c ) of 71 mm. In the entire experiment, the current was varied from 600 to 1220 mA in 15 steps to observe the total deflection time required to achieve Lc . The result showed that the current range can be classified into four different categories based on spring contraction rate as listed in Table 1. Table 1 indicates that for higher current, the contraction rate of spring is higher and precision is lower and vice versa. The infeasible region (category IV) indicates the current range which cannot be utilized in the application because of incomplete contraction over time.

3.2 Nonlinear Exponential Models for Current and Deflection Correlated with Time for Single-Spring Connection The experimental data as discussed in Sect. 3.1 were used to develop a fitted nonlinear exponential mathematical model between current (I) and time (t). The nonlinear exponential fitted Eq. (1) has an adjusted regression value of 0.97442 which clearly indicates the acceptability of the equation with approximately 97.442% fit. Figure 3 depicts Eq. (1) as an exponential curve of negative slope which indicates that with decrease in current, the time required for the same contraction length (L c ) is higher. The slope is higher initially and decreases gradually over time which proves that at

Behaviour of NiTi Based Smart Actuator for the Development …

225

Fig. 3 Current (mA) versus time (s) nonlinear exponential plot

higher current values, the contraction rate is higher as compared to lower current values. Also, at lower currents (I < 700 mA), the slope becomes parallel to time axis which clearly indicates zero contraction rate indicating the infeasible region as discussed in Sect. 3.1. The graph can be used to interpret the time required for full contraction (L f ) at any given current. I = 659.8844 + 683.8789 × exp (−0.00386 × t)

(1)

3.3 Second-Order Quadratic Polynomial Regression Model for Deflection Control of Single-Spring Connection A full quadratic polynomial model was obtained using the experimental procedure, which includes current versus time and deflection versus time data for distinct values of current in Minitab 17. The second-order quadratic polynomial regression model for the response of deflection (d) has been expressed as a function of two process parameters—current (I) and time (t). The following full quadratic polynomial Eq. (2) is obtained with an adjusted regression value of 0.9204 and zero probability which is clear indication of the adequacy for the developed regression equation.   d = −30.6349 + 0.4427 × t + 0.2964 × I + 0.0002 t 2 − I 2 − 0.0008 × t × I (2) Equation (2) is very helpful in correlating the deflection with respect to current and time, the only parameters necessary for the development of SMA spring-based linear actuator. Hence, the time required for a certain deflection based on any fixed current can easily be interpreted based on Eq. (2).

226

D. Singh et al.

Fig. 4 Current versus time plot within deflection range of 30–70 mm

3.4 Predicted Deflection Trend of NiTi SMA Single-Spring Connection The developed quadratic polynomial Eq. (2) presents the deflection trend by plotting the deflection curve in a 2D current—time graph as shown in Fig. 4. The figure shows the deflection trend of limiting displacement as d = 70 and 30 mm. This typical graph depicts the amount of time necessary for a distinct given current to obtain any desired deflection. The curves representing d = 70 mm and d = 30 mm have negative slope which clearly reflects that to attain any desired deflection, the time increases with decrease in current and vice versa. It can also be stated that the slope will become zero after a certain time and the time required for full contraction is infinite which depicts the infeasibility region as discussed in Sects. 3.1 and 3.2.

3.5 Second-Order Quadratic Polynomial Regression Model for Deflection Control of SMA Spring in Series Connection In order to obtain a three DOF planar micro-motion stage, two SMA springs are connected with a pin as shown in Fig. 2. The developed nonlinear polynomial Eq. (2) was applied to control the deflection of springs. Surprisingly, the deflection of the series-connected springs was deviated from the usual as the inherent resistance of the springs acted opposite to each other. The derived single-spring equation failed to describe the deflection behaviour of the series-connected SMA springs. Hence, a series of experiments were conducted again using the same current–time domain. The deflection data of series connection at an interval of ten seconds were noted for distinct supply of current. These data were fed to develop the second-order quadratic polynomial model similar to single-spring connection as described in Sect. 3.3.

Behaviour of NiTi Based Smart Actuator for the Development …

227

ds = 57.513 − 8.674 × I − 18.204 × t − 1.246 × I 2 + 2.126 × t 2 − 1.464 × I × t

(3)

The second-order quadratic polynomial regression model for the deflection (d s ) of series-connected spring has been expressed as a function of two process parameters— current (I) and time (t). The following full quadratic polynomial Eq. (3) has an adjusted regression value of 0.8214 and the probability value less than 0.05 which clearly indicates the significance of the quadratic polynomial equation. Equation (3) is very helpful in correlating the deflection curve or response with current and time for a series connection. Hence, the time required for a certain deflection based on any fixed current can easily be determined using Eq. (3).

4 Conclusion The major conclusions of this study have been outlined below: 1. Based on the contraction rate, the current has been categorized into four categories (I, II, III and IV), each having its own significance. 2. The slope of a constant deflection curve decreases with decrease in current over time and reaches critical current (I < 700 mA), an infeasibility region, below which the deflection is not possible. 3. The contraction rate of SMA spring increases with increase in current and vice versa. Also, the current has an exponential variation with time for any constant deflection. These relationships were best understood based on the developed second-order quadratic polynomial model. The polynomial equation is able to predict the response adequately as checked by ANOVA. 4. Another second-order polynomial model was derived for the series-connected SMA springs which is able to predict the response adequately as checked by ANOVA. Therefore, it is concluded that the polynomial equations obtained by the experiments will help predict the displacement for distinct current values and can further be controlled to stop at the user requirement point (displacement). This experiment helps design a SMA spring-based linear actuator which is lower in cost, friction and noise free. This actuator can be used for various micro-positioning robotic applications. Acknowledgements Authors acknowledge TEQIP-III under National Institute of Technology, Silchar, for financial support.

228

D. Singh et al.

References 1. Drossel WG, Kunze H, Bucht A, Weisheit L, Pagel K (2015) Smart—smart materials for smart applications. Procedia CIRP 36:211–216 2. Miková L, Medvecká-Beˇnová S, Kelemen M, Trebuˇna F, Virgala I (2015) Application of shape memory alloy (SMA) as actuator. METABK 54(1):169–172 3. Spaggiari A, Castagnetti D, Golinelli N, Dragoni E, Scirè Mammano G (2019) Smart materials: Properties, design and mechatronic applications. In: Proceedings of the institution of mechanical engineers, part l: journal of materials: design and applications 233(4):734–762 4. Singh Y, Mohan S (2017) Development of a planar 3PRP planar parallel manipulator using shape memory alloy spring based actuators. In: Proceedings of the advances in robotics (Proceeding AIR’17), 10 5. DesRoches R, McCormick J, Delemont M (2004) Cyclical properties of superelastic shape memory alloys. ASCE J Struct Eng 130(1):38–46

Multi-objective Optimization of Inconel 718 Using Combined Approach of Taguchi—Grey Relational Analysis Manav Sheth, Kunj Gajjar, Aryan Jain, Vrund Shah, Het Patel, Rakesh Chaudhari , and Jay Vora

Abstract Nickel-based alloy such as Inconel 718 is widely used in aerospace, automobiles, gas turbines, nuclear and chemical industries. Inconel 718 is a high-strength temperature resistance which exhibits good resistance to corrosion. In the current paper, Taguchi’s L9 orthogonal array is implemented for wire electrical discharge machining (WEDM) with three factors at three levels. The influence of input variables such as pulse-on time, current and pulse-off time has been investigated on the material removal rate and surface finish. ANOVA analysis has been carried out to check the significance of variables and their effect on output variables. For MRR, all three input parameters are found to be significant pulse-on time and the current is found to have an influence on SR. As Taguchi’s technique can optimize only one objective at a time with no consideration of its effect on another output parameter which may result in either lower production or pitiable quality. To satisfy such conflicting objectives at the same time, an optimum parameter setting is required. Grey relational analysis was used to get an optimal combination of input variables for multiple output variables. The optimal combination of input parameters is found to be pulse-on time 55 µs, pulse-off time 5 µs and current 2 A. Predicted values obtained at an optimal condition using GRA have been validated by experimental trial and show a very close relationship with negligible error. Keywords Inconel 718 · WEDM · ANOVA · Grey relational analysis

1 Introduction A superalloy is categorized as a high-performance alloy which can withstand high temperature and excellent resistance in mechanical and chemical degradations. A nickel-based superalloy, Inconel 718, is having different uses in automobiles, chemical and nuclear industries, gas turbines and aerospace. Inconel 718 possesses M. Sheth · K. Gajjar · A. Jain · V. Shah · H. Patel · R. Chaudhari (B) · J. Vora Department of Mechanical Engineering, Pandit Deendayal Petroleum University, Gandhinagar, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_27

229

230

M. Sheth et al.

high-strength temperature-resistant property which provides high resistance to oxidation and corrosion in addition to the fatigue endurance limit and high creep rupture strength [1]. For obtaining dimensional accuracy along with good surface integrity for aircraft components, aircraft manufacturers are enforced to look for an alternative way for achieving improved surface roughness (SR) and high production rate. The conventional machining of nickel-based alloys exhibits poor machinability due to their low thermal conductivity. In addition, the conventional machining provides unwanted tool wear as such traditional machining consists of carbides and hard abrasive particles which tend to poor surface profile of the final product [1–3]. Thus, non-conventional processes are suitable to overcome the issues related that of conventional machining techniques for nickel-based alloys. Wire electrical discharge machining (WEDM) is widely accepted untraditional machining technique in many industries. It has a capacity to machine any complicated and intricate patterns and shapes with better accurate and precise results [4, 5]. However, WEDM process involves large number input parameters and precise controlling of these parameters which is required to obtain required geometry with subsequent mechanical properties. Agrawal et al. [1] conducted experiment on parametric modelling and optimization for WEDM using RSM which showed that pulse-on time (T on ) is major contributors for obtaining SR. Yang et al. [2] have investigated the single and multi-objective optimization for WEDM process and used Taguchi method in conjunction with the grey relational analysis (GRA) to work on WEDM parameters response on the material removal rate (MRR) and SR. Dabade and Karidkar [3] concluded that T on is the dominant factor which is responsible for SR and MRR conditions. Chaudhari et al. [6] concluded that T on , pulse-off time (T off ) and current are main influencing factors for response variables of MRR, SR and microhardness for WEDM machining of nickel-based alloys. Taguchi’s technique can optimize only one objective at a time with no consideration of its effect on another output parameter which may result in either lower production or pitiable quality. To satisfy such conflicting objectives at the same time, an optimum parameter setting is required. One of the optimization techniques such as GRA is implemented for the optimization of multiple performance variables. Lin et al. [7] concluded that GRA is more simple and easy as compared to fuzzy-based Taguchi’s technique for simultaneous optimization. Rajyalakshmi and Ramaiah [8] applied the grey relational analysis—Taguchi method—which shows an increased value of the MRR and reduced value of SR. In the present work, the effect of different parameters such as T on , T off and current on SR and material removal rate (MRR) has been explored for Inconel 718. Taguchi’s L9 orthogonal array is used for experimentation along with GRA technique for simultaneous optimization of output response. ANOVA was used for determining the effect and contribution of the individual input variables on the individual output responses. Finally, confirmation experimentation was conducted to validate the result obtained from GRA.

Multi-objective Optimization of Inconel 718 …

231

Table 1 Experimental values for MRR and SR along with GRG Run

A (µs)

1

45

2

45

3

45

4

55

5 6

B (µs)

C (A)

MRR (mm3 /s)

SR (µm)

MRR grey relational coefficients

SR grey relational coefficients

Grey relational grade

5

2

0.7794

5.129

10

3

0.6889

6.014

0.605

1.000

0.803

0.535

0.528

0.531

15

4

0.6116

5

3

1.0204

6.987

0.481

0.366

0.423

5.998

0.851

0.532

0.692

55

10

4

55

15

2

0.8612

7.014

0.678

0.363

0.520

0.3639

5.319

0.333

0.830

7

65

5

0.582

4

1.1265

7.324

1.000

0.333

0.667

8

65

9

65

10

2

0.5311

5.904

0.429

0.559

0.494

15

3

0.5210

6.574

0.423

0.418

0.420

2 Experimental Set-up and Experimentation In the present study, a cylindrical bar of Inconel 718 having a diameter of mm with density equals to 8.19 g/cm3 was utilized to machine the specimens, each of mm thickness using a concord wire-cut EDM machine DK7732. Molybdenum wire with 0.18 mm diameter was used as electrode material. Three machining parameters, T on , T off and discharge current are selected on the machining process responses for MRR and SR. Taguchi’s L9 orthogonal array is used for experimentation with three control factors with three levels each. MRR was calculated by using difference in weight before and after the machining of the sample divided by time taken for the machining. The surface roughness measurement was carried out by using Mitutoyo that makes Surftest SJ410 model. Experimentally measured values of MRR and SR for selected nine experimental trials were shown in Table 1.

3 Results and Discussion 3.1 MRR Experimentally measured values of MRR for selected nine experimental trials are shown in Table 1. Higher value of MRR is desirable outcome for obtaining higher productivity. For the experimental data analysis, statistical software Minitab 14 has been used. ANOVA was implemented with 95% confidence level to determine the influence of input parameters on each output variable and is shown in Table 2 for MRR. Factor B (T off ) found to be most significant effect on MRR with 69.77% contribution followed by factor C (Current) and factor A (T on ) with 29.27% contribution and 0.94% contribution, respectively. The value of P should not be greater than 0.05

232

M. Sheth et al.

Table 2 Analysis of variance for MRR Source

DF

SS

MS

F

P

Contribution (%)

A

2

0.004624

0.002312

72.02

0.014

0.94

B

2

0.344527

0.172263

5366.34

0.000

69.77

C

2

0.144537

0.072269

2251.31

0.000

29.27

Error

2

0.000064

0.000032

Total

8

0.493752

0.02 100

R-sq = 99.99%, R-sq (adj) = 99.95%, R-sq (pred) = 99.74%

Fig. 1 Main effect plot for MRR and SR

to keep the particular input parameter significant for the 95% confidence level [9]. All the selected input process parameters are found to be significant for MRR. The model is found to significant for MRR as an extremely close relation was observed between the R-squared values. Figure 1 shows main effect plot for MRR considering the variation in levels of all input parameters. From Fig. 1, with increase current, MRR value is found to be increased due to an increased value of discharge energy, whereas increase in pulse-off time reduces the discharge energy which in turn also reduces MRR. The discharge energy and spark intensity increase with increase in pulse-on time, which will remove more materials and gives higher MRR [5].

3.2 SR In Taguchi method, the SR is in the category of lower the better performance characteristics. ANOVA for surface roughness is shown in Table 3. As seen in Table 3, factor C has a most significant effect on the SR with 87.11% contribution followed by factors A and B with 11.7% and 0.98% contributions, respectively. For SR, factors A and B are found to be significant while factor B is insignificant. The model is found to significant for SR as an extreme close relation was observed between the R-squared values. Main effect plot for SR considering the variation in level of input parameters is shown in Fig. 1. Figure 1 shows that as T on increases, value of SR is

Multi-objective Optimization of Inconel 718 …

233

Table 3 Analysis of variance for SR Source

DF

SS

MS

A

2

0.55554

0.27777

F

B

2

0.04646

0.02323

4.81

0.172

0.98

C

2

4.13596

2.06798

427.82

0.002

87.11

Error

2

0.00967

0.00483

Total

8

4.74762

57.46

P

Contribution (%)

0.017

11.70

0.21 100

R-sq = 99.80%, R-sq (adj) = 99.19%, R-sq (pred) = 95.88%

found to be increased. Because of the T on increases, the discharge energy increases; this in turn increases the rate of melting. This leads to increase in crater size and gives higher SR. Increase in the SR has been observed with an increase in pulse-off time (in the range of 10–15 µs) and beyond 15 µs value of pulse-off time, SR was found to be slightly decreased. The SR seems to increase with increase in current. At higher currents, ionization of deionized water takes place, which leads to high discharge and thermal energy, creating larger and deeper craters, and increasing SR [5]. The optimal combination of input parameters for MRR is obtained as A2 B1 C3 . From the experiments in L9 orthogonal array, run 7 with the combination A3 B1 C3 produced the best MRR of 1.1265 mm3 /s. Further, the optimal combination of input parameters for SR is obtained as A3 B2 C3 . Run 1 with the combination A1 B1 C1 produced the best SR of 5.129 µm. A conflict has been observed for the two quality characteristics for optimal factor levels. This arises a need of optimization of two quality objectives for optimal solution.

3.3 Multi-objective Optimization Using Taguchi’s Grey Method As Taguchi’s technique can optimize only one objective at a time with no consideration of its effect on another output parameter which may result in either lower production or pitiable quality. To satisfy such conflicting objectives at the same time, an optimum parameter setting is required. By using GRA, the problem is improved into a single objective problem. For the optimization of two quality characteristics for Inconel 718 with WEDM process, the following steps are outlined: Step 1 Step 2 Step 3 Step 4 Step 5

Determination of S/N ratio Normalization of S/N ratios for MRR and SR Calculation of grey relational coefficient (GRC) Calculation of grey relational grades (GRG) Calculation of response table for GRG.

234 Table 4 Factor response table

M. Sheth et al. Source

Level 1

Level 2

Level 3

A

0.5856

0.5979

0.5269

B

0.7203

0.5151

0.4750

C

0.6261

0.5477

0.5366

Table 5 Confirmation experiments

Optimal machining parameters Predicted

Experimental

Levels

A2 B1 C1

A2 B1 C1

MRR (mm3 /s)

0.8369

0.8407

SR (µm)

5.208

5.122

By using the above grey relational method, GRC for MRR and SR are obtained along with GRG as shown in Table 1. By using GRG values, factor response tables have been constructed as shown in Table 4. Greater level value means more significance and presents the optimal parameter setting to achieve multiple objectives. Thus, GRA technique shows that for the dual objectives of MRR and SR, the optimal parameter setting of WEDM process is A2 B1 C1 .

4 Confirmation Test Conformation test has been conducted to validate the predicted results obtained from the GRA. Confirmation experiment was conducted at optimal combination of A2 B1 C1 , and values of MRR and SR are determined. Predicted and experimental values are shown in Table 5, and good agreement can be observed between these values which show the suitability of experimental investigation.

5 Conclusions Inconel 718 is used for WEDM process to investigate the effect of pulse-on time, pulse-off time and current on output responses of MRR and SR. All the input variables such as pulse-on time, pulse-off time and current are found to be significant for MRR while pulse-on time and pulse-off time are significant for SR. Through ANOVA, pulse-off time was found to be most significant factor for MRR with 69.77% contribution followed by current with 29.27% contribution and pulse-on time with 0.94% contribution. Further, for SR, current was the most significant factor with 87.11% contribution followed by pulse-on time with 11.70% contribution. The model is found to significant for MRR and SR as an extreme close relation was

Multi-objective Optimization of Inconel 718 …

235

observed between the R-squared values. The optimized process parameter settings using GRA to maximize MRR and to minimize SR were found to be pulse-on time 55 µs, pulse-off time at 5 µs and discharge current at 2 A (A2 B1 C1 ). Validation experiment has been conducted at final optimal parameter setting. Good agreement between predicted and experimental values is observed which shows the suitability of experimental investigation. Output responses of MRR as 0.8369 mm3 /s and SR as 5.208 are obtained at optimal parameter setting of (A2 B1 C1 ).

References 1. Aggarwal V, Khangura SS, Garg R (2015) Parametric modeling and optimization for wire electrical discharge machining of Inconel 718 using response surface methodology. Int J Adv Manuf Technol 79(1–4):31–47 2. Yang C-B et al (2017) Single and multiobjective optimization of Inconel 718 nickel-based superalloy in the wire electrical discharge machining. Int J Adv Manuf Technol 93(9–12):3075– 3084 3. Dabade U, Karidkar S (2016) Analysis of response variables in WEDM of Inconel 718 using Taguchi technique. Proc CIRP 41:886–891 4. Mahapatra SS, Patnaik A (2006) Parametric optimization of wire electrical discharge machining (WEDM) process using Taguchi method. J Braz Soc Mech Sci Eng 28(4):422–429 5. Chaudhari R et al (2019) Multi-response optimization of WEDM process parameters for machining of superelastic nitinol shape-memory alloy using a heat-transfer search algorithm. Materials 12(8):1277 6. Chaudhari R et al Pareto optimization of WEDM process parameters for machining a NiTi shape memory alloy using a combined approach of RSM and heat transfer search algorithm. Adv Manuf 1–17 7. Lin C, Lin J, Ko T (2002) Optimisation of the EDM process based on the orthogonal array with fuzzy logic and grey relational analysis method. Int J Adv Manuf Technol 19(4):271–277 8. Rajyalakshmi G, Ramaiah PV (2013) Multiple process parameter optimization of wire electrical discharge machining on Inconel 825 using Taguchi grey relational analysis. Int J Adv Manuf Technol 69(5–8):1249–1262 9. Chaurasia A, Wankhede V, Chaudhari R (2019) Experimental investigation of high-speed turning of INCONEL 718 using PVD-coated carbide tool under wet condition. In: Innovations in infrastructure. Springer, pp 367–374

The Effect of State Variables on Nucleation of Earthquake Using the Rate and State Friction Nitish Sinha , Arun K. Singh , and Avinash D. Vasudeo

Abstract A better understanding of the nucleation process of an earthquake is of practical importance for early warning and hazard assessment. In present chapter, chaotic nucleation of an earthquake is investigated numerically using spring-mass slider with the rate- and state-dependent friction (RSF) law. The main focus during the numerical simulations is to identify the onset of chaotic motion showing irregular changes in frictional stress as well as slip velocity. It is observed that the chaotic tendency of the sliding system increases with number of state variables in the RSF model. Moreover, the stiffness at which chaos occurs also increases with number of the state variables. Thus, the present study justifies that the RSF laws could also be useful to study the multiscale nature of friction of hard surfaces such as metals and rocks. Keywords Rate and state friction · Chaotic motion · Lyapunov exponent · Earthquake nucleation

1 Introduction Natural phenomena such as earthquakes and landslides are quite common across the world. An early detection of such a potentially disastrous phenomenon is always useful for saving life and loss of properties. It has been established that many crustal N. Sinha (B) Department of Mechanical Engineering, G. H. Raisoni Institute of Engineering & Management, Jalgaon 425002, India e-mail: [email protected] A. K. Singh Department of Mechanical Engineering, Visvesvaraya National Institute of Technology, Nagpur 440010, India e-mail: [email protected] A. D. Vasudeo Department of Civil Engineering, Visvesvaraya National Institute of Technology, Nagpur 440010, India © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_28

237

238

N. Sinha et al.

earthquakes occur because of stick-slip instability at the boundaries of tectonic plates [1–6]. Stick-slip is basically a manifestation of interaction between friction and elastic forces at the slip interface [3, 7, 8]. The rate- and state-dependent friction (RSF) model is often used for explaining the stick-slip events of hard surfaces [3, 5, 7– 9]. This advance friction model is based on the experimental observations, i.e., the friction of rock surfaces at high normal stress (>1 MPa) depends on “slip” velocity as well as “state” variable of the sliding surfaces [2, 4, 6]. The concept of state variable was proposed to elucidate hold time and slip-rate-dependent friction of the sliding surfaces [3, 10]. More than two laws have been proposed for characterizing the state variable [3, 5, 6]. However, only the “slip” law of the state variable shows the chaos [3, 5]. The one-state variable-based rate and state friction (1sRSF) is widely used and also quite successful for elucidating the system stiffness and sliding ratedependent friction [4, 6, 7]. Nonetheless, the 1sRSF does not explain the chaotic motion of the sliding surfaces even after considering the inertia of the sliding system [4]. Consequently, the 1sRSF is further modified by incorporating one more state variable known as the two-state variables-based rate and state-dependent friction (2sRSF). This modified friction model shows the chaotic motion of the sliding surfaces thus signifying the more complex nature of the sliding surfaces [4, 5, 11–13]. Becker [12] simulated numerically the 2sRSF to conclude that irregular changes in stress amplitudes and corresponding slip velocity give rise to the onset of nucleation of earthquake. Motivated by the capability of the state variables to characterize the multiscale nature of the sliding hard surfaces, we have included one more state variable in the 2sRSF law and termed as the three-state variables-based rate and state-dependent friction (3sRSF) model [14]. Moreover, the state variable is related to the history of the asperities contacts at the sliding surfaces [4]. Recently, it has been shown numerically that the effect of inertia, viscous damping, temperature, and normal stress, in general, eliminates the chaotic motion of the sliding system except the normal stress which, in fact, enhances the tendency of chaos under certain conditions [13]. This may be the reason why a very few earthquake faults undergo the nucleation of earthquake via chaotic motion [13]. Nonetheless, the nucleation process of many earthquakes has been detected to be chaotic [15–18]. Rundel et al. [16] have used the statistical physics for studying the multiscale nature of earthquake dynamics. Shelly [17] has reported the evidence of periodic oscillation to chaotic motion in earthquake along the San Andreas Fault near Parkfield California. Sobolev [18] has pointed out that many factors that complicate the earthquake process are heterogeneity, self-similarity of earth materials, and the hierarchical structures of the earthquake faults. According to Ruina [4], frictional stress τ of hard surfaces at fixed normal stress depends on current slip velocity v and number of state variable θi as τ = τ∗ +



θi + A ln

v  v dθi v = − θ . , and + B ln i i v∗ dt Li v∗

(1)

where θi is number (i = 1, 2, 3 . . .) of state variables, A, Bi , L i are related to surface properties. In addition, τ ∗ and v∗ are reference friction stress and shear velocity,

The Effect of State Variables on Nucleation of Earthquake …

239

respectively. Motivated by the chaotic nature of the 2sRSF, Sinha and Singh [14] have derived the governing differential equation of the 3sRSF in the dimensionless form as follows: ⎧ dφ   φ ⎪ ˆ1 + (ρ − ρ1 )θˆ2 + (β1 + ρβ2 + ρ1 β3 − ρ)φ + ρ1 ψ − k − ρ θ = e (1 ) ⎪ 1 ⎪ ⎪ dT ⎪ ⎪ ⎪ + keφ0 ⎨

dψ = k eφ0− eφ . dT ⎪ ⎪ dθˆ1 φ ˆ ⎪ = −e θ1 + β1 φ ⎪ dT ⎪  ⎪ ⎪ ⎩ dθˆ2 = −ρeφ θˆ + β φ 2 2 dT (2) where the nondimensional terms are defined as slip velocity φ, frictional shear stress ψ, state variables θˆ1 , θˆ2 , and θˆ3 , time T, pulling velocity φ0 , spring stiffness k as ∗ ∗ , φ = ln vv∗ , θˆ1 = θA1 , θˆ2 = θA2 , θˆ3 = θA3 , T = vL1t , β1 = BA1 , β2 = ψ = τ −τ A B2 , β3 = BA3 , ρ = LL 21 , ρ1 = LL 13 , φ0 = ln vv0∗ , k = K LA1 . A Noting that the 3sRSF could be easily converted to the 2sRSF just by eliminating the third-state variable θˆ3 . Similarly, the 3sRSF may also be reduced to the 1sRSF again by not considering θˆ2 and θˆ3 in Eq. (2).

2 Results and Discussion Although the linear stability analysis of Eq. (2) is useful for studying stability of the spring-mass slider about steady sliding [7, 8] that analysis fails to predict sliding behavior away from equilibrium or steady state [12]. Further, Eq. (2) is also solved numerically with MATLAB® using ode23s solver for understanding the nonlinear behavior as well. In the numerical simulations, the main focus is to identify the onset of chaos in the form of irregular variation of stress amplitudes and corresponding slip velocity [12, 19]. The results in Fig. 1 illustrate that periodic motion occurs at k = 0.08611 and shows the period doubling at k = 0.08421 but becomes irregular at a particular stiffness termed as chaotic stiffness kch = 0.8410. Change in stiffness signifies the spatial variation of elastic heterogeneity of materials along the earthquake faults. Consequently, there is likely that variation in local stiffness of materials may decrease kch further. Moreover, it is also evident that stress drop also increases with decrease in stiffness in Fig. 1. Erickson et al. [20] have also suggested that the ratio of sudden stress drop to increases in shear stress is accompanied by abrupt change in slip velocity along the earthquake fault. It is thus concluded that the 3sRSF gives rise to rapid nucleation of the earthquake nucleation process. Similarly, the above numerical simulations are also repeated for the 2sRSF and the results are presented in Fig. 2. It is observed that similar to the 3sRSF, the 2sRSF results in chaotic motion via period-doubling bifurcation, but at reduced stiffness kch = 0.0682. The trend of stress drop is also seen similar to the 3sRSF in Fig. 2.

240

N. Sinha et al.

Fig. 1 Stiffness dependence stress amplitudes and corresponding slip velocity using the 3sRSF for fixed values of β1 = 1.0, β2 = 0.84, β3 = 0.38, ρ = 0.048, and ρ1 = 0.034 at initial conditions [0, 0, 0, 0]

Table 1 presents a comparative change in the stress drop and slip velocity between the 3sRSF and the 2sRSF. For instance, stress change ψ ∼ 0.810 at the onset of chaos due to the 3sRSF is smaller than the 2sRSF, that is, ψ ∼ 1.703. Similarly, Table 1 also indicates that change in slip velocity from periodic oscillation to chaotic motion is φ ∼ 0.73 in the 3sRSF which is smaller than φ ∼ 1.1813 in the 2sRSF. Hence, a conclusion is that the 3sRSF predicts a smaller magnitude of stress drop and increase in slip velocity for the nucleation of earthquake compared to the 2sRSF [5, 17, 20, 21]. A possible reason for this observation is attributed to increased number of state variables that take into account the more complex nature of the sliding interface. The similar conclusion is reported that the nucleation activity of an earthquake process increases with increasing number of degree of freedom in the sliding system. Finally, it is believed that the present study could be useful to explain chaotic behavior of the complex sliding surfaces since the state variable is only the parameter that increases the chaotic behavior of the RSF model.

The Effect of State Variables on Nucleation of Earthquake …

241

Fig. 2 Effect of variation on stiffness dependence stress amplitudes and corresponding slip velocity for the 2sRSF for fixed values of β1 = 1.0, β2 = 0.84, and ρ = 0.048 at initial conditions [0, 0, 0, 0]

Table 1 Variation of stress drop and slip velocity of the 3sRSF and the 2sRSF at the onset of chaos Stress drop The 3sRSF

Slip velocity The 2sRSF

The 3sRSF

The 2sRSF

Top

Bottom

Top

Bottom −1.6670

−2.269

−2.180

2.48

−1.390

2.8360

−1.459

−0.373

1.75

−1.017

0.6547

−0.2127

ψ ∼ 0.810

ψ ∼ 1.703

φ ∼ 0.73

φ ∼ 0.273

φ ∼ 1.1813

φ ∼ 1.4543

3 Conclusions The present study establishes that increase in number of the state variables in the rate and state friction law enhances the chaotic nucleation of earthquake process. Moreover, the stiffness at which chaos begins also increases with number of state variables in the rate and state friction law. The stress drop and corresponding slip velocity also increase with number of state variables thus signifying the multiscale mechanism of friction acting along the sliding surfaces.

242

N. Sinha et al.

References 1. Brace WF, Byerlee JD (1966) Stick-slip as a mechanism of earthquake. Science 153:990–992 2. Dieterich JH (1979) Modeling of rock friction: 2. simulation of preseismic slip. J Geophys Res 84(B5):2169–2175 3. Dieterich JH (1979) Modeling of rock friction: 1. experimental results and constitutive equation. J Geophys Res 84(B5):2161–2168 4. Ruina AL (1983) Slip instability and state variable friction laws. J Geophys Res 88(B12):10359–10370 5. Gu JC, Rice JR, Ruina AL, Tse ST (1984) Slip motion and stability of a single degree of freedom elastic system with rate and state dependent friction. J Mech Phys Solids 32:167–196 6. Marone C (1998) Laboratory-derive friction laws and their application to seismic faulting. Ann Rev Earth Planet Sci 26:643–696 7. Rice JR, Ruina AL (1983) Stability of steady frictional slipping. J Appl Mech 50:343–349 8. Persson Bo NJ (2000) Sliding friction physical principal and application, 2nd ed. Springer, Berlin Heidelberg, New York 9. Gu Y, Wong T-F (1994) Nonlinear dynamics of the transition from stable sliding to cyclic stickslip in rock in nonlinear dynamics and predictability of geophysical phenomena. In: Newman WI et al (eds) Geophysical monograph, vol 8. AGU, Washington, D.C., pp 15–35 10. Dieterich JH (1978) Time-dependent friction and the mechanics of stick-slip. In: Rock friction and earthquake prediction. Birkhäuser, Basel, pp 790–806 11. Niu ZR, Chen DM (1994) Period-doubling bifurcation and chaotic phenomena in a single degree of freedom elastic system with a two-state variable law. In: Nonlinear dynamics and predictability of geophysical phenomena. AGU geophysical monograph, vol 83, pp 75–80 12. Becker TW (2000) Deterministic chaos in two state-variable friction slider and effect of elastic interaction. Geocomplex Phys Earthq 120:5–26 13. Sinha N, Singh AK, Singh TN (2018) Effect of inertia, viscosity, temperature and normal stress on chaotic behaviour of rate state friction. J Earth Syst Sci 127:45. https://doi.org/10. 1007/s12040-018-0935-2 14. Sinha N, Singh AK (2016) Linear and nonlinear stability analysis of rate state friction model with three state variables. Nonlinear Process Geophys Discuss https://doi.org/10.5194/npg2016-11 15. Ellsworth WL, Beroza GC (1995) Seismic evidence for an earthquake nucleation phase. Science 268(5212):851–855 16. Rundle JB, Turcotte DL, Shcherbakov R, Klein W, Sammis C (2003) Statistical physics approach to understanding the multiscale dynamics of earthquake fault systems. Rev Geophys 41(4) 17. Shelly DR (2010) Periodic, chaotic, and doubled earthquake recurrence intervals on the deep San Andreas fault. Science 328(5984):1385–1388 18. Sobolev GA (2011) Seismicity dynamics and earthquake predictability. Nat Hazards Earth Syst Sci 11(2):445–458 19. Ampuero JP, Ripperger J, Mai PM (2006) Properties of dynamic earthquake ruptures with heterogeneous stress drop. In: Earthquakes: radiated energy and the physics of faulting, pp 255–261 20. Erickson B, Birnir B, Lavallée D (2008) A model for aperiodicity in earthquakes. Nonlinear Process Geophys 15:1–12 21. Zielke O, Galis M, Mai PM (2017) Fault roughness and strength heterogeneity control earthquake size and stress drop. Geophys Res Lett 44(2):777–783

Finite Element Analysis of Type I and Type II Fracture with PFN Implant—A Comparative Study Sandeep Rathor , Jayamalya Jena, Rashmi Uddanwadikar, and Ashutosh Apte

Abstract The femur or thighbone is the longest and strongest bone, which bears the maximum weight of the human body. There are various types of fracture occurs in femur bone. The intertrochanteric and subtrochanteric fractures are very complex to treat and generally stabilized by proximal femoral nail. The aim of the present research work is to select the proper implant and its respective material during the fixation of intertrochanteric and subtrochanteric fractures of femur. Such type of study helps the orthopedic surgeons to predict the failure of implant. In the present study, the modeling software SolidWorks 17 is used to create the 3D model of the implant. The dimensions for 3D geometric model of implant are taken with the help of Vernier caliper. The FEM software ANSYS 16.2 is used for simulation. The finite element analysis is performed to study the distribution of stress and deformation under onelegged static load boundary conditions. The deformation and stress values on the proximal femur are compared between titanium and stainless steel implant material in case of both subtrochanteric and intertrochanteric fractures of the femur. The bone healing process mainly depends on the stability given to the fracture. The stability is achieved by checking the bone contact surface area at fracture interface surface. The frictional stress and contact pressure at fracture plane are compared by taking both titanium and stainless steel implant material. These contact results are used to estimate the condition of healing process at the fracture interface of intertrochanteric and subtrochanteric fractures. The effect of frictional stress, pressure developed and localized stress on bone healing is also discussed. Keywords Proximal femoral nail · Intertrochanteric · Subtrochanteric · FEA

S. Rathor (B) · R. Uddanwadikar Visvesvaraya National Institute of Technology, Nagpur, India e-mail: [email protected] J. Jena Indian Institute of Technology Roorkee, Roorkee, India A. Apte Sawali Hospital, Nagpur, India © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_29

243

244

S. Rathor et al.

1 Introduction The femur bone is the longest bone in the human body. It consists of an upper end, a shaft and a lower end. The upper end of the femur consists of the head, neck, lesser trochanter, greater trochanter, the intertrochanteric crest and intertrochanteric line as shown in Fig. 1. The greater trochanter is a huge quadrangular prominence positioned at the top part of the junction between the femur’s neck and shaft. The lesser trochanter is present as a conical eminence focused medially and toward the back from the junction between the posterior side of neck and shaft. Intertrochanteric line marks the neck’s junction with the femur. Intertrochanteric crest is the junction of the neck’s posterior side of with shaft of the femur bone.

1.1 Types of Hip Fracture The “hip” is a type of ball-and-socket joint. It allows rotation and bending of the upper leg at the pelvis. The majority of hip fractures occur to the people who are with age 65 or older. Most people are having hip fracture near the upper side of the femur. There are two types of hip fractures. They are explained below. Type I (Intertrochanteric Fracture). An intertrochanteric hip fracture is shown in Fig. 2 which is a specific type of hip fracture. Intertrochanteric means “between the two trochanters” which are bony projections on the thighbone. Generally, intertrochanteric fracture arises between lesser and greater trochanters. Intertrochanteric fractures are the result of such a high energy injury in younger individuals, fall from a height or a motor vehicle accident [1]. Type II (Subtrochanteric Fracture). Subtrochanteric fractures of the femur take place in the proximal region as shown in Fig. 3, whose anatomical definition is controversial and difficult. Subtrochanteric hip fracture is a break in the transverse Fig. 1 Anatomy of the hip bone

Finite Element Analysis of Type I and Type II Fracture …

245

Fig. 2 Intertrochanteric fracture

Fig. 3 Subtrochanteric fracture

plane between the lesser trochanter and the area around 5 cm below from the lesser trochanter [2].

2 Methodology 2.1 Collection of Specimen and Preparation of CAD Model of the Implant The specimen is collected from the diagnosis laboratory of an orthopedic surgeon for better measurement of screw and nail part of the implant. The 3D model of proximal femoral nail (PFN) with a three-holed, screw plate was created with CAD software SolidWorks. The 3D model of the superior screw, inferior screw, distal screw and

246

S. Rathor et al.

Fig. 4 CAD model of implant

PFN nail are also prepared to utilize the geometry and measurements from the best possible estimation by Vernier caliper. The final assembly of the implant is shown in Fig. 4.

2.2 CAD Model of the Femur The three-dimensional solid model of the femur is downloaded from a repository called “GrabCAD,” an online community of professional engineers, and the model is edited by using SolidWorks 17 software as shown in Fig. 5.

Fig. 5 CAD model of the femur

Finite Element Analysis of Type I and Type II Fracture …

247

2.3 Assembly Model of Bone with the Implant During the assembly in the SolidWorks, the femur, PFN implant, distal screw and lag screw are assembled, the positioning of the implant with respect to the bone is done, and the final assembly of bone-implant model is prepared as shown in Fig. 6. Intertrochanteric fracture is created at an angle 45° at the intertrochanteric region giving a 1 mm gap to create fracture as shown in Fig. 7. Subtrochanteric fracture is also created with respect to the transverse plane of the shaft of the femur, and this fracture is created below the lesser trochanter as shown in Fig. 8.

Fig. 6 Bone-implant assembly model

Fig. 7 Intertrochanteric

248

S. Rathor et al.

Fig. 8 Subtrochanteric

2.4 Finite Element Analysis Different linear elastic isotropic material properties were assigned to different regions of the femur like shaft, head, neck, trochanteric region and the implant (stainless steel/titanium). Corresponding elastic constants are given in Table 1. Stainless steel (SS) and titanium (TI) material properties were assigned to the implants. The frictional coefficient of 0.29–0.33 is assigned to all the respective parts of the model [2]. To generate the mesh of the PFN and the proximal femur, ten-nodded tetrahedral elements were used. The implant and bone’s global element edge lengths were 3 and 4 mm, respectively. The no of nodes and elements generated for subtrochanteric fracture is 87,265 and 52,713, respectively Also, for intertrochanteric fracture, the no. of nodes and elements generated is 87,992 and 52,769, respectively. The model is fixed at the distal end of the femur. In this study, a one-legged stance load configuration is used. The load condition consists of one joint reaction force of 730 N, and three muscle retraction force of 300, 188 and 292 N are acted [3]. All the above forces applied to femur bone on ANSYS software are shown in Fig. 9. Table 1 Material properties Properties

Shaft

Head

Neck

Trochanter

SS

TI

Elastic modulus (Mpa)

17,000

900

620

260

200,000

110,000

Poisson’s ratio

0.30

0.29

0.29

0.29

0.29

0.29

Finite Element Analysis of Type I and Type II Fracture …

249

Fig. 9 Loading

3 Results and Discussion The von Mises stress, deformation, frictional stress and the pressure developed at the fracture interface are the outcomes of the analysis. Figure 10a–e describes the results obtained by stress analysis. The results include von Mises stress for subtrochanteric

(a) Superior screw

(d) Femur

(b) Inferior screw

(c) Distal screw

(e) PFN nail

Fig. 10 von Mises stress developed in subtrochanteric fracture with SS implant

250

S. Rathor et al.

fractured femur implantation with stainless steel. Also, deformation, frictional stress and the pressure developed at the fracture interface are shown in Fig. 11a, b and c, respectively. The magnitude of von Mises stresses in different parts of the bone-implant assembly for intertrochanteric and subtrochanteric fractures for both the implant materials is given in Table 2. The comparison with published data [4] is also given in Table 2. The contact status at the fracture interfaces for both implant are given in Table 3.

(a) Deformation

(b) Pressure developed

(c) Frictional Stress

Fig. 11 Deformation, pressure and frictional stress in subtrochanteric fracture with SS implant

Table 2 Maximum von Mises stress induced in the parts of assembly for all the four cases Parts of assembly

Implant material

Intertrochanteric fracture (MPa)

Subtrochanteric fracture (MPa)

Published results (MPa)

Femur bone

SS implant

61.869

63.11

–

TI implant

74.36

63.42

–

Inferior screw

SS implant

109.45

126.24

–

TI implant

97.37

117.10

–

Superior screw

SS implant

114.09

135.11

–

TI implant

110.37

112.16

–

SS implant

127.04

165.11 (maximum in SS)

152.68 (maximum in SS)

TI implant

98.90

123.52 (maximum in TI)

116.11 (maximum in TI)

SS implant

25.20 (minimum in SS)

32.28

27.90 (minimum in SS)

TI implant

17.67 (minimum in TI)

19.31

19.04 (minimum in TI)

Nail

Distal screw

Bold signifies the Maximum and Minimum Values of the stress induced in SS and TI implant materials. Those values are compared with the published data

Finite Element Analysis of Type I and Type II Fracture …

251

Table 3 Pressure and frictional stress for all the four cases Parameter

Implant material

Intertrochanteric fracture (MPa)

Subtrochanteric fracture (MPa)

Pressure

SS implant

5.2157

232.69

TI implant

6.2893

195.90

SS implant

6.8749

208.73

TI implant

7.9341

193.55

Frictional stress

4 Conclusion It is concluded in the past studies that the PFN implant is the minimum invasive implant for Proximal femoral fracture [5]. PFN is proved to be very suitable implant for unstable proximal femoral fracture [5, 6]. It is found that the maximum von Mises stress induced in both implant is within the range of yield stress of SS and TI, respectively. The results obtained are compared and validated with the published data. Stresses developed are greater for SS implant as compared to TI implant, but higher stresses are induced in bone in case of TI implant as compared to SS implant. For subtrochanteric fracture, maximum stress is induced in nail part in both the implant. For intertrochanteric fracture with TI, maximum stress is induced in the superior screw and in nail part in the case of SS implant. It is noticed that the intertrochanteric fracture of the femur with PFN implant has high localized contact stress which does not encourage the early healing of the fracture. Hence, PFN is more suitable for subtrochanteric fracture as compared to intertrochanteric fracture. Ethical Approval This article does not contain any studies with human participants performed by any of the authors.

References 1. Liang C et al (2018) Intertrochanteric fracture: association between the coronal position of the lag screw and stress distribution. Asian J Surg 41(3):241–249 2. Sowmianarayanan S et al (2008) Finite element analysis of a subtrochanteric fractured femur with the dynamic hip screw, dynamic condylar screw, and proximal femur nail implants—a comparative study. Proc Inst Mech Eng [H] 222(1):117–127 3. Fraldi M et al (2010) Topological optimization in hip prosthesis design. Biomech Model Mechanobiol 9(4):389–402 4. Taheri NS et al (2011) Comparative study of two materials for dynamic hip screw during fall and gait loading: titanium alloy and stainless steel. Journal of Orthopaedic Science 16(6):805–813 5. Boldin C et al (2003) The proximal femoral nail (PFN)—a minimal invasive treatment of unstable proximal femoral fractures: a prospective study of 55 patients with a follow-up of 15 months. Acta Orthop Scand 74(1):53–58 6. Chopra B et al (2017) Proximal femoral nail-outcome and complications: a prospective study of 125 cases of proximal femoral fractures. Int J Res Orthop 3(5):973

Postural Evaluation of Construction Labourers Engaged in Excavation Work Using Newly Developed NERPA Method and Its Validation Through REBA and WERA Methods Manoj T. Gajbhiye , Debamalya Banerjee , and Saurav Nandi Abstract Foundation is a preliminary requirement of construction work which is done by excavating soil. The excavation is the process of removing soil from the surface of the earth to form a cavity in the ground. In India, this rigorous and highly diligent task is being done manually by men and women both. These labourers suffer from work-related musculoskeletal disorders (WRMSD) due to extreme physical exertion, excessive force, lifting heavy load and working in awkward posture throughout the day. This study has been carried out to evaluate the levels of pain/discomfort among the excavation labourers. The exposure assessment has been carried out by NERPA, WERA and REBA methods. The labourers’ complaints about pain/discomfort in body parts and risk level score obtained from three methods are found similar. The compatibility of NERPA methods’ worksheet compares with WERA and REBA. Result shows that NERPA worksheet can be implemented for assessment of construction workers postures. Keywords WRMSD · Construction · Excavation · NERPA · WERA · REBA

1 Introduction The first phase of any construction work is its foundation which is base of any house. This foundation is laid down from the ground. The excavation is the digging of soil for making pit inside the ground to lay down the foundation and erect the column. In India, maximum construction work is being carried out manually, and the excavation is an important aspect in initial stage. With development of modern and new technology, construction work still demands manual intervention or physically demanding work. In India, construction industry alone contributes 24.20% fatality M. T. Gajbhiye (B) · D. Banerjee Department of Production Engineering, Jadavpur University, Kolkata, India e-mail: [email protected] S. Nandi Department of Mechanical Engineering, Calcutta Institute of Engineering and Management, Kolkata, India © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_30

253

254

M. T. Gajbhiye et al.

per year due to different reasons of risk factors in construction industry [1]. Many researchers have started working in the construction area and highlighted the issues related to physical exposure among the construction workers and labourers [2–5]. The past researches also indicate the development of assessment tools such as QEC [6], REBA [7], OWAS [8], PLIBEL [9] and many more for measuring the physical risk/discomfort. Many construction workers and labourers are continuously working in unfavourable conditions without any intuition about the physical risks and suffering from WRMSD. Therefore, this study has been carried out with the aims to (1) find the level of pain/discomfort in different body parts of excavation labourers as per complaints of labourers about pain/discomfort in different body parts according to age, (2) verify this pain/discomfort by applying NERPA, WERA and REBA methods and (3) compare the compatibility of NERPA methods’ worksheet in evaluation of risk in construction work.

2 Literature Review 2.1 Novel Ergonomic Postural Assessment Method (NERPA) This method, NERPA, has been developed by Sanchez-Lite et al., in 2013, using a digital human model (DMH) jointly with 3D CAD tool together with motion capture in real time that will be used in 3D virtual environment by incorporating work process, equipment, machineries, tools and human factors. The method has been developed particularly for aeronautical and automobile industries which is the modified version of RULA. The method focused on arms, neck, trunk and wrists and little focus on legs and helps to make adequate decisions in the design and assessment [10].

2.2 Rapid Entire Body Assessment (REBA) Rapid entire body assessment (REBA) is a quick observational and survey method used to assess static and dynamic movements, rapid changing or unstable body posture which leads to the development of work-related musculoskeletal disorder [7].

2.3 Workplace Ergonomic Risk Assessment (WERA) Rahman et al., in 2011, developed this pen-and-paper method to find the WRMSD in construction for evaluation of risk and can be used in any space of workplaces without disturbing the workers. This is quick observational method which was developed to

Postural Evaluation of Construction Labourers Engaged …

255

Table 1 Demographic characteristics of labourers engaged in excavation work Characters

Male (N = 46) Mean ± SD (range)

Age (years)

39.67 ± 10.97 (20–59)

Weight (kg)

59.72 ± 8.39 (45.2–79)

Height (cm)

164.46 ± 5.29 (152–175)

Experience (years)

16.06 ± 10.42 (1–38)

BMI (kg/m2 )

22.01 ± 2.49 (11.20–27.34)

Female (N = 15) Mean ± SD (range) 33.47 ± 7.85 (24–51) 48.09 ± 4.91 (41.5–59.3) 156.15 ± 6.42 (146–166) 12.2 ± 8.58 (1–33) 19.69 ± 1.14 (18.53–22.79)

give screening method to working task for exposure of physical risk factor related with WRMSD. This method comprises of six types of physical risk factors like posture, repetition, force, vibration, contact stress and task duration including five body parts which are neck, shoulders, wrists, lower back and legs. This method needs no special equipment [11].

3 Methodology The study had been carried out for ten days for daily working hours of 8–10 with breaks. 46 male and 15 female labourers engaged in excavation work have been observed, interviewed and recorded from 12 different construction sites whose background characteristics are given in Table 1. The height and weight had been measured using anthropometric scale and standard weighing machine. The personal data like age, daily working hours, procedures and pain/discomfort with severity (like ‘no pain’, ‘occasional pain’, ‘always pain’) and other related problems they faced in daily working hours on the construction sites have been discussed and noted. The labourers were between 20 and 59 years of age, with working experience of 1–38 years. The mean age, weight, height, experience and BMI with standard deviation and range are given in Table 1. The labours’ are categorised in two groups (male and female) while an excavation task is divided into four tasks. Table 2 shows task details, real image and body position while doing the task. The percentage of labourers complaints for pain/discomfort in different body parts are mentioned in Table 3.

4 Results and Discussion In this study, an ergonomic risk score of postures has been evaluated by NERPA, WERA and REBA methods using the postures of labourers while working (Table 2).

256

M. T. Gajbhiye et al.

Table 2 Task carried out by labourers and body parts position Task No. and task details

Real images of task Male

Body position Female

T1 Loosening of soil with the help of pickaxe

1. Shoulder extension 2. Trunk extension/straight, sometimes twist 3. Both arms above shoulder 4. Wrist radial deviation when both arms above shoulder 5. Wrist ulnar deviation when both arms below shoulder 6. Legs flexion (bent): 30°–60°

T2 Collecting excavated soil in pan

1. Trunk flexion while filling pan 2. Both arms below shoulder while filling 3. Wrist ulnar deviation while filling pan (bent), lifting from ground to shoulder and from shoulder to head above 4. Legs flexion (bent): 30°–60°

T3 Lifting of pan/vessel

1. Trunk flexion and sometimes twisted for lifting pan 2. Both arms below shoulder 3. Trunk extension while lifting pan 4. Neck extension while lifting pan 5. Both arms above shoulder 6. Wrist extension (bent) 7. Both legs flexion (bent): >60° 8. Movement to throw soil or pass (continued)

Postural Evaluation of Construction Labourers Engaged …

257

Table 2 (continued) Task No. and task details

Real images of task Male

Body position Female

T4 Throw or dump excavated soil outside pit: Labourer 1:

1. Shoulder extension 2. Both arms above shoulder for passing pan to labourer 2 3. Neck extension 4. Wrist extension >15° (bent)

Labourer 2:

1. Trunk forward flexion to pick pan 2. Trunk lateral bending or twisted: >20° 3. Arms below shoulder while picking pan 4. Arms above shoulder after lifting and throwing 5. One legs flexion (30°–60°) 6. Other leg flexion (>60°)

Table 3 Percentage of labourers’ complaints about ‘always pain’ S. No.

Body parts

Male (n = 46) Always pain

Female (n = 15) No pain

Always pain

No pain

1

Head

0

100

26.67

73.33

2

Neck

10.86

89.14

60

40 6.67

3

Shoulders

93.48

6.52

93.33

4

Chest

28.26

71.74

40

60

5

Arms

97.82

2.18

100

0 13.33

6

Wrist

69.57

30.43

86.67

7

Lower back

100

0

100

0

8

Legs

36.95

63.04

86.67

13.33

9

Knees

26.08

73.92

26.27

73.33

10

Feet, ankle and toe

0

100

13.33

86.67

258

M. T. Gajbhiye et al.

Fig. 1 Percentage of labourers’ complaints about ‘always pain’

These scores are being compared with the labourers’ complaints. To validate compatibility of NERPA methods’ worksheet, the NEPRA methods’ result has been compared with WERA and REBA methods. From the study, it was found that the male labourers have complained ‘always pain’ in wrists (69.57%), shoulders (93.48%), arms (97.82%) and lower back (100%), whereas female labourers has complained ‘always pain’ in neck (60%), wrist and legs (86.67%), shoulder (93.33%), arms and lower back (100%) (Fig. 1). The ergonomic risk score of postures for the four different tasks had been evaluated using NERPA, WERA and REBA methods as given in Table 4. From the scores of all the three methods, it was found that all the tasks were under high risk and labourers were suffering from work-related musculoskeletal disorders. From the score, it can be concluded that shoulders, arms, neck, wrists and lower back were under more risk and need immediate investigations and corrections. The scores obtained from NERPA worksheet had been compared with the WERA and REBA scores. It was found that the risk level scores obtained from WERA and REBA were similar to that of obtained by NERPA method. Also, the body parts, which were under high risk, as complained by labourers and evaluated by NERPA worksheet, WERA and REBA method were found to be same. Hence, NERPA worksheet can be used to evaluate the risk of construction workers using actual 3D visualisation of workstation, working postures and ergonomic design in virtual scenario.

