Light Metals 2023 (The Minerals, Metals & Materials Series) 3031225317, 9783031225314

Citation preview

Edited by STEPHAN BROEK

The Minerals, Metals & Materials Series

Stephan Broek Editor

Light Metals 2023

123

Editor Stephan Broek Kensington Technology Inc. Burlington, ON, Canada

ISSN 2367-1181 ISSN 2367-1696 (electronic) The Minerals, Metals & Materials Series ISBN 978-3-031-22531-4 ISBN 978-3-031-22532-1 (eBook) https://doi.org/10.1007/978-3-031-22532-1 © The Minerals, Metals & Materials Society 2023, corrected publication 2023 This work is subject to copyright. All rights are solely and exclusively licensed 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 Switzerland AG The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland

Preface

I have the great honor and privilege to present the Light Metals 2023 proceedings. After several challenging years, these proceedings finally take us in a positive direction where conferences like TMS’s Annual Meeting can finally be what they should be again: a gathering of accomplished industry professionals sharing science and knowledge in a social environment harnessing the power of human interaction. Our industry is resilient, as illustrated by the number and quality of manuscripts and oral presentations by engineers, scientists, and others represented in this well-referenced bundle of technical articles. The Light Metals 2023 proceedings embody the work of the authors, symposium chairs, session chairs, peer reviewers, and TMS staff who made invaluable contributions to this collective publication. A first thank you goes to the authors of technical papers and posters. It is their engagement and active participation that ultimately make the Light Metals program at the Annual Meeting and this publication an ongoing success. Special acknowledgement goes to the symposium chairs of the Light Metals program and of the five special symposia held in 2023. I thank Pierre Marcellin (Aluminum Reduction Technology), Roy Cahill (Electrode Technology for Aluminum Production), Halldór Guðmundsson (Cast Shop Technology), Julie Lévesque (Aluminum Alloys, Characterization and Processing), and Errol Jaeger (Alumina & Bauxite) for organizing the subjects in the Light Metals program. A further thank you to Mark Dorreen (honorary symposium for Prof. Barry J. Welch), Pernelle Nunez (Aluminum Waste Management & Utilization, and for Aluminum Industry Emissions Measurement, Reporting & Reduction), and Timothy Langan (Scandium Extraction and Use in Aluminum Alloys). All these chairs undertook the bulk of the organization, and that is much appreciated. A constant force behind these efforts is the staff at TMS. In addition to ensuring that deadlines were met, the significant help from Patricia Warren and Trudi Dunlap also safeguarded the society’s guidelines. I extend my sincere appreciation to you both for your hard work and dedication. Finally, I’m grateful to past editors Dmitry Eskin and Linus Perander for their important support throughout the year and to Aluminum Committee colleagues over the years. To be granted the opportunity by my peers to be the editor of the Light Metals 2023 proceedings is a special honor. It has become clear over this past year that incredible changes in our industry lie ahead. The push for sustainability is firm and must deliver solutions within the next few decades that we only can imagine today. This will require scientific breakthroughs, new technologies, and different ways of working together. Platforms like the Light Metals program at the TMS Annual Meeting are invaluable opportunities for sharing the insights and knowledge necessary to spur critical innovations. Students, scientists, researchers, engineers, and other professionals must contribute and collaborate to achieve the crucial changes that will propel our industry, save our climate, and ultimately ensure our collective future.

v

vi

Preface

In 2023 we gathered in San Diego, California for the TMS Annual Meeting. At the time of writing, we have largely overcome the COVID-19 pandemic and travel restrictions have been lifted. But other global events had a profound impact on the 2023 conference since many of our peers have been affected. I can only offer sincere wishes that you all are safe and that we will see each other again soon. Stephan Broek

Contents

Part I

60 Years of Taking Aluminum Smelting Research and Development from New Zealand to the World: An LMD Symposium in Honor of Barry J. Welch

What Makes TMS Special? Let Us Consider a Case Study in Volunteer Excellence: Barry J. Welch . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . James J. Robinson

3

Meeting the Requirements of Potline Customers: The Largely Unmet Challenges Set by Barry Welch to Carbon Anode Producers . . . . . . . . . . . . . . . . Barry Sadler and Alan Tomsett

8

The Need to Respect the Interlink Between Science, Physics, and Cell Design in an Environmentally Responsible Manner: The Next Big Challenge for Aluminium Smelting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Barry J. Welch Anode Quality Optimisation: Industry Learnings from the Research Supervised by Barry Welch . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Alan Tomsett and Barry Sadler Process Recovery to Unlock Power Efficiency Improvement at BSL . . . . . . . . . . . Evan Andrews, Thomas Booby, Murray Ure, and Hao Zhang A Smart Individual Anode Current Measurement System and Its Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Choon-Jie Wong, Jing Shi, Jie Bao, Barry J. Welch, Maria Skyllas-Kazacos, Ali Jassim, Mohamed Mahmoud, and Konstantin Nikandrov

16

24 31

43

Light Metals Research at the University of Auckland . . . . . . . . . . . . . . . . . . . . . . J. B. Metson, R. Etzion, and M. M. Hyland

52

Impact of Aluminium Reduction Cell Parameters on Feeder Hole Condition . . . . Pascal Lavoie and Mark P. Taylor

56

A Dynamic Coupled Mass and Thermal Model for the Top Chamber of the Aluminium Smelting Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Luning Ma, Choon-Jie Wong, Jie Bao, Maria Skyllas-Kazacos, Barry J. Welch, Nadia Ahli, Mohamed Mahmoud, Konstantin Nikandrov, and Amal Aljasmi Following Alumina Dissolution Kinetics with Electrochemical and Video Analysis Tools . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Daniel Marinha, Astrid J. Meyer, Marián Kucharík, Sylvie Bouvet, Miroslav Boca, Michal Korenko, Vladimir Danielik, and Frantisek Simko

67

77

vii

viii

Contents

Monitoring Cell Conditions and Anode Freeze Dissolution with Model-Based Soft Sensor After Anode Change . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Choon-Jie Wong, Jie Bao, Maria Skyllas-Kazacos, Ali Jassim, Mohamed Mahmoud, and Alexander Arkhipov EGA’s First Holistic Mobile Application for Smelter Operations . . . . . . . . . . . . . . Ahmed Al Haddad, Dinesh Kothari, Ghalib Al Falasi, Yousuf Ahli, Sergey Akhmetov, Najeeba Al Jabri, Abdulla Karmustaji, Mustafa Mustafa, and Mahmood Al Awadhi

87

95

Testing Feeding Alumina in Three Channels in a Wide Cell . . . . . . . . . . . . . . . . . 102 Marc Dupuis and Valdis Bojarevics A Pragmatic Model for Bath Temperature Evolution During Alumina Feeding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113 Kurian J. Vachaparambil, Stein Tore Johansen, Asbjørn Solheim, and Kristian Etienne Einarsrud A New Strategy for Transient Heat Transfer Models with Phase Change for the Aluminum Electrolysis Industry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121 Bastien Pansiot, Marc Lebreux, Martin Désilets, Francis Lalancette, Jean-Francois Bilodeau, and Alexandre Blais A Discussion on Thermal Impact of Anode Change in Aluminum Reduction Cell . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128 Zhibin Zhao, Wei Liu, Yafeng Liu, Michael Ren, and Zhaowen Wang Development and Deployment Measures in PLC-Based Pot Control System at Low Amperage Aluminium Reduction Cell . . . . . . . . . . . . . . . . . . . . . . 137 Rajeev Kumar Yadav, Shanmukh Rajgire, Md. Imroz Ahmad, Goutam Das, Ravi Pandey, Mahesh Sahoo, and Amit Gupta Part II

Alumina and Bauxite

Process Simulation with Tertiary Cyclone for Kaolinite Removal from Amazonian Bauxite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149 Allan Suhett Reis, Geraldo Magela Pereira Duarte, Eslyn Neves, Geovan Oliveira, and Thiago Jatobá Granulometry Impact on Digestion Efficiency and Cost-Economics in Alumina Refinery for East Coast Bauxite (India) . . . . . . . . . . . . . . . . . . . . . . . 156 Suchita Rai, M. J. Chaddha, Prachiprava Pradhan, K. J. Kulkarni, M. Panchal, and A. Agnihotri Effect of Thermal Activation Temperature on Pre-desilication of Low-Grade Bauxite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166 Chaojun Fang, Tianrui Cai, Bo Lv, Xiaowei Deng, Jinming Zhang, Zeya Zhao, and Bobing Dong Application of Repeatability and Reproducibility Analysis in the Determination of Available Alumina and Reactive Silica in Bauxites . . . . . . . . . . . . . . . . . . . . . . 171 Paula Lima, Walter Santana, Danielle Matos, Jaqueline Melo, and Janyne Ramos Zero Waste Alumina Production . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178 Casper van der Eijk and Camilla Sommerseth

Contents

ix

Statistical Modelling of Operating Parameters on Bauxite Slurry Hyperbaric Filtration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183 Clara Souza, Antonio Silva, Eduardo Moreira, Enio Laubyer, Fabricia Ferreira, and Raimundo Neto Reduction of GHG Emissions and Increase Operational Reliability Using Immersed Electrode Boiler in an Alumina Refinery . . . . . . . . . . . . . . . . . . . . . . . 191 Rodrigo Neves, Fernando Melo, Everton Mendonça, Erinaldo Filho, and Jeferson Carneiro Steam Grid Stability Using Advanced Process Control and Real Time Optimization in an Alumina Refinery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196 Daniella Costa, Rodrigo Neves, Fernando Melo, João Freitas, Juvenal Sousa, Ediciano Junior, and Danilo Arcodaci Effects of Impurities on the Boiling Point of Bayer Liquor . . . . . . . . . . . . . . . . . . 202 Erwei Song and Erqiang Wang Effects of Different Precursors on the Preparation of b″-Al2O3 . . . . . . . . . . . . . . . 208 Hongsheng Che and Yang Zhang Determination of Unit Cell Parameters of a-Alumina Reference Material . . . . . . . 213 Lin Zhao, Hongsheng Che, Bo Li, and Shuchao Zhang Improvement Seminars: Continuous Improvement and People’s Engagement to Support Sustainability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219 Nathalia Martins, Bruna Dias Cabral, Renan William Costa da Cruz, Raphael Costa, Silene Vendrasco, Jaise Carvalló, Gustavo Silva, Guilherme Brazão, and Karina Trindade Turning Bauxite Residue to Metal Adsorption Materials Through a Low-Cost Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225 Hong Peng, James Vaughan, Shengchun Ma, Sicheng Wang, and Xinyu Tian Hematite and Anatase Conversion to Magnetic Phases During Reductive Re-digestion of Gibbsitic Bauxite Residue . . . . . . . . . . . . . . . . . . . . . . . 230 Paula de Freitas Marques Araújo, Patricia Magalhães Pereira Silva, Andre Luiz Vilaça do Carmo, Fernando Gama Gomes, Adriano Reis Lucheta, Raphael Vieira da Costa, and Marcelo Montini Digestion Efficiency Improvement of Gibbsite-Boehmite Bauxite . . . . . . . . . . . . . . 238 Fengqin Liu, Zegang Wu, Songqing Gu, and Michael Ren Decanter Centrifuge for Dewatering Bauxite Tailings . . . . . . . . . . . . . . . . . . . . . . 246 Camila Botarro Moura, Rafael Alves de Souza Felipe, and Roberto Seno Junior Part III

Aluminum Alloys, Characterization, and Processing

Comparison of TiB2 and TiC Grain Refiners’ Impact on Surface Quality, Edge Cracking, and Rolling Performance of AA5182 DC-Cast Ingot . . . . . . . . . . 253 Josh Lawalin, Pascal Gauthier, and Tao Wang The Influence of Crystallographic Texture Gradients on the Deformation Response of Aluminum Extrusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 258 W. J. Poole, Y. Wang, A. Zang, M. A. Wells, and N. C. Parson

x

Mechanical Properties, Microstructures, and Textures of Cold Rolling Sheets Made from a Low-cost Continuous Cast Al-1.5Cu Alloy with Potential Application in Auto Sheets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262 Xiyu Wen, Yan Jin, and Wei Li Challenges in the Production of 5754 Automotive Alloy Sheet via Twin Roll Casting Route . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 271 Dionysios Spathis, John Tsiros, and Andreas Mavroudis Fabrication of Bright-Rolled Aluminum Suitable for Design Elements in the Automotive Industry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 280 Anita Gründlinger, Peter Johann Uggowitzer, and Josef Berneder Effects of Aging Conditions on Fracture Characteristics of Al–Mg–Si Alloys . . . . 287 Zeynep Tutku Özen, İlyas Artunç Sarı, Anıl Umut Özdemir, Abdullah Kağan Kınacı, Emre Çankaya, Alptuğ Tanses, and Görkem Özçelik Evaluation of EN AW 3003 Aluminium Alloy Homogenization with Specific Electrical Resistivity Measurement . . . . . . . . . . . . . . . . . . . . . . . . . . 294 Maja Vončina, Mitja Petrič, Sebastjan Kastelic, Tilen Balaško, Stanislav Kores, and Jožef Medved The Effect of Octagonal Ingot Shape on AA6xxx Hot Rolling Performance . . . . . 302 Joshua Lawalin, Pascal Gauthier, and Tao Wang The Low-Carbon Production of Wrought Aluminum Alloys Based on Post-consumer Scrap . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 309 Varuzan M. Kevorkijan and Sandi Žist Reduce Inclusion Level Study in Aluminum Slab Products 3XX and 5XX . . . . . . 315 Abdullah Al-Qarni and Bader Dhawi AlMuhana Effect of Iron and Manganese Content on Microstructure and Mechanical Properties of AlSi11 Alloy in Wheels Produced by LPDC-Process . . . . . . . . . . . . 321 Sergey Matveev, Dmitry Moiseev, Tatyana Bogdanova, Roman Vakhromov, and Aleksandr Krokhin Shear Assisted Processing and Extrusion of Unhomogenized Aluminum Alloy 6063 Castings with High Iron Content . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 327 Scott Whalen, Nicole Overman, Brandon Scott Taysom, Md. Reza-E-Rabby, Timothy Skszek, and Massimo DiCiano Solutionization via Severe Plastic Deformation: Effect on Natural Aging in an Al–Mg–Si–(Mn) Alloy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 333 Brian Milligan, B. Scott Taysom, Xiaolong Ma, and Scott Whalen Manufacture of Nano-to-Submicron-Scale TiC Particulate Reinforced Aluminium Composites by Ultrasound-Assisted Stir Casting . . . . . . . . . . . . . . . . . 339 Guangyu Liu, Abdallah Abu Amara, Dmitry Eskin, and Brian McKay Effect of Mn Content on Quench Sensitivity of 6082 Alloys . . . . . . . . . . . . . . . . . . 349 Emrah F. Ozdogru, Aleyna Gumussoy, Hilal Colak, and Isık Kaya Characterization of Aluminum Conductors Steel Reinforced in Overhead Transmission Lines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 355 M. Hassanipour, M. Diago, D. Valiquette, F. Guay, and A. Leblond

Contents

Contents

xi

Mechanical and Electrical Properties of Permanent Steel Mold Cast Eutectic Al-1.8Fe Alloy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 360 S. Liu, A. Hu, A. Dhaif, W. Shen, and H. Hu Effects of the Friction Stir Welding Sliding and Sticking Mechanisms on the Microhardness, Texture, and Element Concentration . . . . . . . . . . . . . . . . . 365 Nicholas Sabry, Joshua Stroh, and Dimitry Sediako Experimental Investigation of the Effect of High-Temperature Shot Peening on the Surface Integrity of 7010-T7452 Aluminum Alloy . . . . . . . . . . . . . 376 Abouthaina Sadallah, Benoit Changeux, Hong-Yan Miao, Anindya Das, Sylvain Turenne, and Etienne Martin Quality Assessment and Features of Microdrilled Holes in Aluminum Alloy Using Ultrafast Laser . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 380 Suman Chatterjee, Abhijit Suhas Cholkar, David Kinahan, and Darmot Brabazon Surface Characterization Methods to Evaluate Adhesive Bonding Performance of 6xxx Automotive Alloys . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 387 T. Greunz, M. Hafner, R. Gruber, T. Wojcik, J. Duchoslav, and D. Stifter Investigation of Resistance of Intergranular Attack for Various Heat Treated 2011 Alloys After Hard Anodizing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 396 Ilyas Artunc Sari, Gorkem Ozcelik, Zeynep Tutku Ozen, and Onuralp Yucel Fundamental Study on Modified Solidification of 1370 and ALSI7 with and without Commercial Grain Refiners . . . . . . . . . . . . . . . . . . . . . . . . . . . . 410 Robert Fritzsch, Amund Ugelstad, Henrik Gobakken, Silje Li, Shahid Akhtar, Lars Arnberg, and Ragnhild Aune Improving the Mechanical Properties of Cast Aluminum via Ultrasonication-Induced Microstructural Refinement . . . . . . . . . . . . . . . . . . . . . . 422 Katherine Rader, Jens Darsell, Jon Helgeland, Nathan Canfield, Timothy Roosendaal, Ethan Nickerson, Adam Denny, and Aashish Rohatgi Microstructural Changes on a Ternary Al–Cu–Si Eutectic Alloy with Different Pre-heated Mold Temperatures . . . . . . . . . . . . . . . . . . . . . . . . . . . . 429 Seung-Hwan Oh, Sung-Soo Jung, and Young-Cheol Lee Nanoparticle-Enhanced Arc Welding of Aluminum Alloys . . . . . . . . . . . . . . . . . . 436 Narayanan Murali and Xiaochun Li Phase Equilibria in Al–Fe Alloys . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 444 Jožef Medved, Maja Vončina, and Jože Arbeiter Secondary Phase Refinement in Molten Aluminum via Low Power Electric Current Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 451 Jonathan Goettsch, Aaron Gladstein, David Weiss, Ashwin Shahani, and Alan Taub Fluidity and Microstructural Analysis of Al–Ni Alloys with Varied Ni Concentrations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 457 Vigneshwar Hari, Dong Xu, Stuart D. McDonald, Zherui Tong, Dongdong Qu, and Kazuhiro Nogita Effect of Ti Addition on the Microstructure and Mechanical Properties of Hypo-Eutectic and Eutectic Al–Si Alloys . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 465 Chandan Choudhary, K. L. Sahoo, Ashok J. Keche, and D. Mandal

xii

Compatibility Study of Polymeric Binders for Aluminum Binder Jet Parts . . . . . . 471 Solgang Im, Rasim Batmaz, Arunkumar Natarajan, and Étienne Martin Material Evaluation Framework of Additive Manufactured Aluminum Alloys for Space Optical Instruments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 478 Zachary J. Post, Walter R. Zimbeck, Steven M. Storck, Floris van Kempen, Gerard C. J. Otter, John D. Boldt, Ludger van der Laan, Steven R. Szczesniak, Ryan H. Carter, Robert K. Mueller, Salahudin M. Nimer, Douglas B. Trigg, Michael A. Berkson, M. Frank Morgan, and William H. Swartz Comparison of Additively Manufactured and Cast Aluminum A205 Alloy . . . . . . 488 Heidar Karimialavijeh, Morteza Ghasri-Khouzani, Apratim Chakraborty, Jean-Philippe Harvey, and Étienne Martin The Role of Ti and B Additions in Grain Refinement of Al–Mn Alloy During Laser Additive Manufacturing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 493 Qingyu Pan, Monica Kapoor, Sean Mileski, John Carsley, and Xiaoyuan Lou AMAG CrossAlloy®—A Unique Aluminum Alloy Concept for Lightweighting the Future . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 500 Florian Schmid, Lukas Stemper, and Ramona Tosone Effect of Alloying Elements on Corrosion Resistance of Quench-Free Al–Ca Alloys for HPDC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 505 Dmitry Fokin, Sergey Matveev, Roman Vakhromov, Dmitry Ryabov, and Aleksandr Alabin Influence of Increased Cu and Fe Concentrations on the Mechanical Properties of the EN AB-42100 (AlSi7Mg0.3) Aluminum Alloy . . . . . . . . . . . . . . . 511 T. Beyer, D. Ebereonwu, A. Koch, P. Decker, A. Kauws, M. Rosefort, and F. Walther Temperature Dependence of Lattice Misfit in Determining Microstructural Evolution of High Temperature High Strength Aluminium Alloys—A 3D Phase-Field Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 520 Dhanish Sidhik and B. S. Sundar Daniel Microstructure and Mechanical Properties of an Al-Mn-Si Alloy Microalloyed with Sn . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 528 Amir R. Farkoosh, David C. Dunand, and David N. Seidman Innovative Approaches in Development of Aluminium Alloys for Packaging Industry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 535 Stanislav Kores, Simon Strmšek, Maja Vončina, and Jožef Medved The Role of Microstructure on Strength and Fracture Anisotropy Effects in Al–Mg–Si Extrusion Alloys . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 542 S. Kordmir, N. C. Parson, and W. J. Poole Comparison of Experimental Test and Finite Element Simulations of Car Crash Boxes Manufactured with Different Aluminum Alloys . . . . . . . . . . . 549 Görkem Özçelik and Melih Çaylak Exploring Semi-solid Deformation of Al–Cu Alloys by a Quantitative Comparison Between Drained Die Compression Experiments and 3D Discrete Element Method Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 558 Te Cheng Su, Meng Chun Chen, Huai Ren Hu, Ying Hsuan Ko, and Ling En Yao

Contents

Contents

xiii

The Role of Through-Thickness Variation of Texture and Grain Size on Bending Ductility of Al–Mg–Si Profiles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 568 P. Goik, A. Schiffl, H. W. Höppel, and M. Göken Anisotropy of Tearing Behavior in AA7075-T6 Sheet at 200 °C . . . . . . . . . . . . . . 578 Daniel E. Nikolai and Eric M. Taleff Evaluating the Earing Amount of Materials Under Various Chemical Composition and Heat Treatment Processes with Finite Element Simulations of Cup Drawing Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 586 Melih Çaylak, Görkem Özçelik, Abdullah Kağan Kınacı, and Koray Dündar Effect of Al-3Ti-1B-1.5Ce Refiner on Microstructure and Mechanical Properties of A356 Aluminum Alloy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 592 Da Teng, Guangzong Zhang, Shuo Zhang, Junwen Li, Yaodong Zhang, and Renguo Guan Effect of Al-Ti-B Refiner on Microstructure and Properties of A356 Alloy by Continuous Rheo-Extrusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 599 Shuo Zhang, Guangzong Zhang, Da Teng, Junwen Li, and Renguo Guan Effect of Annealing Process on Recrystallization Microstructure and Properties of 1235D Aluminum Alloy Sheet . . . . . . . . . . . . . . . . . . . . . . . . . . 605 Wei Tang, Junpeng Pan, Chao Wu, Hongpo Wang, and Zizong Zhu Effect of Thermal Treatment (T5) on Microstructure and Tensile Properties of Vacuum High Pressure Die Cast Al–Si–Mg Alloy . . . . . . . . . . . . . . 612 Henry Hu, Ali Dhaif, and Kazi Ahmed Numerical Simulation of Flow of Liquid in Molten Pool of Twin-roll Casting Rolling 5182 Aluminum Alloy Strip . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 618 Bingxin Wang, Xiaoping Liang, Wenxiong Duan, and Peng Yang Study of the Solidification Behavior and Homogenization Heat Treatment of the Investment-Cast Al–Cu Foams: Experimental and Modelling Investigations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 627 Waleed Mohammed, Mahan Firoozbakht, and Andreas Bührig–Polaczek Part IV

Aluminum Industry Emissions Measurement, Reporting, and Reduction

The Way Towards Zero Carbon Emissions in Aluminum Electrolysis . . . . . . . . . 637 Gudrun Saevarsdottir, Sai Krishna Padamata, Brandon Nicholas Velasquez, and Halvor Kvande Individual Pot Sampling for Low-Voltage PFC Emissions Characterization and Reduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 646 Brian Zukas and Julie Young Determination of PFC with Canister Sampling and Medusa GC–MS Analysis in Comparison to General IPCC Estimation Methods . . . . . . . . . . . . . . . 653 Henrik Åsheim, Morten Isaksen, Ove Hermansen, Norbert Schmidbauer, and Chris Lunder Heavy Metal Emissions through Particulate Matter from Aluminium Electrolysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 662 Fride Müller, Thor Anders Aarhaug, and Gabriella Tranell

xiv

Contents

Verification of Open-Path Dust Laser for Continuous Monitoring of Diffuse Emissions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 672 Lars Moen Strømsnes, Heiko Gaertner, Steinar Olsen, Peter Geiser, and Bernd Wittgens Characterization of Industrial Hydrocarbon Samples from Anode Baking Furnace Off-Gas Treatment Facility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 680 Kamilla Arnesen, Alexandre Albinet, Claudine Chatellier, Nina Huynh, Thor Anders Aarhaug, Kristian Etienne Einarsrud, and Gabriella Tranell Part V

Aluminum Reduction Technology

CFD Modelling of Solidification and Melting of Bath During Raft Formation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 691 Sindre Engzelius Gylver and Kristian Etienne Einarsrud Experimental Investigation of the Alumina Cloud During Alumina Injections in Low- and High-Temperature Conditions . . . . . . . . . . . . . . . . . . . . . . 699 T. Roger, L. Kiss, L. Dion, S. Guérard, J. F. Bilodeau, and G. Bonneau Fundamental Mass Transfer Correlations Based on Experimental and Literature Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 711 Jonathan Alarie, László I. Kiss, Lukas Dion, Sébastien Guérard, and Jean-François Bilodeau Potential of Production Al–Si Green Alloys in AP18 Aluminium Reduction Cell . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 718 Haris Salihagić Hrenko, Anton Verdenik, Branko Juršek, Dragan Mikša, Maja Vončina, and Jožef Medved Effect of Sulfur Content of Carbon Anode on Measuring Current Efficiency of Aluminum Electrolytic Cell by Gas Analysis Method . . . . . . . . . . . . 724 Kaibin Chen, Shengzhong Bao, Fangfang Zhang, Guanghui Hou, Huaijiang Wang, Lifen Luo, and Xu Shi KF Content on Physical and Chemical Properties of Aluminum Electrolysis Electrolyte . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 731 Changlin Li, Shengzhong Bao, Fangfang Zhang, Gang Li, Shilin Qiu, Fei He, Guanghui Hou, and Huaijiang Wang Fundamentals of Pot Ventilation in Aluminum Smelters . . . . . . . . . . . . . . . . . . . . 737 Diego Oitaben, Samaneh Poursaman, and Stephen Lindsay Evaluation of Methodologies for Assessment of SO3 Concentration in Industrial Off-Gas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 743 Thor Anders Aarhaug, Ole Kjos, Morten Isaksen, and Jan Olav Polden Mathematical Modelling of the Desulfurization of Electrolysis Cell Gases in a Low-Temperature Reactor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 749 Arash Fassadi Chimeh, Duygu Kocaefe, Yasar Kocaefe, Yoann Robert, and Jonathan Bernier Improvements to a Mathematical Model Used to Reproduce the Wave and Stream at the Bath-Metal Interface and Assess Their Impact on the Movement of Alumina Rafts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 757 Thomas Richer, Lukas Dion, Laszlo Kiss, Sébastien Guérard, Jean-François Bilodeau, Guillaume Bonneau, and Martin Truchon

Contents

xv

Numerical Investigation of Thermal, Electrical, and Mechanical Behaviour of Aluminium Cell During Preheating Phase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 765 Simon-Olivier Tremblay, Daniel Marceau, Rohini-Nandan Tripathy, Antoine Godefroy, Duygu Kocaefe, Sébastien Charest, and Jules Côté Simplified 3D MHD Model for Quick Evaluation of Aluminium Electrolysis Cell Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 773 Ievgen Necheporenko, Alexander Arkhipov, and Abdalla Alzarouni Achieving Low Pot Failure Rate at Aditya Aluminium . . . . . . . . . . . . . . . . . . . . . 782 Atanu Maity, Venkannababu Thalagani, Deepak Das, Bhanu Shankar, Anish Das, Kamta Gupta, Madhusmita Sahoo, Shanmukh Rajgire, and Amit Gupta Dissimilar Results in Restarting Two Different Potlines . . . . . . . . . . . . . . . . . . . . . 791 María Carolina Daviou, María Alejandra Mollecker Rausch, Ricardo Alonso, and María Fernanda Jaitman Restart of Albras’ Potline 2—Improving Performance and Changing Paradigms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 797 Ana Carolina Guedes, Ana Renata Nunes, Bruno Vasconcelos, Flávio Silva, João Vilckas, Johnson Machado, Márcio Souza, and Michel Pena Application of Cell Retrofit in GP320 Aluminum Reduction Cell Line . . . . . . . . . 808 Zhuojun Xie, Jian Lu, Weibo Li, Song He, and Xingyu Yang The Expanded Industrial Pilot of SAMI’s NCCT+ Technology . . . . . . . . . . . . . . . 817 Xi Cao, Yafeng Liu, Hongwu Hu, Xuan Wang, Jinlong Hou, Wei Liu, Kangjian Sun, Michael Ren, and Pengfei Du The SY500 Pot Technology Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 826 Kangjian Sun, Yafeng Liu, Hongwu Hu, Xuan Wang, Jinlong Hou, Wei Liu, Xi Cao, and Michael Ren Preheat, Start-Up and Early Operation of DX+ Ultra Pots at 500 kA . . . . . . . . . . 831 Mustafa Mustafa, Abdalla Alzarooni, Konstantin Nikandrov, Nadia Ahli, Aslam Khan, Hassan AlHayyas, Marwan AlUstad, and Sajid Hussain Part VI

Aluminum Waste Management and Utilization

Recovery of Value Added Products from Bauxite Residue . . . . . . . . . . . . . . . . . . 841 Himanshu Tanvar and Brajendra Mishra Current Status and Proposed Economic Incentives for Higher Utilization of Bauxite Residue to Enhance Sustainability of the Aluminum Industry . . . . . . . 849 Subodh K. Das and Muntasir Shahabuddin Aluminium Bahrain (Alba) SPL Sustainable Solution from Landfill to Valuable Feedstock “HiCAL30” . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 852 Khalid Ahmed Shareef, Bernie Cooper, Mohsen Qaidi, Nabeel Ebrahim Mohd Al Jallabi, Fuad A. Hussain Alasfor, and Vijay Rajendran Valorization of Treated Spent Potlining in Cement Industry . . . . . . . . . . . . . . . . . 862 Laurent Birry, Jean Lavoie, Victor Brial, Claudiane Ouellet-Plamondon, Hang Tran, Luca Sorelli, and David Conciatori Aluminum Recycling and Recovery of Other Components from Waste Tetra Pak Aseptic Packages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 867 Ilgım Baltacı, Selçuk Kan, Ahmet Turan, and Onuralp Yücel

xvi

Part VII

Contents

Cast Shop Technology

Electromagnetic Priming of Filtration Systems: Pyrotek EM-DF . . . . . . . . . . . . . . 875 Robert Fritzsch, Joseph Whitworth, Paul Bosworth, and Jason Midgley Automated Metal Cleanliness Analyzer (AMCA): Digital Image Analysis Phase Differentiation and Benchmarking Against PoDFA-Derived Cleanliness Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 882 Hannes Zedel, Robert Fritzsch, Shahid Akhtar, and Ragnhild E. Aune Automated Image Analysis of Metallurgical Grade Samples Reinforced with Machine Learning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 890 Anish K. Nayak, Hannes Zedel, Shahid Akhtar, Robert Fritzsch, and Ragnhild E. Aune Characterization of Low- and High Mg-Containing Aluminum White Dross Using Deterministic Image Analysis of EPMA Scans . . . . . . . . . . . . . . . . . . 898 Cathrine Kyung Won Solem, Hannes Zedel, and Ragnhild E. Aune Assessment of Separation and Agglomeration Tendency of Non-metallic Inclusions in an Electromagnetically Stirred Aluminum Melt . . . . . . . . . . . . . . . . 906 Cong Li, Thien Dang, Mertol Gökelma, Sebastian Zimmermann, Jonas Mitterecker, and Bernd Friedrich Microalloying of Liquid Al–Mg Alloy Studied In-Situ by Laser-Induced Breakdown Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 915 Kristjan Leosson, Sveinn Hinrik Gudmundsson, Arne Petter Ratvik, and Anne Kvithyld Hydrogen Absorption of Aluminum-Magnesium Melts from Humid Atmospheres . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 920 Stefan Tichy, Philip Pucher, Bernd Prillhofer, Stefan Wibner, and Helmut Antrekowitsch Influence of Cryolite Content on the Thermal Properties and Coalescence Efficiency of NaCl–KCl Salt Flux . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 928 Veronica Milani, Alicia Vallejo-Olivares, Gabriella Tranell, and Giulio Timelli Oxidation Study of Al–Mg Alloys in Furnace Atmospheres Using Hydrogen and Methane as Fuel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 936 M. Syvertsen, A. Johansson, J. Lodin, A. Bergin, M. Ommedal, Y. Langsrud, and R. D. Peterson Towards the Efficient Recycling of Used Beverage Cans: Numerical Study and Experimental Validation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 942 Nikolaos Chamakos, Malamatenia Koklioti, Theofani Tzevelekou, Athanasia FIampouri, Ioannis Contopoulos, Alexandros Anestis, Grigorios Galeros, Epameinondas Xenos, and Andreas Mavroudis A Novel Green Melt Technology for Aluminum Alloys . . . . . . . . . . . . . . . . . . . . . 949 Kaborson Ke, Xiyu Wen, and Dongjie Ke MagPump . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 955 Oscar A. Perez and Eishin Takahashi Recycling of Aluminum from Aluminum Food Tubes . . . . . . . . . . . . . . . . . . . . . . 960 Sarina Bao, Anne Kvithyld, Gry Aletta Bjørlykke, and Kurt Sandaunet

Contents

xvii

Recent Studies Using HR-TEM on the Fundamental Mechanism of Nucleation of a-Aluminum on TiB2 in TiB D High-Efficiency Grain Refiners . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 967 John Courtenay and Yun Wang A Cellular Automaton Model for Qualifying Current Grain Refiners and Prescribing Next-Generation Grain Refiners for Aluminium Alloys . . . . . . . . 974 G. Salloum-Abou-Jaoude, S. Sami, A. Jacot, and L. Rougier Modelling Contactless Ultrasound Treatment in a DC Casting Launder . . . . . . . . 980 Christopher Beckwith, Georgi Djambazov, and Koulis Pericleous Numerical Analysis of Channel-Type Segregations in DC Casting Aluminum Slab . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 989 Keisuke Kamiya and Takuya Yamamoto Effect of Casting Variables on Mechanical Properties of Direct Chill Cast Aluminium Alloy Billets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 997 S. P. Mohapatra Stability of SiC and Al2O3 Reinforcement Particles in Thermomechanical Processed Direct Chill Cast Al MMnCs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1005 Abdallah Abu Amara, Guangyu Liu, Dmitry Eskin, and Brian McKay TRC Combi Box: A Compact Inline Melt Treatment Unit for Continuous Casting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1015 M. Gorsunova-Balkenhol, M. Badowski, M. Betzing, J. Stotz, and Ø. Pedersen CFD Modeling of Thin Sheet Product Using the Horizontal Single Belt Casting Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1022 Daniel R. Gonzalez-Morales, Mihaiela M. Isac, and Roderick I. L. Guthrie Numerical and Experimental Investigation of Twin-Roll Casting of Aluminum–Lithium Strips . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1031 Olexandr Grydin, Kai-Uwe Garthe, Xueyang Yuan, Jette Broer, Olaf Keßler, Rostislav Králík, Miroslav Cieslar, and Mirko Schaper Segregation Mechanisms and Their Effects on the Aluminium Flat Rolled Products (Sheet/Foil) Produced by Twin Roll Casting Technology . . . . . . . . . . . . 1038 Onur Birbasar, Feyza Denizli, Eda Özkaya, Samet Sevinç, Ali Ulus, and Canan İnel Novel Methods for Roll Texturing: EDT and Sandblast Applications for Aluminium Twin-Roll Cast and Cold Rolling . . . . . . . . . . . . . . . . . . . . . . . . . 1043 Yusuf Özçetin, Onur Birbaşar, Ali Ulus, Koray Dündar, Feyza Denizli, and Canan İnel Characterization of 8006 Aluminium Alloy Casted by TRC Technology with Steel–Steel and Copper–Copper Roll Pairs . . . . . . . . . . . . . . . . . . . . . . . . . . 1050 Feyza Denizli, Onur Birbaşar, Koray Dündar, Yusuf Özçetin, Ali Ulus, and Canan İnel Tailoring the As-Cast Microstructure of Twin-Roll Cast AA3105 Alloy Produced by St/Cu Shell Pair . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1054 Cemil Işıksaçan, Mert Gülver, Hikmet Kayaçetin, Onur Meydanoglu, and Erdem Atar Designing a Safe Casthouse . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1058 Alex W. Lowery

xviii

Contents

Operations Assisting and Predictive Maintenance Tools in Cast Houses: Examples from AMAG Casting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1064 Alexander Poscher, Martin Mönius, Eduard Faschang, and Bernd Prillhofer Counter-Gravity Casting of Al Alloys: Microstructure and Properties . . . . . . . . . 1071 K. Georgarakis, J. Vian, D. Sgardelis, B. Souchon, Y. Chao, K. Konakoglou, M. Stiehler, and M. Jolly Defect Minimisation in Vacuum-Assisted Plaster Mould Investment Casting Through Simulation of High-Value Aluminium Alloy Components . . . . . . . . . . . . 1078 Emanuele Pagone, Christopher Jones, John Forde, William Shaw, Mark Jolly, and Konstantinos Salonitis Part VIII

Electrode Technology for Aluminum Production

Partial Replacement of Coke with Biocoke: Effect of Biocoke Production Temperature on Anode Quality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1089 Belkacem Amara, Duygu Kocaefe, Yasar Kocaefe, Jules Côté, and André Gilbert Method for Calcined Petroleum Coke Evaluation to Improve the Anode Quality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1095 Sheetal Gupta, Suwarna Mahajan, Amit Gupta, and Vilas Tathavadkar Influence of Crusher Type and Particle Shape on the Bulk Density of Blended Shaft and Hearth Calcined Anode Grade Petroleum Coke . . . . . . . . . 1101 Howard Childs, Mike Davidson, and Barry Sadler Managing Green Petroleum Coke Properties Variations on Prebaked Anodes Quality in Aluminium Bahrain “Alba” . . . . . . . . . . . . . . . . . . . . . . . . . . . 1107 Hesham Buhazza, Vasantha Kumar Rangasamy, Nabeel Ebrahim Mohd Al Jallabi, Taleb Al Ansari, Abdulmohsin Hasan Radhi, Francois Morales, and Abdulla Habib Development of an Iron Aluminide Coating for Anticorrosion Protection of Anodic Pins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1117 Henrique Santos, Roberto Seno, Antonio Couto, Alex Fukunaga, and Adriano Francisco New Methods to Determine PAH Emission Dynamics During Electrode Mass Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1124 Ole Kjos, Thor Anders Aarhaug, Heiko Gaertner, Bente Håland, Jens Christian Fjelldal, Katarina Jakovljevic, Oscar Espeland, and Ida Kero Investigation of the Stacking Effects on the Electrical Resistivity of Industrial Baked Anodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1132 Thierno Saidou Barry, Donald Picard, Guillaume Gauvin, Julien Lauzon-Gauthier, and Houshang Alamdari New Generation Anode Baking Furnace: Use of Prefabrication for Additional Conversions at Bell-Bay Plant . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1144 Sandra Besson, David Deneef, Anthony Reeve, Youcef Nadjem, Meaghan Noonan, and Roy Cahill AHEX Full Scale Experiences at TRIMET Aluminium SE . . . . . . . . . . . . . . . . . . 1149 Anders Sørhuus, Vrauke Zeibig, Eivind Holmefjord, Ömer Mercan, and Elmar Sturm Inline Modal Detection System of Anodes and Cathodes Measuring Cracks and Physical Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1156 Dag Herman Andersen and Ole Kristian Brandtzaeg

Contents

xix

Part IX

Scandium Extraction and Use in Aluminum Alloys

Investigations into Optimized Industrial Pilot Scale BR Leaching for Sc Extraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1167 Efthymios Balomenos, Panagiotis Davris, Alexandra Apostolopoulou, Danai Marinos, Elena Mikeli, Aikaterini Toli, Dimitrios Kotsanis, Grigoris Paschalis, and Dimitrios Panias Solvent Extraction of Scandium from Titanium Process Solutions . . . . . . . . . . . . 1173 Dimitrios Filippou, Michel Paquin, Yves Pépin, Mike Johnson, and Niels Verbaan State of the Art Technologies for Scandium Recovery, Purification, and Aluminum-Scandium Alloy Production . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1181 Anne Marie Reyes, Gomer Abrenica, and Ghazaleh Nazari FEA Materials—Aluminum Scandium Master Alloy Production Technology . . . . 1190 Rick Salvucci, Brian Hunt, and Eugene Prahin Scandium Master Alloy Production Via Sulfidation and Vacuum Aluminothermic Reduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1195 Caspar Stinn, Ethan Benderly-Kremen, and Antoine Allanore European Scandium for a Lighter and Greener Future . . . . . . . . . . . . . . . . . . . . 1204 Henk van der Laan, Beate Orberger, Carsten Dittrich, Robin Scharfenberg, Edward Peters, Georges Croisé, Pierre Feydi, Carolin Maier, Richard Schneider, Bernd Friedrich, Yashvi Baria, Konstantinos Sakkas, and Christos Georgopolous Formation of Al3Sc Dispersoids and Associated Strengthening . . . . . . . . . . . . . . . 1207 Thomas Dorin, Lu Jiang, and Timothy Langan Use of Sc to Improve the Properties of AA5083 Cast and Rolled Products . . . . . . 1213 Paul Rometsch, Jerome Fourmann, Emad Elgallad, and X.-Grant Chen Efficiency of Sc for Strengthening and Formability Improvement of 5XXX BIW Sheets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1223 Margarita Nikitina, Aleksandr Gradoboev, Dmitry Ryabov, Roman Vakhromov, Viktor Mann, and Aleksandr Krokhin Effect of Sc and Zr Additions on Dispersoid Microstructure and Mechanical Properties of Hot-Rolled AA5083 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1229 Ahmed Y. Algendy, Kun Liu, Paul Rometsch, Nick Parson, and X.-Grant Chen Effect of Cooling Rate on W-Phase Formation in Al-Cu-Sc Alloys . . . . . . . . . . . . 1238 Austin DePottey, Lu Jiang, Thomas Dorin, Thomas Wood, Timothy Langan, and Paul Sanders Solute Clustering During Natural Ageing in Al-Cu-(Sc)-(Zr) Alloys . . . . . . . . . . . 1247 Lu Jiang, Kathleen Wood, Anna Sokolova, Robert Knott, Timothy Langan, and Thomas Dorin Effect of Zr and Sc on Intermetallic Morphology and Hardening of an Al–Fe Alloy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1252 Suwaree Chankitmunkong, Dmitry G. Eskin, Chaowalit Limmaneevichitr, Phromphong Pandee, and Onnjira Diewwanit Effect of Sc, Zr, and Other REE on the 1XXX Conductive Aluminum Alloy Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1257 Ruslan Aliev, Alexander Gradoboev, Dmitry Ryabov, Roman Vakhromov, Aleksandr Krokhin, and Viktor Mann

xx

Contents

Developing Al-Zr-Sc Alloys as High-Temperature-Resistant Conductors for Electric Overhead Line Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1266 Quan Shao, Emad Elgallad, Alexandre Maltais, and X.-Grant Chen The Development of New Aluminum Alloys for the Laser-Powder Bed Fusion Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1273 Nathan Andrew Smith, Mostafa Yakout, Mohamed Elbestawi, Phil Chataigneau, and Peter Cashin Sustainable Scandium Recovery Method from Metallic 3D Printing Powders . . . . 1284 Bengi Yagmurlu and Carsten Dittrich New Aluminium–Scandium Welding Wires for Additive Manufacturing . . . . . . . . 1289 Thomas Dorin, Lu Jiang, and Andrew Sales Comparative Study of Al-Mg-Ti(-Sc-Zr) Alloys Fabricated by Cold Metal Transfer and Electron Beam Additive Manufacturing . . . . . . . . . . . . . . . . . . . . . . 1295 Jiangqi Zhu, Xingchen Yan, Timothy Langan, and Jian-Feng Nie Dissolution and Development of Al3(Sc, Zr) Dispersoids in Structures Produced Via Wire Arc Additive Manufacturing . . . . . . . . . . . . . . . . . . . . . . . . . 1300 Sonja Blickley, Tori Nizzi, Anna Palmcook, Austin Schaub, Erico Freitas, Tim Langan, Carson Williams, and Paul Sanders Correction to: Fabrication of Bright-Rolled Aluminum Suitable for Design Elements in the Automotive Industry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Anita Gründlinger, Peter Johann Uggowitzer, and Josef Berneder

C1

Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1307 Subject Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1313

About the Editor

Stephan Broek is the president and principle consultant of Kensington Technology Inc. in Burlington, Ontario, Canada. He has a B.Sc. in Chemical Engineering from the University of Applied Sciences in Amsterdam, with a postgraduate diploma in Process Technology from Twente University in Enschede, The Netherlands. For twelve years, he worked at Tata Steel Europe in the engineering group of Danieli Corus. His assignments included process engineering, technology management, and product development. In 2003, Stephan joined Hatch, Mississauga, Canada, where he worked for six years on commercialization and development of technologies. For the next twelve years, Stephan went on to dedicate his work to the global aluminum industry where he is recognized for his know-how in environmental engineering and technologies. He was also actively working on smelter feasibility studies and project management, including the rebuild of potline 2 at PT Inalum. In October of 2021, he joined Boston Metal to provide leadership in the industrial development and deployment of inert anode technology for metal production. This is an exciting and game-changing development that can have a real impact on climate change linked to global metal production. As of March 2023 he is the president and principle consultant of Kensington Technology Inc., a specialized consulting firm primarily focused on the global aluminum industry. Stephan has co-authored over 40 articles and papers and is a regular speaker at international conferences. Over the years, he has been actively involved with TMS, and has served as secretary of the Aluminum Committee for six years. In addition, he has been a subject chair and session chair in several of the TMS annual meetings. Stephan also shares his knowledge within our industry. He created and lectured in several TMS short courses and is a lead instructor of the Potline Scrubber & Fugitive Emissions (PSFE) industrial training course that is part of professional development by TMS.