5 Conclusion The complaint of the labourers and scores of the different methods show that the labourers engaged in excavation work are in high physical risk zone and suffering

4

4

1

7

1

3

11

7

HR

N

LB

L

PS-B

M

F

TS-C

FS

RL

1

HR

7

9

1

1

7

1

5

4

6

1

1

4

HR

> 80%. CFD results are then mapped by standard curve fitting techniques to develop a mathematical model. A good and significant correlation between blood pressure and % area stenosis is found. The developed mathematical model is further used to develop an inexpensive and handy diagnostive tool for preliminary diagnosis of severity of coronary artery disease by clinicians. Keywords Stenosis · Coronary artery disease · Blood pressure

1 Introduction Coronary artery disease (CAD) is one of the leading causes of high blood pressure and heart attack in human. Heart attack or in clinical terms myocardial infarction occurs when blood flow to a part of heart decreases or stops. Physiological importance of stenosis severity cannot be ascertained by just depending on the modern imaging techniques such as angiography, magnetic resonance angiography (MRA) and intravascular ultrasound (IVUS) especially for intermediate area stenosis (AS—it defines blockage in artery based on reduction in lumen cross-sectional area). These modern techniques are used by radiologists to find the exact location and severity of CAD. However, these diagnostive methods are expensive and are also inaccurate for lumen P. Jhunjhunwala (B) · P. M. Padole · S. B. Thombre Visvesvaraya National Institute of Technology, South Ambazari Rd., Nagpur, Maharashtra 440010, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_39

337

338

P. Jhunjhunwala et al.

diameter stenosis (DS—it defines blockage in artery based on reduction in lumen diameter) of 40–70% [1, 2]. Therefore, from a clinician’s perspective, assessment of hemodynamic factors that potentially determine the stenosis severity is important. Hence, the main aim of this study is to assess the rise in blood pressure due to coronial stenosis, calculated using the computational simulation of arterial stenosed models. Further, the relationship between rise in blood pressure and % AS is correlated. The aim of the present study is to develop an inexpensive preliminary diagnostive tool or the prediction of severity of stenosis.

2 Methods 2.1 Coronary Artery Models In order to develop a diagnostive tool which correlates the percentage area stenosis (AS) in coronary artery with increase in blood pressure, sixteen 3-D models of coronary artery are generated. In many previous literatures [3, 4], it is mentioned that the geometry of a perfect coronary artery is nearly symmetrical and cylindrical. In the current study, the three-dimensional arterial models are generated with this assumption. Based on clinical data, the healthy coronary artery is modeled as of 16 mm in length and 4 mm in diameter. In clinical language, blockage in artery is specified in terms of % DS. However, the hemodynamic parameters are the functions of % AS. For the same % DS, the % AS varies depending on the shape and orientation of plaque deposition (refer Table 1). The hemodynamic parameters behave differently for same % DS as % AS can be different. Henceforth in this work, % area stenosis is considered to determine severity of stenosis. Three cases of stenosis model: (1) asymmetric plaque deposition (ASD), (2) one-sided plaque deposition (OSD) and (3) symmetric plaque deposition (SD) as shown in Fig. 1 are considered. For each stenosed model, five cases of diameter stenosis (DS), 25, 50, 66.67, 75 and 83.3% are investigated. In all the three cases of stenosed models, the shape of plaque is assumed to be circular. Although the three stenosed models are generated with same extent of DS, these differ with each other in Table 1 % AS of stenosed artery models corresponding to their % DS

% DS

Stenosed artery model ASD

OSD

SD

25

14.3

19.51

43.75

50

39

49.97

75

66.67

58.4

71.8

89.44

75

68.4

80.44

93.75

83

78.78

88.72

97.11

Development of a Diagnostive Tool for Prediction …

339

Fig. 1 3-D stenosed artery models of 66.67% DS for (1) asymmetric plaque deposition (ASD), (2) one-sided plaque deposition (OSD) and (3) symmetric plaque deposition (SD)

terms of AS. Therefore, these arterial stenosed models result in 15 different models which vary from 14.3 to 97.22% in terms of area stenosis (AS) as shown in Table 1. All the 16 arterial models (healthy artery and 15 stenosed arteries) are designed in SolidWorks 2013. Then, these models are imported to pre-processing program of ANSYSv14.5 for generation of mesh. Grid independency test is performed to avoid any error in computation and to minimize the cost of computation.

2.2 Boundary Conditions and Fluid Properties No-slip boundary condition is executed for velocities of the solid arterial walls. The solid walls of the coronary artery models are assumed to be rigid to simplify the analysis as it does not have significant effect on the output results [5, 6]. At the outlet, a constant 100 mm of Hg gauge pressure which is mean of normal systolic (120 mm Hg) and diastolic (80 mm Hg) blood pressure is imposed. Physiologically, flow of blood inside vessels (artery and veins) is pulsatile in nature. Therefore, at the inlet, a uniform time-varying velocity profile is forced to incorporate pulsatile behavior of blood flow in the simulation. The sinusoidal wave pattern of flow [6, 7] is assumed as a substitute to physiological pulsation during systole. During diastolic phase, there is cut-off of supply from the heart. Therefore, during this phase, the blood flow is assumed to have a fixed velocity of 0.1 m/s. The imposed velocity profile delivers an inflow velocity of 0.5 m/s during peak systole at t = 0.1 s with a fast heartbeat of 120 per minute. The complete cycle is of 0.5 s in which systolic phase is up to 0.218 s, and rest represents diastolic phase. Simulation of two cycles is necessary to achieve a cycle independent solution in the present analysis. Hence, to develop the diagnostic tool, results of the second periodic cycle are only used. To analyze the flow conditions in the coronary artery, blood is considered as a homogeneous and incompressible fluid with specific mass of 1060 kg/m3 as in [8]. The laminar flow has been assumed in the present work since the average Reynolds number ranges from 150 to 800 [9, 10]. Blood is a shear thinning fluid, and hence, in

340

P. Jhunjhunwala et al.

this work, blood is modeled as a non-Newtonian fluid which follows Carreau model [11],  n−1  μ − μ∞ = 1 + (λγ˙ )2 2 μ0 − μ∞

(1)

where μ, μ∞ and μ0 are the effective viscosity, dynamic viscosity at infinite shear rate and dynamic viscosity at zero shear rate, respectively. Characteristic viscoelastic time of the fluid, shear rate and power law index are represented by λ, γ˙ and n respectively. For blood, n = 0.3568, λ = 3.313 s, μ0 = 0.056 kg/ms and μ∞ = 0.0035 kg/ms.

3 Results and Discussion The simulation results are computed with time step size of 0.01 s for pressure distributions in all the 16 artery models for the complete cardiac cycle. In spite of having same % DS, large variations in rise in maximum pressure computed with CFD is observed as the % AS of these models are different. In general, large drops in pressure are observed after stenosis, and pressure has maximum value at the inlet. The simulation results obtained through CFD are then further used to compute rise in systolic and diastolic BP due to increase in stenosis severity. To compute the rise in systolic BP, the maximum pressure obtained at peak systole (t = 0.6 s) for healthy and stenosed models is considered. Then, from the values of pressure obtained for stenosed model, the corresponding % pressure rise to healthy artery case is obtained. Now, if 120 mm Hg (normal systolic BP) is considered as standard baseline for healthy coronary artery, the rise in blood pressure because of increase in stenosis severity can be obtained simply by adding % increase in pressure to baseline pressure of 120 mm Hg. In the similar fashion, to calculate the rise in diastolic BP, maximum pressure obtained during end of cycle (t = 1 s) is used. For diastolic blood pressure, the standard baseline is 80 mm Hg. The obtained and calculated values from CFD analysis for increasing severity of stenosis and corresponding % pressure rise for only healthy and SD stenosed artery models are listed in Table 2 as an illustration. The computed CFD results for % rise in blood pressure are then plotted using standard curve fitting techniques of MATLAB to develop a mathematical equation which relates % rise in blood pressure with % AS. Two different five-degree polynomial equations, y = ax 5 + bx 4 + cx 3 + dx 2 + ex + f, appreciably map the data for % pressure rise at peak systole and diastole with R-square values of 0.93 and 0.9, respectively. As R-square value equals to 1 represents no error, therefore, it can be said that the selected mathematical model can be utilized with confidence for the formulation of diagnostic tool. The constants of selected five-degree polynomial equations for the calculation of rise in pressure (%) at the systolic and diastolic phases are given in Table 3.

Development of a Diagnostive Tool for Prediction …

341

Table 2 Maximum pressure and corresponding % pressure rise values obtained for SD stenosed artery models Area stenosis (%)

Systolic

Diastolic

Maximum pressure (mm Hg)

Pressure rise (%)

Maximum pressure (mm Hg)

Pressure rise (%)

0

102

0

100

0

43.75

103

0.95

100

0.08

75

116

12

101

0.81

89.44

187

45

105

5

93.75

345

70

114

12

97.11

1280

92

164

39

Table 3 Constants of fifth degree polynomial for the best—fit curve Phase

a

b

Systolic

1.489 × 10−6

−0.0003399

Diastolic

0.01717

−1.005

c

d 0.02732

14.82

−0.9274 −71.88

e

f

13

17.68

120.1

22.03

Using these equations and taking into account the errors, a code is developed in the MS-Excel for the formulation of diagnostic tool. The diagnostic tool is developed in the form of a “blockage calculator” as shown in Fig. 2. To use this tool, a user has to just open the excel file in which the tool has been developed and click the button “Click to Calculate.” The developed blockage calculator will pop-out after then. As shown in Fig. 2, to get the prediction of probable % AS as output through this tool, it requires four inputs, namely normal systolic BP, normal diastolic BP, current systolic BP and current diastolic BP. In the developed tool, the normal systolic and diastolic BP represents the blood pressure of a person when he/she is healthy (e.g., 120/80 mm Hg), while current systolic and diastolic BP represents the elevated blood pressure or the blood pressure at which a person is showing discomfort (e.g., 160/85 mm Hg). It can be observed here that this tool can predict the probable % AS just by measuring the blood pressure of a person, and hence, it is patient specific, inexpensive, non-invasive, easy to use, handle and less time-consuming. If normal blood pressure of a person is unavailable in case history, then 120/80 mm Hg can be inserted as a default value.

342

P. Jhunjhunwala et al.

Fig. 2 Developed diagnostic tool in the form of “Blockage Calculator”

4 Conclusion This study proposes using a diagnostive tool “blockage calculator” to estimate the extent of stenosis severity. This quantified tool is based on the CFD study of blood flow through three different stenosed models of same percentage of lumen diameter stenosis with varying cross-sectional lumen area stenosis. This tool is easy to use and can help clinicians in preliminary diagnosis for patients of high blood pressure.

References 1. Pijls NH, Fearon WF, Tonino PA, Siebert U, Ikeno F, Bornschein B, van’t Veer M, Klauss V, Manoharan G, Engstrøm T (2010) Fractional flow reserve versus angiography for guiding percutaneous coronary intervention in patients with multivessel coronary artery disease: 2-year follow-up of the FAME (fractional flow reserve versus angiography for multivessel evaluation) study. J Am Coll Cardiol 56(3):177–184 2. Tonino PA, De Bruyne B, Pijls NH, Siebert U, Ikeno F, vant Veer M, Klauss V, Manoharan G, Engstrøm T, Oldroyd KG (2009) Fractional flow reserve versus angiography for guiding percutaneous coronary intervention. N Engl J Med 360:213–224 3. Jhunjhunwala P, Padole P, Thombre S (2015) CFD analysis of pulsatile flow and non-Newtonian behavior of blood in arteries. MCB Mol Cell Biomech 12(1):37–47 4. Pericevic I, Lally C, Toner D, Kelly DJ (2009) The influence of plaque composition on underlying arterial wall stress during stent expansion: the case for lesion-specific stents. Med Eng Phys 31(4):428–433. https://doi.org/10.1016/j.medengphy.2008.11.005

Development of a Diagnostive Tool for Prediction …

343

5. Bernsdorf J, Wang D (2009) Non-Newtonian blood flow simulation in cerebral aneurysms. Comput Math Appl 58(5):1024–1029. https://doi.org/10.1016/j.camwa.2009.02.019 6. He Y, Duraiswamy N, Frank AO, Moore JE (2005) Blood flow in stented arteries: a parametric comparison of strut design patterns in three dimensions. J Biomech Eng 127(4):637–647 7. Duraiswamy N, Schoephoerster RT, Moore JE (2009) Comparison of near-wall hemodynamic parameters in stented artery models. J Biomech Eng 131(6):061006 8. Karimi S, Dabagh M, Vasava P, Dadvar M, Dabir B, Jalali P (2014) Effect of rheological models on the hemodynamics within human aorta: CFD study on CT image-based geometry. J Nonnewton Fluid Mech 207:42–52. https://doi.org/10.1016/j.jnnfm.2014.03.007 9. Banks J, Bressloff N (2007) Turbulence modeling in three-dimensional stenosed arterial bifurcations. J Biomech Eng 129(1):40–50 10. Valencia A, Solis F (2006) Blood flow dynamics and arterial wall interaction in a saccular aneurysm model of the basilar artery. Comput Struct 84(21):1326–1337. https://doi.org/10. 1016/j.compstruc.2006.03.008 11. Cho YI, Kensey KR (1991) Effects of the non-Newtonian viscosity of blood on flows in a diseased arterial vessel. Part 1: steady flows. Biorheology 28(3–4):241–262

Design, Modelling and Optimization of Artificial Limb for Lower-Extremity Amputees Based on CATIA Smit V. Motghare

Abstract Mechanical engineering applications complement biomedical engineers to attain best performance in prosthesis. Appropriate design of artificial limbs for lower-extremity amputees provides a prosthesis that improves the quality of life. Computer-aided design (CAD) and computer-aided manufacture (CAM) have been used for the development and improvement in performance of artificial limb. Depending on the residual part of limb, the prosthesis may be custom designed. Modified design built of strong but light-weight material and flexible biocompatible interface socket that minimizes the risk of dermatological breakdown of residual limb may be developed. This paper reports designing three-dimensional model in modelling software Computer-aided three-dimensional interaction and application software (CATIA V5) and converting it to a finite element model. The model was imported to Solver ANSYS R16.0 for meshing followed by stress and deformation analysis for stainless steel and carbon fibre. From the obtained results of stress, deformation values and weight, carbon fibre was selected as the material for building principle components of the limb. The FE model may be re-modified and reproduced to achieve higher efficiency. Keywords Prosthetic limb · CATIA · Finite element analysis · Carbon fibre

1 Introduction The earliest successful limb amputation ages back to 484 B.C., when Hegesistratus had to cut his feet to escape from prison. He created a wooden foot for himself to compensate the loss. In 1696, first notable lower-extremity transtibial prosthesis with an unlocked knee, a thigh corset and external hinges was introduced by Verduyn, a Dutch surgeon. Potts in 1816 introduced first transfemoral prosthesis having articulating knee and ankle, the Anglesey leg. Transfemoral artificial limb with polycentric knee, suction socket and multi-articulated foot was introduced by Parmelee in 1863. S. V. Motghare (B) Department of Mechanical Engineering, Government College of Engineering, Nagpur, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_40

345

346

S. V. Motghare

Some historic developments include the solid-ankle cushioned-heel (SACH) foot (1956) and Canadian hip disarticulation prosthesis (1954). Stewart-Vickers in 1960 introduced hydraulic knee known as the hydra cadence leg [1]. Amputation means removal of a limb or part, or organ by surgery. Prosthetic design involves preparing a replacement for the body part lost during amputation with appropriate shape and size [2]. The time between amputation and fitting an artificial limb is at least 6 weeks because definitive fitting requires a stable limb volume maintained for 2–3 weeks [3]. The prosthesis must be functional, comfortable and cosmetic (the prosthetic device should have appearance similar to the anatomical limb) [4]. Prosthetic sockets act as interface between the residual limb and prosthesis [5]. The socket encloses the residual limb tissues, and better enclosure reduces the need of supplemental suspension and provides better control on the prosthesis [6]. To create more functional and comfortable lower-limb prostheses, human– machine interface must be designed properly. This is achieved by the measurement of socket pressure, friction-related phenomena, computational modelling, limb tissue responses to external loads at the interface, etc. However, stresses experienced at the residual body parts have not yet led to sufficient clinical consensus that could essentially modify clinical practice [7, 8]. The tibio-femoral joint of the prosthesis should be free and may require slightest pressure to bend the knee. Cosmetic cover may be fabricated using flexible urethane foam having density 3–4 lb/cu feet to generate its own thin, high density surface, abrasion and puncture resistant and easy to clean surface. Pigments may be applied to the foam system during foaming to produce a cover that has colour similar to the colour of patient’s anatomical limb. An inversion casting process may be implemented to produce a replica of the remaining limb without laborious sculpting to reproduce the anatomical characteristics of a unilateral amputee’s leg. The merger of aerospace materials technology and prosthetics has led to advances in cosmetic covers which when combined with an endoskeletal weight-bearing system provides a prosthesis that satisfies all requirements set forth by CPRD [9]. There are many robotic prostheses but designing and testing new custom prostheses take considerable time, which may be reduced by computational simulations. Computational models of prosthesis are mainly based on finite element methods [10–12]. FE methods have two major advantages, first, full field information on the stress, strain and motion anywhere within the modelled objects can be predicted. Second, it is relatively convenient to do parametric analysis for an optimal design [13, 14]. The purpose of this paper is to design and analyse the performance of the proposed model through computational approach. A number of papers report use of computerized tomography and magnetic resonance imaging scans to develop a realistic model of prosthesis [15]. In the present work, a principle component analysis model (with articulated knee) was designed for transfemoral prosthesis having flexion–extension movement. The model comprises of a socket that may be optimized to fit the residual limb, femoral part with a ball structure and tibial structure with foot at the lower end. Efforts have been made to create an adjustable tibial length and femoral length to suit the patient’s constitution. Computational modelling along with meshing and

Design, Modelling and Optimization of Artificial …

347

measurement of developed shear stress and total deformation for stainless steel and carbon fibre material were performed.

2 Materials and Method 2.1 Computational Modelling Computer-aided design and computer-aided manufacturing (CAD/CAM) technology provide computational modelling of prosthesis to establish hassle-free flexion– extension movement of lower limb. A three-dimensional model of artificial limb for lower-extremity amputees was built in CATIA V5 software as shown in Fig. 1. The socket and femoral parts are shown in Fig. 1a and the tibial structure along with foot in Fig. 1b. The model consists of two parts, (i) adjustable tibial structure with (ankle) foot at the lower end and (ii) adjustable femur with flexible socket at the upper end. For analysis, the 3D model was saved in stp format and sent to finite element software ANSYS.

Fig. 1 Model of lower-limb prosthesis. a Socket and femoral structures. b Tibial structure and foot

348

S. V. Motghare

Fig. 2 Dimensions of various parts

2.2 Model Development Geometry and Mesh Generation A three-dimensional model for lower-extremity amputees was built in CATIA V5 software. The model has a spherical socket and a ball fixed with 2.5 mm horizontal shaft that helps in one-dimensional rotatory motion. The design has tibial length 360 mm adjustable to (360 + 140) mm, diameter 40 mm and femoral length 70 mm adjustable to (70 + 46) mm, diameter 38 mm. The dimensions of various parts of the prosthesis are shown in Fig. 2. For analysis, the 3D model was saved in stp format and sent to finite element software ANSYS. Static structural analysis was assigned to the proposed prosthesis in ANSYS. Stainless steel and carbon fibre material were assigned to principle components of the prosthesis. The model serves to characterize solid support to function as bone and yield information that is otherwise challenging to obtain experimentally. The finite element model analyses the 3D model accurately and its reliability depends upon construction of appropriate geometry, assigned material properties, simulated interactions and constraints. For accurate analysis of different parts of the model, it was converted into finite elements that may be tetrahedral, triangular or hexagonal in shape. The more the elements more is the accuracy in the results. After developing mesh on complete model, the number of nodes and elements formed was found to be 255,771 and 134,659, respectively, shown in Fig. 3.

3 Result and Discussion On the basis of analysis and results, it was observed that the structure made of stainless steel experiences maximum principle stress 0.8539 MPa at tibio-femoral joint,

Design, Modelling and Optimization of Artificial …

349

Fig. 3 Mesh generation

maximum total deformation 0.000183 mm and total weight of principle components 5.2665 kg. On the other hand, carbon fibre experiences maximum principle stress 0.79081 MPa, maximum total deformation 0.000393 mm that lies within the ultimate stress limits and the total weight of the principle components was found to be 2.6 kg, shown in Fig. 4a, b, respectively. On considering carbon fibre material, the percentage reduction of weight was found to be 50.63%. Based on the percentage reduction in weight, carbon fibre was selected as a material for the prosthetic limb;

Fig. 4 a Maximum principle stress. b Total deformation

350

S. V. Motghare

this keeps the weight in permissible limits to maintain stability as well as ease to lift by the amputee limb. During gait, the foot acts as a shock absorber and supports the structure by balancing it. The principle components may be covered with cosmetic cover of flexible polymeric material to impart shape of a normal leg. The purpose of prosthetic knee is to restore normal flexion–extension movement, that is alignment controlled, i.e. knee axis is positioned perpendicular to the weight line, heel strikes to mid stance in such way that an inherently stable knee extension moment is provided. The free movement occurs only in one direction that too is restricted by two springs on the posterior side. The energy required for bending the knee would be very less as the springs does not need to be compressed but only a transverse movement should occur. The client’s ability to voluntarily control the knee during stance would depend on the hip musculature, length of residual limb and stability of the knee unit (prosthesis). A flexible urethane foam cover similar to the anatomical limb may be applied over the proposed design.

4 Conclusion A lower limb was simulated and analysed using finite element analyser (ANSYS R16.0). Total deformation, maximum principle stress and mass of the proposed model were obtained for stainless steel and carbon fibre materials. On analysing results for material optimization, carbon fibre was selected on the basis of percentage reduction in weight of the principle components of artificial limb. Future work will involve reproducing the FE model for experimental validation. Also, design a limb that will show frictionless movement along rotational axes of the tibio-femoral joint to support sitting (on floor) posture and gait.

References 1. Sanders GT, May BJ, Hurd R, Milani J (1986) Lower limb amputations: a guide to rehabilitation. F. A. Davis, Philadelphia 2. Thomas CL (1993) Taber’s cyclopedic medical dictionary, 17th edn. F. A. Davis, Philadelphia 3. Barbara Silver-Thorn M, Standard handbook of biomedical engineering and design. Marquette University, Milwaukee, Wisconsin 4. Shurr DG, Cook TM (1990) Prosthetics and orthotics. Appleton & Lange, East Norwalk 5. Wilson AB (1989) Limb prosthetics, 6th edn. Demos Publications, New York 6. May BJ (1996) Amputations and prosthetics: a case study approach. F. A. Davis, Philadelphia 7. Kaul V (2014) J Biomech Eng 136/011002-1. https://doi.org/10.1115/1.4025692 8. Seelen HAM, Anemaat S, Janssen HMH, Deckers JHM (2003) Effects of prosthesis alignment on pressure distribution at the stump/socket interface in transtibial amputees during unsupported stance and gait. Clin Rehab 17:787–796 9. Krouskop TA, Cosmetic covers for lower-limb prostheses. Bioengineering 10. Donahue TL, Hull ML, Rashid MM, Jacobs CR (2002) A finite element model of the human knee joint for the study of tibio-femoral contact. ASME J Biomech Eng 124(3):273–280

Design, Modelling and Optimization of Artificial …

351

11. Li G, Gil J, Kanamori A, Woo SL (1999) A validated three-dimensional computational model of a human knee joint. ASME J Biomech Eng 121(6):657–662 12. Penrose JM, Holt GM, Beaugonin M, Hose DR (2002) Development of an accurate threedimensional finite element knee model. Comput Methods Biomech Biomed Eng 5(4):291–300 13. Ramaniraka NA, Saunier P, Siegrist O, Pioletti DP (2007) Biomechanical evaluation of intraarticular and extra-articular procedures in cruciate ligament reconstruction: a finite element analysis. Clin Biomech (Bristol, Avon) 22(3):336–343 14. Motghare SV, Motghare SV, Athavale VM, Titermare VP (2019) Modeling and analysis of the brake disc in automobiles using ANSYS. Int J Innov Eng Sci 4(8):62–65 15. Douglas TS, Solomonidis SE, Lee VSP, Spence WD, Sandham WA, Hadley DM (1999) Automatic segmentation of magnetic resonance images of the trans-femoral residual limb. Med Eng Phys 20:756–763

An Experimental Study on Surface Roughness in Slicing Tungsten Carbide with Abrasive Water Jet Machining Ranjan Singh, Virendra Singh, and T. V. K. Gupta

Abstract In the present study, an attempt has been made to slice tungsten carbide rod with an abrasive water jet machining technique. The objective of this work is to evaluate and characterize the surface roughness of the machined surface. Experiments are carried with varying traverse speeds keeping other parameters constant. The surface is examined in detail using a 3D Optical Profilometer to characterize the roughness parameters which include centre line average deviation (Ra ), root mean square deviation (Rq ), total height (Rt ), maximum peak height (Rp ) and maximum valley depth (Rv ). It is observed that except Rv , all other parameters increase with an increase in traverse speed. Keywords Tungsten carbide · AWJM · Traverse speed · Surface roughness

1 Introduction On economic reasons and ecological considerations such as conservation of material and energy resources with waste reduction, there is an increase in the need for adequate limitation of wear and corrosion damage of machines and parts. Similar to this, there is also a demand from industry for wear-resistant materials having heavy tribological properties preferably without lubrication (e.g. chisels, cutting tools, forming dies, punches, etc.). Further, in aerospace and automobile sectors, advanced materials such as engineering ceramics which are ultra-hard, highly erosion/friction resistant and high temperature resistant are more acclaimed and applied [1]. The need for such high wear resistance and hardness materials is used in a variety of applications. Tungsten carbide (WC), composite material also known as cemented carbide, is manufactured by powder metallurgy process. Carbides are known for their high R. Singh · T. V. K. Gupta (B) Visvesvaraya National Institute of Technology, South Ambazari Rd., Ambazari, Nagpur, Maharashtra 440010, India e-mail: [email protected] V. Singh Indian Institute of Technology, Kanpur, Kanpur 208016, India © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_41

353

354

R. Singh et al.

wear resistance and (hot) hardness. This is the reason why WC has gained popularity as a cutting tool material for die and mould making. Hard to cut materials can be easily machined by these carbides, and these carbides are difficult to machine by conventional manufacturing methods. Since the thermal expansion and thermal conductivity are low with a high melting point, brittleness is the major source of machining difficulties which generate high thermal stresses. The micro-cracks enlarge and lead to macro-crack and fragmentation [2]. The available literature reveals that most of the work is accomplished in machining different metallic, non-metallic and composite materials only, but a limited literature was found on machining of carbides by AWJM technique. Ahsan Ali Khan et al. carried machining of carbide grade P25 using AWJM with an objective to study the effect of jet pressure, abrasive flow rate and traverse speed on the smoothness of the surface produced [3]. It was reported that better surface is obtained at higher pressure, lower traverse speed and abrasive flow rate and concluded that at the jet entrance, good surface is obtained than at the bottom of the cut. This is because of the type of cutting wear depth and deformation wear depth [4, 5]. The present work is an attempt to cut WC rod with a high-pressure AWJM which has the advantages of minimum heat generation and no metallurgical changes in the work material, and also, the surface is almost free from residual stresses. Here, investigations are carried on the sliced surface of WC produced by AWJM with different traverse speeds indicating the interaction time between the abrasive jet and work material. Machined surface characterization is carried using a 3D Optical Profilometer.

2 Experimentation 2.1 Material Tungsten carbide material is referred to as cemented carbide since it is cemented to the WC particles with a metallic binder (WC-Co) during the sintering process forming a strong metallurgical bond. The proportion of carbide phase is normally 70–97% of the total weight, and grain size varies between 0.4 and 10 µm. The WC used in the present work is a composition of 94% WC and 6% cobalt (Co), while the grain size is 1 µm. The WC work piece is a square rod of rod size 6 mm × 6 mm × 130 mm. This grade is also called as straight grade cemented carbide. Table 1 gives different mechanical properties of WC-6Co.

An Experimental Study on Surface Roughness … Table 1 Mechanical properties of WC

355

Property

Value

Density

14.9 g/cm3

Hardness

1580 HV30/91.8 HRa

Young’s modulus

630 MPa

Thermal conductivity

80 W/mK

Coefficient of thermal expansion

5.0 × 10−6 K−1

Fracture toughness

6.9 N/mm2 m0.5

2.2 Machining Conditions All the experiments were performed on a 3-axis Abrasive Water Jet Machining Center (Model No. OMAX 2652) which has a cutting pressure of up to 40 KSI. The maximum possible traverse speed is 4000 mm/min with an accuracy and repeatability of ±0.025 mm, respectively. Once the material selection is done and considering the machine parameters for experiments, fixtures are fabricated for holding the work piece. While conducting the experiments, some of the machine parameters are kept unchanged and are shown in Table 2. The experiments were performed at three different quality levels: 1, 2 and 3, respectively, which indicate that as traverse speed decreases, the quality level increases. This makes the difference in quality with the traverse speed of the jet. The traverse speed conditions for different quality levels are given in Table 3. The machined sample is shown in Fig. 1. Table 2 Fixed machining parameters

Table 3 Quality level and corresponding traverse speed

Parameter

Value

Mixing tube diameter

0.7620 mm

Jewel diameter

0.3556 mm

Abrasive flow rate

0.3 kg/min

Abrasive size

80 mesh

Water jet pressure

5 KSI

Stand-off distance

1 mm

Quality level

Traverse speed (mm/min)

1

0.13

2

0.11

3

0.07

356

R. Singh et al.

Fig. 1 WC-Co samples after cutting

2.3 Measurements Surface finish is the nature of a surface as defined by characteristics of lay, surface roughness and waviness comprising of small local deviations of the surface from a perfectly flat ideal surface. The surface roughness includes irregularities of small wavelength caused by cutting tool feed marks or disturbances such as wear, friction or corrosion. Surface roughness influences the wear between sliding surfaces, energy consumption and product life where all of them have an impact on economics of machining. As per ISO 4287:1997 (or IS 15262:2002) standards, the surface roughness is characterized by centre line average deviation (Ra ), root mean square deviation (Rq ), total height (Rt ), maximum peak height (Rp ) and maximum valley depth (Rv ), respectively. All the surface roughness measurements are made using a 3D Optical Profilometer (Model No. Bruker ContourGT-K), and size of the surface scanned for measurement is 0.48 mm × 0.36 mm. Optical profilers are the interference microscopes, well-established accurate surface measurement instruments precisely to measure surface finish and roughness on any shape using the wavelength of light as the ruler.

3 Results and Discussion The sliced samples are measured for surface roughness using a 3D Optical Profilometer. The 2D top surface, 3D view of the sample and the surface profile extracted from the Optical Profilometer for the quality level 1 machined surface at one location are shown in Fig. 2a–c, respectively. Table 3 gives the analytical results obtained after measurements. Each sample is measured at three different locations, and the average of these values is considered for analysis, as given in Table 4. Table 5 gives the average values of the machined samples at different traverse speeds (Table 6). The graphs indicated in Fig. 3a–e show a similar increasing trend of the roughness parameters with increasing traverse speed. The basic reason is that when the traverse speed increases, the abrasive particles get lesser time to cut the material, and their interaction with the material is less, leading to lesser cutting action. So it can be

An Experimental Study on Surface Roughness …

357

Fig. 2 a 2D top surface. b 3D view. c Extracted surface profile Table 4 Analytical results

Parameter

Value (µm)

Ra

2.462

Rp

17.896

Rq

3.098

Rt

34.491

Rv

−16.595

Table 5 Average values of surface roughness parameters for quality 1 sample Roughness parameter

Location 1 (µm)

Location 2 (µm)

Location 3 (µm)

Average (µm)

Ra

3.192

3.280

3.026

3.166

Rp

23.569

28.894

32.941

28.468

Rq

4.110

4.224

4.114

4.1493

Rt

46.523

51.747

57.265

49.135

Rv

−22.954

−22.854

−24.324

−23.377

Table 6 Roughness values at different traverse speeds Traverse speed (mm/min)

Ra (µm)

Rp (µm)

Rq (µm)

Rt (µm)

Rv (µm)

0.13

3.166

28.468

4.149

49.135

−23.377

0.11

3.033

26.474

4.008

48.213

−21.498

0.07

2.209

19.898

2.854

37.301

−17.403

358

R. Singh et al.

Fig. 3 Variation of roughness parameter. a Ra . b Rp . c Rq . d Rt . e Rv with traverse speed

An Experimental Study on Surface Roughness …

359

concluded that to achieve lesser surface roughness on the machined surface of WC, the traverse speed should be as low as possible.

4 Conclusions It is observed from the experiments that an increase in traverse speed increases the surface waviness. For small traverse speeds, the waviness is almost eliminated. Therefore, an AWJ with high cutting capability generates a regular striation structure which is due to external effects such as vibrations and fluctuations are terminated. Acknowledgements The authors thank Head, 4i Laboratory, IIT Kanpur, for permitting us to conduct the experiments and for the support provided in measuring the surface roughness using 3D Optical Profilometer available in Manufacturing Science Laboratory, IIT Kanpur.

References 1. Bonny K, De Baets P, Vleugels J, Van der Biest O, Lauwers B, Liu W (2009) EDM machinability and dry sliding friction of WC-Co cemented carbides. Int J Manuf Res 4(4):375–394 2. Mahamat ATZ, Rani AMA, Husain Patthi (2011) Machining of cemented tungsten carbide using EDM. J Appl Sci 11(10):1784–1790 3. Khan AA, Bin Awang ME, Bin Annuar AA (2005) Surface roughness of carbides produced by abrasive water jet machining. J Appl Sci 5(10):1757–1761 4. Hashish M (1991) Wear modes in abrasive water jet machining. Trans ASME PED 54:141 (Tribological Aspects in Manufacturing, 141, ASME) 5. Momber AW, Kovacevic R (1998) Principles of abrasive water jet machining, 1st edn. Springer, Berlin

Energy Absorption Characteristics of Single and Double-Walled Square Tubes Subjected to Axial Crushing Sanjay S. Toshniwal

and Raghu V. Prakash

Abstract Experimental and numerical study of single and double-wall square tube under axial compression was carried out at a displacement rate of 100 mm/min. Two configurations of double-wall tube, viz., parallel and diamond were explored. During axial compression of double-wall tubes, crumpling takes place either at same end or at opposite ends for the two tubes. Deformation at the same end absorbs 3–5% more energy than opposite end crushing. For the same type of crumpling, diamond arrangement absorbs 5–8% more energy than parallel configuration. Two different lengths of unequal tubes for parallel and diamond arrangement were examined. It was observed that unequal length tubes improve crashworthiness characteristics. Keywords Double-wall tube · Diamond · Crashworthiness · Energy absorption

1 Introduction Thin-wall tubes are widely used in crashworthiness applications. Thin-wall tubes exhibit high energy absorbing capacity during axial loading through a phenomenon termed as progressive buckling. The different modes of progressive collapse in square tubes, namely extensional, symmetric, asymmetric, have been studied analytically and experimentally by Abramowicz and Jones [1]. To increase energy absorption capacity, rib reinforced extruded tubes have been considered [2]. Further, the effect of multi-cornered, multi-cell tubes has been probed by researchers. It was proposed that by increasing the number of corners of a cross section, the energy absorption capacity of thin-walled columns will increase. Apart from this, foam-filled tubes have been considered by researchers [3], and placing one tube inside another has been explored [4] to enhance the energy absorption capabilities. Double-wall tube and interaction between two tubes are a less explored area. This study attempts to understand the crashworthiness characteristics of square tubes and S. S. Toshniwal · R. V. Prakash (B) Department of Mechanical Engineering, Indian Institute of Technology Madras, Chennai, India e-mail: [email protected]; [email protected] © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_42

361

362

S. S. Toshniwal and R. V. Prakash

Fig. 1 Cross section of specimen a single wall tube 1, b single wall tube 2, c parallel arrangement, d diamond arrangement

double-wall tubes with different configurations. For better safety, the structure should absorb as much as energy with minimum peak force possible. Hence, unequal length tubes are examined.

2 Geometrical Details In this study, commercially available extruded square aluminum columns are used. Thin-walled square tubes of two different cross sections, one with 25.4 mm × 25.4 mm and another with 50.8 mm × 50.8 mm (outer dimensions) having a length of 150 mm and 1 mm thickness, are used for this study (Ref. Fig. 1a, b). Prior to this, single wall tubes (SWT) of above two different cross sections are investigated. Two tubes kept one inside other is called double-wall tube (DWT). In this study, two arrangements of aluminum tubes which are parallel (P) and diamond (D) are explored as shown in Fig. 1c, d. In the case of DWT, two tubes can have equal or unequal lengths. DWT with an equal length of 150 mm with parallel and diamond arrangement is examined. For DWT with unequal length (Un), outer tube length is kept constant at 150 mm while two variations of inner tube length, viz., 147 and 153 mm are considered for parallel and diamond arrangement. The cross section, length, arrangement and abbreviations used in this study for simplicity are shown in Table 1.

3 Finite Element Modeling Crushing of specimens was simulated by non-linear explicit finite element code LS Dyna® . For the purpose of modeling, specimen is held between the rigid block and rigid upper plate. The upper plate was modeled using shell element and the lower rigid block was modeled using a brick element. For the modeling of thin-wall tubes,

Energy Absorption Characteristics of Single and Double … Table 1 Specimens name and corresponding lengths and arrangement

363

Specimen name

Cross section (mm)

Length (mm)

Arrangement

SWT1

25.4 × 25.4

150

None

SWT2

50.8 × 50.8

150

None

DWT-P

Two tubes

150

Parallel

DWT-D

Two tubes

150

Diamond

Un-147-P

Two tubes

Inner—147 Outer—150

Parallel

Un-147-D

Two tubes

Inner—147 Outer—150

Diamond

Un-153-P

Two tubes

Inner—153 Outer—150

Parallel

Un-153-D

Two tubes

Inner—153 Outer—150

Diamond

Belytschko-Tsay shell element with five integration points is used. Mesh size of 2 mm was used for square columns, as it gave reasonable convergence of the solution. Piecewise linear plasticity material model for an elasto-plastic material was used for numerical simulation of the specimens. In this material model, the true stress–true plastic strain data obtained from tensile tests were used as input. The rigid block is fixed at the bottom and velocity is given to the upper plate in the downward direction for a specific time. For upper plate, all other degrees of freedom except z-translation degree of freedom are constrained. Automatic nodes to surface contact algorithm were used to model contact between the square column and upper plate, square column and lower block. Automatic single surface contact algorithm is used for each square column of the specimen to depict self-contact between shell elements. For double-wall tube specimens, the automatic surface to surface contact algorithm is used to capture the interaction between two tubes. The static and dynamic friction coefficients used were 0.2 and 0.3. To change the mode of buckling from extensional to symmetric mode (as observed in experiments) an imperfection of the order of 0.1 mm was added to the tube using the Perturbation node command. The crushing characteristic of the tube depends on the way it is manufactured, viz., whether is formed out of metal sheet or extruded [5]. Extruded square columns are used in this study. It was observed that corners are thicker compared to side thickness. Measurements were taken and corner modeling was done. For the rows of two elements at the corners, the thickness is given as 1.54 times thickness in side. Hourglass control was employed to control hourglass mode deformation.

364

S. S. Toshniwal and R. V. Prakash

Fig. 2 a Arrangement for the compressive test, b true stress versus strain curve from tensile test

4 Experimental Investigation 4.1 Tensile Tests To carry out a numerical simulation on the crushing of specimen, data about properties of the material from which tube walls are made are required. For this purpose, tensile specimen is cut from the side of the square tube, parallel to the direction of extrusion. Tensile test specimens are prepared according to ASTM E-08 standard. Tensile tests are done at a displacement rate of 5 mm/min. An axial extensometer is used for strain measurement. Stress is calculated based on the load value measured with the help of load cell. The engineering stress-strain curve is obtained from the test as shown in Fig. 2a. The material has yield strength (σ y ) of 111 MPa, Young’s modulus (E) of 68.59 GPa, Poisson ratio (μ) of 0.3, density (ρal ) of 2700 kg/m3 .

4.2 Compression Tests All compression tests are performed on a 50 kN computerized universal testing machine. For compression testing, two compression platens are used as shown in Fig. 2b. The lateral movement of the specimen is restricted by friction between the specimen and the platen. All the specimens were crushed at a displacement rate of 100 mm/min. Specimens are compressed by a distance of 50 mm. The ends of square tubes are properly machined so that it is perpendicular to the longitudinal axis of the tube and thus do not affect the folding pattern. The force-displacement curve is obtained for each configuration. Energy absorbed by the specimen is calculated

Energy Absorption Characteristics of Single and Double …

365

as the area under force versus displacement curve. Crush force efficiency for each specimen is determined by taking the ratio of mean force of crushing to Peak force of crushing.

5 Results and Discussion Numerical and experimental force versus displacement curves for crushing of SWT1 and SWT2 are shown in Fig. 3. During crushing, plastic hinge is formed first which results in peak load. After hinge formation, bending of tube follows which completes the plastic fold. A symmetric progressive collapse mode is noticed in the tubes. In SWT1, formation of two full plastic folds and an incomplete fold is observed during crushing. In SWT2, one complete fold and next plastic hinge is formed. This corresponds to the second peak force in Fig. 3b. Experimental and numerical deformation of SWT1 is shown in Fig. 4a, b. Deformation in SWT2 is shown in Fig. 4c, d. The basic fold length in SWT2 is more than that of SWT1. Due to smaller fold length, more folds and more plastic hinges are formed in SWT1; hence, it can be said that SWT1 has more specific energy absorption capacity. Figures 4 e-i show various types of deformations in double-wall tubes. In the case of DWT, two types of interactions are observed between the tubes. In one type, two tubes start their progressive collapse at two different ends such that there is no contact between plastic folds of two tubes but a collapsed part of one tube will contact non-collapsed part of another tube as shown in Fig. 4f, i. This is referred as opposite end crushing. Another interaction is when two tubes collapse at same end. In same end collapse case, plastic folds of two tubes contact each other and penetrate one into the another as shown in Fig. 4g, h. It is termed as the same end crushing. In case of DWT-P, same end of crushing specimen absorbs 5.5% more energy compared to opposite end of crushing. In case of same end crushing, due to interpenetration between the folds, there is more interaction and subsequently, resistance offered for crushing increases. Hence, small increase in energy absorption is noticed.

Fig. 3 Experimental and numerical force versus displacement for SWT1 and SWT2

366

S. S. Toshniwal and R. V. Prakash

Fig. 4 a, b Experimental and numerical progressive buckling in SWT1. c, d Experimental and numerical progressive buckling in SWT2, e different plastic hinge formation in diamond arrangementf DWT-D with opposite end crushing, g DWT-D with same end crushing, h DWT-P with same end crushing, i DWT-P with opposite end crushing

In diamond arrangement interaction is much stronger. The interaction starts early in diamond crushing. This results in the formation of the new plastic hinge as shown in Fig. 4e. This plastic hinge is observed in the outer tube in all diamond arrangements of tubes. Due to this new hinge formation for both same end and opposite end crushing, the diamond arrangement absorbs 5–8.7% more energy compared to parallel configuration (Table 2). In DWT crushing, peak force is found to be higher in case of same end crushing compared to opposite end crushing. This effect is more prominent in case of diamond arrangement and it is due to the interaction between two tubes. When the energy absorption of the double-wall tube is compared with the sum of energy absorbed by individual tubes (SWT1 and SWT2), minimum increase of 12% and maximum increase of 25.8% is noticed as shown in Fig. 5a. Crushing of DWTs creates a higher peak force. To decrease the peak force during crushing, unequal length tubes are studied. In unequal length case, there is gradual increase in stiffness. This results in two initial peaks corresponding to two tubes. The peak force of SWT2 is 19.6 kN. This is the minimum possible peak force for double-wall configuration with unequal length. At this peak force, for the maximum energy absorption, two peaks should be of equal height, which is the maximum limit force in crushing. It is found that inner tube (SWT1) with 147 mm length and outer tube (SWT2) with 150 mm length is suitable for this objective. Another configuration

Energy Absorption Characteristics of Single and Double …

367

Table 2 Crashworthiness characteristics of different specimens in the study Specimen

Peak force (N)

Energy absorbed (J)

Crush force efficiency

Mode of interaction

SWT1

10,646

272.49

0.511

No interaction

SWT2

19,673

346.14

0.351

No interaction

SWT1 + SWT2

30,320

618.64

0.408

No interaction

DWT parallel

31,412

735.15

0.461

Same end crushing

DWT parallel

30,683

696.87

0.454

Opposite end crushing

DWT diamond

31,582

758.50

0.480

Opposite end crushing

DWT diamond

36,528

773.20

0.423

Same end crushing

Un-147-P

19,116

632.28

0.661

Same end crushing

Un-147-P

20,354

594.79

0.584

Opposite end crushing

Un-147-D

19,681

663.28

0.674

Opposite end crushing

Un-153-D

26,418

679.83

0.514

Opposite end crushing

Un-153-P

25,151

672.16

0.534

Opposite end crushing

Fig. 5 a Comparison between force versus displacement diagram of SWT1, SWT2, and DWT-D, summation of force of SWT1 and SWT2, b Comparison between force versus displacement diagram of SWT2, Un-147-D, and Un-153-P

368

S. S. Toshniwal and R. V. Prakash

with inner tube length 153 mm and outer tube length 150 mm is also studied. It will have peak force that is between the values of Un-147 and DWT. In this study for all unequal length tubes, the same end crushing mode of interaction was observed only in Un-147-P. For all other configuration opposite end, crushing is observed. As seen in DWT, in this case too, the same end crushing absorbs 6% more energy compared to opposite end crushing. In Un-153 case, the inner tube collapsed first, while in Un-147 outer tube buckled first. In case of Un-147-D specimen, for the same peak force as SWT2, energy absorbed is 91.6% higher. When compared to DWT diamond configuration, peak force decreased by 37.6% and energy absorbed by 12.4% as shown in Fig. 5b. Comparison between Un-153-P and DWT-P suggests a decrease in peak force by 18% while energy absorption is decreased by only 3%. Un-153 tubes displayed increased peak force and corresponding energy absorption All unequal length tubes have better crush force efficiency (CFE) than DWT (Table 2). The numerical model showed less energy absorption compared to experimental results. Peak force values obtained by numerical method correspond well with experimental results. The deformation pattern by numerical and experimental methods was found to be similar. The numerical force versus displacement curves followed the trend of experiments.

6 Conclusion It is found that the energy absorption of the double-wall tube is greater than the sum of energy absorption of two tubes it is made of. This is due to interaction effect between tubes. The minimum increase was 12% for parallel and maximum increase up to 25.8% for diamond. Two types of interactions are observed in double-wall tubes. Same end crushing absorbed about 3–6% more energy than opposite end crushing. Due to formation of plastic hinge for same type of crushing interaction, diamond arrangement absorbs 5–8% more energy than parallel arrangement. In tubes with unequal length, for a small loss in energy absorption, a large decrease in peak force is observed. Unequal tubes have better crush force efficiency than DWT. When compared to DWT, decrease in peak force is around 37% for Un-147 and 18% for Un-153 tubes. Hence, with proper design unequal length tubes can optimize crashworthiness characteristics.

References 1. Abramowicz W, Jones N (1984) Dynamic crushing of axial square tube. Int J Impact Eng 2(2):179–208 2. Prakash RV, Ghogare RS (2014) Energy absorption characteristics of rib reinforced thin-wall extruded structure subjected to axial crush. Int J Veh Saf 7(3/4):345

Energy Absorption Characteristics of Single and Double …

369

3. Baroutaji A, Sajjia M, Olabi AG (2017) On the crashworthiness performance of thin-walled energy absorbers: recent advances and future developments. Thin-Walled Struct 118:137–163 4. Haghi Kashani M, Shahsavari Alavijeh H, Akbarshahi H, Shakeri M (2013) Bitubular square tubes with different arrangements under quasi-static axial compression loading. Mater Des 51:1095–1103 5. Jusuf A, Dirgantara T, Gunawan L, Santosa SP, Putra IS (2017) Corner modelling strategy for finite element impact simulation of extruded square thin-walled column. Procedia Eng 173:1307– 1313

Field Data Analysis Using Work Measurement Techniques in a Packaging Industry Chinmay M. Salkar , Gaurao J. Tapare , Mayank A. Murkute , Chetan R. Zingre , Hansraj A. Mohod , and Vinit S. Gupta

Abstract In this work, an investigation is carried out in a packaging industry, particularly in dispatch section, using work measurement techniques. Operations in the plant were observed, and a standard process flow is designed so as to minimize unnecessary movements and delay. Further, as an outcome of time study, operating cycle time is fixed for both rim packaging and box packaging procedures using line balancing approach. Finally, the daily production target is revised based on the results. The work is aimed at planning the production output for industries having a product mix using time study techniques. Keywords Time study · Operation · Cycle time · Process optimization

1 Introduction In the current industrial scenario, optimization of production process variables plays an important role. This is due to the fact that in mass production and global market approach, products must be offered at competitive price. In order to reduce the overall production costs, it becomes critically important to control variation in product, process, and other manufacturing delays. Therefore, tools such as work measurement, statistical quality control, and lean manufacturing were evolved to support in eliminating wastages of all kind [1, 2]. The use of work study (method study and work measurement) allows the determination of optimum cycle time for individual operations as well as also focuses on overall plant productivity [1, 2]. Method study is a means of enhancing the production efficiency (productivity) of the firm by the elimination of waste movements and unnecessary operations [3]. At the same time, work measurement focuses on gathering data pertaining operation time using adequate sampling for different operations to come out with optimum operation sequencing and cycle time [2, 3]. It also helps to identify the productivity of different C. M. Salkar · G. J. Tapare · M. A. Murkute · C. R. Zingre · H. A. Mohod · V. S. Gupta (B) Department of Mechanical Engineering, S.B. Jain Institute of Technology, Management and Research, Nagpur, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_43

371

372

C. M. Salkar et al.

operators and operating conditions. Work study improves upon the existing process or method and helps in standardization and simplification resulting in appropriate manpower planning, production planning, etc. [1, 2]. Further, lean manufacturing techniques are evolved for shop floor optimization. Lean manufacturing is a methodology that focuses on minimizing waste within manufacturing systems while simultaneously maximizing productivity [4, 5]. The benefits of lean include reduced lead time, reduced operating costs, and improved product quality, to name just a few [6, 7]. Together, all these standard plant optimization techniques help to improve the productivity and reduce the wastages of all kind. In this work, the dispatch section of a medium-scale packaging industry is studied so as to identify the bottleneck operations. This work is carried out at a printing and packaging industry located at MIDC, Hingna region of Nagpur, India, with an annual turnover of 5000 million Indian rupees. During the initial stage, layout of dispatch section is studied, various packaging processes are observed, and process flowchart is prepared. Time study is performed on packaging processes (film wrap stretching process). During time study, bottleneck operation and non-valuable operation and movements are identified. Finally, using line balancing technique and classifying operations into stations, cycle time is evaluated. Section 2 presents layout of dispatch section and process flowchart, Sect. 3 gives the time study, and Sect. 4 presents important conclusions.

2 Dispatch Section: Layout and Process Flow In this section, the layout of dispatch section of the packaging industry is discussed. The section consists of stretching, packing, weighing, and storage as main operations. Figure 1 gives the layout of dispatch section. It can be observed that machines are arranged in a linear pattern, and there is ample free space on the floor. Dispatch is a crucial section in that industry and hence is selected for time study. Tables 1 and 2 give the sequence of operations carried out in packaging of a roll and of a box. It is observed that there is a considerable delay introduced in weighing process due to slip generation. This slip mentions all the parameters of the roll or box and is computer generated. A considerable difference between roll packaging and box packaging operations is that in box packaging operation, additional supporting operations such as preparation of boxes is required.

3 Time Study and Analysis Time study is carried out by classifying operations into workstations and observing operation as well as station time. Three workstations were classified as given in

Field Data Analysis Using Work Measurement …

373

Fig. 1 Layout of dispatch section

Table 1 Machine 1: roll packaging (2-worker) process flowchart S. No.

Process

Symbol

Worker

1

Transport roll pallet to machine

→

1

2

Packaging/stretching roll

O

1

3

Delay in weighing

D

–

4

Weighing

O

1

5

Storage

▼

1

Table 2 Machine 2: box packaging (3-worker) process flowchart S. No.

Process

Symbol

Worker

1

Move pallet to table

O

1

2

Applying tape on box and moving to machine

O

1

4

Packaging

O

1

5

Delay in weighing

D

–

6

Weighing

O

1

7

Transport to storage

→

1

Table 3. A certain number of micro-operations are carried out in each workstation, resulting in total operation time at a particular workstation. Sample sizes of around 22 readings are taken, and average station time and standard deviation are evaluated. Table 4 presents the mean and standard deviation of

374

C. M. Salkar et al.

Table 3 Classification of operations in workstations S. No.

Workstation

Operations

1

1

I. Take box II. Take roll III. Find SR No. IV. Attach core plug V. Fill roll in box

2

2

I. Machine 1 operate II. Stretching operation

3

3

I. Lift box from weight machine and place it to pallet II. Attach slip on box

4

I. PC operator for machine 1 II. Read weight from weight machine screen and generate slip for box III. After completion of pallet generate pallet slip IV. Also gives instruction to worker parallelly

Table 4 Average time and standard deviation for roll packaging and box packaging at each workstation

Item

Station

Mean time (s)

Standard deviation

Roll packaging

One

13.18

0.19

Box packaging

Two

19.77

0.15

Three

20.16

0.10

One

41.29

0.40

Two

36.17

0.05

Three

16.22

0.13

time required at a particular workstation based on the above sample. Based on line balancing concept, the cycle time for roll packaging operation t cr is fixed at 20.16 s and cycle time for box packaging operation t cb is fixed at 41.29 s.

3.1 Analysis This section focuses on the analysis of total production output if a certain percentage of total time is used for box packaging and remaining time is used for rim packaging operation, based on effective working time in a day of 14 h, i.e., 840 min: Box packaging: Considering cycle time of 41.29 s and a pallet consisting of 35 boxes, total time required to prepare a pallet is 24.09 min. Considering additional pallet handling time of 2.67 min as observed, total time required in preparing and handling a pallet for box packaging is 26.76 min. Number of boxes that can be packed in a working day, considering complete time is utilized for same is 1098.65 boxes.

Field Data Analysis Using Work Measurement …

375

Fig. 2 a Production output from box packaging and roll packaging against percentage of time, b total production output against percentage of time (zero on x-axis signifies that hundred percent time is dedicated to box packaging and hundred percent signifies that hundred percent of time is dedicated to roll packaging)

Roll packaging: Operation cycle time for roll packaging is 20.16 s. Considering same assumptions as above and if total time is dedicated to roll packaging, a total of 2034.18 rolls are packed per day. Figure 2a gives total output for box packaging and roll packaging if a certain percent (incremental from 0 to 100%) of working time is utilized for the operation. It can be clearly observed that the output of roll packaging is always higher than box packaging due to low operation cycle time. Figure 2b gives the total production output if total time available is distributed between box packaging and roll packaging. Here, zero percent on time scale signifies that 0% of time is dedicated to roll packaging and 100% of time is utilized for box packaging and vice versa. It is observed that with increase in total percentage of time for roll packaging, total production output increases. This analysis is useful for floor supervisor to fix that during a working day, what proportions of box and rolls can be packed so as to maximize the productivity, simultaneously maintaining the demand constraints.