xxi

Program Organizers

60 Years of Taking Aluminum Smelting Research and Development from New Zealand to the World: An LMD Symposium in Honor of Barry J. Welch Mark Dorreen has his Bachelor and Ph.D. degrees in Chemical and Materials Engineering, and a broad range of leadership experience in both R&D and innovation commercialization. Since late 2022 he has been Research Group Leader–Electrochemical Processing at CSIRO’s Mineral Resources division in Australia. Prior to that he was the Director of the Light Metals Research Centre (a commercial unit within the University of Auckland, focused on technical innovation within the global aluminum industry), followed by some time as the founding CEO of Enpot Ltd (a spinout company from the LMRC). Earlier in his career, Dr. Dorreen had several technical and commercial roles within the aluminum and steel industries in New Zealand. Alumina & Bauxite Errol Jaeger has more than 35 years of experience in the aluminum industry. His background has been primarily in the bauxite and alumina sectors and includes basic and conceptual engineering studies of Bayer process plants, greenfield and brownfield commissioning and start-up planning, project management, and due diligence. Errol has worked with Alcoa, BHP-Billiton, Kaiser Aluminum, Ma’aden Aluminium, Vimetco Management, and SNC-Lavalin.

xxiii

xxiv

Program Organizers

Aluminum Alloys, Characterization and Processing Julie Lévesque (P.Eng., Ph.D.) is an R&D project manager at Quebec Metallurgy Center (CMQ), where she is the leader of the metals forming and assembly axis. As such, she is responsible for projects related to welding and mechanical forming operations, including rolling, deep drawing, wire drawing, stamping, forging, and other solid state forming processes. She holds a master’s degree from Laval University, a Ph.D. from the University of Sherbrooke, and has close to 15 years of experience in the optimization of forming processes in collaboration with industries in the automotive, aerospace, and packaging industry, amongst others. She is an active member of the Aluminium Research Center – REGAL and the director of the Innovation Catalyst in Advanced Manufacturing Processes (college-university-industry cluster). Julie is also the 3rd VP of the Metallurgy and Materials Society of the Canadian Institute of Mining, Metallurgy and Petroleum (MetSoc of CIM). Aluminum Industry Emissions Measurement, Reporting & Reduction and Aluminum Waste Management and Utilization Pernelle Nunez joined the International Aluminium Institute in 2015 and is the Deputy Secretary General and Director of Sustainability. The IAI is the leading global association for the aluminum industry with members involved with bauxite mining, alumina and aluminum production across all major producing regions. In addition to being a spokesperson for the IAI, Pernelle is responsible for the Institute’s broad sustainability work program including its Greenhouse Gas Pathways Working Group and Environment and Energy Committee. Pernelle has led many collaborative industry projects across topics including environmental footprint analyses, sustainable waste management, responsible mining practices, and GHG emissions accounting and pathways. Pernelle holds a MSci in Geology from The Royal School of Mines, Imperial College London, and a Post Graduate Certificate in Sustainable Value Chains from the University of Cambridge. Aluminum Reduction Technology Pierre Marcellin is Principal Advisor in the Rio Tinto Aluminium Technology Solutions (ATS) department in France. This department designs, delivers, and supports aluminum process control supervision and regulation systems, particularly ALPSYS pot control system. Pierre oversees the technical development of ATS solutions and provides assistance to ATS solutions users and delivery projects. Pierre graduated from the Ecole Normale Supérieure (Paris, France) with a Civil Engineering Degree and a post graduate research study in computer science. Pierre has been working for Rio Tinto and its predecessors since 1989 and has held various positions in aluminum reduction process control (in R&D in LRF/France, in ALBA

Program Organizers

xxv

smelter/Bahrain, and St Jean de Maurienne smelter/France), in the computer department (Dunkerque smelter/France), and managing major projects in the St Jean de Maurienne smelter and in ALUVAL (Technical Sales projects). Cast Shop Technology Halldor Gudmundsson received his B.Sc. in Physics in 1986 from the University of Iceland, M.Sc. in Materials Science in 1989 from the University of Virginia, and Master in Light Metals Reduction Technology 2009 from the University of Auckland. Halldor has taught engineering courses on and off for the past 30 years at the University of Iceland. He was an adjunct lecturer in the Mechanical Engineering department from 1993 to1998. He joined Nordural in 1998, and prior to that spent about 10 years doing various consulting work and testing in the field of metallurgy and materials science. His main area of expertise is microstructural analysis using optical and electron microscopy. At Nordural he has had various responsibilities; quality control of products and raw materials, environmental management, process control in potrooms (2005–2010), technical manager of reduction (2010–2014), and product development manager in the casthouse. Electrode Technology for Aluminum Production Roy Cahill has 26 years of experience in the aluminum industry, spanning 3 companies, and has worked across various areas of the industry such as alumina refining, anode production, coke calcination, smelting, and spent potliner treatment systems. He currently is the Carbon Manager for the Smelter Technical Support Team within Rio Tinto Aluminium’s Pacific Operations. The Team is responsible for development of data analysis and visualization tools, process stability, technical training of both new engineers and operators, process technical support, selection and utilization of anode raw materials (coke, pitch, and refractories), and future technical strategies. Roy Cahill has a Ph.D. in Inorganic Chemistry from Texas A&M University. Scandium Extraction and Use in Aluminum Alloys Timothy Langan has extensive experience in all aspects of the commercialization and development of advanced material technologies, ranging from fundamental research up through product engineering and launch. His primary area of technical expertise involves modifying microstructures and surfaces of advanced materials to optimize corrosion resistance and mechanical properties. Dr. Langan is currently working with Sunrise Energy Metals to develop applications for scandium that will be produced from its Sunrise Project in New South Wales, Australia. In this role, Dr. Langan is working with industrial partners to guide, develop, and focus

xxvi

Program Organizers

research efforts on aluminum-scandium alloys at universities including Deakin University, Michigan Technological University, Monash University, and Chongqing University. Dr. Langan has extensive experience with aluminum alloy development. Working at Surface Treatment Technologies (ST2), he developed a family of advanced corrosion resistant weldable scandium containing aluminum alloys. Dr. Langan joined ST2 after working as a Technical Director of Ashurst Technologies Corporation. In this capacity, he was instrumental in developing US markets for materials technology from the Former Soviet Union, particularly Ukraine. Before joining Ashurst, Dr. Langan was Group Leader, Special Programs Group, in the Advanced Alloys Department at Martin Marietta Laboratories in Baltimore, Maryland. In this position he co-invented a family of aluminum-lithium alloys known as Weldalite™ alloys and worked with engineers at Martin Marietta, NASA, and Reynolds Metals to successfully use the alloy to build the Super-lightweight tank, which first flew as the space shuttle main fuel tank in 1998.

Aluminum Committee 2022–2026

Executive Committee 2022–2023 Chairperson Dmitry Eskin, Brunel University, Middlesex, UK Vice Chairperson Stephan Broek, Kensington Technology Inc., Burlington, ON, Canada Past Chairperson Linus Perander, Yara International, Sandefjord, Norway Secretary Kristian Etienne Einarsrud, Norwegian University of Science & Technology, Trondheim, Norway JOM Advisor Anne Kvithyld, SINTEF, Trondheim, Norway Light Metals Division Chair Eddie McRae Williams, Arconic, Pennsylvania, USA

Members-At-Large Through 2023 Houshang Alamdari, Laval University, Quebec, Canada Corleen Chesonis, Metal Quality Solutions LLC, Pennsylvania, USA Mark Dorreen, Auckland, New Zealand Marc Dupuis, Quebec, Canada Les Edwards, Rain Carbon Inc, Louisiana, USA John Grandfield, Grandfield Technology Pty Ltd., Victoria, Australia John Griffin, ACT LLC, Pennsylvania, USA Ali Jassim, Emirates Global Aluminum, Dubai, United Arab Emirates Lorentz Petter Lossius, Hydro Aluminium AS, Øvre Årdal, Norway Eric Nyberg, Kaiser Aluminum Trentwood, Washington, USA Arne Ratvik, SINTEF, Trondheim, Norway Barry Sadler, Net Carbon Consulting Pty Ltd. Victoria, Australia David Sydney Wong, Queensland, Australia

Members-At-Large Through 2024 Stephan Broek, Kensington Technology Inc., Burlington, ON, Canada Dmitry Eskin, Brunel University, Middlesex, UK xxvii

xxviii

Duygu Kocaefe, University of Quebec, Quebec, Canada Johannes Morscheiser, Aleris Rolled Products Germany GmbH, Koblenz, Germany Jayson Tessier, Alcoa Corporation, Quebec, Canada Alan Tomsett, Rio Tinto Pacific Operations, Queensland, Australia

Members-At-Large Through 2025 Linus Perander, Yara International, Sandefjord, Norway Derek Santangelo, Hatch, Quebec, Canada Samuel Wagstaff, Oculatus Consulting, Georgia, USA

Members-At-Large Through 2026 Kristian Etienne Einarsrud, Norwegian University of Science & Technology, Trondheim, Norway Mertol Gökelma, Izmir Institute of Technology, Urla, Turkey Stephen Instone, Speira GmbH, Bonn, Germany Martin Iraizoz, Aluar Aluminum, Puerto Madryn, Argentina Anne Kvithyld, SINTEF, Trondheim, Norway Julien Lauzon-Gauthier, Alcoa Corporation, Quebec, Canada Pascal Lavoie, Alcoa Canada, Quebec, Canada Olivier Martin, Rio Tinto, Saint Jean, France Ray Peterson, Real Alloy, Tennessee, USA Andre Phillion, McMaster University, Ontario, Canada Daniel Richard, Hatch, Quebec, Canada Gudrun Saevarsdottir, Reykjavik University, Reykjavik, Iceland Andre-Felipe Schneider, Hatch, Quebec, Canada Dimitry Sediako, University of British Columbia, British Columbia, Canada Camilla Sommerseth, SINTEF, Mo i Rana, Norway Eddie McRae Williams, Arconic, Pennsylvania, USA Andrey Yasinskiy, Siberia Federal University, Krasnoyarsk, Russia

Aluminum Committee 2022–2026

Part I 60 Years of Taking Aluminum Smelting Research and Development from New Zealand to the World: An LMD Symposium in Honor of Barry J. Welch

What Makes TMS Special? Let Us Consider a Case Study in Volunteer Excellence: Barry J. Welch James J. Robinson

Abstract

The Minerals, Metals & Materials Society (TMS) classically serves as an intersection point of academia and industry, giving experts opportunities to expand their expertise while simultaneously helping other experts expand their own expertise in kind. It is a complexly intertwined yet beneficial set of relationships that supports technology transfer, that improves or reinvents the state of the art, and that thrives on collegiality. Via his volunteer work through TMS, Professor Barry Welch has consistently and tirelessly exemplified all of these qualities and has routinely and positively influenced the work and workforce of the worldwide aluminum industry. This paper presents personal reflections on some specific instances of Prof. Welch’s volunteer activities that have made TMS a better professional society and that have elevated the field and the good of the order in the process. Keywords

JOM




Light Metals Division




TMS




Volunteerism

metals industries have invented and sometimes reinvented themselves many times since 1871, one fact has remained constant: Professional societies have thrived because of volunteer engagement. Within the TMS community, generation after generation of volunteers have lent their energy, ingenuity, experience, and reputations to advance the collective good of the order by leveraging TMS’ power to assemble networks of like-minded volunteers and amplify their collective capabilities. Great associations are great because they have great volunteers, and those great volunteers infectiously inspire other volunteers toward greatness. TMS is a great professional society because of this virtuous cycle. Within TMS, Barry Welch has been a great volunteer. I feel very confident in this assertion because of my first-hand observation of Barry’s volunteer work from the perspective of a professional staff member. I have witnessed him innovating for the Society, giving generously of his time and resources, and inspiring others to engage as well. He has helped me grow as a professional, and he has helped TMS grow as a professional society. What he has not helped is my memory, so apologies in advance for any misremembering on my part. I remember it all like it was yesterday, but my memory of actual yesterday is not always perfect!

Introduction The Minerals, Metals & Materials Society, a.k.a. TMS, has a lengthy and distinguished history that stretches back to 1871 with the founding of what is today called the American Institute of Mining, Metallurgical, and Petroleum Engineers. Over the 152 years since the founding of AIME and the 65 years since TMS was formed within it, much has changed in the metals community, not the least of which is the development of the entire aluminum industry. While the J. J. Robinson (&) The Minerals, Metals & Materials Society (TMS), 5700 Corporate Drive Suite 750, Pittsburgh, PA, USA e-mail: [email protected]

The JOM Years I first became acquainted with Barry Welch in the early 1990s as he was taking on the volunteer role of JOM Advisor from the Aluminum Committee. In JOM parlance, an “advisor” is a volunteer from a technical committee who gathers a suite of four to six papers for publication in JOM; these papers should represent the technical interests of the committee to the full TMS membership. The papers might be invited or could be “over the transom” submissions. It is the job of the advisor to arrange a review of the papers and decide which ones to publish. Still used robustly today, the

© The Minerals, Metals & Materials Society 2023 S. Broek (ed.), Light Metals 2023, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-031-22532-1_1

3

4

advisor system was less than a decade old when Barry became part of the advisor team. He joined at a very meaningful time. During the latter half of the 1980s, JOM (which was then still named Journal of Metals) and the Aluminum Committee had a lively and sometimes contentious relationship. The contentions generally centered on the journal’s too-limited attention to primary aluminum production and cast shop technologies—the stuff that has powered the Light Metals volume year after year since 1971. Passions on the Aluminum Committee to see more representative and expanded coverage in JOM ran deeply. The question came down to this: Why is the technical committee that produces the most technical programming for the Society and producing the most sought-after proceedings volume so proportionately underrepresented in the journal that is supposed to be reflective of the interests of the entire membership? Many philosophical and logistical debates ensued within the volunteer community and among volunteers and staff. How do I know about this? I participated in many of these discussions as I joined TMS to work on the Journal of Metals in 1984; the initial job was to edit manuscripts and transition our manuscript preparation from typewritten papers to digital files. More responsibility soon followed, and it didn’t take long to become Associate Editor, then Managing Editor, and then Editor. These transitions occurred in parallel with the how-to-better-represent-aluminum discussions that were underway with the volunteers, and I became more imbued with each title change. In my experience, solutions to problems come from people working together rather than acting dogmatically. That was certainly the case with JOM. The competing perspectives were harmonized by two bridge-building volunteers representing the Aluminum Committee to the journal as advisors: Wayne Hale and Rod Zabreznik. Rod and Wayne served at times individually as advisors and at times as a duo. They worked hard on behalf of the Aluminum Committee, advocated on behalf of TMS being the global home of the aluminum industry, and expanded my understanding of primary aluminum considerably. What was the harmonious solution? The journal would publish aluminum industry papers quarterly—an aluminum topic in the February, May, August, and November issues. We went from a plan of 1–5 aluminum papers annually to 20 or more. This was a big publishing commitment for the volunteers to fulfil as so many more quality papers would have to be acquired annually. And it was a big commitment from JOM as this meant bigger issues and a bigger expense to TMS as the publisher. (At the time, JOM operated at a deficit to the Society. So, publishing more papers meant increasing the deficit operation.)

J. J. Robinson

Having helped build a new aluminum publishing model, Rod and Wayne began to step back from their focus on JOM and were advisors no more. Rod would go on to chair the Light Metals Division, and Wayne would not only chair the division but serve as TMS President in 2001. With Rod and Wayne’s JOM retirements from collecting and reviewing JOM papers, the question became who should actualize the new publishing model as the next JOM advisor from the Aluminum Committee? Would the model be more vulnerable to failure without the participation of the volunteer architects? Could we actually realize the expanded commitments? The answers took form in the person of Barry Welch. I do not know if any cajoling or arm-twisting was performed in persuading Barry to accept this assignment, but Barry immediately presented enthusiasm and energy in being the next advisor from the Aluminum Committee. He recognized that the new commitments would not be satisfied effectively without effort. We needed quantity and quality. Barry embraced the challenge! And, he set the bar high as he has very high standards for want constitutes a publicationworthy paper. Could the criteria be satisfied? The answer turned out to be “yes” as Barry knew that we could not just wait for more papers to come flooding in unsolicited. Instead, he devised a three-pronged action plan. First, he invited people in his network to contribute papers. Barry has a large network, and he is an exceedingly difficult person to whom to say “no.” By my admittedly subjective impression, the number of high-profile and up-and-coming contributors of papers to the journal went on the ascend. Second, Barry agreed to occasionally author or co-author papers for JOM. Barry writes an exceptional technical paper and overview. This would be leadership by example. Third, Barry thought it important that publication in JOM provides a prestigious opportunity beyond the standard recognition that comes with being accepted for publication. An inspirational model could be taken from the aluminum community, where subject awards are issued for papers published in Light Metals. He set to work with the Aluminum Committee and the Light Metals Division to create a new award: The Light Metals Division JOM Best Paper Award. It was this final development that left me amazed: The journal had accomplished a long and difficult journey from being something that held little to no appeal to members of the aluminum community to being one where the journal was publishing enough quality content that multiple papers could be worthy of award consideration. And speaking of multiple papers, the first award recipient, Manaktala, was recognized for a set of three:

What Makes TMS Special? Let Us Consider a Case Study …

• Nov. 1992: “The Primary Aluminum Industry in the Commonwealth of Independent States—Part I.” • Feb. 1993: “The Primary Aluminum Industry in the Commonwealth of Independent States—Part II.” • May 1993: “The Primary Aluminum Industry in the Commonwealth of Independent States—Part III.” It was a fascinating set of papers and very timely at this moment in history. It was well-deserving of the award, as have the many papers that have followed. The second award recipient was Alton T. Tabereaux for November 1994: “Anode Effects, PFCs, Global Warming, and the Aluminum Industry”—some topics never go out of fashion. Interestingly, Alton’s paper is one of only two bylined papers in JOM history to use the phrase “global warming” in the title. (The other was “Global Warming and the Primary Metals Industry” by David Forrest and Julian Szekely in December 1991.)

Announcement of Manaktala receiving the first Light Metals Division JOM Best Paper Award at TMS1994. (From the LMD Edition newsletter insert in the May 1994 JOM.)

Barry himself would be the primary author on three papers to receive the award (after his service as JOM advisor had concluded, of course): • Feb. 1998: “Cathode Performance: The Influence of Design, Operations, and Operating Conditions.” • May 1999: “Aluminum Production Paths in the New Millennium.”

5

• Nov. 2008: “Applying Fundamental Data to Reduce the Carbon Dioxide Footprint of Aluminum Smelters.” He was also co-author of another of the award-receiving papers: “The Multivariable Model-Based Control of the Non-alumina Electrolyte Variables in Aluminum Smelting Cells” by Fiona J. Stevens McFadden, Barry J. Welch, and Paul C. Austin (February 2006). Fiona was one of Barry’s many excellent students, and he clearly mentored her well as she followed him as JOM advisor from the Aluminum Committee. She ably kept all of the wheels turning and the progress continued. With these initiatives, it seemed that JOM had cracked the code of robustly engaging primary aluminum. No question that many, many volunteers played important roles in this progress, but I will always think of Barry as the anchor runner for the relay team that led to this success. He was the

Alton Tabereaux (right) receives the second Light Metals Division JOM Best Paper Award from Light Metals Division Vice Chair Rodney Zabreznik (left) at TMS1995. (From the LMD Edition newsletter insert in the July 1995 JOM.)

propagator of the good outcomes and the blunting agent of potential problems. He was remarkably generous of himself. Most importantly, he had established a replicable model that future JOM advisors could employ and build upon in their own unique ways. And this all took place before smartphones, low-cost international phone calls, text messages, websites, and email. We used the fax machine as our “miracle” technology!

6

J. J. Robinson

A Passion for Industry and for TMS Barry’s departure from the advisor role did not mean that Barry ceased to volunteer for TMS. The title of this paper cites “volunteer excellence” for a reason. While Barry is well-known and much revered as an educator and shaper of multiple generations of scientists and engineers, I often think of Barry as an “industrial guy” within TMS. Like many professors in the Society’s ranks, he speaks with passion about the interests and needs of the industrial community and has numerous industrial ties. I see him as walking in both worlds and being highly respected in both. I have observed him manifest this duality via his volunteerism in TMS with JOM and beyond. For example, Barry has gone straight to the cell line and taught countless shop-floor and plant-management professionals about optimizing anodes, cathodes, reduction cells, emissions, and the best practices of all of those elements that constitute modern aluminum reduction. I’ve never asked him, but I often wondered if there was a primary aluminum production facility in the world that he has neither visited nor lectured at. I hope that he never disabuses me of my impression that he has been everywhere. Whenever I think of Barry, I am put in mind of American singer/songwriter Johnny Cash who sang in the song “I’ve Been Everywhere,” I’ve been everywhere, man Crossed the deserts bare, man I’ve breathed the mountain air, man Of travel I’ve had my share, man I’ve been everywhere

Very notably as a world traveller while volunteering for TMS, Barry engaged with Halvor Kvande, Alton Tabereaux, and other aluminum industry luminaries in teaching the much-esteemed TMS short course: Industrial Aluminum Electrolysis Course: The Definitive Course on Theory and Practice of Primary Aluminum Production. This course is an exemplar of how TMS volunteers can come together and advance the industrial community. Its instructors have travelled to almost every continent to teach on site at a smelter operation. Some course participants hail from the host operation’s site; others travel in from all points of the compass for the unique experience. All are provided wonderful opportunities to witness plant operations while the instructors present the latest techniques in electrolysis, link theory, and practice and provide tools and information to help the course participants save costs, improve productivity, and manage emissions. Thanks to the volunteer leadership, this course is looked upon as the model continuing education activity within the Society. It both strengthens the workforce and inspires future advancement.

The Industrial Aluminum Electrolysis Course very much advances the TMS mission, which is “to promote the global science and engineering professions concerned with minerals, metals, and materials.” It does this in two ways—the first way is obvious: It provides a platform for our volunteers to develop and deliver a curriculum needed by the community, it helps participants be more informed professionals and more effective metal producers, and it serves as a model of success that other volunteers can emulate in developing their own courses through TMS. The second way is less obvious: The course provides revenue to the Society greater than the cost that is required to conduct the work. That surplus is then used to subsidize the activities by the society that are not self-sustaining (which is most of them). Another way that Barry supported the mission of TMS was by helping to establish a limited-duration award within TMS: “The Vittorio de Nora Prize for Environmental Improvements in Metallurgical Industries.” Vittorio de Nora was a pioneer in the materials processing field and a great Italian technologist. Barry facilitated communication between the de Nora Family and TMS leadership to create an endowment within TMS that would allow the issuance of a $20,000 prize. It was presented six times in the 2010s with the purpose of recognizing outstanding materials science research and development contributions to the reduction of environmental impacts, and particularly greenhouse gas emissions, as applied in global metallurgical industries, especially focused on extractive processing. I thought back on this prize while listening to the recipient of the 2012 award, Jim Yurko of Apple, who presented the all-conference plenary talk at TMS2022. In the talk, he referenced Apple’s collaboration with ELYSIS on employing aluminum produced without greenhouse gasses. What splendid symmetry with de Nora, I thought! One more example of volunteerism in action? Barry has served on the Aluminum Committee many times (of course!) and was its Chair leading up to TMS1998. Many know that the chair of the committee serves as Editor of the annual Light Metals volume, and he was Editor of Light Metals 1998. A big job. It is more than fitting that Barry is a TMS Fellow (Class of 2015). It is the highest honor that TMS bestows on a member. His spot-on citation reads, For significant contributions to the advancement of aluminum smelting technology through pioneering research in aluminum electrolysis cell reactions and fundamental processes followed by outstanding teaching to students and engineering practitioners.

Well said!

What Makes TMS Special? Let Us Consider a Case Study …

7

2014 TMS President Hani Henein (left) presents the TMS Fellow Award to Barry Welch (right).

Closing Thoughts Beyond the grand efforts that I recount in this brief paper, Barry has benefited the aluminum community and TMS through his volunteerism in countless ways. Still, I don’t think that his greatest impact within TMS has been through the formalisms of these high-visibility activities. Similarly, I also don’t think that it has been by presenting at a microphone, or in writing papers, or in building a consensus among peers, or

in chairing a discussion amongst volunteers. Instead, I suspect that his greatest impact as a volunteer has been in the hallways of conferences and in those informal moments after a session’s conclusion—those one-on-one impromptu times when you can seek his advice or opinion, hear his candid thoughts, benefit from some mentorship, kick around an idea, or just chat about how the family is doing. Barry’s impact has been strong in TMS over the decades and his standing as a role model will surely influence current and future generations of volunteers for many decades to come.

Meeting the Requirements of Potline Customers: The Largely Unmet Challenges Set by Barry Welch to Carbon Anode Producers Barry Sadler and Alan Tomsett

Abstract

Barry Welch is well recognized for his contribution to advancing the science and practice of smelting Alumina to Aluminium Metal. He has also made a significant contribution to advancing anode technology, in particular through the work of his students. This is explored further in an accompanying paper. What is probably less well recognized are the significant challenges and opportunities Barry has laid out to anode producers to improve anode quality to meet the increasingly stringent requirements of the Potlines customer. These challenges and opportunities will be outlined and their potential impact described. Keywords

Barry Welch Improvement





Carbon anodes Problems




Aluminum smelting




Introduction The contribution made by Barry Welch to improving the understanding of the science and practise of smelting alumina to aluminium is well known and covered by other speakers at this Honorary Symposium. Barry has also made very significant contributions to advancing the understanding of the key issues related to improving anode performance with some significant practical achievements along the way. The advances achieved through the students Barry

has supervised are covered in a companion paper also presented at this Honorary Symposium [1]. In the present paper, some of Barry’s work over the past 10–15 years on helping improve the understanding of the principles underpinning good anode quality and performance are presented. This includes better defining “what good should look like” when it comes to anodes and some ways to get there—either from Barry or suggested by the Authors. The length of the list of topics (See below) covered in this more recent work may be a surprise to those who only associate Barry with reduction cell operations. In a number of cases, Barry has presented the issues and opportunities he has identified as challenges to anode technologists and manufacturers to basically “lift their game” in order to meet the ever-increasing demands of the potline customer. Indeed one of the challenges he has presented is for anode manufacturers to genuinely see potlines as a customer, and not just the people that take away the rodded anodes (of a quality that producers can “get away with”) and replace them with consumed anode butts (And complain on the odd occasions when quality does not meet the current specifications). As an observation, many of the challenges presented by Barry (which are based on an understanding of the underpinning science and impact on the customer) have not been addressed by the industry, and indeed a significant number appear to have not been given much attention at all, i.e. they have been put in the “too hard basket” or just ignored. In the following, a number of the challenges identified by Barry will be listed and briefly outlined, and in some cases, potential pathways proposed to capture the opportunity or resolve the issue.

B. Sadler (&) Net Carbon Consulting Pty Ltd, Melbourne, Australia e-mail: [email protected] A. Tomsett Rio Tinto Pacific Operations, Brisbane, Australia e-mail: [email protected] © The Minerals, Metals & Materials Society 2023 S. Broek (ed.), Light Metals 2023, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-031-22532-1_2

8

Meeting the Requirements of Potline Customers …

Challenges to Improve Anode Properties and Performance The following list of challenges presented by Barry to anode carbon technologists and producers has been organised into four broad groupings: Mindset, Quality/performance, Design, and Operations.

9

binder pitch and filler coke/butts and to reduce anode carboxy reactivity [2, 5]. There is a limit to how high baking temperatures can be which is set by the onset of significant desulfurisation. Using low-sulfur raw materials is necessary to reduce this limitation. (This is discussed further in Sect. 8.) • High density, but without microcracking that will increase resistivity and affect other properties [2]. This will give less open porosity which will reduce anode consumption rates [5]. • No “Free dust” on anodes sent to Potlines [2], e.g. eliminate Packing Material Accretions (PMA), damaged anode surfaces (mechanical or slumping damage during baking), particle segregation from anode forming, damaged vertical corners, and packing coke in slots (e.g. Fig. 1). All of these anode defects contribute to dusting in cells but can be fixed, i.e. there is no technical constraint on meeting this challenge from Barry. It requires the determination to not accept these defects as unavoidable and to do the work needed to eliminate them.

a. The necessary mindset of anode manufacturers: i. Really recognise that “Potlines is the customer” [2] and stop “blame shifting”. (This is discussed further in Sect. 3) ii. Anodes need to be seen as “value add to aluminum production—an essential component that should be designed for performance, not operating convenience” [2]. iii. Anode producers and potlines should “have the data that quantifies and proves the most common anode problems” [3]. (Anode data systems are discussed further in Sect. 4) iv. (Continually) Ask “How can anodes help the cells perform better?” [4]. ii. Reduce the electrical resistance of anode connections b. The required anode quality and performance: by, for example, not having any distortion or attack of i. Barry has consistently and persistently pushed (or in rod/yoke surfaces that make contact in the electrical his words, “hammered”) the following aspects of circuit [2]. anode quality with a strong theme of producing anodes with a low dusting tendency: • (Very) Low Sodium in anodes achieved by excellence in butt cleaning [2, 5]. The importance of butt cleanliness is well known, however, it is generally not well monitored. It is now possible to get commercially available devices to do this online, but these are not widely installed. The conventional approach of manual visual observations of butts after fine cleaning (shot-blasting), and maybe daily analysis of crushed butt samples, is insufficient to monitor cleaning effectiveness in a way that reflects the importance of excellence in cleaning butts. Barry’s challenge is for a maximum anode sodium level of 200 ppm. In the experience of the Authors, there may only be a handful of smelters that consistently achieve this level, despite it generally being possible to meet this target as long as anodes/ butts have not been impregnated with sodium in the potlines. Failure to meet Barry’s target can be due to operational (e.g. Operator care and attention) or plant equipment limitations, but it is likely that new approaches to butt cleaning will be required to consistently meet the challenge. Innovation is required. Fig. 1 Examples of free dust on anodes. The cause of all of these • (Very) High and consistent baking temperatures to defects is known, as are the solutions. Unfortunately, it is still common reduce the differential in reactivity between coked to see anodes set in cells with these problems. From [2]

10

B. Sadler and A. Tomsett

iii. Barry recommended a specification for anode quality [5] that is intended for all anodes, not just for samples from selected locations in a baking furnace. A target outcome from the specifications is to have less anode dust and help reduce cell energy consumption through excellent anode quality as “Producing anodes with raw materials, formulations, and baking conditions that minimize carbon dust will also help cell efficiency gains by eliminating dust generated spikes” [2]: Baked apparent density

>1.6 g/cm3 (with commensurate low open porosity)

Air permeability

93%

Carboxy reactivity dust

80%

Air reactivity dust

10 shifts and number of cells with high and low metal

E. Andrews et al.

Fig. 7 Number of non-functioning alumina feeders and unscheduled anode changes handed over between shifts

Fig. 6 Alumina feed interval and standard deviation Fig. 8 Number of alumina feeders repaired and unscheduled anode changes actioned per shift

An agreed set of critical KPI’s were established to track the recovery process. Figures 5, 6, 7, 8, and 9 show the basic KPI’s used which included • • • •

Number of cells >0.022 nano-ohm Noise for >10 shifts. Number of cells with high and low metal. Alumina feed interval and standard deviation. Number of non-functioning alumina feeders and unscheduled anode changes handed over between shifts. • Number of alumina feeders repaired and unscheduled anode changes actioned per shift. • Number of suspected spiked anodes per cell based on sample current draw measurements.

Recovery Strategy With the situation changing quickly, it was important to understand the root cause(s) and key drivers to stabilise and then recover Line 3. Figure 10 shows the strategy that was developed from a fundamental analysis of the data and situation with key stakeholders. High levels of cell exceptions, including unscheduled anode changes, unstable new cells, and failed alumina feeders, were hypothesised to be caused by an insufficient understanding of how to operate the G4 cell. Furthermore, insufficient exception cell priority and standards were

Process Recovery to Unlock Power Efficiency Improvement at BSL

35

critical in providing a system to action exceptions and break this cycle. The war room had three key objectives:

Fig. 9 Number of suspected spiked anodes per cell based on sample current density measurements

hypothesised to be contributing to the high levels of exception. It was important to break both cycles to reduce the overall generation of exceptions and a summary of the approaches taken are given in the sections below.

Exception Cell Priority and Standards Prior to the implementation of the war room, a large number of exceptions were not being actioned and this helped perpetuate the cycle of exception generation. The war room was

Fig. 10 Line 3 recovery strategy

• Transparency—have a system to track exception work. This had to be highly visual. At the time, exception work that was not completed (i.e. handed over) was not tracked effectively. • Prioritization—have criteria to organize exceptions into levels of priority. There was an imbalance in the importance placed on scheduled versus unscheduled work, to the point where unscheduled work had decreased to practically nothing on some shifts. Figure 11 shows the method that was used to improve prioritization of exceptions under excursion recovery conditions. • Resource allocation—have a method to allocate resources to exception work. The amount of exceptions being generated far exceeded the resources available for action. With the war room providing a means to manage exceptions, a number of other work standards and process control initiatives were implemented to reduce the rate of exception generation. These were generally “coarse” control type initiatives, and this largely meant moving back to the last known good or reverting to fundamentals or best practice. Some examples are listed below that are representative of this approach:

36

E. Andrews et al. Issue: Unstable Cell CD findings More than 990 deg C Feeder Issues and Multiple AEs Spikes Transition joint problem Burn off

Liquid Level Issue Exposed Stubs Multiple Cathode Noise alarms Required Raises - stable cell

Deformation Air burn

Category:

When to Action: Before scheduled work where possible

Cat. 1

During scheduled set

Cat. 2

After scheduled work is complete

Cat. 3

Fig. 11 Method used for prioritization of exceptions

• Amperage—reduced amperage by 3 kA to help stabilize Line 3. • Turnaround activity—cell replacement activity was temporarily paused in order to increase the ability to action exceptions. • Bath Control—historically reduction Line 3 had used three levers to control bath levels with great effect: – Immediate control by removing or adding liquid bath; – Mid-term “fine” control by adding solid bath from feeders in cell; and – Long-term “coarse” control by adjusting alumina:bath blend in the anode cover. This had drifted with too much emphasis on immediate control and long-term control, with the mid-term control through the bath feeders basically turned off. The reliance on long-term control to keep the line in bath surplus also had some unintended consequences with the percentage of alumina in the cover dropping *10% over a number of years (approx. 55!45%). Observations of “sintering” and “bridges” forming in the crust were made, meaning air was able to get underneath the cover and attack the anodes. Changes were made to increase the percentage of alumina in the cover back to the desired target and increase the use of solid bath additions to get an overall better balance between the 3 levers: • Carbon Dust Control—high levels of carbon dust (70% of all cells as shown in Fig. 12) was resulting in anode-

cathode distance (ACD) squeezing and generation of spikes. At the time, the number of resources allocated to hand skimming was not consistent, and the effectiveness was unclear as the amount of carbon being removed was not tracked. During the set operation if carbon was identified during the cavity cleaning process (“pac-manning”), then the expectation was that the operators identify the skip where the material was placed and it be segregated. These segregated skips and hand-skimming carbon were not returned to the bath process but removed through waste streams. No specific strategy was in place to remove carbon through extra pac-manning. Changes were implemented to ensure consistent skimmer resources, tracking of carbon being removed, targeted approach for hand skimmers, and pac-man skimming of corner stalls. • Current Density Measurements—a strategy was implemented to complete current density measurements (voltage drop on anode rod) on all anodes for a rolling 10-cell group using the project resources. This helped the operations team in finding exception work, but more importantly was used as a barometer for process health. The number of high reading anodes gave a true indication of the spike rate when extrapolated, rather than measuring the whole line. Figure 13 shows that the most prolific type of anode failure were carbon and bath spikes (DEF), and addressing this issue would form a large part of the next phase of the recovery work (100-day plan). Routine current density measurements were also reviewed and

Process Recovery to Unlock Power Efficiency Improvement at BSL

37

Fig. 12 Percentage of G3 and G4 cells with carbon dust by Line 3 Bay

Fig. 13 Pareto chart of why anodes have been replaced

reverted to 24 h post setting operation versus 12 to allow more effective actioning of setting accuracy issues. • Setting Increment—the data from the current draw measurement campaigns also had the additional benefit of providing data to better understand anode problems. Observations had been made from cell controllers that a large number of anode raises were taking place for anodes soon to be replaced. The current draw data indicated that new anodes were not picking up current draw as quickly as desired, meaning that the adjacent anodes, soon to be set, were drawing more current (resulting in raises). The

setting increment for new anodes had been increased in a previous project to combat anodes shattering (21– 27 mm). Consequently, a reduction in the setting increment to 24 mm was made in order to get the desired current draw for new anodes (*80% at 24 h). For a line 3 cell with 40 anodes, each anode carries approximately 2.5% of the current load. Figure 14 shows that after 1 day only 20% of anodes were pulling >80% of full load or 0.02 (2%). • Cover Window—reduction Line 3 had recently changed from 70:30 delayed cover application to 100% cover

38

E. Andrews et al.

80.00%

>0.03 0.025-0.03

60.00%

0.02-0.025 0.015-0.02

40.00%

0.01-0.015 0.3 kWh/kgAl compared to the G3 design at 370 kA. To achieve this, substantial changes were made to the cathode linings, cathode and collector bar designs. Target operating points were also changed in order to capture the full benefit. Prior to the 2019 excursion, there were a mixed population of G3 and G4 cells in operation. A review was conducted to examine how the cells were operated and alternative operating strategies were assessed and trailed. It was found that the G3 and G4 cells were effectively being run the same way and therefore the advantages of the low energy cell design were not being realised and in fact the heat balance was detrimental to cell performance and cell life. Figure 16 shows the metal pad noise in G4 cells was considerably worse than in G3 cells prior to the commencement of the recovery work in February 2019. Anode cover height and metal level were the two key parameters requiring correction on the G4 cells. Lowering the metal level and increasing the cover heat allowed the G4 cells to operate at much reduced total heat input with appropriate heat balance. Monthly cathode voltage drops (CVDs) were

Process Recovery to Unlock Power Efficiency Improvement at BSL

39

Fig. 15 Metal pad noise by cell age and generation type for G3 and G4 cells

Fig. 16 Average metal pad noise by cell design generation

monitored to ensure the CVDs had not deteriorated and that sufficient anode-cathode distance was maintained. Figure 17 illustrates the overall recovery strategy with the use of the operating window. The concept of the operating window and its application have been discussed in previous papers [2, 3]. As shown in Fig. 17, amperage was reduced from January 2019 to help with the stabilisation of the process. The adjustment of anode cover and metal heights effectively lowered the thermal limit, which enabled substantial reduction in cell voltage. As soon as process stability was regained and appropriate heat balance established, amperage was increased to the target level with minimum disruption to internal heat. An 80 mV reduction in G4 cells was achieved

in June 2019 compared to January 2019 at the same amperage with correction of cell heat balance.

Outcome The process recovery achieved over the defined eight-week period was substantial with Line 3 stabilised and a clear pathway forward to all metrics back in control. A key highlight was the reduction in metal pad noise as shown in Fig. 18, all whilst the power applied to cells was reduced as indicated in Fig. 4. Furthermore, Fig. 19 shows the length of time critically noisy cells were remaining noisy was also reduced from typically 5 shifts to under 1 shift. The

40

E. Andrews et al.

Fig. 17 G4 Cell operating window

Fig. 18 Average metal pad noise by Bay for Line 3 nano-ohms

process improvements also had a noticeable impact on workload, a boost in the morale of the workforce, and work standards. Figure 20 shows the 100-day recovery plan that was developed to support the operations team. The critical few initiatives were identified with key metrics for tracking.

There were numerous longer term actions that were effectively parked in stage 1 of follow-up work. Looking back, the recovery work enabled an average cell voltage reduction of 64 mV in 2019 compared to 2018, representing a power efficiency improvement of 0.24 DCkWhr/kg Al. Further stabilisation and fine-tuning of Line

Process Recovery to Unlock Power Efficiency Improvement at BSL

41

Fig. 19 Number of shifts for Metal Pad Noise >0.022 nano-ohms

Line 3 Recovery plan Critical initiative

4-Mar

Cover Strategy

11-Mar

18-Mar

25-Mar

SE1 Roll-out Key metrics: BACI summary for CE/PE/C

1-Apr

8-Apr

15-Apr

22-Apr

29-Apr

Timing to be defined for conversion of remaining 4 sections to new cover standard

War Room Continue focus on exception cell priority Exception Priority key metrics: USC & Feeders actioned per shift, USC & Feeders hander over per shift, number of CD's out of range per shift

Carbon Skimming

Noisy Cells

New G4 Cells

6-May

R14 Skimmers - Targeted skimming 3 months key metrics: Cat 1&2 C dust, amount of C Dust skimmed per day Individual targeting of noisy cells Embed and refine Noise Kicker key metrics: number of cells > 22 microhms noise Robust Strategy

Longer term actions Setting resistance Setting Increment Alarms/deadbands 2 Stage Cover Alumina concentration in Cover Bath control - Feeders G4 Parameters Noise Kicker v2 New Cell Control 100 to 60 to 45 Days CaF2 - Long term trial CCS /Fundamentals Training CVD Strategy AlF3 Mass Balance refine Chemical Analysis (AlF3)

Track progress key metrics: MPN, Bath temperature (must be above 970deg C first 100 days whilst stabilising potlin) Amperage

Go/No go key metrics: Decision to increase based on above KPI's

Line 3 team Day

Fig. 20 Forward 100-day recovery plan for operations

3 into 2020 permitted a full-year performance of 94.2% and 12.96 DCkWhr/kg Al current efficiency and power efficiency respectively.

Key Learnings The Line 3 recovery in 2019 demonstrates the importance of getting the fundamentals right in terms of understanding and operating the process. Furthermore, when faced with a

multi-faceted problem, it is important to get back to basics, identify the root causes, and then support operations with the execution of the plan. It also demonstrates the resource intensiveness and extended time required to recover a Reduction Line that loses stability and control, thereby highlighting the value in focusing first and foremost on stability and reducing variation. The recovery team consisted of highly experienced and motivated team members who worked collaboratively with operations and the Brisbane Smelter Technical Support team

42

to identify quick wins and implement recovery initiatives quickly. At all times, careful diligence was given to change management systems, risk assessments, and syndication of changes. The daily support provided by the site General Manager and senior leaders was pivotal in aligning the recovery work and demonstrating both the urgency and criticality of the situation. Acknowledgements Special thanks to Mansel Ismay, Andrew Karbowiak, Mark P. Taylor, the BSL Operations and Technical teams, and the Rio Tinto Smelter Technical Support team in Brisbane, Australia.

E. Andrews et al.

References 1. Corby C, Zhang H, Lewis M and Roberts, J (2019) Modernisation of Sumitomo S170 Cells at Boyne Smelters Limited, Proceedings of Light Metals 2019. pp. 543-552. 2. Andrews EW, Tomsett A, Hamilton S (2014) Improving Power Efficiency in Power Constrained Aluminium Smelters, 11th Australasian Aluminium Smelting Technology Conference (AASTC), Dubai, 6–11 Dec. 3. Heithcoate K, Tinnoch S, Moratti S, Andrews EW, Illingworth, M (2018) NZAS Step Change Power Efficiency Improvement, 12th Australasian Aluminium Smelting Technology Conference (AASTC), Queenstown, New Zealand, 2–7 Dec.

A Smart Individual Anode Current Measurement System and Its Applications Choon-Jie Wong, Jing Shi, Jie Bao, Barry J. Welch, Maria Skyllas-Kazacos, Ali Jassim, Mohamed Mahmoud, and Konstantin Nikandrov

Abstract

Introduction

This paper discusses a new individual anode current measurement scheme and its applications in real-time monitoring and control of the Hall-Héroult process. While anode current can be directly measured from the anode rod, this approach takes measurements from the anode beam allowing the sensors to remain intact through various cell operations, including anode change. This instrumentation scheme employs smart sensors that are daisy-chained on a common bus for digital data transfer. This approach limits electromagnetic interferences and offers system self-configuration and self-diagnosis, thus allowing for easy maintenance. The system can be configured to work across a broad range of cell technologies. Monitoring anode current distributions helps improve process operation and allows early detection of process faults such as perfluorocarbon co-evolution and blocked feeders. This also offers the ability to monitor process states such as local alumina concentration and bath temperature, along with potential improvements to cell operation and current efficiency. Keywords




Individual anode current measurement monitoring Current distribution




Online

C.-J. Wong
J. Bao (&)
B. J. Welch
M. Skyllas-Kazacos School of Chemical Engineering, The University of New South Wales, Sydney, NSW 2052, Australia e-mail: [email protected]

Measuring individual anode current signals in a Hall-Héroult cell is an attractive option to overcome the lack of observability of local cell variables. Conventionally, only the line current and cell voltage are being continuously measured for monitoring and control purposes [1, 2], but these indicate only the overall cell conditions, such as the average alumina concentration level. The line current and cell voltage cannot detect local cell variations or faults, which are becoming more critical especially as modern cells are designed to operate at higher currents. With anode size growth outpacing the increase in molten electrolyte volume, the assumptions of homogeneity in cell conditions are becoming questionable. Additionally, the anode-cathode distance (ACD) is typically squeezed in modern high-current cells to limit energy consumption, which further exacerbates the uneven cell conditions, due to limited bath mixing arising from the restricted flow. The distribution of anode currents both affects and is affected by the local condition of each anode-cathode path (e.g., bath composition and temperature). This is because, on the one hand, the regulated line current must distribute among all the anode-cathode paths according to the local reversible potential, cell overpotentials, and ohmic potential drop of each path; these are functions of the local path conditions. On the other hand, anode currents drive the local electrolytic reactions and local ohmic heating. Therefore, the measurements of individual anode currents reveal not only the spatial cell information, but also how it will change. This paper first presents a review of measurement schemes found in the literature, followed by a discussion of the new developed smart individual anode current measuring system, along with its applications in monitoring and control of smelter cells.