4 Conclusions This study provides a basis for identifying and analyzing the packaging operations and fixing appropriate cycle time. Operations are classified into workstations and operating cycle time is fixed using line balancing approach. It is clearly established that box packaging requires higher cycle time than roll packaging. At the same time, the study for distributing total available time for box packaging and roll packaging is also conducted in the context of a particular packaging industry. This study helps to establish the appropriate production blend so as to maximize the total output for

376

C. M. Salkar et al.

an industry involved in handling a product mix. However, this study can further be repeated for higher sample size and different packaging industries.

References 1. Telsang M (2013) Industrial engineering and production management. S. Chand Publications, New Delhi 2. Razmi J, Shakhs-Niyaee M (2008) Developing a specific predetermine time study approach: an empirical study in a car industry. Prod Plan Control: Manage Oper 19(5):454–460 3. Bon AT, Daim D (2010) Time motion study in determination of time standard in manpower process. In: 3rd engineering conference on advancement in mechanical and manufacturing for sustainable environment 4. Omagbai O, Salonitis K (2017) The implementation of 5S lean tool using system dynamics approach. Procedia CIRP 60:380–385. https://doi.org/10.1016/j.procir.2017.01.057 5. Hall DJ (1996) The role of creativity within best practice manufacturing. Technovation 16(3):115–121 6. Pauline F, Rich N (2007) The meaning of lean: cross case perception of packaging business in the UK’s fast moving consumer good sector. Int J Logistics Res Appl 10(3):157–171 7. Abdullah F (2003) Lean manufacturing tools and techniques in the process industry with a focus on steel. Ph.D. thesis, University of Pittsburgh

Comparative Study of Nanofinishing of Si (100) Using DDMAF and Allied Processes Kheelraj Pandey , Ajendra Kumar Singh , and Gaurav Raj Pandey

Abstract Monocrystalline silicon wafers find its use in semiconductor industries for a variety of applications. Therefore, its surface finish is of prime importance. The surface finish of Si (100) for semiconductor application should have nanofinishing characteristics that resemble mirror-like surface characteristics of polished Si (100). The present paper outlines the comparative study of nanofinishing of monocrystalline silicon wafers, i.e., Si (100) using double disk magnetic abrasive finishing and allied processes. Starting from the chemical etching of Si (100) wafer in KOH solution and polishing by DDMAF process, the implementation of chemical oxidizers in the polishing region with constant flow rate and application of ultrasonic vibrations, the work highlights the consistent improvement in finishing efficacy. Keywords Magnetic abrasive finishing (MAF) · Double disk magnetic abrasive finishing (DDMAF) · Flexible magnetic abrasive brush (FMAB)

1 Introduction The worldwide application of silicon wafers is used in the production of microchips [1]. Silicon is a hard and brittle material and has hardness above 1000 Vicker units; therefore, its machining is quite difficult to execute. Various machining processes such as chemical mechanical polishing (CMP), abrasive machining, electrolytic inprocess dressing (ELID), and free abrasive machining have been useful in machining the silicon. The experimental studies on magnetic abrasive finishing (MAF) [2–6] have established that smooth finishing can be achieved with minimal surface defects on work materials having stringent properties. Recently, the surface finishing of Si (100) wafer has been done by double disk magnetic abrasive finishing (DDMAF) with chemical assistance [7, 8]. The surface quality of Si (100) was further improved K. Pandey (B) · A. K. Singh Amity University Uttar Pradesh, Lucknow Campus, Lucknow, India e-mail: [email protected] G. R. Pandey M.B.A., Operation Management, IGNOU, New Delhi, India © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_44

377

378

K. Pandey et al.

with the application of ultrasonic vibrations given to the Si (100) workpiece [9]. The comparison of surface roughness presented in the present work will highlight the process capabilities of DDMAF on Si (100) integrated with chemical assistance and ultrasonic approach.

2 Studies on Finishing Si (100) with Different Chemical Processes and DDMAF as a Mechanical Process The DDMAF technique was integrated with chemical processes like chemical etching of Si (100) wafer and the preferential application of chemical oxidizers in the polishing region to enhance the polishing characteristics of Si (100) wafer. The experimental setup of DDMAF and its combination with ultrasonic vibration is shown in Fig. 1a, b, respectively. The DDMAF experimental setup is comprised of primary magnetic disk and secondary magnetic disk, having the dimensions ∅ 25 mm × ∅ 3 mm. In each magnetic disk, four blind holes existed to form a magnetic pole. The silicon workpiece with diameter 100 mm, thickness 525 ± 25 µm, cut and lapped, p-type boron and resistivity of 0.001–1000 -cm was inserted in between the rotating magnetic disks aligned in the vertical direction, placed on the tray, meant for executing the polishing process. The allied processes which are used with the DDMAF are as follows:

Fig. 1 a DDMAF experimental setup [7, 8]. b DDMAF experimental setup integrated with ultrasonic vibration unit [9]

Comparative Study of Nanofinishing of Si (100) …

379

2.1 Surface Roughness of Si (100) Using Chemical Etching Process with DDMAF The integration of chemical etching process with DDMAF is implemented to obtain the polishes surface of Si (100). The concentration of potassium hydroxide (KOH) in 100 ml de-ionized water is selected in the range of 20% (w/w) to 60% (w/w). The solution formed for the aforesaid concentration of KOH is used for etching the Si (100) wafer. The solution formed is warmed at 80 ° C for 40 min of time interval and is allowed to cool, and isopropyl alcohol was added to the KOH solution to restrict the impingement of potassium ions due to the presence of hydroxide ions. Thereafter, the wafer was dipped in the KOH solution for 20 min. The etching procedure in this manner oxidized the silicon atoms from the wafer surface and reduced its hardness and thereby, resulted in the generation of soft surface with respect to the raw wafer. The application of DDMAF process after the etching process on the wafer further reduced the surface roughness of polished wafer. In this direction, the optimum value of surface roughness, obtained by using the GA function of MATLAB, is found to be 18.20 nm on atomic force microscopy scale. The optimum process parameters, namely polishing speed, working gap, abrasive mesh number, and concentration of KOH solution are 175 rpm, 5 mm, 1200, and 60% w/w, respectively [7].

2.2 Surface Roughness of Si (100) Using Chemical Oxidizers with DDMAF The use of chemical oxidizers with DDMAF has further enhanced the surface roughness value of Si (100) wafer to 6.9 nm on atomic force microscopy scale at experimental process parameters namely finishing speed of 175 rpm, working-gap of 5.40 mm, and flow rate of slurry in the polishing region being 10 ml/min. Throughout the experimental procedure, the pH of the slurry was kept at 12.29. The chemical oxidizers in proportionate amount, namely sodium hydroxide (15% w/w), Marshall’s acid salt (1% w/w), Caro’s acid salt (1% w/w), and hydrogen peroxide (0.05% w./w.) are used in slurry of alumina (20% w/w) in 100 ml distilled water. The addition of Marshall’s acid salt in the alumina slurry at pH 12.29 provided elevated oxidizing power, and the addition of Caro’s acid salt generated a hydrophobic over the surface of abrasive that leads to the improvement of surface quality of Si (100); also, the mixing of hydrogen peroxide with slurry created a passivation layer on the wafer surface to protect itself from the attack of hydroxide ions. Thus, the preferential use of chemical oxidizers with DDMAF improves the surface characteristics of polished Si (100) [8].

380

K. Pandey et al.

2.3 Surface Roughness of Si (100) Using Chemo-ultrasonic Machining Approach The induced ultrasonic vibrations during machining operation in the presence of abrasive particles cut complex and non-uniform shapes from the workpiece with extremely high precision. The combined implementation of USM and constant flow rate of chemical oxidizers with DDMAF, the surface roughness value of Si (100) is minimized to 1.5 nm on atomic force microscopy scale. The aforesaid surface roughness of polished Si (100) was obtained at finishing speed of 175 rpm, workinggap of 5.44 mm, and pulse-on-time of 9 s for rate of flow of slurry in the polishing region being 10 ml/min. The use of chemical oxidizers with constant rate of flow of 10 ml/min in the polishing region weakened the atomic bonding of silicon atoms in the wafer. Hence, the cutting force in the form of longitudinal motion of the abrasive particle (F u ) on the peaks of Si (100) generated due to the amplified ultrasonic vibrations further weakened the peaks and the rotation of flexible magnetic abrasive brush (FMAB) on the rotating magnetic disk attached to the spindle of CNC milling machine that sheared the peaks of Si (100) effectively [9].

3 Results and Discussion The surface characteristics of finished Si (100) were carried out on atomic force microscopy scale. The height of the peaks in raw Si (100) originally existed at 5.43 µm. The AFM image of raw Si (100) is presented in Fig. 2a. The peak height in raw Si (100) has been considerably reduced to 266, 96.8, and 59.9 nm by chemical etching process, implementation of chemical oxidizers, and integrated approach of

Fig. 2 AFM images of a raw Si (100) [7–9], b polished Si (100) in chemically etched in 60% KOH at optimum parameters by DDMAF [7]

Comparative Study of Nanofinishing of Si (100) …

381

chemo-ultrasonic with DDMAF, respectively. The surface roughness of 18.20 nm corresponding to the peak height of 266 nm for polished sample was obtained at optimum parameters of 175 rpm polishing speed, 5 mm working-gap, and 1200 mesh size alumina abrasive and 60% w/w KOH concentration [7]. The AFM image of the polished workpiece is presented in Fig. 2b. The use of chemical oxidizers further improved the surface quality of polished silicon wafer in which surface roughness was reduced to 6.9 nm corresponding to the peak height of 96.8 nm at optimum parameters of 175 rpm polishing speed, 5.40 mm working-gap, and flow rate of alumina slurry of 10 ml/min in the polishing region [8]. The AFM image of the polished silicon workpiece for the combined application of chemical oxidizers with DDMAF can be seen in Fig. 3a. The optimum parameters of 175 rpm polishing speed, working-gap of 5.45 mm, and pulse-on-time of 9 s in case of chemo-ultrasonic polishing of Si (100) further reduced the surface roughness to 1.5 nm corresponding to the peak height of 59.9 nm. The AFM image of polished silicon workpiece for chemo-ultrasonic polishing of silicon work material with DDMAF can be revealed from Fig. 3b. The polished samples at optimum polishing parameters for chemical etching process, use of oxidizers, and application of chemo-ultrasonic approach, combined with DDMAF can be seen from Fig. 4b–d. The raw sample of silicon material used for polishing with DDMAF is presented in Fig. 4a. The comparison of surface roughness for the polished sample at optimum experimental parameters can be seen in Fig. 5.

Fig. 3 AFM images of a polished Si (100) using chemical oxidizers with DDMAF at optimum parameters [8], b polished Si (100) using chemo-ultrasonic approach with DDMAF at optimum parameters [9]

382

K. Pandey et al.

Fig. 4 a Raw silicon wafer, b chemically etched and finished silicon wafer at optimal parameters, c finished silicon wafer using chemical oxidizers with magnetic abrasive particles at optimum process parameters, d polished silicon wafer using ultrasonic vibrations along with chemical oxidizers supplied to magnetic abrasive particles at optimum process parameter

4 Conclusions i.

Chemically etching the Si (100) wafer in KOH solution with isopropyl alcohol and finishing by DDMAF process at optimal experimental levels enabled to obtain the surface roughness of 18.20 nm on atomic force microscopy scale. ii. The improvement in polishing quality of Si (100) was noticed with the use of oxidizers, which were permitted to flow in the finishing region at the rate of flow of 10 ml/min. The surface roughness obtained in this case was 6.9 nm. iii. The process efficacy was further improved with the combination of chemical oxidizers and ultrasonic vibration provided to the Si (100) work material. The optimum experimental parameters in this case registered a surface roughness of 1.5 nm.

Comparative Study of Nanofinishing of Si (100) …

383

Fig. 5 Comparison of surface roughness on atomic force microscopy scale of polished Si (100) using different processes integrated with DDMAF

References 1. Sreejith PS, Udupa G, Noor YBM, Ngoi BKA (2001) Recent advances in machining of silicon wafers for semiconductor applications. Int J Adv Manuf Technol 17(3):157–162. https://doi. org/10.1007/s001700170185 2. Jain VK, Kumar P, Behera PK, Jayswal SC (2001) Effect of working gap and circumferential speed on the performance of magnetic abrasive finishing process. Wear 250–251:384–390. https://doi.org/10.1016/S0043-1648(01)00642-1 3. Kala P, Pandey PM (2015) Comparison of finishing characteristics of two paramagnetic materials using double disc magnetic abrasive finishing. J Manuf Process 17:63–70. https://doi.org/10. 1016/j.jmapro.2014.07.007 4. Sihag N, Kala P, Pandey PM (2015) Chemo assisted magnetic abrasive finishing: experimental investigations. Procedia CIRP 26:539–543. https://doi.org/10.1016/j.procir.2014.07.067 5. Yamaguchi H, Shinmura T (1999) Study of the surface modification resulting from an internal magnetic abrasive finishing process. Wear 225–229:246–255. https://doi.org/10.1016/ S00431648(99)00013-7 6. Mulik RS, Pandey PM (2011) Magnetic abrasive finishing of hardened AISI 52100 steel. Int J Adv Manuf Technol 55:501–515. https://doi.org/10.1007/s00170-010-3102-8 7. Pandey K, Pandey U, Pandey PM (2018). Statistical modeling and surface texture study of polished silicon wafer Si (100) using chemically assisted double disk magnetic abrasive finishing. Silicon 11(3):1461–1479. https://doi.org/10.1007/s12633-018-9961-6

384

K. Pandey et al.

8. Pandey K, Pandey PM (2018) Use of chemical oxidizers with alumina slurry in double disk magnetic abrasive finishing for improving surface finish of Si (100). J Manuf Process 32:138– 150. https://doi.org/10.1016/j.jmapro.2018.02.007 9. Pandey, K., & Pandey, P. M. (2019) An integrated application of chemo-ultrasonic approach for improving surface finish of Si (100) using double disk magnetic abrasive finishing. Int J Adv Manuf Technol 103:3871–3886. https://doi.org/10.1007/s00170-01903829-5

Optimization of Thickness of Hollow Punch–Die for Proposed Solar-Assisted Leaf Plate and Cup Making Machine Abhay Nilawar , Pravin Potdukhe , and Deepak V. Bhope

Abstract Leaf plates are traditionally made by hand in Indian villages. These are commonly used for serving food at family, religious and social functions. The laborious craft can now be converted into machine operation to make these containers in elegant shape and sizes and to make the plates much stronger than the normal one by the conventional method of making plates. The punch–die assembly of the traditional machine consists of an electric heater where the effective heating temperature is maintained from 110 to 120 °C. Alternatively, this temperature can also be achieved easily by using solar concentrator to heat the thermic oil which in turn can be used to heat the punch–die assembly. The hot oil will be circulated through hollow space of punch and die so that the green leaf will be heated from both the sides and occupied the required shape of cup or plate in minimum time while moulding in die. The thickness of such punch–die assembly is optimized analytically and also by FE software ANSYSTM . Keywords Leaf plate · Solar energy · Thermic oil · Optimization

1 Introduction The punch–die assembly of the traditional machine consists of an electric heater where the effective heating temperature is maintained from 110 to 120 °C. Alternatively, this temperature can also be achieved easily using the solar energy. Central India falls in a tropical region wherein a sufficient amount of solar heat energy is available for maximum number of sunshine hours and also for a maximum number A. Nilawar · P. Potdukhe (B) · D. V. Bhope Department of Mechanical Engineering, Rajiv Gandhi College of Engineering, Research and Technology, Chandrapur, Maharashtra 442402, India e-mail: [email protected] A. Nilawar e-mail: [email protected] D. V. Bhope e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_45

385

386

A. Nilawar et al.

of days (approx. 300–330 days). So, the solar concentrator can be used to heat the thermic oil. Thermic oil is a suitable fluid for passive heat storage and for supply of heat to the work station at punch–die assembly. Thus, the use of electric heater can be eliminated from traditional machine, and the cost of electrical energy consumption will be reduced to zero. So the above machine can be regarded as a green machine and a boon for rural people to earn their livelihood. The first important aspect is to optimize the design of hollow punch–die assembly. The hot oil will be circulated through hollow space of punch and die so that the green leaf will be heated from both the sides and occupied the required shape of cup or plate in less time while moulding in the die.

2 Literature Review Leaf plate and cup manufacturing is a well-known process in India. Use of banana leaves is a common practice in India to serve breakfast and lunch/dinner during family, social and religious functions. Hence, manufacturing of cups and plates of different size and shapes has emerged in many small-scale industries. The conventional type of machines is high-grade electrical energy based, i.e. electric energy (heaters) is used to heat the punch of die–punch assembly. All the machines available or manufactured in India are electric energy based. Some of them are bench model, stand model, multi-set type model-mass production type, foot press type and hand lever press type. In all of the above machines, either punch is lowered or die is moved upwards. In the majority of machines, punch is heated by means of electric heater. The punch– die assembly used in conventional machines are solid, rugged type, and hence, the design is easy and simple. Literature survey [1, 2] indicates that no researcher has developed solar-assisted type (thermic oil based) leaf plate/cup making machine, and no research paper is cited in the area of solar energy utilization for leaf plate/cup production.

3 Present Design In the present design, both punch and die are heated simultaneously by means of thermic oil heated at solar concentrator. A preliminary leaf cup/plate making machine [2] was developed on conventional (electric energy). In this machine, the die is heated and punch is pressed. The researchers had got an idea of manufacturing the leaf plate along with optimum temperatures needed at punch and die. The temperature of around 120 °C for the duration of one minute is sufficient to produce the leaf plate/cup. It is found that the punch should also be provided with heating arrangement so that both sides of plate are uniformly heated. The thermal energy needed for producing the leaf plate would be easily collected by solar insolation using effective solar

Optimization of Thickness of Hollow Punch …

387

concentrator where thermic oil is heated. The heated thermic oil is circulated in hollow punch and die to produce the leaf plate. The solar-assisted leaf plate making machine (SALPMM) consists of punch, die, thermic oil tank, solar concentrator and valves.

4 Optimization of Thickness of Hollow Punch–Die The main operating parts in the machine SALPMM is hollow punch–die system. The thickness of punch–die system should be proper to get the requisite shape of leaf plate, and it should also be able to withstand the various forces acting during the cup/plate manufacturing operation. Line diagram of punch and die is shown in Fig. 1. In the present research, the optimization of thickness of punch and die is carried out by analytically and by finite element analysis to get a true picture of the effect of forces acting thereupon.

4.1 Analytical Calculations The forces on the hollow punch are shown in Fig. 2. The analytical analysis is as follows. The force per unit length acting on the mean circumference is determined as 2.13 N/mm. The bending stress is calculated for different thickness of plate varying from 10 to 1 mm, and the results are given in Table 1. From the result, it is observed that 3 mm thickness of plate is safe, and 2 mm is also safe, but a factor of safety is less than 2.5, and hence analytically, 3 mm thickness of hollow punch–die plate is found to be most appropriate. Fig. 1 Line diagram of punch and die

Insulated Cover Plate

Hot Oil in Hot Oil

Hot Oil in

Oil out Hollow Punch Plate

Green Leaf

Hollow Die Plate

Hot Oil

Oil out Insulated Enclosure cum support str.

388

A. Nilawar et al.

Fig. 2 Forces on hollow punch

75 Kg = 735.75 N

60

One person with one leg can apply - 25 kg. If operator stands on pedal – 50 kg Leverage = 1.5 ( as b / a = 1.5) Maximum force = 50 x 1.5 = 75 kg= 735.75N

a

b

Table 1 Results form analytical calculations S. No.

Thickness of plate of punch for cup (mm)

Bending stress [MPa (N/mm2 )]

1

10

2.664

2

8

4.182

Allowable stress [MPa (N/mm2 )]

Factor of safety

Remarks

112

42.04

Very Safe

26.78

3

6

7.57

14.79

4

4

17.02

6.58

5

3

30.17

3.71

6

2

67.88

7

1

271.49

Safe

1.65 < 2.5

Unsafe

0.41

4.2 Finite Element Analysis Finite element model is shown in Fig. 3. The pressure load calculations for conical and bottom surfaces of the punch are as follows: Fig. 3 FE model

Pc

Pa

Pb

for pressure calculaƟon it is assumed that, for applied force, pressure at every locaƟon is same. i.e.

Pa = Pb = Pc

Pressure

50mm

Pressure - Pb 60 Pressure Pa

Optimization of Thickness of Hollow Punch …

389

For conical edge surface, Pb = Pa Cos (60°). For axi-symmetry, summation of Pa and Pb is equal to the total pressure. Hence, Pa + Pa Cos (60°) = total pressure and therefore Pa + 0.5Pa = total pressure. Alternatively in terms of force, Fa + 0.5Fa = 1.5Fa = total force. That is 1.5 Fa = 735.75 N, Fa = Ra = 490.5 N will act on the bottom surface and remaining Fb = Rb = 245.25 N will act on conical surfaces. The pressure acting on bottom surface is determined as 0.25 N/mm2, and the pressure on conical surfaces is determined as 0.026 N/mm2 . The analysis is carried out to determine the bending stress in hollow punch plate of different thickness, and the results of analysis by the software are shown in Fig. 4. The bending stresses determined using FE analysis for varying thicknesses are given in Table 2. • For 2 mm thick

• For 3 mm thick

Fig. 4 Stress distribution in hollow punch of different thickness

Table 2 Bending stresses for varying thickness of hollow punch–die plate S. No.

Thickness of hollow punch–die plate (mm)

1

10

2 3 4

Maximum bending stress (N/mm2 )

Maximum deflection

Factor of safety

Remarks

1.796

0.0013

62.36

Very safe

8

2.60

0.001

43.07

6

4.23

0.003

26.47

4

8.21

0.008

13.64

5

3

14.78

0.016

7.57

6

2

31.41

0.047

3.56 > 2.5

7

1

146.84

0.30

0.76

Safe Unsafe

390

A. Nilawar et al.

5 Discussion and Conclusion Optimization of thickness of hollow punch–die is essential. As thickness increases, the resistance to heat flow also increases. Hence, the thickness of hollow punch– die plate should be as minimum as possible so that it can sustain the load without its failure and can transfer maximum amount of heat from oil to green leaf for its moulding to a cup shape. From the results, it is concluded that, The heat transfer rate from the hollow punch to the leaf depends upon the thickness of punch plate. The lesser thickness will enhance the heat transfer rate. So the production time of leaf plate would be less, and the consumption of thermal energy would also be minimum. Hence, optimization of thickness of punch is a vital factor in design of solar-assisted leaf plate making machine. The thickness of hollow punch–die plate determined by analytical method is 3 mm and using FE analysis is 2 mm. The variations in the results are attributed to the approximations involved in analytical and FE analysis. It is evident from the present analysis that the thickness of minimum 2 mm and maximum 3 mm shall be most appropriate for hollow punch and die plate from strength point of view. This research work is further extended for transient and steady-state thermal analysis of hollow punch–die assembly and shall be reported soon.

References 1. Kalita P, Dixit US, Mahanta P, Saha UK (2008) A novel energy efficient machine for plate manufacturing from Areca Palm Leaf Sheath. J Sci Ind Res 67:807–811 2. Potdukhe PA, Nilawar AS (2011) Design, fabrication & performance studies on leaf plate making machine. In: National conference on advances in mechanical engineering, Nov 2011

Development of Briquette Cum Pellet Making Machine Yeshwant M. Sonkhaskar , Gajanan R. Nikhade , Saket Dharmik , Utkarsh Deshmukh , and Pramod Dhote

Abstract Fossil fuels are on the edge of extinction as a very less quantity of them is available now. They also have large ash content, and pollution takes place to a greater extent during combustion. So, there is a need to replace these fuels with something that has the same calorific value and causes less pollution. Biomass briquettes and pellets can be an alternative to coal. These briquettes and pellets are formed by compression of biomass that can be used as fuel. This reduces pollution and provides alternate low-cost fuel. Various machines have been developed for producing briquettes and pellets, but the cost of these machines is very high. After analyzing various machines, a machine that can make both briquettes and pellets through a single setup was not available. So, decision was made to produce a low-cost machine that can produce both briquettes as well as pellets through a single manually driven setup. The extrusion process has been used for producing briquettes and pellets, wherein the forward extrusion process was used for producing the briquettes and the backward extrusion process was used for producing pellets. Analysis, fabrication, and experimentation were carried out on fabricated briquette cum pellet machine. The objective of the experimentation was to find out the density of the briquette and pellet obtained using sawdust and aspen chips as raw material and a binder (bentonite or starch) combination. This machine can reduce farm waste into a useful product by also providing a source of livelihood. Keywords Briquette machine · Pellet machine · Manually driven machine · Scotch yoke mechanism

Y. M. Sonkhaskar (B) · G. R. Nikhade · S. Dharmik · U. Deshmukh · P. Dhote Department of Mechanical Engineering, Shri Ramdeobaba College of Engineering and Management, Nagpur 440013, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_46

391

392

Y. M. Sonkhaskar et al.

1 Introduction In today’s world, humans are mostly reliant on fossil fuels like coal as a primary energy source. These sources have large ash content, and pollution takes place to a greater extent during combustion. So, there is a need to replace these fuels with something that has the same calorific value and causes less pollution. As India is an agricultural country, it has a huge prospective in converting biomass into a renewable form of energy. But most of these loose biomass is directly burnt, which causes extensive air pollution. This can be reduced by using briquetting of biomass. Biomass briquette or pellets are obtained by pressing loose biomass into definite shape and size. The biomass densification leads to increased net calorific value per unit volume, reduced transportation cost, improved bulk density, and moreover handling characteristics [1]. Briquettes generally have a diameter range of 40–90 mm whereas pellets have a diameter range of 5–10 mm [2]. Various types of raw materials like sugarcane residue, sawdust, groundnut shells, maize and cotton stalks, rice husks, paper waste, wood chips, forest residues, etc., can be used for the briquetting process [3, 4]. The ash content, calorific value, moisture content, size of particle, and easy availability [5] are the key factors for the selection of raw materials to get quality briquettes/pellets. Starch, clay, molasses, bentonite, etc., can be used as binders [6]. Materials required for making briquettes and pellets are already available in abundance in rural areas, and this was one of the best solutions to the problem. Briquettes and pellets are not only low-cost fuel, but it can be a source of income, if the briquettes and pellets are sold in the market, thus creating wealth from waste.

2 Briquette Cum Pellet Making Machine The machine is designed to produce briquettes and pellets using a single setup manually by the process of extrusion, wherein the forward extrusion process will be used for getting the briquettes and backward extrusion process will be used for getting pellets. Various parts of the machine include Scotch yoke mechanism, piston, cylinder, and hopper. While designing machine, various mechanisms (like slider crank and toggle mechanism) were considered. The Scotch yoke mechanism best suits requirement for making machine, as this mechanism is easy to operate and linear motion of connecting rod and parts is easy to replace [7]. Due to the advantages of Scotch yoke mechanism, it was finally selected. Various parts like Scotch yoke, piston, cylinder, and hopper were designed according to the requirement [8–10]. Piston design includes six tapered holes for the production of pellets through reverse extrusion. After the finalization of design, fabrication work for the machine was done. Various parts were fabricated and assembled to produce the machine. The CREO model of the machine dimensions is shown in Fig. 1.

Development of Briquette Cum Pellet …

393

Fig. 1 CREO model of the machine

The working of this briquette cum pellet making machine is simple due to the Scotch yoke mechanism. The rotary motion of the Scotch is obtained with the movement of the handle. This rotary motion from Scotch is transmitted to yoke with the help of the movement of a pin attached to it. Reciprocating motion from yoke is transmitted to the piston through the connecting rod. When the raw material is added to the cylinder through the hopper, this reciprocating motion compresses the material. As piston consists of the drilled holes, it allows the formation of pellets through the reverse extrusion process. Briquettes are formed through the forward extrusion process.

3 Experimentation The photographs of different parts of the machine are shown in Fig. 2, and the developed machine is shown in Fig. 3. Experimentation was done to find out the density of the briquette and pellet obtained by this machine, by using various raw material and binder combination [11, 12]. Sawdust was used as raw material as sawdust is easily available, low-cost, and finely ground material. Another raw material used was aspen, which is also easily available. Binders used for experimentation purpose are bentonite and starch along with a certain mixture of water. Clay was used just for the experimentation purpose. In rural areas, rice husk along with cow dung as the binder can be used to produce briquettes and pellets (Fig. 4). During compression stroke, the material inside cylinder is compressed to 2:1 compression ratio. The excess material inside the cylinder after compression comes

394

Y. M. Sonkhaskar et al.

(a) Scotch Disc

(c) Pin Fig. 2 Different parts of the machine

Fig. 3 Actual photograph of the developed machine

(b) Yoke

(d) Piston with holes

Development of Briquette Cum Pellet …

395

Fig. 4 Process flow chart

through the holes drilled on the piston. Holes on the piston are tapered so that the material comes out and is also compressed producing compressed pellets.

4 Results and Discussion Briquette of dimensions as diameter 70 mm and length 50 mm and pellet of diameter 8 mm and length 50 mm were obtained. Experimentation was carried out with different raw material and binder combinations, and it was found that sawdust along with bentonite as a binder has more density for briquette as well as a pellet. The density of briquette with sawdust and bentonite as a binder was 956 kg/m3 , and the density of pellet with sawdust and bentonite as a binder was 679 kg/m3 . Production rate of biomass briquettes and pellets using this setup is 11 kg/h and 2.8 kg/h, respectively (Tables 1 and 2). Table 1 Density of Briquettes

Raw material

Binder

Density (kg/m3 )

Sawdust

Bentonite

956

Aspen chips

Bentonite

798

Sawdust

Starch

936

396 Table 2 Density of Pellets

Y. M. Sonkhaskar et al. Raw material

Binder

Density (kg/m3 )

Sawdust

Bentonite

679

Sawdust

Starch

663

5 Conclusion Analysis of raw materials, the tentative design of components and mechanisms was done. The extrusion process has been used for producing briquettes and pellets, wherein the forward extrusion process was used for producing the briquettes and the backward extrusion process was used for producing pellets. Experimentation includes finding out the density of the briquette and pellet obtained by this machine, by using various raw material and binder combinations. Briquettes and pellets were obtained of diameter 70 mm and 8 mm, respectively. The findings of experimentation state that sawdust along with bentonite as a binder has more density for briquette as well as pellet, among the various raw material and binder combinations used. This machine can help reduce farm waste into a useful product by also providing a source of livelihood.

References 1. Sengar SH, Mohod AG, Khandetod YP, Patil SS, Chendake AD (2012) Performance of Briquetting machine for briquette fuel. Int J Energy Eng 2(1):28–34. https://doi.org/10.5923/j.ijee. 20120201.05 2. Sonkhaskar YM, Deshpande VS, Modak JP (2013) Design and fabrication of dual powered mini pelletizer. Appl Mech Mater 465–466:242–247 3. Saptoadi H (2008) The best bio-briquette dimension and its particle size. AJEE 9(3 and 4):161– 175 4. Shukla S, Vyas S (2015) Study of biomass briquettes, factors affecting its performance and technologies based on briquettes. IOSR J Environ Sci Toxicol Food Technol (IOSR-JESTFT) 9(11):37–44 5. Oladeji JT (2015) Theoretical aspects of biomass briquetting. J Energy Technol Policy 5(3) 6. Križan P, Matúš M, Bábics J, Šooš L, Beniak J (2019) Relationship between raw material composition and pellets physical properties. In: TSME-international conference on mechanical engineering. https://doi.org/10.1088/1757-899X/501/1/012004 7. Shyamalee D, Amarasinghe ADUS, Senanayaka NS (2015) Evaluation of different binding materials in forming biomass briquettes with sawdust. Int J Sci Res 5(3) 8. Mani S, Sokhansanj S, Bi X, Turhollow A (2006) Economics of producing fuel pellets from biomass. Am Soc Agric Biol Eng 22(3):421–426, ISSN: 0883-8542 9. Bhattacharya SC, Augustus Leon M, Mizanur Rahman M (2000) A study on improved biomass briquetting. In: International conference on biomass-based fuels and cooking systems, Pune, India (2000) 10. Sonkhaskar YM, Saluja HS, Srivastava R, Parikh R, Singhai U (2018) Low cost manual briquette making machine using Scotch Yoke mechanism. Int J Innov Res Sci Eng Technol 7(7)

Development of Briquette Cum Pellet …

397

11. Mushiri T, Mugodo P, Mbohwa C (2017) Design of a sawdust pelleting machine. In: International conference on industrial engineering and operations management, Rabat, Morocco, pp 1763–1776 (2017) 12. Sonkhaskar YM, Deshpande VS, Modak JP (2015) Mathematical modeling of pellet making process using human powered flywheel motor. Ind Eng J 8(11):6–10. ISSN: 0970-2555

Towards the Development of Low-Cost Vacuum Setup for Customized Implant Manufacturing Sanjay Randiwe , Dheeraj Bhiogade , and Abhaykumar M. Kuthe

Abstract This study introduces a new vacuum casting setup used to cast customized implants. Setup is equipped with the induction furnace, vacuum pump, top vacuum chamber, thermocouple probe and bottom mold chamber along with Pirani gauge and vacuumized sealed doors and valves mounted. Experiments were conducted with aluminum alloy to investigate the accuracy of the setup. Implant fabrication is carried out with the aid of machining process which is difficult to machine, if the shape is intricate. Casting is one of the best routes to cast the complex geometry with the application of rapid prototyping technology. The proposed setup in this paper for the casting of aluminum under vacuum was designed and developed in-house, equipment opted and setup design after thermodynamic calculation for a vacuum chamber and minimizes required vacuum level for melting of Al. Melting is by induction furnace, metal is directly poured with the help of bottom pouring crucible mechanism mold prepared by zircon sand, and the pattern used for mold making is made on rapid prototype machine having acrylonitrile butadiene styrene (ABS) material. A promising new technique for direct melting and pouring of titanium alloy for medical implants was emerged from this study. This research demonstrated the feasibility of direct melting and casting of the aluminum alloy by vacuum technology. Keywords Customized · Vacuum · Induction furnace · ABS

S. Randiwe (B) Shri Ramdeobaba College of Engineering and Management, Nagpur 440013, India e-mail: [email protected] D. Bhiogade MSPGCL Co., Ltd., Nagpur, India A. M. Kuthe CAD-CAM Centre, VNIT, Nagpur, India © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_47

399

400

S. Randiwe et al.

1 Introduction Today, Ti6 Al4 V titanium is the most widely used material in the biomedical field due to its high mechanical properties, excellent corrosion resistance and excellent biocompatibility. However, titanium is characterized by high thermal reactivity, low thermal conductivity and reduced machinability [1]. It is the most suitable alloy used in biomedical implants in everyday life because of its lightweight as compared to that of stainless steel SS316L (CF3M). The cost of titanium and its alloy is high compared with existing materials such as aluminum alloys and steels. The cost includes the raw materials as well as the processing cost that requires a controlled environment due to the reactivity of titanium. Currently, research is being aimed to reduce the cost of primary titanium [2]. Prior to actual experimental setup, the various studies were carried out to investigate the properties of zircon sand mold which is used to pour the molten metal so that the process becomes very cost effective. The zircon sand is selected as a primary molding material because it is considered to be common foundry sand of the most refractory [3]. Zircon has a fusion point of 2700 K with good flowability, relatively high conductivity and better dimensional stability than graphite. In addition, zircon sand molds have been used for many years in ferrous and nonferrous foundries with good results for heavy walled premium castings. Malone et al. suggested that molten titanium reacts violently with zircon, but later studies by Lang et al. [4] investigated that zircon rammed molds using sodium silicate as binder withstood the attack of molten titanium. But casting of thickness of 12 mm or more produced in the above molds suffered from pinholes and poor details. Koch et al. [5] benefited using sodium silicate in lower quantities and reported very good results in castings up to 100 mm cube size. He also tried a number of mold washes and found that zirconia wash with water or isopropyl alcohol base is very effective for inhibiting metal-mold reactions. Bentonite has been tried as a binder since it produces excellent grain and good dry strength in rammed zircon molds. This combination indicated good results in castings of 150 mm cube size. In the recent study at DMRL, zircon sand was used with CO2 molding method for casting titanium. Here, sodium silicate is used as binder and hardened by CO2 gas which releases colloidal silica as given in the simplified reaction. Among various melting methods, vacuum induction melting (VIM) and vacuum arc remelting (VAR) are widely used for commercial production of NiTi alloys. In VIM, the electrodynamic forces result in the first stirring of the melt and thereby ensuring greater chemical and microstructural homogeneity in the alloy. The main disadvantage of the VIM is carbon contamination from the crucible. VAR uses the consumable electrodes to melt the metal and is carried out under water-cooled copper crucible. Many researchers have focused on machining titanium alloy and achieved the optimal results. The present experimental study focused on the melting of titanium alloy in a vacuum environment with low-cost vacuum casting setup.

Towards the Development of Low-Cost Vacuum …

401

Much effort has been expended in the modification of induction melting furnace for use with titanium. This system uses many advantages, including single-stage melting, uniformity of temperature and composition within the melt and the ability to melt scrap, ingot or sponge. Many researchers have focused on the melting of the titanium alloy in vacuum environment only, and there are many research gaps in the titanium alloy melting. The main aim of the research is to develop a cost-effective and miniature vacuum setup to melt titanium alloy to make the intricate shape such as customized prosthesis for patients in India.

2 Experimental Method 2.1 Experimental Material Ti6 Al4 V alloy is most commonly used in aircraft and biomedical industries. It covers 50% of the whole production of titanium alloys. It is also the most explored and tested titanium alloy with balanced properties such as low density, ductility, good corrosion and oxidation resistance. Standard composition of the ASTM F136 Grade 5 titanium is given in Table 1, and the chemical analysis of the titanium alloy (ASTM F136) along with composition is given in Table 2. Figure 1 shows SEM image of microstructure of work material Ti6 Al4 V. Table 1 Standard composition of the ASTM F136 Grade 5 titanium Standard ASTM F136 Grade 5 Aluminum: 5.5–6.75%

Nitrogen (maximum): 0.05%

Vanadium: 3.5–4.5%

Hydrogen (maximum): 0.0125%

Iron (maximum): 0.40%

Other elements (maximum, each): 0.10%

Oxygen (maximum): 0.20%

Other elements (total): 0.40%

Carbon (maximum): 0.08%

Titanium: balance

Table 2 Tested chemical analysis of the titanium alloy (ASTM F136) S. no.

Sample identity

C%

Al%

V%

Fe%

Ti%

1

Rod (Ti alloy)

0.072

5.77

3.89

0.21

89.46

402

S. Randiwe et al.

Fig. 1 Microstructure of Ti6 Al4 V (light area represents α phase rich in aluminum, and the dark area represents β phase rich in vanadium)

2.2 Experimentation 2.2.1

Common Features of Vacuum Casting Equipment

Some standard features of vacuum equipment. 1. Most vacuum casting equipment is built for mass production of small components. 2. Vacuum casting equipment is semi-automatic with many crucial process steps being automated. 3. The vacuum casting process is often carried out with the lost wax process and the use of permanent molds [6]. 4. An induction furnace with argon gas shielding is used for melting.

2.2.2

Modifications Introduced for This Work

The alloy to undergo vacuum remelting is formed into cylinder typically by vacuum induction melting. This cylinder, refined to as an electrode, is then put into large cylindrical enclosed crucible and brought to a metallurgical vacuum (0.001 mm Hg or 0.1–13.3 Pa). At the bottom of the crucible, there is a small amount of alloy to be remelted which the top electrode is brought close prior to starting the melt. The modification done in our system was to pull the molten metal from the bottom of the crucible to fill the mold cavity which melted under induction heating.

Towards the Development of Low-Cost Vacuum …

403

1. The vacuum casting equipment was designed for small-scale production in factories and research purposes. The furnace thus has a capacity of 5 kg of titanium. 2. There was no reason for making the complicated automation process to the furnace due to its intended usage. Hence, the critical process steps like placing the investment mold made of zircon powder below direct the crucible by positioning it manually. 3. Induction furnace was used instead of resistance furnace.

2.2.3

Material Selection for Vacuum Chamber

The selection of the refractory and insulating bricks was made based on the service requirements of the furnace vacuum chamber. These service requirements were identified as follows: 1. Temperature: A maximum furnace temperature of 2000 °C (3632 °F) was put into consideration. Since titanium melts at 1668 °C (3034 °F), this temperature would allow for the superheating of the charge to be more easily accomplished and also makes the furnace more durable. 2. Size: The furnace volume should be enough to melt a maximum 5 kg of titanium necessary for the casting of small titanium components. The refractory used to be the high alumina R/M or LCC, Al2 O3 % = 90–92, Fe2 O3 % = 0.5 (max), ST = 1100 °C, AT = 1750 °C, Grading = 0–6 mm. 3. Top chamber with door assembly: The material used for the construction of the vacuum chamber was stainless steel. All the mechanical design aspect was taken into account for the development of the top assembly to avoid failure of the chamber under vacuum. The top chamber is mainly used for the charging of the material and stirring of the metal, and eyepiece was provided to observe the condition of the molten metal. The chamber was lined entirely with the refractory to avoid the heat loss. 4. Heating chamber: The central part of the chamber consists of the bottom pouring crucible surrounded by the induction coil. Two electrical connections were directly attached to the middle section where the induction coil was wounded to the crucible. 5. Bottom part of the chamber: Bottom portion of the chamber was designed to serve the purpose of casting. The molds were placed below the bottom pouring crucible for directly pouring the molten metal. 6. Bottom pouring crucible: The specially designed graphite crucibles with zirconium coating were used for the pouring of the molten Ti. This technique helps to avoid complicated tilt-pouring system.

404

S. Randiwe et al.

Fig. 2 Actual experimental setup

2.3 The Vacuum System and Accessories The system consists of: 1. Vacuum pump: The vacuum pump used was having a displacement 50 Hz and 600 LPM (liter per minute capacity). The ultimate vacuum created by the pump was 1 × 10−3 Torr. The pump is primarily an oil-filled positive-displacement rotary vane pump. 2. Pirani gauge: The digital Pirani gauge with display range 0.001–1000 mbar was used to the measure the vacuum level created inside the vacuum chamber. 3. Valve: If the power supply fails or stops, an anti-suck-back device prevents air or oil in the pump to be drawn back into the chamber which is evacuated. When the pump is running, the oil pump-pin pumps oil to the anti-suck-back valve to lift to the predetermined height and allow the oil to pass-through outlet valve of the anti-suck-back device. This process will continue as long as the pump runs. When the pumps stop, the inlet valve of anti-suck terminal devices closes, as this valve is fitted with pre-loaded spring and thus prevent oil to the pump. The vacuum pump, valves and the Pirani gauge were well connected to the vacuum casting system to achieve the higher functionality. The complete setup is shown in Fig. 2.

3 Results and Discussion 3.1 Experimental Casting The material was charged into a crucible at room temperature; the furnace was then closed, oxygen was evacuated in 10−3 Torr. As titanium is very reactive material, it is essential to degas the chamber. Melting took place in a vacuum. The melting temperature reading was noted using immersion-type thermocouple. For Ti6 Al4 V, the casting was poured at temperature set in between 1665 and 1680 °C. The rectangular

Towards the Development of Low-Cost Vacuum …

405

Fig. 3 Rectangular bar of Ti6 Al4 V

bar was cast as shown in Fig. 3 having 100 mm length, 20 mm depth and 20 mm width. After shake out, it observed good surface finish with no significant defect.

3.2 Simulation The casting simulation of rectangular bar 100 × 20 × 20 mm3 was simulated in AutoCAST software, properties assigned to cast Ti were fed into the database, which was obtained from thermophysical properties calculations, which are density 4430 kg/m3 , liquid density 4230 kg/m3 , thermal conductivity 6.70 W/m K, specific heat 526 J/kg K and latent heat 91.8 kJ/kg [7]. Also, for zircon powder mold, properties are derived from the experimental results and software database, which are density 1770 kg/m3 , thermal conductivity 0.98 W/m K, specific heat 735 J/kg K and compressive strength 20 kPa. Moreover, a constant value of 1000 W/m2 K was used for the interface heat transfer coefficient. Total filling time is 2.21 s. The average pouring rate was 0.98 kg/s.

4 Conclusion This article discusses about the aluminum casting for which the low-cost miniature setup is developed with a focus on mechanical properties of aluminum alloy. The experimental trials were carried out to as-cast implant (hip joint) with the mechanical properties of tensile strength 350 Mpa, yield strength 270 Mpa and Young’s modulus of 110 Mpa. Figure 4 shows the hip joint. The vacuum casting equipment constructed was able to replicate the advantages of inclusion-free casting and improved in castings properties produced via vacuum

406

S. Randiwe et al.

Fig. 4 Hip joint

casting process. Earlier, the vacuum casting equipment was constructed with costing of US$18,000. The model can be easily scaled up for large-sized casting by increasing furnace capacity and the size of the molding chamber.

References 1. Ezugwu EO, Wang ZM (1997) Titanium alloy and their machinability—a review. J Mater Process 68(3):262–274 2. Suzuki Y (1998) Fabrication of shape memory alloys. In: Otsuka K, Wayman CM (eds) Shape memory materials. Cambridge University Press, Cambridge, pp 133–148 3. Randiwe SM, Bhiogade DS, Kuthe AM (2017) Investigation & critical analysis of titanium casting in zircon sand mold for customized implant manufacturing. In: Proceedings of 5th International Conference on Powder metallurgy and Advanced Materials (RoPM-AM-2017), Technical University, Cluj-Napoca, Romania 4. Lang RM et al (1954) Development of processes for making Ti casting and other ingots. No. DRDTB-I-12046, Battelle Memorial Institute, Columbus 5. Koch RK, Hoffman JL et al (1977) Casting titanium in zircon sand moulds. IS Bureau of Mines, Report RI-8203, p 44 6. Aremo B, Adeoye MO (2010) A low-cost vacuum equipment for aluminium alloys. Russ J Non-ferrous Metals 51(2) 7. Bhiogade DS, Randiwe SM, Kuthe AM (2017) Study of Hot tearing in stainless steel CF3M during casting using simulation and experimental method. Int J Metal Cast. https://doi.org/10. 1007/s40962-017-0170-7

Simulation Study on Effect of Variable Curvature on the Modal Properties of Curved Cantilever Beams Aqleem Siddiqui , Girish Dalvi , Akshay Patil , and Surabhi Chavan

Abstract Curved beams are widely used in many engineering fields due to the highstrength capacity compared to their straight forms. It is essential that curved beams be analyzed for their modal properties as structures should have the natural frequency away from that which occurs in their working condition to avoid resonance. In the last few decades, many researchers had carried out investigation on several parameters of the curved beam. The present research work proposes a generalized mathematical equation for predicting mode shapes of curved cantilever beams, which take into account the change in length and curvature of the beam. Modeling of curved beam subjected to cantilever boundary condition is done by using ANSYS software. Twelve cases of cantilever beams, each having different curvature and length, were created. The mode shapes of the curved beams were obtained using the normalization method for each mode shape. The effect of the variation of the curvature on the mode shapes and natural frequency was analyzed for the first six transverse modes. It is found that the frequency of vibration and amplitude of vibration increase as the radius of curvature of the beam increases. Curve fitting equations for these mode shapes were obtained using MATLAB software. The generalized equation obtained for the mode shapes of the curved cantilever beams generates the mode shapes which are in good agreement with those obtained from ANSYS. Keywords Modal analysis · Curved beams · Variable curvature

1 Introduction Some of the structures such as arches, arch bridges and long-span roof structure use curved beam elements due to their enhanced strength. The modal analysis of these curved beams plays an important role in analyzing the modal properties which can be used to understand the vibration behavior of the structure. Various researchers have studied the modal properties of curved beams. A. Siddiqui (B) · G. Dalvi · A. Patil · S. Chavan Fr. C. Rodrigues Institute of Technology, Vashi, Navi Mumbai, Maharashtra 400703, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_48

407

408

A. Siddiqui et al.

Lee et al. [1] developed the governing equation for the variable curvature arch. Numerical method is used to calculate the natural frequency and mode shapes of the parabolic, sinusoidal, and elliptic arches. Guierrez et al. [2] obtained the lowest frequency coefficients of symmetrical and unsymmetrical arch-type structures, using geometric parameters coordinate functions and the Ritz method. Kang et al. [3] developed the governing equations of motion for in-plane and out-of-plane vibration of circular arches considering shear deformation using differential quadrature method. Oh et al. [4] derived the governing differential equations for free in-plane vibrations of non-circular arches, including the effects of rotatory inertia and shear deformation. Raveendranath et al. [5] investigated the performance of a curved beam finite element with coupled polynomial distributions for normal displacement and tangential displacement for in-plane flexural vibration of arches. Yang et al. [6] developed the governing differential equations for the free in-plane vibration of curved beams using finite element method. From the literature review, it is found that curvature plays an important role in the modal properties of curved beams. Hence, it is necessary that the effect of curvature on the modal properties be evaluated. The present research work proposes a generalized mathematical equation for predicting mode shapes of curved cantilever beams, which take into account the change in length and curvature of the beam. Modeling of curved beam subjected to cantilever boundary condition is done using ANSYS software. Twelve cases of cantilever beams, each having different curvature and length, are considered.

2 Methodology The methodology for carrying out the research work is shown in Fig. 1.

Fig. 1 Methodology of proposed study

Simulation Study on Effect of Variable Curvature … Table 1 Different cases for study

409

Case I: constant radius, R = 478 mm Length, L

500 mm

750 mm

1000 mm

Case II: constant length, L = 500 mm Radius, R

318 mm,  = 90

286 mm,  = 100

260 mm,  = 110

Case III: constant length, L = 750 mm Radius, R

478 mm,  = 90

428 mm,  = 100

390 mm,  = 110

Case IV: constant length, L = 1000 mm Radius, R

637 mm,  = 90

573 mm,  = 100

521 mm,  = 110

Simulation study was carried out for the various cases of the curved beams, viz. Case I: Constant radius and varying length. Cases II, III, IV: Constant length and varying radius. These cases are shown in Table 1.

3 Simulation of Curved Beams Beam with dimensions 500 mm × 25 mm × 2 mm and radius 478 mm as shown in Fig. 2 was modeled using SOLID45 element in the ANSYS having cantilever boundary condition. The beam is meshed using 5-mm element length, and modal analysis is performed for the extraction of the first six transverse modes for displacement data from ANSYS and plotted in MATLAB as shown in Fig. 3. Fitness function was obtained for these modes using MATLAB, and these equations are polynomials having an order of mode_no + 3. The fitness function for the first six modes of 500 mm length beam is shown below.

Fig. 2 Beam geometry

410

A. Siddiqui et al.

Fig. 3 Modes for case IA

Mode 1 y = −0.0028x 4 − 0.8323x 3 + 1.8784x 2 − 0.0459x + 0.0013 Mode 2 y = 2.4515x 5 − 19.091x 4 + 33.335x 3 − 16.237x 2 + 0.5422x − 0.0115 Mode 3 y = 14.568x 6 − 141.67x 5 + 298.25x 4 − 229.25x 3 + 61.611x 2 − 2.5816x + 0.0413 Mode 4 y = 70.443524x 7 − 921.481243x 6 + 2374.906554x 5 − 2457.268133x 4 + 1132.914128x 3 − 207.222586x 2 + 8.708531x − 0.106581 Mode 5 y = 251.680310x 8 − 4385.064920x 7 + 13,487.062493x 6 − 17,465.066738x 5 + 11,192.265965x 4 − 3546.379710x 3 + 484.654285x 2 − 18.501801x + 0.175822

Simulation Study on Effect of Variable Curvature …

411

Fig. 4 Circular arc geometry

Mode 6 y = 932.472836x 9 − 21,429.260349x 8 + 76,880.476560x 7 − 119,404.463304x 6 + 97,350.557470x 5 − 43,504.651285x 4 + 10,237.820219x 3 − 1098.571839x 2 + 36.621963x − 0.273652 Further, modal analysis of beam of constant radius 478 mm and length 750 and 1000 mm was performed and fitness functions of the mode shapes of each were obtained. Equation to obtain the general curve is obtained using geometrical proof considering the arc T1CT2 of radius R and chord length L as shown in Fig. 4 and to obtain the equation considering any vertical distance Ox at distance x. Ox = OE − DO In ODT1, DO =



R 2 − (L/2)2

In OE E, OE = Ox =

 

R2 − x 2 R2 − x 2 −



R 2 − (L/2)2

(1)

The curve represented using Eq. (1) starts from the negative half of chord length; therefore in order to initiate the curve from zero, shifting operation was performed on the equation after which Eq. (1) becomes

412

A. Siddiqui et al.

Fig. 5 Comparison of original curve and curve generated using Eq. (2)

Ox =



R 2 − [x − (L/2)]2 −



R 2 − (L/2)2

(2)

Figure 5 shows the comparison of actual curve plotted from ANSYS data points and curve plotted using Eq. (2) which is seen to be in good agreement with each other. Similar procedure is followed for the case two of constant length and varying radius, and fitness functions for the mode shapes were obtained and observed to be similar. Thus, the polynomial equations were generalized for the mode shapes for any length and radius expressed as: y=



R2 − x 2 −



R 2 − (L/2)2 + F

m 

Ci X i

(3)

i=0

where R Radius of curvature L Chord length C Coefficient of equation

x L    θ R F = × 1 − cos 5 2 m = mode_no + 3 X=

The values of C i for first six modes are shown in Table 2. The mode shape plot for the first six modes obtained from the generalized mode shape Eq. (3) is shown in Fig. 6. These plots from the generalized mode shape equation are seen to be similar in nature as that of the mode shape obtained from the

Simulation Study on Effect of Variable Curvature …

413

Table 2 Values of C i for first six modes Mode

1

2

3

4

5

6

C0

0.003

−0.0211

0.048

−0.13064

0.222163

−0.33876

C1

−0.0939

0.9289

−3.2768

9.983031

−21.5728

41.21816

C2

1.9322

−19.14

69.159

−215.015

525.9834

−1167.58

C3

−0.5592

38.474

−257.3

1171.207

−3829.52

10,814.94

C4

−0.285

−21.366

337.8

−2568.61

12,175.17

−46,173.3

C5

–

2.105

−162.03

2538.184

−19,278.9

104,359.9

C6

–

–

16.546

−1034.18

15,290.54

−13,0136

C7

–

–

–

99.26852

−5290.29

86351.91

C8

–

–

–

–

429.0183

−25,883.1

C9

–

–

–

–

–

1792.927

Fig. 6 First six modes from generalized equation

ANSYS for the beam of radius 478 mm and length 500 mm. Similarly, mode shape plot for the lengths 750 and 1000 mm can be tested.