J. Shi
A. Jassim
M. Mahmoud
K. Nikandrov Emirates Global Aluminium, Jebel Ali Operations, P.O. Box 3627 Dubai, UAE © The Minerals, Metals & Materials Society 2023 S. Broek (ed.), Light Metals 2023, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-031-22532-1_6

43

44

C.-J. Wong et al.

Measurement Principles There are generally two methods of measuring anode currents as described in the literature. One method, which the development described in this paper employs, is based on the ‘isometric voltage drop’ principle. In this method, current I flowing through a conductor is determined by measuring the voltage drop V across a known, short distance L on said conductor. For a conductor with a resistivity that changes linearly with temperature T, this is given by V¼

ðq1 T þ q2 ÞL I; A

ð1Þ

where A is the surface area through which current flows, and q1 and q2 are the coefficients specific to the conductor material for calculating resistivity. Table 1 summarises the developments of an anode current measurement system employing the isometric voltage drop principle. In its simplest form, the voltage drop on each anode rod is directly measured. The raw data can be immediately converted into anode currents by a local microprocessor and/or sent to a remote computer for later processing and analysis. However, due to the nature of the Hall-Héroult process where the carbon anodes are continuously consumed, one or two anode assemblies must be replaced every few days. This necessarily means that the measuring probes must be detached and re-attached to the

new anode rods, which subjects the wirings and the rods to damages due to excessive handling in addition to inconveniencing the process operators. Keniry and Shaidulin [4] suggest taking the measurements from the anode beam. However, in this approach, the potentials (cf. potential difference) on the beam were measured, which Yao [15] argues that this can lead to unreliable measurement of all the anode currents even if only one potential measurement is faulty. Nonetheless, this scheme inspired Li et al. [19], Yang et al. [20], and Yao [15] to measure the voltage drops on the anode beam, from which beam currents can be determined. The anode current is then the difference in the two neighbouring beam currents. Another method of measuring anode currents is based on the ‘Hall effect’ principle. In this method, the anode current I is determined by measuring the magnetic field strength B around the anode rods induced by the current, with [24] B¼

l0 I; 2pr

ð2Þ

where l0 is the magnetic field constant and r is the distance between the centre of the conductor to the Hall sensor. However, this method is susceptible to magnetic field interference from currents flowing in neighbouring anodes, risers, and crossovers, as well as those in neighbouring cells. Hence, to minimise this interference, researchers have proposed the use of multiple Hall sensors for each anode [25, 26] (e.g., the

Table 1 Review of instrumentation systems based on isometric voltage drop Author

Smelter

Year

Methodology summary

Holmes et al. [3], Keniry and Shaidulin [4]

Alcoa, United States

1969

P-225 cells were designed for a high degree of automation. The anode currents were monitored by six Mod Comp II computers and controlled by independently adjusting the elevation of the 16 anode pairs

Potocnik and Reverdy [5]

Reynolds Metals, Germany

1973

Connectors were plugged into the anode rods to measure voltage drops, which were sent to a local microcomputer at each cell. The processed results were sent to the main computer system. The system was abandoned after 6–9 months due to excessive damages to anode rods and wires

Langon and Varin [6]

Aluminium Pechiney, France

1975

Four prototype 280 kA cells were constructed, of which two can adjust each of 20 anode pairs automatically (presumably with the use of anode current readings), in addition to conventional main beam adjustments. Individual anode adjustment was later abandoned due to high superstructure cost, and the other two conventional cells performed just as well by improving operation and control

Huni [7]

Alcan, Canada

1987

A-275 prototype cells were designed with individual anode drives. Voltage drops on the flexible current leads which connected the anode rods to the busbar were measured and processed by a computer at each cell

Barnett [8]

Reynolds Metals

1988

Voltage drop and temperature on each anode rod were encoded to ASCII characters for transmission on a RS-485 multidrop network. The data was then converted for a RS-232 multidrop network for radio-transmission to a remote computer

Keniry et al. [9]

Comalco, Australia

1992

Voltage drops on anode rods were filtered, conditioned, and amplified. A local microprocessor then digitalised and transmitted the voltage signals to a computer via the fibre-optic link, which then calculated the current. The system could sample every two minutes for 30 days, or at 50 Hz for 20-s bursts (continued)

A Smart Individual Anode Current Measurement System …

45

Table 1 (continued) Author

Smelter

Year

Methodology summary

Rye et al. [10]

Elkem Aluminium, Norway

1998

The anode rods were individually connected to the central beam with a flexible. The voltage drops on these flexibles were measured and converted to anode currents. During normal operation, the anode current distribution was represented as a bar graph with a LED matrix, but not recorded

Panaitescu et al. [11, 12]

ALRO SA, Romania

2000

For visualising metal pad waves, the voltage drops between the bimetal plate and stubs on each anode were measured with the acquisition system DAQ642. On a cell with 16 anodes, 64 stub voltage readings were recorded on a hard disk as well as visualised in real time

Shaidulin et al. [13, 14]

RUSAL, Russia

2005

Each anode rod was fitted with an electric load measuring transducer, and the signals were sent and collected by a control and data acquisition unit. The anode current can then be calculated

Keniry and Shaidulin [4], Yao [15]

RUSAL, Russia

2008

Due to frequent damages and inconvenience arising from the frequent disconnection of the sensor from the anode rods, this design measured the potentials on the anode busbar, relative to a common ground. The anode currents could be determined simultaneously accounting for the cell superstructure design

Cheung et al. [16], Cheung [17]

DUBAL, United Arab Emirates

2013

Voltage drops across spring-loaded contact pins on anode rods were amplified and transmitted to a NI CompactRIO 9024 acquisition system at 10 Hz, while the temperature was also measured and transmitted once every minute. These data were then transferred to a remote computer via an Ethernet connection

Qiu et al. [18]

Yunnan Aluminium, China

2013

Voltage drops on anode rods were measured and transmitted to a converter module which transfers the data to a remote system for data processing and analysis

Li et al. [19], Yang et al. [20]

Jinlian Aluminium, China

2015

Voltage drops and temperatures on the beam were collected at 1 Hz by wires riveted to the beam. A collection unit with a browser then transmitted and displayed the anode currents

Bao et al. [21]

DUBAL, United Arab Emirates

2017

Voltage drops on the anode busbar and their corresponding temperatures were sent to a NI CompactRIO 9082 acquisition system at up to 200 Hz. A new file was created for these raw data every 3 min, which were then uploaded to a cloud storage (Dropbox) for further analysis

Huang et al. [22], Yin et al. [23]

Zunyi Aluminium, China

2021

Voltage drops and temperatures on anode rods were digitalised with AD7705 and transmitted via RS-485 bus to a main controller powered by STM32 for display and control. The data can be transmitted to the factory production process control system

system described by Evans and Urata [27, 28] has five Hall sensors per anode), which drives up the capital costs of the system. It also requires solving a complicated model describing the cell design for good measurement accuracy [25], which can be computationally costly—experiences at TRIMET [29] and Alouette [30] show that the data has to be archived and processed with a cloud computing service (AWS). Hence, the system described in this work focuses on the isometric voltage drop principle.

New Measurement Scheme with Smart Sensors Based on years of experience and learning from successful deployments of anode current monitoring systems in our previous projects [15, 16], this work further develops an in-house, improved individual anode current measurement system with superior signal quality and system robustness.

A schematic of this measurement system is shown in Fig. 1. Measurements of the voltage drop and temperature are taken from the anode beam, and so the installation will remain intact through cell operations such as anode change and beam raising. The sensors are daisy-chained together in a bus topology network, while a communication unit collects the data and transmits it to a computer server. This system samples data at 2 Hz under normal operation, and a higher sampling frequency of up to 100 Hz can be enabled on-demand for special studies or for diagnostic purposes. The measurement probes can be installed on either the top or the side of the anode beam, as the scheme is flexible.

Exceptional Signal-to-Noise Ratio Designed to scale for future large cells with more anodes, distributed processing forms part of the core design principle. Each sensor installed at each signal pickup location has

46

C.-J. Wong et al.

Comms. Unit

Power and CANbus

Network

Sensor

Sensor

Sensor

Beam

Server Anode

Anode

Fig. 1 Schematic of measurement scheme with smart sensors

its own microprocessor and can function autonomously under the instructions of the communication unit. All the sensors can individually and automatically calibrate their readings to reject voltage offset and gain offset errors, as well as independently adjust their instrument amplifier gains. This achieves a high signal-to-noise ratio since the signal magnitude varies greatly depending on the locations on the beam. The sensors and the communication unit transfer signals digitally on the bus line. The analogue-to-digital converter of each sensor uses ‘oversampling and decimation’ method [31, 32] to increase the effective resolution—the sensors achieve a better representation of the input signal by measuring at above 250 Hz before subsequent filtering and downsampling to the required sampling frequency.

System Robustness The digital signals are transmitted through shielded twisted pair cables to minimise the impacts of electromagnetic interference. The multi-node bus topology also reduces the cabling required, in contrast to other designs with a star topology, where analogue signals from all the measurement points are transmitted to a single cell-level microprocessor with long cables which inadvertently work like antennas. Using a single multiplex bus for all nodes reduces the cross-talks between cables, while also allows easier maintenance of the system as sensors can be added or removed easily from the daisy-chain. This bus uses the controller area network (CAN) interface protocol, which has seen increased adoption in industrial applications partly due to its robustness in harsh electrical environments [33]. As CANbus protocol transmits data with voltage differential between two bus lines (see Fig. 2), the bus is not susceptible to inductive spikes or other noises due to interference. Additionally, CANbus has built-in error detection capabilities, with each node tracking transmission

errors to prevent faulty devices from interfering with the bus traffic [34]. CANbus also has a major advantage over other protocols, such as the RS-485, in how it handles messages. In a RS-485 system, if multiple nodes are transmitting data simultaneously, the signal performance will suffer degradation possibly leading to hardware damages [33]. In contrast, the CANbus protocol can resolve this issue by queuing messages according to the priority ranking of each message. To ensure timely data sampling and delivery, especially during the on-demand high-frequency sampling, all the sensors and the communication unit are time-synchronised. Timer-based interrupts are used to precisely control the sampling interval, while timestamping of all data packets accounts for transmission delays.

Ease of Maintenance The measurement system has self-configuration and self-diagnosis capabilities. Each sensor in the CANbus has a unique addressable identifier, which is used by the communication unit to both identify the source of data packets and target an instruction. As such, the system can automatically detect and report changes in identifiers observed (for example, a missing identifier signifies sensor failure). In some cases, the system can be automatically configured (for example, observing a missing identifier followed shortly by a new identifier signifies that a sensor has been manually replaced). With the use of identifiers, each sensor can send and receive diagnostic information and instructions. To indicate system health, each sensor also measures and sends its board temperature to the communication unit, in addition to the beam temperature. Each sensor also has an integrated watchdog timer [35] for monitoring microprocessor responsiveness and for ensuring that the CANbus connection is alive. The firmware of the communication unit and the sensors can also be remotely managed by an IT department, owing to the bootloader-program architecture. The bootloader checks the integrity of the firmware at boot time, and if it fails (such as when a firmware update process was interrupted), an older known-good version of the firmware can be restored. The communication unit also has a physical reset switch for on-site interactions—a quick press reboots the unit and a long press factory-resets the system.

Field Test of Proposed Measurement System The system was deployed in Emirates Global Aluminium, Jebel Ali Operations to verify the reliability and ease of use of the smart individual anode current measurement system

A Smart Individual Anode Current Measurement System …

47 16

CAN High

14

CAN Low

12

Time

Anode Current (kA)

Differential V

Volts

Interference

10 8 6 4

1

0

1

0

1 Time

Anode Change Events 2 0 Dec 06, 2021

Dec 20, 2021

Jan 03, 2022

Fig. 2 CANbus data transmission with voltage differential Fig. 4 Anode current measured continuously through cell operations

anode service life, the fluctuations caused by cell operations including the changing of other anodes, material feeding patterns, and anode beam adjustments are also visible.

Applications of Anode Currents Measurements As previously explained, the anode current individually reveals information on local cell conditions. Additionally, since all the anode currents are coupled due to the parallel electrical connection, the distribution of anode currents also provides information on the relative path resistance of all anode-cathode paths, which allows monitoring and control of spatial variations in cell conditions. Fig. 3 a Sensors installed on the anode beam (enclosed in the box for protection), b communication unit mounted at cell duct-end, c a prototype human-machine interface beside the cell

Individual Anode Drives

developed in this paper under real conditions. For the 400+ kA cell design, each cell has 44 sensors or nodes in its CANbus, in addition to a communication unit that can simultaneously connect and transmit data to both a production server and a development server, via standard TCP/IP on a dedicated virtual local area network (vLAN) set up by the EGA IT department. The installation (see Fig. 3) was finished in late 2021 and the system was continuously tested for more than 8 months. Data can be collected continuously as the sensors remain intact during the numerous cell operations that were conducted in this period, including for anode change operations. For example, Fig. 4 shows the current of a particular anode which is measured by the system as it undergoes two anode change operations in the span of a month. Throughout the

As summarised in Table 1, individual anode currents were utilised to independently adjust the elevation of anodes since at least 1969. It has been experimentally shown that the low setting of an anode causes current overloading, which also results in uneven overheating of anode stubs, leading to uneven thermal expansion and possibly carbon cracking [36]. As the anode currents determine the rate of local material consumption and the rate of local heat generation, this current imbalance also exacerbates the spatial non-homogeneity in the cell. Overloading or underloading of anode currents can be determined by comparing the measured anode current with the target anode current (total line current divided by the number of anodes). The individual anode drives on the anodes can then be independently adjusted to bring the anode current closer to target values [3, 6, 7]. For automatic

48

controls, the elevation adjustments must consider the nature of anode current re-distribution and the state the cell is operating in, such that the anode adjustments do not lead to instability.

Limiting Perfluorocarbon Emissions and Preventing Anode Effects Perfluorocarbons (PFCs), particularly CF4 and C2F6, are greenhouse gases that start to co-evolve when the electrical potential across the anode exceeds the reversible potentials for PFC reactions of 1.83–2.75 V [37]. PFCs are generated in large quantities during an undesirable phenomenon associated with the depletion of alumina, called the ‘anode effect.’ In this event, the anodic gas production reaction changes, leading to dewetting and partial passivation of anodes, as the anode currents fall [37, 38]. As anode current imbalance grows, the PFC co-evolution propagates to all other anodes. For this reason, anode current measurements can be used to characterise and allow early detection/mitigation of anode effects [4, 10, 38–41]. The studies also extend to detecting faults in feeders or crust-breakers [30, 42], since they also contribute to alumina depletion. In recent decades, it was also established that PFCs are emitted even under typical operating conditions; these are termed ‘low voltage PFCs’ (LV-PFCs) since they are emitted in the absence of anode effects and the associated characteristic high cell voltage. Studies found that LV-PFCs contribute a significant portion towards total PFC emissions [43–45], with estimates ranging as high as 93% of total smelter emissions [46]. It may be possible to identify LV-PFCs formation with anode current signals.

Identifying Anode Faults Both the time and frequency domain analyses of anode current signals can be used to identify local cell conditions, including fault identification in the region of the carbon anode. Experiments [10, 42] show that the anode current can indicate the formation of an anode spike, which lowers the cell current efficiency. The frequency domain can be obtained by performing a windowed Fourier transform of the anode current signals. Low-frequency noises (0.02–0.05 Hz) are typically caused by thermal imbalance and magnetohydrodynamics instability, while high-frequency ones are typically related to the anode-metal shorting arising from anode faults such as slippage, cracking, and spikes [4]. The study of the bubble

C.-J. Wong et al.

dynamics (0.5–2 Hz) also yields information on the presence of slots as anode ages, the onset of anode effects, and if an anode slippage occurred [40].

Understanding, Modelling, and Improving Anode Setting Carbon anodes are regularly replaced in commercial cells as their service lives end. This operation, called the ‘anode setting,’ introduces significant disturbances to the mass and thermal balance, as the bath freezes around the cold new anodes and impedes the flow of anode currents. The gradual warming up of the new anodes and the dissolution of bath freeze contribute to the local energy deficits introduced by this operation. Anode current uptake trajectories of young anodes were used to study the freeze dissolution process and anode conditions at various stages while the anodes are heating up [17, 40]. The measurement of anode currents also allows coupled mass and thermal balance dynamic models to be proposed based on the first principles [47, 48] for monitoring/control studies of a cell undergoing anode setting operation. The use of anode current signals in preheating studies [49, 50] has been proposed to minimise the mass and thermal perturbations introduced by the operation.

Spatial Distribution Monitoring and Control While the anode currents individually reveal the local cell conditions, the distribution of anode currents as a whole indicates the extent of spatial variations in the cell. The monitoring and control of cell homogeneity have become increasingly important as designs of modern cells and retrofitting of existing cells focus on increased productivity; the changes include reduced bath volume per line current amperage, reduced number of alumina feeders per line current amperage, impeded bath mixing due to reduced ACD/cell voltage for energy savings, and bigger anodes taking longer to heat up after replacements [37]. As the anode current distribution also indicates how local electrolytic reactions and local ohmic heating will vary in the future, it can be used in conjunction with model-based state/parameter observers (e.g., Kalman Filters) to estimate local process variables such as alumina concentration, ACD, bath flow velocity, and bath temperature [51–56]. Controllers, either model-based or data-based, have also been developed to improve cell homogeneity by minimising the imbalance of anode currents [57–60].

A Smart Individual Anode Current Measurement System …

Conclusion It is becoming increasingly critical to maintain the homogeneity in cell conditions as modern cell designs and retrofitting of existing cells pursue higher production rates and lower energy consumptions at the cost of cell stability (e.g., squeezing of ACD leading to increased shorting risks). There is wide recognition that an anode current measurement is a promising approach for revealing spatial cell conditions, for the purpose of process monitoring/control and fault detection/mitigation. A successful application requires a stable, reliable instrumentation system for accurate and continuous signal analysis. In this paper, a new individual anode current measurement scheme with smart sensors is proposed. The signals are sampled continuously from the anode beam, uninterrupted by cell operations, and are amplified by each sensor to achieve a high signal-to-noise ratio. The system was designed to be robust and easy to maintain, with features including self-configuration, self-diagnosis, and remote system management. An industrial field test was conducted to verify the system’s performance under environments of elevated temperature, strong magnetic field, and EMI conditions, as well as corrosion and dust. As the system is designed in-house, the sensors can be mass manufactured at a reasonable cost without having to purchase costly off-the-shelf data acquisition systems, such as a CompactRIO. This instrumentation system is designed to work with various cell designs, with the option to retrofit it onto existing cells. Acknowledgements The authors acknowledge the financial and technical support from Emirates Global Aluminium and Jebel Ali Operations, as well as thank Mr. John Lam for the technical expertise.

References 1. Homsi P, Peyneau J-M, Reverdy M (2000) Overview of Process Control in Reduction Cells and Potlines. Paper presented at the Proceedings of the TMS Light Metals, Nashville, Tennessee, USA. 2. Gao YS, Taylor MP, Chen JJJ, Hautus MJ (2011) Operational and control decision making in aluminium smelters. Advanced Materials Research. 201–203(2011):1632–1641. 3. Holmes GT, Fisher DC, Clark JF, Ludwig WD (1980) Development of Large Prebaked Anode Cells by Alcoa. Paper presented at the Proceedings of the TMS Light Metals, Las Vegas, Nevada, USA. 4. Keniry J, Shaidulin E (2008) Anode signal analysis—the next generation in reduction cell control. Paper presented at the Proceedings of the TMS Light Metals, New Orleans, Louisiana, USA. 5. Potocnik V, Reverdy M (2021) History of Computer Control of Aluminum Reduction Cells. Paper presented at the Proceedings of the TMS Light Metals, Orlando, Florida, USA.

49 6. Langon B, Varin P (1986) Aluminium Pechiney 280 kA Pots. Paper presented at the Proceedings of the TMS Light Metals, New Orleans, Louisiana, USA. 7. Huni JPR (1987) A-275—Individual anode control. Paper presented at the Proceedings of the TMS Light Metals, Denver, Colorado, USA. 8. Barnett WM (1988) Measuring Current Distribution in an Alumina Reduction Cell. Patent number 4786379. 9. Keniry JT, Barber GC, Taylor MP, Welch BJ (2001) Digital processing of anode current signals: an opportunity for improved cell diagnosis and control. Paper presented at the Proceedings of the TMS Light Metals, New Orleans, Louisiana, USA. 10. Rye KA, Königsson M, Solberg I (1998) Current Redistribution Among Individual Anode Carbons in A Hall-Heroult Prebake Cell at Low Alumina Concentrations. Paper presented at the Proceedings of the TMS Light Metals, San Antonio, Texas, USA. 11. Panaitescu A, Moraru A, Panaitescu I (2000) Visualisation of the Metal Pad Waves in the Aluminum Reduction Cell with Pre-baked Anodes. Paper presented at the Proceedings of the TMS Light Metals, Nashville, Tennessee, USA. 12. Panaitescu A, Moraru A, Panait N, Dobra G, Munteanu N, Cilianu M (2001) Experimental Studies on Anode Effects by the Visualisation of the Molten Aluminium Surface Oscillations. Paper presented at the Proceedings of the TMS Light Metals, New Orleans, Louisiana, USA. 13. Shaidulin E, Gusev A, Vabischevich P (2005) Method of controlling aluminum reduction cell with roasted anodes. Patent number 2303658C1. 14. Shaidulin E, Gusev A, Vabischevich P (2007) Method of controlling aluminum reduction cell with prebaked anodes. Patent number 11/592557. 15. Yao Y (2017), Process Monitoring, Modelling and Fault Diagnosis in Aluminium Reduction Cells. Ph.D. thesis, School of Chemical Engineering, Faculty of Engineering, the University of New South Wales, Australia. 16. Cheung CY, Menictas C, Bao J, Skyllas-Kazacos M, Welch BJ (2013) Frequency response analysis of anode current signals as a diagnostic aid for detecting approaching anode effects in aluminum smelting cells. Paper presented at the Proceedings of the TMS Light Metals, San Antonio, Texas, USA. 17. Cheung C-Y (2013), Anode Current Signals Analysis, Characterization and Modeling of Aluminum Reduction Cells. Ph.D. thesis, School of Chemical Engineering, Faculty of Engineering, the University of New South Wales, Australia. 18. Qiu Z, Ji F, Li Y, Yu Q, Li L, Li C (2013) Online measurement and data analysis device for anode current distribution of aluminum electrolysis cell. Patent number CN203080085U. 19. Li J, Yang S, Zou Z, Zhang H (2015) Experiments on Measurement of Online Anode Currents at Anode Beam in Aluminum Reduction Cells. Paper presented at the Proceedings of the TMS Light Metals, Orlando, Florida, USA. 20. Yang S, Zou Z, Li J, Zhang H (2016) Online anode current signal in aluminum reduction cells: measurements and prospects. JOM. 68(2):623–634. 21. Bao J, Welch BJ, Akhmetov S, Yao Y, Cheung C-Y, Jassim A, Skyllas-Kazacos M (2017) Method of monitoring individual anode currents in an electrolytic cell suitable for the Hall-Heroult electrolysis process. Patent number WO 2017/141135 A1. 22. Huang R, Li Z, Cao B (2021) Design and Implementation of Online Detection System for Anode Current Distribution in Aluminum Reduction Cell. Paper presented at the China Automation Congress (CAC). 23. Yin Y, Wang J, Cui J, Xiao G, Xu Z, Zhang S, Wang F (2016) Aluminium cell positive pole distribution electric current precision

50

24.

25.

26. 27.

28.

29.

30.

31.

32.

33.

34.

35.

36.

37.

38.

39.

40.

C.-J. Wong et al. measurement appearance with self calibration function. Patent number CN205501431U. Wieser C, Helmbold A, Glzow E (2000) A new technique for two-dimensional current distribution measurements in electrochemical cells. Journal of Applied Electrochemistry. 30(7):803– 807. Urata N, Evans J (2010) The determination of pot current distribution by measuring magnetic fields. Paper presented at the Proceedings of the TMS Light Metals, Seattle, Washington, USA. Hung OK (2000) Anode and Cathode Current Monitoring. Patent number 6136177. Evans JW, Urata N (2011) Technical and operational benefits of individual anode current monitor. Paper presented at the Proceedings of 10th Australasian Aluminium Smelting Technology Conference, Launceston, Tasmania, Australia. Evans JW, Urata N (2012) Wireless and non-contacting measurement of individual anode currents in Hall-Heroult pots; experience and benefits. Paper presented at the Proceedings of the TMS Light Metals, Orlando, Florida, USA. Lützerath A, Evans JW, Victor R (2014) On-line monitoring of anode currents: Experience at TRIMET. Paper presented at the Proceedings of the TMS Light Metals, San Diego, California, USA. Dion L, Lagac C-L, Evans JW, Victor R, Kiss LI (2015) On-line monitoring of individual anode currents to understand and improve the process control at Alouette. Paper presented at the Proceedings of the TMS Light Metals, Orlando, Florida, USA. Baker B (2011) How delta-sigma ADCs work, Part 1, in Data Acquisition, Texas Instruments Incorporated. https://www.ti.com/ lit/an/slyt423a/slyt423a.pdf?ts=1660200202417. Accessed 11 August 2022. Microchip Technology (2019) SAM D/L/C—Oversampling and decimation feature. https://microchipsupport.force.com/s/article/ SAM-D-L-C—Oversampling-and-decimation-feature. Accessed 11 August 2022. Gee R (2020) CAN vs. RS-485: Why CAN is on the move, in Delivering Robust Communications, Maxim Integrated, Editor. https://www.maximintegrated.com/content/dam/files/design/ technical-documents/white-papers/can-wp.pdf. Accessed 11 August 2022. Total Phase (2019) What is CAN Bus Protocol? https://www. totalphase.com/blog/2019/08/5-advantages-of-can-bus-protocol/. Accessed 11 August 2022. ABLIC (2020) What is a watchdog timer (WDT)? https://www. ablic.com/en/semicon/products/automotive/automotive-watchdogtimer/intro/. Accessed 11 August 2022. Peterson RW (1976) Temperature and voltage measurements in Hall cell anodes. Paper presented at the Proceedings of the TMS Light Metals, Las Vegas, Nevada, US. Wong D, Welch B, Nunez P, Dion L, Spirin A (2019) Latest progress in IPCC methodology for estimating the extend of PFC greenhouse gases co-evolved in the aluminium reduction cells and challenges in reducing these emissions. Paper presented at the Proceedings of the International ICSOBA Conference, Krasnoyarsk, Russia. Tarcy G, Tabereaux A (2011) The initiation, propagation and termination of anode effects in Hall-Heroult cells. Paper presented at the Proceedings of the TMS Light Metals, San Diego, California, US. Zarouni A, Reverdy M, Zarouni AA, Venkatasubramaniam KG (2013) A study of low voltage PFC emissions at DUBAL. Paper presented at the Proceedings of the TMS Light Metals, San Antonio, Texas, USA. Cheung C-Y, Menictas C, Bao J, Skyllas-Kazacos M, Welch BJ (2013) Characterization of individual anode current signals in

41.

42.

43.

44.

45.

46.

47.

48.

49.

50.

51.

52.

53.

54.

55.

56.

57.

aluminum reduction cells. Industrial & Engineering Chemistry Research. 52(28):9632–9644. Zarouni A, Al Zarouni A (2011) DUBAL’s experience of low voltage PFC emissions. Paper presented at the Proceedings of 10th Australasian Aluminium Smelting Technology Conference, Launceston, Tasmania, Australia. Puzanov II, Zavadyak AV, Klykov VA, Makeev AV, Plotnikov VN (2016) Continuous monitoring of information on anode current distribution as means of improving the process of controlling and forecasting process disturbances. Journal of Siberian Federal University. Engineering & Technologies. 9 (6):788. Marks J, Bayliss C (2012) GHG measurement and inventory for aluminum production. Paper presented at the Proceedings of the TMS Light Metals, Orlando, Florida, USA. Marks J, Nunez P (2018) Updated factors for calculating PFC emissions from primary aluminum production. Paper presented at the Proceedings of the TMS Light Metals, Phoenix, Arizona, USA. Wong DS, Fraser P, Lavoie P, Kim J (2015) PFC emissions from detected versus nondetected anode effects in the aluminum industry. JOM. 67(2):342–353. Li W, Chen X, Yang J, Hu C, Liu Y, Li D, Guo H (2012) Latest results from PFC investigation in China. Paper presented at the Proceedings of the TMS Light Metals, Orlando, Florida, USA. Wong C-J, Yao Y, Bao J, Skyllas-Kazacos M, Welch BJ, Jassim A (2021) Modeling Anode Current Pickup After Setting. Paper presented at the Proceedings of the TMS Light Metals, Orlando, Florida, USA. Wong C-J, Yao Y, Bao J, Skyllas-Kazacos M, Welch BJ, Jassim A, Mahmoud M, Arkhipov A (2021) Modelling of Coupled Mass and Thermal Balances in Hall-Heroult Cells During Anode Change. Journal of the Electrochemical Society. 168(21):123506. Jassim A, Akhmetov S, Welch B, Skyllas-Kazacos M, Bao J, Yao Y (2016) Studies on anode pre-heating using individual anode signals in Hall-Heroult reduction cells. Paper presented at the Proceedings of the TMS Light Metals, Nashville, Tennessee, USA. Wong C-J, Yao Y, Bao J, Skyllas-Kazacos M, Welch BJ, Jassim A (2020) Study of heat distribution due to ACD variations for anode setting. Paper presented at the Proceedings of TMS Light Metals, San Diego, California, USA. Jakobsen SR, Hestetun K, Hovd M, Solberg I (2001) Estimating alumina concentration distribution in aluminium electrolysis cells. IFAC Proceedings Volumes. 34(18):303–308. Hestetun K, Hovd M (2005) Detecting abnormal feed rate in aluminium electrolysis using extended Kalman filter. IFAC Proceedings Volumes. 38(1):85–90. Yao Y, Cheung C-Y, Bao J, Skyllas-Kazacos M (2015) Monitoring local alumina dissolution in aluminum reduction cells using state estimation. Paper presented at the Proceedings of the TMS Light Metals, Orlando, Florida, USA. Yao Y, Cheung C-Y, Bao J, Welch BJ, Skyllas-Kazacos M, Akhmetov S, Jassim A (2017) Method for estimating dynamic state variables in an electrolytic cell suitable for the Hall-Heroult electrolysis process. Patent number WO 2017/141134 A1. Yao Y, Bao J (2018) State and parameter estimation in Hall-Heroult cells using iterated extended Kalman filter. IFAC-PapersOnLine. 51(21):36–41. Wong C-J, Yao Y, Bao J, Skyllas-Kazacos M, Welch BJ, Jassim A, Mahmoud M (2021) Discretized Thermal Model of Hall-Heroult Cells for Monitoring and Control. IFAC-PapersOnLine. 54 (11):67–72. Shi J, Yao Y, Bao J, Skyllas-Kazacos M, Welch BJ (2020) Multivariable Feeding Control of Aluminum Reduction Process Using Individual Anode Current Measurement. IFACPapersOnLine. 53(2):11907–11912.

A Smart Individual Anode Current Measurement System … 58. Shi J, Yao Y, Bao J, Skyllas-Kazacos M, Welch BJ, Jassim A, Mahmoud M (2021) A New Control Strategy for the Aluminum Reduction Process Using Economic Model Predictive Control. IFAC-PapersOnLine. 54(11):49–54. 59. Shi J, Yao Y, Bao J, Skyllas-Kazacos M, Welch BJ, Jassim A, Mahmoud M (2022) Advanced Model-Based Estimation and

51 Control of Alumina Concentration in an Aluminum Reduction Cell. JOM. 74(2):706–717. 60. Wang R, Bao J, Yao Y (2019) A data-centric predictive control approach for nonlinear chemical processes. Chemical Engineering Research and Design. 142:154–164.

Light Metals Research at the University of Auckland J. B. Metson, R. Etzion, and M. M. Hyland

Abstract

The University of Auckland has had a more than 40-year history of research contributions supporting the light metals industries. This research programme was initiated with the appointment of Professor Barry Welch in Chemical and Materials Engineering but subsequently broadened to embrace academics across a range of disciplines, particularly in chemical sciences. Work was initially focused on the aluminium and alumina industries but has extended at times into lithium, magnesium and titanium. A defining characteristic of this work has always been close engagement with the light metals industries. However, over time, the nature of this interaction has evolved. Of note has been convergence on a range of industry-defining and challenging issues where cross-sector collaborations, often working closely with government agencies, have progressively displaced individual projects addressing technical advantage. Environmental footprint and response to a dynamic energy environment are prominent within this common ground. These trends will be examined along with potential future directions for this research relationship. Keywords

Universities




Light metals research




Professor Welch

Introduction The light metals occupy an increasingly important position in the periodic table. Location towards the top of the table makes them abundant, while their utility and recyclability J. B. Metson (&)
R. Etzion
M. M. Hyland The University of Auckland, 22 Princes St, Auckland, 1010, New Zealand e-mail: [email protected]

make them increasingly important materials in a range of applications. Aluminium and Magnesium have been widely used in construction, packaging etc. and are critical in light-weighting of transport fleets, while titanium has been a key metal in aerospace, transport and bio-medical application. The ores from which raw materials for smelting are sourced are abundant and widely distributed across the globe leading to similarly widely distributed industry centres. Although the embedded energy in their manufacture is high, for aluminium and magnesium in particular most of this embedded energy is recovered when the metal is recycled. This leads for example to the remarkable statistic that an estimated 75% of aluminium ever made is still in use [1]. In an energy constrained world, a singular focus on the energy of production, while important, ignores the more significant issue of the lifetime energy cost of these materials. Provided target compositions can still be achieved, and recycling the metals ensures maximum utilization of embedded energy and these materials become a critical contribution to a greener economy and environment. Despite these advantages, the metals still present major challenges in manufacture and utilization in an energy-constrained and increasingly environmentally conscious world. For aluminium, the relative maturity of the Hall-Heroult and Bayer processes’ central technologies, dating from the 1880s, presents a particular problem. At a time of incremental optimization of established technologies operating at very large scale, there are pressing demands for step-change improvements in energy efficiency, environmental footprint and resource utilization. This has presented fertile ground for a range of related research themes both in support of these current technologies and in the exploration of more radical approaches to aluminium and alumina production. Research and innovation are key to the future of the light metals.

© The Minerals, Metals & Materials Society 2023 S. Broek (ed.), Light Metals 2023, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-031-22532-1_7

52

Light Metals Research at the University of Auckland

The University of Auckland Research at the University of Auckland in support of the light metals industries was largely initiated by the arrival of Professor Barry Welch’s in 1980, the same year the first edition of the seminal “Aluminium Smelter Technology” by Grjotheim and Welch was published [2]. This began a sustained period of growth in fundamental and applied research, largely in support of the international aluminium and alumina industries. The 1970s and 80s saw a major expansion of the aluminium industry in Australasia. In New Zealand, Comalco’s construction of the Tiwai Point Aluminium Smelter, completed in 1971, had made the primary aluminium industry strategically important to New Zealand [3]. This was part of major growth in the aluminium industry in the wider region, in part attributed to the contraction of the industry in Japan following the oil shock. This generated both a significant demand for skills, including graduate engineers and a need for research and technology support to sustain this growth. The research model established by Professor Welch at the University was distinguished by a combination of excellence in basic research, often with a direct line of sight to application. Projects were frequently coupled with significant time in smelters and, on occasion, alumina refineries, for most of the many students and their supervisors engaged with these research programmes. This combination nurtured strong industry relationships, depth and flexibility in researchers and potentially short pathway to the application of knowledge. Of note was the breadth of experimental capability established at the University during this time, with major furnace and electrochemical capabilities, able to simulate aspects of the aluminium reduction industrial environment at a sufficient scale to provide meaningful insights. This capability was supported by a range of leading-edge characterization facilities. The Research Centre for Surface and Materials Science was established in 1988 and also hosted by the Department of Chemical and Materials Engineering. The Centre housed both electron microscopy facilities and New Zealand’s first X-ray Photoelectron Spectrometer (XPS) instrument, offering a unique surface analytical capability very quickly harnessed in support of such industry-connected research programmes. The supporters of this XPS purchase included Comalco, NZ Steel, and New Zealand’s Defence Scientific Establishment. As the Auckland program grew, it embraced an increasingly diverse group of academic researchers, including Professors Chen, Hyland, Metson, Muller-Steinhagen, Taylor and a steadily expanding range of interests and capabilities. Work covered a remarkable span of research topics, including:

53

• • • • • • • • • • • • • • • • •

Alumina dissolution and the resistance curve. Bubble dynamics and impacts on current efficiency. Anode reactivity and carbon consumption. Mechanisms of cathode degradation. Degradation of sidewall refractories. Heat transfer through cell cover. Operational and control decision-making for smelters. Alternative technologies for aluminium production. Emissions and the mechanisms of HF generation and adsorption on alumina. Dry scrubbing and the performance of aluminas. Sulfur gases and impacts on dry-scrubbing. Impacts of impurities in reduction cells. Alumina calcination and resulting performance in the smelter. PFC generation. Potroom dust and the implications for environmental emissions. Power modulation of reduction cells. Catalytic destruction of PFCs.

More than 30 Ph.D. students have completed their degrees on these projects. The experimental capabilities have also been harnessed in commercial materials testing, particularly in cathode degradation, alumina dissolution and general alumina characterization. The close relationship of Professor Welch and others with Comalco, the operator of the Tiwai Point smelter, was highlighted by the very significant role University researchers played in support of the major upgrade of the smelter completed in 1996. A major catalyst for the growth of research at Auckland through the 1980s and 90s was undoubtedly the technological ambition of Comalco in aluminium reduction. Work on drain cathode cells [4, 5], the development of Torbed dry-scrubbers [6, 7], amongst other initiatives, provide some of the more challenging examples. This close relationship also saw the appointment at the University of Auckland of now Professor Margaret Hyland as the Comalco Lecturer in Materials Science and Engineering in 1991. Of note also has been long-term relationships with a range of other industry players such as Aluminium Pechiney in the formative years, Hydro Aluminium, Trimet Aluminium, EGA, Rio Tinto, Outotec, SAMI, GAMI and others. This also generated particularly strong relationships with other University based research providers including NTNU, UNSW and the REGAL consortium. The formation of the Light Metals Research Centre in 2002 was led by Professors Welch and Metson and triggered by the retirement of the former and the desire to build on the momentum already established. The subsequent recruitment of the former Auckland graduate Professor Mark Taylor as

54

the Director saw a major build in activity, with a particularly fruitful relationship with the rapidly emerging Chinese industry and addressing the increasing survival challenges of European based smelters. Dr Mark Dorreen took over in this role from 2014, however the years following the Global Financial Collapse saw increasingly tight conditions in the industry, with flat metal prices prompting limited research activity, contraction of the industry in Australasia and refocusing the University of Auckland efforts.

Education As with other universities engaged in research with the industry, part of the Auckland model has also been to support the development and training of industry personnel. In addition to contributions to numerous TMS short courses, the University of Auckland Post-Graduate Certificate in Aluminium Reduction Technology and the follow-on Masters programmes have been offered in a succession of venues including New Zealand, Australia, several venues in the Middle East and Germany. These courses are centred on research-based teaching and learning, drawing on research leaders and expertise from across the international industry. The format of the courses reflected a pre-Covid world and has typically been based around a three-week residential block course located close to a smelter to allow observation and practical experience of a range of processes. This practical approach has been used both in formal post graduate qualifications such as the PG Cert. and in training smelter teams worldwide on production process fundamentals, benchmarked operation and process control strategies.

Evolution of Research Needs in the Global Industry While Light Metals research at the University of Auckland has now passed through several generations of researchers, the industries themselves have also seen a major structural change through this same period. Of particular note has been the consolidation of the historical industry through mergers and, with this, the contracting numbers of major industry centres of research excellence, particularly in North America, Europe and Australia. In parallel, the industry has seen expansion elsewhere driven by power costs, with the emergence of accompanying major research and technology capabilities, particularly in China and the Middle East. This period has seen the rise of a new geographical distribution of both producers and technology suppliers. This change has resulted in a more limited and concentrated research base with significant dependence on research

J. B. Metson et al.

institutions offering complementary skills and facilities. This breadth has also reflected a wider range of industry challenges and access to the more diverse range of disciplines needed to address them. Industry challenges such as environmental impact, reduced energy consumption, and managing power insecurity also represent “common good” research where industry partnerships, frequently with government involvement, are needed to make real progress. The collaboration between the University of Auckland and governmental agencies has led to several projects that addressed fundamental issues within the aluminium industry. A US EPA-funded initiative has resulted in developing a fluoride emissions manual [8] produced in Mandarin and English and working on understanding PFC emissions and ways to reduce them. A collaboration of Auckland University with Australian Universities funded by the Australian Government via CSIRO led to a cluster of projects that aimed to find innovative breakthrough solutions to significant problems of the production process by focusing on three themes: cell design and operation, alternate processes/breakthrough technologies, and process control. The New Zealand Government’s support has also helped test innovative power modulation technology [9] in the Tiwai Point smelter and ultimately deployment in Europe. The strategy of sourcing innovation and complementary capabilities outside of the industry has followed a similar pathway to developments in other sectors. Perhaps the lead has been shown by the pharmaceuticals sector, where the cost of in-house research capacity, particularly in drug discovery, has become near unsustainable. Industry instead seeks to harness the power and breadth of smaller, nimbler innovator companies and particularly Universities [10]. Tralau-Stewart et al. [11] highlight the role of universities in this drug discovery process and argue for new partnership models that leverage public funding where common areas of interest allow a win–win outcome. It could be argued that the initiatives above, particularly in emissions control, already demonstrate the application of this model to the aluminium industry. There is little doubt that in parallel with company specific initiatives, the collaborative model will be increasingly prominent in the industry.

Conclusions The University of Auckland has had a long and distinguished history of engagement with the light metals industries, particularly in the smelting of aluminium and the manufacture and properties of metallurgical alumina. Built on the pioneering work of Professor Barry Welch, this has provided more than 30 years of fundamental insights, research support and innovative technologies to a steadily evolving industry. This evolution has been structural in

Light Metals Research at the University of Auckland

terms of the major companies, technological, including in terms of models of research collaboration and increasingly geographic, such that in some parts of the world the industry continues to face major challenges to its very existence.

55

6. 7.

References 1. https://international-aluminium.org/work_areas/recycling/ accessed 31/07/2022. 2. K. Grjotheim and B.J. Welch (1980) Aluminium Smelter Technology: A Pure and Applied Approach, Aluminium-Verlag. Second edition published in 1987. 3. Grant, David. Aluminium smelter, Tīwai Point. teara.govt.nz. Te Ara - The Encyclopedia of New Zealand. 4. B J Welch JOM, 51 (5) (1999), pp. 24–28. Aluminum Production Paths in the New Millennium. 5. G.D. Browne et al., “TiB2 Coated Aluminium Reduction Cells: Status and Future Directions of Coated Cells in Comalco,” Proc.

8.

9.

10.

11.

6th Aust. Al Smelting Tech. Workshop, ed. B.J. James, M. Skyllas-Kazacos, and B.J. Welch (Sydney, Australia: U. of N.S.W. and RACI, 1998), pp. 499–508. Torbed dry scrubber-Register, J. (1995). Tiwai’s Upgrade - Thank Sweat and the ECA. New Zealand Engineering, 50(10), 6–9. https://www.nzas.co.nz/files/1235_20160718112004-1468797604. pdf. Nursiani Tjahyono, Yashuang Gao, David Wong, Wei Zhang & Mark P. Taylor. Fluoride Emissions Management Guide (FEMG) for Aluminium Smelters. Light Metals 2011 pp 301–306. Lavoie, P., Namboothiri, S., Dorreen, M., Chen, J.J.J., Zeigler, D. P., Taylor, M.P. (2011). Increasing the Power Modulation Window of Aluminium Smelter Pots with Shell Heat Exchanger Technology. In: Lindsay, S.J. (eds) Light Metals 2011. Alexander Schuhmacher, Oliver Gassmann & Markus Hinder (2016) Changing R&D models in research-based pharmaceutical companies. Journal of Translational Medicine volume 14, Article number: 105. Cathy J. Tralau-Stewart, Colin A. Wyatt, Dominique E. Kleyn and Alex Ayad. (2009) Drug Discovery Today Volume 14, Numbers 1/2 January 2009 REVIEWS.