4 Conclusion The generalized equation obtained for the mode shapes of the curved cantilever beams generates the mode shapes which are in good agreement with those obtained from ANSYS. Hence, generalized equation obtained using the current methodology can be used for generation the mode shapes of a curved cantilever beam.

414

A. Siddiqui et al.

References 1. Lee BK, Wilson JF (1989) Free vibrations of arches with variable curvature. J Sound Vib 136(l):75–89 2. Gutierrez RH, Laura PAA, Rossi RE, Bertero R, Villaggi A (1989) In-plane vibrations of noncircular arcs of non-uniform cross-section. J Sound Vib 129(2):181–200 3. Kang K, Bert CE, Striz AG (1995) Vibration analysis of shear deformable circular arches by the differential quadrature method. J Sound Vib 353–360 4. Oh SJ, Lee BK, Lee IW (1999) Natural frequencies of non-circular arches with rotatory inertia and shear deformation. J Sound Vib 219(1):23–33 5. Raveendranath P, Singh G, Pradhan B (2000) Free vibration of arches using a curved beam element based on a coupled polynomial displacement field. Comput Struct 78:583–590 6. Yang F, Sedaghati R, Esmailzadeh E (2008) Free in-plane vibration of general curved beams using finite element method. J Sound Vib 318:850–867

Variation in the Properties of Spot Weldments of Cold Rolled Mild Steel Welded with Filler Metal by Annealing Treatment Sushil T. Ambadkar

and Deepak V. Bhope

Abstract Filler metal addition has been verified as an effective way to refine the mechanical behaviour of cold rolled mild steel in resistance spot welding. Negligible quantity of filler metal if added to the spot weld is found to improve mechanical properties of spot weldments, if no variation in the composition of base metal and filler metal is allowed. Looking at practical applications, the sensitivity of the resistance spot welding process with filler metal to variation in annealing treatment was experimentally investigated. Filler metal quantity was from 30 to 70 mg and annealing heat treatment was kept between 30 and 70%. The material of the filler metal is the same as that of base metal and was added at the centre of overlap in lap joint. The experimentation was carried out by spot welding of specimen varying filler metal from 30 to 70 mg maintaining annealing cycle constant at 30%. The experimentation was then repeated for different annealing cycles. For 30–70% of anneal cycle, 30 mg quantity of added filler metal yielded maximum breaking point and plasticity ensuring optimum load-bearing and energy absorption capability under the button pullout mode. Keywords Spot welding · Filler metal · Failure mode · Anneal · Cold rolled mild steel

1 Introduction For selecting materials for cars, car designers today seek materials with best stiffness, mass reduction, safety performance, mass saving, formability, weldability, corrosion resistance, fatigue resistance, cost, and environmental factors. Spot welding is the S. T. Ambadkar (B) Department of Mechanical Engineering, Government College of Engineering, Chandrapur, Maharashtra, India e-mail: [email protected] D. V. Bhope Department of Mechanical Engineering, Rajiv Gandhi College of Engineering, Research and Technology, Chandrapur, Maharashtra, India © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_49

415

416

S. T. Ambadkar and D. V. Bhope

preferential method of fabricating automotive structure components because of its less cost and maximum weld speed. A vehicle’s structural behaviour depends in part on the fabricated-joint structural uniformity. Mild steel is a good material for car body as it offers reduction in weight, renewed crash performance, good mechanical properties and good manufacturability. Different techniques/methods are being verified to improve spot weld mechanical performance by improving microstructure and phases of the weldments, by changing welding techniques, welding of different metals as base metals and innovative welding designs. Spot welding of dissimilar base metals and resulting changes in microstructure and subsequent mechanical performance is being verified. Effects of the filler metal and its quantity on failure mode are discussed on 1.5 mm thick CRM steel [1]. Newer material like advanced high strength steels (AHSS), offers the potential for improvement in vehicle crash performance without extra weight increase. Currently, two types of advanced high strength steels are being used in the automotive industry. One is the dual-phase (DP) steel in which mechanical properties are controlled by the martensite volume fraction and the ferrite grain size [2, 3]. Dual-phase steels and transformation-induced plasticity (TRIP) steels are finding increased use in automotive bodies due to a combination of high strength and high ductility and its weldability. Also, it has been stated that spot welding of AHSS steels has issues of weldability problems due to the relatively higher alloying level in these steels [4]. Resistance spot welding of TRIP steel yields inconsistent interfacial failure or partial interfacial failure modes indicating inferior mechanical performance [5]. Weld behaviour of mild steel has been well explored, the same cannot be said about advanced high strength steels. It has been proven that strength and crash performance of traditional mild steel in spot welding can be enhanced with filler metal [1]. The performance of resistance spot welds, in general, can be enhanced by altering the microstructure by applying a post-weld heat treatment. Thus, local post-weld heat treatment called process annealing was introduced in weld cycle. Annealing is used to induce ductility, relieve internal stresses, refine structure and is ideal for improving mechanical properties. One of the efforts is investigated by varying filler metal to be added to spot weldments, varying annealing energy levels and verifying its response to mechanical testing.

2 Experimental Procedure The material, i.e. cold rolled mild steel sheets used in this study is specifically for automotive body applications and selected sheet had 1.25 mm thickness. The chemical composition of this material is shown in Table 1. Tensile testing of welded specimen was carried out to obtain tensile stress–strain curves of the CR MS sheets. A Resistance spot welder, K J Thermoweld make with specifications as mentioned in Table 2 was used to spot weld the specimen (Fig. 1).

Variation in the Properties of Spot Weldments …

417

Table 1 Composition of CR mild steel Sample identity

C%

Si%

Mn%

P%

S%

Cr%

Mo%

Ni%

Cold rolled mild steel strips

0.047

0.011

0.25

0.0060

0.0181

0.036

0.0070

0.0286

Table 2 Spot welder specification [1] Machine model K J Thermoweld

Unit

TSP 30

Rating KVA @ 50% duty cycle

KVA

30

Max. available current (short circuit) @ throat depth: 460 mm

K Amp

15

mm

0.3 + 0.3

Optimum weldability{@ 460 throat depth} Minimum Maximum

mm

1.5 + 1.5

Machine dimensions (approx. L × W )

mm

1020 × 520

Height

1500

Net wt. (for 460 mm throat) (+30 kg for 200 mm throat (approx.)

kg

450

Shipping wt. (for 460 throat) (+30 kg for 200 throat) (approx.)

kg

525

Fig. 1 Resistance spot welding machine

418

S. T. Ambadkar and D. V. Bhope

The cycle in the spot welder can be controlled with the help of parameters as mentioned in Table 3. The machine was microcontroller based and energy levels were programmable. The dimensions of the spot weld specimen were 70 mm by 25 mm. The thickness of CRMS sheet as mentioned was 1.25 mm thick. The electrode material was made C15000 (copper alloy) and it had a face diameter of 5 mm. The coolant, i.e. water flow-rate in the electrode was at 4 l/min (Fig. 2). The spot welding schedules and parameters for experimentation are shown in Table 3. The purpose of experimentation was to determine the effect of annealing on the addition of filler metal in relation to mechanical behaviour. The composition of filler metal was the same as that of base metal and it was added at the centre of overlap. The extent of filler metal added in experimentation was from 30 to 70 mg for each annealing energy level from 30 to 70%. The schedule followed is mentioned in Table 3 and was maintained unchanged with addition of filler metal and varying annealing Table 3 Spot welder parameters and setting variables Function of soft button as indicated

Red indication

Description

Setting parameter

Squeeze time

SQZ

0–99 cycle

4

Preheat time

W1

0–99 cycle

0

Energy level

W1

0–99% programmable

0

Cool (I) time

C1

0–99 cycle

34

Slope for W2

SLP

0–99 cycle

2

Weld time

W2

0–99 cycle

37

Weld energy (% heat setting)

W2

0–99% programmable

20

Cool (II) time

C2

0–99 cycle

28

Anneal time

W3

0–99 cycle

10

Energy level

W3

0–99% programmable

30–70%

Hold time

HLD

0–99 cycle

5

Off time

OFF

0–99 cycle

5

Fig. 2 Test coupon

Variation in the Properties of Spot Weldments …

419

Fig. 3 Spot welded and tested specimen

energy level. Specimens were tested in tensile tests and load–displacement curve was plotted to obtain breaking capacity. The failure mode was also observed and was interpreted by visual inspection. Some of the spot weld specimens were sectioned through the cross-section perpendicular to the length direction and mounted for macrostructure and microstructure observations following standard metallographic procedures (Fig. 3).

3 Result and Discussion Investigations are carried out by adding a small quantity of filler metal from 30 to 70 mg in spot weldments. The annealing energy level is varied from 30 to 70%. Changes in is strength, plasticity is then determined using load–displacement curve (Figs. 4 and 5). Interfacial fracture and nugget pullout are two distinct failure modes observed during tensile testing (Fig. 6). Table 4 shows, for 30–70% of annealing energy levels, filler metal with 30 mg quantity yield maximum breaking strength and plasticity. Based on this, it can be concluded that filler metal with 30 mg quantity produces

Fig. 4 Variation of displacement at max load (mm) and average breaking strength (kN) on Y-axes with filler metal quantity (mg) on X-axes at 30–70% annealing energy level

420

S. T. Ambadkar and D. V. Bhope

Weld Metal

Heat Affected Zone (HAZ)

Fig. 5 Specimen No. 135, weld metal and HAZ (filler metal 30 Mg, anneal 60%) weld metal showing Widmanstatten ferrite along with normal ferrite (whitish) and pearlite (blackish), HAZ showing tempered martensite (black) and ferrite

Fig. 6 Failure modes in spot weld specimens: button pull out and interfacial

optimum tensile performance, at all annealing levels. Most desirable button pull out mode of failure is found in the 30 and 40 mg of filler metal with gradual transformation to interfacial mode. This eventually is true at all annealing levels. Load carrying capacity and energy absorption capability were found to be more for weldments welded with 30 mg filler metal failing under the button pullout mode. Maximum nugget diameter of 4.71 mm is obtained at 30% annealing corresponding to addition of 30 mg filler metal but its further addition from 40 to 70 mg leads to a gradual reduction in nugget diameter. At all annealing energy levels, 30 mg filler metal yielded maximum nugget diameter with uniform reduction. Not much variation is visible in nugget diameter. For all filler metal and annealing combinations, the properties seem to be governed by recrystallization annealing at 30% annealing and full annealing at 40–70% annealing level.

4 Effect of Annealing on Phases in Microstructure of Weldments with Filler Metal In the present study, filler metal from 30 to 70 mg is added with annealing variation from 30 to 70%. One way of improving the performance of resistance spot welds is to change the microstructure by selecting a suitable post-weld heat treatment. A localized post-weld heat treatment can be accomplished by inducing in process annealing [6]. Annealing relieves internal stresses, refines structure and induce ductility. The joint region for 30% annealing consisted of three separate zones: (i) fusion

Variation in the Properties of Spot Weldments …

421

Table 4 Results obtained with filler metal addition and annealing Specimen No.

Filler material by weight (mg)

Breaking load (kN)

Annealing variation (%)

Type of failure

Displacement at max load (mm)

Nugget diameter (mm)

135

30

10.6

30

Button pull out

3.7

4.71

136

40

10.46

Button pull out

3.6

4.43

137

50

10.1

Button and interfacial

3.55

4.17

138

60

9.8

Button pull out

3.6

3.90

139

70

9.5

Button pull out

3.56

3.91

145

30

10.5

Button pull out

3.8

4.51

146

40

10.4

Button pull out

3.7

4.39

147

50

10.2

Button and interfacial

3.6

4.2

148

60

9.78

interfacial

3.7

3.91

149

70

9.49

interfacial

3.6

4.01

155

30

10.55

Button pull out

3.85

4.52

156

40

10.42

Button pull out

3.72

4.39

157

50

10.2

Button and interfacial

3.7

4.19

158

60

9.7

interfacial

3.8

3.78

159

70

9.5

interfacial

3.7

4.06

165

30

10.61

Button pull out

3.87

4.76

166

40

10.48

Button pull out

3.8

4.61

167

50

10.3

Button and interfacial

3.79

4.2

168

60

interfacial

3.7

3.98

169

70

interfacial

3.69

3.95

175

30

Button pull out

3.86

4.71

40

50

60

9.68 9.42 10.6

70

(continued)

422

S. T. Ambadkar and D. V. Bhope

Table 4 (continued) Specimen No.

Filler material by weight (mg)

Breaking load (kN)

176

40

177

50

178

60

179

70

Annealing variation (%)

Type of failure

Displacement at max load (mm)

Nugget diameter (mm)

10.44

Button pull out

3.8

4.59

10.3

Button and interfacial

3.76

4.01

9.68

interfacial

3.72

3.8

9.52

interfacial

3.7

3.91

area also called weld nugget; adjacent to it is (ii) heat-affected zone and at the end (iii) base metal. The material in the heat-affected zone must have faced maximum temperature and must have cooled by cooling rate varying inversely with respect to its distance from the fusion zone. The heat-affected zone phases near the fusion line consist of martensite, ferrite. The treatment seems to be low-temperature tempering which is relieving internal stresses thus improving toughness. The strength obtained is found to be maximum for 30 mg filler metal with gradual reduction with respect to filler metal. The fusion zone boundary phases of specimens produced tempered martensite profile along with a significant fraction of ferrite. Tempering of martensite in the specimen is synonymous with tempering of steels as mentioned by Chuko and Gould [6]. The distant microstructure from fusion zone consisted of ferrite and pearlite phase. With an increase in filler metal and annealing temperature, a gradual drop in strength and plasticity is observed. Any martensite formed after the welding might have retransformed to austenite during high-temperature annealing with the formation of new grains (recrystallization). This can be attributed to increased carbon percentage due to the addition of filler metal locally, reducing critical transformation range. Fast cooling rates were observed after re-austenitization and, again, predominantly martensite was reformed which resulted in reduction in plasticity and strength as confirmed by Baltazar Hernandez et al. [7].

5 Effect of Annealing with Addition of Filler Metal on Mechanical Properties of Resistance Spot Weldments Interfacial fracture and nugget pullout are observed failure modes as shown in Fig. 6. At 30–70% of annealing, addition of 30 mg of filler metal lead to maximum breaking strength and elongation. Hence, it is concluded that the optimum tensile performance, based on the peak load and energy absorbed at failure, is achieved with 30 mg filler metal. Pull out mode of failure is found in the 30 and 40 mg of filler metal with smooth and gradual transformation to interfacial mode. Pullout failure in the specimen can

Variation in the Properties of Spot Weldments …

423

be attributed to low-temperature tempering and subsequent superior properties of boundary martensite due to stress relieving and grain refinement on fusion zone and HAZ irrespective of the addition of filler metal. The strength of weldment at 30 mg filler metal at all annealing levels is found to be the same. 30 mg of filler metal improves load carrying capacity and energy absorption capability in weldments and hence fails under the button pullout mode [7].

6 Effect of Annealing with Addition of Filler Metal on Nugget Diameter of Resistance Spot Weldments As confirmed in first experimentation, the addition of 30 mg of filler metal gives nugget diameter of 4.71 mm, which is followed by reduction in nugget diameter for 40–70 mg which may be due to less heat generation and reduced contact resistance reducing the heat/temperature generated at the interface at lower current. The same result is observed for 30–70% annealing levels in cycle, at 30 mg of filler metal and all annealing levels. The columnar grain growth seems transformed to refined grain morphology during low-temperature annealing (tempering) which is confirmed in the fusion zone and HAZ microstructure shown by Fig. 5. The above results indicate that the grain morphology developed during the spot welding is transformed into new grains upon annealing, suggesting grain recrystallization and tempering of martensite. The process appears to be recrystallization annealing [8]. The addition of filler metal appears to have contributed to increasing nugget size and annealing temperature. Not much variation is visible in nugget diameter, but it is not lesser than nugget diameter corresponding to 30 mg of filler metal. For all filler metal and annealing combinations, the properties seem to be governed by full annealing process, due to which lamellar pearlite and ferrite were visible in fusion zone. The results indicate that at 40–70% of annealing, the grain morphology developed during weld cycle, transformed into new grains upon annealing [6].

7 Summary and Conclusion The microstructure and the mechanical behaviour of spot-welded cold rolled mild steel was successfully modified by adding filler metal and utilizing annealing heat treatment cycle in resistance spot welding. The main results in this study are listed as follows: 1. From experimentation, it is found that maximum breaking strength is obtained with 30 mg filler metal for all annealing levels from 30 to 70%. Gradual drop in breaking strength is visible with a further addition of filler metal. It is concluded

424

S. T. Ambadkar and D. V. Bhope

that due to a localized increase in carbon percentage with an increase in filler metal, critical temperatures are reduced promoting recrystallization. 2. Even though nugget diameter is found to be varying with filler metal as observed earlier, the properties seem to be governed by annealing process. It is concluded that annealing does not influence nugget diameter as observed, but it does influence phases existing in weld metal and HAZ which is reflected in change in properties. 3. 30% mg of filler metal yield good combination of breaking strength and plasticity. 4. Predominant failure mode with annealing is button pullout indicating superior mechanical properties.

References 1. Ambadkar ST, Bhope DV (2018) Resistance spot welding of cold rolled mild steel with filler metal. In: Advanced manufacturing and material science. Springer, Cham, pp 63–73 2. Rathbun RW, Matlock DK, Speer JG (2003) Fatigue behavior of spot welded high-strength sheet steels. Welding J 82(8):207-s–218-s 3. Matlock DK, Krauss G, Ziaebrahimi F (1984) Strain hardening of dual phase steels: an evaluation of the importance of processing history. In: Krauss G (ed) Deformation, processing, and structure. ASM International, Materials Park, pp 47–87 4. Smith S, den Uijl N (2006) Resistance spot welding of advanced high strength steels for the automotive industry. In: The 4th international seminar on advances in resistance welding. Wels, Austria, pp 30–60 5. Khan MI, Kuntz ML, Biro E, Zhou Y (2008) Microstructure and mechanical properties of resistance spot welded AHSS. Mater Trans JIM 49(7):1629–1637 6. Chuko WL, Gould JE (2002) Development of appropriate resistance spot welding practice for transformation-hardened steels. Welding J 81(1):1-s–7-s 7. Baltazar Hernandez VH, Okita Y, Zhou Y (2012) Second pulse current in resistance spot welded TRIP steel—effects on the microstructure and mechanical behavior. Welding J 91(10):278-s– 285-s 8. Kou S (2003) Welding metallurgy, 2nd edn. Wiley-Interscience, London, p 232

Comparison of Metro Track Vibration with Federal Transit Administration Limits Chaitanya V. Bhore , Atul B. Andhare , and Pramod M. Padole

Abstract Structures located in the vicinity of at grade (surface) and elevated metro rail are subjected to vibration induced by metro movements. Instruments with a very high level of precision such as MRI, X-Ray and microscopes are sensitive to external vibrations. Externally induced vibrations may affect overall performance of such instruments. Also, considering the case of human comfort, noise produced by such vibrations above audible limit causes discomfort in adjacent buildings, residents and industries. It is thus important to measure quality and quantity of vibrations transferred from metro track to ground. These measurements can be compared with FTA limits and respective measures may be taken. In this paper, frequency and amplitude of vibration are measured on the metro track and compared with the limits provided in Federal Transit Administration (FTA) for human comfort. While the metro train was in moving condition, track vibration was measured at a fixed point. Measurements were taken for three different positions of metro train with respect to the measurement point, such as when metro train is (a) approaching, (b) above and (c) crossed the measurement point. amplitude and frequency content was also found for all the three conditions. These experimental results are studied and discussed in this paper, and they will provide the basis for excitation to be considered for analysing vibrations of nearby structures or instruments. Keywords Metro train · Structures · Track · Vibration · Federal Transit Administration

C. V. Bhore (B) · A. B. Andhare · P. M. Padole Department of Mechanical Engineering, Visvesvaraya National Institute of Technology, South Ambazari Rd., Ambazari, Nagpur, Maharashtra 440010, India e-mail: [email protected] URL: http://vnit.ac.in/ A. B. Andhare e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_50

425

426

C. V. Bhore et al.

1 Introduction Exponential growth of population takes place when there is industrialisation in certain area. Same thing occurred in Nagpur city in the past few years. Increased traffic resulted in increased traffic, and therefore, need for new transport facility became important. One of the best and efficient urban transport in today’s time is metro rail. Metro rail has shown its impact on reduction in traffic and ease of transport in many big cities in India such as Kolkata, Mumbai and Hyderabad. Therefore, in Nagpur city, metro rail has been constructed. With population of about 2.6 million, Nagpur is the fourth largest city in the state of Maharashtra, India. Maximum speed and acceleration of metro train is 1.1 m/s2 and 95 km/h respectively, according to its design. Metros and subways attract urban construction of buildings, offices, etc., in the vicinity of their transport route, as they are easily accessible. Metro trains on surface and under ground (subway trains) induce vibration in structures near to its tracks. These induced vibrations hamper the operation of equipments and instruments which are highly sensitive to external vibration. Precision equipments and devices are used in medical laboratories and precision manufacturing facilities, which are advanced technologies in today’s world are installed in facilities near metro rail. Most of the instruments which are sensitive to external vibrations such as MRI fall have sensitivity falling in the range of 8–100 Hz. Along with effect on vibration-sensitive equipment, comfort for humans is also compromised in office and resident buildings. Frequency range between which the audible noise and feelable vibrations lie are 20– 20,000 Hz and 8–80 Hz, respectively. It is thus imperative to reduce these vibrations, by understanding and quantifying their levels so that the occupant or operations inside the building are not affected severely. Adam and Estroff [1] studied and showed that Rayleigh vibration waves travel (propagate) close to soil surface and effectively gets transferred to adjacent structures or buildings. Anderson [2] studied and measured vibration in buildings in the vicinity of railway track which were subjected to vibration produced by motion of surface trains. 5–50 Hz is the frequency range for vibration waves so as to propagate through ground as well as perceptible to human senses. Xia et al. [3] studied effect of vibration due to distance from foundation and train’s speed on six-storied building near Guangzhou station in Beijing. Such studies help designers to design and eliminate vibrations in areas in the building which are highly sensitive to vibration. This study aims to measure and analyse quality and quantity of vibrations produced on metro track. Vibrations produced by metro train on track are measured and compared with the limits provided in FTA [4] vibration criteria. These findings provide basis for base excitation, which is necessary in designing of vibration mitigation models and hence minimising the induced vibration to a limit which is comfortable for humans as well as vibration-sensitive equipments.

Comparison of Metro Track Vibration with Federal …

427

2 Measurement Set-Up The measurements were carried out with the help of OROS data acquisition and vibration analyser. It was carried out on track near Airport South metro station, Nagpur city, Maharashtra, India. This portion of metro rail is on ground which makes it easy to measure track vibration. Figure 1 shows schematic diagram of measurement set-up showing details of various components in it. Accelerometer position on the track base is such that it will measure vibration amplitude in vertical direction. Strong magnetic base at accelerometer end maintains firm contact with metro rail track. Figure 1 shows the schematic of the measurement set-up and Fig. 2 shows the photo of actual measurement set-up. As shown in Fig. 1 accelerometer is connected to DAQ system. This signals through DAQ are communicated to the computer through computer bus, which can be interpreted by computer software. For frequency analysis, fast Fourier transform (FFT) was used. The measurements were taken for 6 s interval when the train was in motion. Data were acquired in three different conditions when metro train is (a) approaching, (b) above and (c) crossed the measurement point. Figure 2 shows actual measurement set-up of track vibration showing accelerometer, data acquisition system, battery and laptop.

Fig. 1 Schematic diagram of experimental set-up

Fig. 2 Measurement set-up for measuring vibration of the track

428

C. V. Bhore et al.

Table 1 Root mean square (RMS) values of amplitude for track vibrations for different conditions Case Condition of measurement RMS (mm/s2 ) (w.r.t measurement point) a b c

When the train is approaching When the train is above When the train has crossed

2296.2 3789.9 756.1

Fig. 3 Spectrum of signal when the train is approaching the measurement point

3 Results Measured vibration results were obtained through OROS software in terms of acceleration (m/s2 ) versus time (s) graph. Table 1 shows root mean square values of acceleration amplitude (m/s2 ) of track vibration when the train is at different positions from the measurement point. Fast Fourier transform (FFT) is used to convert time domain signal to frequency spectrum. With the help of this frequency spectrum, one can pinpoint maximum amplitudes for its respective frequency. Frequency spectrum of all the conditions are found. Figure 3 shows spectrum of measured time domain signal when train is some distance away and approaching the measurement point. It was observed that when metro train is some distance away and approaching the measurement point, maximum amplitude of acceleration occurred at 1600 Hz with magnitude of 525 (mm/s2 ). Similar graphs were plotted for remaining two case as shown in Fig. 3, and maximum acceleration amplitudes corresponding to their frequencies were found. They are shown in table.

4 FTA Criteria Federal Transit Administration can be considered as a manual which provides guidelines for reviewing and preparing the vibration and noise-related factors affecting surrounding environment caused due to different modes of transport. FTA gives reg-

Comparison of Metro Track Vibration with Federal …

429

ulation related to environmental impact of most common transit projects providing different levels of environmental analysis. In this study, train-induced vibrations to the ground and nearby structures are compared with limits given in FTA. Vibration data of train-induced vibration is processed as the regulations in FTA as this is a transient event in which vibration amplitude varies according to the position of train passing-by [4]. Wave signals obtained through train vibration were analysed in frequency domain using fast Fourier transform (FFT). For analysis peak, frequency in different frequency band was recorded. FTA has provided guidelines for conversion of acceleration (a) (mm/s2 ) to velocity (v) (mm/s) and further to decibel (dB) scale (velocity level) (vdB) as given in Eqs. (1) and (2) v=

a i(2π f ) 

v vdB = 20 log vref

(1)  (2)

where vref is 5 × 10−8 m/s as reference velocity, f is frequency in Hz and velocity v in m/s. As the speed of train may be different from what is mentioned in the FTA guidelines, adjustment for different train speed must be incorporated and is given by Eq. (3) [4]   speed (3) adjustment (dB) = 20 log speedref where for comparison of vibration speed, speedref is used. If the measured speed is lower, then we must increase the dB level for comparison with speedref dB level. Reference speed is defined as the most common speed in one set of measurements. Structures and humans are having some sort of response to ground-borne vibrations, some of which vibrations sources are illustrated in FTA [4]. Many different applications are having their specific range of vibration, but in this case, we are interested in vibration level ranging from 50 to 100 vdB. For human comfort as well as many instruments which are vibration-sensitive, the threshold specified is 65 vdB. In case of highly sensitive equipment, such as electron microscope and lithography equipment, below 48 vdB and 54 vdB, respectively, are considered. But when we consider extremely sensitive equipments, vibration of 42 vdB is considered as safe. Using Eq. (1), peak accelerations from Table 2 are converted to velocity. These velocities are further converted to velocity levels in decibels using Eqs. (2) and (3). Peak vibration level found at different train position with respect to measurement point shown in Table 3. The values obtained as shown in Table 3 are around 10–15 vdB less than the limits provided in FTA. Limits provided in the FTA are given for objects which are more than 15 ft away from the track. As the vibration levels directly measured on the track are under limit, it can be said that the vibration at distance more than 50 ft which is the minimum distance for having a structure near railway track, will also be under limit.

430

C. V. Bhore et al.

Table 2 Maximum acceleration amplitude and corresponding frequencies for all three cases Case Condition of Frequency (Hz) Acceleration (mm/s2 ) measurement (w.r.t measurement point) a b c

When the train is approaching When the train is above When the train has crossed

525

1600

2117

8100

390.5

7950

Table 3 Peak vibration velocity levels (db) and corresponding frequencies at different positions of train Case Condition of Velocity level (vdB) Frequency (Hz) measurement (w.r.t measurement point) a b c

When the train is approaching When the train is above When the train has crossed

56.23

1600

58.40

8100

43.88

7950

5 Conclusion Vibration induced due to train motion to the ground adjacent to its track was measured and studied. Maximum value amplitude of acceleration was found out for three cases of metro train position with respect to measurement point. As the metro rail in Nagpur is yet to start in its full capacity, therefore, its speed in the trial period lies in the range of 25–30 km/h. According to FTA guidelines, the velocity amplitudes of vibration corresponding to their measured acceleration amplitudes at their respective frequencies were calculated. On comparing with the FTA limits, it shows that they are in limit and are around 10–15 vdB less than which was suggested. Therefore, it can be said that there will be no disturbance to vibration-sensitive equipment as well as to human comfort in nearby buildings.

References 1. Adam M, Von Estorff O (2005) Reduction of train-induced building vibrations by using open and filled trenches. Comput Struct 83(1):11–24 2. Anderson DC (1994) Engineering prediction of railway vibration transmitted in buildings. Environ Eng 7(1)

Comparison of Metro Track Vibration with Federal …

431

3. Xia H, Chen J, Wei P, Xia C, De Roeck Guido, Degrande Geert (2009) Experimental investigation of railway train-induced vibrations of surrounding ground and a nearby multi-story building. Earthq Eng Eng Vib 8(1):137–148 4. Hanson CE, Towers DA, Meister LD (2006) Transit noise and vibration impact assessment. Technical report

Effect of Moisture Content and Fiber Orientation on the Mechanical Behavior of GFRP Composites Alok Behera, M. M. Thawre, Atul Ballal, Prathamesh Babrekar, Pratik Vaidya, Satya Vijetha, and Tushar Sawant

Abstract The glass fiber reinforced polymer (GFRP) has provided better strength to weight ratio for various structural applications. Due to their non-corrosive nature, they are particularly suited for corrosive environments where materials like steel can cause problems. The aim of this research is to determine the combined effect of moisture and fiber orientation on the static strength of GFRP composites. Accordingly, twenty-four layered woven glass fiber/epoxy laminated composite was fabricated using the autoclaving technique. Moisture absorption and mechanical tests were carried out with two samples with different fiber orientation, i.e., 0°/90° and +45°/−45°. The specimens accelerated moisture aged with 70 °C tap water immersion, and the percentage change in moisture content was calculated up to saturation point. Then, the tensile test was carried out using the Instron universal testing machine with a constant crosshead speed of 1 mm/min. Higher tensile strength was observed for 0°/90° laminate with fiber allied parallel to the loading direction. Tensile strength was decreasing post-moisture absorption as compared to virgin laminates. After the mechanical tests, fracture analysis of tested specimens was carried out using scanning electron microscopy (SEM) which revealed a reasonable post aging fiber–matrix bonding degradation. Keywords GFRP · Moisture aging · Fiber orientation · Tensile strength · SEM

1 Introduction Glass fiber reinforced polymer (CFRP) composite structures are widely employed as aerospace and automobile components to reduce structural weight [1–3]. Composite structures have gained importance in many weight-sensitive applications due to their superior properties over metals. However, the use of GFRP composite laminate is subjected to its physical and mechanical performance in the desired environmental A. Behera (B) · M. M. Thawre · A. Ballal · P. Babrekar · P. Vaidya · S. Vijetha · T. Sawant Department of Metallurgical & Materials Engineering, Visvesvaraya National Institute of Technology, Nagpur, Maharashtra 440010, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_51

433

434

A. Behera et al.

condition [4, 5]. The application sector involving high-temperature humid environment requires static and dynamic analysis of composite in that particular condition. The main advantages of using fiberglass-reinforced thermosetting resin to produce marine structures include lightweight structures, lower fabrication and raw material cost, ease of fixing additional devices and ease of use [6]. Even though composites have interesting mechanical properties, they are also characterized by disadvantages such as their sensitivity to temperature and environment, which may influence their mechanical properties and reduce the stiffness of the structure [7, 8]. This paper examines the effect of long-term high-temperature water aging and fiber orientation on static behavior of GFRP composite laminates.

2 Materials and Methodology A twenty-four layered woven glass fiber/epoxy laminated composite was fabricated using the autoclaving technique. The actual fiber/matrix volume fraction was evaluated using the burning method as per ASTM D2548-68 standard. The specimen was kept in a furnace to degrade the matrix and calculate the fiber volume fraction from the leftover reinforcement. The specimens were hygrothermal aged up to saturation point using tap water at 70 °C. The increase in weight was measured in a regular interval, and percentage water absorption was evaluated using Eq. 1. %Moisture content =

Wf − Wi × 100% Wi

(1)

where W f = final weight of the composite after a particular time interval and W i = initial weight of the composite both in grams. The tensile tests of both virgin and moisture aged specimen were carried out using the Instron universal testing machine (Make: INSTRON, Model: 8802, Capacity-250KN). The tests were carried out as per ASTM D3039 standard with a constant crosshead speed of 1 mm/min. The tensile tests of moisture aged specimens were carried out as soon as it reached a moisture saturation point. Finally, the fracture morphology and mechanism were analyzed using scanning electron micrographs (Make: JEOL, Model: 6380). for both virgin and moisture aged specimens. Due to the non-conducting nature of the composite, the sample was coated with palladium using a coater.

Effect of Moisture Content and Fiber Orientation …

435

Fig. 1 Samples before (a) and after (b) the burn-off test

3 Results and Discussions 3.1 Burn-Off Test The burning method (ASTM D2548-68) has been considered as a simple and effective technique to determine the volume fraction of cured resin composite materials. A furnace was preheated to 600 °C, and the specimen (20 mm × 10 mm × 4.5 mm) was inserted into it placing it in a crucible. Prior to that, the mass of the specimen and the crucible together was determined. The crucible was placed till all the epoxy was degraded due to high temperature leaving behind only reinforcement. Then, the crucible was removed from the furnace and placed carefully on a gram scale after it reaches room temperature. A detailed image of sample before and after the burn-off test is shown in Fig. 1a, b, respectively. The volume fraction of fibers was observed to be 0.55, and weight fraction was 0.6. The mechanical properties of GFRP composite laminate are related to the mechanical properties and fractional volume of each phase, matrix and reinforcement. Amount of fiber in a fiber-reinforced composite directly corresponds with the mechanical properties of the composite [9]. Adding too little fiber reinforcement in the composite will actually deteriorate the properties of the material.

3.2 Moisture Absorption Test The moisture aged tensile testing specimens of dimension 130 mm × 10 mm were cut from the composite sheet. These samples were weighed with an accuracy of three decimal places before aging which was their initial weight W i . These were kept in a beaker with water and ensured that the samples were fully dipped into the water throughout the aging duration. The trend of moisture absorption of the glass fiber reinforced epoxy composites at 70 °C is shown in Fig. 2.

436

A. Behera et al.

Fig. 2 Weight gained by the composite versus time

Series 1–6 in Fig. 2 represent six static strength specimen which had minor variation in weight before aging. It can be noted that all the six samples showed a linear slope increase for initial 4–5 weeks which was due to capillary action along the fibers. Then, there was a minor weight gain or loss in the moisture absorption curve. The initial rate of moisture absorption was 1%, and it was maximum after two months. Time to reach the saturation point was about two months. This rate of moisture uptake was dependent on the temperature, relative humidity, exposure time and mechanical load. This absorbed moisture adversely affects the performance of laminate in a number of ways like (1) dimensional changes (swelling); (2) reduction in the glass transition temperature and (3) variation in mechanical and physical properties [5]. The effect of moisture is to cause hydrolytic breakdown of the fiber–matrix interface resulting in a loss in the efficiency of load transfer between the matrix and the fiber reinforcement [8].

3.3 Tensile Testing The tensile strength of aged samples was observed to be decreased by about 50% as shown in Fig. 3. The aged samples failed at half the load value as compared to virgin samples as revealed from true stress vs true strain curves given in Fig. 4a–d. This shows that there was a decrease in strength due to moisture absorption, irrespective of the fiber orientation, i.e., 0°/90° and +45°/−45°. Even the percentage degradation was very similar to each other. Fibers–matrix bonding degradation was the primary concerning issue. The loss of tensile strength was dependent on exposure time, temperature and degree of humidity [4]. On initial exposure to a high-temperature water environment, the rate of fiber degradation is relatively rapid.

Effect of Moisture Content and Fiber Orientation …

437

Fig. 3 Tensile strength of virgin and aged laminates

Fig. 4 True stress versus true strain curve for a virgin 0°/90°, b Aged 0°/90°, c virgin +45°/−45°, d aged +45°/−45° laminates

438

A. Behera et al.

Fig. 5 Scanning electron micrographs of a virgin 0°/90°, b virgin +45°/−45°, c aged 0°/90°, d Aged +45°/−45° laminates

3.4 Fracture Morphology Analysis The fracture surfaces of both virgin and aged tensile test specimens were analyzed using a scanning electron microscope. Figure 5a illustrates individual fibers separated showing clearly the 0° and 90° orientation of fibers [9]. Similar layered wise tearing of fibers was observed in virgin +45/−45 laminate as shown in Fig. 5b. This shows that the fiber failure was in the direction of their orientation, i.e., 45° direction. The fiber pull-out in brittle mode and fiber breakage was observed [10]. The moisture aged specimen showed major delamination and weak interface caused due to the moisture uptake as shown in Fig. 5c [6]. Matrix erosion was also observed in saturated moisture aged specimen as shown in Fig. 5d.

4 Conclusions In the present investigation, the effect of moisture absorption characteristics and its effect on tensile properties of glass fiber reinforced epoxy composite were investigated. Few major conclusions from the work are listed below

Effect of Moisture Content and Fiber Orientation …

439

1. The burn-off test analysis showed the fiber content of the sample to be 55% 2. There was a significant 50% decrease in tensile strength of aged 0°/90° laminate as compared to virgin, whereas the difference was not that huge in +45°/−45° specimen. 3. The thickness of the sample exposed to moisture was increased due to absorbed moisture by epoxy. 4. The mode of failure was brittle matrix dominated with fiber pull-out, matrix erosion and delamination in both virgin and aged specimens.

References 1. Menail Y, El Mahi A, Assarar M, Redjel B, Kondrats A (2009) The effects of water aging on the mechanical properties of glass-fiber and Kevlar-fiber epoxy composite materials. Mechanika 76:28–32 2. Behera A, Dehury J, Thaware MM (2019) A comparative study on laminated and randomly oriented Luffa-Kevlar reinforced hybrid composites. J Nat Fibers 16:237–244. https://doi.org/ 10.1080/15440478.2017.1414653 3. Behera A, Thawre MM, Ballal A (2018) Effect of matrix crack generation on Fatigue life of CFRP multidirectional laminates. Mater Today Proc 5:20078–20084. https://doi.org/10.1016/ j.matpr.2018.06.374 4. Poodts E, Minak G, Zucchelli A (2013) Impact of sea-water on the quasi static and fatigue flexural properties of GFRP. Compos Struct 97:222–230. https://doi.org/10.1016/j.compstruct. 2012.10.021 5. Fang Y, Wang K, Hui D, Xu F, Liu W, Yang S, Wang L (2017) Monitoring of seawater immersion degradation in glass fibre reinforced polymer composites using quantum dots. Compos Part B Eng 112:93–102. https://doi.org/10.1016/j.compositesb.2016.12.043 6. Wang Z, Zhao XL, Xian G, Wu G, Singh Raman RK, Al-Saadi S (2017) Durability study on interlaminar shear behaviour of basalt-, glass- and carbon-fibre reinforced polymer (B/G/CFRP) bars in seawater sea sand concrete environment. Constr Build Mater 156:985–1004. https:// doi.org/10.1016/j.conbuildmat.2017.09.045 7. Malpot A, Touchard F, Bergamo S (2016) Influence of moisture on the fatigue behaviour of a woven thermoplastic composite used for automotive application. Mater Des 98:12–19. https:// doi.org/10.1016/j.matdes.2016.02.123 8. Katunin A, Gnatowski A, Kajzer W (2015) Evolution of static and dynamic properties of GFRP laminates during ageing in deionized and seawater. Adv Compos Lett 24:47–52 9. Behera A, Dupare P, Thawre MM, Ballal AR (2019) Effect of fatigue loading on stiffness degradation, energy dissipation, and matrix cracking damage of CFRP [±45]3S composite laminate. Fatigue Fract Eng Mater Struct 42:2302–2314. https://doi.org/10.1111/ffe.13065 10. Behera A, Thawre MM, Ballal A (2019) Failure analysis of CFRP multidirectional laminates using the probabilistic weibull distribution model under static loading. Fibers Polym 20:2390– 2399

Experimental Investigation and Simulation of Modified Evaporative Cooling System Manju Lata and Dileep Kumar Gupta

Abstract This paper presents design and fabrication of modified evaporative cooler for producing cooled air without an increase in humidity. An experimental investigation has been carried out in Ahmedabad, India, and subsequently, a mathematical model is developed for the analysis of heat and mass transfer involved during the process. The model is validated with the measured experimental data, and further, the validated model has been used to analyze the performance of the system for Ahmedabad weather conditions throughout the year. The result shows that the minimum and maximum reduced ambient temperature obtained using a modified evaporative cooling (MEC) system is 6.52–34.31 °C and the maximum temperature drop is 17.85 °C. This system is made of kraft paper and aluminum sheet and it can give a better result than the direct and indirect evaporative cooler. The air can be cooled here lower than the wet-bulb temperature without an increase in humidity, and subsequently, it is an economical solution that can provide better comfort. Keywords Modified evaporative cooler · Counter-flow configuration · Effectiveness

Nomenclature As C h hm H Hl H wv

Surface area (m/s) Specific heat capacity at constant pressure (J/kgK) Heat transfer coefficient (W/m2 K) Mass transfer coefficient (m/s) Enthalpy (J/kg) Latent heat of vaporization of water at 0 °C (J/kg) Enthalpy of the water vapour at water film temperature (J/kg)

M. Lata · D. K. Gupta (B) Institute of Infrastructure Technology Research and Management (IITRAM), Ahmedabad, Gujarat, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_52

441

442

k m˙ t W W˙ evap

M. Lata and D. K. Gupta

Thermal conductivity (W/mK) Mass flow rate (kg/s) Temperature (K) Humidity ratio (kg/kg of dry air) Water evaporation rate (kg/s)

Subscript a db dp in l out p s w wb w_d wf wv

Average Dry bulb Dew point Inlet Wall Outlet Primary air Supply air Working air Wet bulb Dry working air Water film Water vapour

Greek 1 ρ

Effectiveness Density (kg/m3 )

1 Introduction Technological development and increased population growth along with the improvement in living standards resulted in enlarging demands comfort in life. Therefore, the demand for air conditioning systems is continuously increasing which leads to large electricity consumption and also affects the environment both directly (i.e., use of harmful refrigerant) and indirectly (i.e., electric consumption). However, there is a large population who cannot afford air conditioning; hence, they have to depend on the evaporative cooler, as it is economical. The conventional evaporative cooling system can give satisfactory cooling but the uncontrolled humidity is the major disadvantage of the system, which creates discomfort. One of the ways to solve the above

Experimental Investigation and Simulation of Modified …

443

problem is to use an indirect evaporative system, which can control the humidity, but it reduces the cooling and effectiveness compared to the direct evaporative system. In order to overcome these restrictions, it needs to be redesigned and modified for better comfort. There are recent developments of evaporative coolers, which are summarized below. Zhao et al. [1] presented a theoretical study on an indirect evaporative cooling system with counter-flow arrangement. The wet-bulb effectiveness (WBE) and dewpoint effectiveness (DPE) were found up to 1.3 and 0.9, respectively, at operating conditions of 28 °C dry-bulb temperature (DBT) and 20 °C wet-bulb temperature (WBT). Riangvilaikul et al. [2] have presented an experimental investigation on counter-flow dew-point evaporative cooler for dry, moderate, and humid climates. The range of WBE and DPE obtained was 92–114% and 58–84%, respectively. Subsequently, the theoretical model has also been presented, and the range of WBE and DPE achieved was 106–109% and 65–86%, respectively [3]. Later, Jradi et al. [4] presented on the cross-flow system, where WBE and DPE were found 112% and 78%, respectively, with 30 °C DBT and 50% relative humidity (RH) and 0.33 working to intake airflow ratio. Heidarinejad et al. [5] have done study on indirect evaporative cooler with a single and double stage, using a finite difference method. The results show 50% improvement in WBE with two-stage system. Duan et al. [6] developed and tested regenerative evaporative cooler with counter-flow arrangement for different operating conditions and found 31% and 40% improvement in WBE and energy efficiency ratio (EER), respectively. Khalid et al. [7] presented a study of counter-flow indirect evaporative cooler for hot, moderate, and humid climates. It was reported that WBE and DPE vary from 104–120 to 68–87% at inlet air temperature 25–45 °C. Furthermore, Liu et al. [8] concluded from the parametric study for two-dimensional numerical model that with increase in air ratio leads to higher temperature drop and cooling capacity. Further, Wang et al. [9] presented a numerical study on dew-point counter-flow indirect evaporative cooler. The optimum geometrical and operating conditions were reported. It has been observed from the literature that, a very limited study has been reported in Indian climatic conditions. In present work, a counter-flow MEC system has been designed, fabricated, and experimentally tested with local climatic conditions. A thermodynamic model has been developed and simulated using MATLAB® and validated with experimental data. This validated model has been used to analyze the year-round performance of the system using hourly ambient air condition (DBT and RH) data (Taken from the IITRAM weather station, installed at IITRAM, Ahmedabad, India) for Ahmedabad city.

2 Experimental Work The MEC system is designed and fabricated in the laboratory. The schematic diagram of MEC is shown in Fig. 1. The aluminum-coated kraft paper is used as a sheet material for making heat and mass exchanger unit (HMU). Thickness of aluminum

444

M. Lata and D. K. Gupta

Fig. 1 Schematic diagram of the MEC system

sheet is 0.2 mm while kraft paper’s thickness is 0.27 mm. Dry and wet channel is of 1000 mm * 550 mm * 1000 mm (length * Width * Height). The dry and wet passage gap is 5 mm with perforation diameter of 5 mm. The total number of 40 dry and 41 wet passages is used in HMU. There are three streams of air, i.e., air flowing in dry passages (product air), wet passages (working air), and the final output air ready to supply (supply air) as shown in Fig. 2, which pass between the dry and wet passage (aluminum-coated kraft paper) alternatively. A fraction of product air, after the dry passage, is allowed to pass through the wet passage from the top end of each sheet, and it is controlled by the perforation made in each sheet. Only sensible heat transfer takes place from product air to working air in the dry passage, and it divides into two streams at the top end of the dry passage, i.e., one stream as working air to wet passage and another as supply air. The working air gets pre-cooled before entering the wet passage and it flows in reverse direction of the product air, which enhances the effectiveness due to a higher temperature gradient between dry and wet passage. In the wet passage, both latent and sensible heat transfers take place. The working air comes in contact with wetted kraft paper, and evaporation of water takes place by absorbing the heat from it, which cools the wet surface, and working air, subsequently, allows sensible heat transfer from product air to working air. Finally, humidified air is discharged to the atmosphere from bottom of the wet passage.

Exhaust air

Inlet air

Wet Passage Water film Dry Passage

Working air

Product air

Supply air

Fig. 2 Counter-flow HMU airflow direction and its assembly

Experimental Investigation and Simulation of Modified …

445

3 Mathematical Model A mathematical model has been developed to study the process of heat and mass transfer, involved in the HMU. The heat and mass transfer equation is solved using the finite difference equation and simulated with the help of MATLAB® . The following assumptions have been made for the modeling: (a) negligible heat loss between system and surrounding, (b) interface temperature between working air and water film is water temperature, (c) the uniform properties and velocity of the fluid in a control volume, (d) uniform distribution of water in the wet channel, and (e) negligible thermal resistance. Energy balance equation for dry passage     m˙ p Cp tp_ in − tp_ out = h p dAs tp_ a − tl_ a

(1)

Energy balance and mass balance equation for wet channel     m˙ w Ww_ in − Ww_ out = h m dAs ρwf_ a − ρw_ a

(2)

  m˙ wf_ in − m˙ wf_ out = m˙ w Ww_ in − Ww_ out

(3)

      m˙ w Hw_ out − Hw_ in = h w dAs twfa − twa + Hwv h m dAs ρwf_ a − ρw_ a

(4)

Overall energy balance equation for dry and wet passage       m p C p tp_ in − tp_ out + m˙ w Hw_ out − Hw_ in = m wf Cwf twf_ in − twf_ out   − m˙ w Ww_ in − Ww_ out Cwf twf (5) Equations 1–5 have been solved simultaneously using the iterative method, and the cooling performance of MEC is evaluated by εwb =

tdb,in − tdb,out tdb,in − twb,in

(6)

εdp =

tdb,in − tdb,out tdb,in − tdp,in

(7)

4 Model Validation The mathematical model has been validated with the real-time data taken for the month of June 2018 (i.e., peak summer conditions) from 11.00 a.m. to 10.00 p.m.

446

M. Lata and D. K. Gupta

0.014

15 10

RAT(Simulation) 0.012 RAT(Experimental) TD(Experimental) 5 TD(Simulation) Ambient humidity ratio 0 0.01 36 37 38 39 40 Ambient air temperature ( °C)

0.019 0.018

0.8

0.017

0.6

0.016

WBE(simulation) 0.015 0.4 WBE(Experimental) 0.014 DPE(Experimental) 0.2 DPE(Simulation) 0.013 Ambient humidity ratio 0.012 0 36.5 37 37.5 38 38.5 39 39.5 40

Ambient Humidy ratio (kg/kg)

0.016 20

0.02

1 Effectiveness

25 Temperature ( °C)

1.2

0.018 Ambient Humidity ratio (kg/kg)

30

Ambient air temperature (°C)

Fig. 3 Comparison of experimental and simulation data by varying ambient air conditions

(IST) at Ahmedabad, Gujarat, India. It is observed that the ambient air temperature (AAT) and humidity ratio (HR) during the experimentation period varies from 34.4 to 39.5 °C, and 0.0206 and 0.0145 kg/kg of dry air, respectively. The minimum OAT is 34.4 °C with 0.0204 kg/kg (HR), whereas the maximum AAT is 39.5 °C with 0.0151 kg/kg (HR). Figure 3 depicts the comparison of experimental and simulated results, which includes the variation of reduced ambient temperature (RAT), temperature drop (TD), WBE, and DPE with different AAT and ambient HR. The operating range of AAT and ambient HR varies between 36.8 and 39.5 °C and 0.0173– 0.0151 kg/kg of dry air, respectively. A very good agreement between the simulated and experimental results has been found.

5 Results and Discussions

Bin hours

The validated mathematical model of the MEC system is used to analyze the yearround performance of the system for Ahmedabad city using hourly ambient air condition (DBT and RH) data. Figure 4 shows the variation in bin hours with respect to the ambient air conditions throughout the year. Ambient air condition is represented 40 35 30 25 20 15 10 5 0

0

100

200

300

400 500 600

700

800 900 1000 1100 1200 1300 1400 1500

Ambient air condition Fig. 4 Bin hours at ambient air conditions

Experimental Investigation and Simulation of Modified …

447

Reduced ambient temperature ( °C)

45 35

RAT

TD

25 15 5 -5 0

100

200

300

400

500

600

700

800

900 1000 1100 1200 1300 1400

Ambient air condition Fig. 5 Reduced ambient temperature at ambient air conditions

on X-axis, and Y-axis represents bin hours. Annually, DBT and RH vary between 8–44 °C and 23–100%, respectively. Figure 5 shows the variation of reduced ambient air temperature and temperature drop, using the MEC system at different ambient air conditions throughout the year as shown in Fig. 4. It has been observed that minimum and maximum reduced ambient temperature obtained by using the MEC system is 6.52 °C at the ambient condition of 8 °C DBT and 78% RH, and 34.31 °C, at the ambient condition of 42 °C DBT and 54% RH. However, maximum temperature drop obtained is 17.85 °C at 40 °C and 24% RH. It has also been observed that the performance of the system is better with the dry and hot climate. However, the performance is always better than the conventional system, as it reduces temperature even below the WBT, without increasing humidity. The system ensures that the maximum output temperature of air never goes beyond the 34.3 °C with the respective humidity throughout the year in Ahmedabad’s climatic conditions.

6 Conclusions Experimental investigation and thermodynamic study have been carried out for the MEC system having a counter-flow configuration of HMU. The year-round performance of the MEC system has been predicted, for Ahmedabad weather conditions. It is concluded that the maximum temperature drop obtained using the MEC system is found 17.85 °C at 40 °C and 24% RH. The proposed system can provide better comfort than the conventional configurations, throughout the year. This system can also be used to supply secondary air for any other cooling system of HVAC application.