Impact of Aluminium Reduction Cell Parameters on Feeder Hole Condition Pascal Lavoie and Mark P. Taylor

Abstract

Introduction

Aluminium reduction requires alumina to be fed and dissolved into electrolyte. The feeder hole condition may impact severely the ability to dissolve the added alumina. In this study, two sets of observations were made on a large sample of aluminium reduction cells to understand the feeder hole condition progress during feed events and determine which reduction cell parameters affect the feeder hole condition and the progression pathway of the feeding event sequence. The results show that feed rate has a major impact on the initial feeder hole condition and is determinant to the pathway of the feeding sequence. Higher superheat, lower excess AlF3 and longer time since the last anode change increased the probability of finding feeder hole conditions conducive to dissolution. The observations of blocked feeder holes were found to be linked to abnormal process conditions or a mechanical issue. Changes made to the feeding strategy between the two observation sets led to a significant improvement of the feeder hole condition. The proportion of opened feeder holes increased from 12 to 41%, resulting in a more than 50% reduction of the anode effect frequency smelter-wide over an 18 month period. Keywords





Alumina Dissolution hole Sludge




Feeder




Superheat




Feeder

The work reported in this paper is part of independent research from Pascal Lavoie conducted during completion of his Ph.D. at the University of Auckland, prior to joining Alcoa. P. Lavoie (&) Alcoa Corp., Quebec, Canada e-mail: [email protected] M. P. Taylor The University of Auckland, Auckland, New Zealand e-mail: [email protected]

In modern electrolysis cells alumina is fed through point feeders which periodically add a volumetric dose equivalent to between 1 and 2 kg into the cell. Most of these feeders consist of two devices, as shown in Fig. 1. The first device, a “breaker”, is an air actuated steel shaft coming down to break the electrolyte crust that may have formed between feeding events. The second device, a “feeder”, delivers a fixed volumetric dose of alumina to the feeder hole [1]. The condition of the feeder hole may impact the dissolution rate of the alumina added. Point feeders are normally designed to feed into molten bath in order to maximise the effective contact area between the alumina particles and the electrolyte [3]. However, feeder holes have been observed to operate in a push feed mode [4] where the alumina from the last feeding event lodges in the feeder hole and is pushed into the bath by the breaker at the next feeding event. This has the potential to greatly limit the effective contact area between the alumina and bath, and hinder alumina dissolution leading to sludging and operational difficulties [5]. Some smelters may intentionally operate the feeder hole in such closed state to reduce gaseous emissions or conserve heat but at the detriment to dissolution. In this article, the authors describe the factors and mechanisms that may affect the feeder hole condition. The results of a study based on experimental observations of feeder hole conditions and selected cell parameters are presented. Finally, the implications of controlling cell parameters to maximise alumina dissolution are discussed.

Factors That May Affect Feeder Hole Condition In a previous article, a list of factors that may affect the feeder hole condition and the mechanisms that in turn would affect alumina dissolution have been presented [4].

© The Minerals, Metals & Materials Society 2023 S. Broek (ed.), Light Metals 2023, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-031-22532-1_8

56

Impact of Aluminium Reduction Cell Parameters …

57

In addition to the cell parameters measured on the day of the observations, the following elements were noted:

Fig. 1 Point feeder schematic. Adapted from [2]

The hypothesis developed in [4] is that by carefully controlling the feeder hole condition, through the cell parameters management, alumina dissolution can be aided. This first linkage in Fig. 2 is the focus of the present paper, while the second linkage was also investigated but will be subject of further publications from the Ph.D. thesis of the primary author [6]. The actual relationship between these cell parameters previously hypothesised as factors impacting dissolution coupled with observations of feeder hole conditions is therefore examined in the present paper. The cell parameters studied are the following: • • • •

Bath level. Metal level. Total liquid level. Bath Temperature. In the present study, a Heraeus FiberLab.1 • Liquidus Temperature. In the present study, also the FiberLab synchronously with bath temperature. • Superheat. As defined in the aluminium industry, the difference between bath temperature and liquidus temperature. • Excess AlF3. In the present study, also using the FiberLab.

Measurement of bath temperature by optic fiber and analysis of the cooling curve inferring bath composition. www.heraeus.com.

1

• Last anode changed An aluminium reduction cell consists of a multitude of anodes. Anodes changed in the vicinity of the feeder hole have the potential to cool the bath locally as they are introduced at room temperature and will absorb sensible and latent heat of freezing from the bath as they heat up to the operating temperature. • Feed phase A cell undergoes a feed cycle imposed by the cell controller to regulate the alumina concentration [7]. As a consequence, the sensible heat in the feeder hole region gradually depletes during overfeed (enriching of the alumina concentration in the bath), as more cool alumina is introduced at a higher rate than stoichiometrically required for aluminium production. Conversely, the sensible heat of the bath in and around the feeder hole increases during the underfeed phase, depending on how far above and below these feed rates are from the theoretical requirement of the cell the feed rate [7]. This variation in temperature due to the feed cycle can be significant and is observed globally throughout the cell as well as locally in the feeder holes, as is represented in Fig. 3. In the present study, the feed cycle is schematically represented in Fig. 4.

Feeder Hole Behaviour Description and Characterisation Observation Sequence For each feed event observed, the feeder hole condition and behaviour throughout that event was recorded following semi-quantitative, ordinal characterisation scales describing the state or behaviour of the feeder hole during the event time sequence below: 1. Initial state/Type of break. The state of the feeder hole immediately before initiation of a feeding event (0–10 s). 2. Response to breaking. The behaviour of the feeder hole during the first phase of a feeding event, just after the breaker is actuated (0–5 s). 3. Response to feeding. The behaviour of the feeder hole during the second part of a feeding event, just after the alumina dose is delivered by gravity to the bath surface (0–10 s). 4. Closing state. The state of the feeder hole, approximately 10 s after the alumina dose has been delivered.

58

P. Lavoie and M. P. Taylor

Cell Parameters

Effect on feeder hole condition

Feeder Hole Conditions

Effect on dissolution

Alumina Dissolution

Fig. 2 Link between cell parameters and alumina dissolution

4. Open feeder hole In the open feeder hole state, the molten surface of the electrolyte can be seen and is free of solid material in this initial stage.

Response to Breaking Four behaviours have been defined to describe the response of the feeder hole following the breaker actuation

Fig. 3 Example of extensive bath temperature variation measured at the end of a cell throughout the feeding cycles [7]

Feeder Hole Condition State Definition The following feeder hole condition states have been defined for the purpose of the study. They consist in an ordinal scale of the feeder hole condition. Under the hypothesis concerning the feeder hole condition affecting dissolution, the scale ranges from the least favourable to the most favourable for alumina dissolution. These states can be visualised in Fig. 5. 1. Blocked feeder hole A blocked feeder hole consists of a crusted electrolyte with a significant accumulation of alumina in the hole resulting in the inability of the breaker to break the crust and push the material in molten bath. 2. Push feeder hole A push feeder hole consists of a crusted electrolyte surface with a piled quantity of alumina. As opposed to the blocked feeder hole, the breaker action successfully pushes the material in the molten bath, with material present in the hole after the breaker retraction. 3. Break feeder hole Molten bath surface cannot be seen in the initial state, its surface is covered by crusted electrolyte with powder alumina on top or not. The breaker operation breaks the crust and pushes the material in the molten bath underneath. The molten bath surface can be seen after the breaker retraction.

1. No Response The breaker comes down and may push the material inside the molten bath or not. In either case, there is no dynamic change notable within the feeder hole once the breaker retracts. 2. Particle Aeration The breaker action leads to the evacuation of process gases throughout a small opening, leading to alumina particles being aerated upwards. 3. Bubbling A bubbling feeder hole consists in a semi-solid electrolyte surface with a small quantity of alumina present on the top. Gaseous emissions can be observed creating a bubbling effect in the alumina present on the top. 4. Open Same definition as with the feeder hole state.

Response to Feed Three behaviours have been defined to describe the reaction in the feeder hole when the alumina dose is delivered. Note that number 2 has been omitted in the sequence to conserve a quantitative scale consistent with other states or behaviours. 1. On Material The alumina dose falls on the crust or alumina already present in the feeder hole. No gas evolution or movement of the material can be seen. 3. Bubbling Same definition as with the feeder hole condition.

Impact of Aluminium Reduction Cell Parameters …

59

160%

OA

Time between feed relative to theoretical

140%

120%

OB 100%

80%

UA 60%

AE 40%

20%

0% 0

20

40

60

80

100

120

140

Time

Fig. 4 Feed cycle diagram showing the four feeding rates used by the controller for the cell technology under study (OA, the highest federate; OB, at slight overfeed; UA, at slight underfeed; AE, the lowest feed rate)

Fig. 5 Ordinal scale of feeder hole states. Going from the least (1) to the most (4) favourable to alumina dissolution

60

P. Lavoie and M. P. Taylor

4. Open The molten surface of the electrolyte can be seen but a floating alumina raft could be present. However, the wave movement is noticeable and the alumina raft can be seen dissolving or floating away.

Experimental Observations In two separate series of experiments, a relatively large number of feeding events in reduction cells of the same technology were observed, 66 and 70 events respectively. For each set of experiments the breaking/feeding states and responses defined above were characterised. The series of observations in each experiment occurred over a timeframe of approximately two hours. Between these sets of process characterisations, some of the operating conditions were changed. Adjustments were made to the feeding strategy and a specific abnormality detection algorithm implemented to detect and prevent sludging of the cells similar to that described by Mulder et al. [8]. The potline was however operated with the same amperage, working cycle and other parameters targets remained similar. The observations were made at random timeframes during the feed cycle. The cells were observed sequentially. They were not selected randomly but were within a group where the cell parameters (Temperature, Liquidus, XsAlF3) had been measured within the 8-h shift, and where no other physical cell operations were occurring.

Results State and Behaviour Progression Throughout the Feeding Sequence During a feeding event, the feeder hole condition can evolve along different pathways, progressing from the initial state to the closing state. In order to assess the progression likelihoods of the feeder hole state and behaviour throughout the feeding sequence, a probability tree was built. The

observations were compiled to represent the likelihood of progression to each branch of the next step, schematically shown in Fig. 6. The evaluation of individual links along the feeding sequence is detailed in the primary author’s Ph.D. thesis [6]. Examples of the probability tree evaluation are provided in this paper to illustrate the methodology and the conclusions presented below. As a first example, differences can be seen on Fig. 7, when comparing the proportion of initial state by feed phase between the two observation sets. During the first set of observations, the likelihood of finding a feeder hole state detrimental to dissolution (1 “blocked” or 2 “push”) was the highest (A), except during the lowest feeding rate phase AE (B), where it was more likely to be beneficial to dissolution (3 “bubbling” or 4 “open”). However, it was equally likely to find a feeder hole state beneficial to dissolution in all feed phases during the second set of observations (C), except when feeding at the highest rate OA (D), where the likelihood tended slightly towards a detrimental state. This improvement in the feeder hole condition from observations set 1 to set 2 can be attributed to the adjustments made to the feed strategy by the smelter management, aimed at improving dissolution. These included restricting feed rates during overfeed and the addition of an online abnormality detection of excessive alumina feed against normal demand through time. Although the occurrence of detrimental feeder hole states was marginally reduced in set 2 from 56 to 50%, the increase in open feeder holes from 12 to 41% mostly came from improvement in the “Break” state condition (Not explicitly shown on Fig. 7). As a consequence, a significant reduction in the sludging behaviour is observed in the cells under study. The alumina cusum amplitude, a soft sensor for the extent of undissolved alumina settled at the bottom of the cell [9] reduced by approximately 500 kg (99% confidence). This improvement in feeder hole condition and reduction in sludging behaviour has yielded a large decrease in anode effect frequency of approximately 45% observed smelter-wide and sustained over one year. This last observation is significant and will be discussed later in the paper.

4 Initial State / Type of break

1

Response to break

2

Bath condition at feed

Fig. 6 Analysis of progression of state and behaviour through the feeding sequence

3

Response to feed / Closing state

Impact of Aluminium Reduction Cell Parameters …

61

Chart of Initial state n 20

56%

C A

A

41%

D

29%

B

C

0% 0% 0% 0%

0%

1 Blocked 2 Push 3 Break 4 Open

1 Blocked 2 Push 3 Break 4 Open

1 Blocked 2 Push 3 Break 4 Open

4 OA

1 AE

2 UA

50%50%

1 Blocked 2 Push 3 Break 4 Open

5%

1 Blocked 2 Push 3 Break 4 Open

0% 0%

Set 2

0%

15%

4 OA

10%

0%

C

11%

3 OB

1 Blocked 2 Push 3 Break 4 Open

50% 40%

Set 1

Date

0%

2 UA

Feed rate

8%

0%

12% 12%

9%

1 Blocked 2 Push 3 Break 4 Open

0

Initial state

33%

35%

1 Blocked 2 Push 3 Break 4 Open

58%

5

45%

3 OB

10

1 AE

Initial state n

15

44%44%

Fig. 7 Initial feeder hole state distribution between feed phase and observation sets

Progression from Initial State to Response to Break (Link 1, Fig. 6) As a second example, Fig. 8 shows the effect of the feed rate and the initial state on the response to break. Because of low occurrence, the initial condition “Blocked” and the feed rate “OB” were omitted from this graph. As can be expected, having an open feeder hole condition initially was never affected by the breaker actuation. When starting with a detrimental state (“push” in this case), the likelihood of obtaining a detrimental response to break is high (77%). However, the response to break improves with lower feed rates, as the modal behaviour moves from “1 no response” in OA to “2 particle aeration” in UA and “1 no response” becomes exceptional in AE feed, the lowest feed rate. Feeder hole condition can therefore improve during lower feed rate phases, probably because there is more time to replenish and increase sensible heat in the feeder hole between each feed event. Such behaviour has also been demonstrated in a previous study and can be observed in Fig. 9 [7]. Finally, when the initial condition is “3 Break”, there is by definition a response to break. In the majority of observations (85%), the response to break is favourable, having modes at either “3 Bubbling” or “4 Open”. Contrary to the general trend found within the study, with the initial

condition “3 Break” the lowest feed did not lead to the most favourable response. This could be due to differential bath temperature and liquidus temperature evolutions or to the time between breaks allowing a stronger crust to form, getting a higher proportion of “2 Particle Aeration” than with higher breaker operation frequency.

Progression from Response at Break to Bath Condition at Feeding (Link 2, Fig. 6) As a final example, Fig. 10 shows the progression from the response at break to the bath condition at feeding. In all cases where no response occurred at break (A), the alumina dose was delivered onto material without perceivable movement of alumina in the feeder hole, maintaining the detrimental state. Similarly, a feeder hole that was open (B) invariably had the alumina dose delivered into the molten bath. When the breaker operation was partially successful (C), resulting in aeration of particles, the condition at feed was mostly detrimental. The alumina would most likely be delivered onto a pile of material (62%). The feeder hole condition had however improved between breaking and feeding, moving into bubbling 33% of the time and even opening completely in 5% of the cases. Performing ordinal

62

P. Lavoie and M. P. Taylor

Chart of Response to break n 100%

54%

15 100%

10

39%

26% 57% 20% 67% 63%

5

100% 29%

1 No R 2 Aeration 3 Bubble 4 Open

1 No R 2 Aeration 3 Bubble 4 Open

1 No R 2 Aeration 3 Bubble 4 Open

2 Push

4 OA 3 Break

4 Open

1 No R 2 Aeration 3 Bubble 4 Open 2 Push

0 0 0

1 No R 2 Aeration 3 Bubble 4 Open

1 No R 2 Aeration 3 Bubble 4 Open 4 Open

0

4 Open

1 No R 2 Aeration 3 Bubble 4 Open

0 0 0

1 AE 3 Break

Feed rate

0

0

1 No R 2 Aeration 3 Bubble 4 Open

Initial state

20%

12%

2 Push

Response to break

11%

60%

1 No R 2 Aeration 3 Bubble 4 Open

11%

0

13% 9%

14%

25%

2 UA 3 Break

Response to break n

20

Fig. 8 Effect of feed rate and initial state on response to break. Overfeed “3 OB” and initial condition “1 Blocked” omitted due to low occurrence

Fig. 9 Sensible heat increasing during underfeed and decreasing during overfeed [7]

logistic regression allows identification of cell parameters that affect the outcome likelihoods [10]. Four factors were determined to be statistically significant (90% confidence) leading to a better feeder hole condition: • Superheat. For each additional degree of superheat, the odds of feeding on material are reduced by 40%. • Lower XsAlF3. For each 1% increase in XsAlF3, the odds of feeding on material double.

• Bath level. Surprisingly, as bath level increases, the odds of feeding on material increased as well. Closely examining the parameters however, revealed that the trend had been driven by three observations of low bath levels that also had low XsAlF3 and enough superheat. This suggests that the other favourable parameters helped overcome low bath levels to maintain the feeder hole condition at a better state. • Feed rate. The statistical significance of this factor is only marginal as the number of data points are small, especially at OB (feed rate 3). Therefore, there may be a trend that as the feed rate increases, and the probability of feeding on the material increase as well. The statistics cannot however quantify the likelihoods accurately. Finally, when the breaker was successfully operated (D), obtaining a bubbling response to break this state remained until feeding in the vast majority of observations. In exceptional cases (2 occurrences or 9%), the bubbling behaviour had stopped after feeding occurred. In one case, the superheat was only 4 °C which could explain the issue. In the other case, the parameters do not seem to explain the outcome.

Impact of Aluminium Reduction Cell Parameters …

63

Chart of Condition at feed n 100%

50

Condition at feed n

B 40

C

30

A

D

62% 91%

20

17 33%

10 0

0

1 Material

4 Open 3 Bubbling

4 Open

0

4 Open

3 Bubble 3 Bubbling

9%

1 Material

4 Open

2 Aeration 3 Bubbling

1 Material

0

4 Open

Response to break

0

1 No R 3 Bubbling

Condition at feed

1 Material

5%

0

Fig. 10 Effect of response to break on bath condition at feeding

Discussion Effects Cell Parameters Affecting the State Progression The complete analysis [6], too long for this publication, shows that the alumina feeding rate does have a significant influence on the initial feeder hole condition. As the feed rate is increased, the likelihood of feeding alumina into a feeder hole in a detrimental state increases. This has repercussions throughout the feeding sequence as the initial feeder hole condition influences greatly the pathway of the condition through the subsequent feeding steps. The mechanism by which this occurs is probably the gradual depletion of sensible heat during the overfeed period through feeding before local sensible heat has had time to recover [4]. This suggests that restraining the range of feed rates to ensure the local sensible heat is replenished before the next feed event is a key to maintaining the feeder hole condition. In contrast, the study found marginal statistical influence of very low feed rates to drive a detrimental condition at feeding and response to feed. It is possible that as the time between the crust break increases, the thickness and strength of the crust formed between feed events also increase, creating more difficulty in breaking. In that case, restricting the

range of aggressive underfeed (i.e. AE feed in this study) or actioning the breaker at a higher frequency than the feeder would help maintaining the breaking ability. The statistical analysis showed that two cell parameters were clearly impacting the feeder hole condition throughout the feeding sequence. The first is superheat, governing the available sensible heat for dissolution. Maintaining enough superheat is therefore critical in maintaining the feeder hole condition. The second is Excess AlF3, impacting the alumina solubility and therefore the mass transfer gradient away from the dissolving alumina is needed for its dissolution. Finally, anode change also impacted the ability to maintain feeder hole condition. In the case of this study, the anode position itself did not appear to have statistical significance, but the time since the last anode change did. Optimisation of additional energy input to the cell after anode change could be necessary to compensate for this effect.

Blocked Feeder Holes Obtaining a blocked feeder hole condition should be considered a process failure as it is abnormal to get the condition and it stems from assignable causes of variation. In all 3 recorded occurrences, the condition was observed during OA, the over feed phase where sensible heat in the feeder hole is likely to be at its lowest as was shown

64

in Fig. 9. Examining the cell parameters of these 3 cases leads to the conclusion that low liquid levels, preventing the breaker from reaching the liquid (low bath in 1 case, low metal in another case) and mechanical issues (leaky feeder in the last case) provoked a blocked feeder hole condition. A blocked feeder hole remained blocked in the 3 cases recorded, all observed during an overfeed phase. It is possible that the hole condition could improve during an underfeed phase, but from experience and other observations [11], it is more likely to require physical attention from a worker to unblock or possibly a special breaking treatment triggered by the automatic controller if this condition can be detected automatically.

Improving the Hole Condition to an Open Hole During the two sets of observations, no observations were made where a feeder hole condition terminated its feeding sequence in an open state if the initial condition was different than open. Since open feeder holes were frequently observed, there are certainly situations under which a nonopen feeder hole would improve to an open state but these are yet to be determined.

Fig. 11 Smelter-wide anode effect frequency reduction

P. Lavoie and M. P. Taylor

Impacts to Smelter Performance Changes to the feed strategy between set 1 and set 2, made partially as a result of the data collected here in set 1, yielded a significant performance increase to the smelter, with the Anode Effect Frequency reduced from 0.094 AE/cell/day to a level of 0.042 AE/cell/day. This represents a drastic reduction of more than 50%. There are always multiple factors contributing to the outcome of overall cell performance, but the anode effect frequency is directly related to the effectiveness of the alumina feeding process (Fig. 11).

Limits of Study and Comparative Observations with Other Technologies As the observations were made randomly with regards to the feed cycle, there are cases which would have been classified under the subsequent feed phase because they were made shortly after a phase transition. Therefore, there could possibly be a categorical bias for a small proportion (1,000 RB grains in “8Cu simulation”, a ⌀24  12 mm cylindrical box was created for constraining generated 1,273 spherical discrete elements with a diameter of 1700 ± 10 lm. Those elements were evenly distributed after reaching a dynamic equilibrium state using a simple linear contact model [24]. The centroids of spherical elements then served as points for Voronoi tessellation to generate an equiaxed *100% packing fraction assembly comprised of RBs similar to the Refs. [18, 28]. Desired packing fraction (82% for Al–Cu at 583 °C) was reached by applying erosion operation on each RB element. In the “15Cu simulation”, 1,289 two-circle grains (represented as a clumped sphere, CS, in Fig. 2) with an equivalent spherical diameter of 1345 ± 368 lm were also constrained inside a virtual ⌀24  12 mm cylindrical box with a packing fraction of 60%, and the same linear contact model was used to equilibrate this assembly. The “switching model” step started by replacing the linear contact model with Burgers model for all RB-RB contacts and the Hertz model for RB-W contacts in the “8Cu simulation”. Adjustment of g1;n shown in Table 2 for compression simulation in 583 °C was due to the consideration of temperature-dependent flow stress of the solid aAl [27]. In addition, the difference in mass densities between the liquid phase and solid phase for semi-solid Al–Cu was accounted for by introducing a buoyancy field ð1  qL =qS Þg. Low friction coefficients (0.01 for RB-RB and 0.0 for

Exploring Semi-solid Deformation of Al–Cu Alloys by a Quantitative Comparison …

561

Fig. 2 DEM modeling and simulation flow chart Table 2 A summary of the contact model parameters used in Hertz and Burgers contact model, including Maxwell and Kelvin parts. Parameters of shear components in Burgers contact model were set the same as those of normal components

Temperature 540 °C and 583 °C

Parameters

Value

Shear modulus—Hertz contact model (G)

1:71  10

Poisson’s ratio—Hertz contact model (m)

0:384

Normal contact stiffness—Kelvin part (k2;n )

1:55  107

583 °C

Pa

N=m N
s=m N=m

2:4  10

6

Normal contact stiffness—Maxwell part (k1;n )

2:5  10

7

Normal damping coefficient—Maxwell part (g1;n )

1:15  106

N
s=m

Normal damping coefficient—Maxwell part (g1;n )

5

N
s=m

Normal damping coefficient—Kelvin part (g2;n ) 540 °C

Unit 10

RB-W) were specified since low l helps create a stable high packing fraction assembly [4, 19]. After cyclic compression at a slight strain three times similar to Ref. [29], the dynamic equilibrium was again reached. In the “15Cu simulation”, Burgers model and Hertz model with the same set of parameters were applied to CS-CS contacts and CS-W contacts, respectively. The total contact forces between the upper layer of the “15Cu simulation” assembly and the top wall drop down very quickly. Virtual “8Cu simulation” and “15Cu simulation” samples were then subject to axial compression by the ⌀24 mm mobile top wall at a constant displacement rate duz =dt ¼ 120 lm=s. This corresponds to the engineering compressive strain rate e_ ¼ 0:01 s1 .

8:48  10

Results and Discussion Microstructural Evolutions of Al–Cu Alloys Figure 3 shows Al–Cu microstructure under SEM-BSE mode in different processing states. In the as-cast condition, the microstructure of the Al–15Cu specimen is more dendritic due to thermosolutal instability [30] and contains more aAl + Al2Cu eutectic compared with the Al–8Cu specimen. After globularization heat treatment, grains become coarse and globular due to Ostwald ripening [31]. The grain size of equiaxed aAl also rises significantly from 50–100 µm to

562

T. C. Su et al.

Table 3 A summary of microscale and macroscale simulation settings in this study

Microscale

Macroscale

“2Ws + 2RBs”

“2Ws + 2Ss”

“8Cu simulation” 1,273 grains in a cylindrical box

“15Cu simulation” 1,289 grains in a cylindrical box

Size [lm]

Two cubes in contact a ¼ 1410

Two spheres in contact d ¼ 1750

d ¼ 1700  10 Voronoi tessellation using sphere centroids

d ¼ 1615  368 two-sphere grain

Eff. Gravity [m/s2]

None

1:075 (upward)

Friction Coefficient [–]

lRBRB ¼ lRBW ¼ 0:1

lRBRB ¼ 0:01; lRBW ¼ 0

Fig. 3 The SEM-BSE microstructure of Al–8Cu and Al–15Cu alloys in different processing states

300–450 µm. The globularized equiaxed aAl grains naturally become rounded polyhedral in Al–8Cu. After drained die compression on globularized Al–8Cu and Al–15Cu

specimens at 583 °C, liquid expulsion results in the near disappearance of the aAl + Al2Cu eutectic structure that was occupying aAl interstitial space.

Exploring Semi-solid Deformation of Al–Cu Alloys by a Quantitative Comparison …

Figure 4 shows the EBSD IPF-Z maps of Al–8Cu and Al–15Cu specimens after globularization and drained die compression. Some small aAl grains are distributed along grain boundaries between large aAl grains in compressed samples, which may be explained as the occurrence of multiple metallurgical phenomena during late-stage compression: (i) partial cohesion, (ii) continuous dynamic recrystallization, and (iii) grain growth during cooling. First, some thin liquid films separating contacting grains were replaced with solid–solid interfaces by grain boundary coalescence [32]. The grain boundaries can serve as regions for forming sub-grain enclosed by dislocation cells and increased grain boundary misorientation due to sub-grain rotation known as continuous dynamic recrystallization (CDRX) [33]. The post-compression cooling with a limited cooling rate of 20 °C/min may be responsible for grain growth under the temperature range 480–548 °C [34]. It will be valuable to conduct more EBSD investigations on drain-compressed Al–Cu samples to elucidate the deformed aAl EBSD IPF maps as a function of process parameters and Cu concentrations.

Fig. 4 EBSD inverse pole figure of aAl phase in Al–8Cu and Al– 15Cu alloys

563

Burgers Contact Behavior in Microscale DEM Simulations Figure 5a compares stress–strain curves between “2Ws + 2 RBs” microscale DEM compression and previous measurements of compressing high-temperature (T=T m  1) Al– 4.5Cu alloy with compressive strain rate 0.01 s−1 and 0.001 s−1 by Braccini et al. [27]. The stress in this DEM simulation was calculated by the resistance force acting on the top wall divided by a square of the size length, a2 , whereas the strain was defined as the downward displacement of the top wall divided by the initial height of the sample, 2a. The iterative process and sensitivity analysis have shown that g1;n in Burgers model is linearly proportional to the flow stress under the constant e_ condition. On the contrary, k1;n , g2;n , and k2;n parameters do not influence the flow stress and slightly affect the yield behavior. Besides, g1;n in “2Ws + 2RBs” microscale DEM 0.001 s−1 compression was set 7:1  106 N
s=m to reproduce the experimental result, which was *6.3 times higher than “2Ws + 2RBs” 0.01 s−1 compression.

564

Besides, the contact force–contact overlap relationship during compression is primarily a function of discrete element shape. As shown in Fig. 5b, although the “2Ws + 2 RBs” and “2Ws + 2Ss” were set to have the same simulation loading conditions, the repulsive contact force rises much slower but reaches a higher plateau in “2Ws + 2Ss” than that in “2Ws + 2RBs”. The shape sensitivity of discrete elements suggests that DEM can effectively simulate grain rearrangement and grain-level strain localization with various particle shapes.

Visualization of Deformation Microstructure for DEM Simulations Figure 6 includes snapshots of discrete grain assemblies undergoing constant e_ z compression. Grain displacement along the z-axis is more negative for grains near the top wall. This is because the descending top wall contacted the upper layer of particles, and those particles pushed the lower layer of particles downward. Some grain-level heterogeneous phenomena are also visualized in Fig. 6. For example, grains in the “15Cu simulation” with an initial loose packing fraction of 60% became buoyant and floated. After grains packing into a new solid network, the repacked assembly was then subject to compression from the top wall. However, some particles in “8Cu simulation show high instantaneous velocity magnitude at the later stage of simulation (uz ¼ 746 lm), suggesting the occurrence of calculation instability. Further work should address this issue, possibly by understanding the effects of other Burgers contact parameters (k1;n , g2;n , and k2;n ). The bottom row of Fig. 6 shows the contact force chain network. With the increase of the compressive strain, the contact network gradually becomes denser, and the contact force magnitude increases. But at the same time, the contact force network of “8Cu simulation” has several thick main chains and thin branches, which is a typical phenomenon of granular materials [19]. In contrast, the “15Cu simulation” does not show well-defined main chains, which is worth checking its evolution by further compression.

T. C. Su et al.

die compression experiments in this study, and macroscale DEM simulations with g0S ¼ 82% and g0S ¼ 60%. It is found that the Al–8Cu semi-solid compression experiment from Drezet et al. shows higher stress in the solid fraction range 84–95% than measurement in this study. This can be realized because the globularized grains used in this research can rearrange easily, so the grain interlocking and stress resistance developed slower. On the contrary, the stress resistance of Al–8Cu at the late-stage deformation in this study is higher than the measurement from Drezet et al. [21], which can be attributed to the higher strain rate (0.01 s−1) and the development of a solid network in the globularized assembly. On the other hand, the Al–15Cu specimen with an initial solid fraction of 60% in this study has lower stress resistance than Al–8Cu until late-stage deformation. Still, it has higher stress in the solid fraction range of 65–96% than the measurement by Drezet et al. This suggests that the presence of liquid in Al–15Cu may act some pressures on the piston under 0.01 s−1 strain rate compression. Further work includes checking filter mesh size, compressive strain rate, and initial microstructure effects on the stress–volumetric solid fraction curves using the same apparatus. Figure 7b shows the relationship between interstitial liquid (void) ratio e ¼ ð1  gS Þ=gS and compressive stress is shown in the logarithmic scale known as e  logp 0 projection plane in CSSM [17]. We can see that all Al–8Cu (initial solid fraction at 82% or 84%) and Al–15Cu experiments (initial solid fraction at 60% or 62%) tend to converge into a straight one-dimensional normal compression line (1D NCL). It is also noted for the “8Cu simulation” that the facets for each RB are perfectly smooth in DEM. Still, the surface of aAl grains in Al–8Cu can be curved initially, giving rise to even softer contact behavior in the experiment. Nevertheless, the “8Cu simulation” shows yield behavior near the possible 1D-NCL in Fig. 7b. In comparison, the “15Cu simulation” using clumped two-sphere grain approach did not show a significant stress rise in Fig. 7a until 71% solid fraction, and its compression curve in e  0 logp diagram again yields near to the possible 1D-NCL. It will be valuable to couple with the computational fluid dynamics (CFD) approach in 3D to consider the liquid pressure effect, especially for “15Cu simulation”.

Comparison of Stress–Strain Response in Experiments and Simulations

Conclusions

Figure 7a summarizes stress–solid fraction curves of semi-solid Al–8Cu and Al–15Cu among experimental measurements with a compressive strain rate of 0.001 s−1 by Drezet et al. [21] (Al–Cu alloys were totally remelted and re-solidified to 555 °C, corresponding to Al–8Cu at 84% solid fraction and Al–15Cu at 62% solid fraction), drained

• After heat treatment, the Al–8Cu and Al–15Cu alloys became equiaxed polyhedral-like and globular, respectively, and the amount of Al2Cu became few after drained die compression. The change of grain size distribution after drained die compression suggests the occurrence of dynamic recrystallization.

Exploring Semi-solid Deformation of Al–Cu Alloys by a Quantitative Comparison …

565

20

Stress [MPa]

16

12

8

Braccini ε = 0.01 s-1 • DEM ε = 0.01 s-1 η1 = 1.15×106 N·s/m

4

Braccini ε = 0.001 s-1 • DEM ε = 0.001 s-1 η1 = 7.10×106 N·s/m

•

•

0 0.00

0.01

Strain

0.02

0.03

Contact Force [N]

40

30

20

Sphere (S) ε = 0.01 s-1 • Rigid Block (RB) ε = 0.01 s-1 • Sphere (S) ε = 0.001 s-1 • Rigid Block (RB) ε = 0.001 s-1 •

10

0 0

20

40

60

80

100

120

Contact Overlap [µm] Fig. 5 a Stress–strain curves of “2Ws + 2RBs” microscale DEM compression simulation and comparison with experimental measurements by Braccini et al. [27]. b Comparison of “2Ws + 2RBs” and “2Ws + 2Ss” microscale DEM compression behavior

Fig. 6 Snapshots of a macroscale DEM compression “8Cu simulation” and b macroscale DEM compression “15Cu simulation”. The bottom row shows the contact force network: the red color bar levels showing RB-RB or CS-CS contact force magnitude are on a logarithmic scale, and a thicker chain represents the greater force. The RB-W or CS-W contacts are represented by magenta

566

T. C. Su et al.

30

20

Interstitial Liquid Ratio [- ]

Stress [MPa]

25

0.7

Al-8Cu (DEM) Al-8Cu (Exp) Al-8Cu (Drezet) Al-15Cu (DEM) Al-15Cu (Exp) Al-15Cu (Drezet)

15 10 5

0.6

0.5 0.4 0.3 0.2 0.1

0 0.7

0.8

0.9

1.0

Possible One dimensional NCL

0.6

0.0 -3 10

Al-8Cu (DEM) Al-8Cu (Exp) Al-8Cu (Drezet) Al-15Cu (DEM) Al-15Cu (Exp) Al-15Cu (Drezet) -2

10

Volume Fraction of Solid [-]

-1

10

0

10

1

10

2

10

Stress [MPa]

Fig. 7 Comparisons of a compressive stress–volumetric solid fraction relationship and b interstitial liquid (void) ratio—compressive stress in logarithmic scale of semi-solid Al–8Cu and Al–15Cu alloys among drained die compression tests in this study (Exp), experimental measurements by Drezet et al. [22], and analogs DEM simulations

• The stress–strain curve of constant strain rate compression simulation on two cubic rigid blocks is mainly affected by the viscous damping component in the Maxwell part of Burgers contact model, and the contact force rises slower while compressing on two virtual spheres than on two cubic rigid blocks even when the same contact model parameters are used. • While visualizing the simulated deformation microstructure, the gradient of grain displacement along the z-axis from downward to upward along the top-to-bottom direction can be found. Meanwhile, the macroscale DEM simulations show discrete behavior, such as uneven instantaneous velocity magnitude and contact force chain. • The stress resistance of globularized and partially remelted Al–8Cu at the early stage of drained die compression is lower than in the past study using partially solidified Al–8Cu, suggesting the effect of primary aAl morphology on force transmissions. On the other hand, the higher stress resistance of globularized and partially remelted Al–15Cu in this study than in literature at the early stage of deformation may be attributed to the strain rate effect, resulting in some contribution of resistance from liquid pressure. The “interstitial liquid (void) ratio vs. compressive stress shown in the logarithmic scale” diagram shows that all drained die compression experiments tend to converge into a straight one-dimensional normal compression line at the later stage of compression.

Acknowledgements The authors acknowledge the financial support of the National Science and Technology Council (MOST 109-2222-E-002-005-MY3), technical support from the Instrumentation Center at NTU for EPMA experiments (MOST 110-2731-M-002-001, EPMA000300) and the Instrumentation Center at NTHU for ICP experiments (MOST 110-2731-M-007-001, ICP000200).

References 1. Otarawanna S, Laukli HI, Gourlay CM, Dahle AK (2010) Feeding mechanisms in high-pressure die castings. Metall. Mater. Trans. A 41A(7):1836-1846 2. Gourlay CM, Dahle AK (2007) Dilatant shear bands in solidifying metals. Nature 445:70-73 3. Meylan B, Terzi S, Gourlay CM, Dahle AK (2011) Dilatancy and rheology at 0-60% solid during equiaxed solidification. Acta. Mater. 59(8):3091-3101 4. Yuan L, O'Sullivan C, Gourlay CM (2012) Exploring dendrite coherency with the discrete element method. Acta. Mater. 60 (3):1334-1345 5. Kareh KM, Lee PD, Atwood RC, Connolley T, Gourlay CM (2014) Revealing the micromechanisms behind semi-solid metal deformation with time-resolved X-ray tomography. Nat. Commun. 5(4464):1-7 6. Gourlay CM, Meylan B, Dahle AK (2008) Shear mechanisms at 0-50% solid during equiaxed dendritic solidification of an AZ91 magnesium alloy. Acta. Mater. 56(14):3403-3413 7. Gourlay CM, Laukli HI, Dahle AK (2007) Defect band characteristics in Mg-Al and Al-Si high-pressure die castings. Metall. Mater. Trans. A 38A(8):1833-1844 8. Otarawanna S, Gourlay CM, Laukli HI, Dahle AK (2009) The thickness of defect bands in high-pressure die castings. Mater. Charact. 60(12):1432-1441 9. Gras C, Meredith M, Hunt JD (2005) Microdefects formation during the twin-roll casting of Al-Mg-Mn aluminium alloys. J. Mater. Process. Technol. 167(1):62-72 10. Flemings, MC (1974) Solidification Processing. McGraw-Hill, New York 11. Chen CP, Tsao CYA (1997) Semi-solid deformation of non-dendritic structures. 1. Phenomenological behavior. Acta. Mater. 45(5):1955–1968 12. Kim MS, Kim SH, Kim HW (2018) Deformation-induced center segregation in twin-roll cast high-Mg Al–Mg strips. Scripta. Mater. 152:69-73 13. Su TC, O'Sullivan C, Nagira T, Yasuda H, Gourlay CM (2019) Semi-solid deformation of Al-Cu alloys: a quantitative comparison between real-time imaging and coupled LBM-DEM simulations. Acta. Mater. 163:208-225

Exploring Semi-solid Deformation of Al–Cu Alloys by a Quantitative Comparison … 14. Su TC, O'Sullivan C, Yasuda H, Gourlay CM (2020) Rheological transitions in semi-solid alloys: in-situ imaging and LBM-DEM simulations. Acta. Mater. 191:24-42 15. Cai B, Lee PD, Karagadde S, Marrow TJ, Connolley T (2016) Time-resolved synchrotron tomographic quantification of deformation during indentation of an equiaxed semi-solid granular alloy. Acta. Mater. 105:338-346 16. Schofield, A, Wroth, P (1968) Critical state soil mechanics. McGraw-Hill, New York 17. Altuhafi FN, O’Sullivan C, Sammonds P, Su TC, Gourlay CM (2021) Triaxial compression on semi-solid alloys. Metall. Mater. Trans. A 52:2010–2023 18. Sistaninia M, Terzi S, Phillion AB, Drezet JM, Rappaz M (2013) 3-D granular modeling and in situ x-ray tomographic imaging: a comparative study of hot tearing formation and semi-solid deformation in Al-Cu alloys. Acta. Mater. 61(10):3831-3841 19. O'Sullivan, C (2011) Particulate discrete element modelling: a geomechanics perspective. Spon, London 20. Cundall PA, Strack ODL (1979) A discrete numerical model for granular assemblies. Géotechnique 29(1):47-65 21. Drezet J, Ludwig O, M Hamdi M, Fjaer H, Martin CL (2004) FEM modeling of the compressibility of partially solidified Al-Cu alloys: comparison with a drained compression test. In: Tabereaux, AT (ed) Light Metals 2004. The Minerals, Metals & Materials Society, Charlotte, North Carolina, p 655–660 22. Ludwig O, J.M. Drezet, Martin CL, Suery M (2005) Rheological behavior of Al-Cu alloys during solidification: constitutive modeling, experimental identification, and numerical study. Metall. Mater. Trans. A 36A(6):1525-1535

567

23. Nguyen TG, Favier D, Suery M (1994) Theoretical and experimental study of the isothermal mechanical-behavior of alloys in the semisolid state. Int. J. Plasticity 10(6):663-693 24. PFC-Particle Flow Code 6.0 (2018) http://docs.itascacg.com/ pfc600/pfc/docproject/index.html. Itasca Consulting Group, Inc., Minneapolis, USA 25. Lommen S, Mohajeri M, Lodewijks G, Schott D (2019) DEM particle upscaling for large-scale bulk handling equipment and material interaction. Powder Technol. 352:273-282 26. Ganesan S, Poirier DR (1987) Densities of aluminum-rich aluminum-copper alloys during solidification. Metall. Trans. A 18(4):721-723 27. Braccini M, Martin CL, Tourabi A, Brechet Y, Suery M (2022) Low shear rate behavior at high solid fractions of partially solidified Al-8 wt.% Cu alloys. Mater. Sci. Eng. A 337(1–2):1–11 28. Eliáš J (2014) Simulation of railway ballast using crushable polyhedral particles. Powder Technol. 264:458-465 29. Jiang MJ, Konrad JM, Leroueil S (2003) An efficient technique for generating homogeneous specimens for DEM studies. Comput. Geotech. 30(7):579-597 30. Dantzig JA, Rappaz, M (2009) Solidification. EPFL Press, London 31. Lifshitz IM, Slyozov VV (1961) The kinetics of precipitation from supersaturated solid solutions. J. Phys. Chem. 191):35-50 32. Rappaz M, Jacot A, Boettinger WJ (2003) Last-stage solidification of alloys: theoretical model of dendrite-arm and grain coalescence. Metall. Mater. Trans. A 34(3):467-479 33. Li J, Wu X, Cao L, Liao B, Wang Y, Liu Q (2021) Hot deformation and dynamic recrystallization in Al-Mg-Si alloy. Mater. Charact. 173:110976 34. Dennis J, Bate PS, Humphreys FJ (2009) Abnormal grain growth in Al–3.5Cu. Acta. Mater. 57(15):4539–4547

The Role of Through-Thickness Variation of Texture and Grain Size on Bending Ductility of Al–Mg–Si Profiles P. Goik, A. Schiffl, H. W. Höppel, and M. Göken

Abstract

Introduction

Requiring a high strength and concurrently a high ductility in materials is generally a demand for opposing properties in dislocation slip deforming materials, such as Al–Mg–Si wrought alloys. However, these are essential mechanical properties for safety parts in the mobility sector. While the strength of Al–Mg–Si wrought alloys is mainly governed by the state and density of the secondary precipitates, the deformation behavior and ductility are affected by both precipitates and crystallographic texture. The deformation during extrusion leads to the formation of characteristic textures in the bulk, which are distinct to a plane-strain deformation, and a peripheral coarse grain (PCG) layer beneath the surface. This PCG layer can have a detrimental effect on the bending ductility, which assesses the crashworthiness. However, an appropriate texture in the bulk can counteract the detrimental effect of PCG and increases the bending ductility at high strengths. Subsequently, based on EBSD investigations of bending deformed microstructures, a way to enhance bending deformation capability in Al–Mg–Si profiles is proposed. Keywords















AA-6xxx Extrusion EBSD Crystallographic texture Peripheral coarse grain Plate bending Ductility

P. Goik (&)
H. W. Höppel
M. Göken Department of Materials Science and Engineering, Friedrich-Alexander-Universität Erlangen-Nürnberg, Institute I: General Materials Properties, Erlangen, Germany e-mail: [email protected] A. Schiffl Hammerer Aluminium Industries Extrusion GmbH, Ranshofen, Austria

Profiles from Aluminum alloys are used in mobility applications such as cars, busses, and trains as structural parts for safety components. For these applications, alloys must exhibit a high strength to withstand high forces in case of crash, as well as to utilize maximum lightweight potential. Additionally, to ensure maximum energy absorption in case of crash, safety components must exhibit a ductile deformation behavior [1]. However, strength and ductility are competing mechanical properties, so that for technological applications a best compromise needs to be found [2]. The class of AA-6xxx or Al–Mg–Si describes heat-treatable alloys in which artificial aging leads to precipitation of a sequence of different metastable Mg-Si phases that lead to an increase in strength and hardness, depending on the precipitation state [3, 4]. The characteristic development of strength over aging duration at the beginning consists of an increase of strength during precipitation formation. In this under-aged state, dislocations cut through coherent b”-MgSi. At peak-aging, maximum strength is achieved which decreases with further aging time as precipitates undergo Ostwald ripening, marking the overaged state, where dislocations bow around precipitates by the Orowan mechanism. However, ductility behaves contrary to strength during aging. Consequently, when a structure requires high strength and ductility, a tradeoff between these material parameters is needed [3]. Ductile deformation is further affected by large primary phases (constituent particles) >1 µm at which cracks initiate by braking and decohesion of the matrix under load [5, 6]. In technological Aluminum alloys such constituent particles form intermetallic phases of Al-X-Si and Al-X, where X denotes one or more of the alloying elements Fe, Mn, Cr, Zr, V, and other metallic elements [7–10]. Among these elements Fe is exceptional with a very low solubility beneath solidus temperature of < 0.05 wt.-% [9], thus bound to precipitate in large AlFeSi-phases during casting. Initially

© The Minerals, Metals & Materials Society 2023 S. Broek (ed.), Light Metals 2023, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-031-22532-1_77

568

The Role of Through-Thickness Variation …

large constituent particles (11) and often makes recycling a colossal challenge. The Canadian bauxite residue used in this work consists of Fe (21.4%), Al (10.2%), Na (6.9%), Ca (2.3%), Si (5.0%), and Ti (3.7%) in the form of hematite (Fe2O3), boehmite (Al(OOH)), gibbsite (Al (OH)3), sodium aluminum silicate hydrate (Na8Al6Si6O24(OH)2(H2O)2), calcium carbonate (CaCO3), quartz (SiO2), and rutile (TiO2) phases. The chemical analysis and XRD spectrum of the bauxite residue feed sample are shown in Fig. 1a and b. The pH value of the bulk bauxite residue sample was determined as 10.5, reflecting its high basic

Recovery of Value Added Products from Bauxite Residue

843

Fig. 1 Feed characterization: a Chemical analysis, b XRD, c SEM (25,000 magnification), d SEM-EDS (elemental mapping)

nature. The SEM analysis and EDS elemental mapping of the feed sample shown in Fig. 1c and d depict the presence of sub-micron sized particles with a complex association of different elements in the bulk sample. Mild acid washing was carried out to separate alkali fractions from bauxite residue.