448

M. Lata and D. K. Gupta

References 1. Zhao X, Li JM, Riffat SB (2008) Numerical study of a novel counter-flow heat and mass exchanger for dew point evaporative cooling. Appl Therm Eng 28:1942–1951 2. Riangvilaikul B, Kumar S (2010) An experimental study of a novel dew point evaporative cooling system. Energy Build 42:637–644 3. Riangvilaikul B, Kumar S (2010) Numerical study of a novel dew point evaporative cooling system. Energy Build 42:2241–2250 4. Jradi M, Riffat S (2014) Experimental and numerical investigation of a dew-point cooling system for thermal comfort in buildings. Appl Energy 132:524–535 5. Heidarinejad G, Moshari S (2015) Novel modeling of an indirect evaporative cooling system with cross-flow configuration. Energy Build 92:351–362 6. Duan Z, Zhan C, Zhao X, Dong X (2016) Experimental study of a counter-flow regenerative evaporative cooler. Build Environ 104:47–58 7. Khalid O, Butt Z, Tanveer W, Rao HI (2017) Design and experimental analysis of counter-flow heat and mass exchanger incorporating (M-cycle) for evaporative cooling. Heat Mass Transfer 53:1391–1403 8. Liu Y, Li JM, Yang X, Zhao X (2018) Two-dimensional numerical study of a heat and mass exchanger for a dew-point evaporative cooler. Energy 168:975–988 9. Wang L, Zhan C, Zhang J, Zhao X (2019) Optimization of the counter-flow heat and mass exchanger for M-Cycle indirect evaporative cooling assisted with entropy analysis. Energy 171:1206–1216

Effects of Different Vegetable Oils and Additives in Gearbox Operation and its Condition Monitoring Anupkumar Dube

and M. D. Jaybhaye

Abstract There is significant worldwide attention in recent time on condition-based intensive care of gear transmission systems across the globe from both industries and academia. The reason behind this is an effective CBM will always extend the life span of the rotating equipment and helps in reducing maintenance cycles. For the mechanical component’s good working cycle, different entities such as lubrication oil properties, operating speed, load condition, lubrication [oil and additives], temperature, and wear component are the major parameters which play contributory role. Vegetable oil as lubricants has many good and useful physicochemical properties. Vegetable oils possess high lubricity, viscosity Index, flash point, and low losses in evaporative. In this research work, a comparative behavior study of a designed worm gear test rig is done between selected commercial gear oil (HP EP 90) and natural edible and non-edible vegetable oils (used as gear oils) (natural coconut and castor oil) with and without additives (garlic oil and rapeseed oil). The temperature analyzing techniques were used to describe the performance of worm gear system as a function of load applied on the worm, and the temperature responses of a worm gear are plotted with varying loading conditions for the selected oils and additives. A mechanical closed-loop test rig is developed for carrying out this work. Keywords HP EP 90 · Coconut oil · Castor oil · Garlic oil · Rapeseed oil · Worm gear box · Oils · Additives

1 Introduction For the proper operation and compliance of worm gear, thermal stability plays an important role and it primarily depends on lubricating oil and additives. Temperature has direct effect on the lubrication oil viscosity, and however, the oil viscosity as well significantly affects the processes in the sliding zone, and hence the temperatures. A. Dube (B) · M. D. Jaybhaye Department of Production Engineering and Industrial Management, College of Engineering Pune, Pune, Maharashtra 411005, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_53

449

450

A. Dube and M. D. Jaybhaye

The optimum lubrication for the chosen lubricating oil can be achieved only within the suitable range of the operating temperature. Given a much higher proportion of sliding relative to rolling movement, worm gears have far more friction between the teeth sides than in the case of cylindrical and cone gears, which results in a significantly lower efficiency [1]. It is known that the efficiency of a worm pair depends on lubricant, sliding speed, load, and temperature [2]. Ruggieroa et al. [3] analyzed hydro-treated vegetable oil from raw rapeseed oil and raw Jatropha carcass oil. They have observed that the selected oils showed good physicochemical properties and can favorably be used as lubricant feedstock. Thottackkad1 et al. [4] in their research work have explored the possibility of use of vegetable base oils as automobile lubricants. They have concluded that as a lubricant, friction-reduction properties of coconut oil are enhanced by the addition of CuO nanoparticles to a moderate concentration. Mannekote and Kailas [5] in their research work have evaluated coconut and palm oil as lubricants in four-stroke engine. They have observed that antiwear properties of fresh vegetable base oils and engine oil were comparable at ambient conditions. Li et al. [6] evaluated natural garlic oil (NGO) as an extreme pressure additive for lubricating oils. They have demonstrated that NGO could provide superior loadcarrying ability in the selected base fluids than the conventional extreme pressure additive SIB. Jayadas et al. [7] experimentally evaluated the effect of an AW/EP additive on the tribological performance of coconut oil. They have observed that the addition of the AW/EP additive has shown substantial reduction in wear with coconut oil as 2T oil. Wu et al. [8] evaluated tribological properties of castor oil. Results observed showed that castor oil as the additive could efficiently reduce the friction and wear. Shanhua et al. [9] claimed that a small amount of ionic liquid can postpone the peak coefficient of friction and decrease the wear behavior of castor oil. Penga et al. [10] given detailed information on the dependent and independent roles of vibration and wear debris analyses in predicting and diagnosing machine faults.

2 Vegetable Oils as Lubricants Mineral oils are petroleum by-products and have very poor biodegradability, and they are major source of environmental pollution. Oils of natural ester (vegetable oils) have capability to act as a base for the environmentally friendly lubricating oils. They have good lubricity, biodegradation capability, viscosity versus temperature characteristics, low evaporation capacity, etc. Vegetable oils have high viscosity index (VI). Table 1 [11] provides details regarding rating of vegetable oil [rapeseed oil]properties compared with mineral oils ratings: A stands for excellent, B stands for

Effects of Different Vegetable Oils and Additives … Table 1 Vegetable oil and mineral oil properties ratings

Table 2 Coconut and rapeseed oil important properties

Oil property

451 Mineral oil

Vegetable oil

VI

D

B

Pour point

E

C

Thermal stability

D

D

Biodegradability

D

A

Lubricating property

C

A

Toxicity

C

A

Cost relation with mineral oil

–

B

Oil nature

Viscosity at 40 °C, (cSt)

Viscosity at 100 °C, (cSt)

Viscosity index ISO 2009

Coconut oil

27.7

6.1

175

Rapeseed oil

38

–

215

very good, C stands for good, D stands for moderate, and E stands for poor. The main important property of vegetable oil which is completely different from mineral oils is the carbons may or may not be saturated with hydrogen. Because of the presence of triglyceride units, vegetable oils are more polar compared with the mineral oils also triglycerides have a higher affinity to metal surfaces. The structure of the triglycerides of vegetable oils provided good properties of a lubricant. The important properties of comparison of the coconut oil and rapeseed oil are mentioned in Table 2 [11]. Biolubricants can act as potential substitute lubricants in different applications. One of the most effective properties of the biolubricant is their fast biodegradability and good VI. An increased number of studies have shown that the biolubricants have very good potential to be used as alternative lubricant. There are many researches going on in this area worldwide but still there is need of a systematic research to evaluate and validate the tribological behavior of different biolubricants for different applications and under different working conditions. As mentioned in Table 1, the biolubricants have advantages over the mineral oils in comparison with the important oil properties. Biolubricants have few disadvantages which are as well mentioned in Table 1. Various researches have worked on the possibility of using vegetable oils as lubricants using various methods of evaluation. Table 3 demonstrates the work done with a comparison done with the standard SAE reference lubricant [12].

452

A. Dube and M. D. Jaybhaye

Table 3 Vegetable oil-based biolubricant evaluation methods Natural oils

Reference oil

Investigation method

Results

Coconut oil

SAE20W50

Four-ball tester

Less coefficient of friction, better lubricity property Higher anti-wear properties, good oxidation and anti-corrosion properties, strong stability of the lubricant film

Castor oil

Refined mineral oil

Four-ball tester

Greater VI, cheaper and eco-friendly, lower volatility, lower deposit forming tendencies

3 Selection of Oil Blends Researchers have also shown that natural oils can also be used as an additive (like rapeseed oil, palm oil, etc.) as an antiwear additive and extreme pressure additives. Antiwear additives protect the sliding machine parts from wear and extreme pressure additives and create a thin film which prevents welding and seizure of contacting surfaces during working life of the sliding components. Based on the literature survey and availability in the market, natural garlic oil and rapeseed oil are used as an additive for the selected base oils that is coconut oil and castor oil. Literature survey suggests that natural garlic oil can act as a good lubrication oil additive. It has pale yellow color and has refractive index as 1.566, specific gravity (20 °C) as 1.081, acid value (mgKOH/g) as 1.6, and peroxide value (meq/kg) 2.03. Base oils and additives both are natural oils and are completely soluble without any solid particles remains in the blends. Table 4 shows the blends which are selected for this study. The behavior of the selected blends is compared with the standards HP EP 90 gear oil. Physical properties of selected natural oils are listed in Table 5 [13].

4 Test Rig Setup A worm gear test rig system is designed for carrying out this research work. Figure 1 shows the components of the designed test rig. The designed test rig apparatus contains a model gear box with half HP motor attached to it. A system is designed for applying loads at the output shaft of the gear box using hanger bearing. A square base is designed for this test rig. Rubber bushings are provided at the four legs of the rig to avoid any vibrations due to uneven resting of the test rig. In the designed test rig for applying load on the output shaft of the gear box assembly, a load cell and hanger bearing arrangement is used. The load cell and the hanger bearing are attached with each other using a threaded rod. A bolt is present at the top end of the load cell attachment plate. On tightening, the bold load gets applied

Effects of Different Vegetable Oils and Additives …

453

Table 4 Selected oil blends S. No.

Oil name

Oil type

Additive name

Blend name

% of additive

1

HP EP 90

Commercial gear oil

With no additive added

EP90 01

N/A

2

Coconut oil

Raw edible oil

With no additive added

Coconut 01

0

3

Castor oil

Raw non-edible oil

With no additive added

Castor 01

0

4

Coconut oil

Raw edible oil

Garlic oil as an EP additive. Rapeseed oil as a friction modifier additive

Coconut 02

10% [5% garlic oil + 5% rapeseed oil]

5

Castor oil

Raw non-edible oil

Garlic oil as an EP additive. Rapeseed oil as a friction modifier additive

Castor 02

10% [5% garlic oil + 5% rapeseed oil]

Table 5 Physical properties of selected vegetable oils Physical property

Coconut oil

Castor oil

Rapeseed oil

Iodine value

8–11

83–86

98–105

Pour point (°C)

12.7

−21

−15

Cloud point (°C)

13.1

−18

−2

Kinematic viscosity at 40 °C (mm2 /s)

27

251

35

Flash point (°C)

266

229

246

Density at 15 °C (g/cm3 )

0.918

0.960

0.912

on the output shaft of the gear box. The applied load can be measured using an output display unit attached with the load cell assembly. The output display is a commonly used weighing scale which displayed applied load in KG unit. The selected load cell is calibrated using a KG weight block. On keeping the KG blocks on this system, the output shows exactly same quantity in the display unit scale. Digital output devices are used for viewing the motor RMP value, gear box oil temperature, and load applied on the out rod in KG. For measuring accurate temperature of entire system, a smart vibration sensor QM42VT2 with bi-axial accelerometer and temperature measurement capability was selected. This is a plug and play type vibration sensor and does not need a separate data logger. All the electronics including filters and ATD converter are packed in a single unit that can be mounted on any parts for which vibrations and temperature are to be measured using a special mounting bracket and RS485 to USB connector.

454

A. Dube and M. D. Jaybhaye

Fig. 1 Test rig assembly components

5 Method of Experimentation For the current experimentation, gear box temperature and gear box oil temperature are selected as parameters. The test rig is operated for 70 min and applied load is increased by 10 kg after each 10 min run. Readings were taken after every 10 min. Two different readings were taken to check if there is any sudden change in the readings and then average of 2 readings is taken as final reading. A test plan was developed for taking different readings at the selected locations. Readings were taken for each selected five oil blends mentioned in Table 4. For recording gear box temperature responses, the Banner vibration–temperature sensors are used, and for recording gear oil temperature, RTD sensor with a digital display is used. The output load applied on rod readings was taken using a load cell mechanism with weighing scale display.

6 Results Graphs are generated for the temperature changes for the selected locations with respect to the varying loading condition for each selected oil blends. The readings are compared with the standard HP EP 90 commercial gear oil. Figure 2 shows gear oil temperature vs load applied. It has been observed that the temperature changes are less for coconut, castor additive, and coconut additive and higher for castor oil as compared with the standard HP EP 90 oil. Coconut and its additive perform way better than all. Figure 3 shows gear box temperature vs load applied. It has been observed that the temperature changes are less for coconut and coconut additive and higher for

Effects of Different Vegetable Oils and Additives …

455

Fig. 2 Gear oil temperature versus load applied

Fig. 3 Gear box temperature versus load applied

castor oil and almost same for castor additive as compared with the standard HP EP 90 oil. Coconut and its additive perform way better than all.

7 Conclusions For the designed test rig and for the selected oil blends and for the selected loading conditions, it can be concluded that coconut additive blend has very good temperature absorbing properties as compared with other selected oil blends. Performance of coconut additive blend is better than that of the coconut oil blend without any additive. [For maximum load condition coconut additive shows 14% less temperature while coconut oil shows 9% less temperature compared to the standard HP EP 90 oil]. Castor oil has somewhat less temperature observing properties as compared with other selected oil blends. Castor oil with additive performs better than that of the castor oil without additive. Gear box housing and gear oil temperature are almost same throughout the test. The selected oil blends have potential to be used as lubricating oils with proper additives added in them.

456

A. Dube and M. D. Jaybhaye

References 1. Opali´c M (1984) Prilogistraživanjuopteretivostibokovapužnih kola pužnihprenosnika. In: Contribution to study carrying capacity flanks of worm wheel of worm gear, Ph. D. thesis. Faculty of Mechanical Engineering and Naval Architecture, Zagreb, Croatia 2. Szeri AZ (1998) Fluid film lubrication—theory and design. Cambridge University Press, New York 3. Ruggieroa A, D’Amato R, Merola M, Valášek P (2017) Tribological characterization of vegetal lubricants: comparative experimental investigation on jatropha curcas L oil, rapeseed methyl ester oil, hydrotreated rapeseed oil. Tribol Int 109:529–540 4. Thottackkad MV, Perikinalil RK, Kumarapillai PN (2012) Experimental evaluation on the tribological properties of coconut oil by the addition of CuO nano particles. Int J Precision Eng Manufact 13(1):111–116 5. Mannekote JK, Kailas SV (2011) Experimental investigation of coconut and palm oils as lubricants in four stroke engine. Tribol Online 6(1):76–82 6. Li W, Jiang C, Chao M, Wang X (2014) Natural garlic oil as a high-performance, environmentally friendly, extreme pressure additive in lubricating oils. ACS Publication, pp 798–803 7. Jayadas NH, Nair KP, Ajithkumar G (2007) Tribological evaluation of coconut oil as an environment-friendly lubricant. Tribol Int 40:350–354 8. Wu X, Zhao Q, Zhang M, Li W, Zhaoa G, Wang X (2014) Tribological properties of castor oil tris (diphenyl phosphate) as a high-performance antiwear additive in lubricating greases for steel/steel contacts at elevated temperature. Royal Soc Chem 4(97):54760–54768 9. Shanhua Q, Xuliang C, Liguo L, Qingzhong L (2016) Tribological properties of the castor oil affected by the additive of the ionic liquid [HMIM] BF4. J Tribol 138(1) 10. Penga Z, Kessissogloub NJ, Coxa M (2005) Study of the effect of contaminant particles in lubricants using wear debris and vibration condition monitoring techniques. Int J Sci Technol Friction 258(11–12):1651–1662 11. Gnanasekaran D, Chavidi VP (2018) Vegetable oil based bio lubricants and transformer fluids applications in power plants, materials forming, machining and tribology. Springer, Berlin 12. Mobarak HM, Mohamad N, Masjuki HH, Kalam MA, Al Mahmud KAH, Habibullah M, Ashraful AM (2014) The prospects of bio lubricants as alternatives in automotive applications. Renew Sustain Energy Rev 33:34–43 13. Karmakar G, Ghosh P, Sharma BK (2017) Chemically modifying vegetable oils to prepare green lubricants. Lubricants

Study and Analysis of Various Parameters of Bio-mechanization Plant Deepak Patil , Rahul Barjibhe, Lakhan Meghani , Omkar Nanaware , Tejas More , and Aditya Pujari

Abstract Among the recent challenges in India, one is the extensive consumption of non-renewable energy sources due to rising demands of fuels and environmental issues. India is an agriculture-based country with the second-highest population in the world. Converting the waste material into the energy and use of renewable sources is the best way to serve the energy needs of the population. The biogas production in India is around 2.07 billion cubic meters per year which are targeted to increase up to 30 billion cubic meters per year by using various advanced technologies. To extract more methane percentage, various plants are approaching toward pressure swing adsorption (PSA) technology. PSA technology increases the percentage of methane by reducing carbon dioxide and other gases’ percentage. In this paper, we have studied and analyzed various parameters of a bio-mechanization plant which is based on segregated organic municipal waste for the production of bio-CNG and organic fertilizer and is working on the vapor pressure swing (VPSA) technology. The effect of season, temperature, waste obtained, etc. on bio-CNG production has been researched. Keywords Bio-CNG · Pressure swing adsorption · Methane · Waste

1 Introduction The most important factor for India to reach the status of developed countries is the renewable source of energy. The annual LPG consumption in India is 101.4 kg and 119.3 kg for rural and urban households, respectively [1]. Biogas is a renewable energy source that is generated by the action of the methanogenic bacteria on the municipal solid waste (MSW), industrial wastes, hotel waste, animal and agricultural wastes. Biogas as a renewable energy source contributes 14% of the total renewable D. Patil (B) · R. Barjibhe Shri Sant Gadge Baba College of Engineering and Technology Bhusawal, Bhusawal, India e-mail: [email protected] L. Meghani · O. Nanaware · T. More · A. Pujari D. Y. Patil College of Engineering, Akurdi, Pune, India © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_54

457

458

D. Patil et al.

energy produced in India [2]. It has been estimated that India has biogas production potential of 40,734 Mm3 /year from different organic wastes [3]. More than 127,486 tons per day of MSW is been generated [4]. National Biogas and Manure Management Program (NBMMP) organization of India is taking efforts to install biogas plants in rural and semi-urban areas in various states of India [5]. Biogas is a mixture of various gases such as methane, carbon dioxide, hydrogen sulfide, nitrogen, and water vapors. The percentage of methane (CH4 ) is in a range of 50–70% and that of carbon dioxide (CO2 ) is 30–50%. Small traces of nitrogen (N2 ) and oxygen (O2 ) are present at concentration of 0–3% and water vapors (H2 O) at concentration of 5–10% [6]. The purification of biogas can be done by the various technologies. The percentage extraction of methane depends on the technology being used. The technologies used are cryogenic separation of gases, physical and chemical adsorption, membrane-based technology, water scrubbing, chemical scrubbing, and other biological methods of separation [7, 8]. Water scrubbing is the oldest technology which is mostly used in India in old biogas plants. The percentage of methane obtained from water scrubbing is about 58% which is less and needs to be improved [9]. The biological method of methane production is an upgrading technology and research is being carried out to improve the bio-mechanization efficiency [6]. Various methods have been tested for the reduction of carbon dioxide in biogas by the use of alkali metal compounds. Biogas plants are looking forward to some modern techniques which improve methane percentage up to maximum level by reducing carbon dioxide for bio-CNG. Pressure swing adsorption (PSA) is a promising technology which improves the methane percentage up to 85% by reducing carbon dioxide into methane [10]. The purification of biogas depends on the various parameters. For example, in summer and winter, the amount of methane produced is different [11]. Also, the temperature and type of waste affect the rate of purification in different seasons [12]. In VPSA, biogas is compressed at high pressure and then fed to the column. The column consists of adsorbents like activated carbons, zeolite molecular sieves, etc. which have high porosity. Methane-rich gas is obtained at top of the column and CO2 is removed. Hydrogen sulfide adsorption is irreversible so it is separated first. As the adsorbent gets saturated with CO2 after several passages, it has to be regenerated. It can be easily done by reducing the pressure up to vacuum pressure. This method yields highly pure form of bio-CNG, and hence, its working was studied in bio-mechanization plant based on VPSA. The bio-mechanization plant of Indore has been researched by our team. The name of the plant is “Design, Construction, Supply, Installation, and Commissioning of 15 TPD Bio Mechanisation Plant-Based on Segregated Organic Municipal Waste for Production of Bio-CNG And Fertilizer” and the work is executed by Mailhem Ikos Environment Private Limited. The plant is situated at Kabit Khedi, Indore. Various readings were taken to plot the graphs. The recordings obtained were analyzed to come across some crucial conclusions.

Study and Analysis of Various Parameters …

459

2 Methodology As briefed earlier, the potential of food waste for biofuels was realized and research was carried out. Various technologies were studied and it was found out that PSA and water scrubbing technologies are effective in India. The most recent plant with advanced technology setup in 2018 by the Government of India is in Indore, Madhya Pradesh. It purifies biogas with the use of vapor pressure swing adsorption (VPSA) technology. The plant supplies bio-CNG to be used by the local buses by collecting local hotel waste. We visited the plant on June 8–9, 2019, to know more about VPSA technology. The data collected included waste intake and bio-CNG filled for three months (from February 2019 to April 2019) (Fig. 1). The data collected was conditioned for parameter selection. A quantitative approach was used and analysis was done after parameter selection and correlation. This research paper is basic, exploratory, and inductive. The data is primary, quantitative, and descriptive. It is fixed, field research with longitudinal studies and non-probability sampling. Based on the waste-related data, the quantity of useful waste was taken into consideration. The record of CNG filled in the vehicles was collected. The data is organized in a pivot table in Excel, in order to extract dataset. We plotted graph for waste rejected and net waste for a particular period. Also, to establish the effect of season on the biogas purified and temperature variation, graph was plotted on the corresponding parameters. The flow of materials and plant description: Daily waste is collected from local hotels by municipal garbage trucks and brought to the plant. The waste is weighed on the weighing machine, which is capable of measuring tons of weight. Then the waste is dumped in an area allocated for dumping near to the feeding conveyor, on which the waste is manually segregated by the workers. Segregation is required as the waste comprises both biodegradable and non-biodegradable types of waste. From the feeding conveyor, after the separation, the waste comes on to shredding table where a 5Hp motor shreds the biodegradable waste into finely chopped particles of the waste. Recycled water from the digester is being used in the shredder for proper mixing of the biodegradable wastes. From the shredder, the slurry of bio-digester goes into a pre-digester where it is stored for a

Fig. 1 Flowchart for methodology

460

D. Patil et al.

while before inserting it into the primary digester. The larger sized particles which are not chopped into shredder are again brought back onto the shredder table for shredding. Organic waste compost is obtained from the solid–liquid separator which is useful as a fertilizer in farms. From the pre-digester, the digest is pumped inside a slurry tank, also called primary digester and secondary digester, with the help of two 5HP motors. The digest is injected from the top in the slurry tank. The anaerobic bacteria digest the waste and produce slurry mixed up of some raw biogas. An agitator helps to reduce the formation of foam/scum. Slurry form of waste is injected into the Mailhem Ikos MUASB digester to produce the purified form of methane gas after the pre-digester. The digester has a temperature and pH level sensor to maintain the temperature and pH level of the digester within prescribed limits. The recycle and sludge lines are present to reuse the highly viscous digestate present in the digester. From the digester, the lines go to the storage of the raw biogas. Moisture trappers on the lines help to reduce the moisture content in the raw biogas. The balloon has a capacity of 375 m3 each to store the raw biogas into it. From the balloon, the raw biogas is let into the VPSA purification system wherein biogas is purified to bio-CNG (Fig. 2). Along the way, a mass flow meter is used to measure the amount of raw biogas and a flare unit is inserted to flare the gas into the atmosphere if the purification system is not in working condition. A blower helps to increase the mass flow rate of the raw biogas and increase its capacity of purification. The raw biogas is purified inside the purification system (VPSA technique) and is regularly monitored by gas chromatography. After the purification, the purified gas is compressed by a compressor so as to fill in the tank for storage purpose. Further, the gas is dispensed through a CNG dispenser and filled in government buses.

Fig. 2 Plant layout

Study and Analysis of Various Parameters …

461

3 Observations and Results The observed readings of CNG produced, net waste, waste rejected and atmospheric temperature is taken and analyzed to plot the effect of temperature variation on CNG production, effect of used MSW variation on CNG production, and variation of the MSW and waste rejected. Figure 3 shows the variation of the MSW and waste rejected. The MSW generated in the first half of March summed to the value of approximately 160,000 kg. It shows that more the waste generated, more care is to be taken to reject the useless waste to produce good quality CNG. Figure 4 shows the variation of CNG produced from the MSW in the interval of seven days. It can be observed that the CNG production is directly proportional to the amount of solid waste fed. The month of March shows the highest waste generation which leads to increased production of CNG which is approximately equal to 1800 kg. The readings taken were used for the ratio of CNG produced (kg) per kg waste and are plotted in graphical form against the atmospheric temperature. Figure 5 shows the effect of temperature variation on the ratio of the amount of purified CNG produced to used MSW. It can be concluded that there is a seasonal effect on the purified CNG produced via V-PSA, which increases with an increase in the temperature from winter to summer season.

Fig. 3 Weekly variation of net waste and rejected waste

Fig. 4 Effect of used waste on CNG produced

462

D. Patil et al.

Fig. 5 Effect of temperature on purified CNG per MSW

4 Conclusion With the change in season, the quality of biogas is altered. As temperature increases, quality of biogas increases, thereby quality of bio-CNG. The quality of biogas affects the VPSA purification process. It can be concluded that seasonal effect occurs on the purification system and bio-CNG produced per kg of waste increases as temperature increases from winter to summer. Acknowledgements Special thanks to Mailhem Ikos Environment Pvt. Ltd. which is a global waste management conglomerate offering customized solutions in solid waste treatment and management. The plant is a joint venture between two leading waste management firms Mailhem Engineers Pvt. Ltd., India, and Lhotellier Ikos Group, France. All the guidance and mentoring during the tenure of project was provided by research and development department.

References 1. Kadam R, Panwar NL (2017) Recent advancement in biogas enrichment and its applications. Elsevier 2. Overview of Biogas in India. https://www.globalmethane.org/documents/16.Ag2.1.Dhussa. pdf. Accessed 04.09.19 3. Rupnar AK, Jain S, Panwar NL (2018) Biogas in India: potential and integration into present energy systems. Int J Curr Microbiol Appl Sci 07(07):2175–2186 4. CPCB India (2012) Central pollution control board. Government of India 5. Ministry of New and Renewable Energy [National Biogas and Manure Management Programme (NBMMP)]. https://mnre.gov.in/biogas. Accessed 25.07.19 6. Angelidakia I, Treua L, Tsapekosa P, Luoc G, Campanarob S, Wenzeld H, Kougiasa PG (2018) Biogas upgrading and utilization: current status and perspectives. Elsevier 7. Prussi M, Padella M, Conton M, Postma ED, Lonza L (2019) Review of technologies for Biomethane production and assessment of Eu transport share in 2030. Elsevier 8. Ong MD, Williams RB, Kaffka SR (2014) DRAFT comparative assessment of technology options for biogas clean-up. In: Public interest energy research (PIER) program draft interim project report, Oct 2014 9. Olugasa TT, Oyesile OA (2015) Design and construction of a water scrubber for the upgrading of biogas. J Fundam Renew Energy Appl 05(05):1–6

Study and Analysis of Various Parameters …

463

10. Magomnang AASM, Maglinao AL, Capareda SC, Villanueva EP (2018) Evaluating the system performance of a pressure swing adsorption (PSA) unit by removing the carbon dioxide from biogas. Indian J Sci Technol 11(17):1–17 11. Shukla PV, Bhalerao TS, Ingle ST (2010) Effect of seasonal variation on biogas production from different food waste. J Pure Appl Microbiol Res Gate 4(1):333–337 12. Ramaraj R, Unpaprom Y (2016) Effect of temperature on the performance of biogas production from duckweed. Chem Res J 1(1):58–66

Robust Sliding Mode Controller (RSMC) for an Omniwheeled Mobile Robot with Uncertainties and External Perturbations Mohammad Saad , Uddesh Tople , Amrapali Khandare , and Zeeshan Ul Islam

Abstract Mobile robots at present are used extensively in the industrial and commercial sectors. There is a need for a robot that can easily maneuver in the sparse space at the warehouse, stores, etc. Omnidirectional robots have an upper hand over conventional mobile robots as they need not change their orientation while maneuvering. In this paper, we are putting forth the design of robust controller for omniwheeled mobile robot (OWMR). In the beginning, kinematic and dynamic modeling of OWMR was done; then, PID and sliding mode controller (SMC) were implemented on the OWMR. The proposed controller tracks the trajectory with greater accuracy as compared to other standard controllers. Keywords Control methods · Robust controller · Sliding mode controller · Mobile robotics · Trajectory tracing · Omni wheels

1 Introduction Omnidirectional mobile robots are capable of maneuvering in all directions without changing its orientation on a flat surface. One of the omnidirectional mobile robot is OWMR. Omni wheels consist of small rollers on the circumference which are perpendicular to the rotational direction of the wheel [1]. The OWMR is a squareshaped robot with omni wheels mounted at the four corners of the square base making M. Saad · U. Tople (B) · A. Khandare · Z. Ul Islam Visvesvaraya National Institute of Technology, Nagpur, India e-mail: [email protected] M. Saad e-mail: [email protected] A. Khandare e-mail: [email protected] Z. Ul Islam e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_55

465

466

M. Saad et al.

45◦ with the side of the base (Fig. 1). The adjacent wheels are kept perpendicular to each other, while opposite wheels are made parallel to each other. Kinematic and dynamic modeling of OWMR was done by taking voltage as an input to motors, whereas most of the previous researches were focused on the torquebased approach. Kinematical modeling [2, 3] was done considering different frames, and transformation matrices were used to obtain the kinematic equations in the world frame. Dynamic modeling was done by using the Newton–Euler method taking into account all external perturbations and uncertainties. The dynamic model of the robot is nonlinear. As the external disturbances and uncertainties are bounded but unknown, the standard linear controllers like PID cannot track the trajectory accurately. In this case, sliding mode controller [4] is very effective and gives much better results in trajectory tracking. It consists of nominal as well as the discontinuous controller. The nominal part of the controller is used to achieve the trajectory, while discontinuous part of the controller makes the OWMR robust [5]. The paper has been divided into six sections. We have dealt with the kinematic model of the OWMR in Sect. 2. In Sect. 3, we have discussed the dynamic model of OWMR. Section 4 focuses on the controller design for OWMR. Behavior of OWMR while using standard controllers and SMC has been shown in Sect. 5 with simulation results. The paper will end with the conclusion in Sect. 6.

2 Kinematic Modeling Figure 1 shows the top view of OWMR. The system is studied from six different frames. Owi X wi Ywi represents the frame assigned to four wheels of the robot, where i is referred to the ith no. of wheel. Robot moving frame attached with robot platform is represented by Or X r Yr , and the world frame is represented by Oq X q Yq . Pr = [xr yr φr ]T is the position vector of the robot in Or , where xr and yr are used to represent the position of the robot and φr represents the rotational orientation of robot [6]. Pq = [xq yq φq ]T is the position vector of the robot in Oq . P˙ r = [x˙r y˙r φ˙ r ]T represents the velocity vector in Or . The velocity vector Pr in terms of [θ˙1 θ˙2 θ˙3 θ˙4 ]T , where θ˙i represents the velocity of the i th wheel [2], is given by: ⎡ ⎤ ⎡ ⎤ ⎤ θ˙1 ⎡ x˙r 0 1 0 −1 ⎢ ˙ ⎥ ⎣ y˙r ⎦ = R ⎣ 1 0 −1 0 ⎦ ⎢θ2 ⎥ (1) 2 −1 −1 −1 −1 ⎣θ˙3 ⎦ φ˙ r 2a 2a 2a 2a θ˙4

Robust Sliding Mode Controller (RSMC) for an Omniwheeled Mobile Robot …

467

Fig. 1 Schematical top view of mobile robot

3 Dynamic Modeling The free-body diagram of all the forces acting on the robot is shown in the Fig. 2. Fi represents the driving force due to the actuator on the ith wheel of the robot [7]. It is given as Fi = αu i − β R θ˙i . α and β are coefficients of motors and are calculated using datasheet of motor and u i is the controller input. τ is the net torque due to all the forces acting on the robot. Fex is the unknown external force acting on the robot at an angle ψ with Yr at height b from top corner of the robot. Applying Newton’s second law of translational and rotational motion, i.e., F = Ma and τ = I α in Or q q and using rotational transformation matrix [6, 8] P˙ q = Rr (φ)P˙ r and Fq = Rr (φ)Fr and rearranging , we get ˙ 3X 1 + g(φ)3X 4 u(t)4X 1 + h(Fex , φ, ψ)3X 1 + ξ(t)3X 1 x(t) ¨ 3X 1 = f (x) where ⎡ 1 ⎤ − M Sφ − M1 Cφ M1 Sφ K 1 x˙q + K 2 y˙q ⎢ f (x) ˙ = ⎣ K 3 x˙q + K 4 y˙q ⎦ , g(φ) = α ⎣ M1 Cφ − M1 Sφ − M1 Cφ a a a K 5 φ˙ q ⎡

I

⎡ ⎢ h(Fe x, φ, ψ) = ⎣

1 F S(ψ − φ) M ex 1 F C(ψ − φ) M ex 1 F (bC(ψ) I ex

− (b − a)S(ψ))

I

I

⎤ 1 Cφ M ⎥ 1 Sφ ⎦ , M a I

⎤ ⎥ T ⎦ , ξ(t) = [ξx ξ y ξφ ]

(2)

468

M. Saad et al.

Fig. 2 Free-body diagram of the OWMR

⎡

Cφ Sφ Rrq (φ) = ⎣−Sφ Cφ 0 0

⎤ 0 0⎦ 1

where C(φ) = cos(φ), S(φ) = sin(φ), C(ψ) = cos(ψ) and S(ψ) = sin(ψ) K1 =

1 ((2β M

K2 = K3 = K4 =

− Bx )C 2 (φ) − (2β + B y )S 2 (φ))

1 (4β M

1 ((2β M

− Bx + B y )CφSφ

− Bx )S 2 (φ) − (2β + B y )C 2 (φ))

K 5 = 1I (4βa 2 − Bz ) and Bx , B y and Bz are linear coefficients of friction in x-, y- and z-directions, respectively, and ξ(t) is the uncertainty in the environment.

4 Controller Design The main objective for controller design is to track the desired trajectory in the presence of bounded but unknown external perturbations and uncertainties. OWMR is a system of the type multiple inputs multiple outputs (MIMO). In this case, robust nonlinear controller is appropriate rather than the standard linear controller.

Robust Sliding Mode Controller (RSMC) for an Omniwheeled Mobile Robot …

469

Sliding Mode Controller: Let Pd (t) = [xqd yqd φqd ]T is the vector of desired ˆ Pq (t) − Pd (t). Let s be the sliding trajectory. Pˆ is the error vector given by P= d n−1 ˆ surface defined by s = ( dt + λ) P. The value of n = 2, as we are considering till second derivative of P [3], i.e. P¨ ˆ˙ + λ P(t) ˆ s = P(t)

(3)

Equation of the best estimated continuous or nominal controller (uˆ nom ) is derived by equating the first derivative of Eq. (3) with zero and is given by ˆ˙ ¨ˆ + λ P(t) s˙ = P(t) =0 ˆ˙ uˆ nom = g −1 (φ)[x¨qd − f (x) ˙ − h(Fex , φ, ψ) − ξ(t) − λ P(t)]

(4) (5)

whereas equation of discontinuous part of controller (uˆ dis ) is given by uˆ dis = −g −1 (φ)[k sgn(s)].

(6)

where sgn(s)=1 if s ≥ 0 and sgn(s) = −1 if s ≤ 0, k is switching gain and is given by k ≥ F + η. From (2), f(x) ˙ + h(Fex , φ, ψ) + ξ(t) is the dynamics of the OWMR ˆ + ξ(t) ˆ be the dynamics of the ˆ Fˆex , φ, ψ) with unknown uncertainties, let fˆ(x) ˙ + h( OWMR with best estimated values of uncertainties. ˆ − ξ(t)| ˆ ˆ Fˆex , φ, ψ) ˙ − h( and η is the constant F = | f (x) ˙ + h(Fex , φ, ψ) + ξ(t) − fˆ(x) which is inversely proportional to reaching time from initial position to the desired position [4], and hence, selecting proper values for η is very important. Using (5) and (6), the final equation for robust sliding mode controller is ˆ˙ − ksgn(s)] (7) ˙ − h(Fex , φ, ψ) − ξ(t) − λ P(t) u = g −1 (φ)[x¨qd − f (x)

5 Simulation Results To verify the efficacy and accuracy of the proposed controller, trajectory tracking was performed. The physical dimension and parameter values for the robot were taken as mass of OWMR is M = 5 kg, moment of inertia Iq = 0.0806 kgm2 , friction force on the wheels Bx = B y = Bz = 0.2 Ns/m, motor coefficients α = 0.087 N/V and β = 10 kg/s,length of the robot a = 0.22 m, radius of wheel R = 0.05 m. The results of error as deviation from the trajectory are quantified in the form of integral square  error (ISE =  2 dt) , integral average error (IAE = | |dt), integral time average Error (ITAE = t| |dt), where is error magnitude and t is time. (Table 1). The smaller values of ISE, IAE and ITAE for RSMC indicate that the controller has very

470

M. Saad et al.

Table 1 Tracking performance comparison Controller height ISE (m)

RSMC

PID



ex ey





ex ey



0.000499 0.0184

2.2591 17.8358





IAE (m)

ITAE (m)



0.2415 1.7724







17.27 50.554



23.109 169.4

1660 4850





small response time to external perturbations and disturbances. The equation of the trajectory ∀t ∈ (0s, 190s) xqd =

3 sin(t/30) 9 sin(t/30) cos(t/30) yqd = φqd = 0 1 + cos2 (t/30) 1 + cos2 (t/30)

(8)

and the bounded but unknown uncertainties are given as ξx (t), ξ y (t), ξz (t) ∈ (−1, 1), , π ) and b ∈ (0, a). The parameters of the SMC controller is Fex ∈ (−3, 3), ψ ∈ ( −π 2 2 obtained ⎡ after ⎤ repetitive estimation, ⎡ ⎤ and the one giving most robust behaviors is 500 300 λ = ⎣0 5 0 ⎦ η =⎣0 4 0⎦ 005 003 The initial position and orientation of the OWMR are [xn yn φn ] = [0 0 0]T . Trajectory tracking of robot with SMC and PID is shown in Fig. 3. It was observed that the trajectory tracking when SMC is applied on the robot, is far more accurate than that of the trajectory tracking by PID. Figures 4 and 5 show the error in trajectory tracking of robot in x- and y-directions with respect to time. The error in x- and ydirection using SMC is almost zero, while error in PID is more prominent [9]. 4

smc pid desired

3 2

metres

1 0 -1 -2 -3 -4 -4

-3

-2

-1

0

metres

Fig. 3 Trajectory tracing plot

1

2

3

4

Fig. 4 Plot of error in x versus time

error(metres)

Robust Sliding Mode Controller (RSMC) for an Omniwheeled Mobile Robot … 0.25 0.2 0.15 0.1 0.05 0 -0.05 -0.1 -0.15 -0.2 -0.25

471 smc-x pid-x

0

20

40

60

80

100

120

140

160

180

200

Fig. 5 Plot of error in y versus time

error(metres)

time(s) 0.3 0.2 0.1 0 -0.1 -0.2 -0.3 -0.4 -0.5 -0.6 -0.7

smc-y pid-y

0

20

40

60

80

100

120

140

160

180

200

time(s)

6 Conclusion and Future Aims In this paper, a robust sliding mode controller for OWMR has been proposed. Kinematic equations are developed. Dynamic equations are derived in the presence of unknown external perturbations and uncertainties. The second-order sliding mode control law is developed considering external disturbances and uncertainties. The proposed controller efficacy has been tested for a reference trajectory. The simulation result proves that the proposed controller excels in comparison with the PID controller. Addition of static and dynamic boundary layer in SMC can be considered for improving the performance of SMC by trading off with accuracy. Adaptive control law can also be developed for unbounded uncertainties.

References 1. Oliveira HP, Sousa AJ, Paulo Moreira A, Costa PJ (2008) Precise modeling of a four wheeled omni-directional robot. Proc Robot 2. Muir PF, Neuman CP (1986) Kinematic modeling of wheeled mobile robot, Carnegie Mellon University 3. Muir PF, Neuman CP (1990) Kinematic modeling for feedback control of an omnidirectional wheeled mobile robot. In: Cox IJ, Wilfong GT (eds) Autonomous robots vehicles. Springer, New York, NY 4. Edwards C, Spurgeon SK (1998) Sliding mode control theory and application, CRC Press Taylor and Francis Group 5. Slotine J-JE, Li W (1991) Applied non linear control, Prentice Hall 6. Mittal RK, Nagrath IJ (2015) Robotics and control, Tata Mcgraw Hill Publication

472

M. Saad et al.

7. Alakshendra V, Chiddarwar S (2016) Adaptive robust control of Mecanum wheeled mobile robot with uncertainties. Non Linear Dyn Springer 8. Muir PF, Neuman CP Kinematic modeling for feedback control of an omnidirectional wheeled mobile robot. In: 1987 IEEE international conference on robotics and automation (ICRA) 9. Control Tutorials for MATLAB and Simulink by University of Michigan, Carnegie Mellon University, University of Detroit Mercy

The CFD Analysis of Convection Heat Transfer with Magnetic Field in the 2D Domain Using OpenFOAM Ranjit J. Singh and Trushar B. Gohil

Abstract The present numerical study focused on the enhancement and regulation in the heat transfer with bifurcation in the flow along with improvement in the average temperature (T avgE ) of enclosure with the application of magnetic field. The magnetic field-based in-house solver is developed using open-source CFD toolkit OpenFOAM. The electric potential formulation with Boussinesq approximation is employed in the present solver to analyze the buoyancy-driven natural convection flow with the magnetic field. The buoyancy force is constant in the fluid by maintaining the Rayleigh number of 106 . The impact of the force of magnetic field on the fluid stream and heat transfer rate is reported. The magnetic field is applied in terms of Hartmann number of Ha = 0, 10, 25, 50, 75, and 100. It is noticed that the intensity adjusts the strength and orientation of Lorentz force in the domain and alters the flow pattern as well as regulates the heat transfer. The detail discussion on the impact of the magnetic field on the isotherms, streamlines, and the run-time average Nusselt number is reported. Keywords OpenFOAM · Magnetic field · Lorentz force · Natural convection

1 Introduction The natural convective heat and fluid stream inside the closed cavity is one of the important areas of the industrial application such as cooling of electronic devices, heat dissipation in lubrication system, heat exchangers in nuclear industries, solar energy receiver, solar desalination method, melting and solidification of molten metal [1–4]. The fluid motion is brought in the cavity by the density variation as the consequence of the temperature difference in the domain. The MHD is one of the attentive R. J. Singh · T. B. Gohil (B) Mechanical Engineering Department, Visvesvaraya National Institute of Technology Nagpur, South Ambazari Road, Nagpur, Maharashtra 440010, India e-mail: [email protected] R. J. Singh e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_56

473

474

R. J. Singh and T. B. Gohil

applications in the case of natural convection flow, as it assists in regulating and suppressing the flow features and heat transfer performance in the cavity even at same Ra [5]. Thus, the application of additional magnetic force on the system retards the buoyancy force which reduces the heat transfer [6]. The alignment of the magnetic field on the system also plays a significant role in the convectional roll formation, which further brings more unsteadiness in the fluid [7, 8]. In the case of unrestricted convection flow with the exposure of the magnetic field, the above literature shows that that the presence of Lorentz force refrained the heat transfer and the flow of fluid in the domain. However, the present study reports the favorable role of the existence of Lorentz force for boosting the heat transfer and regulating the fluid flow in the cavity at low strength of magnetic field. In this numerical analysis, in-house developed code on the open-source CFD toolkit OpenFOAM is used for MHD-based buoyancy driven. The detail discussion of relevant quantities such as run-time average Nusselt number, local Nusselt number, streamlines, isotherms, the average temperature in the enclosure (T avgE ), Lorentz force strength and its direction in the domain is reported.

2 Mathematical Formulation and Numerical Scheme The present solver couples the Navier–Stokes equation with Poisson’s electric potential equation and energy equation to obtain the velocity distribution, temperature variation, and Lorentz force in the domain. The Lorentz force and Boussinesq approximation are added in the Navier–Stokes equation as a source term to show its influence on the flow physiognomies. The fluid is supposed to be incompressible, viscous, and electrically conducting in nature. The complete set of the partial differential equation is taken from reference [6]. The non-dimensional parameters incorporated in the solver that govern the fluid flow under the influence of the magnetic field are: 

σ Ra = Gr · Pr μ gβ(T − Tref )(2L)3 ν Gr = Pr = ν2 α

Ha = B(2L)

where Ha is the Hartmann number, which measures the power of the magnetic field imposed on the system. Ra is the Rayleigh number, Gr is the Grashof number, and Pr is the Prandtl number. The characteristic length (2L) is considered as a bottom length of the cavity. Direct numerical simulation (DNS) is applied in OpenFOAM tool to solve the complete sets of Eqs. (1–6) [6]. Collocated finite volume method is used to solve all terms in the equations. The PIMPLE algorithm is set in the solver to solve the entire sets of the equation with two outer correctors and pressure correctors which assist in refining convergence of velocity and pressure field at every interval.

The CFD Analysis of Convection Heat Transfer …

475

The Rhie and Chow interpolation scheme [9] is used to couple the pressure and velocity. The first-order accurate implicit Euler scheme is supposed to discretize the time derivate terms. The second-order central difference method is implemented to discretize the convection and diffusion terms of the governing equations.

3 Problem Definition and Grid Independence Test The gradually constricted cavity from the top and the flat bottom surface is used as a computational domain. The characteristic length is taken as the base length of the surface (2L), and the opposite vertical adiabatic surface is of length (L). The geometry details, coordinates system, and mesh distribution in the present study are shown in Fig. 1. All computations are performed at a constant Rayleigh number of Ra = 106 and at Pr = 0.71 [10]. The grid independence study is performed on the three different sizes, Grid 1 = 100 × 200, Grid 2 = 125 × 250, and Grid 3 = 150 × 300, at Ra = 106 and Ha = 50. The average temperature of the enclosure (T avgE ) is calculated for all three grids as shown in Fig. 2b. The T avgE variation for all grids overlaps with each other; therefore, Grid 2 is fine enough to capture the flow physics near the boundary layers and in the central core region. Hence, Grid 2 is adopted for the entire simulation in the present study.

4 Results and Discussion The validation test case is considered for two-dimensional MHD-based natural convection flow in the rectangular domain and obtained data is compared with reference [6, 11]. The geometrical size, aspect ratio, and boundary conditions are maintained as mentioned in the given reference [6, 11]. The streamwise normalized velocity (U x /U x(max) ) obtained in the present case is compared with the result mentioned in the reference [6, 11] as shown in Fig. 2a, and the comparison trend shows a good →

→

→

agreement with published data [6, 11]. The Lorentz force ( F = j × B ) is the consequence of the interaction among the working fluids with the magnetic field. Due

Fig. 1 a Geometrical details of constricted cavity

476

R. J. Singh and T. B. Gohil

Fig. 2 a Validation of the result obtained from present solver with reference [6, 11], b the grid independence test

to the heated bottom surface, fluid movement triggered and formation of convection rolls in the domain are observed. Now, the fluid interacts with the magnetic field (B) to produce electric potential and subsequently electric current. In the two-dimensional computational domain, fluid flows in x-direction (U x ) and y-direction (U y ). The imposed magnetic field  → → is in x (Bx ). Hence, the electric potential (∂φ ∂z ≈ U × B ) and electric current →

( j ) are developed in the z-direction [6–8]. However, the present analysis is twoz

dimensional; the distribution of an electric current and an electric potential variation in the cavity is not available in this study. The orientation of Lorentz force is induced in the reverse direction of the flow of the fluids, which affects the flow pattern and hence heat transfer. Figure 3 shows the streamlines variations in the enclosure for the different intensity of the magnetic field applied in the direction parallel (Bx ) to the heated surface. The streamlines for non-MHD free convection flow (Ha = 0) in the

Fig. 3 Streamlines variation for various Ha

The CFD Analysis of Convection Heat Transfer …

477

Fig. 4 Isotherms variations at various Ha

enclosure has two convection rolls and two primary vortices at the top and bottom parts of the cavity as shown in Fig. 3a. The sharp constriction of the top surface of the enclosure is responsible for the flow separation in the top and lowest curves of the cavity. Hence, the fluid gets cooled in the mid-span of the top surface, and after that, it starts gravitating leaving eddies behind at top corners of cavity. Similarly, at the bottom corners of enclosure, eddies are generated for non-MHD flow (Ha = 0) as shown in Fig. 3a. The vortices at the bottom corners of the cavity are first eliminated with the application of magnetic field at Ha = 10. Although, the strength of Lorentz force is not sufficient to cop with the primary vortices at the top corners as shown in Fig. 3b–d. As the intensity of the magnetic field (Ha = 50) increased, secondary vortices near the top surfaces start emerging; this is due to the dissimilar orientation of the Lorentz force with respect to the heated surface. The dissimilar variation of Lorentz force suppresses the fluid toward the bottom surface and pulls the fluids toward the top corners. The increase in the intensity of the magnetic field (Ha = 75 and 100), two convectional rolls are formed and occupy the entire domain without any traces of smaller vortices as shown in Fig. 3e, f. Figure 4 shows the isotherms variation in the enclosure. It is observed from Fig. 4a for non-MHD flow; the plume is developed from the center of the cavity and bifurcated into two equal halves from the top converged point. Consequently, for Ha = 0, the interaction area for the cooling is lesser as the fluid becomes denser and starts gravitating leaving the cold surface without occupying the entire upward slope. However, the plume formation is shifted near the adiabatic walls as soon as the magnetic field is imposed on the system (Fig. 4), which is observed from the streamline plots (Fig. 3). The shift in the position of plume generation increases the heat dissipation area. When the hot fluid started following the cold surface and gravitates along with the downward slope of the wall, it increases the contact area for cooling as shown in Fig. 4b–f. The furthermore increase in the magnetic strength (Ha = 50 − 100), the fluid is occupying the entire cavity into two equal halves without any traces of eddies as observed in the streamlines plots (Fig. 3e, f). Hence, the average heat dissipation area of fluid gets increased. However, the flow velocity of the fluid in the domain

478

R. J. Singh and T. B. Gohil

Fig. 5 Times-averaged Nu variation for various Ha

is highly suppressed by the presence of the magnetic field; hence, the rate of heat absorption and dissipation gets retarded at the higher Ha. Figure 5 illustrates the run-time average Nusselt number (N avg ) variation over the hot surface. It has been observed that the Nuavg upsurge at Ha = 10 compared to Ha = 0 (Fig. 5a); this is because, at the small intensity of Ha, the Lorentz force assists the fluid to dissipate heat in the entire area of cavity as the buoyancy force is dominant over viscous force and Lorentz force. The flow of fluid becomes unsteady at Ha = 25 (Fig. 5b); this is happening because the Lorentz force starts dominating over buoyancy force and it tries to pull the fluid according to its nature of flow and direction. The unsteady behavior at Ha = 25 keeps the Nuavg same as to Ha = 0. Further, with the increase in the Bx magnetic field, Nuavg begins to drop down as the strength of the induced Lorentz force in the Bx magnetic field increases with the rise in the Ha and has direct opposing nature to buoyancy force which suppress the flow motion and heat transfer as shown in the Fig. 5c. For these cases, all reported results are time-averaged.

5 Conclusions The numerical analysis of magnetic and non-magnetic-based convection flow is performed in the constricted rectangular cavity heated at the bottom part, and the top constricted surface is maintained cold. The buoyancy force is kept constant by keeping Ra = 106 . The non-MHD convection flow (Ha = 0) has two primary convection rolls and two eddies at the top and bottom corners of the cavity. It has been perceived from the results that the secondary flows (eddies) are eliminated first when the magnetic field is imposed on the system. The Bx magnetic field produces the Lorentz force acting parallel to the gravitational force; hence, the convection flow and Nuavg are severely suppressed under this regime at higher Ha. However, the low intensity of magnetic field is found to be favorable for the regulation of the heat transfer by maintaining the unsteady in the flow at Ha = 10.

The CFD Analysis of Convection Heat Transfer …

479

References 1. Chiang H, Kleinstreuer C (1991) Analysis of passive cooling in a vertical finite channel using a falling liquid-film and buoyancy-induced gas vapour flow. Int J Heat Mass Transfer 34:2389– 2394 2. Payvar P (1991) Laminar heat transfer in the oil groove of a wet clutch. Int J Heat Mass Transfer 34:1791–1798 3. Dayem AMA (2006) Experimental and numerical performance of a multi-effect condensation– evaporation solar water distillation system. Energy 31:2710–2727 4. Das D, Roy M, Basak T (2017) Studies on natural convection within enclosures of various (non-square) shapes—a review. Int J Heat Mass Transfer 106:356–406 5. Gajbhiye NL, Eswaran V (2015) Numerical simulation of MHD flow and heat transfer in a rectangular and smoothly constricted enclosure. Int J Heat Mass Transfer 83:441–449 6. Singh RJ, Gohil TB (2019) The numerical analysis on the development of Lorentz force and its directional effect on the suppression of buoyancy-driven flow and heat transfer using OpenFOAM. Comput Fluids 179:476–489 7. Singh RJ, Gohil TB (2019) Influence of the presence of the lorentz force and its direction on the suppression of secondary flow in two different orifices: a numerical study using OpenFOAM. J Appl Mech 12(3):751–762 8. Singh RJ, Gohil TB (2019) The numerical analysis on the variation of electric potential, electric current and Lorentz force with its influence on buoyancy-driven conjugate heat transfer and fluid flow using OpenFOAM. Fusion Eng Des 148:111300 9. Ferziger JH, Peric M (2002) Computational methods for fluid dynamics, 3rd edn. Springer, Berlin, pp 247–251 10. Chatterjee D, Mondal B, Halder P (2014) Mixed convective transport in a vertical lid-driven cavity including a heat conducting rotating circular cylinder. Numer Heat Transfer Part A 65:48–65 11. Garandet JP, Alboussiere T, Moreau R (1992) Buoyancy driven convection in a rectangular enclosure with a transverse magnetic field. Int J Heat Mass Transfer 35:741–748

Design of a Remote Racking Module for Racking Operation Alex Sherjy Syriac

and M. R. Rahul

Abstract Circuit breaker is a device used to disrupt the flow of current when an abnormal condition occurs in switchgear. Withdrawal-type circuit breakers have the provision for displacing the circuit breaker in and out of the switchgear compartment while inspection and maintenance operations. This paper discusses a new mechanism for displacing the circuit breaker during racking operation by stationing the operator at a remote location. The first part deals with the design of lead screw and variation of various thread parameters on the stress distribution and performance. The later part of the paper deals with the development and analysis of the novel racking mechanism. Keywords Circuit breaker · Racking mechanism · Lead screw

1 Introduction Locations such as industries, offshore oil platform, refineries, and hospitals employ multiple circuit breakers, typically located within a circuit breaker cabinet. The installation and removal of circuit breaker involves the engagement and disengagement of contacts of the contact breaker with a power bus within the circuit breaker cabinet. Conventionally, this installation and removal is performed manually by a technician. One of the main issues faced by the operator working in the electrical industry is the sudden explosion of high energy arc due to proper methods of engaging of the breaker, handling of tools, etc. During the service and inspection period, these circuit breakers are often dislodged from the electrical system. The manual racking action necessitates that the operator stands directly in front of the breaker. As the breaker is methodically extracted from the bus, several potential problems may occur. This creates a vulnerable surrounding to the operator engaging with the breaker. As a safety precaution, the electrician often wears equipment to protect oneself from fatal injuries while working with the electrical breaker. This heavy equipment often A. S. Syriac (B) · M. R. Rahul Visvesvaraya National Institute of Technology, South Ambazari Rd., Ambazari, Nagpur, Maharashtra 440010, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_57

481

482

A. S. Syriac and M. R. Rahul

reduces the dexterity and hinders the movement of the operator. The personnel are advised to keep a minimum distance with working with the breaker. The most easy and feasible solution is to keep the worker out of the flash boundary around the breaker by maintaining a safe distance. These devices pose advantages over manual operation like protects operator, creates a safe workplace around the breaker, reduces errors due to manual racking, and preserves integrity of interlocks, and self-stopping system helps to stop abnormal conditions. This apparatus allows the insertion and withdrawal of electric components while the operator is outside the safe boundary. They are designed specifically to remove the personnel from proximity to the circuit breaker in operation. The device provides a portable, motorized device that can be hand-carried to the work location and affixed to the circuit breaker without the need for any modification to the breaker or its enclosure. Hawkins et al. [1] designed a remote racking module along with a structural member for mounting. Bower et al. [2] and Vaill et al. [3] developed a mechanism that interfaces with the interlock system in switchgear as well as assembly for fixture thereby facilitating displacing of circuit breaker. Varanasi et al. [4] designed a model for a system consisting of a lead screw taking into the inertia as well as damping effects.