Bauxite Residue Neutralization Direct leaching of bauxite residue in acid solution requires an additional reagent because a part of it is consumed to neutralize the alkali left behind from the Bayer process. Neutralizing bauxite residue with HCl reduces the acid demand during the second stage of leaching with oxalic acid. Oxalic acid is relatively more expensive than mineral acids; therefore, neutralization with other mineral acids reduces the overall reagent cost. HCl and H2SO4 were selected as mineral acids to neutralize bauxite residue. The dissolution value of different elements using HCl and H2SO4 during neutralization is shown in Fig. 2a and b. It was found that Na removal depends upon the concentration of H+ ions in the solution; therefore, 1 M HCl solution resulted in a similar result as with 0.5 M H2SO4. The bauxite residue slurry pH was reduced from *10.5 to *2.5, along with the dissolution of approx. 38% solid mass with HCl and 27% with

H2SO4. Both the acids completely dissociated the sodium aluminum silicate phase at the optimized conditions and separated more than 90% Na, 40–45% Al, 60% Si and less than 10% Fe. The key difference between the two acids is the limited separation of Ca with H2SO4 due to the formation of insoluble CaSO4. The leach solution from both HCl and H2SO4 wash consists of a high amount of Al (*4500 ppm), along with Si (*4400 ppm), Na (4000 ppm), and Ca (500– 3000 ppm). The pH adjustment of the leach solution can be employed to recover silica, alumina, and calcium values, and the remaining solution can be recirculated for acid washing. Considering the high recovery of Na and Ca with HCl, neutralization experiments were performed with HCl. The neutralized bauxite residue was further subjected to leaching with oxalic acid to recover Fe and V values.

Production of Magnetite from Neutralized Residue The high chelating ability of oxalic acid provides a high affinity for forming an aqueous complex with metal species such as Fe and V [20]. Oxalic acid is a diprotic acid and dissociates into HC2O4− and C2O42− with a pKa of 5.6  10–2 and 6.2  10–5, respectively. The ionization depends on the solution pH, with C2O4− being the dominant

844

H. Tanvar and B. Mishra

Fig. 2 Dissolution of different elements during neutralization with a sulfuric acid, b hydrochloric acid

species above pH 4 and HC2O4− in the acidic range [21]. The oxalate ion (C2O42−, HC2O4−) have (two, one) oxygen atoms with unshared electrons and can form a coordination bond with a metal ion (such as Fe and V) and form a ring, resulting in a complex metal ion species. The chemical reaction for the leaching of Fe and V oxide with oxalic acid is shown in Eqs. (1)–(4). Based on our previous work and literature findings, the activation energy for the dissolution of hematite using oxalic acid was determined as 100– 150 kJ/mol in the temperature range of 65 to 95 °C [19, 22, 23]. The leaching kinetics of hematite in oxalic acid is highly influenced by the temperature. The temperature and oxalic acid concentration during leaching was fixed at 95 °C and 2 M, respectively, whereas the leaching time and solid to liquid ratio were varied as shown in Fig. 3. The dissolution value of key elements and concentration of resulting leach liquor is shown in Fig. 3a and b, respectively. High Fe leaching was obtained at a low solid to liquid ratio and

extended leaching duration, where approximately 85.4% Fe, 77.9% V, 16.9% Sc, 4.2% Ti, 12.7% Si, and 48.5% Al was dissolved after 2.5 h using 10% pulp density. The ferric oxalate leach liquor was further subjected to photochemical reduction using UV light to precipitate and recover Fe as ferrous oxalate. Ferric oxalate ([Fe(C2O4)3]3−) is a photochemically active complex that undergoes spontaneous photochemical reduction to form ferrous oxalate [24, 25]. The reduction mechanism proceeds through the formation of carbon dioxide (CO2•−) and oxalate (C2O4•−) radical anion [24, 25]. Electron transfer from ligand (oxalate) to metal (Fe3+) produces a C2O4•− anion in the solution, which further dissociates into CO2 and CO2•−. The CO2•− can further reduce another Fe3+ ion or combine with CO2•− anion to form C2O42− in the solution. The Fe(II) complex is further hydrolyzed and precipitated as ferrous oxalate (FeC2O4
2H2O). The photochemical reduction was performed at an optimized duration of 6 h using a 100 W UV

Fig. 3 a Dissolution of different elements during leaching with oxalic acid, b concentration of solution at different oxalic acid leaching conditions

Recovery of Value Added Products from Bauxite Residue

845

Fig. 4 Characterization of magnetite product: a XRD and chemical analysis, b SEM micrograph

lamp [19]. After photochemical reduction, more than 99% of the Fe in the solution was recovered as ferrous oxalate precipitate, whereas V was retained in the solution. The filtered solution consists of approx. 58.7 mg/L V which can be recovered through solvent extraction using Cyanex 923, and Aliquat 336 [26]. The ferrous oxalate precipitate was decomposed at a high temperature to produce magnetite [27]. Thermal decomposition was performed in a low oxygen atmosphere at 500 °C for 1 h. A low oxygen atmosphere was maintained using a continuous flow of N2 in the tube furnace to restrict the oxidation of Fe+2 to Fe+3. The XRD spectrum, chemical analysis, and SEM micrograph of the magnetite product are shown in Fig. 4a and b. The magnetite product is characterized by more than 98% purity, with magnetite being the dominant mineral phase. The SEM analysis depicts the presence of porous particles of variable geometry. High purity magnetite possesses application in magnetic resonance imaging, targeted drug delivery systems, photo magnetics, and black pigment material.
3 þ Fe2 O3ðsÞ þ 6C2 O2 4ðaqÞ þ 6HðaqÞ ! 2 FeðC2 O4 Þ3 ðaqÞ   þ 3H2 OðlÞ DG25 C ¼ 212:75 kJ=mol

ð1Þ

þ Fe2 O3ðsÞ þ 2HC2 O 4ðaqÞ þ 6HðaqÞ    ð2Þ þ ! 2½FeHC2 O4 2ðaq Þ þ 3H2 OðlÞ DG25 C ¼ 87:35 kJ=mol


2 þ V2 O5ðsÞ þ 5HC2 O 4ðaqÞ þ HðaqÞ ! 2 VOðC2 O4 Þ2 ðaqÞ   þ 3H2 OðlÞ þ 2CO2ðgÞ DG25 C ¼ 326:05 kJ=mol

ð3Þ


2 þ V2 O5ðsÞ þ 5C2 O2 4ðaqÞ þ 6HðaqÞ ! 2 VOðC2 O4 Þ2 ðaqÞ   þ 3H2 OðlÞ þ 2CO2ðgÞ DG25 C ¼ 448:05 kJ=mol

ð4Þ

Sulfation Baking for Titanium and Scandium Recovery The leach residue obtained after leaching with oxalic acid constitutes about 27% of starting bauxite residue weight and consists of approx. 10.3% Si, 15.1% Ti, 8.9% Fe, 13.8% Al, and 0.0048% Sc. It is worth mentioning that Ti and Sc were concentrated in the residue up to fourfold after the recovery of major elements. Separation of Fe before an attempt to recover Sc and Ti presents better selectivity, reduced acid demand, and easy separation in downstream processing. The oxalic leaching residue was thoroughly mixed with different amounts of sulfuric acid, followed by baking at 300 °C for 1 h. The baking temperature was selected based on the decomposition behavior of sulfuric acid at high temperatures and its reactivity [28]. Sulfuric acid decomposes to gaseous sulfuric acid, SO3, and H2O in the temperature range of 127–427 °C. The second stage of decomposition proceeds with endothermic reduction of SO3 to SO2 and is observed at a temperature of more than 750 °C [29]. The main advantage of sulfation baking over direct leaching includes reduced acid demand, fast reaction kinetics, and a high reaction rate. The dissolution of different elements at different acid to residue ratio is shown in Fig. 5a. With an increase in acid dosage from 0.5 mL/g to 1 mL/g, the dissolution of all elements increased. The highest dissolution of Ti, Al, and REEs was obtained at acid to residue ratio of 1.3 mL/g, resulting in the dissolution of 42.7% Si, 90.2% Ti, 87.3% Fe, 93.2% Al, and 64.5% RE. The dissolution of RE was limited to 65% even at the high dose of acid, therefore, 1.3 g/mL was considered optimal. The leach liquor obtained after sulfation baking and water leaching consist of 12.31 g/L Ti, 7.21 g/L Al, and 7.19 g/L Fe as major components. The recovery of Ti was carried out

846

H. Tanvar and B. Mishra

Fig. 5 a Dissolution of different elements during sulfation baking and water leaching at different sulfuric acid to residue ratio (300°, 1 h), b SEM micrograph and sample photograph of titanium dioxide product

by thermal hydrolysis by boiling the solution at 100 °C for 4 h [30]. Fe(III) was first reduced to Fe(II) with iron powder before thermal hydrolysis to prevent coprecipitation of Fe. The filtered solution after thermal hydrolysis contains 55.10 ppm Ti, 9080.00 ppm Fe, 7190.20 ppm Al, and 4.1 ppm Sc. At the same time, more than 96% of the Ti was recovered as TiO2 precipitate. The final solution can be further processed through solvent extraction to recover the rare earth concentrate of Sc. The SEM micrograph and sample photographs of the TiO2 product are shown in Fig. 5b. The purity of TiO2 precipitate was determined as more than 99%, with minor impurities of Fe and Al.

Comprehensive Flowsheet The summarized process flow diagram for the recovery of different products from bauxite residue is shown in Fig. 6. The proposed process consists of three-stage processing to selectively recover magnetite, alumina, and titanium dioxide as major products with the high market value from bauxite residue. The critical elements, including Sc and V, are retrieved in the liquid stream after the extraction of major elements and can be further recovered through solvent extraction. Hydrochloric acid, oxalic acid, and sulfuric acid are utilized as chemical reagents for the dissolution of (Al, Si, Ca), (Fe, V), and (Ti, Sc), respectively. Whereas precipitation from solution, photochemical reduction, and hydrolysis are adopted to recover the dissolved species selectively. The following flowsheet provides high recovery and purity of final products. Based on the material balance, the processing of one ton of bauxite residue will result in the

production of 254 kg magnetite, 53 kg alumina, and 62 kg titanium dioxide. Critical elements including Sc and V are recovered into the solution containing 58.7 ppm V and 4.1 ppm Sc generating 1200.0 g V2O5 and 10.9 g Sc2O3, respectively.

Conclusions Based on the research work carried out in this study, a comprehensive flowsheet for recovering valuable products from bauxite residue is presented. The process includes neutralizing bauxite residue with HCl to separate alkali fractions and obtain neutralized bauxite residue. The leach liquor from the neutralization stage is further processed to recover alumina, silica, and calcium values. The neutralized solid is leached with oxalic acid to selectively dissolve Fe and V. Neutralization of bauxite residue before oxalic acid leaching reduces the acid consumption and provides a high purity product. The oxalic leaching liquor is subsequently reduced with UV light to precipitate ferrous oxalate, further decomposed to magnetite in an inert atmosphere. The residue from leaching with oxalic acid is processed through sulfation baking, water leaching, and hydrolysis process to recover Ti as TiO2 and generate a liquid stream with dissolved rare earth elements. The major constituents of bauxite residue are recovered in the form of magnetite (98% purity, 95% recovery), alumina (98% purity, 31% recovery), and titania (98% purity, 87% recovery). Based on the experimental work and theoretical analysis, the proposed process results in near-zero waste discharge and presents an excellent opportunity for industrial-scale processing of bauxite residue.

Recovery of Value Added Products from Bauxite Residue

847

Fig. 6 Summarized process flow diagram for recovery of different products from bauxite residue

Acknowledgements The authors are thankful to Mr. Glenn Yee for the fellowship he instituted at the Worcester Polytechnic Institute. Thanks are due to the NSF Center for Resource Recovery and Recycling with their technical support through Global Minerals Recovery, LLC.

References 1. Habashi, F. (2016) A Hundred Years of the Bayer Process for Alumina Production, in Essential Readings in Light Metals. 85– 93. 2. Healy, S. (2022) Sustainable Bauxite Residue Management Guidance, S. Healy, Editor., International Aluminum Institute. p. 92. 3. Tsakiridis, P.E., S. Agatzini-Leonardou, and P. Oustadakis (2004) Red mud addition in the raw meal for the production of Portland cement clinker. J Hazard Mater 116(1–2) 103–110. 4. Ruyters, S., et al. (2011) The red mud accident in ajka (hungary): plant toxicity and trace metal bioavailability in red mud contaminated soil. Environ Sci Technol 45(4) 1616–1622. 5. Hammond, K., et al. (2013) CR3 Communication: Red Mud—A Resource or a Waste? Jom 65(3) 340–341. 6. Evans, K. (2016) The History, Challenges, and New Developments in the Management and Use of Bauxite Residue. Journal of Sustainable Metallurgy 2(4) 316–331.

7. Borra, C.R., et al. (2016) Recovery of Rare Earths and Other Valuable Metals From Bauxite Residue (Red Mud): A Review. Journal of Sustainable Metallurgy 2(4) 365–386. 8. Khairul, M.A., J. Zanganeh, and B. Moghtaderi (2019) The composition, recycling and utilisation of Bayer red mud. Resources, Conservation and Recycling 141 483–498. 9. Borra, C.R., et al. (2015) Smelting of Bauxite Residue (Red Mud) in View of Iron and Selective Rare Earths Recovery. Journal of Sustainable Metallurgy 2(1) 28–37. 10. Cardenia, C., E. Balomenos, and D. Panias (2018) Iron Recovery from Bauxite Residue Through Reductive Roasting and Wet Magnetic Separation. Journal of Sustainable Metallurgy 5(1) 9– 19. 11. Mishra, B., A. Staley, and D. Kirkpatrick (2002) Recovery of value-added products from red mud. Mining Metallurgy and Exploration 19 87–94. 12. Archambo, M.S. and S.K. Kawatra (2020) Utilization of Bauxite Residue: Recovering Iron Values Using the Iron Nugget Process. Mineral Processing and Extractive Metallurgy Review 42(4) 222– 230. 13. Reid, S., et al. (2017) Technospheric Mining of Rare Earth Elements from Bauxite Residue (Red Mud): Process Optimization, Kinetic Investigation, and Microwave Pretreatment. Sci Rep 7(1) 15252. 14. Ding, W., et al. (2022) Efficient Selective Extraction of Scandium from Red Mud. Mineral Processing and Extractive Metallurgy Review 1–9.

848 15. Ujaczki, É., et al. (2019) Recovery of Gallium from Bauxite Residue Using Combined Oxalic Acid Leaching with Adsorption onto Zeolite HY. Journal of Sustainable Metallurgy 5(2) 262–274. 16. Akcil, A., et al. (2017) Overview On Extraction and Separation of Rare Earth Elements from Red Mud: Focus on Scandium. Mineral Processing and Extractive Metallurgy Review 39(3) 145–151. 17. Borra, C.R., et al. (2015) Leaching of rare earths from bauxite residue (red mud). Minerals Engineering 76 20-27. 18. Agrawal, S. and N. Dhawan (2021) Investigation of mechanical and thermal activation on metal extraction from red mud. Sustainable Materials and Technologies 27. 19. Tanvar, H. and B. Mishra (2021) Hydrometallurgical Recycling of Red Mud to Produce Materials for Industrial Applications: Alkali Separation, Iron Leaching and Extraction. Metallurgical and Materials Transactions B 52 3543–3557. 20. Taxiarchou, M., et al. (1997) Dissolution of hematite in acidic oxalate solutions. Hydrometallurgy 44(3) 287–299. 21. Panias, D., et al. (1996) Thermodynamic analysis of the reactions of iron oxides: Dissolution in oxalic acid. Canadian Metallurgical Quarterly 35(4) 363–373. 22. Lee, S.O., et al. (2006) Study on the kinetics of iron oxide leaching by oxalic acid. International Journal of Mineral Processing 80(2–4) 144–152.

H. Tanvar and B. Mishra 23. Salmimies, R., et al. (2012) Acidic dissolution of hematite: Kinetic and thermodynamic investigations with oxalic acid. International Journal of Mineral Processing 110–111 121–125. 24. Mangiante, D.M., et al. (2017) Mechanism of Ferric Oxalate Photolysis. ACS Earth and Space Chemistry 1(5) 270–276. 25. Ogi, Y., et al. (2015) Ultraviolet photochemical reaction of [Fe (III)(C2O4)3]3− in aqueous solutions studied by femtosecond time-resolved X-ray absorption spectroscopy using an X-ray free electron laser. Structural Dynamics 2(3). 26. Liu, Z., et al. (2020) Separation and recovery of vanadium and aluminum from oxalic acid leachate of shale by solvent extraction with Aliquat 336. Separation and Purification Technology 249. 27. Angermann, A. and J. Töpfer (2008) Synthesis of magnetite nanoparticles by thermal decomposition of ferrous oxalate dihydrate. Journal of Materials Science 43(15) 5123–5130. 28. Narayanan, R.P., N.K. Kazantzis, and M.H. Emmert (2017) Selective Process Steps for the Recovery of Scandium from Jamaican Bauxite Residue (Red Mud). ACS Sustainable Chemistry & Engineering 6(1) 1478–1488. 29. Schwartz, D., et al. (2000) A Kinetic Study of the Decomposition of Spent Sulfuric Acids at High Temperature. Industrial & Engineering Chemistry Research 39(7) 2183–2189. 30. Grzmil, B.U., D. Grela, and B. Kic (2008) Hydrolysis of titanium sulphate compounds. Chemical Papers 62(1) 18–25.

Current Status and Proposed Economic Incentives for Higher Utilization of Bauxite Residue to Enhance Sustainability of the Aluminum Industry Subodh K. Das and Muntasir Shahabuddin

Abstract

Introduction

The three-stage production chain of aluminum from ore to casting is complex and consumes massive amounts of energy and input materials. The process also results in undesirable by-products, such as bauxite tailings and bauxite residues or red mud. From a holistic perspective, the overall sustainability of the aluminum industry is dependent on several factors that go beyond the recent focus on the low carbon drive across the companies and regions. Globally, some 3 billion tons of bauxite residue (red mud) are now stored in massive waste ponds or dried mounds, making it one of the most abundant industrial wastes on the planet. Alumina refining plants generate an additional 150 million tons each year. The industry produces 0.5–2 tons of bauxite residue per ton of aluminum. Although actively globally pursued, the current utilization of bauxite residues is less than 4%. In the author’s published analysis, mitigation and utilization of bauxite residue are the number one sustainabilityenhancing factor. The objectives of this paper are:

The utility and prevalence of aluminum in modern society belie the century-long impacts of its caustic, crimson byproduct: bauxite residue. Aluminum production begins with the Bayer process: where bauxite, a red ore containing a slew of metal oxides (including iron, aluminum, and titanium), is hydrothermally extracted in sodium hydroxide so the alumina can be selectively removed. From there, the Hall-Héroult process electrolytically reduces the alumina into aluminum metal. The remaining oxides and sodium hydroxide left from the Bayer process leave exorbitant amounts of a high-pH, low-value red slurry dubbed “bauxite residue” or “red mud”—in amounts up to twice the mass of the resulting aluminum produced [1]. The operative “low value” of bauxite residue is caused by its high causticity and iron oxide content—which are expensive to process and result in low-value products. This has led to neither valorization of bauxite residue’s component minerals nor its management as a waste product.

1. Review of bauxite residue utilization status 2. Assess the feasibility and prospects of current efforts underway 3. Suggest economic incentives to encourage higher utilization. Keywords





Aluminum Sustainability recovery Bauxite residue




Recycling and secondary

S. K. Das (&) Phinix, LLC, Clayton, MO, USA e-mail: [email protected] M. Shahabuddin Worcester Polytechnic Institute, Worcester, MA, USA

Problem Description Over 3 billion tons of bauxite residues have been amassed in holding ponds which grow by 150 million tons per year [2]. The global alumina industry is expected to have 7 to 8 billion tons of red mud inventory by 2040. These ponds take up significant real estate and require upkeep to prevent spills [3], the damage of which is illustrated by the over 120 casualties caused by drowning and chemical burns during the Timföldgyár alumina plant spill in Ajka, Hungary [4], alongside the resultant, and well-studied, destruction of terrestrial and aquatic ecosystems in its fallout [5, 6]. The most recent disaster of bauxite pond leakage happened in Hydro’s Alunorte refinery in Brazil in 2018 [7]. This phenomenon—of large bauxite residue deposits—is not geographically limited. Alumina refineries exist across

© The Minerals, Metals & Materials Society 2023 S. Broek (ed.), Light Metals 2023, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-031-22532-1_112

849

850

nearly every continent, with at least 200,000 tons of alumina produced on each one. Their prodigious output of residues necessitates their disposal in accompanying red mud “ponds” and dry stacks, each of which comes with huge maintenance costs and the potential risks of their failure: which can be catastrophic displays, as seen in Ajka, or smaller leakages, as seen in Hydro’s Alunorte refinery in Brazil. In either case, the destruction caused by improper waste management has led to quantifiable economic damage and irreparable human impact. Ultimately, mitigating effective disposal and management are key to aluminum’s sustainable future.

Current Efforts Historically, the management of bauxite tailings was completely limited to storage and discarding. The major methods for bauxite residue disposal differed only in their location and solids loading. 1. Aquatic disposal, for example, forgoes the land required to house the tailings, but causes destructive changes in pH and dissolved minerals in aquatic ecosystems. Regulatory pressure has nearly eliminated this option. 2. Land-based disposal has thus become the most popular option, where waste solids content governs cost-benefit analyses. Lower solids content “lagooning” (1700 °C, softening temperature >1450 °C under loading, compression strength of >90 MPa and heating

951

shock of 25 times from 1100 °C to room temperature. In addition, the bricks with disperse 3D micropores has a special feature, that is, degree of porosity of >30%, and gas can go through with disperse micropores out of the bricks and the small bubbles were formed in Al liquid. The porous plug is not wetting with aluminum liquid and has a large wetting angle; The porous plug can prohibit aluminum liquid through (not easily plugging). During the porous plug material manufactured, the nanoparticles with rare earth elements were added into size to improve bonding strength of interfaces and granules, so that the loose/porous bricks first were obtained. The bricks were put into high-temperature furnace with N2 atmosphere, and then SiC will convert into SixNy in the sintering processing. The bricks with dispersal porous pores and no leaking and plugging for aluminum liquids were made. Based on the result from the flow simulation, the construction of a furnace for melt aluminum alloys were designed as shown in Fig. 1a–c. The dispersal porous bricks (see Fig. 1) were put on bottom of the furnace, which formed a porous plug bed and the inert gas (Ar) can be blown through the porous plugs made by the bricks into the aluminum liquids. Programmable Logic Controller (PLC) was used to control volume and pressure of the gas as

Fig. 1 Constructions of a furnace and porous plugs for melt aluminum alloys

952

K. Ke et al.

well as the change of frequency and amplitude of pulses. The technology without chemical resolvent not only can decrease difference of temperatures in different locations in the furnace but also can make distribution of uniform alloying elements in melt furnace by heat and mass transfer in 3D convection.

Evaluation of Al Melt Quality Measurement of Hydrogen Content The HYCAL Hydrogen Determinator from EMC for on-line evaluation of hydrogen content during ingot casting trials was used. The trials were carried out by the FMT team at FMT’s cast house on a melt furnace before and after degassing as well as after the FCC filter while running alloy AA5052. Content of hydrogen can finally be controlled range of 0.2–0.28 ml/(100 gAl) in the ingots.

Analysis and Determination of Inclusions in Aluminum Liquid Because of the customer requested, melt trials were carried out by using the Prefil Footprinter of development from FMT for inclusion analysis. The metal flow distributions for casting drops from the processing without chemical resolvent were tested. All PODFA samples were analyzed at the lab of FTM. The Inclusions in Al liquid were at the level of less 0.282 mm2/kg Al. Aluminum oxide films are thin and at a level of *No. of 60.

Homogeneity of Temperature Field in Furnace Temperatures at 11 locations on the 100 mm depth to the surface of Al liquid and the 100 mm height to bottom of the furnace with 50-ton capacity were monitored. The locations in the furnace were set as shown in Fig. 2a. Measuring time kept at 3 min for each measurement. The measuring result

Homogeneity of Composition of Aluminum Alloy Melts Optical Emission Spectrometry (SPECTRO MMAXx Optical Emission Spectrometer) was used to determine approximate chemistry of samples taken at each location in the melt furnace according to ASTM 1251-11 Standard Test Method for Analysis of Aluminum and Aluminum Alloys by Spark Atomic Emission Spectrometry. Composition at 11 locations on the 100 mm depth on the surface of Al liquid and the 100 mm height to bottom of the furnace with 50-tons capacity were also determined. The measuring result was listed below Table 1. The liquid of aluminum alloys has perfect homogeneity of the composition. In addition, the measuring result of homogeneity of composition in casting start, medium, and ending were listed in Table 2. The good stability of chemical composition can be observed during casting.

Main Characteristic of the Green Project The melting process of no chemical resolvent can avoid some residual harmfully chemical substance in dross of aluminum alloys. The door of a furnace often opened & closed did not need in melting processing because of no spraying and adding the resolvent as well as skimming operation. Heat lost remarkably reduced in duration of melt operation. Each porous plug has an independent sub-system with pipes which connect to Ar gas station. The system was constituted by PLC in control cabinets and can provide independently intelligential parameters for each porous plug. The bubbling plug bed formed with all porous plugs and intensive bubbles (see Fig. 3) would catch hydrogen & inclusions and float upward based on the third law

Exit

Plug

Door (a)

100mm to surface of liquid 100mm to bottom of furnace

Temperature, C°

Fig. 2 Locations measured in the furnace in (a); measuring result in (b)

was shown in Fig. 2b. The difference in temperatures at the locations is less than 2.0 °C.

Location

(b)

A Novel Green Melt Technology for Aluminum Alloys

953

Table 1 compositions of different locations in the furnace (wt.%) Location, top

Si

Fe

Cu

Mn

Mg

Cr

Ti

Location, bottom

Si

Fe

Cu

Mn

Mg

Cr

Ti

1

0.080

0.234

0.018

0.065

2.51

0.188

0.015

1

0.079

0.238

0.018

0.065

2.50

0.188

0.0102

2

0.082

0.241

0.018

0.066

2.53

0.187

0.015

2

0.081

0.244

0.0182

0.065

2.48

0.188

0.0101

3

0.082

0.237

0.018

0.065

2.52

0.189

0.015

3

0.081

0.239

0.0181

0.065

2.51

0.189

0.099

4

0.080

0.241

0.018

0.065

2.54

0.188

0.015

4

0.080

0.24

0.0183

0.065

2.53

0.186

0.0103

5

0.081

0.239

0.018

0.065

2.52

0.188

0.015

5

0.081

0.235

0.0182

0.065

2.51

0.187

0.0099

6

0.079

0.236

0.018

0.065

2.49

0.189

0.015

6

0.080

0.239

0.0184

0.065

2.48

0.189

0.0103

7

0.080

0.234

0.018

0.065

2.51

0.188

0.015

7

0.079

0.235

0.0182

0.065

2.50

0.188

0.0102

8

0.080

0.243

0.018

0.065

2.54

0.189

0.015

8

0.081

0.241

0.0183

0.065

2.53

0.189

0.0101

9

0.081

0.241

0.018

0.065

2.51

0.188

0.015

9

0.080

0.242

0.0181

0.065

2.50

0.187

0.01

10

0.081

0.237

0.018

0.065

2.49

0.187

0.015

10

0.081

0.235

0.0182

0.065

2.51

0.188

0.00990

11

0.081

0.242

0.018

0.065

2.48

0.189

0.015

11

0.080

0.241

0.0181

0.065

2.47

0.187

0102

12

0.079

0.239

0.018

0.065

2.51

0.187

0.015

12

0.079

0.243

0.0182

0.065

2.52

0.188

0.01

13

0.079

0.234

0.018

0.065

2.51

0.189

0.015

13

0.080

0.238

0.0182

0.065

2.50

0.189

0.0101

Table 2 Compositions and temperature starting, medium and ending casting Location

Wt.% of main alloying elements Si

Fe

Cu

Mn

Mg

Cr

Ti

Temperature of Al liquid/°C

Casting start

0.0788–0.0812

0.237–0.241

0.0180–0.0181

0.0651–0.0654

2.48–2.51

0.188–0.189

0.0146–0.0148

716–718

Medium of casting

0.0788–0.0812

0.240–0.242

0.0180–0.0181

0.0651–0.0654

2.47–2.49

0.188–0.189

0.0185–0.0187

712–713

Casting ending

0.0795–0.0805

0.241–0.243

0.0180–0.0181

0.0651–0.0654

2.45–2.47

0.186–0.188

0.0186–0.0188

701–703

Fig. 3 Porous plugs with intensive bubbles

954

(PV = constant) of thermodynamics. The process was realized and has advantages in industrial scale as follows: (a) There are no evil-smelling odors due to use Ar gas instead of Cl, F, and ammonia. Those operators don’t need to be afraid for health and environment. The cost of environmental protection will be saved. (b) Total dross after the melting processing can reduce 50%. The dross easily meets requirements of industrial exhaust in any country. The dross can be treated without toxic and harmful substance. The processing from melting start to casting end is in dynamic equilibrium state in all the time. The casting will start immediately if chemistry of aluminum alloys meets requirement of Al alloys because of no spraying powder, skimming and settling treatment, and so on. Total melt time will decrease by 1 h more. The processing can save natural gas per ton aluminum alloy than more 5m3/per cast drop in current conditions. The Al liquid included hydrogen of about 0.5 ml/ (100 gAl) before Ar gas going through the porous plugs. After treatment using the porous plug bed, the melting process can control H content in the range of 0.2–0.28 ml/ (100 gAl), from current melting process not used chemical resolvents. It was seen that the melting process without chemical resolvents can obtain a lower level of H content in aluminum liquid. The melting process shows that the dynamic equilibrium of aluminum alloy liquid under temperature and gravitational fields can obtain uniform alloying, that is, no segregation of composition. This is a foundation to produce high-quality aluminum alloy ingots. No gases including Cl and F were produced and exhausted into the atmosphere, that is, the green melting process. The bottom of a furnace was kept flat and the

K. Ke et al.

porous plugs have no plugging or slog due to heavy compounds and inclusions with bubbles to float towards to surface of aluminum liquid. Acknowledgements The authors thank the work of engineers of the Fuzhou Metal-new High Temp. Tech. Co. Ltd and support of Fuzhou University, Fujian Institute of Research on the Structure, Chalco Ruimin Co., Ltd, and Guangdong Hoshion Aluminum Co., Ltd. In addition, the authors thank support of EMC Hycal Limited in United Kingdom in hydrogen determination and MS & T Consulting in USA in PODFA measurement.

References 1. Beland, Guy, Dupuis, Claude and Riverin, Gaston. “Rotary Flux Injection: Chlorine-Free Technique for Furnace Preparation.” Light Metals (1998) pp 843–847. 2. KE Dongjie, CHEN Qun, LI Yuhang, ”Technical research on online degassing purification mechanism of molten aluminum”, Light Alloy Fabrication Technology, 2012 Vol. 40, Issue 8, p 13-22. 3. WANG Shichong, KE Dongjie, “Study on bubble shape during water simulation experiment of rotary impeller spray process”, Light Alloy Fabrication Technology, 2010,38(08) p 8-12. 4. WANG Shichong, KE Dongjie, CHEN Xiao, CHEN Qun. “Study on dynamic water simulation experiment for purification of molten aluminum alloys by rotary impeller”, Light Alloy Fabrication Technology, 2010,38(03) p 18–24. 5. KE Dongjie, LU Guimin, Richard J., “Discussion on the development of green melting technology for aluminum alloys”, Processings of the 13th National Light Alloy Processing Academic Exchange Conference,Hangzhou, Jiangsu province, P. R. China, 2005: p 7–13. 6. The Puur Gas Injection System – http://www.sparref.com/ uploadedFiles/File/Puur_System_Brochure.pdf. 7. ZL201810880452.1, inventing patent of P. R. China, “A porous plug used bottom of melt furnace with gas stirring as well as a metal melt furnace”. 8. ZL201810557056.5, inventing patent of P. R. China, “A metal melt method and furnace”. 9. NR.112018002481, inventing patent of Germany, “A metal melt method and furnace”.

MagPump Oscar A. Perez and Eishin Takahashi

Abstract

Keywords

TST is a California-based company leading the way in manufacturing aluminum ingot, billet, and slab worldwide for over 76 years. TST is committed to providing the highest quality products. TST is committed to continuous improvements in its casthouse operation and has installed the world’s first and only permanent magnet-based pump, MagPump™, from Zmag (Japan). MagPump is designed to be a direct replacement for traditional mechanical pumps in side well (multi-chamber) furnaces without requiring furnace modification. MagPump is powered by zPMC™ (Zmag Permanent Magnetic Circuit) which generates virtual impellers to pump molten aluminum. Unlike mechanical pumps, MagPump does not have physical impellers and therefore is a nearly hands-free system, with no consumable parts, and reduced downtime. MagPump is also capable of other applications such as gas injection (e.g., Chlorine, Nitrogen, Argon) and scrap submersion. Operator safety has been increased as the necessity to approach the pump well has been minimized.

Molten metal pumping innovation




Magnetic stirring




Metal pump

Introduction TST Inc. has been leading the way since 1946 in manufacturing aluminum ingot, billet, and slab worldwide with five melting/holding furnaces. TST is committed to providing high-quality products that meet or exceed customer specifications. TST has set a goal to revolutionize how the company melts aluminum. This goal is achieved by adhering to an effective quality management system based on continual improvements and always searching for the technology to improve TST’s quality production and manufacturing cost. The MagPump technology addresses many of the industry challenges that companies are presently faced with. The design and operation of the MagPump maximizes operation and productivity and does not compromise quality.

O. A. Perez (&) TST Inc, 11601 S Etiwanda Ave, Fontana, CA 92337, USA e-mail: [email protected] E. Takahashi Zmag America, Ltd., 1011 NW Glisan Street, Unit 202, Portland, OR 97209, USA © The Minerals, Metals & Materials Society 2023 S. Broek (ed.), Light Metals 2023, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-031-22532-1_127

955

956

O. A. Perez and E. Takahashi

MagPump—The Design

1. MagPump = MagPump Engine (zPMC) + MagPump Housing (refractory case) MagPump Engine is powered by zPMC. zPMC is optimized to bring the best pump performance to a side well furnace. zPMC’s magnetic fields are inherently existing (permanent) with the MagPump Engine. Since the field is permanent, there is no need to use electricity to generate magnetic fields as in the case of EMS (electromagnetic stirrer) and EMP (electromagnetic pump)—common technologies in the industry.

2. Virtual impeller using magnetic fields There is no physical impeller with MagPump. Magnetic fields are used to pump molten aluminum, and this means that MagPump is a hands-free system. 3. MagPump Housing’s pump room The pump room is big and empty for increased performance. Molten aluminum is directed into the pump room from the bottom.

MagPump

957

4. Gas injection

8. Vortex/chip submersion

Gas can be injected from the bottom or in a path close to the discharge port to promote shearing and dissolution.

MagPump is designed to work with a vortex system.

5. Air blower

MagPump Installation

A small industry-standard air blower is used to protect zPMC from molten aluminum heat. MagPump does not generate heat in its magnetic circuit (zPMC) as in the case of EMS and EMP.

Zmag technology was truly plug-and-play; it was installed in the existing mechanical pump layout without the need for any modifications. TST also used the same VFD used by mechanical pumps with minor setting adjustments only for the MagPump™ motor to operate efficiently. Once MagPump was running to meet TST’s operating standards, it was handsfree, with zero maintenance assistance after installation.

6. Controls and electricity consumption One VFD runs MagPump. The total electricity consumption of MagPump is about 10 kW.

MagPump Operation 7. No furnace modification In most cases, MagPump does not require any furnace modification. “Plug and Play.”

During the 6 weeks of trial, TST had six alloy changes. Even when alloy changes were not conducted, the bath was always nearly emptied on most casts (dry hearth or batch operation), TST cast an average of twice a day; however,

958

during the six weeks, the bath was emptied once per day due to TST’s production schedule. 1 time per day * 5 days * 6 weeks = 30 times, 2.1 Million pounds of good metal was produced. TST has a second side well furnace next to the MagPump trial furnace; this furnace is a mirror image of the trial furnace. While TST had no issues at all with MagPump, TST did have challenges with the mechanical pump in the second furnace with wear and tear along with other unplanned issues. TST lost an average of 3.5–4 h of heat and production during the replacement and inspection of mechanical pumps which then required the addition of extra resources to assist in our second furnace. This lost production had the following impact: Average 4 h downtime per event/failure  4 events = 16 lost production hours. If the average melt rate is 11,000 pounds per hour, 11,000 pounds per hour melt rate  4 h downtime per event  4 events = 176,000 pounds of lost production. This is not including lost opportunity during alloy changes which required the raising of the mechanical pumps due to the metal level being too low for the mechanical pump to operate at an optimal RPM. Extrapolating from the 6 week trial: One year has 52 weeks. During the 6 weeks, mechanical pumps failed 4 times. 52/6 weeks  4 events/failures = 35 events/failures expected per year. 11,000 pounds per hour melt rate  4 h downtime  35 events/failure per year = 1,540,000 pounds lost production. In keeping with TST’s goal of being environmentally responsible, MagPump also leads to a dramatic reduction in greenhouse gas: At an average of 2,300 BTU per pound, TST would have used/wasted 2,300 BTU per pound  1,540,000 pounds = 3,542,000,000 BTU. LB—CO2/mmBTU is 117, so 117  3,540,000,000/ 1,000,000 = 414,414 pounds or 188 tons per year of CO2 reduced with MagPump. Another benefit of MagPump is that the only inspection required was a simple check of the temperature display for MagPump’s zPMC™ (Zmag Permanent Magnetic Circuit). A two second walk through and check of the displays was enough to let TST know MagPump was performing at peak performance. Additionally, Zmag’s MagPump effectively minimized dross build-up in the pump well. Dross is typically a side effect of mechanical pumps. MagPump continuously ran at the set RPM of about 800. The metal level in the furnace did not affect MagPump’s performance; MagPump did not need to be raised when the bath had a low heel since it pulled metal from underneath, not from the top. MagPump eliminated the risk of splashing when the metal level was low, which contributed to TST‘s commitment to providing a safe work environment for TST’s associates.

O. A. Perez and E. Takahashi

MagPump Productivity TST’s present experience and most notable observation with operators of MagPump was observing the metal stirring movement that appeared to happen under the surface of the metal while maintaining a consistent temperature and constant charge rate. This resulted in a reduction of dross by not disturbing or breaking the metal surface. The effectiveness, reliability, and low maintenance of MagPump dramatically reduced the time needed by furnace operators to care for it, allowing operators to focus on other tasks.

MagPump Safety Contributions MagPump has no moving impellers, and its low maintenance reduces the operator’s exposure to molten metal. By having no moving parts submerged in the metal, it eliminates the risk of metal splashing. At TST, ensuring employee safety is always at the forefront of the decision-making process. MagPump provides peace of mind and priceless value, reducing the variable of uncertainty and provides stability that enhances productivity.

Zmag Zmag was founded in Japan in 1990 and since its inception, it has committed itself to introducing cutting-edge magnetic engineering solutions. Zmag’s original market was the sorting industry for scrap recycling, and its magnetic separators (e.g., eddy current separation) have been widely adopted by major companies in Asia. In the molten metal industry, Zmag introduced the world’s first permanent magnet-based stirring system, MagStir™, which is powered by the Zmag Permanent Magnetic Circuit (zPMC™) technology. MagStir comes with varieties of designs, customised for different types of furnaces, and is widely adopted by casthouses around the world. Zmag also introduced Typhoon™, another zPMC-based chip submerging system widely adopted by automotive and die-cast companies in Japan, Asia, and other markets.

North American Market In the North American market, the majority of casthouses operate side well furnaces equipped with (graphite) mechanical pumps. A mechanical pump comes with a rotating impeller shaft, which is submerged in molten aluminum and pumps aluminum in the side well and out to the main hearth. Flows from a mechanical pump help to quickly

MagPump

melt scrap that is charged in the charge well and/or chips that are charged in a vortex system next to a pump well. A mechanical pump is also used to inject gasses such as Argon, Nitrogen, and Chlorine to treat molten aluminum. Mechanical pumps have been the de facto standard technology in the industry for the past many decades.

Casthouse and Mechanical Pump Challenges A mechanical pump is submerged in molten aluminum. Physical submersion in molten metal creates a great deal of unavoidable challenges for casthouses. For example, some mechanical pump direct costs are: 1. Unpredictable running costs (consumable parts) A rotating impeller shaft may be cracked or damaged due to dross balls, rocks, broken refractory, etc. A mechanical pump’s support posts are also submerged in molten aluminum and oxidization thins and weakens the posts despite being protected by a special coating. The gas injection pipe and base are also required to be replaced when damaged or worn out. The timing of failure cannot be known and following Murphy’s Law, failures usually occur at the worst possible time. 2. Maintenance shop and spare pumps Since a mechanical pump needs to be repaired frequently, a multiple number of mechanical pumps are usually kept in the maintenance shop. Spare mechanical pumps need to be heated in a preheat chamber before being submerged in molten aluminum to avoid thermal shock. 3. Maintenance personnel

959

Hidden/indirect mechanical pump costs: 1. Lost production, revenue, and profit A casthouse typically spends an average of 4 h to replace a mechanical pump. This long process is required because thermal shock needs to be avoided. Slow submergence is a must. The impact of lost production, revenue, and profit varies among casthouses; however, it can be calculated using a simple formula. Loss of profit (US $) = melt rate (T/h)  4 h * # of downtime per year  aluminum profit (US $/T). 2. Dross generation A mechanical pump’s rotating impeller shaft breaks the surface of molten aluminum and causes dross generation in the pump well. Cleaning dross around the rotating shaft and posts is not easy, safe, or operator friendly. 3. Metal splash and operator safety A typical mechanical pump draws aluminum from the top of the base, and if a furnace sees a low heel and/or batch operation, a mechanical pump’s RPM must be lowered so that it does not splash molten aluminum. Splashing is a serious safety concern. Furthermore, in a low heel condition, furnace production will suffer (which may be considered as downtime at some casthouses) because the mechanical pump runs at a lower RPM in order to reduce splashing, rather than at the pump’s optimal RPM. 4. Wasted gas consumption and CO2 emissions

Typically, one or two dedicated maintenance personnel are required to maintain the plants' mechanical pumps.

When a mechanical pump is replaced, the furnace’s burners keep running in order to maintain bath temperature over the full replacement operation (4 h on average). This wastes natural gas and emits CO2.

4. Operator safety

5. CO2 emissions related to graphite production

When a mechanical pump fails, operators need to lift it out of molten aluminum and submerge a replacement pump. Safety is always a big concern for casthouses since any time operators are around molten aluminum, there is the possibility of an accident. The industry’s average annual mechanical pump budget per furnace is estimated to be about US $200,000. This does not include the cost of spare mechanical pumps, which is about US $27,000 per pump (on average, depending on size and other factors).

CO2 is also emitted when graphite is produced. Reduction of use of graphite due to reduction in consumable parts is good for the environment and industry. In order to solve the above-mentioned industry challenges, Zmag committed itself to invent MagPump™ that works as a direct replacement of a mechanical pump in a side well furnace. MagPump is designed to work as a pump rather than a stirrer.

Recycling of Aluminum from Aluminum Food Tubes Sarina Bao, Anne Kvithyld, Gry Aletta Bjørlykke, and Kurt Sandaunet

Abstract

Introduction

Aluminium is applied in food packaging due to its preservative capability. However, food residue and the fact that most packaging is thin gauge material, 50– 250 µm, makes recycling challenging. In Northern Europe, processed cheese, caviar, and mayonnaise are popular items stored in toothpaste-shaped tubes. This paper focuses on the evaluation of the recyclability of these Al tubes from the aspect of tube thickness, user habits, food residue, benefits of pre-treatment, all in regard to yield. Food residue reduced the yield from around 88% (non-used empty) to 57% (with 3% food residue), and down to 34% (with 16% food residue). For comparison, the influence of beverage residue on yield was also studied. The influence of beverage residue is minimal, even neglectable after drying the can. The results also show that the influence of food residue on the yield is larger than that from decoating. The producer change in tube wall thickness did not influence the recycling yield considerably. However, a thinner tube makes it easier to be emptied. This together with that thinner tube uses less Al to protect the same amount of food implies that the thinner tube can be regarded as more environmentally friendly. Keywords








Aluminium recycling Aluminium food tubes yield Decoating Food residue




Metal

S. Bao (&)
A. Kvithyld
K. Sandaunet SINTEF Materials and Chemistry, N-7465 Trondheim, Norway e-mail: [email protected] G. A. Bjørlykke Kavli AS, Sandbrekkeveien 91, 5225 Nesttun, Norway

Al is often applied in packaging due to its formability, strength, and protective quality. The Al packaging flows in Norway with an overview and challenges was reported in [1]. Also, a laboratory method for evaluating the quality and yield of the Al fraction from a Norwegian material recovery facility has been reported previously [2]. This work focus on the Al food tubes recycling. In Northern Europe, processed cheese, caviar, and mayonnaise are stored in toothpaste like Al tube to extend food conservation. This tube is made from an Al slug, containing for example 1xxx alloy with Al > 99.5%, in a size of a coin by impact extrusion. The slug is pressed at a high velocity with extreme force into a desired form by a punch. Impact extrusion is often applied to manufacture tubes, cans, and technical parts in the automotive, food, cosmetics, pharmaceutical, and electrical industries. After extrusion a protective barrier is sprayed inside the tube. Ink is dried on the outside of the tube on a white paint, which is resistant to most solvents and sun damage. Afterwards food can be inserted, and the gasket is sealed [3]. Analyses from disposed household glass and metal shows that there is 16.2% food residue left in tubes after use [4]. The coating is essential for food protection, but at the same time, it challenges the recycling process. If the coating is not removed before remelting, the content can hamper the recycling during remelting [5]. Thermal treatment is a standard industrial practice known as decoating [6]. However, burning of coating can increase oxidation and lead to environmentally unfriendly off-gas [7]. Oxidation can lead to metal loss and dross formation. Compaction [8] has been suggested as a method for preventing oxidation by reducing the surface area [9]. The fundamental processes of decoating is described in [10]. 10 min heat treatment at 550 °C has been chosen in the current work. Al scrap can be remelted in reverberatory furnace, rotary furnace, crucible, or electrical furnaces in the industry [11].

© The Minerals, Metals & Materials Society 2023 S. Broek (ed.), Light Metals 2023, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-031-22532-1_128

960

Recycling of Aluminum from Aluminum Food Tubes

Salt flux can be favorable when melting scraps of packaging. The salt layer can protect the metal underneath from further oxidation. It can extract impurities, inclusion, and oxides from the melt and extract metal entrapped in the dross. Beside reduced toxic properties and low vapor pressure, fluorides are often added into the salt to enhance coalescence. Fluorides help strip and break up the oxide layers so that imprisoned metal droplets can be accessed, and coalescence improved [12, 13]. After the sheet is milled, beverage cans are produced from forming and ironing operation [14]. Then washed, dried, painted externally, and coated internally before filling in the beverage [15]. Can lids (for example AA5182 with up to 5% Mg) are manufactured from a different alloy than can body (AA3004 or AA3104 with up to 1.4% Mg). In this work, the influence of beverage residue on recycling of Used Beverage Can (UBC) will be briefly discussed to compare with that of food residue in Al tube. Classical studies performed by Rossel, [16] have demonstrated the role of the thickness on metal loss. The thinner the scrap the higher the metal loss in particular for thickness under 20 mm as shown in Fig. 1. As expected, metal loss increases with temperature (Fig. 2). Particularly alloy with higher Mg content (AlMg4.5) notes intensified metal loss.

Experimental In this study Al tubes with two different gauge thicknesses were supplied by a food processor. The tubes were both non-used empty tubes and tubes filled with processed cheese as shown in Fig. 3. Filled beverage cans were provided by an Al company. First, food was emptied from the used tubes by various methods, i.e., tube presser, hand roll, hand squeeze, squeezing key, and even more creative ways as stepping on, and scrolling on the table edge as shown in Fig. 4.

961

Fig. 2 The melt loss for various Al alloys as a function of temperature. Original figure from [16]

Fig. 3 Filled food tubes in the left picture (thicker gauged tube with white label), and empty non-used tubes in the right picture (thicker gauged tube to the left with blue color)

Fig. 4 Several ways to empty the food tubes. a tube presser, b tube emptied with tube presser, c tubes stepped on and rolled on the table edge, and d tubes rolled with squeezing key

Fig. 1 The melt loss for scrap thickness at 750 °C. Original figure from [16]

Secondly, the tube’s mass was registered, and gauge thickness was characterized by light microscopy. Third, the Al tubes and cans were placed in a muffle furnace and heated up to 550 °C over approximately 30 min, and held at 550 °C for 10 min. After this decoating step was completed, white paint remained for tubes, while cans have changed the colour as shown in Fig. 5. Fourth, to investigate the thermal decoating of lacquered Al scrap in detail, a Linkam TS 1500 hot stage with a Leica DMLM microscope was used. A 10 mm scrap piece

962

S. Bao et al.

a Thin

Thick

b

c

Fig. 5 Before and after decoating of Al tubes a, and before b and after c decoating of UBC

Fig. 7 Measurements of coatings and gauge thickness of the thin tube body. From left to the right, outer protective layer ink, (white) paint, Al, and inner protective barrier is 3.1 µm, 8.6 µm, 144.1 µm, and 8.3 µm, respectively

Table 1 Weight of Al tubes Used tube Weight (g)

Non-used tube

Thick

Thin

Thick

Thin

14.0 ± 0.1

12.5 ± 0.1

13.5 ± 0.1

12.7 ± 0.1

Results and Discussion Characterization of Al Food Tubes

Fig. 6 Al tubes remelted in salt

from Al tube with the letter Ø was heated with 30 °C/min up to 700 °C and cooled to room temperature in 20 min in air. The same temperature profile was used in a thermogravimetric (TG) furnace in synthetic air, where the weight loss during decoating were measured as a function of time and temperature. Fifth, as shown in Fig. 6, the Al samples were immersed into approximately 150 g melted salt, which contains 45% NaCl, 45% KCl, and 10% Na3AlF6, in a teacup-sized crucible in copper induction coil furnace. Both salt and the Al droplets were cast into a mould, after stirring and complete remelting. The salt was then crushed at cold conditions, and Al droplets were collected, soaked in water for overnight, dried, and weighted to determine the yield.

From optical microscope analysis, as an example shown in Fig. 7, the shoulder and gauge thickness are 649 ± 7 µm and 145 ± 4 µm for thin tubes, and 841 ± 25 µm and 167 ± 7 µm for thick tubes. Thus, the thick tube has an extra 192 µm Al on the shoulder and 22 µm more Al on the body part of the tube. After emptying the food in tubes by various methods, the tubes were washed, dried, and mass registered. As listed in Table 1, these two designs of thin tubes (Fig. 3) need 0.8– 1.5 g less Al than the thick tubes.

Squeezing Out Food from Al Tube Considering the variety of customers, a child of 5–7 years and an adult were engaged to squeeze out the cheese with a tube presser (Fig. 4a). As shown in Table 2, at least 5 g food residue would be left inside the tube, which is mostly located around the shoulder area. It is reasonable to see slightly more food residue for a child. Food residue with hand roll, hand squeeze, and squeezing key was bigger than when using a tube presser, stepping on, or rolling on the table edge. The food residue for thick and thin tubes is compared in Fig. 8. There is always slightly less, 0.4–4.3 g, food residue in thin

Recycling of Aluminum from Aluminum Food Tubes Table 2 Food residue inside the tube emptied with tube presser Thick tube

Thin tube

a

963

Man-force

Full mass (g)

After press (g)

Food residue (g)

Food residuea (%)

Child

188.2

20.8

6.7

3.8

Adult

190.1

20.3

6.2

3.5

Adult

188.2

19.5

5.6

3.2

Child

187.6

18.2

5.7

3.3

Adult

186.8

17.9

5.3

3.0

Adult

186.9

17.6

5.0

2.9

Considering the net weight of cheese 175 g

Fig. 8 Food residue in thick and thin used tubes with tube presser, stepping on, or rolling on the table edge

6.0 Weight loss, [%]

tubes with various methods. This means it is easier to squeeze out food from thin tubes. The highest food residue amount in this study is 6% and only a third of what was found in the analysis from household waste [4]. This means this study achieved likely the lowest limit of customer habit. More food residue should be expected in everyday life.

4.0 2.0 0.0

Thermal Decoating The weight loss due to decoating is shown in Fig. 9 with 4.4, 3.8, and 1.5% for non-used thick tubes, non-used thin tubes, and UBC, respectively after preheating at 550 °C. The weight loss for the tubes is the same (within the uncertainty) and larger than that of the beverage can. The variation is due to coating components and thickness. In Fig. 10, the fundamental processes described in the literature are confirmed for decoating. The first volatilization and pyrolysis of the coating (scission phase) has a peak

Thick tube Thin tube

UBC

Fig. 9 Weight loss due to decoating

around 430 °C and the pigments and carbon residue are shown in the bottom left picture. The following combustion peak occurs around 550 °C (combustion phase) and only inorganic material remains on the surface as shown in the bottom right picture. The white residue is considered to be pigmented TiO2. Pictures of starting materials and end are shown above the graphs in Fig. 10.

964

S. Bao et al.

Fig. 12 Recycling yield of empty non-used tubes

Fig. 10 The weight loss with temperature during decoating and corresponding observation under a hot stage microscope

Remelting and Recycling Remelting the tubes and beverage cans in salt with high content of Na3AlF6 had, as expected, enhanced the coalescence of the metal. Only 5 out of 23 experiments had more than one merged drop. Figures 11 and 12 present the recycling yield, which is calculated by metal out divided by

material into the salt. Food residue in tube reduces the recycling yield from 88% (empty non-used tube) to around 57% (with 3% food residue), even down to 34% (with 16% food residue). The result in Fig. 12 indicates that the recycling yield of the decoated tube is slightly higher than that of the non-decoated tube for non-used tube (94–88%). However, such an effect is not obvious for the used tube (Fig. 11). Taking the coating into the denominator of yield calculation, the yield was slightly reduced for empty non-used tubes from 94 to 90%, and for used tubes with 3% residue from 52 to 49%. This indicates that decoating reduces the contamination and thus increases yield, 3–4% in the current case. However, food residue has decreased the yield from the 90 to the 50%, that is by as much as 40%. With a thickness of several hundreds micrometer, when the shoulder and wall increase for 192 and 22 µm, it does not seem to change the recycling yield for non-used Al tubes (Fig. 12). But thick used tube has slightly larger recycling rate than thinner (Fig. 11). Meanwhile UBC has much higher recycling yield than tubes (Fig. 11). This highlights that the influence of beverage residue is much less than that of food residue, even neglectable after drying.

Parameters Influence Al Tube Recycling

Fig. 11 Recycling yield of tubes and UBC with residue

Thickness. From Fig. 1 we see that metal loss for alloy of 99.5% Al at 750 °C is around 5% for the gauge of 150 µm, and 4% for shoulder around 650 µm. The weight loss in this work is slightly larger, 6%, for non-used decoated tube. Food residue One reason why recycling food tube is more difficult than that of UBC is food residue, which reduces the yield largely. Meanwhile, food residue produces large amount of gas, and has to be handled with care when

Recycling of Aluminum from Aluminum Food Tubes

recycled. Shredding, or slowly heating before entering molten metal is necessary. The lacquer thickness is around 5% for tube body and 1% for shoulder for both thin and thick tubes, but slightly higher for thin tubes. Decoating reduces the contamination from lacquer and thus increases yield, but, as we observed, food residue reduces the yield more significantly. Alloy. As known, alloying element Mg can be readily selectively oxidized during remelting due to its affinity for oxygen than Al. Thus, UBC alloys, AlMg4.5 (can top) and AlMn1Mg1(can body) give almost up to 3 times more weight loss than a 99.5% pure Al which is popular for Al tube extrusion (Fig. 2). However, the recycling yield is 30– 40% lower for Al tubes in this work (Fig. 11), that is dominated by the food residue in Al tubes. Recycling furnace. Different types of furnaces for melting Al scrap can be used, depending on the initial metal content in the scrap, type and content of impurities, geometry of the scrap, frequency of change in the alloy composition, operating condition, energy cost, and desired product quality [11]. For example, rotary furnace is more common in Europe due to high energy cost. While gas reverberatory furnace, which operate with a lower energy efficiency and requires lower capital cost, covers the 95% of Al scrap in the United States. Considering these features, the laboratory crucible furnace used in this work has less yield than reverberatory furnace. The yield basically depends on the recycling method as described in the introduction. These experiments simulate part of a rotary furnace practice with salt. The industry uses less salt and has generally higher yields than what is usually obtained in small-scale laboratory experiments. It is experienced that the recycling yield in the industry of contaminated food packaging also can be low.

Conclusions Al food tubes and UBC were remelted and recycled in a laboratory scale teacup-sized Cu coil induction furnace with salt. Two Al tubes, thick and thin, were investigated. When the shoulder and wall thickness increase respectively 192 and 22 µm, thinner tubes require 0.8–1.5 g less Al alloy than the thick tubes. This thickness change does not seem to influence the recycling yield considerably. All the methods used to empty the Al tubes resulted in that the thinner tube had less food residue. The food left in the tubes lowers the recycling yield significantly. Thus, a thinner Al tube is more environmentally friendly.

965

Decoating reduces the contamination and thus increases yield, but food residue has a more dominating effect on yield. UBC has a much higher recycling yield than Al tubes. The influence of beverage residue is less than that of food residue, even neglectable after drying. Acknowledgements This research was carried out as part of the Norwegian Research Council (NRC)—funded IPN Project (296276) Alpakka-Circular Aluminium Packaging in Norway. It includes the following partners: Norsk Hydro, Metallco Aluminium, Norsk Metallgjenvinning, Infinitum, O. Kavli, NTNU, and SINTEF. Funding by the industrial partners and NRC is gratefully acknowledged. Ingrid Hansen and Kjersti Øverbø Schulte, SINTEF are acknowledged for hot microscopy study, and discussion and methods for emptying the tubes and packaging, respectively. Thanks are also given to Kavli and Nork Hydro for supplying Al tubes and UBC.

References 1. M. S. A. K. Magnus Skramstad, “Aluminium Packaging Flow in Norway,” Trondheim, Apr. 2021. 2. S. Eggen, K. Sandaunet, L. Kolbeinsen, and A. Kvithyld, “Recycling of Aluminium from Mixed Household Waste,” in Minerals, Metals and Materials Series, 2020, pp. 1091–1100. doi: https://doi.org/10.1007/978-3-030-36408-3_148. 3. “How It’s Made” featuring Montebello Packaging. Accessed: Aug. 10, 2022. [Online Video]. Available: https://www.youtube. com/watch?v=8yOqFbygVAQ. 4. Sveinung Bjørnerud, “Plukkanalyser glass/metall 2020,” 2020. 5. S. Capuzzi, A. Kvithyld, G. Timelli, A. Nordmark, E. Gumbmann, and T. A. Engh, “Coalescence of Clean, Coated, and Decoated Aluminum for Various Salts, and Salt–Scrap Ratios,” Journal of Sustainable Metallurgy, vol. 4, no. 3, pp. 343–358, 2018. 6. T. A. Engh, G. K. Sigworth, and A. Kvithyld, Principles of Metal Refining and Recycling. Oxford University Press, 2021. 7. Solveig Høgåsen, “The effect of compaction and thermal treatment in the recovery of coated aluminium scrap through FTIR off-gas analysis and remelting in molten heel,” Norwegian University of Science and technology, Trondheim, 2022. 8. A. Vallejo-Olivares, S. Høgåsen, A. Kvithyld, and G. Tranell, “Effect of Compaction and Thermal De-coating Pre-treatments on the Recyclability of Coated and Uncoated Aluminium,” in Light Metals 2022, Springer, 2022, pp. 1029–1037. 9. C. Hamers and A. Jessberger, “Aluminium Cycle: Machining, Briquetting, Melting,” Global Recycling, Mar. 2018. https:// global-recycling.info/archives/2354 (accessed Aug. 10, 2022). 10. A. Kvithyld, C. E. M. Meskers, S. Gaal, M. Reuter, and T. A. Engh, “Recycling light metals: Optimal thermal de-coating,” JOM, vol. 60, no. 8, pp. 47–51, 2008, doi: https://doi.org/10.1007/ s11837-008-0107-y. 11. S. Capuzzi and G. Timelli, “Preparation and melting of scrap in aluminum recycling: A review,” Metals, vol. 8, no. 4. MDPI AG, Apr. 01, 2018. doi: https://doi.org/10.3390/met8040249. 12. M. S. S. Besson, A. Pichat, E. Xolin, P. Chartrand, and B. Friedrich, “Improving Coalescence in Al-Recycling by Salt Optimization,” in Proceedings of EMC, 2011, p. 1.

966 13. R. D. Peterson, “Effect of salt flux additives on aluminum droplet coalescence,” in Proceedings of the 2th International Symposium on Recycling of Metals and Engineered Materials, 1990, pp. 69– 84. 14. D. A. Doutre, “LiMCA and its Contribution to the Development of the Aluminum Beverage Container and UBC Recycling,” 2011.

S. Bao et al. 15. PE Americas, “Life Cycle Impact Assessment of Aluminum Beverage Cans,” 2010. [Online]. Available: www.pe-americas. com. 16. H. Rossel, “Fundamental investigations about metal loss during remelting of extrusion and rolling fabrication scrap,” Light Metals, pp. 721–729, 1990.

Recent Studies Using HR-TEM on the Fundamental Mechanism of Nucleation of a-Aluminum on TiB2 in TiB D High-Efficiency Grain Refiners John Courtenay and Yun Wang

Abstract

Recent studies using HR-TEM, (High-Resolution Transmission Electron Microscopy) at BCAST, Brunel University, London, on the mechanism of nucleation of aaluminium on TiB2 in commercial high-efficiency grain refiners have shown that efficiency as measured by the Opticast Nucleation test can be directly related to the extent to which the TiB2 particles have successfully formed a monoatomic layer of TiAl3 on their basal plane. This factor was found to be predominant over other factors such as particle size distribution and average particle size. As a result of this research, it is now possible to produce a range of ultra-high efficiency grain refiners which enable addition rates to be further reduced to levels as low as 0.15 kg/t resulting in higher cost savings and the particle count being significantly reduced in the liquid metal leading to as cast metal quality being improved. Keywords

Aluminum




Grain refinement




Efficiency

Introduction Grain refinement during the casting of engineering alloys is usually desirable, since it results in not only a fine and equiaxed grain structure but also a significant reduction of casting defects, which in turn leads to an improved engineering performance [1]. As a common practice in the metal

J. Courtenay (&) MQP International Ltd, Solihull, UK e-mail: [email protected]

casting industry, grain refinement is usually achieved by chemical inoculation through the addition of grain refiners. Among a series of Al-Ti-B-based grain refiners developed for Al-alloys, Al-5Ti-1B (all compositions are in wt.% unless otherwise specified) master alloy has been the most widely used grain refiner, which contains potent TiB2 particles for enhancing heterogeneous nucleation of a-Al grains [2]. Since the introduction of Al-Ti-B-based grain refiners at the beginning of the 1950s [3], significant effort has been made to optimize their performance largely through a trial-and-error approach and to understand the grain refining mechanisms, with both the theory and practice being extensively reviewed [4, 5]. Various researchers recognized that AI3Ti was a much stronger nucleant than TiB2; however, its presence could not be substantiated because it was not thermodynamically stable for the hypo peritectic case. For example, Johnsson [6] indicated that the aluminides probably take less than 1 min, but definitely less than 5 min to dissolve at a holding temperature of 775 °C. Jones and Pearson suggested that all aluminides dissolved in molten Al in less than 30 s. Although the dissolution time depends on both the holding temperature and the size of the AI3Ti particles, the resulting consensus is that AI3Ti dissolves rapidly above the Al liquidus and hence the peritectic reaction is not thermodynamically feasible. For AI3Ti to be responsible for grain refinement during the solidification of hypoperitectic alloys, other factors have to be operational. Vader et al. [7] and Backerud et al. [8] proposed the peritectic hulk theory in the early 1990s. This theory recognized that AI3Ti is a more potent nucleant than TiB2 and attempted to explain how the borides increase the stability of aluminides. It was suggested that the borides form a shell around the aluminides and slow down the dissolution of the aluminides. The aluminides eventually dissolve and leave a cell of liquid with approximately the peritectic composition. The peritectic reaction can then take place to form the a-Al.

Y. Wang BCAST Brunel University, Uxbridge, England © The Minerals, Metals & Materials Society 2023 S. Broek (ed.), Light Metals 2023, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-031-22532-1_129

967

968

However, a number of researchers (e.g., Refs. [1–13] investigated the Al-rich comer of the Al-Ti-B phase diagram and showed that boron had virtually no effect on the Al-Ti phase diagram. Therefore, the nucleation process cannot be explained purely based on theories that attempt to modify the conditions for the peritectic reaction to occur. Later in pioneering TEM carried out by Schumacher and Greer [10, 11] using a metallic-glass technique, it was claimed that the presence of an AI3Ti coating phase on the TiB2 was essential for them to be potent. In the first phase of this work, experimental evidence for the existence of an atomic monolayer of (112) AI3Ti twodimensional compound (2DC) on the (0001) TiB2 surface of commercial Al-Ti-B-based grain refiners was reported [12, 13]. In the second phase of the work, HR-TEM studies of grain refiners with varying performance as measured by the Opticast Nucleation Test have been able to demonstrate that efficiency is related to the extent to which the TiB2 particles have been successfully coated.

J. Courtenay and Y. Wang

3 mm diameter discs. The discs were then hand ground to a thickness of less than 70 µm, followed by ion beam thinning using a Gatan PIPS facility.

Results Sample 1 50% efficiency A total number of 37 TiB2 particles were examined for the 50% efficiency grain refiner. Six particles were electronically thin enough and thus the status of their (0001) surfaces is clarified decisively by HR-TEM. Five of the six particles had no Al3Ti 2DC layer, and one exhibited a partial Al3Ti 2DC layer on its basal plane (Fig. 1). An orientation relationship (OR) between TiB2 and Al grain is not observed in 50% sample, although limited particles were examined (Fig. 2).

Study Using HR-TEM of the Extent of Layer Formation in Samples of Grain Refiner with 50 and 123% Efficiencies Having established that nucleation can only occur when an AI3Ti monolayer is present on the surface of the TiB2 particles, the next phase of the work centered on examining the possibility that the wide variations in grain refiner performance observed using MQP’s Opticast nucleation test measurements of efficiency might be due to differences in the extent to which the TiB2 particles in the grain refiner had successfully formed an Al3Ti layer. Samples of commercial Optifine Al-3Ti-B production provided by MQP were examined by HR-TEM as follows:

Sample 1 This batch had been rejected for acceptance as Optifine because the specific efficiency measured by the Opticast Nucleation test was only 50% compared to the target efficiency of 100%.

Sample 2 This batch was from experimental production aimed at achieving substantially higher efficiency and had an efficiency measured at 123%. To prepare thin foils for conventional HR-TEM, and HRSTEM examinations, slices from the rods of the grain refiner master alloys were mechanically ground and cut into

Fig. 1 TEM bright field images showing the morphology of the TiB2 particles in the low-efficiency (50%) grain refiner master alloy. Both the particles are being viewed along their [1 1 −2 O] direction

Recent Studies Using HR-TEM on the Fundamental …

969 The extent of Al3Ti 2DC on layer formation on TiB2 particle is a priority factor in determining the refining efficiency.

The results can be shown in a histogram as follows: Possibility of 2DC Ti3 AI in grain finers 1.00 Q.)

0.80

tu)

2 0.60 C

Q.) Q.)

0.40

a..

0.20

0.00

Fig. 2 No Al3Ti 2DC layer is seen on the surface of TiB2

123 %GR

50%GR

Grain Refiner

5

Sample 2 123% efficiency

Development of New Super Efficiency Grain A total of 33 samples were examined and among them, eight Refiner-Optifme 5:1125 particles were electronically thin enough and thus the status of their (0001) surfaces was clarified by HR-TEM (Fig. 3). Seven of the eight particles were confirmed to have the Al3Ti 2DC layer, and one had no Al3Ti 2DC layer. In summary, in this phase, we have shown the following:

Following promising results with experimental production of an Optifine 3:1 with higher efficiency, MQP decided to concentrate further development on a new version based on the Al-5Ti-B system.

HR-TEM analysis shows the probability of TiB2 particles which nucleated Al grains is significantly lower in 50% than in 123% sample.

Fig. 3 TEM bright field images showing the morphology of the TiB2 particles in the high-efficiency (123%) grain refiner master alloy. Both the particles are being viewed along their 1 1 −2 0 direction

Fig. 4 a High-resolution TEM image showing a well-defined orientation relationship (OR) between TiB2 and Al grain observed in highefficiency (123%) sample Orientation Relationship (OR): (0001)[11 −20]TiB2//(111)[0 −1 l]Al. b Fast Fourier transformation (FFT) pattern obtained from Fig. 4a showing the well-defined orientation relationship is (0 0 0 1)[1 1 −2 0]TiB2//(1 1 1)[0 −1 l]Al

970

Samples as follows from trial production were provided for HR-TEM examination to quantify the degree of particle coating, SEM analysis to study morphology and particle size distribution, and macro analysis to review the rod structure.

J. Courtenay and Y. Wang

Thus, it was concluded, albeit on a limited number of particles observed, that the degree of TiBD particle coating with the Al3Ti atomic monolayer was significantly higher in the new Optifine 5:1 125 product compared to the earlier experimental Optifine 3:1 which had an efficiency of 50%

• Batch 0417 W 5:1 (Not Optifine) 63% efficiency. • Batch 6878 T Optifme 5:1 125 125% efficiency. In sample 6878T, the much-expected orientation relationship (ORl) is observed in the Optifine 5:1 125, indicating its high grain refining efficiency (Fig. 5b).

Results of HR-TEM and SEM Analyses HR-TEM Analysis—Batch 6878T Optifine 5:1125 120% Nine TiB2 particles from batch 6878T were successfully prepared for analysis and examination showed that all were fully covered with a Al3Ti2DC layer indicating 100% particle coating of the 125% efficiency Optifine 5:1 125.

Examination of Particle Morphology and Particle Size Distribution of Optifine 5:1125 Samples of the rod from trial production were examined at low magnification to compare the structure in the longitudinal section between the successful Optifine 5:1 125 (125% efficiency) batch used in the previous HR TEM examination with a rejected batch that had exhibited only 63% efficiency when measured on the Opticast Nucleation test. A more uniform microstructure at low magnifications for the high-efficiency grain refiner batch 6878T was observed compared to that for the low-efficiency batch 0417TW. Under examination by SEM at 20,00 k x, a clear difference in morphology can be observed. In the low-efficiency sample, evidence of significant amounts of salt reaction products can be seen around the periphery of the TiB2 particles whilst in the high-efficiency sample, the TiB2 particles are more clearly developed being clean and smooth with only minor amounts of salt reaction product present. Measurement of particle size distribution From Figs. 9 and 10, it can be concluded that TiBD particles in the two samples are of similar average size dD and size distribution a−.

Fig. 5 a The much expected orientation relationship (OR1) is observed in the Al-5Ti-B, indicating its high grain refining efficiency. b The well-defined orientation relationship (OR): (0 0 0 1) [1 1 −2 O] TiB2II(1 1 1)[0 −1 l] Al is again confirmed by the Fast Fourier transformation (FFT) pattern obtained from Fig. 5a

Alloy

Batch

Efficiency (%)

Size distribution (dD µm)

(J)

Particle coating AITi3

5:1

6878T

125

0.43

0.44

100%

5:1

0417

63

0.42

0.54

Fig. 6 HR-TEM lattice image indicates that the Al3Ti 2DC layer is readily observed on the basal surface of TiB2 particles in the Al-5Ti-1B master alloy (Optifine 5:1 125)

Recent Studies Using HR-TEM on the Fundamental …

971

Fig. 7 Low-magnification examination of longitudinal sections from rod

6878T Al-5Ti-1B (125%), d0=0.43µm, =0.44

Number percentage, %

10

8

d0 = 0.43 mm  = 0.44

6

3613 particles

4

2

Fig. 8 High-magnification examination of longitudinal sections

0417W Al-5Ti-1B (63%), d0=0.42µm, =0.54

0 0.0

0.5

1.0

1.5

2.0

2.5

Particle size, m Fig. 10 Particle size distribution for 6878 T Al-5Ti-1B (125%), d0 = 0.43 µm, cr = 0.44

Summary

Fig. 9 Particle size distribution for 0417 W Al-5Ti-1B (63%), d0 = 0.42 µm, cr = 0.54

• The microstructure of the high-efficiency Al-5Ti-1B alloy (6878 T, 125%) is more uniform than that of the lowefficiency one (0417 W, 63%). • The low-efficiency alloy shows more severely agglomerated TiB2 clusters than the high-efficiency one. • Salt reaction products are readily found to remain in the low-efficiency alloy. • In spite of their difference in refining efficiency, the TiB2 particles in the three Al-5Ti-1B rods show similar average size and size distribution.

972 Table 1 Histogram showing % of particles coated with Al3Ti 2DC layer versus grain refiner efficiency

J. Courtenay and Y. Wang Grain refiner

Addition rate (kID’ton)

Alloy

Standard 3/1

1, 0.5, 0.2

AAA 1050

Optifine 3/1 Optifine 5/1

• TEM reveals that, so far, all the nine TiB2 particles examined in the high-efficiency refiner are covered with Al3Ti 2DC, indicating its high refining efficiency (125%).

This results in a reduction of the addition rate of 85% yielding even higher cost savings.

Discussion Results for Super High-Efficiency Optifine 5:1125 Compared to Optifine 3:1 Laboratory trials were conducted using AA1050. In the trial, the performance of three grain refiners with significantly different efficiencies—including standard 3/1, Optifine 3.1, and Optifine 5.1:125—were tested at addition rates from 0.2 kg/t up to lkg/t; see Table 1. The alloy was melted in a resistance furnace and the temperature was held at 720 ± 5 °C throughout the testing period. The performance of each grain refiner was tested using the Opticast equipment. Figure 11 shows the effect of addition rate versus grain size. The red hatched line indicates that Optifine 3:1 provides the same grain size at an addition rate of 0.28 kg/t compared to the standard grain refiner at an addition rate of l kg/t. This results in a reduction of the addition rate of more than 70%, which, in practice, will reduce the cost of grain refinement by 50%. Respectively, the yellow hatched line shows that Optifine 5:1 125 provides the same grain size at an addition rate of 0.15 kg/t compared to the standard grain refiner at 1 kg/t.

Fig. 11 Performance of Optifine 5:1125 on AAA 1050 compared to Optifine 3:1 and standard grain refiner

In their 2015 paper on the “Grain refining mechanism in the Al-Ti-B system” Fan et al. [13] remarked in the summary of the discussion that “suitable size, size distribution, and adequate number density of TiB2 particles in the Al-5Ti-1B grain refiner can also be contributing factors to effective grain refinement. The Al-Ti-B-based grain refiners were developed over 60 years, mainly by trial and error, and their degree of optimization is surprisingly high. It is anticipated that further improvement of grain refining efficiency will be very difficult, if at all possible”. This was the state of the art at that time. Although by using advanced techniques such as HR-TEM, HRSTEM, and super STEM, it had been possible to establish that in terms of the mechanism it was the formation of an ABTi 2DC monolayer that enables normally low nucleation ability TiB2 particles to nucleate successfully; the quantitative effect of this and the other contributory factors such as particle size distribution on grain refiner efficiency had not been studied. However, with the development of a rapid, reliable, accurate, and reproducible testing method capable of mass application in the production process, it is now possible to generate accurate data on how efficiency varies with a variety of factors. The TP-1 test has been the industry standard for testing for more than 50 years but without modification and the use of multiple samples, it has severe statistical limitations. It is also time-consuming to operate and requires a large melt volume. These practical limitations have hindered the measurement and control of efficiency. In this study, the use of the Opticast Nucleation test has made it possible to correlate the enhanced understanding of the nucleation mechanism, in terms of the extent to which the Al3Ti2DC layer is present on TiB2 particles, directly to the efficiency of the grain refiner expressed in grains/mm3/ ppm of boron, or for practical purposes kg/t of grain refiner addition to produce a given grain size for different alloys. At the time of the 2015 study although it was considered that commercial grain refiners were quite well optimized, it has been demonstrated subsequently that this is not the case

Recent Studies Using HR-TEM on the Fundamental …

and wide variations in efficiency exist between different batches of standard grain refiners supplied. If we take Optifine 3:1 as the current standard for high-efficiency grain refiner with a relative efficiency of 100%, then although many batches of standard grain refiner may have relative efficiencies between 50 and 80%, there are always batches with efficiencies as low as 30% present and this means that application rates in practice have to be based on this value if cracking of cast product is to be avoided. Now with the application of the latest technology, it is possible to produce a commercial supply of grain refiner specified as either 100 or 120% efficiency representing in the best case an improvement of 85% in efficiency or in practical terms a reduction in addition rate from l kg/t down to 0.15 kg/t with strong benefits in terms of cost reduction and improvement of quality.

973

7.

8. 9. 10.

120% efficiency. TEM examination of nine particles from a sample of batch 6878 T has revealed that all nine TiB2 particles were covered with an Al3Ti 2DC layer, indicating its high refining efficiency (125%). The microstructure of the high-efficiency Al-5Ti-1B alloy (6878 T, 125%) is more uniform than that of the low-efficiency one (0417 W, 63%). The low-efficiency alloy shows more severely agglomerated TiB2 clusters than high-efficiency one. Salt reaction products are readily found to remain in the low-efficiency alloy. In spite of their difference in refining efficiency, the TiB2 particles in both Al-5Ti-1B batches show similar average size and size distribution.

References Conclusions 1. Whilst TiB2 particles can nucleate a-Al, they have only low potency, and to nucleate effectively require the formation of a (112) Al3Ti 2DC monolayer on the (0001) TiB2 surface. 2. Al3Ti 2DC is stable in concentrated Al-Ti solution but is unstable and dissolves in dilute Al-Ti solution. 3. Excess Ti a affects the effectiveness of Al-Ti-B grain refiner by the formation of Al3Ti 2DC to increase the potency of TiB2 particles and the provision of free Ti in the inoculated melt to cause columnar- to-equiaxed transition. 4. Studies of grain refiner batches with varying efficiencies have shown that efficiency is proportional to the extent to which the TiB2 particles exhibit an Al3Ti 2DC layer. 5. Specifically, in a 50% efficiency batch, only 20% of TiB2 exhibit a Al3Ti 2DC layer, whilst in a 123% batch, 80% of TiB2 particles exhibited an AI3Ti 2DC layer. 6. A new grain refiner has been developed, Optifine 5:1 125 based on the Al-5Ti-B system with a target of

1. G.K. Sigworth, in: ASM Handbook, Casting. Metals Park,vol. 15, ASM, OH, 2008, p. 255. 2. B.S. Murty, S.A. Kori, M. Chakraborty, Int. Mater. Rev. 47(2002) 3. 3. A. Cibula, J. Inst. Met. 80 (1951) 1. 4. D.G. McCartney, Int. Mater. Rev. 34 (1989) 247. 5. G.P. Jones, J. Pearson, Metal!. Mater. Trans. B 7 (1976) 223. 6. M. Johnsson. A critical survey of the grain refining mecha- nisms of Al (PhD thesis), Stockholm University, 1993. 7. M. Vader, J. Noordegraaf, P.C. Van Wiggen, in: E.L. Rooy (Ed.), Light Metals, TMS, Warrendale, PA, 1991, pp. 1123–1130. 8. L. Backerud, P. Gustafson, M. Johnsson, Aluminum 67 (1991) 910. 9. G.K. Sigworth, Metal!. Mater. Trans. A 15 (1984) 277. 10. P. Schumacher, A.L. Greer, J. Worth, P.V. Evans, M.A. Kearns, P. Fisher, A.H. Green, Mater. Sci. Tech. 14 (1998) 394. 11. P. Schumacher, A.L. Greer, Mater. Sci. Eng. A 181 (1994) 1335. 12. Z. Fan, Y. Wang, M. Xia, S. Arumuganathar, Acta Mater. 57 (2009) 4891. 13. Z. Fan, Y. Wang, Y. Zhang, T. Oin, X.R.Zhou, G.E. Thompson, T. Pennycook and T. Hashimoto, Acta Mater. 84 (2015) 303.

A Cellular Automaton Model for Qualifying Current Grain Refiners and Prescribing Next-Generation Grain Refiners for Aluminium Alloys G. Salloum-Abou-Jaoude, S. Sami, A. Jacot, and L. Rougier

Abstract

Introduction

In cast aluminium products, small equiaxed grains reduce the risk of hot tears and shrinkage porosities by facilitating the liquid feeding in the interdendritic liquid. Although grain refinement in aluminium alloys is well known and widely used, grain size control is still not always guaranteed industrially. This depends largely on the nature and fabrication quality of the grain refiner rod. In this work, we developed a cellular automaton model to establish a clear link between the grain refiner type/nature and grain refining efficiency while accounting for the principal physical phenomena affecting grain refiner performance: grain refiner nature, nucleant size distribution, recalescence, solute suppressed nucleation zone. At TMS2020 [1] we experimentally highlighted the inconsistencies in grain refiner performance between different producers and in batches of the same producer. This model helps in qualifying grain refiners and would serve as a prescriber for designing next-generation grain refiners with superior efficiency. Keywords




Grain refiner Solidification Cellular automaton model




Aluminium alloys

G. Salloum-Abou-Jaoude (&)
S. Sami Constellium Technology Center C-TEC, Parc Economique Centr’alp, 725 Rue Aristide Bergès, CS10027 Voreppe, France e-mail: [email protected] S. Sami
L. Rougier ESI Group, 70 Rue Robert, 69006 Lyon, France A. Jacot Calcom ESI SA, Route Cantonale 105, 1025 St-Sulpice, Switzerland




Grain size control during casting of aluminium alloys have always been essential to ensure sound cast quality and guarantee the desired properties of the final product. A non-reliable solidification microstructure control, reduces production yield, resulting in significant value loss. This also increases the plants runaround scrap in an industry where recycling is key. A plant is more than ever seeking to reduce its runaround scrap to maximise external scrap consumption resulting in reduced carbon footprint of the alloy. A well-controlled equiaxed grains size [1, 2], helps in increasing production yields by reducing the risk of hot tearing [3] and shrinkage porosities in cast products. This is a results from enhanced liquid feeding between the solidifying a-Al grains [4]. Although grain refinement in aluminium alloys is well known from the fundamental and experimental point of view, grain size control is still not always guaranteed at the industrial scale. This depends largely on the nature and fabrication quality of the grain refiner rod. There are still debates in the industrial community about what makes a grain refiner less or more efficient. During the last decades there has been great interest to quantitatively predict solidification microstructures. Several modelling techniques have emerged such as multiphasemodels [5–7] and cellular automata [8] which can provide fields of various microstructural quantities on the scale of an entire cast component. On a smaller scale, the phase-field method can now address the growth of columnar or equiaxed grains with a very high level of details about the solid morphology and the distribution of the solute elements in the microstructure including the effect of fluid flow [9, 10]. In spite of this progress, the prediction of grain size remains a modelling challenge. The reason is that the average grain size in a microstructure is the result of a complex competition between nucleation and growth and is governed by phenomena taking place at very different length scales.

© The Minerals, Metals & Materials Society 2023 S. Broek (ed.), Light Metals 2023, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-031-22532-1_130

974

A Cellular Automaton Model for Qualifying Current Grain Refiners …

A major progress in the understanding of grain refinement and quantification of the grain size has been the free growth theory of Greer [11]. One of the outcomes of the theory is that the particle size distribution of the inoculant can be translated into a distribution of nucleation undercoolings. Combining this with a proper growth model, the grain size can be predicted [12]. Models predicting the grain size normally also include a calculation of a thermal recalescence, which will stop nucleation, leaving the less potent particles unused. Another phenomena to also take into consideration, is the solute suppressed nucleation zone (SSNZ), it is well described by Easton and StJohn [13]. Predicting the grain size using a direct representation of the individual grains requires a relatively large computation domain to have enough grains to carry out meaningful statistics. For grain refined alloys, computation domains must typically be in the order of 1 mm to correctly predict the grain size in a given region of the casting. While state-of-the art implementations of the phase-field method [9] would probably allow to address such volumes, the computation cost of the technique remains very high, and prevents any implementation in numerical schemes where such local microstructure calculations would be coupled with heat transfer on the process scale. Another choice would be mean-field approaches, which have been applied with success [8]. A direct description of the grains is however desired when non-random spatial distributions of nucleant particles are analysed, including the complex multiple interactions through their diffusion fields. For these reasons, an envelope model was considered. In this approach, only the grain envelope is directly represented whilst the internal solid morphology is simplified. In 2020 an experimental method was published [1] allowing to establish a direct link between the nucleant particle size distribution and grain refining efficiency of AlTiB and AlTiC grain refiners during casting of 7xxx alloys. In this work we aim at developing a cellular automaton model that can use as input the experimentally measured grain refiner nucleant particle size distribution and predict the grain refining efficiency during casting of aluminium alloys.

Model Description The model used in this work is based on the latest CAFE model renovation CAFE2G of ProCAST [15], originally developed at EPFL [8]. The model supports parallel computations, it uses a grid of cells to track the nucleation and growth of grains nucleating either at the mould surface or in the bulk. Nucleation is assumed to be heterogeneous and athermal. The model incorporates the main mechanisms that determine the final grain size: the size and spatial

975

distribution of nucleant particles, the solute diffusion field around growing grains, thermal recalesence and site capture by SSNZ. The model has been designed to be compatible with a resolution of heat flow on the process scale. For the purpose of this work, the model was revised and a new version was created allowing to take into consideration the principal physical phenomena affecting grain refiner performance: 1. Experimentally measured nucleant particle size distribution of TiB2 and TiC based grain refiners 2. Growth kinetics a. Globular growth b. Dendritic growth c. Globular to dendritic transition More details about the model can be found in [15]. The details for the physical laws used in the latest version of the model are described elsewhere [16]. In this paper, we focused on one 7xxx alloy refined at an addition rate of 1 kg/t with two grain refiners on AlTiB and another AlTiC. The 7xxx alloy properties considered for the presented calculations are deduced from a pseudo binary approximation of the alloy taking into consideration all the alloying elements Zn, Mg, Cu, Zr, Fe, Si, Ti, Mn. The parameters introduced to the model are: • • • • • • • • •

Liquidus temperature Composition Diffusion coefficient of the elements in the liquid state Liquidus slope Partition coefficient Eutectic temperature Latent Heat Heat capacity at constant pressure Gibbs Thomson Coefficient

Experimental Data Analysis To validate the model, we used the experimental procedure explained in TMS2020 [14]. Solidification experiments were carried out in the Cold Finger grain refiner test apparatus on both Al-5Ti-1B and Al-3Ti-0.15C grain refiners. In order to simulate precisely the solidification conditions, the characterisation of temperature distribution in the cold finger test during solidification of 7xxx alloys was also carried out (Fig. 1). The TiB2 and TiC particle size distributions were also quantified using a scanning electron microscope (SEM). Commercial grain refiner (cylindrical) rods were cut

976

G. Salloum-Abou-Jaoude et al.

Fig. 1 Image to the left shows the location of three thermocouples for thermal analysis of the Cold Finger solidification test. The image on the right shows the temperature curves during solidification of a 7xxx alloy with a liquidus at around 630 °C

Fig. 2 Graph to the left shows size distribution of TiB2 and TiC particles in commercial AlTiB and AlTiC grain refiners. The green rectangle shows the range identified using SEM on non-etched samples. The images to the right show examples of the microstructures and the segmentation of the quantified particles on the macro-etched samples

longitudinally along the central axis and polished, then examined under SEM. In order to have precise particle size description, the samples were quantified once under SEM in the as-polished state, and then a second under time after macro-etching in order to reveal the particle agglomerates. The procedure is detailed elsewhere [14], and the results are presented in Fig. 2. The grain refiner particle size distributions were later fit with lognormal laws. Both the lognormal fit parameters and the temperature curves deduced experimentally were used as inputs in the grain refiner numerical model to predict the grain size after solidification.

Results In this paper, we focused on calculations done under 0.8 K/s cooling rate simulating the Cold Finger test at a distance of 25 mm from the top. A 7xxx alloy was refined at an addition rate of 1 kg/t with two grain refiners, one AlTiB and another AlTiC. First, the effect of grain refiner nature will be presented in a globular kinetic growth mode. Then a globular to dendritic criterion will be introduced into the model and its effect on grain size will be investigated and compared with measured

A Cellular Automaton Model for Qualifying Current Grain Refiners …

experimental results. In the end, the effect of particle agglomerates will be discussed and its introduction into the model will be described and future perspectives are presented.

Globular Growth Kenetics In this section, a round of numerical calculations is presented. The calculations are done considering globular growth kinetic for the grains. The simulated box is 0.3 cm3. Figure 3 presents the cooling curves (full lines) and the grain number (dashed lines) during solidification. One can see that a lower undercooling is reached with TiC grain refiner that that of TiB2 grain refiner. This observation is awaited since TiC grain refiner has smaller average particle size and narrower distribution than that of TiB2 Fig. 2.

977

The final grain size is deduced and compared to the experimentally obtained grain size in Fig. 4. It seems that the calculations predict well the tendency of grain refinement for TiB2 and TiC. Furthermore, the TiB2 calculation result seems to be an excellent prediction of the grain size since the calculated value is in the range of the experimental error bar. For TiC refinement, the calculated grain size is outside the experimental error bar.