2 Study and Analysis on the Lead Screw of Withdrawal Part of the Circuit Breaker Withdrawal-type circuit breakers for medium voltage application usually are placed over a withdrawal part. The circuit breaker moves with the help of a lead screw inside the withdrawal part. The vital element in transmission of power from the remote racking device to the breaker for displacement is the lead screw. This session discusses the lead screw design and the effect of thread parameter on the stress and deformation [5]. As per the various forces acting on the lead screw, the lead screw is designed as shown in Table 1. The various thread parameters like thread pitch, nominal diameter, etc., play an important role in power transmission. The following section discusses the influence of thread pitch on the stress distribution on the screw and the frequency of the system. The study is done by considering square thread profile. Table 1 Thread parameters of the lead screw designed

Parameters of the lead screw drive

Value

Units

Material of the lead screw

Structural steel

–

Root diameter of screw d c

26

mm

Pitch diameter of screw p

3.5

mm

Nominal diameter of screw d

30

mm

Screw length

370

mm

Design of a Remote Racking Module for Racking Operation

483

Fig. 1 Effect of thread pitch of square thread profile on stress distribution and deformation

Fig. 2 Stress and displacement plot of square profile lead screw of pitch 3.4 mm at position of nut 102 mm from front end

2.1 Influence of Thread Pitch Variation on Stress Distribution The stress distribution is evaluated for the lead screw for different thread profiles such as square, acme, and trapezoidal with pitch ranging from 2.5 to 5.5 mm. The influence of pitch variation on stress values at various positions of the nut during its travel on the square profile lead screw is shown in Fig. 1. The stress plot and deformation plot for the square profile lead screw of pitch 3.5 mm are shown in Fig. 2. It is observed that stress in the lead screw decreases as the position of nut moves toward the center. Later, there is a slight increase in the stress value as the nut further travels to the other end. Another observation is that the variation of stress values is less for all pitch when the nut is at the center of the lead screw.

2.2 Influence of the Screw Pitch on the Natural Frequency The effect of thread pitch on the system’s frequency is studied by considering the square thread profile of pitch varying from 2.5 to 6.5 mm. It is observed that the

484

a

A. S. Syriac and M. R. Rahul

b

Fig. 3 Dynamics analysis of square profile lead screw of pitch 3.5 mm a Mode 5 and b Mode 6

system’s frequency is not affected much by the change in thread pitch. The trend seen in all the different pitch considered is that there is a small drop in frequency as we increase pitch. Further, the value of frequency is seen to increase as we move to pitch 6.5 mm. The mode shape of the lead screw is shown in Fig. 3.

3 Racking Mechanism: A New Concept A new system has been developed to be used for rack-in and rack-out of the circuit breaker. The breaker placed inside a compartment should allow the racking-in and racking-out movements behind a closed door. This new racking truck brings many improvements to existing functions and integrates new features linked with the position of the circuit breaker. The main components of the mechanism are confined inside a box. This novel racking module brings many improvements to features linked to the present racking solutions. The basic system consists of a mechanism for mounting, a device to actuate racking and controller. The mounting mechanism is designed to fasten the racking device onto the panel of the breaker. This mechanism is developed in a way that no modifications are to be made on the front panel of the breaker to accept the device. The structure of the mounting mechanism is robust and shaped for easy handling by the user as shown in Fig. 4. The device can be engaged and dislodged by magnetic means. The provision Fig. 4 CAD model of the mounting mechanism

Design of a Remote Racking Module for Racking Operation

485

Fig. 5 Prototype of the proposed racking mechanism

for angular rotation of the device helps for precise mounting of the device in front of the breaker door. The main component of the system is the racking device which helps to actuate the racking screw of the withdrawal part of the circuit breaker for maneuvering of the circuit breaker forward and backward. The entire components are housed inside a box. The box is attached to the mounting mechanism for the operation. The rotation of the racking screw is done using the selected motor with the required torque. The exact torque for rotation is calculated to displace the breaker between the connected and disconnected position. The torque value selected must help in maintaining the integrity of the interlocks in the withdrawal device as well as overcome the weight of the breaker along with friction. Various safety devices are present which help to sense the overtorque condition. To keep the operator out of the arc flash boundary, the device is designed to control from outside the flash boundary by the means of a remote console module. The controller contains buttons to configure the racking device in the racking-in and raking-out operation. In case of any abnormal condition senses by the system, emergency button is present to disable the racking operation immediately. The proposed mechanism is shown in Fig. 5. To test the performance of the device for various stress induced in the system, vertical deformation due to the weight of the device, analysis of the device is carried out in ANSYS Workbench. The boundary conditions for the study are shown in the figure. The static analysis boundary conidtion is shown in Fig. 6a. Modal analysis of the remote racking device is carried out to study the various types of vibration induced in the system which operates as shown in Fig. 6b.

4 Conclusion This paper discusses a novel racking mechanism that allows the medium voltage circuit breaker to be racked in or out by using a remote racking device that can be operated from a remote location. This proposed method of racking stations the operator outside the arc flash boundary thereby protecting the operator from fatal

486

A. S. Syriac and M. R. Rahul

Fig. 6 a Racking mechanism device with boundary condition and b modal analysis of the device depicting Mode 1 (bending about Z-axis)

injuries. The mechanism consists of a motor to rotate the racking screw for displacing the breaker, unit for gripping the remote racking device as well to operate the device from a remote location.

References 1. Hawkins et al (2011) Circuit breaker racking apparatus, systems, and methods of using same. US patent 8, 553, 394, B2, 17 Mar 2011 2. Bower et al (2012) Remotely—operated switchgear racking device and a mounting method for the same. US patent 2012/0199450 A1, 8 Feb 2012 3. Vaill et al (2005) Racking device and power module therefor. US patent 7,019,230 B1, 3 Aug 2005 4. Varanasi KV, Nayfeh SA (2004) The dynamics of lead-screw drives: low-order modeling and experiments. J Dyn Syst Meas Control 126:388–396 5. Dragoni E (1994) Effect of thread pitch and frictional coefficient on the stress concentration in metric nut-bolt connections. J Offshore Mech Arct Eng 11:21–27

A Coupled Heat Transfer and Artificial Neural Network Based Model for Accelerated Direct Cooling of Steel Plate Sagar Dave , Sirshendu Chattopadhyay , and Deepak Gupta

Abstract This article proposes a novel approach for controlling the accelerated cooling of hot steel plates in the existing plate mill at AM/NS India, Hazira. Difference in top and bottom surface temperature or uneven temperature distribution at the exit of accelerated cooling equipment leads to the property variation and shape deformation. Therefore, the model aims to predict top and bottom surface temperature of the plate for better online operational control. The control algorithm is a fundamental heat transfer calculation coupled with statistical artificial neural networks (ANN). The model prediction shows an excellent agreement with the plant measured data and almost 90% of the plates are within error range of ±20 ◦ C. Keywords Accelerated cooling · Mist spray cooling · Plate mill · Heat transfer calculation · Artificial neural network

1 Introduction Thermo-mechanical controlled process (TMCP) is widely used in hot rolled mills to produce high strength and toughness in steel plates. TMCP is a combination of controlled rolling and accelerated cooling (ACC). During ACC, the plate needs to be cooled at high cooling rates to achieve the desired properties. However, in industrial scale, applying high cooling rate uniformly over the steel plate can be a challenge, specially, if the steel plate is wide and long. Non-uniform cooling leads to property variation and shape deformation. AM/NS India operates a state-of-the-art 5 m wide plate mill with accelerated direct cooling (ADCO) facility. ADCO is equipped with the horizontal slit-type air-assisted water sprayers for uniform cooling over the width of steel plates. Due to large number of operational parameters and various compositions of steel grades, it is very difficult S. Dave (B) · S. Chattopadhyay · D. Gupta Research and Development, ArcelorMittal Nippon Steel India Limited (AM/NS India), Hazira, Surat, Gujarat 394270, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_58

487

488

S. Dave et al.

to estimate the ADCO exit temperature manually. Therefore, a model is required in order to establish controlled cooling. The developed model must be accurate and fast for online implementation in the plant. The cooling intensity of any ACC equipment is decided by the surface heat transfer coefficient (HTC) or surface heat flux [6]. The HTC for water mist spray cooling of the hot plate principally depends on local spray mass flux density over the plate [1, 5, 7]. Therefore, the model must optimize the water flow rate and plate speed for particular steel grade and thickness which ultimately gives the uniform temperature distribution after cooling. Rest of this article is structured as follows: The detail of mist spray cooling configuration in ADCO machine is given in Sect. 2. The model development and prediction statistics are discussed in Sect. 3. Plant scale implementation and testing of the model are illustrated in Sect. 4. Finally, the work is concluded in Sect. 5.

2 Mist Spray Cooling Configuration The schematic diagram of ADCO machine at AM/NS India Hazira facility is shown in Fig. 1. It contains 6 modules of mist spray cooling arrangement (module 1–6). Each module has 4 top and 4 bottom air-assisted slit-type sprayers. Total length of each module is 4 m in which the sprayers are placed at equidistance of 1 m. Width of the sprayers is 5.1 m. Maximum water flow rate through a module is limited to 1000 m3 /h which is controlled by electro-pneumatic valves. The operating water pressure is 1.5 bar for module 1–6. Air pressure and flow in each module are set constant as 780 mm of water column and 55,000 ∗ 2 m3 /h, respectively. Apart from that one additional module (module 0) with high-pressure spray nozzles is installed for intense cooling of the plate. The maximum water flow rate through module 0 is the same as module 1–6. However, operating pressure is set as 7 bar for air-water mist spray. Three scan pyrometers are installed for plate surface temperature measurements before and after ADCO machines. The first pyrometer (P1) measures top surface temperature of the plate at ADCO entry. At ADCO exit, the second (P2) and the third (P3) pyrometers are positioned to measure the top and bottom surface temperature of the plate, respectively. The plate temperature measurement inside the cooling system

Fig. 1 Schematic diagram of ADCO machine

A Coupled Heat Transfer and Artificial Neural Network …

489

is very difficult due to radiation effect of high mass flux density mist spray and steam generation over the plate.

3 Model Development and Prediction Statistics Earlier modeling approach was to combine heat transfer calculation with the phase transformation model of steel [3, 4, 8]. However, for accurate prediction of phase fraction, the detailed characterization of each steel grade is required at the rolling temperature. Hence, instead of pure fundamental model, an hybrid model of the basic heat transfer calculation and artificial neural networks is proposed in the present study. The first principle heat transfer calculation without incorporating metallurgical phase transformation model forms a base of the model prediction, while ANNs help in compensating the error in basic heat transfer calculation and online implementation.

3.1 Heat Transfer Calculation For fast prediction, heat conduction in length and width direction is assumed much smaller than the thickness direction. Therefore, unsteady one-dimensional heat diffusion equation with the moving boundary condition is used to predict the top and bottom surface temperature of the plate.  ρC p

∂T ∂t

 =

∂ ∂z

  ∂T k + Q˙ ∂z

(1)

where T is the temperature (K), ρ is the density (kg/m3 ), C p is the specific heat (J/kg K) and k is the thermal conductivity of the plate (W/m K). Q˙ is the heat source due phase transformation (kJ/kg). It is assumed that density of the plate depends only on the chemistry, and it remains constant. The variation of thermal conductivity with respect to temperature is illustrated in Eq. (2) [2]. k = 59.92 − 0.0221T − 5.4 × 10−5 T 2 + 4.3 × 10−8 T 3 C p = (1 − α)(C p )Austenite + α(C p )Transformed Phase

(2) (3)

The specific heat of the plate varies significantly with the change in plate chemistry, temperature and phase transformation. Since the phase transformation model ˙ is not incorporated in the heat transfer calculation (neglecting the source term Q), transformed phase fraction (α in Eq. (3)) is assumed to follow sigmoidal curve from Ar3 to Ar1 (critical temperatures) (cf. Fig. 2a). C p values of ferrite, pearlite and bainite are assumed to be same. Figure 2b also shows the variation of specific heat for low-carbon steel in working temperature range [8].

490

S. Dave et al.

)

0.5

1400

Austenite Ferrite, Pearlite, Bainite 2 Surface Heat Flux (kW/m )

-x

Cp Value (J/kg K)

Transformed Phase Fraction (α)

1200 1 1000

800

600

0 Ar

1

Ar

3

Temperature (°C)

(a)

400 300

M0 Top M0 Bottom M1 to M6 Top M1 to M6 Bottom

1200

0.75

162*G

1000 800 120*G0.7 250*G0.3

600 400

140*G0.42

200 0

400

500

600

700

Temperature (°C)

(b)

800

900

0

2

4

6

8

10

12

16

14

Spray Mass Flux Density (G)(kg/m2s)

(c)

Fig. 2 Heat transfer calculation parameters a assumption for transformed phase fraction (for C p value), b specific heat variation for low-carbon steel [8], c heat flux profile for different modules

Equation (1) is discretized using finite difference method. Thermal profile of the plate critically depends on the initial and boundary condition applied for the numerical calculation. The plate temperature is assumed to be uniform at ADCO entry, and it is initialized with the pyrometer reading (P1). Total heat flux from the top and bottom surface of the plate is dominated by convection flux and radiation flux due to high mass flux density of water spray and working temperature range of ADCO machine. The moving boundary condition (changes with respect to time) applied in the present simulation is illustrated in Eq. (4).   ∂ T  ∂ T  4 4 4 = qtop ˙ + σ (Ts − T∞ ) and − k = qbottom ˙ + σ (Ts4 − T∞ ) −k ∂z top ∂z bottom (4) where Ts is the plate surface temperature (K), T∞ is temperature of water (K), σ is the Stefan–Boltzmann constant (5.67 × 10−8 W/m2 K4 ) and  is emissivity of the surface (as a function of Ts [4],  = ((T + 273)/1000)[(0.125(T + 273)/1000) − ˙ are the convective heat flux from the top surface and 0.38] + 1.1). qtop ˙ and qbottom bottom surface of the plate, respectively. For high impingement density and high surface temperature regime, the dominant parameter for calculating heat flux rate is the spray water flux density [1, 5, 7] due to Leidenfrost effect. Figure 2c shows the convective heat flux profile considered in the present study as a function of water mass flux density, and it is extracted from the last two-year plant data using data fitment and inverse calculation. For air cooling zones, the value of total heat flux can be determined using equation q˙air = h air (Ts − T∞ ). Empirical correlation is used for calculating overall heat transfer coefficient [3] which is depicted in Eq. (5). h air = 4.5u 0.926 +

−3038 + 0.935(Ts + 273) + 2.14 × 10−8 Ts4 (Ts + 273) − 20

(5)

A Coupled Heat Transfer and Artificial Neural Network … Present Model (top) Present Model (bottom) . . Q (top) Sun et al.(2002) without Sun et al.(2002) without Q (bottom) . Sun et al.(2002) .with Q (top) Sun et al.(2002) with Q (bottom)

950 850

15

800 750

10 5

700 650

x is compensated by ANN

600

x

550 0

1

2

3

4 5 Time (sec)

(a)

6

7

8

0 60

Temperature (°C)

900

TFCT Error BFCT Error

20

% Plates

1000

491

Error in Temperature Prediction (°C)

(b)

Fig. 3 Results of heat transfer calculation a transient calculation for R O T configuration [8], b error in temperature prediction

First, the developed model has been validated with the results of Sun et al. [8] for their R O T configuration (cf. Fig. 3a). In present ADC O setup, the model is dedicated to predict top finish cooling temperature (TFCT) and bottom finish cooling temperature (BFCT) at the ADC O exit. Total 5000 plates data which include broad range of thicknesses (10–60 mm) and more than 60 grades of the steel plate (0.05– 0.24 %C), are used to train the model. Results of unsteady heat transfer calculation are summarized in Fig. 3b. Temperature prediction error shown in Fig. 3b is closely related with the variation in plate chemistry. To compensate this error due to the phase transformation source term, ANN-based model (ANN1) is coupled with the present heat transfer calculation.

3.2 ANNs Training Multilayer perceptron (MLP) networks with three layers (input-output and one hidden layer) are trained to develop the coupled model by applying the second-order Broyden–Fletcher–Goldfarb–Shanno (BFGS) algorithm. Different combinations of nonlinear activation functions (like logistic, tangent hyperbolic, exponential, sine, etc.) and node weights are tuned during the learning process in order to improve the output prediction. The input parameters used for training and testing of the network (ANN1) are the plate chemistry (%C, %Mn, %Mo, %Nb), plate speed, length/width ratio and plate entry temperature. The prediction statistics of the coupled model are detailed in Fig. 4. A comparison of predicted top and bottom surface temperature with the pyrometers reading at ADC O exit is illustrated in Fig. 4a. The prediction error in TFCT and BFCT is less than ±20 ◦ C for 91.56% and 95.42% of the plates, respectively (cf. Fig. 4b). It shows that after coupling ANN with the basic heat transfer calculation, the results are in excellent agreement with the plant measurements.

S. Dave et al. 30

TFCT BFCT ± 20°C

700

TFCT Error BFCT Error

25

650

% Plates

600

20 15 10 5

550

0 500 500

550

600

650

700

50

Predicted Cooling Temperature (°C)

492

Error in Temperature Prediction (°C)

Measured Cooling Temperature (°C)

(a) Predicted vs Measured FCT

(b) Temperature prediction error

Fig. 4 Prediction statistics of the coupled model

4 Plant Scale Implementation and Testing Two more networks were trained for the online implementation of the model. ADC O entry temperature is predicted with the help of ANN2 as the model has to run before actual cooling starts for its dynamic performance. Apart from that there is inbuilt dynamic water control system to modify water flow rate for head, tail and body of the plate, and it is very difficult to use instantaneous flow for the heat transfer calculation. Therefore, ANN3 is introduced to predict averaged water flow rate per module using head and tail set points which are fixed by the operator. Input parameters for ANN2 and ANN3 are as follow: (1) ANN2: plate chemistry (%C, %Mn, %Mo, %Nb, %Ti), target thickness and finish rolling temperature (2) ANN3: plate speed, head flow rate, tail flow rate and module index.

Fig. 5 Model performance testing

60

TFCT BFCT

Deviation (°C)

40 20 0 -20 -40 -60 520

550

580

610

640

670

Measured Pyrometer Reading (°C)

700

A Coupled Heat Transfer and Artificial Neural Network …

493

After successful installation, the model has been tested for 300 plates for online operation. Figure 5 shows the deviation in TFCT and BFCT prediction from the measured pyrometer readings which is inline with the model claim.

5 Conclusion A hybrid model comprising fundamental heat transfer calculation and statistical artificial neural network has been developed and tested for the top and bottom surface temperature prediction of hot steel plate after accelerated cooling. It was found that combining the first principle heat transfer model with ANN model for estimating error due to variation in chemical composition of the steel has given better results. Prediction statistics of the coupled model indicates an excellent agreement with the plant measured data. The TFCT and BFCT prediction error fall within band of ±20 ◦ C for almost 91% and 95% of the plates respectively. The model online prediction after plant scale implementation is inline with the model claim which indicates the applicability of this type of hybrid and computationally efficient model in the steel industries.

References 1. Al-Ahmadi H, Yao S (2008) Spray cooling of high temperature metals using high mass flux industrial nozzles. Exp Heat Transf 21(1):38–54 2. Darken L, Gurry R (1953) Physical chemistry of metals, vol 582. McGraw-Hill book company, Inc., New York, Toronto, London 3. Edalatpour S, Saboonchi A, Hassanpour S (2011) Effect of phase transformation latent heat on prediction accuracy of strip laminar cooling. J Mater Process Technol 211(11):1776–1782 4. Han HN, Lee JK, Kim HJ, Jin YS (2002) A model for deformation, temperature and phase transformation behavior of steels on run-out table in hot strip mill. J Mater Process Technol 128(1–3):216–225 5. Hnizdil M, Chabicovsky M, Raudensky M, Lee TW (2016) Heat transfer during spray cooling of flat surfaces with water at large reynolds numbers. J Flow Control Measur Vis 4(03):104 6. Lee P, Raudensky M, Horsky J (2013) Development of accelerated cooling for new plate mill. Ironmaking Steelmaking 40(8):598–604 7. Liang G, Mudawar I (2017) Review of spray cooling-part 2: high temperature boiling regimes and quenching applications. Int J Heat Mass Transf 115:1206–1222 8. Sun C, Han H, Lee J, Jin Y, Hwang S (2002) A finite element model for the prediction of thermal and metallurgical behavior of strip on run-out-table in hot rolling. ISIJ Int 42(4):392–400

Effect of Air Distribution on Cooling of Photovoltaic Panel and Its Performance Someshwar S. Bhakre and Pravin D. Sawarkar

Abstract Increase in operating temperature of photovoltaic (PV) cells decreases its conversion efficiency and power output. In the present study, numerical simulations were carried out to find the proper design of diffuser so as to get uniform distribution of air along with the PV panel. Uniform distribution of air plays a vital role in decreasing thermal stresses which reduces the formation of hot spots, and ultimately it increases the life of the PV panel. The CFD results show that the new design of diffuser with three inner deflector plates and curved side walls maintain uniformity of air at inlet of PV panel. The experimental study with new diffuser indicates a significant reduction in the surface temperature of PV panel which results in increase in power output and efficiency. Keywords A diffuser · Uniform cooling · Photovoltaic (PV)

1 Introduction Solar energy is one of the most abundantly available resources on earth which can be used in various applications such as power generation systems, heating and cooling system, and photovoltaic cell systems. A solar cell device is used to convert irradiance into electricity through the photovoltaic effect. The solar cells convert only 15–20% of incoming irradiance into useful electrical energy and rest of it is converted into heat which increases the surface temperature of the PV panel [1]. For 1 °C rise in temperature of PV cells above the working limit reduces conversion efficiency by 0.4–0.65% [2]. Thus, it becomes necessary to keep the temperature of the PV panel close to the standard operating temperature. Several cooling techniques have been investigated to reduce the temperature of the PV panel such as water cooling, phase change material, heat pipe cooling, and refrigerant cooling, etc. [2–5]. The nonuniform temperature on the surface of the PV panel produces thermal stresses (hot S. S. Bhakre (B) · P. D. Sawarkar Department of Mechanical Engineering, Visvesvaraya National Institute of Technology Nagpur, South Ambazari Rd., Ambazari, Nagpur, Maharashtra 440010, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_59

495

496

S. S. Bhakre and P. D. Sawarkar

spots). Thus, uniform cooling of the PV panel is required to reduce thermal stresses which lead to an increase in the life of PV panel [6]. Tonui and Tripanagnostopoulos et al. [7] studied the performance of PV/T system using thin flat metal sheet suspended at the back wall of an air channel. The experimental data and result of theoretical model are in good agreement. The suggested modification improves electrical and thermal efficiencies. This system is tested with and without glass cover in natural and forced convection [8]. One of the important conclusions of above study is that setting glass cover on PV panel increases thermal efficiency but decreases electrical efficiency. Popovici et al. [9] presented a numerical model to reduce the temperature of PV module at different configuration of heat sinks by changing angles between fins and base plate. Improving efficiency of PV panel has been proposed using aluminum plate heat sink under natural convection and forced convection [10, 11]. Nabil et al. [12] presented a novel geothermal air cooling system coupled with a photovoltaic system for improving the efficiency of the PV panel. In the current study, the effect of a newly designed diffuser with a photovoltaic panel on the uniformity of air distribution was investigated at a different flow rate. Also, the electrical efficiency and power output of a PV panel with a varied flow rate were compared with the reference module.

2 Geometrical Modelling and Numerical Simulation A two-dimensional geometrical model of the diffuser is created to study the distribution of air velocity. Figures 1 and 2 show the computational domain considered using ICEM CFD (curved sidewalls with three inner deflector plates) and unstructured generated mesh for the considered domain, respectively. The diffuser dimensions are varied to obtain a uniform distribution. The dimensions of the computational domain for the case considered are given in Table 1. Examination was carried out under the assumption of laminar flow in a photovoltaic system. The objective was to determine the air flow pattern and effect of heat transfer on the system at varied air flows. The finite volume approach was used Fig. 1 Computational domain of diffuser

Effect of Air Distribution on Cooling of Photovoltaic Panel …

497

Fig. 2 Mesh generated for diffuser

Table 1 Dimensions of computational domain

Parameters

Dimensions

Panel dimension

0.655 × 0.455 m

Height of diffuser

0.2275 m

Radii of sidewalls

0.275 m

Diffuser outlet

0.455 m

Number of air deflectors

3, 4

to solve the given numerical fluid flow problem in ANSYS Fluent 16. A pressurebased solver and ambient temperature of 27 °C (300 K) chosen for a case to obtain simulation results. For uniform distribution of air, a diffuser with three and four deflector plates with curved sidewalls was validated [13]. A diffuser with three deflector plates and curved sidewalls shows a more uniform distribution of air. Figure 3a, b shows the effect of three and four deflector plates on the distribution of air, respectively. Figure 4 shows the magnitude of velocity at the exit of a diffuser with three deflector plates.

Fig. 3 a Effect of three deflector plates on the distribution of air, b effect of four deflector plates on the distribution of air

498

S. S. Bhakre and P. D. Sawarkar

Fig. 4 Comparison of velocity magnitude results at the exit of the diffuser (three deflectors)

3 Experimental Setup The diffuser with three deflector plates with curved sidewalls was fabricated for uniform distribution of air as per the dimension given in Table 1. The experiments were performed to investigate the effect of diffuser on the performance of the PV panel. All experiments were performed at VNIT Nagpur, India (21.1458° N, 79.0882° E) in April 2019 under sunny weather. A schematic and photograph of experimental setup were shown in Figures 5 and 6, respectively. Fig. 5 Schematic of experimental setup

Effect of Air Distribution on Cooling of Photovoltaic Panel …

499

Fig. 6 Photograph of experimental setup

Fig. 7 Variation of solar radiation and ambient temperature of PV panel over a day

3.1 Data Reduction The maximum power and electrical efficiency of the photovoltaic (PV) panels are calculated using Eqs. (1) and (2), respectively. Pmax = Vmax × Imax ηelectrical pv = where I Pmax V max I max Ac

solar radiation (W/m2 ), maximum power (W), maximum voltage (Volt), maximum current (Ampere)   C/s area of PV panel m2

Pmax Vmax × Imax = Pi I × Ac

(1) (2)

500

S. S. Bhakre and P. D. Sawarkar

The temperature of the PV panel was measured using k-type thermocouple for every 30 min. Irradiation was measured using solar meter. The rheostat was adjusted to obtain maximum voltage and maximum current. The current and voltage are measured using a multimeter.

4 Results and Discussion The experiments were performed to investigate the performance of a PV panel using a new design of diffuser. The performance of the PV panel mainly depends on the solar radiation incident on the surface of PV panel. Figure 7 shows the variation of solar radiation and ambient temperature over a day. Figure 8 shows the variation of the surface temperature of PV panel over a day for reference module and at different flow rates of air (1 and 2.5 m/s). The maximum surface temperature observed was 71 °C at 12 noon for the reference module. For 2.5 m/s of air flow rate, the surface cooled down to 64 °C. The maximum temperature drop of 7 °C was observed compared to the reference module. Figure 9 shows maximum power of 25.74 W and 26.69 W are obtained at 12 noon when solar radiation is maximum at the air flow rate of 1 m/s and 2.5 m/s, respectively, compared to 24.82 W for reference module. The use of a higher flow rate (2.5 m/s) increases the power output by 1.87 W. Figure 10 shows that the solar radiation increases, the PV panel will receive higher heat flux that causes the PV panel to be overheated, which results in drop of efficiency. The study shows that, if we compare the results of two flow rates, average percentage increase in efficiency for 1 m/s and 2.5 m/s is 10.55% and 15.33%, respectively, compared to 7.64% in the reference PV panel. The careful examination of Figures 8 and 10 shows that the maximum efficiency is obtained when surface temperature of PV panel is the lowest (at 10:00 am). Fig. 8 Variation of surface temperature of PV panel over a day

Effect of Air Distribution on Cooling of Photovoltaic Panel …

501

Fig. 9 Variation of power output of PV panel versus time of the day

Fig. 10 Variation of electrical efficiency of PV panel versus time of the day

5 Conclusions For uniform distribution of air, different geometries of the diffuser are simulated. The result of simulation shows that among all the geometries, diffuser with curved sidewalls and three deflector plates is the best geometry to be considered for uniform distribution of air velocity. The performance of the PV panel with diffuser increases with an increase in air velocity. The average electrical efficiency of PV panel is increased by 15.33% and power output is increased by 1.87 W at a flow rate of 2.5 m/sec.

502

S. S. Bhakre and P. D. Sawarkar

References 1. Rahman MM (2015) Effects of various parameters on PV-module power and efficiency. Energy Convers Manag 103:348–358. https://doi.org/10.1016/j.enconman.2015.06.067 2. Stropnik R, Stritih U (2016) Increasing the efficiency of PV panel with the use of PCM. Renew Energy 97:671–679. https://doi.org/10.1016/j.renene.2016.06.011 3. Moharram KA, Abd-Elhady MS (2013) Enhancing the performance of photovoltaic panels by water cooling. Ain Shams Eng J 4:869–877. https://doi.org/10.1016/j.asej.2013.03.005 4. Anderson WG (2008) Heat pipe cooling of concentrating photovoltaic cells. IEEE 7(8):978– 985. https://doi.org/10.1109/PVSC.2008.4922577 5. Sajjad U, Amer M (2019) Cost effective cooling of photovoltaic modules to improve efficiency. Case Stud Therm Eng 14:100420. https://doi.org/10.1016/j.csite.2019.100420 6. Baig H, Heasman KC, Mallick TK (2012) Non-uniform illumination in concentrating solar cells. Renew Sustain Energy Rev 16(8):890–909. https://doi.org/10.1016/j.rser.2012.06.020 7. Tonui JK, Tripanagnostopoulos Y (2008) Performance improvement of PV/T solar collectors with natural air flow operation. Sol Energy 82:1–12. https://doi.org/10.1016/j.solener.2007. 06.004 8. Shahsavar A, Ameri M (2010) Experimental investigation and modeling of a direct-coupled PV/T air collector. Sol Energy 84:1938–1958. https://doi.org/10.1016/j.solener.2010.07.010 9. Popovic G, Sebastian VH (2016) Efficiency improvement of photovoltaic panels by using air cooled heat sinks. Energy Procedia 85:425–432. https://doi.org/10.1016/j.egypro.2015.12.223 10. Mays A (2017) Improving photovoltaic panel using finned plate of aluminum. Energy Procedia 119:812–817. https://doi.org/10.1016/j.egypro.2017.07.103 11. Soliman MN (2019) An experimental study of the performance of the solar cell with heat sink cooling system. Energy Procedia 162:127–135. https://doi.org/10.1016/j.egypro.2019.04.014 12. Elminshawy N (2019) Performance of PV panel coupled with geothermal air cooling system subjected to hot climatic. Appl Therm Eng 148:1–9. https://doi.org/10.1016/j.applthermaleng. 2018.11.027 13. Amanlou Y, Hashjin TT (2018) Air cooling low concentrated photovoltaic/thermal (LCPV/T) solar collector to approach uniform temperature distribution on the PV plate. Appl Therm Eng 141:413–421. https://doi.org/10.1016/j.applthermaleng.2018.05.070

Numerical Investigations of Photovoltaic Phase Change Materials System with Different Inclination Angles Tushar Sathe, A. S. Dhoble, Sandeep Joshi, C. Mangrulkar, and V. G. Choudhari

Abstract Excess unused solar radiation falling on the PV system rises its temperature and drops the electrical conversion efficiency. Phase change materials can be effectively used for the PV thermal management systems due to its high latent heat. In the current research work, effect of inclination angle on the thermal performance of PV PCM system is investigated. The inclination angle is increased from 15° to 90° with constant PCM thickness of 30 mm at 1000 W/m2 of incoming solar radiations. It is observed that the PCMs can be effectively used for the PV thermal management systems. The increase in inclination angle of PV PCM system reduces the time required for melting PCM and increases the PV surface temperature. Keywords Boussinesq approximation · Photovoltaic · Phase change materials

1 Introduction The natural resources for fulfilling ever increasing demand for heat and electricity across the globe are on the verge of extinction. Solar energy has been considered as one of the prominent alternatives for natural resources due to its abundant availability, free of cost and probability to efficiently replace the use of natural fuels for the production of heat or electricity. Solar energy from the sun can be effectively converted into electricity with photovoltaic (PV) solar cells. PV solar cell is a very mature technology now; however, it has a serious issue of decreasing its efficiency T. Sathe (B) · A. S. Dhoble · V. G. Choudhari Department of Mechanical Engineering, Visvesvaraya National Institute of Technology, South Ambazari Rd., Ambazari, Nagpur, Maharashtra 440010, India e-mail: [email protected] S. Joshi Department of Mechanical Engineering, Shri Ramdeobaba College of Engineering and Management, Nagpur 440013, India C. Mangrulkar Department of Mechanical Engineering, Bajaj Institute of Technology, Wardha, India © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_60

503

504

T. Sathe et al.

with the increase in its surface temperature. Solar PV panels generally convert 16– 20% of the incoming solar radiation into electricity and most of the excess unused solar radiation causes to increases the temperature of PV cells; which affects its performance [1]. The solar PV cell efficiency is measured at the standard test conditions of 25 °C, AM1.5 and the incoming solar radiations of 1000 W/m2 . The increase of 1 °C in the surface temperature of PV solar cells above 25 °C causes to decrease the efficiency of PV solar cells around 0.45–0.6% and thus, PV thermal management systems are considered to be one of major challenge across the globe especially, in the country like India with very hot climate in most of the time around the year. During the last 2–3 decades various active and passive thermal management systems have been investigated by various researchers; though, the best possible thermal regulation systems are still not available. The PCMs may be organic or inorganic and have a common tendency to absorb heat during charging and releasing the absorbed heat during the discharging period [2]. Phase change materials (PCMs) have been used as thermal energy storage systems in the past but, due to its property of high latent heat and maintaining a constant temperature during phase change; it can be efficiently combined with the PV systems [3]. Choubineh et al. [4] performed experimental investigations of PV systems with salt hydrates as a PCM for the applications in the building storage systems. They found a 3.7–4.3 °C reduction in the PV panel surface temperature. Karthick et al. [5] experimentally investigated the effect of inorganic Glauber salt as a PCM on the performance of building-integrated PV PCM system. The results showed a 12% reduction in PV cell temperature compared to the conventional PV system. The CO2 mitigation over the lifetime of the developed PV PCM system was observed to be 1.74 tCO2 . Su et al. [6] performed on-site experimental analysis of concentrated PV systems. The performance of the PV water cooling and PV PCM system was compared. They concluded that PCM cooling was a more effective technique for the performance improvement of CPV-T systems. Ma et al. [7] performed mathematical modelling and sensitivity analysis of PV system integrated with PCM. They developed a 1D resistance model and observed that with every 100 W/m2 increase in solar radiation causes a 5 °C increase in the peak PV temperature, however, they suggested that the PCM melting temperature should be slightly higher than the surrounding ambient temperature. Lu et al. [8] investigated the performance of building-integrated concentrated PV systems with the fins and PCMs. A 12% improvement in PV efficiency was observed with the developed fin-based PV PCM system. It is observed from the previous literature that the PCM can be effectively utilized for the thermal regulation of PV systems. Huge research work has already been carried out to analyze the performance of PV system with PCM; however, it is noticed that the numerical analysis of PV systems integrated with PCMs need to be performed more effectively as most of the previous literature is based on vertical building-integrated PV system and very few studies are available with inclined PV systems. It is well known that the PV systems are mounted with different inclination angles depending on the geographical location on the earth. In the present work, numerical investigations are carried out with a PV PCM system at different inclination angles. The numerical model is validated first with

Numerical Investigations of Photovoltaic Phase Change Materials …

505

the past experimental results available in the literature. The effect of temperature on the specific heat and thermal conductivity is considered during the analysis. Natural convection plays a major role during inclined PV PCM system; thus, the effect of natural convection is also considered during numerical analysis using Boussinesq approximations. The inclination angle of the PV PCM system is varied from 15° to 90° and the effect of inclination angle on the different performance parameters are analyzed.

2 Methodology In the current research work, a polycrystalline PV panel is considered for the numerical analysis integrated with the PCMs. The PV panel consists of five layers and at the back of the PV panel, the PCM container is attached with the aluminum metal plate as a top side as shown in Fig. 1. For the numerical analysis few assumptions were considered as heat loss is neglected from the bottom and the sidewall, solar flux is distributed uniformly and the PV panel contact resistance is neglected. The melting process of PCM inside the PCM container is modelled using enthalpyporosity technique explained in Voller and Prakash [9]. The governing equations used during the numerical analysis are given in Table 1. The developed numerical model is validated first with the experimental results obtained by Kamkari et al. [11] with different input temperatures of 60 and 70 °C applied to the vertical PCM tank as shown in Fig. 2. The grid independence study and time-independent study has also been carried to analyze the effect of changing time step size and grid size on the numerical results. Three different time step size with different number of nodes is investigated and finally, 18,620 number of nodes is selected for the analysis with 0.1 s of time step size. The temperature variations of the PV surface, PCM melt fraction inside PCM container and velocity of PCM during Fig. 1 Numerical model of the system

506 Table 1 Governing equations used in the numerical analysis [10]

T. Sathe et al. Boussinesq approximations Continuity equation

ρ = ρl /(β(T − Tl ) + 1) (1)  → ∂ρ + ∇. ρ − u = 0 (2) ∂t

 −   →−  ρ→ u + ∇. ρ − u→ u = − → → → μ∇ 2 − u −∇ p +ρ − g + S (3)  − →  ∂ ∂t (ρ H ) + ∇. ρ u H = ∇.(k∇T ) (4) − → (1−γ )2 − → u (5) S =A ∂ ∂t

Momentum equation

Energy equation − → The source term S in the momentum equation

mush γ 3 +ε

Where, β thermal expansion coefficient, Amush mushy zone constant, H enthalpy

Fig. 2 Validation of numerical model with experimental [11]

1.0

Liquid fraction

0.8 700C 0.6

600C

0.4

Numerical simulation Experimental results

0.2

0.0 50

0

50

100

150

200

Time (min)

250

300

350

phase transition is investigated at different tilt angle of the system with 1000 W/m2 of incoming heat flux. ANSYS Fluent 16 is used for the numerical analysis. SIMPLEC scheme is used for the pressure velocity coupling and PRESTO is used for the pressure spatial discretization. Energy, velocity and continuity residuals were selected to be 10–8, 10–4 and 10–4, respectively. Piecewise linear changes are used to accommodate change in specific heat as suggested by Biwole et al. [12]. RT25 is selected for the analysis with the melting point of 26.6 °C. The ambient temperature is selected to be 20 °C.

Numerical Investigations of Photovoltaic Phase Change Materials …

507

3 Results and Discussion PCM melting is greatly affected by inclination angles due to changes in a natural convection current. Inclination angles of PV system are different for different geographical locations and thus to get overall understanding of PCM melting, inclination angles are varied from 15° to 90° in the present work. Figure 3 shows the variation of different parameters of PV PCM systems with respect to time at different inclination angles. Figure 3a exhibits the changes in the PCM temperature inside the PCM container. It can be observed that the temperature of PCM increases with the increase in the inclination angle. For the smaller inclination angle, the heat transfer from the aluminum surface to the PCM is mainly with conduction; however, the natural convection current is quite low due to smaller velocity magnitude as shown in Fig. 3d. As the inclination angle increases the natural convection dominates the conduction heat transfer and it increases the heat transfer to the PCM from aluminum plate. Figure 3b shows the PV surface temperature with respect to time. It can be seen that, as the inclination angle increases the PV surface temperature is also increases. As, 360

150 300 450 600 900

340

330

0

Temperature of PV surface(0C)

Temperature of PCM (0C)

350

320

310

300

15 300 450 600 900

350

340

330

320

310

300

290 0

50

100

150

200

250

0

300

50

100

(a)

250

300

150 300 450 600 900

0.0005

Velocity magnitude(m/s)

Melt fraction inside PCM container

0.8

200

(b)

150 300 450 600 900

1.0

150

Time (min)

Time (min)

0.6

0.4

0.2

0.0004

0.0003

0.0002

0.0001

0.0000

0.0 0

50

100

150

Time (min)

(c)

200

250

300

0

50

100

150

200

250

300

Time (min)

(d)

Fig. 3 a Temperature of PCM, b temperature of PV surface, c melt fraction, d velocity magnitude

508

T. Sathe et al.

t = 40min

t=80min

t=120min

t=160min

t=200min

t=240min

Fig. 4 Temperature variations in PV PCM system with respect to time

at a lower inclination angle, the heat transfer is quite low, the time required to melt the PCM is also more and thus it takes more time to convert into liquid Fig. 3c. The velocity magnitude of PCM inside the PCM container is more in 90° system and it decreases with the decrease in the inclination angle as shown in Fig. 3d. Figure 4 shows the contours of temperature variation with respect to time at 45° of inclination angle. The temperature of the PV PCM systems is observed to be more on the topmost corner of the system compared to the bottom side during the analysis and it is not distributed equally. Thus, it is suggested that some passive systems like extended surfaces can be attached on the top part of the PV PCM systems to extract excess heat from the outside. The extended surfaces can also be placed on the heated surfaces to make sure even distribution of temperature throughout the heated surface.

4 Conclusions Inclination angle plays a major role in the design of PV PCM system. In the current research work, effect of inclination angle on the thermal performance of PV PCM systems is investigated. It is observed that as the inclination angle increases the time required to melt the PCM decreases and PV surface temperature increases. Thus, it can be concluded that optimum PCM thickness needs to be selected for the required inclination angle of PV panel during the design of PV PCM system.

References 1. Browne MC, Norton B, Mccormack SJ (2015) Phase change materials for photovoltaic thermal management. Renew Sustain Energy Rev 47:762–782. https://doi.org/10.1016/j.rser.2015. 03.050 2. Barnes FS, Levine JG (2011) Large energy storage systems handbook. CRC Press 3. Nazir H, Batool M, Bolivar Osorio FJ, Isaza-Ruiz M, Xu X, Vignarooban K et al (2019) Recent developments in phase change materials for energy storage applications: a review. Int J Heat Mass Transf 129:491–523. https://doi.org/10.1016/j.ijheatmasstransfer.2018.09.126 4. Choubineh N, Jannesari H, Kasaeian A (2019) Experimental study of the effect of using phase change materials on the performance of an air-cooled photovoltaic system. Renew Sustain Energy Rev 101:103–111. https://doi.org/10.1016/j.rser.2018.11.001

Numerical Investigations of Photovoltaic Phase Change Materials …

509

5. Karthick A, Murugavel KK, Ramanan P (2018) Performance enhancement of a buildingintegrated photovoltaic module using phase change material. Energy 142:803–812. https://doi. org/10.1016/j.energy.2017.10.090 6. Su Y, Zhang Y, Shu L (2018) Experimental study of using phase change material cooling in a solar tracking concentrated photovoltaic-thermal system. Sol Energy 159:777–785. https:// doi.org/10.1016/j.solener.2017.11.045 7. Ma T, Zhao J, Li Z (2018) Mathematical modelling and sensitivity analysis of solar photovoltaic panel integrated with phase change material. Appl Energy 228:1147–1158. https://doi.org/10. 1016/j.apenergy.2018.06.145 8. Lu W, Liu Z, Flor JF, Wu Y, Yang M (2018) Investigation on designed fins-enhanced phase change materials system for thermal management of a novel building integrated concentrating PV. Appl Energy 225:696–709. https://doi.org/10.1016/j.apenergy.2018.05.030 9. Voller VR, Prakash C (1987) A fixed grid numerical modelling methodology for convectiondiffusion mushy region phase-change problems. Int J Heat Mass Transfer 30:1709–1719. https://doi.org/10.1016/0017-9310(87)90317-6 10. Kamkari B, Amlashi HJ (2017) Numerical simulation and experimental verification of constrained melting of phase change material in inclined rectangular enclosures. Int Commun Heat Mass Transfer 88:211–219. https://doi.org/10.1016/j.icheatmasstransfer.2017.07.023 11. Kamkari B, Shokouhmand H, Bruno F (2014) Experimental investigation of the effect of inclination angle on convection-driven melting of phase change material in a rectangular enclosure. Int J Heat Mass Transfer 72:186–200. https://doi.org/10.1016/j.ijheatmasstransfer.2014.01.014 12. Biwole PH, Eclache P, Kuznik F (2013) Phase-change materials to improve solar panel’s performance. Energy Build 62:59–67. https://doi.org/10.1016/j.enbuild.2013.02.059

Edge Feature Based Classification of Breast Thermogram for Abnormality Detection Shawli Bardhan

and Sukanta Roga

Abstract Nowadays, breast cancer is the most commonly occurring cancer among women. Early diagnosis of such disease improves the survival rate. Mammography is considered as the gold standard for the diagnosis of breast disease. But a frequent examination of the breast through mammography increases the chance of cancer occurrence due to radiation effect. However, taking advantage of non-radiating nature of thermography, it can be used for regular screening of breast region temperature distribution noninvasively. The presence of an inflammatory region in a breast thermogram is a signal of abnormality. In a thermogram, the inflammation is represented with higher gray level pixels. Observing this, in this study we aim to classify sick and healthy breast thermograms by comparing the edge features of the images. For the elimination of unwanted edges, the anisotropic diffusion is used as the filtering process of thermograms before edge detection. The outcome of the Artificial Neural Network-based classification using edge features as input generates 93.3% accuracy, with an area under the curve value 0.92, sensitivity 100%, and specificity 85.7%. Keywords Thermogram · Edge detection · Feature extraction · Classification

1 Introduction Breast cancer is the most commonly occurring cancer among women. In India, the occurrence rate of breast cancer in a female is 25–32% among all types of cancers [1]. Also among all cancer, worldwide it is the second most common type of cancer [2]. 97% of survival rate can be achieved through early detection and treatment of S. Bardhan Department of Computer Science and Engineering, Tripura University (A Central University), Suryamaninagar, Agartala, Tripura 799022, India e-mail: [email protected] S. Roga (B) Department of Mechanical Engineering, Visvesvaraya National Institute of Technology, South Ambazari Rd., Ambazari, Nagpur, Maharashtra 440010, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_61

511

512

S. Bardhan and S. Roga

breast cancer [3]. Different imaging procedures are present for breast abnormality diagnoses such as mammography, ultrasound imaging, magnetic resonance imaging, etc. But those techniques contain the drawback of radiation effect, invasive nature of examination, patient discomfort, etc. Infrared breast thermography is a noninvasive, radiation-free method of breast screening that can detect abnormality in the early stage of breast disease. The thermography technique is based on receiving the infrared radiation emitted from the breast skin surface related to internal inflammation and represents it in image format. In general, the inflamed regions are represented with higher gray level pixel intensities in the thermogram. Analysis of the thermogram can help in early stage abnormality detection. Sathish et al. [4] performed local energy features of wavelet sub-bands based contralateral asymmetry analysis and achieved 91% accuracy. The left and right breast thermogram asymmetry is measured for abnormality detection by Kirubha et al. [5]. Garduño-Ramón et al. [6] also performed image processing methods for contralateral breast asymmetry analysis and achieved 0.8684 and 0.8943 rate of sensitivity and specificity, respectively, for the detection of breast abnormality. The abovementioned recent studies show that contralateral asymmetry analysis is the most commonly used method for breast abnormality detection. But the method fails if the temperature of both the breast increases due to the presence of abnormality. Focusing on the limitation of existing methods, here in this study, we perform the classification of sick and healthy breast thermograms by extracting holistic edge features of the image.

2 Methodology The method targets to classify the thermograms for abnormality detection. One group consists of healthy thermograms and another is of sick thermograms of breast. The methodology is dependent on edge features and performed through seven different steps as shown in Fig. 1. The rest of the section detailed steps of the methodology.

Fig. 1 Overall flow diagram of breast thermogram classification

Edge Feature Based Classification of Breast Thermogram …

513

Fig. 2 Preprocessing (input thermogram → cropped thermogram → blue channel thermogram)

2.1 Dataset Collection The experiment has been performed over the publically DMR dataset [7]. The dataset contains thermograms of sick breast and healthy breast. We collected 51 thermograms and among those 23 belong to sick category and rest are from healthy groups. The first image of Fig. 2 demonstrates an example of the collected sick thermogram.

2.2 Preprocessing The experiment concentrates on the pixel of thermogram breast area. But the thermograms contain unwanted regions like arm, stomach, logo, temperature bar, etc. Information from those unwanted regions may generate errors in analysis and classification. Therefore, in the preprocessing step, the breast region is manually cropped. In the next step of preprocessing, the blue channel from the cropped RGB thermogram has been selected as this channel contains maximum information about inflammation spreading. The example related to the outcome of preprocessing is shown in Fig. 2.

2.3 Filtering The white patches in the thermogram, in general, represents an increase of temperature due to tumor or any other abnormality. The extraction of edge features from the white region and its analysis is the main aim of the method for abnormality detection. But the presence of noise in the thermogram and frequent change of intensity will generate excessive region oriented boundary edges. Therefore, filtering has been applied before region oriented boundary edge detection. The anisotropic diffusion by Perona-Malik is used here for filtering purposes [8, 9]. This is basically a denoising mechanism based on partial derivatives. The texture smoothing/blurring is performed through this due to the application of the law of diffusion. The advantage of the filtering is that, the smoothening is only happened in between homogeneous regions due to the use of threshold function and information of edges are preserved.

514

S. Bardhan and S. Roga

Fig. 3 Outcome of filtering using anisotropic diffusion and canny edge detection

Therefore, the noise of images is removed without affecting edge details and as a result, unwanted edges are removed. To control the diffusion mechanism, the method requires to give the iteration number, conduction coefficient, and speed of diffusion. Here we use 50 iterations with conduction coefficient value 30 and speed of diffusion 0.25. The output will preserve the information of high contrast edges for abnormality analysis. The outcome of filtering is shown in the middle image of Fig. 3 below where input is the blue channel thermogram.

2.4 Edge Detection In this step, the region base high contrast edges are detected. For that, the canny edge detection mechanism has been applied over the outcome of filtering mechanism [10]. Though canny uses the Gaussian filtering for smoothing base noise removal, it does not diminish the edges as the high contrast edges are preserved through anisotropic diffusion-based filtering. After smoothing, the gradient magnitude is computed for finding out the edges. In the third step of canny, non-maximal suppression performed to eliminate pixels that do not represent the edge of a region. In the final step, hysteresis thresholding is applied depending on the image gradient for edge detection. The last image of Fig. 3 shows the outcome of canny edge detection over the outcome of anisotropic diffusion.

2.5 Feature Extraction The outcome of edge detection is a binary image. Therefore, binary image-based features are extracted from the edge outcome. Total 11 shape features are extracted: area, Euler number, perimeter, convexity, eccentricity, orientation, solidity, major axis length, minor axis length, elongation, and circularity [11]. All those mentioned 11 features are extracted from each breast thermogram present in both the healthy and sick group. The extracted features are fed to classifier for training and testing based system development for breast thermogram abnormality identification.

Edge Feature Based Classification of Breast Thermogram …

515

2.6 Classification For classification, Artificial Neural Network (ANN) is used in this study with twolayer feed-forward channel. The sigmoidal activation function is applied to generate nonlinearity in the network. Here we used 40 hidden neurons for optimal training of the network through trial and error mechanism. 30% of total images are sent for testing purposes of the system where each represented with 11 features. The output layer divides the input images into two classes: one is healthy group and another represents sick thermograms. The Support Vector Machine (SVM) classifier is also used for the same classification for the comparative study of classification.

2.7 Validation The validation is performed for the outcome of the system measurement. The classifier outcome (sick or healthy) of each thermogram used for testing is matched with the actual status of the breast thermograms.

3 Result The measurement of the ability of a system is performed depending on the quantitative outcome. Here, also to analyze the system performance, we calculate the accuracy of both the ANN and SVM classifier. The ANN classifier provides maximum accuracy of 93.3% with sensitivity 100% and specificity 85.7% as shown in Fig. 4b through the confusion matrix of classification. Also, the ROC curve of ANN classifier (given in Fig. 4a) shows that the area under the curve(AUC) value of classification is also high, which is 0.92 which indicates an acceptable performance of the system. We also performed the same classification using the SVM classifier with linear, polynomial, and radial basis function kernel. The polynomial kernel provides maximum accuracy of 86% using SVM classifier with AUC 0.82 as shown in Fig. 4c. Though the performance of SVM classifier is also acceptable, ANN provides the highest accuracy of the same classification.

4 Conclusion In past research, the classification of breast thermograms is performed through contralateral asymmetry analysis of two breast regions. The method contains the possibility of failure if both the breasts contain a similar distribution of temperature. Here in this study, we classify breast thermograms by extracting edge oriented features

516

S. Bardhan and S. Roga

Fig. 4 a ROC curve of ANN classifier with AUC: 0.92; b confusion matrix of ANN classifier; c ROC curve of SVM classifier with AUC: 0.82

from sick and healthy thermogram images. The anisotropic diffusion base filtering is applied over the blue channel of color thermogram for noise cancellation. The outcome of the analysis shows that use of the edge-based extracted features in the ANN classifier provides maximum accuracy of 93.3%, with sensitivity 100% and specificity 85.7%. Acknowledgements Authors would also like to thank Dr. M. K. Bhowmik, Assistant Professor, Department of Computer Science and Engineering, Tripura University, Suryamaninagar 799022, Tripura, India for his support during knowledge development in the area of thermal image processing, image feature extraction, and classification.