Globular to Dendritic Transition In order to be more precise, the authors investigated a criterion for switching the growth of the grains from globular to dendritic kinetics [16]. A spherical grain destabilizes when its radius exceeds a value multiple of the critical radius of free growth. Thus, based on this hypothesis, the model was

Fig. 3 Cooling curves (full lines) of the calculated 7xxx solidification refined with TiB2 and TiC based grain refiners. The dashed lines indicate the number of grains as a function of time

We can see from Fig. 3 that the number of grains starts increasing just before reaching the lowest value of undercooling for both TiB2 and TiC cases. The number of grains reaches its maximum at the maximum of undercooling at which point recalescence happens and no further nucleation events happen. This number of grains depicts the final grain size after solidification, we can already see that for this alloy at the calculated cooling rate using the globular growth kinetic, TiB2 grain refinement seemed to be more efficient that TiC.

developed to take into consideration the transition from a globular to a dendritic form. Figure 5 illustrates the adapted globular to dendritic transition criterion. The dotted and full lines represent respectively the critical radius of globular to dendritic transition (Rgd) for both TiB2 and TiC based grain refiners. The dashed lines represent the maximum calculated grain size (Rmax) during globular solidification. We can see that for the TiB2 case, the Rdg and Rmax curves never cross, that means that the Globular to Dendritic criterion is never met and the

978

Fig. 4 Comparison between the calculated and the experimental final average grain size of both TiB2 and TiC refined cases taking into consideration only globular growth kinetics

G. Salloum-Abou-Jaoude et al.

Fig. 6 Comparison between the calculated and the experimental final average grain size of both TiB2 and TiC refined cases taking into consideration the Globular to Dendritic transition

globular to dendritic growth kinetics when the local conditions are met. Figure 6 shows that the calculated grain size for the case of TiB2 does not change when comparison with Fig. 4. This is expected since Fig. 5 shows that Globular to Dendritic transition is never attained for this calculated case with TiB2. On the other hand, in the case of TiC, the calculated grain size drops down to 190 µm from the initially calculated 220 µm (Fig. 4) when only globular growth is considered. This shows that taking into consideration the globular to dendritic transition and switching the growth kinetic during growth is key for precise grain size prediction.

Conclusions In this work, a new version of cellular automaton CAFE model was developed. This allowed to take into consideration: Fig. 5 Globular to dendritic transition criterion illustrated by the intersection of the Rgd curve with the Rmax for both grain refiners

growth will continue in globular fashion for the TiB2 case. One the other hand, for the TiC case, the Rgd and the Rmax curves meet just before the solidification time of 1 s. This means that the globular grains start to destabilise and grow following dendritic growth kinetics, this affects the nucleation phenomena still active since recalescence for the case of TiC did not yet start at this time (Fig. 3). The model was adapted to include this transition and grain by grain the growth kinetic was switched from

• Experimentally measured grain refiner particle size distributions. • The solidification conditions of the Cold Finger grain refiner test. • The globular to dendritic transition criterion. The numerical model showed that the predicted grain size of 7xxx alloy solidification experiments refined with TiB2 and TiC based grain refiners have good match with experimental data. Including in the model a globular to dendritic transition improved the quantitative grain size prediction especially for the TiC based grain refiner.

A Cellular Automaton Model for Qualifying Current Grain Refiners …

The foreseen steps are: • Adapt the model to take into consideration the physics behind nucleation and growth of grains on particle agglomerates • Exploit this model to: Discriminate between low and high efficiency grain refiners. Design next-generation grain refiners with superior efficiency.

References 1. Jackson KA, Hunt JD, Uhlmann DR, Seward III TP (1966) Transactions of The Metallurgical Society of the American Institute of Mining, Metallurgical and Pet. Trans TMS-AIME 236:149–58 2. Nguyen-Thi H, Reinhart G, Mangelinck-Noël N, et al (2007) In-Situ and Real-Time Investigation of Columnar-to-Equiaxed Transition in Metallic Alloy. Metall Mater Trans A 38:1458–1464. https://doi.org/10.1007/s11661-007-9170-1 3. Rappaz M, Drezet J-M, Gremaud M (1999) A new hot-tearing criterion. Metall Mater Trans A 30:449–455. https://doi.org/10. 1007/s11661-999-0334-z 4. Moldovan P (2007) Microporosity Formation in DC cast 5083 Alloy. TMS Light Met 733–737 5. Wang C, Beckermann C. Mat Sci Eng A 171:199 6. Appolaire B, Combeau H, Lesoult G. Mat Sci Eng A 487:33 7. Wu M, Ludwig A (2009). Acta Mater 57:5621

979 8. Gandin C-A, Desbiolles J-L, Rappaz M, Thevoz P (1999) A three-dimensional cellular automation-finite element model for the prediction of solidification grain structures. Metall Mater Trans A 30:3153–3165. https://doi.org/10.1007/s11661-999-0226-2 9. Boettinger WJ, Warren JA, Beckermann C, Karma A (2002) Phase-Field Simulation of Solidification. Annu Rev Mater Res 32:163–194. https://doi.org/10.1146/annurev.matsci.32.101901. 155803 10. Takaki T, Sakane S, Ohno M, et al (2018) Competitive grain growth during directional solidification of a polycrystalline binary alloy: Three-dimensional large-scale phase-field study. Materialia 1:104–113. https://doi.org/10.1016/j.mtla.2018.05.002 11. Greer A, Bunn A, Tronche A, et al (2000) Modelling of inoculation of metallic melts: application to grain refinement of aluminium by Al–Ti–B. Acta Mater 48:2823–2835. https://doi.org/ 10.1016/S1359-6454(00)00094-X 12. Quested T, Greer A (2004) The effect of the size distribution of inoculant particles on as-cast grain size in aluminium alloys. Acta Mater 52:3859–3868. https://doi.org/10.1016/j.actamat.2004.04. 035 13. Easton M, StJohn D (1999) Grain refinement of aluminum alloys: Part I. the nucleant and solute paradigms—a review of the literature. Metall Mater Trans A 30:1613–1623. https://doi.org/10. 1007/s11661-999-0098-5 14. Salloum-Abou-Jaoude G, Jarry P, Celle P, Sarrazin E Effect of Nucleant Particle Size Distribution on the Grain Refining Efficiency of 7xxx Alloys | SpringerLink. https://link.springer. com/chapter/https://doi.org/10.1007/978-3-030-36408-3_134. Accessed 9 Sep 2022 15. Jacot A (2020) A cellular automaton approach for the prediction of grain size in grain refined alloys. IOP Conf Ser Mater Sci Eng 861:012061. https://doi.org/10.1088/1757-899X/861/1/012061 16. Rappaz M, Dantzig JA (2016) Chapter 11: Macro-and microstructures. In: Solidification, 2nd ed. pp 489–491

Modelling Contactless Ultrasound Treatment in a DC Casting Launder Christopher Beckwith, Georgi Djambazov, and Koulis Pericleous

Abstract

Ultrasonic processing can be performed without a vibrating probe by electromagnetic induction with a suitable frequency where resonance conditions can be established. This contactless method is suitable for high-temperature or reactive metal alloys providing purity of the melt and durability of the equipment. Hydrogen bubbles coming out of solution grow by rectified diffusion, and larger bubbles escape from the top surface leading to degassing. Violent collapses of the remaining smaller bubbles help grain refinement. In this study, the application of a contactless ‘top-coil’ device to continuous casting via a launder is considered. Resonance is achieved by the positioning of baffles on either side of the coil. Electromagnetic forces also cause strong stirring, increasing residence time. The process is modelled using time domain and frequency domain methods, and results for the proposed setup are compared with data obtained for the immersed sonotrode. Accuracy and sensitivity to process and model parameters are discussed. Keywords





Ultrasonic melt processing Numerical modelling Acoustic cavitation Acoustic resonance




Introduction Techniques of improving the quality of light alloy metal billets are of high importance, as reducing trapped hydrogen through degassing, grain refinement, and dispersion of metal clusters have been linked with improvements in mechanical C. Beckwith (&)
G. Djambazov
K. Pericleous Computational Science and Engineering Group, University of Greenwich, 30 Park Row, London, 10 9LS, UK e-mail: [email protected]

properties including tensile strength and ductility [1]. One such method that has been the subject of a significant amount of recent research is the application of ultrasound (UST) while the melt is still in its liquid phase before casting [2]. Traditionally this is performed with the use of an immersed mechanical vibrating sonotrode [2, 3], and the high acoustic pressures directly next to the sonotrode surface result in the rapid growth and then collapse of bubbles through inertial cavitation. Unfortunately, inertially cavitating bubbles attenuate the sound field through thermal and viscous losses, in addition to acoustic radiation, which prevents cavitation from happening further away from the sonotrode surface. An alternative approach which has been the subject of recent research has been to replace the mechanical sonotrode with an AC induction coil [1, 4–6]. This approach relies on resonance to build suitable pressures, and as a result can trigger cavitation deep into the melt, far away from the liquid surface. This could have a number of advantages, including the repositioning of the active zone for maximum processing efficiency (for example, just above the liquid–solid interface during casting, or through the creation of multiple active zones at the antinodes of a standing wave). This is in addition to other, already well established benefits of contactless processing which include the reduction of contamination due to sonotrode erosion, which eventually also results in reduced cost as traditional processing requires frequent sonotrode tip replacements [6]. In addition, contactless treatment also allows for the processing of high temperature (Ni, Fe, Cu) or reactive (Ti, Zr) melts which cannot be processed using a mechanical sonotrode. However, existing work mostly implements contactless UST in small scale experiments in a crucible, and work implementing the treatment in a practical casting process is limited. One study [4] performed initial numerical simulations demonstrating how contactless UST might work with the coil placed directly in the hot top of a DC caster, and showed that it was possible to establish a fixed resonant frequency during casting, due to the impedance mismatch at

© The Minerals, Metals & Materials Society 2023 S. Broek (ed.), Light Metals 2023, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-031-22532-1_131

980

Modelling Contactless Ultrasound Treatment in a DC Casting Launder

the liquid–solid interface. Electromagnetic stirring then provides suitable mixing, transporting dendrite fragments which might lead to an evenly refined microstructure. An alternative to processing directly on the hot top is to instead process further up in the launder. Previously, this has been investigated [3, 7, 8] for a mechanical sonotrode, and microstructure grain analysis has shown that this approach can result in more evenly refined grains. Processing in the launder is also less prone to macrosegregation caused by the acoustic streaming jet, and could potentially also be induced by the similar electromagnetic jet induced with the contactless method. This work attempts to develop this idea further, with full 3D simulations of the fluid flow and solidification, and for the first time attempting to apply contactless UST further up in the casting process, directly in the launder instead of in the hot top.

Problem Description DC casting of aluminium alloy billets has been carried out at the Advanced Metal Casting Centre (AMCC) of the Brunel Centre of Advanced Solidification Technology (BCAST). These experiments have so far used mechanical sonotrodes, as described in Fig. 1a. These casts used AA6XXX series aluminium with an addition of 0.25 wt% Zr but without the addition of an AlTiB grain refiner. The diameter of the cast billets measured 152 mm. Results from these experiments have previously been presented in [8] with a 5-kW magnetostrictive transducer (Reltec) driven at 17.3 kHz used to power the sonotrode, which has a diameter of 40 mm. The sonotrode tip was immersed 12 mm below the melt surface, oscillating with a peak-to-peak amplitude of 30 lm. Grain analysis showed that by applying UST in the launder, grain size decreased twofold and the presence of feathery grains was suppressed. The presence of partitions in the launders was also linked with increased acoustic pressures and more efficient UST, with the maximum acoustic pressure being twice as high as that without partitions, and RMS pressures increasing by 50%. This was linked with an additional grain refinement of approximately 10% compared to the same experiments without partitions. For simulations investigating the potential effect of replacing the sonotrode with an AC Induction coil, the partitions also increase the potential for resonance by adding additional geometry for sound reflection, as the location of these partitions can be modified to target particular acoustic frequencies. Numerical simulations are carried out assuming the same experimental conditions and partition configuration as the experiments with a sonotrode, but with the induction coil immersed into the liquid as shown in Fig. 1b.

981

Modelling Approach A number of authors have presented models for the frequency domain calculation of the acoustic field including the effect of inertially cavitating bubbles [9–11]. In our previous paper [5], the model of [9] was modified and a nonlinear Helmholtz equation for cavitation problems including the effect of a background source term F I representing the Lorentz force was obtained. A description of the method is given in this section, but for a full description including derivation, please see the cited work. The equations governing the propagation of the sound field are given in Eqs. (1) and (2), where P represents the complex acoustic pressure field, and k2m is a modified wave number due to the attenuation and change of speed of sound that exists in the presence of inertially cavitating bubbles. The nonlinear Helmholtz equation includes terms which allow for the variation in density at material boundaries (e.g. containing walls), in addition to the density change that occurs during solidification.
 1 FI k2 r rP  ð1Þ þ mP¼0 q q q k2m ¼

x2 c



AðPÞ B ð PÞ i j Pj jPj

ð2Þ

The dispersion coefficients A and B can then be calculated using Eq. (3) [9], which considers only the change in void faction over the last acoustic period. q x2 A¼ l p

Z

2p 0

@b q x2 sinsds; B ¼  l @s p

Z

2p 0

@b cossds @s ð3Þ

where b ¼ 4=3pr 3 N is the void fraction and N is the number of bubbles. The void fraction must be computed with a bubble dynamics simulation, and many choices could be suitable for solving the time evolution of bubbles, for example [12, 13]. Here, the Keller-Miksis Equation (KME) as given in [14] is used due to first order compressibility and acoustic radiation terms and is shown here in Eq. (4).



 _ R_ € þ 3 R_ 2 1  R RR C 3c 2
 R_ 1 Rd ¼ 1þ þ ½pl  pðtÞ c ql c dt

1

ð4Þ

where pl represents the liquid pressure at the liquid gas interface and is defined by Eq. (5), where re is the surface tension, l is the liquid viscosity, and pg is the pressure in the

982

C. Beckwith et al.

Fig. 1 a A typical setup for ultrasonic melt treatment in the launder using a mechanical sonotrode. b The alternative setup uses an AC induction coil. Partitions are placed in the launder for both cases

gas at the interface, which can be assumed to follow the adiabatic equation of state [15] given by Eq. (6). 2re 4lR_  R R
3c R0 pg ¼ pg0 R

pl ¼ pg 

ð5Þ ð6Þ

Here, c ¼ 1:4 is the polytropic exponent, and pg0 the initial gas pressure in the bubble. A background pressure pðtÞ ¼ p0 ð1  AsinðxtÞÞ accounts for both the sinusoidal acoustic pressure with dimensionless amplitude A, and atmospheric pressure. It is important to run the single bubble model for more than one acoustic period so that simulation can converge to a harmonic solution, and the number of cycles needed increases with frequency and driving amplitude [9]. In the simulations in this paper, 500 cycles are chosen as the cut-off point at which if a harmonic solution has not been obtained, interpolation with cubic splines is used instead. N is assumed to be a smoothed stepwise function W ðjPjÞ centred on the Blake pressure, with a smoothing distance equal to Pblake  Prd , the difference between the Blake threshold and the rectified diffusion threshold, the acoustic pressure required for bubbles to begin growing. Under this pressure, hydrogen bubbles begin dissolving back into solution and do not significantly influence the acoustic field. Assuming that the driving frequency is far from the bubble resonant frequency, which is generally the case in metal processing. The resulting threshold is given by Eq. (7) [5], where C i is the concentration of hydrogen in the bulk fluid, C 0 is the saturation concentration, g is the polytropic coefficient. i 2 1  x2 =x20 þ b2 ðx2 =x20 Þ ð1 þ 2r=R0 P1  Ci =C0 Þ

  ¼ ð3 þ 4KÞðCi =C0 Þ  34 ðg  1Þð3g  4Þ þ ð4  3gÞK ð1 þ 2r=R0 P1 Þ 

P2rd

qR20 x20

2 h 

ð7Þ

For the fluid flow simulation, Eqs. (8) and (9) describe the continuity and momentum equations. An additional term F s is included to model the additional stirring due to the influence of a background field. For the top mounted induction coil, this term represents the induced Lorentz forces, the mean component of which drives the main fluid flow. For an immersed sonotrode, this term can be used to represent acoustic streaming, which has been shown in previous work [3, 7, 8]. The Lorentz forces induced by the coil are calculated from a separate simulation using Comsol’s Magnetic Fields solver. Solidification at the mould is then included through the use of a continuum approach described in [16] and previously used in [7]. The continuum approach adds a Carman–Kozeny momentum sink term Sd that forces the fluid velocity to the background velocity vref . . This was implemented using a custom OpenFOAM solver based on the included “buoyantPimpleFoam” solver (a combination of the PISO and SIMPLE algorithms). The electromagnetic source terms were exported from Comsol in CSV format and then interpolated onto the OpenFOAM mesh. r
v¼0 q0

ð8Þ

@v þ q0 r
ðvvÞ ¼ rp þ l0 r2 v þ q0 gbT ðT  T ref Þ @t þ Sd þ F s ð9Þ

where q0 and l0 are the fluid density and the dynamic viscosity. Turbulence is included in the model, using the k-Omega-SST turbulence model. The system is closed by the energy balance, as given in Eq. (10). qcp

@T þ qcp r
ðvT Þ ¼ r
ðkrT Þ @t

@gl þ r
ðvgl Þ  q0 Lf @t

ð10Þ

Modelling Contactless Ultrasound Treatment in a DC Casting Launder

where cp is the specific heat, T is the temperature, and k is the thermal conductivity, Lf the latent heat of fusion, and gl the volume fraction of liquid. In the slurry, the effective dynamic viscosity leff can be calculated from the Stefanescu formula [18, 19] given in Eqs. (11, 12), where ll is the liquid viscosity, f s is the solid fraction, and f c is the dendrite coherency point, chosen to be 0.3.
leff ¼ ll

1 1  F l f s =f c

2

1

F l ¼ 0:5  ð1=pÞtan ð100ðf s  f c ÞÞ

ð11Þ

hc ¼

 0 1=3 1:67  105 þ 704T
Q ;  0 1=3 ðDT x Þ3 1:67  105 þ 704T
Q þ 20:8DT ;

Table 1 Model properties of liquid aluminium. Properties obtained from [15] Electrical conductivity (S/m)

4e6

Relative permeability

1

Relative permittivity

1 −1

Casting velocity (m s )

0.0023

Inlet temperature (K)

1003.15

Liquidus temperature (K)

929.2

Solidus temperature (K)

757.4

−1

Latent heat (J kg )

ð12Þ

At the water spray, the heat transfer can be described by a Fourier boundary condition with a heat flux function depending on the average temperature T between the surface of the billet and the bulk fluid [21]. Including the effect of nucleate boiling above a critical point qc ¼ 3910DT 2:16 the heat transfer coefficient takes the form of Eq. (13). (

983

375,696.0

Density (kg m−3)

2375 −1

Speed of sound (m s )

4600

Thermal expansion coefficient (K−1)

2.3  10−5

Kinematic viscosity (m2 s−1)

5.5  10−7

Table 2 Properties of the copper induction coil Electrical conductivity (S/m)

5.998e7

if qc  qi

Relative permeability

1

if qc \qi

Relative permittivity

1

ð13Þ

−3

Density (kg m )

8700

At the free surface, a surface radiation boundary condition is used and is given by Eq. (14), where
¼ 0:3 is the surface emissivity and r ¼ 5:6708  108 the Stefan– Boltzmann constant.   r T 4amb  T 4 rT ¼ ð14Þ k

Results Following the solution procedure described in [5], target acoustic Eigenfrequencies are first calculated from a linear model (achieved by setting N = 0), and are then used as an initial guess for the resonant frequency including the effect of cavitation. The frequency is then adjusted until resonance is achieved in the non-linear case. Target frequencies were taken only if they were below 30 kHz, to target specific modes that could be obtainable by the AC coil. A full list of material properties for the liquid aluminium is given in Table 1, and the properties of the copper induction coil are given in Table 2. In the acoustic simulation, a sound soft (P ¼ 0Þ boundary condition is used on the top surface, and sound hard boundaries are used elsewhere. At the inlet and outlet of the launder, a perfectly matched layer was used to prevent reflections, in an attempt to focus on resonant modes present due to partitions and to prevent unnatural resonant

Fig. 2 Obtained Eigenfrequencies from linear theory potentially obtainable by the AC coil. a, b The alternative setup uses an AC induction coil. Partitions are placed in the launder for both cases

frequencies that might occur due to not simulating the entire launder. Figure 2 shows two chosen eigenfrequencies for the launder with partitions. The two Eigenfrequencies occurred at 21564 and 23862 Hz, corresponding to electrical frequencies of 10782 and 11931 Hz and high pressure regions were located close to the partitions. For the rest of the results in this work, the 21564 Hz mode will be the target mode that we will use. The nonlinear Helmholtz solver is then used to compute the acoustic field in the presence of cavitating bubbles. The

984

C. Beckwith et al.

Table 3 Hydrogen bubble properties [15] Bubble density N

5e7

Surface tension (N m−1)

0.860

Vapour pressure (Pa)

0

Specific heat capacity (J kg–1 K−1)

717

Bulk temperature (K)

1013.15

Ambient bubble radius R0 (m)

10  10−6

properties of these hydrogen bubbles are given in Table 3. Due to the reduction in speed of sound that occurs with an increase in bubble volume fraction, the resonant frequency is slightly lower than that of the eigenfrequency study, and the closest frequency is calculated to be 21320 Hz, a shift of 244 Hz. Results for this case are shown in Fig. 3b. Peak pressures induced by the coil were approximately 180 kPa, well above the threshold for cavitation for 10 micron hydrogen bubbles in liquid aluminium (calculated to be approximately 154 kPa [5]). The traditional mechanical sonotrode is capable of reaching much higher pressures, up to 600 kPa directly below the sonotrode surface, but the active region is isolated to the area directly under the sonotrode, with acoustic shielding preventing higher pressures elsewhere in the domain. This can be seen in Fig. 3a. The top coil by comparison resulted in an active zone much deeper below the surface, located in the gap between the downstream partition and the base of the launder. Figure 4 shows the induced magnetic field from the AC induction coil at an electrical frequency of 10660 Hz. The coil interacts with induction currents in the melt to repel the free surface, ensuring that no contact is made with the metal. At the electromagnetic skin layer, the magnetic flux density reaches a peak of 0.11 T, comparable to that in previous work in a crucible [5]. The interacting magnetic and electric fields result in a force on the aluminium F ¼ J  B. This force has a mean component given in Fig. 4b which drives

Fig. 3 The induced acoustic field including the effect of inertially cavitating bubbles, using an immersed sonotrode operating at 17300 Hz (a) and a top mounted induction coil operating at 10660 Hz (an acoustic frequency of 21320 Hz) (b). The black line in both figures indicates the active processing zone above the Blake threshold

the bulk fluid flow, and is used as the source term F s in Eq. (9), and a harmonic oscillating component given in Fig. 4c, which is responsible for the acoustic field source term F I . The magnitude of the mean part was found to be lower than the amplitude of the harmonic part, with peak amplitudes of the order 2e6 and 6e6 respectively. The fluid flow induced by the sonotrode reached a peak velocity of approximately 1 m/s directly below the sonotrode, with the acoustic streaming jet having a velocity closer to 0.6 m/s. Flow patterns for this case can be seen in Fig. 5a. The top coil on the other hand, which does not rely on acoustic streaming but instead the electromagnetic Lorentz forces, reached a peak velocity of approximately 2.5 m/s, 2.5 what was obtained by the sonotrode. This results in very strong mixing and is likely to result in a very even distribution of processed particles, which has previously been linked with improved uniformity in grain size across the final billet [7]. This could then further be improved by designing a geometry such that the resonant mode, and therefore the active processing region, be located to increase residence time before mixing, as in these simulations the active zone primarily acts downstream, which could limit the effectiveness of treatment. The resulting sump profile showing the solidification of the billet in the mould and temperature contours are given in Fig. 5c.

Integrated Cylindrical Vessel An alternative geometry using a cylindrical vessel integrated into the launder is presented in this section. This concept hopes to improve the efficiency of processing by encouraging a central resonant mode, with partitions placed to restrict flow in and out of the isolated cylindrical vessel. A diagram of this concept is given in Fig. 6. Initial simulations were carried out on the cylindrical vessel, carried out in 3 stages. Initially, an electromagnetics solver computes the magnetic and electric fields, then the

Modelling Contactless Ultrasound Treatment in a DC Casting Launder

985

Fig. 4 a Induced magnetic field from the AC induction coil at 10660 Hz, in addition to the magnitude of the resulting mean Lorentz force (b) and harmonic component (c)

fluid flow, followed by the acoustics. For the electromagnetics and free surface model, Maxwell’s equations in electromagnetic induction approximation are solved numerically via a formulation based on the electric field vectors (real and imaginary parts) and Biot-Savart integral [17]. The 3-turn top coil was modelled as 3 axisymmetric rings each carrying a current of 1700A at 14,580.4 Hz. This frequency was chosen as the lowest for which the periodic component of the Lorentz force induces acoustic resonance in the metal volume with a radius of 96.4 mm and a height of 100 mm. The coils are lowered from their initial positions where the lowest turn is 5 mm above the metal surface in steps of 1.25 mm and the computational mesh is deformed according to the mean pressure exerted by the Lorentz force until a total displacement of 17.5 mm (Fig. 7) is reached— just before the hydrostatic pressure of the liquid metal under the deformed free surface becomes too high for the generated Lorentz force to balance. Both the mean and the periodic components of the Lorentz force vectors are saved for the next stages. The flow of the liquid metal is then caused by the mean component of the Lorentz force. Figure 8 shows the result of

a CFD simulation with a k−e turbulence model carried out in the PHYSICA code. The calculated maximum velocity in the downward jet is 0.38 m/s, comparable to typical flow patterns induced by an immersed sonotrode. This is less than the 2.5 m/s jet obtained by the frequency domain simulation in the previous section, but this can largely be explained by the coil operating at 1700A as opposed to 2000A, which also reduces the induced Lorentz forces. Acoustic cavitation of any hydrogen bubbles formed in the melt is driven by the periodic part of the electromagnetic Lorentz force. It is simulated with a bespoke piece of software [5] capturing the time-dependent behaviour of both the acoustic field and the bubbles dispersed in the metal volume. 32,366 acoustic cycles were simulated with time step 1.75  10–7 s, taking about 5 h on a 32-processor workstation. The acoustic algorithm [18] needs a regular cartesian grid, so the depression of the free surface is not modelled within the acoustic part, but the total volume of liquid metal is retained. The equilibrium bubble radius, i.e. the radius without acoustic excitation, is assumed to be 7 lm and the bubble concentration is 1  107 m−1. Acoustically soft (zero pressure amplitude) boundary conditions were

986

C. Beckwith et al.

Fig. 5 The fluid flow induced by the acoustic streaming generated traditional mechanical sonotrode (a), Lorentz forces generated by the induction coil (b), and the sump profile in the mould with temperature contours (c)

Fig. 6 Launder concept which aims to improve contactless UST processing efficiency by integrating an additional cylindrical vessel into the launder

assumed on all sides of the liquid metal volume. The simulated acoustic field as shown in Fig. 9 (for RMS pressure p) shows a spheroidal zone with an approximate radius of 2 cm around the centre of the domain which is above the Blake threshold for inertial cavitation (130 kPa RMS in the

Fig. 7 Vertical central cross-section of the liquid metal volume and top coil showing the deformed free surface, electric field contours and amplitude of the periodic component of the Lorentz force vectors; ‘real part’ means in phase with the driving coil electrical current

Modelling Contactless Ultrasound Treatment in a DC Casting Launder

987

sustain resonance under the bubble action which alters the effective speed of sound in the liquid metal. This is important for practical applications as automatic frequency tuning will be needed.

Conclusions

Fig. 8 CFD result for the mean flow driven by the steady component of the Lorentz force; contours—turbulent viscosity, vectors—flow velocity

A coupled model for simulating the acoustic field in the presence of bubbles, electromagnetic induction from an AC coil, and fluid flow has been presented and has been applied to the potential application of contactless UST processing in the launder. It has been shown numerically that it is possible to obtain pressures required for processing as the Blake threshold was exceeded deep into the launder close to the partitions. This mode was obtained at 10660 Hz electrical frequency, corresponding to an acoustic frequency of 21320 Hz, which is obtainable by the coil. In addition, induced flow velocities were 2.5 higher in the launder when using the Lorentz force, as opposed to traditional UST processing with a mechanical sonotrode. This could have benefits for enhanced mixing and could be used to evenly distribute processed particles that would result in an even distribution of refined grains in the final billet. However, as this was not tested further theoretical and experimental work needs to be carried out to determine the exact effect that this will have on grain refinement. In addition, an alternative launder concept has been suggested that might be better suited for contactless UST. The new concept integrates a cylindrical vessel into the launder, and a resonant mode can then be calculated which results in a large active zone directly in the middle of the vessel. Induced flow velocities for this system are comparable to that obtained by traditional UST with a sonotrode but needed a higher acoustic frequency of 29132 Hz to obtain.

References

Fig. 9 Vertical and horizontal sections of the acoustic cavitation field showing the RMS acoustic pressure averaged over 0.1 s just before the end of the simulated 1.11 s

simulated case) and is the necessary condition for cavitation treatment of the melt. A Lorentz force frequency in the acoustic simulation was set at 29,132 Hz, but its amplitude was taken from the electromagnetic results. One can see the prescribed frequency is not exactly twice the electrical (as expected theoretically) and the slight shift is needed to

1. Dybalska, A.; Caden, A.; Griffiths, W.D.; Nashwan, Z.; Bojarevics, V.; Djambazov, G.; Tonry, C.E.H.; Pericleous, K.A. Enhancement of Mechanical Properties of Pure Aluminium through Contactless Melt Sonicating Treatment. Materials 2021, 14, 4479, doi:https://doi.org/10.3390/ma14164479. 2. Eskin, G.I.; Eskin, D.G. Ultrasonic Treatment of Light Alloy Melts, Second Edition; CRC Press, 2014; ISBN 978-1-4665-7798-5. 3. Subroto, T.; Eskin, D.G.; Beckwith, C.; Skalicky, I.; Roberts, D.; Tzanakis, I.; Pericleous, K. Structure Refinement Upon Ultrasonic Melt Treatment in a DC Casting Launder. JOM 2020, 72, 4071– 4081, doi:https://doi.org/10.1007/s11837-020-04269-3. 4. Tonry, C.E.H.; Bojarevics, V.; Djambazov, G.; Pericleous, K. Contactless Ultrasonic Treatment in Direct Chill Casting. JOM 2020, 72, 4082–4091, doi:https://doi.org/10.1007/s11837-02004370-7.

988 5. Beckwith, C.; Djambazov, G.; Pericleous, K.; Tonry, C. Comparison of Frequency Domain and Time Domain Methods for the Numerical Simulation of Contactless Ultrasonic Cavitation. Ultrasonics Sonochemistry 2022, 89, 106138, doi:https://doi.org/10. 1016/j.ultsonch.2022.106138. 6. Pericleous, K.; Bojarevics, V.; Djambazov, G.; Dybalska, A.; Griffiths, W.D.; Tonry, C. Contactless Ultrasonic Cavitation in Alloy Melts. Materials 2019, 12, 3610, doi:https://doi.org/10. 3390/ma12213610. 7. Beckwith, C.; Djambazov, G.; Pericleous, K.; Subroto, T.; Eskin, D.G.; Roberts, D.; Skalicky, I.; Tzanakis, I. Multiphysics Modelling of Ultrasonic Melt Treatment in the Hot-Top and Launder during Direct-Chill Casting: Path to Indirect Microstructure Simulation. Metals 2021, 11, 674, doi:https://doi.org/10.3390/ met11050674. 8. Subroto, T.; Eskin, D.G.; Beckwith, C.; Roberts, D.; Tzanakis, I.; Pericleous, K. Effect of Flow Management on Ultrasonic Melt Processing in a Launder upon DC Casting. In Proceedings of the Light Metals 2022; Eskin, D., Ed.; Springer International Publishing: Cham, 2022; pp. 649–654. 9. Trujillo, F.J. A Strict Formulation of a Nonlinear Helmholtz Equation for the Propagation of Sound in Bubbly Liquids. Part I: Theory and Validation at Low Acoustic Pressure Amplitudes. Ultrasonics Sonochemistry 2018, 47, 75–98, doi:https://doi.org/10. 1016/j.ultsonch.2018.04.014. 10. Louisnard, O. A Simple Model of Ultrasound Propagation in a Cavitating Liquid. Part I: Theory, Nonlinear Attenuation and Traveling Wave Generation. Ultrasonics Sonochemistry 2012, 19, 56–65, doi:https://doi.org/10.1016/j.ultsonch.2011.06.007. 11. Jamshidi, R.; Brenner, G. Dissipation of Ultrasonic Wave Propagation in Bubbly Liquids Considering the Effect of

C. Beckwith et al.

12.

13.

14.

15.

16.

17.

18.

Compressibility to the First Order of Acoustical Mach Number. Ultrasonics 2013, 53, 842–848, doi:https://doi.org/10.1016/j. ultras.2012.12.004. Plesset, M.S. The Dynamics of Cavitation Bubbles. Journal of Applied Mechanics 1949, 16, 277–282, doi:https://doi.org/10. 1115/1.4009975. Löfstedt, R.; Barber, B.P.; Putterman, S.J. Toward a Hydrodynamic Theory of Sonoluminescence. Physics of Fluids A: Fluid Dynamics 1993, 5, 2911–2928, doi:https://doi.org/10.1063/1. 858700. Keller, J.B.; Miksis, M. Bubble Oscillations of Large Amplitude. The Journal of the Acoustical Society of America 1980, 68, 628– 633, doi:https://doi.org/10.1121/1.384720. Harkin, A.; Nadim, A.; Kaper, T.J. On Acoustic Cavitation of Slightly Subcritical Bubbles. Physics of Fluids 1999, 11, 274–287, doi:https://doi.org/10.1063/1.869878. Voller, V.R.; Prakash, C. A Fixed Grid Numerical Modelling Methodology for Convection-Diffusion Mushy Region Phase-Change Problems. International Journal of Heat and Mass Transfer 1987, 30, 1709–1719, doi:https://doi.org/10.1016/00179310(87)90317-6. Djambazov, G.; Bojarevics, V.; Pericleous, K.; Croft, N. Finite Volume Solutions for Electromagnetic Induction Processing. Applied Mathematical Modelling 2015, 39, 4733–4745, doi: https://doi.org/10.1016/j.apm.2015.03.059. Lebon, G.S.B.; Tzanakis, I.; Djambazov, G.; Pericleous, K.; Eskin, D.G. Numerical Modelling of Ultrasonic Waves in a Bubbly Newtonian Liquid Using a High-Order Acoustic Cavitation Model. Ultrasonics Sonochemistry 2017, 37, 660–668, doi:https://doi.org/ 10.1016/j.ultsonch.2017.02.031.

Numerical Analysis of Channel-Type Segregations in DC Casting Aluminum Slab Keisuke Kamiya and Takuya Yamamoto

Abstract

In the direct chill casting of aluminum alloys, the stripe-shaped segregation called channel-type segregations is formed in the slab, but the mechanism of their formation is not clear, and the casting conditions under which the segregation is minimized have not been found. In this study, it is reported that a numerical simulation model for the segregation have been developed, and the segregation distribution in the Al–Mg alloy slab was numerically analyzed. As a result, the segregation similar to that observed in actual slabs was reproduced on numerical analysis. This simulation results showed that the channel-type segregations could be suppressed by colliding the strong down flow with the solidification front. Keywords

Aluminum analysis




DC casting




Segregation




Numerical

Introduction In the semi-continuous casting process called direct chill (DC) casting of aluminum alloys, the segregation of solute concentration occurs in the slab. Negative segregation is observed in the center region of the slab, and stripe-shaped

K. Kamiya (&) UACJ Corporation, 3-1-12 Minato-Ku, Chitose Nagoya, 455-8670, Aichi, Japan e-mail: [email protected]

segregation, called channel-type segregation is observed around the center region of the slab (Fig. 1). Since these segregations cause the changes in the mechanical and chemical properties of the final product, they have to be controlled. However, their formation mechanism remains unclear up to now. In response to such problems, we have attempted to clear the formation mechanism of channel-type segregation by numerical simulation [1]. Figure 2 shows the results of numerical analysis of solute concentration in DC casting billet. This is the first example in which the channel-type segregation was reproduced numerically in the DC casting model for aluminum alloys. This simulation results allow us to explain in detail how the channel-type segregation is formed during the casting process. Figure 3 shows the simulation results of Mg concentration around the solidification front in DC casting billet [1]. First, as the molten metal solidifies, a zone with high concentration of Mg (a in Fig. 3) is created at the solidification front, and the concentration of Mg causes occurrence of the upward solutal buoyancy flow along solidification front (b) resulting in the solidification delay (c). As a result, Mg is transferred from the low concentration zone toward the high concentration zone of delayed solidification due to Mg partition (d) between the solid and liquid phases. In this way, the low concentration zones (e) are formed below the high concentration zones. This segregation cycle is repeated, resulting in the formation of channel-type segregation. As described, the mechanism of channel-type segregation has been cleared, however, casting conditions to suppress its occurrence have not yet been found. Therefore, the aim in this paper is to find a casting conditions to suppress the channel-type segregation in DC casting slab of aluminum alloys by the numerical simulation.

T. Yamamoto Tohoku University, 6-6-02 Aza-Aoba, Aramaki, Aoba-ku Sendai, 980-8579, Miyagi, Japan e-mail: [email protected] © The Minerals, Metals & Materials Society 2023 S. Broek (ed.), Light Metals 2023, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-031-22532-1_132

989

990

K. Kamiya and T. Yamamoto

Numerical Simulation Method Governing Equation The numerical simulation model in this study is based on the model proposed by Vreeman et al. [2], and the modified model by Fezi et al. [3], which basically consists of four basic governing equations for the balance of momentum, mass, enthalpy and chemical species concentration with solidification.

Fig. 1 Schematic view of DC casting and observed segregation on etched slab sample

Fig. 2 Numerical analysis result of Al-2.5mass%Mg billet [1]

Fig. 3 Formation process of channel-type segregation shown by numerical analysis [1]

Numerical Analysis of Channel-Type Segregations in DC Casting Aluminum Slab

Srigid ¼  lKl qq ðu  us Þ

Each equation is given as
 @ ðquÞ ll þ rðquuÞ ¼ rp þ r
ru @t ql  qgðbT ðT  T 0 Þ þ bC ðC  C 0 ÞÞ

l

ð1Þ

þ ð1  PÞSSlurry þ PSrigid @q @t

þ r
ðquÞ ¼ 0


 @ ðqhÞ k þ r
ðquhÞ ¼ r
rh @t cps
 k þr
r ð hs  hÞ cps

991

ð2Þ

ð7Þ

where ls is the effective solid viscosity, f is the mass fraction, and K is the permeability, which is described by the following Blake–Kozeny model [6]. The solid velocity us is modeled as    aðqs ql Þd2g ð8Þ us ¼ ð1  PÞ u þ f l g þ Pucast 18l l

where dg is the diameter of floating solid particle, and ucast is the casting velocity. ð3Þ

Simulation Models

 r
ðqðu  us Þðhl  hÞÞ @ ðqC Þ @t

þ r
ðquC Þ ¼ r
ðal ql Dl rC Þ þ r
ðal ql Dl rðCl  CÞÞ  r
ðqðu  us ÞðCl  CÞÞ ð4Þ where q is the density, u is the flow velocity, t is the time, p is the pressure, l is the viscosity, g is the gravitational acceleration, bT is the volume thermal expansion coefficient, bC is the volume solutal expansion coefficient, T 0 is the reference temperature, C 0 is the reference composition, P is the packing variable, SSlurry is the source term for mushy zones, Srigid is the source term for solid zones, h is the enthalpy, k is the thermal conductivity, cp is the heat capacity of solid, C is the composition, a is the volume fraction, D is the diffusion coefficient, the subscripts s and l indicate the solid and liquid phases, respectively. The model used of the slurry and solid parts was that proposed by Plotkowski and Krane [4], Coleman and Krane [5]. The packing variable, P; can be written as      acs as P ¼ min max 1  Da ;0 ;1 ð5Þ s acs is the critical volume fraction of solid phase, and Das is the steepness of phase transition. If as is greater than 0 and less than acs , then P is 0. If as is greater than acs and less than 1, then P is 1. The source term for mushy and solid zones can be written by
 qfs rus þ r
ð
SSlurry ¼ r
ll ls ð1  aÞrus Þ  r 
 q l  qfs ð u  u s Þ ð u  us Þ
ql ð6Þ

The physical properties and operating parameters are shown in Table 1. In this numerical simulation, Al-5.0mass%Mg alloy was used to simplify the numerical simulation model. Table 2 shows the calculation conditions. In this study, we focused on distributor geometry and casting speed and investigated the effect of casting speed in calculation #1–#3 and distributor in calculation #A–#C. Figure 4 shows the schematic view of the distributors, calculation domain, and boundary conditions. In this study, we devised three different distributor models. Distributor model (A) is the default distributor model that supply molten metal to the width direction of slab. The geometry of distributor in model (B) and (C) is square and supply molten metal to width, thickness, and depth directions. Distributor model (C) has a smaller diameter inlet into which molten metal inflows than the other models. The sizes of each distributor model are shown in Table 3. As for the boundary conditions, in all distributor models, the distributor surface through which the molten metal passes was set to occur a pressure drop based on the Darcy law. Primary cooling and heat extraction from the bottom block were calculated based on Newton’s cooling law, and both using a constant heat transfer coefficient. Secondary cooling with coolant was also calculated based on Newton’s cooling law, and the heat transfer coefficient was set so that the value depends on the surface temperature of slab. Figure 5a shows the numerical grid used in this study. The slab size was 400  800 mm, to reduce the calculation time the calculation domain was limited to a quarter-symmetrical region of the slab. The size of grid cell in each direction was set to 5 mm. Figure 5b shows schematic drawing of dynamic grid motion. In DC casting, typical casting length of produced slabs is several meters, so in the present study, the dynamic grid motion was implemented by adding new grid cells.

992 Table 1 Physical properties and operating parameters for numerical simulation

K. Kamiya and T. Yamamoto Parameters

Marks

Values

Unit

Density of melt

ql

2350

kg/m3

Density of solid

qs

2650

kg/m3

Heat capacity of melt

cpl

1180

J/kg K

Heat capacity of solid

cps

1000

J/kg K

Thermal conductivity of solid

ks

140

W/m K

Kinematic viscosity

m

5.47  10–7

m2/s

Thermal expansion coefficient

bT

6.90  105

1/K

Solutal expansion coefficient

bC

3.20  10

Eutectic temperature

Te

723

K

–

Eutectic concentration

Ce

0.38

–

Partition coefficient

kp

0.47

–

Latent heat

L

389,000

J/kg

Gravitational acceleration

g

9.81

m/s2

Critical solid volume fraction

acs

0.30

–

Smooth parameter of packing fraction

Das

0.05

–

Secondary dendrite arm spacing

k

5.00  10–5

m

Diameter of floating particle

dg

7.50  10–5

m

Averaged solid viscosity

ls

6.45  10

Casting speed

ucast

50, 70, 90

Table 2 Calculation conditions

Fig. 4 Schematic view of the distributors, calculation domain, and boundary conditions

–1

–3

Pa s mm/min

Casting speed (mm/min)

Distributor model

Calculation #1

50

(A)

Calculation #2

70

(A)

Calculation #3

90

(A)

Calculation #A

70

(A)

Calculation #B

70

(B)

Calculation #C

70

(C)

Numerical Analysis of Channel-Type Segregations in DC Casting Aluminum Slab Table 3 Size of each distributor

993

Marks

Inlet (Umm)

Width thickness  depth (mm)

(A)

60

275  105  30

(B)

60

105  105  20

(C)

30

105  105  20

Fig. 5 Schematic view of a numerical grid and b dynamic grid motion

Results

Discussion

Figure 6 shows the numerical simulation result of time variation in temperature and Mg concentration distribution within a slab at calculation #2. It can be seen that stripe-shaped negative segregation is formed successively along the solidification front in the thickness section as casting proceeds. Figures 7 and 8 show the results of Mg concentration distribution within a slab for the same casting length, and the graph plots the normalized Mg concentration by the average concentration in the thickness section of each slab. In Fig. 7, it is compared the difference of Mg concentration distribution each casting speed. As far as the standard deviation of Mg concentration is concerned, the casting speed and channel-type segregation strength is tiny correlations, and it is not expected to suppress the segregation by controlling casting speed. In Fig. 8, the differences in Mg concentrations distribution for each distributor model are compared. The result of calculation #B has a smaller segregation concentration than calculation #A, but the segregation appears in a width cross section too. The result of calculation #C has an even smaller segregation concentration than calculation #B, and the segregation is suppressed in both the width and thickness cross sections.

Figure 9 shows the concentration distribution of Mg and flow fields in the molten sump. At calculation #A, it can be seen that the molten metal forms circular flow that flows from distributor to width direction of slab, next flows downward along the solidification front to center of slab, finally flows upward along the solidification front to thickness direction of slab. This circular flow allows molten metal to flow in the same direction as the upward solutal buoyancy flow at the solidification front in thickness direction. Therefore, the high Mg concentration zone is widely distributed at the solidification front, and the partitioning of solute is promoted, resulting in pronouncing the channel-type segregation in the thickness cross section. At calculation #B, molten metal flows in both the thickness and width directions, so circular flow as # A does not occur, but it can be seen that the downward flow separate at bottom of molten sump and exerts an upward flow in both the thickness and width direction. As a result, it is considered the high Mg concentration zone occur and the channel-type segregation was formed in both thickness and width cross section. At calculation #C, inlet size is smaller than the other models, resulting in a higher inflow velocity.

994

K. Kamiya and T. Yamamoto

Fig. 6 Time variation in temperature and Mg concentration at calculation#2

Fig. 7 a Mg concentration distribution within a slab at each casting speed, and b the graphs of the normalized Mg concentration by the average concentration in the thickness section of each slab

As a result, circular flow does not occur and downward flow against upward solutal buoyancy flow occurs throughout molten sump. Therefore, it is considered that the solute was diffuse in molten sump without collecting at the solidification front and the channel-type segregation did not occur in both thickness and width cross section.

Conclusion In this study, channel-type segregation in DC casting of aluminum slabs was reproduced by numerical analysis, and the following results were obtained.