Edge Feature Based Classification of Breast Thermogram …

517

References 1. Breast Cancer India Pink Indian statistics. Available at http://www.breastcancerindia.netlbc/ statistics/stat_global.htm 2. Resmini R, Araujo A, Conci A, Silva L, Moran M (2019) Number of Texture unit as feature to breast’s disease classification from thermal images. In: Proceedings of IEEE/ACS international conference on computer systems and applications, AICCSA, 1–6 Nov 2018. https://doi.org/ 10.1109/AICCSA.2018.8612826 3. Ng EYK (2009) A review of thermography as promising non-invasive detection modality for breast tumor. Int J Therm Sci 48(5):849–859 4. Sathish D, Kamath S, Prasad K, Kadavigere R (2019) Role of normalization of breast thermogram images and automatic classification of breast cancer. Vis Comput 35(1):57–70. https:// doi.org/10.1007/s00371-017-1447-9 5. Kirubha SPA, Anburajan M, Venkataraman B, Menaka M (2018) A case study on asymmetrical texture features comparison of breast thermogram and mammogram in normal and breast cancer subject. Biocatal Agric Biotechnol 15:390–401. https://doi.org/10.1016/j.bcab.2018.07.001 6. Garduño-Ramón MA, Vega-Mancilla SG, Morales-Henández LA, Osornio-Rios RA (2017) Supportive noninvasive tool for the diagnosis of breast cancer using a thermographic camera as sensor. Sens Switz 17(3). https://doi.org/10.3390/s17030497 7. Silva LF, Saade DCM, Sequeiros GO, Silva AC, Paiva AC, Bravo RS et al (2014) A new database for breast research with infrared image. J Med Imaging Health Inform 4:92–100. https://doi.org/10.1166/jmihi.2014.1226 8. Perona P, Malik J (1990) Scale-space and edge detection using anisotropic diffusion. IEEE® Trans Pattern Anal Mach Intell 12(7): 629–639 9. Gerig G, Kubler O, Kikinis R, Jolesz FA (1992) Nonlinear anisotropic filtering of MRI data. IEEE Trans Med Imaging 11(2):221–232 10. Canny John (1986) A computational approach to edge detection. IEEE Trans Pattern Anal Mach Intell 6:679–698 11. George YM et al (2014) Remote computer-aided breast cancer detection and diagnosis system based on cytological images. IEEE Syst J 8(3):949–964. https://doi.org/10.1109/jsyst.2013. 2279415

Analytical Approach to Develop a Robust Mechanism for On-Orbit Gimballing of Satellite Antenna V. Sri Pavan RaviChand , Anoop Kumar Srivastava, Abhishek Kumar, H. N. Suresha Kumar, and K. A. Keshavamurthy

Abstract Mechanism for gimballing of antenna on-orbit is a critical appendage on any satellite. It is necessary for directed beam data transfer from satellite to ground stations. For near-real-time data transfer, the dual-axis steering of the antenna is required, which demands a two-axis steering mechanism for precise and accurate pointing of antenna. This paper details the development of a robust antenna gimbal mechanism based on a modular configuration, which fulfills varied satellite requirements, using iterative analytical studies based on finite element analysis (FEA). The configuration of the mechanism has been made robust by ensuring high natural frequency and sufficient margins for static and dynamic loads across variants of the configuration. Different configurations of the mechanism integrated with various types of antennas have been evaluated by varying specific modular parameters. A systematic overview of the approaches adopted in this process of development of robust mechanism with optimized mass, high stowed frequency and high reliability has been presented. Keywords Gimbal · Analytical · Robust mechanism

1 Introduction Mechanisms used in space applications are mission critical and need to be evaluated on various dimensions like modal, quasi-static loading and frequency response analyses during configuration studies. Analytical methodology using finite element analysis (FEA) is used to estimate characteristics of the system that are used as input for the improvement of system configuration via iterative studies. Mechanism for dual-axis gimballing of the antenna is a key spacecraft appendage that enables the communication between earth stations and satellite. It provides twoaxis steering of the antenna to ensure hemispherical coverage and ground station V. S. P. RaviChand (B) · A. K. Srivastava · A. Kumar · H. N. Suresha Kumar · K. A. Keshavamurthy Spacecraft Mechanisms Group, U.R.Rao Satellite Centre, ISRO, Bengaluru 560017, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_62

519

520

V. S. P. RaviChand et al.

tracking during data transmission. These gimballing requirements vary significantly across the satellites based on mission objectives. In this regard, making the mechanism configuration robust enough using modularity so as to cater to the requirement of various antenna sizes and rotation angles is highly appreciable. This paper presents the systematic methodology adopted using FEA studies and parametric analyses to realize a robust mechanism configuration with modular features for dual-axis gimbal of varied antenna systems. This has been achieved by modifying one module of the mechanism, called hold-down restraint unit that helps in meeting stiffness characteristics of the mechanism, while retaining all other modules of the mechanism configuration as common modules.

2 Requirements of Antenna Gimbal Mechanism Satellite-level requirements for positioning an antenna on-orbit are based upon the communication system, envisaged between the ground stations and satellite, which depends upon various parameters that can be broadly classified as below: a. Mission parameters like orbital height, inclination, field of view, pointing accuracy, data transmission (real time or non-real time), ground passes, steering rate. b. Environmental constraints like extreme temperatures, vibration loads during launch, radiation impacts on-orbit. c. RF specifications in the form of frequency band, spot beam focus, RF power. Based on the satellite-level requirements, mechanical system-level requirements are derived. These requirements can be classified as: a. Drive system parameters, to provide torque for gimbal along two axes, based upon global parameters like steering rate, vibration loads, pointing accuracy [1]. b. Environment support system considerations in the form of stiffness characteristics, stress distribution, moments along gimbal axes that depend on global parameters like launch vibrations, on-orbit thermal gradients, geometry of antenna. c. RF transmission system considerations like waveguide or cable management unit for contactless data transmission during steering depending on global parameters like orbital height, inclination, frequency band, RF power [2]. From the above, it is evident that multiple variants of mechanism configuration are required to meet the mechanical requirements corresponding to varied satellite demands. However, the realization of different mechanism configurations is not warranted from a practical point of view. Thus, an optimized solution in the form of a mechanism configuration that caters to the varied satellite requirements is essential. In this regard, the development of a modular mechanism configuration, where with

Analytical Approach to Develop a Robust Mechanism …

521

modification of specific modules, the mechanism configuration can cater to the varied gimbal requirements, is worth considering. Advantages of modular configuration include: a. Benefit of catering to multiple requirements with minimum changes. b. Ease of testing, as it can be carried out module wise. c. Optimization of time, as modules can be produced in batches that can be used as per satellite requirements. Development of such robust mechanism configuration demands analytical approach toward iterative parametric studies in order to identify as well as minimize the specific modules that determine variants of mechanism configuration.

3 Robust Mechanical Configuration for Dual-Axis Antenna Gimbal Mechanism The mechanism for antenna gimbal is configured with several sub-systems like motor, gear, RF transmission system and hold-down system to fulfill the mechanical requirements. The prime objective in the development of this mechanism configuration is to fulfill the varied antenna gimballing requirements across satellites with minimum modification of the mechanism configuration. Toward this, the mechanism is divided into several modules where the majority of these modules are the same across the variants of mechanism configuration. For instance, identical high torque capability drives are used for both the gimbal axes in all the variants. Also, the mechanism configuration has been made keeping in view the requirements from different multidisciplinary systems involved in satellite development. For instance, mechanism can suit X-band as well as Ka-band rotary joints that are determined by the communication systems. Thus, robustness in the configuration of mechanism for antenna gimbal has been enabled by adoption of modular approach and multidisciplinary perspective during the mechanism development process. Analytical approach has been adopted to analyze the mechanism configuration module wise, using iterative parametric studies. This enabled the development of variants of mechanism configuration to fulfill the varied satellite demands, with minimum changes in the configuration. For instance, the gimballing mechanism for 1-m-diameter antenna and 0.35-m-diameter antenna is identical except the holddown system, as shown in Fig. 1. This characteristic of robustness in mechanism configuration enables quicker design iterations and effective parametric analyses in order to address larger and diverse antenna gimbal requirements.

522

V. S. P. RaviChand et al.

Common modules – Drive modules Antenna Variable modules – Hold-down unit with satellite interface structure

Patch antenna gimbal mechanism

Parabolic dish antenna gimbal mechanism

Fig. 1 CAD representation of two variants of gimbal mechanism

3.1 Details of the Analytical Approach Adopted for Development of the Robust Antenna Gimbal Mechanism Spacecraft mechanism systems are determined by diverse mechanical characteristics owing to the fact that they need to act as stiff structures during launch vehicle phase while they need to deploy with predefined degrees of freedom on-orbit. This necessitates a host of studies including kinematic simulations, static analyses and dynamic response evaluations. Adoption of analytical approach during these studies enables the development of robust mechanism configuration that can cater to diverse satellite requirements. Some insights have been gained from such attempts that have been carried out with respect to interplanetary mechanisms [3] and self restraint gimbal mechanism [4]. The details of the approach along with their application for devising one of the variants of mechanism configuration have been presented.

3.1.1

Studies Carried Out Under the Analytical Approach

These studies are carried out sequentially in an iterative manner during the entire mechanism development process, i.e., from mechanism conceptualization phase till on-orbit performance review stage, so as to minimize modifications across different variants of gimbal mechanism configuration while meeting the respective satellite mission requirements. a. Determination of geometric shape for gimballing Kinematic simulation is carried out to obtain the configuration of mechanism in the form of geometric shape for steering along the two gimbal axes. It is also used to carry out the interference studies during steering. b. Evaluation of dynamic loads during gimballing

Analytical Approach to Develop a Robust Mechanism …

523

After ensuring that all the components are seamlessly rotating without any interference, dynamic simulations are carried out to estimate loads on the joints by simulating drive speeds, acceleration and inertia parameters. c. Estimation of stiffness characteristics of mechanism Satellite-based systems are primarily driven by stiffness characteristics. Hence, the configuration of mechanism needs to be evaluated with respect to stiffness by carrying out normal mode analysis with requisite boundary conditions. This FEA provides details on dynamic characteristics of the mechanism with respect to mode shapes, natural frequency values and participating mass details. Based on the FEA results, the improvements with respect to mechanism configuration in the form of providing stiffeners, etc. can be carried out iteratively. d. Evaluation of strength parameters of system Spacecraft mechanisms are characterized by intricate geometry so as to withstand various eccentric loads like extreme thermal gradients and large launch accelerations. In this regard, quasi-static FEA, for inertia loading of 30 or 20 g based on launch vehicle specifications, is carried out to evaluate stress distribution and joint forces in the system. Accordingly, the improvements in configuration, by way of providing fillets, can be carried out to reduce stress concentrations and high joint loads. e. Evaluation of amplitude responses of system during vibration Integrity of the mechanism system under harmonic vibrations across a range of frequencies, determined based on mission specifications, needs to be ensured so as to avoid resonance. This can be ensured by evaluating amplitudes within the mechanism using frequency response FEA and accordingly reconfiguring the system to have beta values within environmental specification limits. f. Estimation of responses at satellite level Mechanism system’s fundamental frequency should be kept away from the satellite frequencies so as to avoid resonance phenomena. This can be achieved by evaluating amplitudes at the spacecraft level using coupled load analysis. This coupled load FEA also provides the input loads for the mechanism system. g. On-orbit performance analysis The functioning of mechanisms on-orbit is contingent upon the body rates experienced by the satellite. To evaluate on this account, analysis from multi-body dynamics perspective is carried out, and the corresponding data is correlated with the actual on-orbit satellite data received, by carrying out fast Fourier transform (FFT) on body rate plots, to evaluate the overall system performance. The analytical approach has been applied for devising a mechanism for dualaxis gimbal of parabolic dish antenna from the available mechanism for dual-axis gimbal of a patch antenna, as shown in Fig. 2. The variations with respect to satellite requirements for these configurations are that the mass and size of parabolic dish antenna are about twofold more than that of the patch antenna. Also, the stowing

524

V. S. P. RaviChand et al.

Hold-down structural interface

Patch antenna gimbal mechanism

Parabolic dish antenna gimbal mechanism

Fig. 2 Finite element analysis on two variants of antenna gimbal mechanism

angles are different for both cases. Accordingly, there needs to be variations on mechanical requirements to accommodate stiffness characteristics. Based on the above-mentioned analytical approach, configuration of mechanism for gimbal of parabolic dish antenna has been devised from that of the mechanism for gimbal of the patch antenna, as shown in Fig. 2, by appropriate modification. It is to be noted that both the mechanism configurations have the same modules except for hold-down unit and structural interface of mechanism with satellite. Using analytical studies, the hold-down system module related to patch antenna gimbal mechanism configuration has been modified to meet stiffness requirements of parabolic dish antenna, while the structural interface base has been modified to meet stowing angle requirements. It is important to note that both these mechanisms are planned to be used on agile low mass satellites of nearly the same structural frequency, thereby the fundamental frequencies of both these variants need to be almost the same in spite of differences in mass or geometry of respective antennas. For the given case study, adoption of the systematic methodology enabled devise of both the variants of mechanism configuration, shown in Fig. 2, as per requisite specification of around 70 Hz fundamental frequency with minimal modification of mechanism configuration. Thus, the adoption of the systematic analytical approach during detailed parametric studies enables the development of a robust configuration of modular antenna gimbal mechanism that can cater to the varied satellite demands.

4 Conclusion This paper brings out a systematic methodology using analytical approach for devising a robust configuration of mechanism for dual-axis gimbal of the antenna to cater to the varied satellite requirements. This is achieved by adopting a modular approach with multidisciplinary perspective during mechanism development process based on

Analytical Approach to Develop a Robust Mechanism …

525

iterative analyses and parametric studies. The satellite-level requirements as well as the corresponding mechanical requirements related to antenna gimbal mechanism have been presented. A typical case study related to two variants of the antenna gimbal mechanism configuration has been highlighted. Details of various analytical studies involved in the life cycle of the mechanism configuration process have been highlighted, and their outputs have been defined. The pivotal role played by analytical studies for the development of spacecraft mechanisms has been presented.

References 1. Kumar A, Srivastava AK, Viswanathan N (2015) Design and development of an electromechanical drive for spacecraft mechanisms—a systematic approach. In: 9th national symposium and exhibition on aerospace and related mechanisms (ARMS 2015), ISRO Satellite Centre, Bengaluru, India, 30–31 Jan 2015 2. Kumar A, Viswanathan N, Ramakrishna MV, Bhat NC (2007) A two axis fine tracking mechanism for satellite antenna. In: 13th national conference on mechanisms and machines (NaCoMM07) 3. Suresha Kumar HN, Narendra S, Nirmal AV (2008) Mechanisms for Chandrayaan-1. INSARM, Bangalore Chapter E Newsletter 2(4) 4. Srivastava AK, Kamboj A, Sharma G, Nagaraj BP, Shankara A, Suresha Kumar HN, Keshava Murthy KA (2018) An innovative, self-restrained, high torque, dual gimbal mechanism for antenna pointing. J Spacecraft Technol

Impact of Rock Abrasivity on TBM Cutter-Discs During Tunnelling in Various Rock Formations N. N. Sirdesai , A. Aravind , and S. Panchal

Abstract With the introduction of tunnel boring machines (TBMs), the rate of development and construction of tunnels has increased across the world. Breakdowns, repairs and replacements that occur in a TBM have to be addressed within the tunnel. Of the several components within a TBM, the cutter-picks within the cutter-head undergo maximum wear and tear, thereby requiring periodic replacements. In order to optimize the consumption of cutter-picks, it is important to analyse the forces involved in the cutter and the properties that govern rock–cutter interactions. One such property is the abrasivity, which can be determined by CERCHAR abrasivity index (CAI) tests. Research suggests that the value of CAI is largely dependent on the mineralogical, physical and mechanical properties of rocks. Therefore, this study aims to perform experimental analysis of the various properties and their corresponding effects on granite and to determine the respective changes in CAI. Keywords CERCHAR · Abrasivity · TBM · Tunnel

1 Introduction Mechanized tunnelling has improved the rates of excavation and exploration along the last few decades. This was achieved through the introduction of various rock-cutting machines like tunnel boring machine (TBM), road-headers, etc. Economics of the tunnelling operation is largely dependent on the machine availability. One of the most important components of a tunnelling machine is the cutting mechanism, which in the case of a TBM is the cutter-head that comprises several cutting discs, also known as cutter-picks arranged in a specific and concentric pattern, as illustrated in Fig. 1. Since the cutter-picks undergo continuous wear and tear due to rock interaction, their condition determines the efficiency of the cutting operation. Frequent replacements of the worn-out picks have to be carried out within the tunnelling site which increases the N. N. Sirdesai (B) · A. Aravind · S. Panchal Department of Mining Engineering, Visvesvaraya National Institute of Technology, Nagpur 440010, Maharashtra, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_63

527

528

N. N. Sirdesai et al.

Fig. 1 Types of tunnel boring machines (TBMs) and their respective cutter-head

total project cost and duration. Studies suggest that the consumption of cutter-discs and its rate is mainly governed by the strata (geological) and machine parameters. While the machine parameters are within our control, the type of geological formation and its characteristics should be studied in order to design and deploy the proper type of TBM. One such characteristic is the abrasivity which accounts for gradual loss of the cutting tool due to pick–rock interaction. The physico-mechanical properties of the rock surface being cut and the stresses involved in the pick–rock interaction also play a major role in the total wear [1]. Of the several tests developed for analysing the abrasivity of rocks, the CERCHAR abrasivity index (CAI) is one of the most widely used tests due to its simplicity. The test was developed in 1973 by Laboratoire du Centre d’Etudes et Recherchers des Charbonages de France and involves the formation of a scratch on the surface of a rock sample with the help of a steel stylus. The original design of the equipment involved making a 10-mm scratch on a stationary rock specimen within one second by means of a pin that carried a static load of 70 N [2]. However, a modification to the design is proposed by West [3], wherein a stationary pin with a 70 N static load creates a scratch on a rock sample that moves at a velocity of 1 mm/s. The test is conducted for 10 s. Thereafter, the wear to the steel stylus is examined in order to calculate the CERCHAR abrasivity index (CAI) [4]. The CAI of the rock can be used to estimate the cutter-pick consumption with the help of predefined charts and equations [5]. The magnitude of wear depends entirely on the rock type, and therefore, in this study, various physico-mechanical properties of a granitic rock have been analysed with respect to its CAI.

2 Materials and Methods 2.1 Sample Collection and Preparation In this study, granitic rock samples were used for testing the dependence of physicomechanical properties on its abrasivity. The samples were obtained from the core specimens sent to the Department of Mining Engineering, Visvesvaraya National Institute of Technology (VNIT), by Central Institute of Mining and Fuel Research

Impact of Rock Abrasivity on TBM Cutter-Discs During Tunnelling …

529

(CIMFR). The core samples were cut to appropriate lengths with the help of a core cutter. Care was taken to ensure that the length and attributes of the specimen were in accordance with the standards and methods as specified by the American Society for Testing and Materials (ASTM) and International Society for Rock Mechanics (ISRM), respectively [6–9]

2.2 Physico-mechanical Properties Cylindrical specimens of the sample were used for testing the compressive (σ C ) and tensile (σ T ) strengths. Samples with a length-to-diameter ratio of 2:1 were tested in a compressive testing machine, and the axial strain was recorded with the help of strain gauge which was installed on the sample before the test. The load and strain data was recorded with the help of DataTaker 80G data acquisition system. The tensile strength was analysed using the indirect Brazilian testing method. The cylindrical samples for the tensile test had a length-to-diameter ratio of 0.75:1. The elastic modulus (E) of the samples was measured by analysing the stress–strain plots. It was ensured that the linear section of the plot was selected for the modulus measurement. Additionally, it was ensured that the corresponding stress window of the linear section amounted to at least 20% of the peak stress. This helped in obtaining representative data of the elastic modulus. Furthermore, the porosity of the samples was tested using the buoyancy and saturation techniques. The limitations of this technique are that it can only be applied to rocks that are non-swelling in nature, and that the technique only measures the effective porosity. In this technique, the samples were submerged in water for 24 h before measuring the saturated mass (M SAT ), submerge-saturated mass (M SUB ) and the oven-dried mass (M S ). The porosity (Φ) was calculated using Eq. (1). φ=

MSAT − M S × 100 MSAT − MSUB

(1)

2.3 CERCHAR Abrasivity Index (CAI) The CERCHAR abrasivity set-up used for this study is illustrated in Fig. 2. The setup, as developed by West [3], is a modification of the original design, wherein the surface of the rock sample moves across a stationary pin [1]. The test was conducted in accordance with the recommendations prescribed by the ASTM standard D762510 and the ISRM suggested method [4, 8]. The steel stylus with a Rockwell hardness of 54–56 and a 90° conical tip was used in this study. A static load of 70 N was applied to make a scratch across a predefined path. The rock sample was moved across the stationary pin at a velocity of 1 mm/s. In order to attain statistical coherence, five

530

N. N. Sirdesai et al.

Fig. 2 a CERCHAR set-up, b steel stylus and c scratched sample

scratches were made across each sample using re-sharpened styluses, as suggested by Käsling and Thuro [2]. The damage to the tip was observed and measured using an optical microscope with a 30× magnification. The change in diameter at the tip (DW ) was recorded to an accuracy of 0.01 mm to calculate the CAI as illustrated in Eq. (2). CAI = 10 × DW

(2)

3 Results and Discussions The results of the physico-mechanical and abrasivity analysis have been enlisted in Table 1. The correlation between the physico-mechanical properties and CAI has been plotted in Fig. 3. As seen in this figure, there exists a linear correlation between the strength and elastic properties, and porosity with CAI. The equations and correlation coefficients are illustrated in Eqs. (3–6). CAI = 0.02 · σC + 2.15, R 2 = 0.98

(3)

Table 1 Results of experimental analyses Sample

CAI

σC (MPa)

σT (MPa)

E (GPa)

Φ (%)

G1-1

2.972

38.339

12.930

72.532

0.990

G1-2

2.922

36.524

11.459

65.121

1.195

G1-3

2.865

33.189

9.360

59.547

1.478

G1-4

2.881

34.451

10.223

61.789

1.244

G1-5

2.962

37.845

11.687

72.125

1.001

Impact of Rock Abrasivity on TBM Cutter-Discs During Tunnelling …

531

Fig. 3 Correlation between a compressive strength, b tensile strength, c Young’s modulus and d porosity with respect to CAI

CAI = 0.03 · σT + 2.55, R 2 = 0.91

(4)

CAI = 0.01 · E + 2.39, R 2 = 0.98

(5)

CAI = −0.23 · φ + 3.19, R 2 = 0.91

(6)

The equation states that except for porosity, the CAI increases with an increase in strength and elasticity. Similar correlations have been suggested by several researchers. For instance, Tripathy et al. [10] studied the correlation between abrasivity and physico-mechanical properties of several Himalayan rock samples using statistical and soft-computing tools. The results of their analyses suggest that an increase in the mechanical properties of the rock increases the abrasivity of the rock. This can be attributed to the increase in the resistance provided by the rock to the formation of a scratch. Additionally, the rocks formed by relatively tougher minerals offer high resistivity to deformation. Research conducted by Deliormanlı [11] and Hamzaban et al. [12] have also suggested similar findings. Further, Alber [1] performed a rigorous experimental analysis to correlate the effect of in situ stresses with abrasivity. The results suggest that an increase in the porosity of a rock sample tends to decrease the sample’s resistance to wear and tear. This makes it easier to create a fracture (scratch) within such rock types. Therefore, porosity has a negative correlation with CAI.

532

N. N. Sirdesai et al.

4 Tool Wear Prediction Using CAI Several studies have been conducted to classify the abrasivity of rocks based on its CAI value. The studies suggest that the CAI values can be used for predicting the wear and tear of tool bits and pick consumption in mechanized excavators such as road-headers and TBMs [5, 13, 14]. For instance, Thuro and Plinninger [14] proposed a graphical representation to predict the rate of consumption of a 45-mm button drill bit as a function of CAI, as represented in Fig. 4(a). The study suggests that rocks with high abrasivity (≥3.0) have an adverse effect on the condition of drill bit. Further, the plot suggests that rocks belonging to the metamorphic and igneous variety tend to be highly abrasive when compared to sedimentary rock types. The rock samples used in this study have an average CAI of 2.92, thereby suggesting that if a 45-mm button drill bit were to be used for drilling into such rocks, the tool bit would have to be replaced after approximately ~300 m of drilling. Such correlations help in developing techno-economic models for any drilling operation. Further, Girmscheid [15] suggested a graphical correlation between the UCS and CAI to predict the wear and tear of the cutter bits of a road-header. Figure 4b illustrating the Girmscheid graphical correlation provides an insight on the cutter-bit lifecycle for a road-header and provides an estimation on the volume of loss upon the excavation of 1 m3 of fresh solid rock. Based on this correlation, if a road-header were to be used to excavate a tunnel in a granitic rock formation similar to the one used in this study, the magnitude of wear would be 0.3 picks per 1 m3 of excavation. This suggests that the pick needs to be replaced after excavating 3.3 m3 of rock. In order to estimate the loss of cutter-disc while excavating a tunnel using TBM, Gehring [16] suggested a 17-inch cutter-disc should be replaced if the loss of weight is about 3500 gm. In order to simplify the disc replacement protocol, Gehring [16] correlated the specific ring weight loss VS (mg/m-rolling distance) as function of the CAI, as illustrated in Eq. (7). VS = 0.74 × CAI1.93 (mg/m)

Fig. 4 Estimation of a 45-mm button drill bit lifetime and b cutter-pick wear

(7)

Impact of Rock Abrasivity on TBM Cutter-Discs During Tunnelling …

533

Based on this equation, the loss in weight for a 17-inch cutter-disc fitted onto a TBM which is excavating through a granitic rock formation similar to that analysed in this study would be 5.85 mg/m. This suggests that the cutter-discs should be replaced after roughly ~600 km of travel (rolling distance). It should be noted that the rolling distance travelled by the cutter-discs of a TBM depends entirely on their position and distance from the centreline. A disc at the peripheral section would cover larger rolling distance to that near the centreline for a unit length of excavation. This suggests that the peripheral cutter-discs would endure higher wear and tear than that of central discs. Similar to the example illustrated by Alber [1], if TBM with a 2.5-m cutter-head were to be used for constructing a tunnel within a granitic rock and if the advance per revolution is 1 mm, then the peripheral cutter-discs would have to be replaced after roughly ~40 m of advance. Such correlations help in estimating and planning the maintenance schedule for the equipment.

5 Conclusion In this study, the physico-mechanical properties of a granitic rock were correlated with the abrasive properties. The abrasivity of the rock was analysed by CERCHAR abrasivity method (CAI). The results suggest that CAI has a direct correlation with the strength properties and a negative correlation with porosity. This can be attributed to the fact that rocks with superior mechanical properties provide greater resistance to deformation. Additionally, porosity tends to decrease the mechanical strength, and therefore, porous rocks are easier to deform and are low abrasive in nature. Abrasivity can be used for the techno-economic analysis of several excavation processes. The CAI values of granitic rock samples analysed in this study were used to predict the loss and tool consumption of drill bits, road-header picks and TBM cutterdiscs. Further studies can be conducted to develop the pre-existent techno-economic models for Indian mining and geotechnical scenarios.

References 1. Alber M (2008) Stress dependency of the Cerchar abrasivity index (CAI) and its effects on wear of selected rock cutting tools. Tunn Undergr Space Technol 23(4):351–359 2. Käsling H, Thuro K (2010) Determining rock abrasivity in the laboratory. Rock mechanics in civil and environmental engineering—Proc EUROCK:425–428 3. West G (1989) Rock abrasiveness testing for tunnelling. Intl J Rock Mech Mining Sci Geomechanic Abs 26(2) 4. Alber M, Yaralı O, Dahl F, Bruland A, Käsling H, Michalakopoulos TN, Cardu M, Hagan P, Aydın H, Özarslan A (2013) ISRM suggested method for determining the abrasivity of rock by the CERCHAR abrasivity test. In: The ISRM suggested methods for rock characterization, testing and monitoring: 2007–2014. Springer, pp 101–106

534

N. N. Sirdesai et al.

5. Plinninger RJ, Kasling H, Thuro K (2004) Wear prediction in hardrock excavation using the CERCHAR abrasiveness index (CAI). In: Schubert (ed.) EUROCK 2004 & 53rd Geomechanics Colloquium, VGE, pp 1–6 6. ASTM (2008) D3967-08. Standard test method for splitting tensile strength of intact rock core specimen. ASTM International, West Conshohocken, PA. www.astm.org, https://doi.org/10. 1520/d3967-08 7. ASTM (2014) D7012-14E: standard test methods for compressive strength and elastic moduli of intact rock core specimens under varying states of stress and temperatures. Annual Book of ASTM Standards. ASTM International, West Conshohocken, PA 8. ASTM (2010) D7625-10, Standard test method for laboratory determination of abrasiveness of rock using the CERCHAR method, ASTM International, West Conshohocken, PA. www. astm.org, https://doi.org/10.1520/d7625-10 9. ISRM (1981) Rock characterization testing and monitoring—ISRM suggested methods. Pergamon Press for the Commission of Testing Methods 10. Tripathy A, Singh T, Kundu J (2015) Prediction of abrasiveness index of some Indian rocks using soft computing methods. Measurement 68:302–309 11. Deliormanlı AH (2012) Cerchar abrasivity index (CAI) and its relation to strength and abrasion test methods for marble stones. Constr Build Mater 30:16–21 12. Hamzaban M, Memarian H, Rostami J (2018) Determination of scratching energy index for Cerchar abrasion test. J Mining Environ 9(1):73–89 13. Plinninger RJ, Spaun G, Thuro K (2002) Predicting tool wear in drill and blast. Tunnels & Tunneling International Magazine:1–5 14. Thuro K, Plinninger R (2003) Hard rock tunnel boring, cutting, drilling and blasting: rock parameters for excavatability. In: Proceedings of the 10th ISRM Congress. International Society for Rock Mechanics and Rock Engineering 15. Girmscheid G (2000) Construction management and construction techniques in tunneling. Berlin Ernst and Sohn 16. Gehring K (1995) Leistungs-und Verschleißprognosen im maschinellen Tunnelbau. Felsbau 13(6):439–448

Tool Condition Prediction Using Acoustic Signal Processing and Learning-Based Methods Pranjali S. Deole

and Priya M. Khandekar

Abstract In the work presented here, signal processing techniques are used for condition monitoring of boring tool using acoustic signals. To accomplish this work, 108 experimental runs are performed to diagnose the tool life of the boring operation, by using two different sensor data. The acoustic signals are captured from the tool which is performing boring operation on mild steel materials. The acoustic signature being the main signal for study, vibration signals have also been measured for comparison. Four sets of boring experiments were performed. After the data had been collected for all runs of boring operation, signal processing methods are applied for doing the analysis of captured signal. The acoustic signals are captured from the boring tool and further analyzed by using learning-based methods for determining the condition of boring tool. Keywords Boring · Acoustic signals · Signal processing · Tool condition monitoring

1 Introduction The problem of tool failure like tool wear and tool breakage is becoming more and more important in the manufacturing industry. In metal cutting operation, such as boring, the cutting tool becomes dull after a certain working time and requires changing. The consequences of tool wear while machining may change the dimensions, tolerances, and surface finish of the part or product. So, measures for avoiding this must be taken in early stage before the tool breakage or it effects on the quality of the product in order to reduce cost and time [1]. Use of signals such as acoustic emissions and vibrations that are influenced by the boring operation can be used effectively for monitoring the condition of boring tool. The existing approaches may P. S. Deole (B) · P. M. Khandekar Shri Ramdeobaba College of Engineering & Management, Nagpur 440013, India e-mail: [email protected] P. M. Khandekar e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_64

535

536

P. S. Deole and P. M. Khandekar

not be capable of predicting the tool life and monitoring the tool health effectively [2]. This necessitated the need for an in-process tool condition prediction approach, which may be used while the machining process is going on and which is based upon the signals that are believed to be influenced by the machining process [4]. So, an approach to determine the condition of boring tool using acoustic signals is presented here.

2 Methodology Four sets of boring experiments were performed. After the data had been collected for all runs of boring operation, signal processing methods are applied for doing the analysis of captured signal. Figure 1 shows the graph between tool life (mm2 ) and tool wear (gm). The tool wear is measured as a loss of weight during each boring operation run. The tool life is measured in terms of the circumference of the workpiece machined. The circumference is given as
DL where the length of the work piece is fixed which is taken as 125 mm. The diameter is getting changed after machining so the tool life is taken in terms of the circumference of the workpiece getting bored. It is observed from the graph that as the tool wear is increasing, there is a spontaneous decrease in the life of tool which is true for all four sets of experiments. The experiments are conducted to collect the data of acoustic and vibration signals, and the corresponding output of weight loss of tool and roughness of the surface machined is recorded [3]. It was found that the tool is worn out at experiment no. 39. This is found from the fact that there was a reduction of 0.05 gm of the total weight of the cutting tool. Figure 2 represents the tool wear and the tool worn out. It can be observed from the graph that tool worn out took place at 39th experimental run.

Fig. 1 Experimental setup for developing TCM using acoustic signals

Tool Condition Prediction Using Acoustic …

537

Fig. 2 Remaining useful life versus tool wear

As the same insert was used to carry out all the experimental runs, the insert started wearing out after carrying out experimental runs. As a consequence of which the remaining useful life of the cutting tool started reducing. Figure 2 shows that when the tool wear reduces, the remaining useful life of the tool increases gradually. The red mark in the graph shows that there is a failure of tool at that point.

3 Processing of Signals Acquired During Experimentation The acoustic signal [6] data captured during boring operation is processed using an appropriate ANN signal processing technique. Vibration signals are too acquired and processed for comparison in the later part of this research [5]. The acoustic signals captured through microphone are analyzed [3]. The raw data undergoes de-noising, amplification, and conditioning for effective analysis of the signal.

3.1 Selection of Appropriate Wavelet for De-noising The wavelet families that are selected and optimal wavelet de-noising parameters are estimated based on SNR criterion. These wavelet families represent the comparison of acoustic signal de-noising performance based on different levels as shown in Fig. 3.

538

P. S. Deole and P. M. Khandekar

Fig. 3 Comparison of acoustic signal de-noising performance based on different wavelets

3.2 Feature Extraction from Acoustic Signals The preprocessed signal acoustic signals contain a large volume of information that needs to be broken into meaningful information. The information so extracted is the feature of the signal and is generally processed in the time domain to obtain its statistical characteristics, in the frequency domain to analyze the power of signal distribution over a range of frequencies, and in the time–frequency domain to obtain spectrum of non-stationary signals. The time domain features which contain important information are extracted from the signal. As the time domain features reflect only changes in signal with respective time, there is a need to extract features in the frequency domain also. Every de-noised acoustic signal is further processed using wavelet packet transform to extract the time–frequency domain features. The signal was decomposed into 6 levels, and 64 nodal points were determined using WPT toolbox of MATLAB. Coefficients of these nodal points were further estimated. For every array of coefficients, features estimated are logarithmic energy, skewness, kurtosis, threshold crossing rate, pulse width, and effective value.

4 Key Feature Selection for Acoustic Emissions 52 features are extracted from every de-noised acoustic signal in time, frequency, and time–frequency domains. It is desirable that the features should be optimized according to a criterion such as Pearson’s correlation coefficient (PCC), R-square statistical methods, and Pearson’s chi-squared test. The features for which PCC is greater than 0.9, R squared statistics is greater than 0.9 and Pearson chi squared test for 95% significance level are selected. The features of effective value are extracted

Tool Condition Prediction Using Acoustic …

539

for the said three methods. The features from Pearson’s correlation coefficient are selected for further analysis, as the Pearson’s correlation coefficient is the best method for feature selection. Hence, 12 features for acoustic emissions are selected.

5 Model Developed for Proposed Approach For developing a model for tool life estimation and estimating remaining useful life of the tool, the artificial neural network is used [7, 8]. There are a wide variety of networks depending on the nature of information processing carried out at individual nodes, the topology of the links, and the algorithm for adaptation of link weights. Multilayered perceptron (MLP) neural network and radial basis function (RBF) neural networks are used in the present work for developing the model for tool life and remaining useful life estimation [7]. There are 17 total inputs to neural network, i.e., cutting speed, feed rate, depth of cut, hardness of material, tool slenderness ratio, and 12 features of acoustic emission. The expected output of neural network is surface roughness, tool wear, and remaining useful life of tool. Table 1 shows the parameters used for neural network for both multilayer perceptron (MLP) and radial basis function (RBF). By doing the error comparison between MLP and RBF for different error performances, it is found that R-square prediction is higher, i.e., 0.9861 in RBF. So, RBF is best suited for the analysis of the acoustic signals. Table 1 Parameters used for neural network Parameters Learning rate

MLP 0.1

RBF 0.1

Learning algorithm

Levenberg–Marquardt

K-mean clustering + backpropagation

No. of layers

5 (1 input–3, hidden–1 output layer)

3 (one in each layer)

No. of neurons in each layer

17 neurons—input layer 60,63,65 neurons—hidden layer 3 neurons—output layer

17 neurons—input layer 40 neurons—hidden layer 3 neurons—output layer

No. of samples for training

108

108

No. of samples for testing

18

18

No. of samples for validation (fresh set of experiments were performed by changing parameters)

27

27

540

P. S. Deole and P. M. Khandekar

Fig. 4 Measured tool wear versus predicted using MLP and RBF

6 Experimental Validation For validation of developed neural network model, a fresh set of experiments were conducted and performance parameters were measured. Figure 4 gives the performance of MLP and RBF for measured tool wear and the one predicted with MLP and RBF. The validation of the proposed model is done by performing a new set of experiments with some fixed set of process parameters to observe the prediction of tool wear using the two ANN approaches.

7 Result and Conclusion The average error in the prediction of tool wear is 0.0334 (gm) for MLP and 0.021 gm in RBF, whereas the percent average error in the prediction of tool wear is 1.58 in MLP and 0.0168 in RBF. So, the RBF ANN architecture is preferred over MLP. It is found that there is a similar trend in the change of surface roughness (microns) measured for validation experimental runs for RBF neural network. It is observed from the validation experiments clearly that MLP is not a good performing neural network for surface roughness. The features proved good enough to predict the tool condition and remaining useful life of the tool with radial basis function neural network being more accurate learning-based method for determining the condition of boring tool.

Tool Condition Prediction Using Acoustic …

541

References 1. Schneider G (2010) Cutting tool applications chapter 10: boring operations and machines. Boring increases the inside diameter of a hole, and achieves three things: sizing, straightness, and concentricity. American Machinist. 27 July 2. Singh S (2011) Handbook of mechanical engineering, 2nd edn. S. Chand publication, pp 807–885 3. Tana CK, Irvinga P, Mba D (2007) A comparative experimental study on the diagnostic and prognostic capabilities of acoustics emission, vibration and spectrometric oil analysis for spur gears. Mech Syst Sig Proc 21:208–233 4. Dongre PR, Chiddarwar SS (2013) Tool condition monitoring in various machining operations and use of acouastic signature analysis. Int J Mech Eng Robot (IJMER) 1(1). ISSN (Print): 2321–5747 5. Orhan S, Osman A, Camus-cu N, Aslan E (2007) Tool wear evaluation by vibration analysis during end milling of AISI D3 cold work tool steel with 35 HRC hardness. NDT and E Int 40:121–126 6. Mathew MT, Pai PS, Rocha LA (2008) An effective sensor for tool wear monitoring in face milling: acoustic emission. Sadhana 33(3):227–233 7. Xu C, Fan X, Luo W (2009) Research of tool wear based on radical basic function network. In: Proceedings of the international multi conference of engineers and computer scientists, vol 1, March 18–20 2009. Hong Kong, pp 1–6 8. Zhang C, Yao X, Zhang J, Jin H (2016) Tool condition monitoring and remaining useful life prognostic based on a wireless sensor in dry milling operations. Sensors 16(6–795):1–20

Finite Element Simulation of Ballistic Response of Metallic Sandwich Structures with Aluminium Foam Core Nikhil Khaire , Vivek Bhure, and Gaurav Tiwari

Abstract Acknowledging the huge applications and growing interest in understanding the behaviour of sandwich structure utilizing metal foams as a core material, the present study analysed numerically the ballistic performance of sandwich structures. Deformable plates of Aluminium 5005-H34 alloy material of thickness of 0.60 mm with deformable core made of CYMAT foam of 25 mm thickness and constant span of 100 mm were considered for numerical analysis against normal impact of rigid hemispherical projectile of length (L) 25.50 mm and diameter (d) 7.50 mm using ANSYS/LS-DYNA. Utilizing isotropic constitutive Deshpande–Fleck material model and orthotropic honeycomb material model capable of simulating fully anisotropic behaviour of foams, the foams were modelled using available data from the literature. Residual velocity, ballistic limit, energy absorption and failure mode were analysed. The developed model was successful in capturing the physical phenomenon using complex material models and obtained numerical results were in good agreement with the observed results from the available literature survey. Keywords Ballistic performance · Foam · Sandwich structure

1 Introduction The performance of sandwich structure under impact loading has been studied widely [1–6] and characterization of their behaviour with varying conditions of ballistic impact, viz. configurations of individual components, boundary conditions, thickness, shape and size of the targets, and nose shape, incidence angles, size, impact velocity of projectiles are reported. The output parameters such as residual velocity, ballistic limit, failure modes and deformations observed locally and globally and damage initiation in sandwich structure; work and nature of work amount absorbed in the components are also reported in the literature. The parameters specific to the sandwich structure such as stress versus strain characteristics, load versus indenter N. Khaire (B) · V. Bhure · G. Tiwari Visvesvaraya National Institute of Technology, Nagpur 440010, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_65

543

544

N. Khaire et al.

depth and shock enhancement under varied rates of loading have also been presented in the literature. These cellular structures have been studied using all three techniques experimental, numerical and analytical/empirical methods. Zhao et al. [1] experimentally investigated the performance of sandwich panels against hemispherical projectile by carrying out inverse perforation tests where instead of free-flying projectile, the sandwich is made to impact on a stationary perforator. Hou et al. [2] experimentally determined the performance of sandwich panels with Al5005-H34 alloy face sheets and CYMAT foam cores against projectiles with hemispherical, conical and flat-shaped projectile. Effect of core density, projectile nose shapes, impact velocities of the projectiles and thickness of face sheets and core were studied. Hanssen et al. [3] studied the effect idealized bird strike on double sandwich panels, and LS-DYNA code was employed to carry out impact on sandwich panels. A model based on continuum damages theory was employed for the panel sheets, and phenomenological Deshpande–Fleck model was employed for the foam core. Mohan et al. [4] suggested that the load vs. penetration depth of the indenter revealed that foam material provides resistance by cell wall crushing under the impactor nose and tearing of cell wall near the edge of the impactor. Increasing the thickness of foam and incorporating face plate increased energy absorption of the sandwich panels. Elnasri et al. [5] calculated the top piercing force required for top face sheet penetration by RPPL shock model and found to be good agreement with experimental result. Su et al. [6] studied ballistic response of curved sandwich shells under normal/oblique impact including different curvatures, projectile nose shapes and angle of impact of the projectile. Graded foam core and thicker face sheets were proven to increase the ballistic resistance of the sandwich structure. For aluminium foams and sandwich structures, it is evident that complete finite element analysis considering complex behaviour of sandwich structure under impact conditions requires a detailed constitutive model for both face plates and core as based on the review of the literature. Hence, this study attempted to combine the two detailed constitutive models: Johnson–Cook material model for face plates and Deshpande– Fleck foam material model for core. The focus of this study is to describe the ballistic performance against projectiles under different conditions of impact.

2 Finite Element Modelling Deformable plates of Aluminium 5005-H34 alloy material of thickness of 0.60 mm with deformable core made of CYMAT foam of 25 mm thickness and constant span of 100 mm were considered for numerical analysis against normal impact of rigid hemispherical projectile of length (L) 25.50 mm and diameter (d) 7.50 mm. Figure 1 shows the geometrical details of individual parts. The face plates, core and projectile were meshed by using 8-noded brick elements. To save computational effort, only quarter of the sandwich was considered for the impact. The region near the impact zone was meshed finer when compared to the outer region towards periphery. The central circular region of radius 10 mm was meshed by keeping aspect ratio of solid

Finite Element Simulation of Ballistic Response …

545

3.75 mm

25.50 mm

(c) (a) (d)

(b)

Fig. 1 Finite element model for parametric study: a assembly of the parts, b hemispherical projectile, c enlarged view of primary impact zone, d number of elements over thickness

Table 1 Material constants for Al 5005-H34 [7] (MAT_SIMPLIFIED_JOHNSON_COOK) Elastic constants and density E (GPa)

ν

ρ

65.00

0.3

2700

(kg/m3 )

Yield stress and strain hardening

Strain rate hardening

A (MPa)

B (MPa)

n

ε0 °(s−1 )

C

147

60

0.9

0.001

0.003

elements close to 1. The sides of elements were kept as 0.50 mm in the primary impact zone and across the thickness of 1 mm; 6 elements were taken as shown in Fig. 1d for face plates. The core was modelled with the same element size, and 21 elements were taken across the thickness. The total elements in the primary impact zone for each layer were 246 for face plate while 5376 for the core. Material model used for the face sheet and foam has been shown in Tables 1 and 2.

3 Result and Discussion 3.1 Residual Velocity The sandwich target of face plate thickness of 0.60 mm and core thickness of 25 mm was impacted with hemispherical projectile, and residual velocities upon impact at various impact velocities are given Table 3 and Fig. 3. The numerical values are presented and compared with the experimental values. It is observed that higher impact velocities, the numerical model is in good agreement with the experimental results while the model numerically over predicts the residual velocities at lower impact when compared to the experimental values (Fig. 2 and Table 4).

ν

0.30

E (GPa)

0.890

486

ρ (kg/m3 ) 2.12

α 139

γ (MPa) 0.35

εD 56

α2 (MPa)

Table 2 Material constants for 18% CYMAT foam [5] (MAT_154_DESHPANDE_FLECK_FOAM) 2.32

β

11.3

σ p (MPa)

0.1

εcr

546 N. Khaire et al.

Finite Element Simulation of Ballistic Response …

547

(a)

(b)

(c)

(d)

Fig. 2 Residual velocity variation with impact velocity and perforation energy. a and b indicate experimental and numerical values obtained for bare plates and core for residual versus impact velocity and perforation energy respectively. c and d indicate experimental and numerical values obtained for sandwich structure for residual versus impact velocity and perforation energy, respectively [2]

3.2 Ballistic Limit The ballistic limit for sandwich structure and bare individual parts was obtained by both experimental and numerical methods and is given in Table 4. As indicated by Fig. 3, the simulation gives slightly lower value for ballistic limit when compared with the experimental results. Excellent agreement is observed for bare plate and foam core, whereas for full sandwich structure, the numerical simulation gives 4.3% lower ballistic limit than calculated by the experimental values.

548

N. Khaire et al.

Table 3 Experimental and numerical results for sandwich structures individual parts Target configuration

Impact velocity (m/s)

Residual velocity (m/s)

Perforation energy (J)

Experimental [2]

Numerical

Experimental [2]

Numerical

Sandwich

186.34

143.34

142.73

49.28

50.45

181.82

139.65

140.07

47.87

47.24

174.22

124.29

128.86

54.99

50.72

157.73

113.30

113.36

44.29

42.33

106.31

46.84

50.07

36.47

35.21

98.88

24.08

27.90

37.48

31.63

77.76

0

0

24.79

27.90

Face plate only

105.93

102.23

102.39

3.20

3.06

Core only

103.45

65.85

68.08

26.42

25.18

Diameter = 100 mm, Hemispherical projectile (Length = 25.50 mm, Diameter = 7.5 mm) Fig. 3 Ballistic limit obtained by experimental and numerical methods [2]

Table 4 Experimental and numerical ballistic limit results for various configurations

Target configuration

Ballistic limit Experimental [2]

Numerical

Sandwich

105

100.50

Face plate only

27.76

27.15

Core only

79.79

77.90

Finite Element Simulation of Ballistic Response …

549

4 Conclusion A finite element model was prepared to carry out the validation and understand the behaviour of sandwich structures, and numerical results were compared against experimental results available from literature and following were the major findings of the study: • It is observed that higher impact velocities, the numerical model is in good agreement with the experimental results, and a difference of 0.05% was observed while the model numerically over predicts by the maximum difference of 6.9% between the residual velocities at lower impact velocities when compared to the experimental values. • The prepared numerical model was able to successfully capture the physical phenomenon, and failure modes were in good agreement with the experimental observations. The ballistic limit obtained numerically was compared against the experimental values. A good agreement was observed for bare plates, and bare foams, while there was a difference 4.50 m/s between experimental and numerical ballistic limit which accounts for 4.3% error. Acknowledgements This research was fully supported by SERB/DST under project number DST/SERB ECR/2016/001440 for providing resources for numerical simulations and experimentation. We thank our colleagues from Visvesvaraya National Institute of Technology, Nagpur, who provided insight and expertise that greatly assisted the research.

References 1. Zhao H, Elnasri I, Girard Y (2007) Perforation of aluminium foam core sandwich panels under impact loading—An experimental study. Int J Impact Eng 34(7):1246–1257 2. Hou W, Zhu F, Lu G, Fang DN (2010) Ballistic impact experiments of metallic sandwich panels with aluminium foam core. Int J Impact Eng 37(10):1045–1055 3. Hanssen AG, Girard Y, Berstad T, Langseth M (2006) A numerical model for bird strike of aluminium foam-based sandwich panels. Int J Impact Eng 32(7):1127–1144 4. Mohan K, Yip TH, Idapalapati S, Chen Z (2011) Impact response of aluminum foam core sandwich structures. Mater Sci Eng 529:94–101 5. Elnasri I, Zhao H (2016) Impact perforation of sandwich panels with aluminum foam core: a numerical and analytical study. Int J Impact Eng 96:50–60 6. Su B, Zhou Z, Zhang J, Wang Z, Shu X, Li Z (2015) A numerical study on the impact behavior of foam-cored cylindrical sandwich shells subjected to normal/oblique impact. Lat Am J Solids Struct 12:2045–60 7. Bendarma A, Jankowiak T, Łodygowski T (2014) Experimental and numerical analysis of the aluminum alloy AW5005 behavior subjected to tension and perforation under dynamic loading. Theor Appl Mech. 55(4):1–15

Crushing Behavior of Thick Circular High Strength Aluminum Tube Against Quasi-static Axial Loading Vivek Patel , Sanket Suresh Kalantre, Gaurav Tiwari , and Ravikumar Dumpala

Abstract The present study investigates the crushing performance of circular tube, made of high strength aluminum alloy AA-7005 and AA-7075 by numerically and experimentally. The tubular structures were exposed to quasi-static axial loading with variation in length (51, 68 and 85 mm) while diameter (34 mm) and thickness (3.55 mm) kept constant. The uni-axial tension tests for both the material were carried out to explore the behavior of stress–strain which was used as input for numerical simulations. The quasi-static compression test was conducted on Instron compressive testing machine while for simulation finite element code LS-Dyna was used. The crashworthiness parameters such as initial maximum peak load, crash force efficiency, energy absorption capacity, and specific energy were found from the obtained deformation behavior of structures. It is found that the tubular structures made of AA-7005 show higher crash force efficiency whereas the structures made of AA-7075 absorb significantly higher energy during the collapse. Keywords Thick tubular structures · Crashworthiness · Quasi-static axial loading · LS-Dyna

1 Introduction The vehicles are now being widely used for transport purposes. Increasing number of such vehicles has led in a rise in number of crashes, reflecting a greater safety requirement. This caused the designer to explore the energy absorbing component that mitigates the effect of crash or collision by absorbing the impact energy. The energy absorbers are designed in such way that it undergoes progressive collapse in order to absorb maximum amount of impact energy. Concertina mode of deformation reflect a good characteristic for a thin wall tube structures compare with the diamond and mixed mode. Starting phase of research suggested some theoretical model for V. Patel (B) · S. S. Kalantre · G. Tiwari · R. Dumpala Department of Mechanical Engineering, Visvesvaraya National Institute of Technology, Nagpur, Maharashtra 440010, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_66

551

552

V. Patel et al.

(a)

(b)

Fig. 1 True stress–strain plot of: a AA-7005 and b AA-7075

both concertina mode [1] applicable for a small crushing length and diamond mode [2] relevant for a large deformation. Earlier lots of studies [3–5] have been done on conventional circular thin walled tubular structure. Other than circular, researchers [6–10] have explored different cross-sections such as triangles, squares, rectangles and polygons. In actual crash incident, the structures collapse under different loading condition like axial, lateral or bending. In such crash event the impact is dynamic during which large distortions occur in a short time. The study of thick tubular structures is limited. In this paper, a thick circular tubular structure made of aluminum alloy AA-7005 and AA-7075 is introduced. The crashworthiness performance has studied in terms of deformation modes and loaddisplacement curves against quasi-static loading by experimentally and numerically.

2 Experimental Details 2.1 Tensile Tests The AA-7075 and AA-7005 tubes were used for experiment and numerical test. To obtain material properties, a tensile test was performed on the samples that were cut from the tube as per ASTM-E8 standard. Figure 1 shows the true stress–strain curve.

2.2 Compression Testing The specimens of the required dimension were cut from both grade of high strength aluminum tubes, see Fig. 2. The prepared specimen having three height variations (51, 68 and 85 mm) while diameter and thickness kept constant as 34 mm and 3.55 mm respectively.

Crushing Behavior of Thick Circular High Strength …

553

Fig. 2 Specimens for testing having height: a H = 85 mm, b H = 68 mm and c H = 51 mm

Table 1 Specification of specimen Specimen id

Material

Height (mm)

Diameter (mm)

Thickness (mm)

AA-05-01

AA-7005

51

34

3.55

AA-05-02

68

AA-05-03 AA-75-01

85 AA-7075

51

AA-75-02

68

AA-75-03

85

Table 1 shows the specification of all specimens. The complete set-up for compression testing machine with a load capacity of 5000 kN is shown in Fig. 3a. The lower part of machine is fixed while upper part moving downward with a constant rate of quasi-static loading of 6 mm/min.

3 Numerical Modeling The finite element code LS-Dyna [11] is used to investigate the crushing performance of thick tubular structures. The complete part of the numerical set-up shows in Fig. 3b. The modeled tubular structure compressed between two rigid plates. The bottom plate was fixed while upper plate moving axially downwards at a rate of 6 mm/min. The circular tube was taken as a solid element of EQ-2, having an element size of 1 mm. The piecewise linear plasticity material model was assigned to the tube, while the upper and lower plates were allocated as rigid properties. The contact between the tube and plate taken as automatic surface to surface while for tube, automatic single surface was assigned in order to minimize the effect of interpenetration. The friction

554

V. Patel et al. (a)

(b)

Fig. 3 Testing arrangement: a Experimental set-up and b numerical set-up

coefficients 0.3 and 0.2 considered for the static and dynamic friction, respectively, between two contact surfaces.

4 Result and Discussion 4.1 Crashworthiness Parameters The load-displacement curve describes the structural behavior during the collapse. The all crashworthiness parameters [Peak Force (PF), Energy absorption (E a ), and Crash Force Efficiency (CFE)] defining the feasibility of the tubular structure are calculated on the basis of load-displacement curve. Figure 4 shows the loaddisplacement curve of tubular structure made of AA-7005 and AA-7075 for all six specimens. It can be observed that PF values for AA-7075 specimens are higher than that of AA-7005 specimens. The higher thickness of specimens resulted in the formation of only one-fold while compressing. The PF indicates the resistance offered by structure at initial state of collapse. After attained maximum value it decreases as the plastic hinge formed. The obtained graph from both experimental and numerical observation was approximately close that indicate the correctness of numerical models. When compressing the structure, it absorbs energy (E a ) that is calculated by incorporating region under the load-displacement curve. The obtained value of experimental and numerical are compared in Table 2. The percentage error for peak load varies 0.4–14% while for energy it is in the range of 0.4–16%. The specific energy absorption (SEA) is defined as the amount of energy per unit mass absorbed by specimens

Crushing Behavior of Thick Circular High Strength …

555 (b)

(a)

AA-05-01

AA-05-02

AA-05-03

AA-75-01

AA-75-02

AA-75-03

Fig. 4 Load-displacement plot of: a AA-7005 and b AA-7075 tubular structures

and it should be higher. The fracture failure was observed in AA-75-03 specimens that reduce the structures’ ability to absorb impact energy. The CFE, defined as the ratio of mean force to the peak force, value close to unity is desirable. It can be observed that CFE value for AA-7005 specimens is higher than that of AA-7075 specimens.

556

V. Patel et al.

Table 2 Load comparison of samples that tested experimentally and numerically Sample

PF (kN)

Error %

Exp.

Num.

AA-05-01

129.89

140.98

AA-05-02

147.87

170.02

AA-05-03

154.69

179.12

13.01

AA-75-01

204.75

210.12

AA-75-02

219.38

245.48

AA-75-03

234.04

236.42

0.46

(a)

E a (J)

Error %

SEA (J/kg)

CFE

Exp.

Num.

7.86

3120.47

3426.35

8.92

72.9

0.86

13.63

4026.09

4239.07

5.02

67.28

0.78

4550.42

4568.39

0.42

58.57

0.77

2.55

3705.08

3814.87

2.87

81.16

0.80

10.63

4625.77

4902.85

5.65

77.82

0.70

2867

2469

16.02

31.65

0.93

(b)

(c)

Fig. 5 Compressed specimen: a AA-7005, b AA-7075 and c compared deformation mode obtained through experimental and simulation

4.2 Deformation Mode For axial compression concertinas are the best deformation mode for higher energy absorption. But it is only observed in very thin circular columns. With a change in geometry and thickness, deformation mode changes to diamond folds. Figure 5 shows the deformation behavior of tested samples. The double folding happens for shorter height while as an increase in height it deforms from the middle that led to buckling. Fracture failure was observed in AA-7075 specimens.

5 Conclusion Crashworthiness performance of AA-7005 and AA-7075 tubes subjected to axial loading was investigated experimentally and numerically. The concluding remarks are as follows:

Crushing Behavior of Thick Circular High Strength …

557

• Increase in peak crushing force was observed for increase in length of specimens. For AA-7005 specimen, the peak load increased by 19.09%, when length increased from 51 to 85 mm whereas for AA-7075 it increases by 14.71%. • The crush force efficiency of AA-7005 specimens was consistently better than the AA-7075 specimens. • The total amount of energy absorbed by AA-7075 specimens was 16.81% higher than AA-7005 specimens of the same dimensions. • AA-7005 specimens had a complete plastic collapse, while AA-7075 specimens had a plastic collapse, resulting in an undesirable fracture failure. • Although AA-7075 has a higher strength than AA-7005, the tubular structure made of AA-7005 performs better than the structure made of AA-7075.

References 1. Alexander JM (1960) An approximate analysis of the collapse of thin cylindrical shells under axial loading. Q J Mech Appl Math XIII:10 2. Pugsley SA, Macaulay M (1960) The large-scale crumpling of thin cylindrical columns. Q J Mech Appl Math 8:1 3. Andrews KRF, England GL, Ghani E (1983) Classification of the axial collapse of cylindrical tubes under quasi-static loading. Int J Mech Sci 25:687–696 4. Jones N, Abramowicz W (1985) Static and dynamic axial crushing of circular and square tubes. In: Proceedings of symposium on metal forming and impact mechanics, Pergamon Press, Oxford, p 225 5. Karagiozova D, Jones N Dynamic effects on buckling and energy absorption of cylindrical shells under axial impact. Thin Wall Struct 6. Reddy et al (2015) Multi-cornered thin-walled sheet metal members for enhanced crashworthiness and occupant protection. Thin Wall Struct 94:56–66 7. Abbasi et al (2015) Multiobjective crashworthiness optimization of multi-cornered thin-walled sheet metal members. Thin Wall Struct 89:31–41 8. Reyes A, Langseth M, Hopperstad OS (2003) Square aluminum tubes subjected to oblique loading. Int J Impact Eng 28(10):1077–1106 9. Rossi A, Fawaz Z, Behdinan K (2005) Numerical simulation of the axial collapse of thin-walled polygonal section tubes. Thin Wall Struct 43(10):1646–1661 10. Zhang X, Huh H (2010) Crushing analysis of polygonal columns and angle elements. Int J Impact Eng 37(4):441–451 11. Hallquist J (2006) L.-D.K.U.s. Manual, Livermore Software Technology Corporation, Google Scholar, 2007

Estimation of Burr Dimensions Using Image Processing for Robotic Deburring Rohini Y. Bhute and M. R. Rahul

Abstract This paper discusses an effective method to identify the burr location for 2D and 3D workpiece. The proposed method uses an image processing technique to estimate the dimension of the burr from an image. A set of image processing algorithms are developed for estimating the location as well as the dimension of the burr. The burr dimensions are verified using coordinate measuring machine. The burr data generated are used for planning the deburring trajectory. ABB IRB 120 robot is used to validate the result experimentally. The result clearly shows the effectiveness of this approach. Keywords Burr recognition · Image processing · Deburring

1 Introduction A burr is a raised edge or an unwanted piece of material that has to be removed by the process called deburring. Though there are conventional methods to remove the burr, automating the deburring process is still a challenging task. The main challenge for automating the deburring process is the identification of burr location and its dimension. Though locating the burr is critical, very few efforts are reported in the literature. One such attempt by Ryuh et al. [1] uses parametric modeling for extracting burr parameter from the original workpiece. However, this method was computationally time-consuming. To overcome this, Li et al. [2] generated trajectory for robotic-assisted grinding by automatically scanning the workpiece. However, scanning has to be done every time the product changes. To overcome the above drawbacks, Jinno et al. [3] and Princely and Selvaraj [4] proposed a programming method using computer vision and force feedback. However, estimation of burr R. Y. Bhute (B) Dassault Systèmes Solutions Lab, Pune 411057, India e-mail: [email protected] M. R. Rahul Visvesvaraya National Institute of Technology, South Ambazari Road, Ambazari, Nagpur, Maharashtra 440010, India © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_67

559

560

R. Y. Bhute and M. R. Rahul

dimensions from the image data remains a challenge. In view of this, Idaku [5] developed a method to acquire three-dimensional image from an advanced structure light projection at a high frame rate. However, this method requires a sophisticate camera setup which makes it impractical to implement in an industrial scenario. The major drawback of these methods is: (1) The burr location and its dimensions cannot be estimated. (2) The deburring trajectory planning has to be done manually. Therefore, these methods are not effective in finding the burr location and its dimensions. In this paper, identification of burr location and burr dimension estimation for 2D and 3D workpiece is discussed. The outline of the paper is as follows: Sect. 2 discusses the methodology for burr dimension estimation. Section 3 discusses the trajectory planning for robotic deburring, its simulation, and experimental validation. The result of the proposed method is discussed in Sect. 4. Finally, Sect. 5 discussed the conclusion drawn from the work.

2 Methodology The proposed method consists of two phases: In phase one, burr thickness estimation for 2D workpiece is done and phase two estimates the burr thickness and height for a 3D workpiece. The images are captured using a CCD camera placed at a fixed height above the workpiece. In the first phase, a subtraction algorithm and thickness algorithm are developed to extract the burr details. In the second phase, the image from a mirror setup is employed to estimate the thickness and height of the burr. The estimated burr data points are used to generate the deburring trajectory. ABB RobotStudio software is used for simulating the deburring trajectory. The simulated deburring path is verified on ABB IRB 120 robot with a dummy tool. The workflow diagram for the proposed method is shown in Fig. 1.

2.1 Burr Thickness Estimation for 2D Workpiece The thickness of burr for 2D workpiece is estimated by using two algorithms, namely the subtraction algorithm and thickness calculation algorithm. The subtraction algorithm identifies the burr present in the workpiece, whereas the thickness calculation algorithm estimates the burr thickness in pixel values. The subtraction algorithm uses two images, a masterpiece image and a machined piece image as shown in Fig. 2. The images captured using a CDD camera is cropped, filtered, and resized to make them suitable for subtraction. The masterpiece image is subtracted from the machined piece image, which will give data of the burr part. These data are used further for calculating burr thickness. Once the burr is identified, the burr thickness is estimated using thickness algorithm. The algorithm uses a novel unit matrix search method to estimate the burr thickens in terms of pixel values. The pixel values are calibrated to get the actual

Estimation of Burr Dimensions Using Image …

561

Fig. 1 Algorithm for the proposed method

value of the burr in mm. The burr image is converted to binary image such that ‘1’ represents the presence of burr and ‘0’ represents the standard part or no burr. It is observed that the thickness of the burr is equivalent to a maximum size of the order of unit matrix found in the binary image of the burr part. The results generated by this method are validated by measuring the thickness of the same burr on CMM.

562

R. Y. Bhute and M. R. Rahul

Fig. 2 Subtraction technique

2.2 Burr Thickness and Height Estimation for 3D Workpiece A mirror setup as shown in Fig. 3a is used to estimate the burr thickness and height of the workpiece. The setup consists of four mirrors placed at 45°, which help in taking side images of the burr specimen. Thus, an original three-dimensional workpiece can be transformed into a two-dimensional image. The image captured using CDD camera is sectioned into five parts namely Lt (Left), Rt (Right), Up (Upper), Lo (Lower), and C (Central) as shown in Fig. 3b. Up, Lo, Lt, and Rt part of the image help in estimating the burr height and part C in the image helps in estimating the thickness of the burr using subtraction and thickness algorithm.

Fig. 3 a CAD model of 3D setup. b Image from CCD camera

Estimation of Burr Dimensions Using Image …

563

3 Trajectory Planning—Simulation and Experiment The main objective of this phase is to select the data points of the burr part for robotic deburring. Once the thickness of the burr is estimated for 2D workpiece, a systematic approach for data point collection is carried out for trajectory planning. To determine the number of passes (n) for robotic deburring, thickness of the burr is divided by the radii of the tool. From the reference edge, all the X-coordinate and Y-coordinate for the ‘n’ passes are stored and an optimized trajectory is generated for deburring process. The thickness and height data are used to plan the trajectory for 3D workpiece. Prewitt edge detection algorithm is used to detect the inner edge of the workpiece. Then, the number of passes for the deburring is determined by considering the diameter of the tool. However, multiple edges are detected for 3D workpiece. To find the optimal deburring trajectory, various correcting techniques like backtracking coordinate elimination, cure smoothening, and curve fitting are carried out. This ensures the accuracy of the planned trajectory. The data points generated for deburring for both 2D and 3D workpiece are with respect to the camera coordinate system. These data points are transferred to robot coordinate system using a transformation matrix. The transferred data points are given to ABB RobotStudio software for simulation. The simulation shows that the generated trajectory is tracked accurately by the robot. The verified data points are converted into RAPID code, which can be directly given to ABB IRB 120 robot controller. To test the feasibility of the proposed method, experimental validation using dummy tool is carried out on ABB IRB 120 robot. The experimental setup consists of a CCD camera, MATLAB for data processing, RobotStudio software for simulation and RAPID code generation, and ABB IRB 120 robot. The workpiece was fixed on a dummy fixture, and the deburring process was carried out. The experiment shows that the robot tracked the deburring trajectory accurately.

4 Result and Discussion To estimate the burr thickness and location, various existing methods like pattern matching, canvas overlapping, and edge detection techniques were used. The pattern matching method matches the same intensity points of the two images but fails to match the correct points, whereas canvas overlapping overlaps the two image and identifies the unmatched part in the image, and however, this method failed as it replaced the old canvas with a completely new canvas. The third method, edge detection technique, identified the complete edges and provided good information of burr but was insufficient to calculate the burr thickness. The subtraction and thickness algorithm identified the burr as well as estimated the burr thickness from the image

564

R. Y. Bhute and M. R. Rahul

Fig. 4 Burr identification by image processing

Fig. 5 Trajectory tracked by the robot

(Fig. 4). The burr dimensions estimated were verified using coordinate measuring machine, and the average error of 0.8 mm was observed in the dimension. The simulation result using ABB RobotStudio clearly shows that the robot tracked the deburring path accurately (Fig. 5).

5 Conclusion The proposed method discusses an effective technique to identify the burr location and estimate its dimension for 2D and 3D workpiece. The subtraction algorithm, thickness algorithm, and height algorithm can effectively estimate the burr dimensions. An effective trajectory planning for the deburring process can be done using the data points generated by the algorithms proposed in this paper. The simulation and experimental results clearly show the effectiveness of the proposed method.

Estimation of Burr Dimensions Using Image …

565

References 1. Ryuh B-S, Pennock GR (2006) Robot automation systems for deburring. In: Industrial robotics: programming, simulation and applications, p 15 2. Li CJ, Park CH, Kyung JH, Chung GJ (2011) Study on teaching path reconstruction algorithm based direct teaching and playback method. In: Eighth international conference on ubiquitous robots and ambient intelligence (Incheon 2011) 3. Jinno M et al (1999) Teaching-less robot system for finishing workpieces of various shapes, Kyongju, South Korea, 17–21 Oct 1999 4. Princely FL, Selvaraj T (2014) Vision assisted robotic deburring of edge burrs in cast parts. In: 12th Global Congress on Manufacturing and Management (GCMM) 5. Idaku I (2012) A coded structured light projection method for high-frame-rate 3D image acquisition. In: Ventzas D (ed) Advanced image acquisition, processing techniques and applications, 14 Mar 2012

Study of Structural and Mechanical Behaviour of Severe Plastically Deformed Al–Mg(AA 5052) Alloy Processed by Constrained Groove Pressing Technique Jaya Prasad Vanam, Vinay Anurag Potnuri, and Sree Vidya Sravya Nallam Abstract This paper discusses the effects of Constrained Groove Pressing (CGP) on Aluminum–Magnesium(AA 5052) alloy specimens at room temperature. CGP is one of the severe plastic deformations technique (SPD) by which Ultra FineGrained (UFG)/Plane metallic materials can be processed. A comprehensive study is made on the structural and mechanical properties of the Aluminum specimen before and after constrained grooves pressing. The entire process is simulated in AFDEX CAE Software. Further, simulated results of differently oriented workpieces (90° and 180°) which are processed by the CGP technique are compared. It is found that most of the properties are superior to Aluminum samples such as Yield Strength, Ultimate Tensile Strength, Hardness, Strain rate, etc. are found to be better for the CGP processed specimen. The results are discussed with respective graphs. Keywords Constrained Groove Pressing (CGP) · AFDEX (Adviser for metal forming process Design Expert · UFG (Ultra fine-grained) materials · SPD (Severe Plastic Deformation Technique) · AA5052 alloy

1 Introduction Ultra fine-grained (UFG)/Plane metallic materials are mostly processed by Severe plastically deformation technique (SPD) because often processing by these techniques lead to superior mechanical properties [1, 2]. Various such techniques which can impose intense plastic strain are Equal Channel Angular Pressing (ECAP), High Pressure Torsion (HTP), Multi Axial Forging (MAF), Constrained Groove Pressing (CGP), Cyclic Extrusion and Compression (CEC), and so on. Out of these, CGP is exclusively purported for processing plane metallic materials [3] and in CGP process nearly uniform strain to specimen by deforming material in between pair of corrugated dies and flat dies can be imposed. CGP is a forming technique in which metal is subjected to intense plastic deformation through repeated dominant shearing and J. P. Vanam (B) · V. A. Potnuri · S. V. S. Nallam Department of Mechanical Engineering, UCEK, JNTUK, Kakinada, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_68

567

568

J. P. Vanam et al.

pressing (flattening) of plate [3]. Repetition of process is needed to obtain desired mechanical properties and micro structural changes [3, 4] and development of UFG structure in metals and alloys has been observed [5, 6]. Also, number of passes and design of dies play a major role in achieving desired fine-grained microstructure which leads to superior mechanical properties of sheet samples. By repeating CGP, a cumulative plastic strain will be accumulated within the sample without affecting the initial plate or sheet dimensions. Hence, the aim of our work is to understand the study the effect of process parameters like number of passes, orientation of work piece, and die design on the material (AA 5052) processed under the CGP technique.

2 Experimentation Procedure 2.1 Material Preparation and CGP Experiment In the present study, AA 5052 alloy sheet with sample size 85 × 85 × 3 mm3 was used. Further sample was deformed in between corrugated and flat dies alternatively as per Fig. 1 (Table 1). In the first pass, the die is placed in between the corrugated die Fig. 2a with the slightest offset and deformed in between the dies. Then the deformed sheet is removed and kept in between the flat pair of dies and now it will be flattened as per Fig. 2b. After first flattening, now the sample is rotated by 90°/180° about the axis which is normal to plane of the sample specimen. Here by rotating the specimen, the un-deformed areas are in between the dies for better stress distribution [7]. Further then flattened work piece is again deformed in between the corrugated dies as per Fig. 2c. Followed by flattening the deformed sample after pass 3 in next consecutive pass Fig. 2d. Theoretical equations for calculating shear and effective strain accumulated after each pressing induces an engineering strain [1] Fig. 1 Schematic of die showing flat specimen between corrugated dies

Study of Structural and Mechanical Behaviour of Severe … Table 1 Chemical composition of AA 5052 alloy

569

Material

Composition (%)

Mg

2.2–2.8

Cr

0.15–0.35

Fe

0.0–0.40

Si

0.0–0.25

Others

0.0–0.15

Cu

0.0–0.10

Zn

0.0–0.10

Mn

0.0–0.10

Other (each)

0.0–0.05

Al

Remainder

Fig. 2 a First pressing of corrugated specimen, b Second pressing of flat specimen, c Third pressing of flat specimen with 90° rotation, d Fourth pressing of specimen with flat dies

γx y = H/T = tan θ

(1)

where H, T and θ are groove height, width and groove angle, respectively. Shear strain is given as εx y = γx y /2

(2)

570

J. P. Vanam et al.

Since CGP is assumed as a pure shear deformed under plane strain condition, correspondingly, the effective strain [4]  εeff =

4(εx y)2 tan θ = √ 3 3

(3)

Therefore, the total effective strain accumulated in a CGP sample pressed by n passes is presented as 2 tan θ εtotal = n √ 3

(4)

Our corrugated dies were designed and manufactured with groove angle θ is 45° and with equal groove width (T ) and height (H) which theoretically√concludes the engineering shear strain γ xy = 1, and the effective strain εeff = 1/ 3 ≈ 0.58 1st pass). At the end of first flattening (Pass 2), the sheet has a uniform strain magnitude of 1.16 [8].

3 Results and Discussion 3.1 Simulation Dies are modelled in CATIA V5 R20 and Simulation (FEA) is carried out in AFDEX (Advisor for metal forming process Design Expert) software which is for the first time incorporated for present work simulation which is a forming tool which widely used in industries. AFDEX reduces the complexity of explicit analysis of ANSYS and continuous simulation of DEFORM 3D, Entire Process was simulated seamlessly with minimum effort.

3.2 Simulation Results Mechanical properties when work piece is oriented 90° and 180° at the end of the fourth pass.

3.2.1

Orientation of Work Piece Is 90°

Here in this case after the first flattening, i.e., after the second pass, the work piece was rotated by 90° [9] to deform the un-deformed areas of specimen which are strained in particular direction in the earlier pass. After the four passes, different mechanical

Study of Structural and Mechanical Behaviour of Severe …

571

Fig. 3 a Effective strain (90°), b hardness (90°), c tensile strength (90°)

properties are varied in the following trend. These are the respective simulation results after all four consecutive stages don in AFDEX software. Mechanical properties varied in the following fashion. Effective strain rate at the end of four stages is 4.49 from Fig. 3a. In the study conducted earlier by Satheesh Kumar and Raghu [10] strain rate can be up to 5.8. Hardness variation in the above software is given by Brinell hardness load in the simulated studies variation in terms of kgf/mm2 can be converted to N/mm2 by multiplying above result with 9.8. Total hardness variation after four passes increased by 1.42 times from Fig. 3b and according to Kia-Huai Yang and Ze-Chang hardness may increase up to 1.51 [6] times the initial hardness value. Strength varied up to 194 mpa as showed in Fig. 3c. Here the load local min and max value represents respective yield strength and Ultimate tensile strength at that particular pass. Both Yield and Ultimate strength are showed improvement at the end of every pass which is discussed in the coming graphs.

3.2.2

Orientation of Work Piece Is 180°

In this case after the first flattening, i.e. second pass the specimen will be rotated 180° [6] normal to plane of the specimen to deform the un-deformed areas in the specimen which are present in the earlier passes. Variation of mechanical properties, in this case, are as follows: effective strain rate after pass 4 is 4.6 in this case as per Fig. 4a. Hardness value varied in terms of Brinell hardness load in AFDEX and after pass 4 hardness is increased by 1.41 times. The variation of each pass is plotted as a graph and will be discussed with the help of graphs.

572

J. P. Vanam et al.

Fig. 4 a Effective strain (180°), b hardness (180°), c tensile strength (180°)

Yield strength and ultimate tensile strength are improved significantly after four passes and they are shown in Fig. 4c and variation is discussed in the coming graphs.

3.3 Graphical Variation Mechanical property variation of AA 5052 Specimen which is processed by CGP with work piece orientation at 90° is given by Fig. 5a Effective strain rate at the end of pass 2 is 1.19 and effective strain variation from pass 3 to pass 4 is 2.1. By this, it can be observed that the strain variation is more from pass 3 to pass 4 when compared to pass 1 to pass 2 because the material is strain hardened more after the second flattening, i.e. pass 4. Total effective strain at the end of CGP is 4.49. Brinell hardness load variation from Fig. 5a can be observed as the increase in load variation is more linear until pass 2, then variation is more dynamic until the end of CGP. Strengths, both Yield and Ultimate tensile strength, are given by strength variation is post-processing of AFDEX. The yield strength keeps on increasing after every pass but variation is more linear but Ultimate tensile Strength is variation is linear up to pass 2 then it is more dynamic until the end of CGP process. Mechanical property variation of AA5052 specimen with 180° specimen orientation after pass 2 is given by Fig. 5b. Effective strain rate variation is similar to case of 90° but it is slightly higher in the case of 180° specimen orientation which can be observed by comparing Figs. 3a and 4a. Brinell Hardness load variation is similar in both the cases but hardness in more dominant in case of 180° orientation because it is 200% more when compared to 90° oriented specimen which can be observed in Fig. 5a, b.

Study of Structural and Mechanical Behaviour of Severe …

(a)

AA 5052 Property Variation under CGP work piece orientation 90 o

800

Mechanical Property Variation

573

700 600 Yield Strength(Mpa)

500

Brinell Hardness Load (N/sq.mm)

400

Effective strain Rate

300

Ultimate Tensile Strength(Mpa)

200 100 0 0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

CGP Stages

Mechanical Properties Variation

(b)

AA 5052 Property Variation under CGP with Work piece orientation 180o

1600 1400 1200

Yield Strength (Mpa)

1000

Brinell Hardness load(N/sq.mm)

800

Effective Strain rate

600

Ultimate Tensile Strength(Mpa)

400 200 0 0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

CGP Stages Fig. 5 a Mechanical properties variation for 90° orientation of work piece, b mechanical properties variation for 180° orientation of work piece

Yield strength variation is nominal and increasing after every pass until pass 4 in the 180° oriented CGP process but the intensity of yield strength is 150% more in case of 180° orientation. In Fig. 5a, b, yield strength at the end of pass 4 of 90° is almost identical to yield strength at the end of pass 2 of 180° case of specimen. Ultimate tensile strength variation is linear up to pass 3 then variation is more dynamic from pass 3 to pass 4 in 90° specimen orientations. In the 180° Specimen orientation is the ultimate tensile strength variation is linear up to pass 1 but after that improvement in UTS is very dynamic from pass 2 to pass 4 which can be observed from Fig. 5a, b. So after studying the graphical variation of above two simulated cases, the intensity of property improvement is significant in the case of 180° than 90° even at almost similar strain rates, From the simulated results, in the case of 90° specimen orientation, improvement in properties is limited because in the pass 3 s corrugation, i.e.

574

J. P. Vanam et al.

deformation is perpendicular to pass 1 so there is more a chance that improvement in properties is being cancelled out in this orientation but in case of 180° specimen orientation, the second corrugation i.e., pass 3 is parallel to pass 1 which is the first corrugation.

4 Conclusion In this current work, AA 5052 alloy sheets were subjected to an SPD method called CGP and cumulative strains imposed after four passes on the sheet is 4.6. Also, the effect of CGP on Mechanical Properties variation like Yield Strength, Ultimate Tensile Strength (UTS), and Hardness was investigated with respect to our simulation and experimental results. Then the main findings of our study can be presented as follows: 1. Ultimate Tensile Strength of AA 5052 alloy is increased by increasing number of passes in the CGP process. Maximum rate is during the initial stage (first pressing) and further rate of increase is paced down with increasing deformation in case of both 90° and 180° orientation of work piece. 2. The improvement of Yield and UTS of sample specimen is more dominant is 180° orientation of work piece which has been rotated to deform un-deformed zone of sample specimen rather than the 90° oriented work piece because in this case, the deformation is perpendicular to its earlier case, so it is limiting the property improvement like in case of 180° oriented work piece. 3. The average Brinell hardness load increased by a factor of approximately 1.42 after the second flattening or pass 4. 4. Strain rate for 90° is varied with 5% when compared to 180° so strain rate is almost similar for both the specimen orientation.

References 1. Wang ZS, Guan YJ, Wang GC, Zhong CK (2015) Influences of die structure on constrained groove pressing of commercially pure Ni sheets. J Mater Process Technol 215(1):205–218 2. Valiev RZ, Alexandrov IV (2000) Nanostructured materials produced by severe plastic deformation. Logos, Moscow 3. Dobatkin SV (2000) Severe plastic deformation of steels: structure, properties and techniques. In: Lowe TC, Valiev R (eds) Investigation and applications of severe plastic deformation, Kluwer, Netherlands, vol 3, p 13 4. Shin DH, Park JJ, Kim YS, Park KT (2002) Constrained groove pressing and its application to grain refinement of aluminium. Mater Sci Eng 328A:98–103 5. Huang JY, Zhu YT, Jiang H, Lowe TC (2001) Microstructures and dislocation configurations in nanostructured Cu processed by repetitive corrugation and straightening. Acta Mater 49:1497– 1505

Study of Structural and Mechanical Behaviour of Severe …

575

6. Yang K, Zou Z, Zeng J, Chen W (2018) Microstructures evolution and mechanical properties of 5052 aluminium alloy processed by constrained groove pressing. Int J Comput Mater Sci Surf Eng 7(3):4 7. Krishnaiah A, Chakkingal U, Venugopal P (2005) Applicability of the groove pressing technique for grain refinement in commercial purity copper. Mater Sci Eng 410–411(A):337–340 8. Nagaraju KN, Sunil AR, Sachin K, Sujay H, Siddesha H, Anand Kumar S (2018) Influence of constrained groove pressing passes and annealing characteristics on the mechanical properties of 6061 aluminum alloy 9. Soon Fong K, Jen Tan M, Lan Ng F, Danno A, Wah Chua B (2017) Microstructure stability of a fine-grained AZ31 magnesium alloy processed by constrained groove pressing during isothermal annealing. J Manuf Sci Eng 139(8):081007 10. Satheesh Kumar SS, Raghu T (2011) Tensile behaviour and strain hardening characteristics of constrained groove pressed nickel sheets. Mater Des 32:4650–4657

Shear Rate Dependent Frictional Behavior of the Granular Layer Pawan Kumar Soni

and Arun K Singh

Abstract The present experimental study investigates shear rate dependent frictional properties of the granular layer between two hard surfaces. Slide-free-slide (SFS) experiments were performed on the layer in direct shear mode. It is observed that static stress increases with both normal stress as well as shear velocity. The Mohr–Coulomb (MC) failure criterion is used for determining adhesive stress as well as coefficient of friction of the sliding interface. Both components of friction increase with shear velocity. Their scaling laws, in terms of shear velocity, reveal that the Coulombic friction is more pronounced over the adhesive friction and these results are also justified. Keywords Granular material · Static stresses · Slide-free-slide (SFS) test · Adhesive stress · Coefficient of friction

1 Introduction A granular material is a system of discrete particles and the physical and mechanical properties of such material depend on distribution and voids among the particles [1]. Granular materials find applications in sugar, salt, food grains, and pharmaceutical industries and the examples include sand, soil, rock, gouges, etc. [2]. Motivated by the interesting frictional properties of the granular solids, we perform the experiments for understanding the frictional properties of the granular layer in low shear velocity conditions. The aim of the present work is to understand the shear rate dependent frictional characteristics of the granular solid sandwiched between two hard surfaces. Krantz has performed sliding tests on the mixture of sand and cement, sand and clay and dry quartz sand and reported that the coefficient of friction and cohesion increase with density of the materials [3]. Schellart has also studied frictional behavior of sugar grains, glass particle, and quartz and established that shape such as P. K. Soni (B) · A. K. Singh Department of Mechanical Engineering, Visvesvaraya National Institute of Technology, South Ambazari Rd., Ambazari, Nagpur, Maharashtra 440010, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_69

577

578

P. K. Soni and A. K. Singh

rounding and sphericity of the granular particles, instead of their size, determine their frictional properties [4]. Duttine et al. investigated the frictional properties of the glass and natural sand particles having varying sizes and their observations show the rate-independent sliding behavior [5]. Recently, van den Ende et al. have experimentally determined the cohesion and frictional properties of an earthquake fault during the interseismic period [6]. Nasuno et al. have explored experimentally the stick-slip behavior of glass particles under the condition of low sliding velocity [7]. Geminard et al. have reported that sliding velocity has a negligible effect on steady friction of the saturated granular layer consists of the glass beads [8]. Lieou et al. proposed a mechanistic model for the gouge materials subject to acoustic vibrations and shear deformation [9]. Morgan used the particle dynamics method for simulating the frictional behavior of the gouge layer for validating the rate and state friction laws [10]. Dorostkar et al. have studied the effect of liquid contain on frictional properties of wet granular layer [11]. Frye and Marone have found in their experiments that coefficient of friction is independent of relative humidity, but at high humidity, there is a transition from velocity weakening to velocity strengthening as shear velocity increases [12]. Guo and Julia have experimentally studied the effect of size and shape of the particle of granular material on frictional properties of granular material [13]. Marone has reviewed the frictional behavior of gouge materials in view of the rate and state friction laws [14]. Further, if friction increases as sliding velocity increases, the phenomenon is known as velocity strengthening (VS) process. While friction decreases as sliding velocity increases, this is known as velocity weakening (VW) process and stick-slip instability occurs in the VW regime [14]. In the present study, friction experiments were performed in VS regime to understand the steady frictional properties of thin granular layer. According to Mohr–Coulomb (MC) friction criterion, shear strength arises due to the cohesion and Coulomb friction [14, 15]. In other words, frictional stress τ = c + σn μ, where σn and μ are effective normal stress and coefficient of friction, respectively. Further, c is generally characterized as cohesive stress [16]. However, in the case of sliding interface forming dissimilar solids for instance soil and rock surfaces, the MC failure criterion takes the form as τ = a +σn μ, where a is adhesion and μ is corresponding interfacial coefficient of friction [15]. The significance of this simple model is that role of both Coulombic, as well as non-Coulombic friction (adhesive) could be determined between the granular material and rock surface in a single test. The objective of the present study is to understand the effect of normal stress and shear rate on frictional behavior of granular layer.

2 Experimental Set Up Figure 1 shows the schematic sketch of the experimental set up used in the present study. The experimental set up is based on principle of direct shear sliding [15]. The lower surface is fixed while the upper block acts as the slider. Test sample consists of 10 g of black cotton soil having grain size of 425 micron sheared between two

Shear Rate Dependent Frictional Behavior of the Granular Layer

579

Normal stress, σn

Fig. 1 A schematic sketch of the experimental system employs for friction of the granular layer in direct shear test [15]

Constant Sliding Velocity, v0

Rock solid

Frictional stress, τ

Granular layer

Rock solid substrate

rough granite blocks. In order to maintain the thickness of the layer, a fixed amount of granular material was spread over a fixed area (30.0 cm2 ) of the substrate. The normal load was varied in the experiment in range of 10–50 N. A load cell (50 N) connected with data acquisition system (DAS) was used for measuring the friction force at the sampling rate of 10. Although the hold time was fixed for 60 s, the external shear velocity V0 was varied from 0.2 to 1.0 mm s−1 in direct shear test using slide-free-slide (SFS) test [15].

3 Results and Discussion Figure 2 presents the relation between static frictional stress τs versus normal stress σn . The linear relationship between τs versus σn is linear, thus validating the MC criterion. At the same time, the slope of the straight signifies coefficient of friction, which intercepts adhesive stress. It is also observed from the plots in Fig. 2 that τs increases with v0 at a fixed σ n . Figure 3a presents the scaling law between adhesive stress as and shear velocity vo for static friction as as ~ v00.012 and Fig. 3b, on the other hand, shows the scaling Fig. 2 Plot showing static frictional stress τs versus effective normal stress σn for varying shear velocity v0

580

P. K. Soni and A. K. Singh

Fig. 3 shows the scaling law between adhesive stress as versus v0 as as ~ v00.012 in (a) as well as static coefficient of friction μs versus v0 as μs ~ v00.1244 in (b)

law between static coefficient of friction μs and shear velocity v0 for as μs ~ v00.124 . It is seen that as increases with increase in v0 . Experiment results also show the exponent of as is lower in magnitude than corresponding μs . In other words, frictional properties of the wet granular layer still arises mainly due to the Coulomb stress (hard solid friction) than the adhesion process. The rate-dependent behaviors of the granular solids could be explained using the Arrhenius theory of rate reaction. This theory predicts the variation of shear strength as logarithm of shear velocity [14]. Future studies include hold time-dependent adhesion and frictional properties also validating the rate and state friction laws on granular materials at low normal stress [6, 15].

4 Conclusion The present experiments show that the magnitude of adhesion and friction coefficient increases with shear velocity. Static stress follows the increasing trend with normal stress. The exponent of adhesive stress with sliding velocity is larger in magnitude than corresponding to the coefficient of friction. The reason for this observation may be attributed to less surface energy of the granular layer. Acknowledgements This work is supported by NRDMS-DST, Government of India, through project no. NRDMS//02/43/016(G).

Shear Rate Dependent Frictional Behavior of the Granular Layer

581

References 1. Campbell CS (1990) Rapid granular flows. Annu Rev Fluid Mech 22(1):57–90 2. Wornyoh EY, Jasti VK, Higgs CF (2007) A review of dry particulate lubrication, powder and granular materials. J Tribol 129(2):438–449 3. Krantz RW (1991) Measurements of friction coefficients and cohesion for faulting and fault reactivation in laboratory models using sand and sand mixtures. Tectonophysics 188(1-2):203– 207 4. Schellart WP (2000) Shear test results for cohesion and friction coefficients for different granular materials, scaling implications for their usage in analogue modelling. Tectonophysics 324(1–2):1–16 5. Duttine A, Tatsuoka F, Kongkitkul W, Hirakawa D (2008) Viscous behavior of unbound granular materials in direct shear. Soils Found 48(3):297–318 6. Van den Ende MPA, Niemeijer AR (2019) An investigation into the role of time-dependent cohesion in interseismic fault restrengthening. Sci Rep 9(1):9894 7. Nasuno S, Kudrolli A, Bak A, Gollub JP (1998) Time-resolved studies of stick-slip friction in sheared granular layers. Phys Rev E 58(2):2161 8. Géminard JC, Losert W, Gollub JP (1999) Frictional mechanics of wet granular material. Phys Rev E 59(5):5881 9. Lieou CK, Elbanna AE, Carlson JM (2016) Dynamic friction in sheared fault gouge: implications of acoustic vibration on triggering and slow slip. J Geophys Res Solid Earth 121(3):1483–1496 10. Morgan JK (2004) Particle dynamics simulations of rate-and state-dependent frictional sliding of granular fault gouge. In: Computational earthquake science part I, pp 1877–1891 11. Dorostkar O, Guyer RA, Johnson PA, Marone C, Carmeliet J (2018) Cohesion induced stabilization in stick slip dynamics of weakly wet, sheared granular fault gouge. J Geophys Res Solid Earth 123(3):2115–2126 12. Frye KM, Marone C (2002) Effect of humidity on granular friction at room temperature. J Geophys Res Solid Earth 107(B11):ETG-11 13. Guo Y, Julia KM (2004) Influence of normal stress and grain shape on granular friction, Results of discrete element simulations. J Geophys Res Solid Earth 109:B12 14. Marone C (1998) Laboratory-derived friction laws and their application to seismic faulting. Annu Rev Earth Planet Sci 26(1):643–696 15. Koupouli NJ, Belem T, Rivard P, Effenguet H (2016) Direct shear tests on cemented paste backfill–rock wall and cemented paste backfill–backfill interfaces. J Rock Mech Geotech Eng 8(4):472–479 16. Debasis D, Kumar VA (2016) Fundamentals and applications of rock mechanics. PHI Learning Pvt. Ltd.

Mathematical Overview on Omnidirectional Spherical Wall Traversing Robot Yogesh Phalak, Rajeshree Deotalu, Onkar, and Sapan Agrawal

Abstract Since the past few decades, many mechanisms are being developed for wall climbing robots. Robotic systems having omnidirectional surface traversing ability independent of its inclination require complex morphology transformation for a floor to wall transition as well as perfect adhesion on the vertical surface. This paper depicts the development of the mathematical model for omnidirectional maneuverability of the robot. The novel design of this robot is newly aimed, and its motion is mathematically modeled. Keywords Omnidirectional · Spherical robot · Mechanical design · Mathematical model

1 Introduction Robots having ability to climb the vertical surfaces are categorized into wall climbing robots. Wall climbers have been developed across the globe for surveillance, inspections, flaw detection, cleaning, maintenance, etc. The proposed mechanism has a 2 DOF bi-propeller plane which allows getting any direction of thrust on to the wall which helps it to move freely and remain attached to the surface. As the working principle of the proposed Omni-directional Spherical Wall Traversing Robot is based on propeller-based thrust, hence its performance is independent of surface parameters, thus outspreading the competence of wall climbers. The proposed design is resulting in robustness and simplified mobility of the robot.

Y. Phalak (B) · R. Deotalu · Onkar Visvesvaraya National Institute of Technology, Nagpur, India e-mail: [email protected] S. Agrawal Worcester Polytechnic Institute, Worcester, MA, USA e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_70

583

584

Y. Phalak et al.

Due to supersymmetry of outer spherical shell and omnidirectional navigation capabilities, this robot has major applications in duct maintenance, inspection of skyscrapers, maintenance including cleaning and painting of chimneys in radioactive (hazardous) environment.

2 Historical Overview Various designs of wall climbing robots have been developed in the past few years. These robots are primarily built keeping in mind that they can be used for rescue, wall inspection, and firefighting [1]. Some of them were also used in nuclear plants, construction industries, military areas, etc. [2]. The challenging part of these robots is the adhesion mechanism which sticks the bot on the surface of the wall. One technique is to use vacuum suction to produce the adhesive force. However, this method fails miserably if the surface is uneven and rough or if there are obstacles on the path. This is because, in order to have maximum contact force the surface must be smooth, any loose contact will make the bot to fall down [3]. The other idea is to use magnetic/electromagnetic wheels on the bot, but then, this will work only on iron, steel, and other ferromagnetic structures. Another mechanism is to use dry elastomer adhesives on the legs of the bot [4]. However, this technology is currently not mature enough to be used for climbing robots. Lastly, there is one mechanism in which the thrust generated by the propellers is used to drive the robot on the surface of the wall. Comparing with other mechanisms, the bot moves fastest in this one. It does not depend on the properties of the wall and can work on uneven surfaces also. Including a pair of coaxial propellers mounted on a gimbal structure increases the degrees of freedom of the bot and helps in the steering of the bot on the wall [5].

3 Mathematical Overview 3.1 Variables and Parameters • Independent Input Variables: ω = angular velocity of propellers. θin = inner angle of gimbal. θout = outer angle of gimbal. • Dependent Variables: f = magnitude of the thrust given by by propellers. ζ (α, β, γ ) = thrust hindrance factor function. α, β, γ = inclination angles with x-,y-, and z-axes, respectively. F = overall magnitude of the thrust. f r = required force of friction.

Mathematical Overview on Omnidirectional Spherical Wall …

585

• Other Parameters and Variables: μws = friction coefficient between wall and outer sphere. f (x) = trajectory path function. φ = slope angle of tangent to f(x) at position (x = x0 ), i.e. tan−1 ( f  (x)). δ = angle of thrust in x-y plane.

3.2 Angular Speed and Thrust of Propellers The coaxial bi-propeller mechanism consists 1045 (10 × 4.5”) ABS Propellers (1CW + 1CCW-1 pair Black) and a couple of A2212 2200KV BLDC Motors. Thrustangular velocity characteristics are given by the Eq. 1 (Fig. 1) F = 0.109 × 10−6 × ω2 − 210.6 × ω + 0.154

(1)

Perforations are made onto the spherical shell to pass the air required for producing thrust. As propulsion thrust is directly proportional to the area of the propeller plane, perforated shell hinders some factor of the thrust; the thrust hindrance function ζ is introduced to include this problem into account. Due to non-uniform thrust hindering factors into the space of the spherical robot, the thrust hindrance function will depend upon space angular space coordinates, i.e., ζ (α, β, γ ). Therefore, the thrust vector is given by (2) F = f · ζ (α, β, γ ) · [cos(α) cos(β) cos(γ )]T

Fig. 1 Thrust versus angular velocity characteristics

586

Y. Phalak et al.

3.3 Direction of Thrust and Gimbal Angles Assuming equilibrium position of the sphere during the motion heading in a direction making angle φ with the positive x-axis, the frictional force fr depends upon wall and sphere surface material characteristics. μws is the kinetic friction coefficient between two touching surfaces of wall and robot [6]. On resolving into three-axis components, the resulting thrust is given by the equation. ⎤ ⎤ ⎡ fr · cos(φ) cos(α) F = f · ζ (α, β, γ ) · ⎣cos(β)⎦ = ⎣fr · sin(φ) + mg ⎦ − μfrws cos(γ ) ⎡

(3)

Transformation matrix for rotation of plane with angle θ about axis with unit vector [l m n]T is given by, ⎡

⎤ l 2 · (1 − cθ ) ml · (1 − cθ ) nl · (1 − cθ ) ⎣lm · (1 − cθ ) m 2 · (1 − cθ ) nm · (1 − cθ )⎦ ln · (1 − cθ ) mn · (1 − cθ ) n 2 · (1 − cθ ) ⎡

⎤ cθ −n · sθ m · sθ cθ −l · sθ ⎦ + ⎣ n · sθ −m · sθ l · sθ cθ The 2DOF gimbal mechanism is used to control the direction of the thrust. This mechanism is made of two rotating motors with perpendicular axes of rotation. Let, angle of rotation of inner motor be θin about axis [1 0 0]T , and an outer motor is a θout about axis [0 0 1]T ; hence, transformation matrices of both individual rotations are given by: ⎡ ⎤ 1 0 0 T rin = ⎣0 cos(θin ) −sin(θin )⎦ (4) 0 sin(θin ) cos(θin ) ⎡

T rout

⎤ cos(θout ) −sin(θout ) 0 = ⎣ sin(θout ) cos(θout ) 0⎦ 0 0 1

Hence, thrust vector with direction controlled by gimbal angles is given as: ⎡

⎤ cos(θin ) × sin(θout ) F = f · ζ (α, β, γ ) · ⎣cos(θin ) × cos(θout )⎦ sin(θin )

(5)

Mathematical Overview on Omnidirectional Spherical Wall …

587

⎡

⎤ fr · cos(φ) = ⎣fr · sin(φ) + mg ⎦ − μfrws

(6)

3.4 Motion Analysis Let the robot is initially located at point (x1 , y1 ) in the x-y plane which is supposed to reach point (x2 , y2 ). Therefore, a linear velocity of the bot is taken proportional to the instantaneous Euclidean distance between its position and the destination. As it is an accelerated motion (with acceleration ‘a’) pseudo force (ma) is taken opposite to the direction of motion. Kinetic frictional force fr is opposing the motion. Let

= |x2 − x| and η be the proportionality constant between velocity and distance (Fig. 2).  |vx y | ∝ (x2 − x)2 + (y2 − y)2 ∴ vx y = η · sec(φ) × [sin(φ) cos(φ)]T

(7)

∴ a = η2 · sec(φ) × [sin(φ) cos(φ)]T

(8)

Therefore, required thrust vector is:

Fig. 2 Motion of spherical bot from (x1 , y1 ) to (x2 , y2 )

588

Y. Phalak et al.

Fig. 3 Closed-loop control system for OsWalT

⎡

⎤ fr · cos(φ) − mη2

F = ⎣fr · sin(φ) + mg − mη2 · tan(φ)⎦ − μfrws

(9)

Considering above derived mathematical transformations, one experiment has been conducted in the MATLAB/Simulink environment with parameters taken as η = 1; (x1 , y1 ) = (0, 0); (x2 , y2 ) = (1, 1); fr = 40; μws = 0.8; mg = 29.415; hence, resulting plots of required thrust magnitude and directions are as given in (Fig. 3).

4 Closed-Loop Control of the System Arduino Nano V3.0 CH340 Chip is used as CPU of the system. Figure 3 shows the overview of the closed-loop control system of the bot. The angular velocity of the thruster is measured using a Hall effect sensor (BMES A3144 Hall Effect Magnetic Sensor Module) and given as a feedback PID controller used [3] with self tuning ability given in [7]. Similarly, two PID controllers are used to control gimbal angles. Tilt angles are calculated using potentiometers attached to the motor shafts and given as a feedback to the PID controller. Pairs of servomotors in gimbal are operated synchronously with opposite angels.

5 Future Work In future, we will redesign improved prototype closely resembling the proposed model with spherical shell and will have experiments on vertical walls by applying

Mathematical Overview on Omnidirectional Spherical Wall …

589

the proposed angular control of the gimbal mechanism. Furthermore, in addition to the remote control by human operations, the robot will be also equipped with autonomous functions such as obstacle avoidance and path tracking. Acknowledgements The research work of this project is supported and funded by IvLabs, Robotics lab of Visvesvaraya National Institute of Technology, Nagpur, India.

References 1. Nishi A (1991) A wall climbing robot using propulsive force of propeller. In: 5th international conference on Advanced Robotics. Robots in Unstructured Environments, vol 1. Pisa, Italy, 320–325. https://doi.org/10.1109/ICAR.1991.240633 2. Briones L, Bustamante P, Serna MA (1994). Wall-climbing robot for inspection in nuclear power plants. Proceedings IEEE international conference on Robotics and Automation, vol 2, 1409–1414. https://doi.org/10.1109/ROBOT.1994.351292 3. Liu G, Liu Y, Wang X, Wu X, Mei T (2016) Design and experiment of a bioinspired wall climbing robot using spiny grippers. In: 2016 IEEE international conference on Mechatronics and Automation. Harbin, 665–670 4. Murphy MP, Sitti M (2007) Waalbot: an agile small-scale wall-climbing robot utilizing dry elastomer adhesives. IEEE/ASME Trans Mechatron 12(3) 5. Tanaka Y, Nozaki K, Ioi K (2017) Motion control of a wall climbing robot with coaxial propeller thruster. In: 2017 2nd IEEE international conference on Intelligent Transportation Engineering (ICITE), 2017, 360–364 6. Schmidt D, Hillenbrand C, Berns K (2011) Omnidirectional locomotion and traction control of the wheel-driven, wall-climbing robot. Cromsci Robot 29:991–1003 7. Babu VM, Das K, Kumar S (2017) Designing of self tuning PID controller for AR drone quadrotor. In: 2017 18th international conference on Advanced Robotics (ICAR), Hong Kong, 167–172, 2017. https://doi.org/10.1109/ICAR.2017.8023513. 8. Lee G, Kim H, Seo K, Kim J, Kim HS (2015) Robotics and autonomous systems multitrack: a multi-linked track robot with suction adhesion for climbing and transition, vol 72. https://doi. org/10.1016/j.robot.2015.05.011 9. Ma B, Liu R, Zhang R, Tian L, Nashunbuhe (2007) Design of wall climbing robots with transition capability. In: IEEE international conference on Robotics and Biomimetics, 15–18 December

Finite Element Analysis of Ballistic Impact on Monolithic and Multi-layered Target Plate with and Without Air Gap Rohit Kumar, Manoj Kumar, and Pramod Kumar

Abstract In the present work, Finite Element Analysis is applied to analyze the ballistic impact on 1100-H14 aluminum and weldox 460 E steel multi-layered plate using FEM package ABAQUS/CAE explicit. A conical projectile is projected on the plate with different velocities. The 1100-H14 aluminum plate is model using Johnson–Cook material modeling and the Bao-Wierzbicki failure model is used for fracture. The Cut-off on negative triaxiality has been incorporated in the present work. The values of material parameters such as elastic and plastic are taken from the literature and the projectile is assumed to be rigid. Fracture pattern of the plate, residual velocity, and velocity drop on different thicknesses of plates are calculated. The number of petals formed in the plate after the fracture has been reported and the maximum deformation experienced by the plates are studied. It is found that the fracture pattern by the numerical analysis on the plate is almost similar to the experimental result as reported in the literature. And also the ballistic performance of multi-layered metal plates with and without spaced, subjected to impact by blunt projectile is investigated by numerical simulation. Further, the effect of the air gap on ballistic resistance is investigated. Ballistic limit velocities of layered plates are decreased with multi-layered target. The residual velocity, ballistic limit velocity and perforation time are determined. The result also showed that there is a consistent increase in ballistic resistance of target as the number of layered is increased. Keywords Finite element analysis · Ballistic impact · Bao-Wierzbicki failure model · Multi-layered

1 Introduction Ballistic impact is the high-velocity impact on a target by a projectile having a low mass and size. The damage in such an effect within fractions of millisecond. It has a moderately more predominant impact on the object than a lower force is applied R. Kumar · M. Kumar (B) · P. Kumar Dr. B. R. Ambedkar National Institute of Technology, Jalandhar, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 V. R. Kalamkar and K. Monkova (eds.), Advances in Mechanical Engineering, Lecture Notes in Mechanical Engineering, https://doi.org/10.1007/978-981-15-3639-7_71

591

592

R. Kumar et al.

over a relatively longer timeframe. A large number of works have been carried out with impact modeling, impact damage assessment and the assessment of post-impact residual properties. Borvik et al. [1] demonstrated the impact of nose shape (blunt, hemispherical, conical) projectile on the impact of weldox 460E steel plate of thickness 12 mm. They found that ballistic cut-off speed for hemispherical and conical were found near one another yet for blunt shot is lower. Kurtaran et al. [2] simulated the ballistic effect on plate of 2 mm thickness by projectiles with hemispherical rounded nose of 7.62 mm distance across moving with speeds in the range of 500, 1000, 1500 m/s. They saw that the influence of thermal softening effect must consider for reproduction of better outcomes. Gupta et al. [3] considered the effect of shot nosed shape and aluminum plate of thickness in the scope of 0.5–3 mm both experimentally and numerically. It was seen that ogival projectile were the best penetrators for exceptionally thin plate thickness and blunt projectile for 3 mm thick plate thickness. Teng et al. [4] have shown the ballistic performance of doubled layered steel defensive plates against projectile effect utilizing finite element analysis by conical-shaped projectiles. Iqbal et al. [5] analyzed the effect of projectile nose shape. Aluminum plates were influenced by steel projectiles having double nose shape viz. conico–blunt, blunt–blunt, blunt–conico. It was seen that failure mode was exposed to nose shape, and was particularly not exactly the same as that saw with single nose projectile. Kumar et al. [6] examined simulation using continuum damage mechanics (CDM) model on ballistic effect of steel target impacted by rigid projectiles. The continuum damage model was made by them realized in ABAQUS/Explicit through a client-specified material model subroutine (VUMAT). The attain of temperature and strain rate were incorporated. Zhang et al. [7] thought about the ballistic impact of monolithic and multi-layered steel plates affected by blunt rigid projectile. They examined the impact of air gap between layers, the number, order and thickness of layers on the ballistic resistance. Corran et al. [8] done a no. of tests on the multi-layered in contact plate and found that layered plates in contact were superior to relative monolithic plate. Gupta et al. [9] drove a movement of perforation trail of single- and three-layered metallic objectives with practically identical thickness against level, ogival, and hemi-circular nosed steel projectile. The results exhibited that when the amount of layers were exhausted, the speed drop was viewed as higher than one record of single plate. Ogival nosed shots were viewed as the most profitable penetrator of layered objective. Hemi-circular nosed shot requires the most outrageous vitality for perforating the target.

2 Methodology An impact fracture is related with adiabatic heating that leads to an expansion in temperature. This again causes a thermal softening of the material undergoing the impact load. The Johnson–Cook material model is used for material modeling to incorporate the effect of temperature on the flow stress. Bao and Wierzbicki fracture model is used for fracture of the material.

Finite Element Analysis of Ballistic Impact on Monolithic …

593

    The Johnson−Cook plasticity model σ = A + Bεn 1 + C ln ε∗ 1 − (T ∗ )m Strain Rate r σ = Flow stress; ε = equivalent plastic strain ε* = Reference ; T* = TTm−T , A, Strain Rate −Tr B, n, C and m are material constants where T m is the melting temperature, T r is the reference temperature, i.e., 298 K and T is the working temperature. Bao and Wierzbicki fracture model: The proportional strain to break and the normal stress triaxiality for each case, the fracture locus strain versus stress triaxiality space was built in Fig. 1 by Bao and Wierzbicki [10]. In the scope of negative stress triaxiality, the proportionate strain to fracture diminishes with stress triaxiality. Crack happens because of void formation in tension test specimen which is in scope of high-stress triaxiality. On account of negative triaxialities, an appropriate class of functions with a vertical asymptote was developed by Wierzbicki and Werner based on experimental results. Using this expression, the best mathematical expression is developed as

 ε f = 0.1225 ×  ε f = 1.9 ×

σh σeq

2

σh 1 + σeq 3

−0.46 

− 0.18 ×

ε f = 0.15 ×



σh σeq

−1

σh σeq

− 0.33