Numerical Analysis of Channel-Type Segregations in DC Casting Aluminum Slab

995

Fig. 8 a Mg concentration distribution within a slab at each distributor model, and b the graphs of the normalized Mg concentration by the average concentration in the thickness section of each slab

Fig. 9 a Flow field in molten sump and b Mg concentration distribution

1. Channel-type segregation occurs at different locations in slab depending on the geometry of distributor. 2. Channel-type segregation is more pronounced in the high Mg zone formed by the molten metal upward flow at solidification front. 3. By reducing the area of molten metal inlet, the molten metal inflow velocity increase and strong downward flow occur throughout molten sump. Therefore, Mg concentration at the solidification front is suppressed and channel-type segregation can be reduced.

References 1. Yamamoto T, Kamiya K, Fukawa K, Yomogida S, Kubo T, Tsunekawa M, Komarov S (2021) Numerical Prediction of Channel-Type Segregation Formation in DC Casting of Al–Mg Billet. Metallurgical and Materials Transactions B. 52:4046–4060. 2. Christopher JV, Matthew JMK, Frank PI (2000) The effect of free-floating dendrites and convection on macrosegregation in direct chill cast aluminum alloys: Part I: model development. Int. J. Heat Mass Transf.43:677–686.

996 3. Fezi K, Plotkowski A, Krane MJM (2016) Macrosegregation modeling during direct-chill casting of aluminum alloy 7050. Numer. Heat Transf. A Appl.70:939–963. 4. Plotkowski A, Krane MJM (2016) A continuum grain attachment model for simulations of equiaxed solidification. Comput. Mater. Sci. 124:238–248.

K. Kamiya and T. Yamamoto 5. Coleman J, Krane MJM (2020) Influence of liquid metal feeding on the flow and macrosegregation in DC casting. Mater. Sci. Technol 36:393–402. 6. Stefanescu DM (2008) Science and engineering of casting solidification. Springer, New York.

Effect of Casting Variables on Mechanical Properties of Direct Chill Cast Aluminium Alloy Billets S. P. Mohapatra

Abstract

Microstructure of DC cast 6000 series Aluminium alloy billets developed during casting and homogenization largely determine the performance of end product. Apart from chemical composition, casting parameters like casting speed, casting temperature, and cooling water flow rate influence microstructure development during solidification. There is a need to understand the effect of each of these variables on the mechanical properties of a DC cast Aluminium alloy billet. In the present investigation, the effects of casting speed, cooling water flow rate, and casting temperature on Yield strength and work hardening behaviour of an AA6063 alloy billet have been discussed. An empirical relationship has been formulated to understand the impact of these operating parameters on mechanical properties of the casting. Keywords

Aluminium




Microstructure




Mechanical properties

Microstructure of a DC cast billet is largely determined by three factors: chemical composition, casting parameters, and homogenization treatment. The microstructure variables of interest are: grain size, dendrite arm spacing (DAS), amount, and morphology of intermetallic particles. Each of these variables determines hot deformation behaviour and mechanical properties of the part. The mechanical properties of most cast alloys are seen to be strongly dependent on SDAS (Secondary Dendrite Arm Spacing). SDAS is controlled by solidification time whereas grain size is dependent on many different factors like composition, nucleation rate, grain refiner addition, etc. Billet manufacturers give more importance to SDAS at mid radius as a quality parameter. The effect of casting parameters on the variation of SDAS and consequently on mechanical properties of AA 6063 DC cast billet are analyzed in the present study. The most vital factor for the formation of structure and quality of any casting is cooling rate. The spacing between secondary dendrite arms is a function of cooling rate alone. Cooling rate reflects the heat extraction rate during casting. CR ¼

Introduction Direct chill cast 6xxx series Aluminium alloys are widely used as extruded parts in construction and automotive sectors. Formation of the structure during casting and homogenization play a major role in determining the final product quality. Understanding various aspects of the manufacturing process and using them in developing qualitative and quantitative models has become a necessity. Such models must be able to predict the structure & properties of the cast product.

Tl  Ts Tf

ð1Þ

where Tl and Ts are liquidus and solidus temp, Tf is the time taken by the solidification front to move from the liquidus to the solidus line. The time taken by a single particle to travel in the mushy zone from liquidus and solidus isotherms can be traced from a thermal model. Generally the structure is refined by increasing the heat extraction rate, i.e., by increasing cooling rate. A lot of studies have been made on the microstructure of binary alloys in order to determine the inter dependence of DAS and solidification parameters [1–4].

S. P. Mohapatra (&) DGM (R&D), Nalco Research & Technology Center, Bhubaneswar, Odisha, India e-mail: [email protected] © The Minerals, Metals & Materials Society 2023 S. Broek (ed.), Light Metals 2023, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-031-22532-1_133

997

998

S. P. Mohapatra

Secondary dendrite arm spacing (k) which has an impact on the billet properties, is known to vary inversely with cooling rate during solidification as per formula [5, 6], k ¼ ATcn

ð2Þ

where ‘A’ is a constant for a particular casting technology, alloy, and size of casting. ‘Tc’ is the cooling rate. The exponent ‘n’ lies in between 0.2 and 0.4 for most of the Aluminium alloys [5, 6]. Many researchers have derived expressions in terms of local solidification time Tf as k ¼ CTfm

ð3Þ

The exponent ‘n’ used in Eq. 2 has been summarized for a few Aluminium alloys in the literature [7]. A number of studies have pointed out the effect of microstructure and particularly of dendrite arm spacing upon mechanical properties [8–13]. P. R. Goulart et al. [14] developed experimental-based expressions which correlate the ultimate tensile strength with SDAS for Al-Si hypoeutectic alloys. Their experiment was conducted with a 100 mm  190 mm  140 mm steel mold in a laboratory by taking Al—9 wt % Si alloys. The study of Devdas et al. [15] on the prediction of SDAS has been significant. They have shown the variation of SDAS with respect to casting speed, super heat of metal and size of billet through mathematical modelling. There are hardly any studies carried out to illustrate the effect of casting variables on mechanical properties of AA6063 DC casting. Experimental investigation is adopted in this work to find out relationship between casting variables with SDAS. Subsequently experiments were carried out to model the effect of casting variables on mechanical properties of the AA 6063 DC cast billet.

The pouring temp was varied from 695 °C to 725 °C, water flow was varied from 68 to 100 lpm, and casting speed from 102 to 126 mm/min during different casting campaigns, which were taken as experimental cases for this study. The following cases were experimented in order to identify the influence of process parameterTable 1. A typical dendritic microstructure and SDAS of a 6063 DC cast billet are shown Fig. 1. The sample for analysis of SDAS was taken from the mid radius point of a sound billet. The samples were prepared by following the standard procedures of metallography. Optical microscopic analysis was carried out to measure SDAS by line intercept method and individual counting method. Average value of 10 numbers of readings was taken as the reference value. This is a standard procedure followed by Wislei Riuper Osorio and others and Eduardo Netto De Souza and others [16, 17] to calculate SDAS. In order to understand the effect of casting parameters on mechanical properties of the billet five samples were collected from billets produced from five different sets of conditions mentioned in Table 2. The samples were tested under compression to determine the properties of the materials. The specimens are of cylindrical in the shape of 30 mm diameter and 40 mm height. The specimens were cut to size by turning, using kerosene as cutting fluid. All specimens before experimentation were annealed in the boiling water for a period of two hours. The dimension was so chosen that the cylindrical specimen should avoid buckling during testing. The specimen had oil grooves turned on both ends to entrap lubricant during the compression process. After annealing the specimen in boiling water for two hours, its ends were adequately lubricated with lithium-based grease lubricant (commercially available SKF LGMT 3IN1 general purpose grease).

Experimental The experimental work for the study was carried out at the Billet Casting Facility of National Aluminium Company Ltd (Nalco) Smelter plant, India. The three important casting process parameters which significantly affect the structure of billet are casting speed, metal pouring temp & water flow rate. In order to understand the quantitative relation between casting parameters and SDAS, experimental studies were conducted. The parameters were varied one by one & influence of each variable was quantified. Studying the effect of one process variable on structure could be possible only if other variables are kept constant during casting. The trials were conducted in different campaigns to generate 9 different cases for the complete study.

Experimental Determination of Stress–Strain Characteristics In order to plot the stress–strain diagram, specimens were tested under uniaxial compression on the INSTRON®600KN hydraulic pressing machine. To avoid the rate effect, the movement of the punch was adjusted to approximately 1 mm per minute. The compressive load is recorded at every 0.5 mm of punch travel. After compressing the specimen to about 10 mm it was taken out from the sub-press, re-machined to a cylindrical shape of diameter 30 mm with the oil grooves again turned on both ends and tested in compression. This process continued till the specimen was reduced to about 15 mm in height.

Effect of Casting Variables on Mechanical Properties of Direct Chill Cast Aluminium Alloy Billets Table 1 Experimental cases for microstructural study

Cases

Water flow, (lpm)

Casting speed, (mm/min)

Pour temp, (° C)

1

78

102

705

2

78

110

706

3

78

126

705

4

78

126

705

5

78

125

712

6

78

125

725

7

68

126

706

8

78

126

705

9

100

125

705

999

For different casting speed

For different pouring temp

For different water flow

Fig. 1 Micrograph of AA 6063 billet, a bright field SEM image identifying a grain, b dark field SEM image identifying SDAS

Table 2 Experimental cases for deformation study

Sample no

SDAS, (lm)

Speed V, (mm/min)

Water flow F (lpm)

Casting Temperature T, (° C)

1

13

138

135

690

2

15

137

120

702

3

16

127

130

702

4

18

126

100

698

5

20

126

78

705

Results and Discussion The variation of measured SDAS values with individual casting parameters are depicted Fig. 2. It is evident that as speed increases, cooling rate also increases proportionately. The rise in cooling rate raises heat extraction rate for a particular grain in the entire solidification zone. This prevents grain coarsening and ensures finer grain size. The dendrite arm spacing decreases as the grain size is finer.

Higher melt temperature will increase the size of SDAS as noticed from the above figure. If the casting temperature is very high, the liquid in the center of the billet will remain above the liquidus temperature for a long time. As a result, most of the crystals will melt again soon after they depart from the cast wall. Only the remaining crystals can extend near the cast wall to form the chill zone and a thin shell composed of randomly oriented small equiaxed grains will be formed on the surface of the billet. Consequently grains will remain longer time in liquid and coarsen.

1000

S. P. Mohapatra

SDAS vs Metal temp

SDAS vs Speed 27 25 23 21 19

R² = 0.9944

17 15 90 95100105110115120125130135140

(a) SDAS vs Speed

SDAS vs Water flow 23

22.5 22 21.5 21 20.5 20 19.5

R² = 0.9709

22 21 20 19

R² = 0.8459

18 700

710

720

730

(b) SDAS vs Temp

60

80

100

(c) SDAS vs Water flow

Fig. 2 Variation of SDAS with casting parameters

On the other hand, if the casting temperature is lower, all of the liquid will cool fast below the liquidus temperature and the crystals will move fast into the melt. Then a fine equiaxed billet structure will be produced. Again a caster will never want to run at higher melt temperature as it will increase the energy use and the risk of bleed out. The water flow has less effect on the maximum heat that can be extracted, since after a certain flow rate, the heat transport mainly depends on the conduction through solid portion of the billet rather than to the outside water. It is also observed that increasing water flow has less impact on cooling rate beyond a certain flow rate. SDAS is found to be decreasing when water flow rate is increased as the time spent by a particle in the mushy zone is reduced.

Regression Model It is seen from the relation between SDAS and casting parameters that SDAS varies linearly with all three parameters. The correlation coefficient in each case is around 0.9, which indicates a strong linear relationship. A multivariate linear regression analysis with a confidence level 95% has been taken up to correlate all the three casting variables with SDAS. Considering the fact that the rate of change of each parameter is independent of the other variable, the composite regression model developed is as per the following equation: SDAS ¼ 17:47  0:169  V  0:0993  L þ 0:0464  T ð4Þ where V, L and T are casting speed in mm/min, water flow in lpm and T is in °C respectively. The summary output of the analysis is presented in Table 3. The correlation coefficient of the analysis is 0.988. This indicates a strong relationship.

Table 3 Regression analysis summary output Regression statistics Multiple R

0.9940463

R2

0.9881281

Adjusted R2

0.9810049

Standard error

0.2430962

Observations

9

The measured values of SDAS and values calculated from the model are compared in Table 4. It can be seen from the table that the model is able to predict the SDAS value within a reasonable range of variation. This model is valid for a specific range of casting speed from 102 to 126 mm/min, pouring temp from 705 to 725 °C and water flow rate from 68 to 100 lpm. This study can be extended for large ranges of casting parameters. The casting speed is found to have the maximum impact on SDAS followed by the water flow and then metal casting temperature. The model was also tested for five different cases undertaken to study the deformation behaviour of the billets with respect to casting variables (Table 5). The standard errors are also seen to remain within acceptable range, so the model is validated. In metal forming, the plastic region plays an important role because metal is plastically and permanently deformed in this process. The stress–strain relationship for a metal exhibits elasticity below the yield point and strain hardening after that. In the plastic region the metal’s behavior is expressed by the equation; r ¼ A 2n

ð5Þ

The mechanical properties of the specimens tested in compression can be characterized by their yield stress r, their strength coefficient, (A) and work hardening exponent (n).

Effect of Casting Variables on Mechanical Properties of Direct Chill Cast Aluminium Alloy Billets Table 4 Comparison between measurement and model prediction

Table 5 Model validation

Experimental cases

SDAS as per regression model output (equation-4), (µm)

SDAS as per Measurement values, (µm)

Standard error with respect to regression model, (%)

1

25.13

25

0.79

2

23.82

23

3.73

3

21.06

20

5.40

4

21.06

20

5.40

5

21.56

21

2.9

6

22.16

22

1.08

7

22.10

22

0.82

8

21.06

20

5.40

9

19.05

19

0.66

Experimental cases

SDAS as per regression model output (Eq. 4), (µm)

SDAS as per Measurement values, (µm)

1

12.7585

13

1.857692

2

14.9738

15

0.174667

3

15.6708

16

2.0575

4

18.6332

18

−3.51778

5

21.1426

20

−5.713

The result is shown in Table 6. The characteristic equations for the different billets obtained from the compression test are summarized below. With SDAS 13 lm : r ¼ 241:0e0:19

ð6Þ

With SDAS 15 lm : r ¼ 229:0e0:16

ð7Þ

With SDAS 16 lm : r ¼ 195:0e0:14

ð8Þ

With SDAS 18 lm : r ¼ 193:0e0:13

ð9Þ

With SDAS 20 lm : r ¼ 163:0e0:10

ð10Þ

Table 6 Compression test result

1001

Standard error with respect to regression model, (%)

To determine the uniaxial yield stress of the billet material for any given reduction R, the corresponding strain imparted to the billet material during the compression process must be known beforehand. For the same reduction, the strain imparted to the billet during the practical forming processes is always higher than that imparted to the billet during compression because of the redundant work done during the forming processes. Following the principles laid by Johnson and Mellor [18], the above strain in the present case is calculated from the empirical relation,
 1 e ¼ 0:8 þ 1:5 In ð11Þ 1R

Sample no

SDAS, (lm)

Speed V, (mm/min)

Water flow F, (lpm)

Casting Temperature T, (°C)

Strength coefficient (A), (MPa)

Work hardening exponent, (n)

Yield strength, (MPa)

1

13

138

135

690

241

0.19

224.0

2

15

137

120

702

229

0.16

214.9

3

16

127

130

702

195

0.14

184.2

4

18

126

100

698

193

0.13

183.0

5

20

126

78

705

163

0.10

156.3

1002

S. P. Mohapatra

where R (for the present calculation, a severe situation, R = 0.45 was taken) is the fractional reduction. In Aluminum forming process, general reduction ratio 45% is considered as the limiting value. The average yield stress of the compressed billet is then calculated by dividing the area under the stress–strain curve with the corresponding value of the abscissa. The magnitude of the strain is calculated from Eq. (11) and the magnitude of the average yield stress ry for different specimens is determined in the above manner. Figure 3 shows the variation of the mechanical properties as a function of SDAS. It can be seen from the figure that microstructural refining enhances the properties of the material. The yield strength and the strength coefficient are highest in the sample with SDAS of 13 µm. The sample with largest SDAS has shown a work hardening coefficient as 0.1 and yield strength of 156.3 MPa. The findings are matching with the findings of Q. Wang for Aluminum alloy A 356 and A 357 [19]. It is clearly seen that the yield strength decreases as SDAS is increased. The refining of spacings and consequently the grains enhance the blocking resistance to slipping of dislocations which causes deformation. So a material having less SDAS will exhibit higher strength. The work hardening coefficient ‘n’ is found to be decreasing when SDAS is increased. This is also in agreement with the result of Defeng et al. [20] for different types of Aluminum alloys, A 319/356/357. Since ‘n’ defines the working range of a material, the deformation range of a material is shortened for a coarse grain size material. The formability of a material is dependent on strain gradients in the deformed metal. Strain gradient arises when deformation is not uniform. Since the most highly strained region will have the highest hardness, the load is passed on to the neighbouring elements for a metal with a high ‘n’. This forces them to strain more and by doing so the strain

True Stress, MPa

300

gradient is reduced [21]. As a result, higher formability can be achieved by the metal with higher ‘n’. The ductility of the metal is also influenced by SDAS. In a coarse grain with large SDAS, the eutectic particles become coarser and longer. A particular homegenisation treatment will have less impact on a coarse grain than fine grain structure material as the diffusion distance is increased in a coarse grain material. So large and elongated particles in the metal will exhibit rapid development of particle stress and thus low ductility [22, 23]. So increasing the cooling rate refines the grain and also the eutectic particles, thereby improving the strength as well as formability of the material. It is seen that the work hardening exponent, ‘n’ which influences the formability of material is influenced by SDAS and strength coefficient A, which are in turn affected by the casting parameters. Correlation between the deformation behavior of the material represented by the strain hardening coefficient and yield strength with its casting conditions, represented by SDAS is depicted in Fig. 4. The variation of work hardening exponent with SDAS and yield strength for the present investigation can be expressed by an equation as follows: n¼

r0:799 y 56:3  SDAS0:764

ð12Þ

In order to derive the value of yield strength with different SDAS and ‘n’ the following figure will be helpful (Fig. 5). The relationship is expressed in the following equation: ry ¼ 347:2  SDAS0:379  n0:845

ð13Þ

The ultimate aim of a caster is to select his casting conditions for the desired quality of the product. In order to correlate the casting parameters like casting speed, pouring temp, and water flow rate to the final objective of the product, i.e., strength and formability, regression analysis was done. An empirical model is proposed for the strength hardening coefficient and yield strength of the product.

200

100

SDAS=13µm SDAS=16µm SDAS=18µm SDAS=15µm SDAS=20µm

0 0.0

0.2

0.4

0.6

0.8

1.0

True Strain

Fig. 3 Stress–strain curve for five different samples

Fig. 4 Variation of work hardening coefficient with SDAS (µm) and Yield strength (MPa)

Effect of Casting Variables on Mechanical Properties of Direct Chill Cast Aluminium Alloy Billets

1003

Acknowledgements The author would like to sincerely thank Nalco management for allowing him to carry out the experiment in the plant and also giving permission to present the findings.

References

Fig. 5 Variation of ry (MPa) with SDAS (µm) and work hardening coefficient

Proposed Empirical Model This model is applicable to only a specific technology (Wagstaff Air-slip) of billet casting and for a specific range of variables experimented. The applicability can be translated to any technology and any range of variables through experiments in line with the present work. n ¼ 1:3  1015

V 1:93 F 0:555 T 7:45

ð14Þ

V 1:74 F 0:256 T 2:6

ð15Þ

ry ¼ 2:96  105

Conclusion Increasing cooling rate reduces SDAS and thereby improves the yield strength of a material. The work hardening coefficient is decreased with an increase in SDAS, so that ductility of the material is diminished. The casting variables like casting speed and cooling water flow rate improves the formability of the metal whereas melt temperature has a negative impact. Similarly, the strength of the metal is improved by both casting speed and water flow and is lowered by metal casting temperature. An attempt has been made to propose an empirical model for the prediction of work hardening coefficient and yield strength based on the three casting variables. It is envisaged that such a model could help the casting operators to fine tune the casting variables as per the requirement of product behavior (in line with casting technology permission values). The advantage of such a model is that the casting operator does not have to be knowledgeable on mathematical modeling. The control of dendritic as-cast microstructures by regulating operating parameters can be used as an alternative way to produce components with better mechanical properties.

1. Measurement technique, Light Metals, TMS, USA: 883–889. 2. H. Jacobi and K. Schwerdtfeger (1976) Dendritic Morphology of Steady state Unidirectionally Solidified Steel, Metall. Trans. A, Vol. 7A: 811–820. 3. K. P. Young and D.H. Kirkwood (1975) The Dendritic Arm Spacing Arms Spacing of Aluminum-Copper Alloys Solidified Under Steady-State Conditions, Metall. Trans. A, Vol. 6A: 197–205. 4. J. A. Spittle and D. M. Lloy (1979) Dendritic Arms Spacing in Hypoeutectic Pb-Sb Alloys Directionally Solidified Under Steady and Non-Steady Conditions, Proc. Of International Conference on Solidification and Casting, The Metals Soceity, London: 15–20. 5. M. O. Shabani and Ali Mazahery (2011) Microstructural prediction of cast A256 alloy of a function of cooling rat, Journal of Materials, TMS, USA:132–136. 6. D. G. Eskin, V. I. Savran, and L. Katgerman (2008) Effects of Melt Temperature and Casting Speed on the Structure and Defect Formation during Direct-Chill Casting of an Al-Cu Alloy”, Metallurgical And Materials Transactions A Volume 36A: 1965–1976. 7. J. Sengupta, D. Maijer, M. A. Wells, Cockroft and A. Laruch (2003) Mathematical Modelling Of Water Ejection and Water Incursion During The Start Up Phase Of The Dc Casting Process, Light Metals, TMS, USA: 841–847. 8. Wang Mengjun, Gan Chunlei, Liu Xinyu(2005) Deformation Behavior Of 6063 Aluminum Alloy During High-Speed Compression, Journal Of Wuhan University Of Technology, Vol. 20 No. 3: 40–43. 9. H. Cao And M. Wessén , “Effect of Microstructure on Mechanical Properties of As-Cast Mg-Al Alloys (2004) Metallurgical And Materials Transactions a Vol. 35 A, January: 309–319. 10. C. H. Caceres, C.J. Davidson, J. R. Griffiths,C. L. Newton (2002) Effects Of Solidification Rate And Ageing On The Microstructure And Mechanical Properties Of AZ91 Alloy” Materials Science and Engineering A325: 344–355. 11. Wislei R. OsoRio, Pedro R. Goulart, Givanildo A. Santos, Carlos Moura Neto, and Amauri Garcia (2006) Effect of Dendritic Arm Spacing on Mechanical Properties and Corrosion Resistance of Al 9 Wt Pct Si and Zn 27 Wt Pct Al Alloys, Metallurgical And Materials Transactions a Vol. 37A: 2525–2038. 12. Suyitno, D.G. Eskin, V.I. Savran, And L. Katgerman (2004) Effects of Alloy Composition and Casting Speed on Structure Formation and Hot Tearing during Direct-Chill Casting of Al-Cu Alloys”, Metallurgical And Materials Transactions A Vol. 35 A: 3551–3561. 13. I. Ebrahimzadeh, G. H. Akbari (2007) Comparing Of Mechanical Behavior And Microstructure Of Continuous Cast And Hot Worked Cuzn40al1 Alloy”, IJE Transactions B: Applications Vol. 20, No. 3, December: 249–256. 14. Pedro R. Goulart, Jos. E. Spinelli, Wislei R. Osrio, Amauri Garcia (2006) Mechanical Properties as a Function Of Microstructure And Solidification Thermal Variables Of Al–Si Castings”, Material Science and Engineering A421: 245–253. 15. C. Devdas, Ivo Musulin, Olaf Celliers(1992) Prediction Of Microsrtucture Of Dc Cast 6063 Billets and Its Effect On Extrusion Processes, International Conference on Extrusion Technology: 121–128.

1004 16. Doru M. Stefanescu, G. Upadhya, and D. Bandyopadhyay (1990) Heat Transfer-Solidification Kinetics Modelling of Solidification of Castings, Metallurgical Transactions A Volume 21: 997–1005. 17. M. J. Couper, B. Rinderer, M. Cooksey (2004) A Simplified Empirical Model of Precipitation Strengthening in 6000 Al-Mg-Si Extrusion Alloys, Materials Forum Volume 28: 159–164. 18. W. Johnson and P. B. Mellor (2003) Engineering Plasticity Van-Nostrand Reinhold, London. 19. Q. G. Warg, (2003) Microstructural effects on the tensile and fracture behavior of Aluminum casting alloys A356/357 inch, Metallurgical and Materials Transaction A, Vol-34A:2887–2889.

S. P. Mohapatra 20. M. O. Defeng, H. Zhengfei, L. Xiaoshan (2010) Effect of Microstructure on Strain Hardening Exponent Prediction of Cast Aluminum Alloy, Acta Metallurgica Sinica, VOl-46, No-2:184– 188. 21. G. E. Dieter (2001) Mechanical Metallurgy, McGraw-Hill Books, UK. 22. Q. G. Wang (2004) Plastic Deformation Behavior Of Aluminum Casting Alloys A356/357 Inch, Metallurgical and Materials Transaction A, Vol-35A: 2707–2718. 23. C. H. Caceres and J. R. Griffiths (1996) Damage by the Cracking of Silicon Particles in an 7Si-0.4Mg Casting Alloy, Acta Metallurgica, Vol. 44, No. 1: 25–33.

Stability of SiC and Al2O3 Reinforcement Particles in Thermomechanical Processed Direct Chill Cast Al MMnCs Abdallah Abu Amara, Guangyu Liu, Dmitry Eskin, and Brian McKay

Abstract

Introduction

In this study, aluminium alloys reinforced by alumina (Al2O3) and silicon carbide (SiC) were investigated. Lab-scale DC cast billets with a diameter of 80 mm were cast by ultrasound-assisted stir-casting technology. The billets were subsequently thermo-mechanically processed. The distribution of the particles in the matrix was analysed before and after thermomechanical processing. It was found that hot rolling improved the distribution of the reinforcement particles in the matrix. Reactions between the reinforcement particles and the matrix were also investigated, and their implications studied. It was found that alumina reinforcement reacted with magnesium (Mg) in the alloy to produce spinel and SiC reinforcement oxidised producing a layer of silicon oxide (SiO2) around the particles. The consequences of these reactions are discussed. Whilst theoretical calculations of free energies for different carbides suggested that transition metals in the alloys should substitute the silicon in SiC, no reactions were observed in the physical experimentation. This was attributed to the weight percentages of the transition metals in the alloy being too small. These results are discussed and analysed further in this study, with plans for future experimentation. Keywords










Al MMnC DC casting SiC particles Thermomechanical processing




Al2O3 particles

A. Abu Amara
G. Liu (&)
D. Eskin
B. McKay Brunel Centre for Advanced Solidification Technology, Brunel University London, Uxbridge, UB8 3PH, Middlesex, UK e-mail: [email protected] A. Abu Amara e-mail: [email protected]

The synthesis of lightweight materials with elevated mechanical properties has long been a challenge for engineers and material scientists. Materials with such characteristics are known to be highly desirable in transport applications, especially the aerospace and automotive industries. As legislative restrictions continue to grow around CO2 emissions from vehicles and aircraft, Original Equipment Manufacturers (OEMs) are seeking to adopt the use of new materials in their designs. This comes with the knowledge that a 10% reduction in the vehicle weight can improve the fuel economy by 6–8% [1]. Furthermore, the electrical revolution is pushing new specifications and requirements for Battery Electric Vehicle (BEV) technology, whilst the aerospace industry has already begun implementing designs for hydrogen fuel cell aircraft [2]. The introduction of new technologies in the form of electric vehicles and hydrogenpowered aircraft will undoubtedly cause the OEMs to continue searching for new material technologies. The addition of ceramic nano-particles as a reinforcement to aluminium alloys has proven to be a legitimate method for enhancing the material properties [3]. In recent years, there has been a growth in the interest of these materials due to their desirable mechanical properties. Many studies have shown that aluminium metal matrix composites (AlMMC) have superior mechanical properties when compared to their base alloys [3–9]. Nevertheless, there are still many challenges to overcome in the field of Metal Matrix Composites (MMCs). Reactions between the reinforcement and the matrix can occur, such as alumina reinforcement producing spinel phase (MgAl2O4) [10]. The growth of spinel causes the viscosity in the melt to increase which has negative implications on the castability of the material [10]. SiC reinforcement, on the other hand, has the potential to oxidise when in the melt and produces an SiO2 layer around the particles. Unlike spinel phase, the SiO2 has positive implications as this layer enhances the

© The Minerals, Metals & Materials Society 2023 S. Broek (ed.), Light Metals 2023, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-031-22532-1_134

1005

1006

wettability, allowing for better incorporation of particles [11]. Moreover, the addition of carbide reinforcement particles to the aluminium alloys containing transition metals can lead to chemical reactions which substitute the metal in the reinforcement with the transition metal. For example, an addition of SiC reinforcement to an alloy containing Zr can lead to a chemical reaction between the reinforcement and the Zr in the alloy, resulting in the formation of ZrC particles rather than SiC [12]. Transition metals in alloys have been known to make substitutions in ceramic particle additions used for grain refinement. An example of this is the Zr poisoning effect. This effect which is reported in titanium carbide (TiC) particles reduces the grain refinement of TiC considerably [13]. The mechanism of the Zr poisoning effect is not completely agreed upon in the literature; however, it is generally accepted that the poisoning occurs as a result of Zr interacting with the nucleating particles such as Al3Ti and TiC from the grain refiner [14]. The thermodynamic stability of the compounds can be used to determine whether or not a reaction will take place between the transition metals in the alloy and the carbide reinforcement. A carbide Ellingham diagram is presented in Fig. 1 which compares the Gibbs free energies of carbides at different temperatures [15]. It can be seen from the diagram that there are a number of transition metal carbides which exhibit lower free energies than silicon carbide, which is one

Fig. 1 Carbide Ellingham diagram

A. Abu Amara et al.

of the reinforcement types used in this study. This means that these carbides are likely to appear in the material instead of the SiC particles under suitable conditions. In this study, work is carried out to determine which reactions (if any) will occur between the matrix of an aluminium alloy and two types of reinforcement particles: Al2O3 and SiC. The base aluminium alloy and the composite materials are produced via DC casting then thermomechanically processed. Theoretical calculations of the free energies of the compounds are calculated to determine which carbides are likely to form in place of SiC. The implications of the reactions observed are discussed.

Methods Experimental Material Synthesis 6 kg of aluminium alloy was charged in an electric furnace and heated to 730 °C. At the same time, 150 g of Al-20 wt% SiC master-alloy powder was preheated to 200 °C in an electric oven. This amount corresponded to an addition level of 0.5 wt% SiC of the 6 kg base alloy. This master-alloy powder was produced by mechanical alloying and supplied by MBN Nanomaterialia (Italy). The master-alloy powder was introduced to the melt through mechanical stirring with a rotational speed of 400 RPM for a period of 10 min. Next, ultrasound was applied to the melt to further disperse the reinforcement particles. The same process was repeated for the base aluminium alloy, without any reinforcement addition. Also, it was repeated for the aluminium alloy with an addition of 150 g of Al-20 wt% Al2O3 master-alloy powder which corresponded to 0.5 wt% Al2O3. Once the ultrasound was applied to the melt for 10 min at a frequency of 17 kHz and a power of 3.5 kW, the melt was direct chill (DC) cast through an 80 mm diameter mould, at a pouring temperature of 700 °C. The casting speed was set to 230 mm/min. The 80 mm diameter billet can be seen in Fig. 2b. Aluminium alloys, which are widely used in the automotive industry [16], are typically DC cast into round billets for the purpose of heat treatment and subsequent mechanical processing [17]. Lab-scale DC casting was performed in this study to replicate the process used in industry. Thermo-Mechanical Processing The homogenisation process was carried out at temperatures between 500 and 550 °C for a period of 3–4 h. The purpose of this step is to dissolve the Mg2Si phase and to transform the b-AlFeSi (b-Al5FeSi) to a-AlFeSi (a-Al12(FeMn)3Si) [18, 19]. Sarafoglou et al. [18] observed the dissolution of the Mg2Si gradually over the course of 3 h at 540 °C in a 6xxx alloy.

Stability of SiC and Al2O3 Reinforcement Particles in Thermomechanical Processed Direct Chill Cast Al MMnCs

1007

Fig. 2 a Casting setup, b 80 mm diameter billet, c rectangular samples obtained from billet, and d hot-rolled samples

Rectangular samples were then obtained from the centres of the 80 mm diameter billets, for the purpose of hot rolling. These can be seen in Fig. 2c. The rectangular samples were rolled at a temperature of 350 °C, from a thickness of 13 mm down to a thickness of 2.5 mm (*80% reduction) as can be seen in Fig. 2d. This was achieved via eight passes through the rolling mill. A T6 heat treatment was then performed. For the solution treatment step, the samples were heated to temperatures between 500 and 25 °C and then quenched in water. The samples were then left to naturally age (at room temperature) for one day. Finally, the samples were aged at temperatures in the range of 150–160 °C for 15 h. The homogenisedrolled samples were solution treated and aged according to temperatures and times taken from Smithells Metals Reference Book [20].

Characterisation Samples were obtained from the as-cast and as-rolled homogenised castings. These samples were ground and polished in preparation for microstructural analysis. The microstructures of the castings were then studied via Optical Microscopy (OM). The use of Optical Microscopy allowed for the observation of nano-particle clusters in the matrix as well as secondary intermetallic phases. The measurement tool in the OM was also used for comparing the sizes of different clusters of reinforcement particles. A Zeiss Supra 35 field-emission scanning electron microscope (FESEM), equipped with energy-dispersive X-ray spectroscopy (EDS), was also used to analyse the size of particles and microstructure of the cast materials. EDS was used to obtain the compositions of the reinforcement particles. This was important for confirming that these were added reinforcement. Compositions of the intermetallic phase were also analysed using EDS. The interfaces between the reinforcement and the matrix were studied to determine if there had any chemical reactions between the particles and the matrix.

Gibbs Free Energy Calculation The Gibbs free energy was calculated using the following equation: DG ¼ DH  ðT  DSÞ

ð1Þ

where DG is the change in the Gibbs free energy (kJ * mol−1), DH is the change in the enthalpy (kJ * mol−1), DS is the change in entropy (kJ * K−1), and T is the temperature in Kelvin [21]. The change in enthalpy and entropy values was taken from the National Institute of Standards and Technology Database [22]. These were used to calculate the change in Gibbs free energy for SiC and the other carbides that were likely to form as a result of alloying elements in the alloy. These carbides are zirconium carbide (ZrC), titanium carbide (TiC), and chromium carbide (Cr23C6). An example calculation is shown below for the silicon carbide: DG ¼ 29:02  ð1000  0:06404Þ ¼ 35:02 kJ=mol In this example, the DH is equal to 29.02 (kJ/mol) and the DS is 0.06404 (kJ/mol * K) [22]. The temperature used in the calculation is 1000 K which is approximately equivalent to 730 °C. This is the temperature at which the particles were added and processed in the aluminium melt (using mechanical stirring and ultrasonication). The remaining free energy values are presented in Results.

Results Free Energies of Transition Metal Carbides As previously mentioned in Sect. 2.2, the free energies of formation were calculated for ZrC, TiC, and Cr23C6 as presented in Table 1. Table 1 shows that the three transition metals present in the alloy have lower free energies than SiC. This means that

1008 Table 1 Calculated theoretical free energies of formation

A. Abu Amara et al. Carbide

SiC

ZrC

TiC

Cr23C6

Free energy of formation (kJ/mol)

−35

−56

−46

−983

the reactions of all of the shown carbides are more spontaneous than the reaction of SiC. This suggests that, in theory, these carbides can form in place of the SiC particles.

Base Alloy’s and Composite Materials’ Microstructure As-Cast Microstructure The microstructure of the casting containing alumina reinforcement showed that the particles were spread across all of the matrix. In some areas, the particles can be seen clustering as in Fig. 3a, but generally the particles were not adhering to each other or agglomerating. Individual particles can be seen, as indicated, in Fig. 3b. The SiC castings however revealed large agglomerates greater than 100 lm in diameter as shown in Fig. 3c. It can be seen from Fig. 3d that the particles are not only clustering, but also coalescing, as if still bonded by the van der Waals forces of attraction. The SEM micrographs and their corresponding EDS spectra shown in Fig. 4 were used to confirm that these particles are alumina and SiC through the use of EDX point analysis. For the alumina particles, two main aluminium and oxygen peaks can be seen with smaller magnesium and silicon peaks also visible in the spectra shown in Fig. 4d. This confirms that the particles present are Al2O3 with potential reactions with magnesium to produce spinel phase. This will be discussed in further detail below.

Fig. 3 Optical micrographs of a macro-distribution of alumina particles, b micro-distribution of alumina particles, c macro-distribution of SiC particles and d micro-distribution of SiC particles (magnified from c)

For the SiC particles, two main silicon and carbon peaks can be seen with a smaller aluminium peak also visible in the spectra in Fig. 4e. This confirms that the particles are silicon carbide. The existence of the aluminium peak may be a result of some of the backscattered electrons penetrating below the particle into the Al matrix. There were no chemical reactions observed between the SiC reinforcement particles and the transition metals present in the alloy, even though the theoretical calculations suggested that some reactions should take place between the transition metals in the alloy and the reinforcement. Reasons for this will be outlined in the discussion section. The alumina particles in Fig. 4a can be seen ranging in sizes between 0.2 and 2 lm, whilst the majority of the SiC particles in Fig. 4c are smaller than 100 nm in diameter. It is difficult to tell the exact sizes of these particles due to the limitation of the machine; however, TEM will be used in future studies to further characterise the SiC particles.

Rolled and Heat-Treated Microstructure Figure 5a shows optical micrographs of the as-cast base alloy with 0.5 wt% alumina. It is clear that the particle dispersion in the rolled samples shown in Fig. 5b, c is better than that of the as-cast state (in Fig. 5a). It can be said that the rolling has caused the clustered alumina particles to disperse further in the matrix. A similar effect can be seen in Fig. 5d which represent the further dispersion of SiC particles due to rolling.

Stability of SiC and Al2O3 Reinforcement Particles in Thermomechanical Processed Direct Chill Cast Al MMnCs

1009

Fig. 4 a SEM micrograph of alumina clusters, b SEM micrograph of individual alumina particles, c SEM micrograph of a SiC agglomerate, d EDX spectra of an alumina particle, and e EDX spectra of a SiC particle

1010

A. Abu Amara et al.

Fig. 5 Optical micrographs of composite materials, a as-cast base alloy with 0.5 wt% alumina before rolling, b rolled base alloy with 0.5 wt% alumina, c magnified from b, e base alloy with 0.5 wt% SiC

Homogenously distributed, fine-sized dispersoid particles were observed in the heat-treated rolled samples. The particles can be seen in Fig. 6. Particles can be seen ranging in size from approximately 100 nm up to almost 1 lm. The dispersoids were observed in the base alloy casting and both of the composite material castings (0.5 wt% alumina and 0.5 wt% SiC). The EDX spectra shown in Fig. 6c was taken at one of the dispersoid particles shown in Fig. 6a. There is a main peak and further smaller peaks at Mn, Fe, and Si. This suggests that the dispersoid particles observed are a-Al15 (MnFe)3Si2. This is a common dispersoid phase in the wrought aluminium alloys [23]. Another common dispersoid phase in the wrought aluminium alloys is the Al3Zr; however, it is likely that these are the smaller particles observed in Fig. 6a, b [23]. Due to their relatively fine size, these nano-sized particles were difficult to characterise using the EDX on the SEM employed. It is planned to study these particles in further detail using TEM, in future work.

Discussion Particle Distribution The alumina reinforcement particles that showed a better distribution than the SiC particles can be attributed to the difference in size between the particles. The alumina particles exhibited diameters in the range of 0.2–1 lm. Even larger alumina particles were observed, in some cases reaching up to 2 lm in diameter. On the other hand, the SiC particles often had diameters of less than 100 nm. This means that these particles had larger van der Waals forces than the alumina particles, which made them more difficult to disperse in the matrix. It has been reported in previous studies that the stir-casting technique is not effective for the purpose of dispersing particles in the nano-meter range [24]. This is why the use of ultrasound is generally implemented

Stability of SiC and Al2O3 Reinforcement Particles in Thermomechanical Processed Direct Chill Cast Al MMnCs

1011

Fig. 6 SEM micrographs of rolled samples a base alloy dispersoids b composite material dispersoids and c spectra from dispersoid particle

in addition to mechanical stirring. However, the acoustic streaming and cavitation effect generally tend to occur directly below the ultrasonic sonotrode [7]. This means that in large volumes of melt, not all of the melt will benefit from the use of ultrasound. Choi et al. [25] discussed the use of ultrasonication for the purpose of distributing nano-particles on a large scale. In the study, it was suggested that in order for the ultrasonication to be effective in large volumes of melt, all of the melt should be passed through a chamber with an active ultrasonic sonotrode. This method guarantees that almost all of the melt will at some point travel below the sonotrode and experience acoustic streaming and cavitation. This method may be experimented with in the future. Previous studies have shown that alumina has poor wettability with aluminium [26] and this also contributed to the fact that there was no significant increase in mechanical properties. One of the key strengthening mechanisms mentioned is the load-bearing effect and this requires good adhesion between the particle and the matrix. Therefore, this strengthening mechanism has a very small effect when using alumina particles as a reinforcement due to the poor wettability. This means that using different types of particles as a reinforcement can result in greater increase in strength due to better wettability. An example of this is silicon carbide. Bao et al. [27] showed that the wetting angle of an Al–SiC system is approximately 79°, whilst the wetting angle of an Al–Al2O3 system is approximately 97°. However, the

relatively small size of the silicon carbide particles used in this study caused them to agglomerate excessively which is likely to have had an impact on the properties in the as-cast state. The improved distribution of particles in the matrix due to rolling is reported in the literature in several studies [28, 29]. Song et al. [28] found that 85% rolling reduction at 300 °C increased the distribution homogeneity of AlN reinforcement particles in the matrix. This in turn contributed to an increase in the strength and ductility of the material. However, Bisht et al. [30] reported observing cracks in the rolled materials due to agglomerated reinforcement particles. The material was rolled at 200 °C with 60% reduction. It is likely that the lower temperature contributed towards the cracking of the material, hence higher temperatures should be used in the cases of rolling composite materials with agglomerated particles.

Chemical Reactions Between Reinforcement and Matrix The EDX spectra in Fig. 4d shows a small oxygen peak, which is likely due to the oxidation of the SiC particles during the synthesis process. The oxidation of SiC is used as a technique to treat SiC particles so that the processability of aluminium metal matrix composites containing SiC reinforcement can be improved [11]. The generation of a

1012

continuous SiO2 layer allows for the incorporation of the particles in the molten aluminium [11]. This is because the degrading reaction of Al4C3 formation is replaced by alternative interfacial reaction which can enhance the wettability [11]. The EDX spectra in Fig. 4e shows the smaller Mg and Si peaks in addition to the main Al and O peaks, suggesting the appearance of spinel phase. The appearance of spinel has been reported when alumina reinforcement is used. For example, Forn et al. [31] reported observing a layer 1.5 lm in diameter with an irregular shape around the reinforcement particles. XRD was used to confirm the presence of MgAl2O4. The author pointed out that the appearance of this phase is detrimental to the mechanical properties as the decohesion between the matrix and the reinforcement phase, according to the fractography study, might be attributed to the spinel creation in the interface [31]. It was suggested that the cohesion between the reinforcement and the matrix can be improved through a T6 heat treatment. This was one of the reasons why a T6 heat treatment was carried out in this study. No chemical reactions were observed between the SiC reinforcement and the transition metals present in the alloy. This can be attributed to two main contributing factors. Firstly, the holding time of the material in the furnace may have been insufficient for the reactions to take place. It was mentioned previously that as the reinforcement was added to the melt, the melt was stirred continuously for 10 min. The melt was then left to sit for approximately 5 min before the ultrasonic treatment was applied for 10 min. This means that since the incorporation of the reinforcement, the total time in which the melt and the reinforcement was in the furnace was approximately 25 min. Considering that the reinforcement was added in the form of a master powder, some time would have been needed for the aluminium in the master powder to melt, releasing the SiC particles. So, it is likely that 25 min was not enough time for the reactions to take place. Secondly, the compositions of transition metals in the alloy may have been too low for any reaction to take place. None of the transition metals present in the alloy had any wt % greater than 0.15%. In the case of Zr poisoning, the effect had been observed in 7xxx series alloys [32]. 7xxx series alloys typically contain more Zr than the aluminium alloy used in this study as will be discussed later. Furthermore, Bunn et al. [33] reported observing the Zr poisoning effect in an alloy containing 5 at% Zr. This suggests that the 0.15 wt % Zr in the alloy was also insufficient for any reaction to take place with the reinforcement. The presence of transition metals such as Zr, V, and Cr in the wrought aluminium alloys is important for the purpose of solid-state reactions with the soluble elements during the homogenization process [23]. Typically, these reactions

A. Abu Amara et al.

produce the nano-sized hard particles known as dispersoids [23]. The thermally introduced stable dispersoids (during homogenisation) are an effective way to retard the dislocation motion and inhibit recrystallisation [34]. Zr can be found in greater weight percentages in 7xxx alloys compared to other wrought alloys. Yet, the poisoning effect has been observed with 7xxx series alloys containing as little as 0.10% Zr [32]. Some other wrought alloys contain up to 0.2 wt% Zr, meaning that this certainly has the potential for producing an effect similar to Zr poisoning. These substitution chemical reactions in the melt will be exothermic reactions due to negative free energies. This will cause the cooling rate to be slower in certain localised areas or pockets in the alloy. These areas will have coarser grain sizes and this risks the introduction of weak spots in the material, hence reducing mechanical properties. Further experimentation will be carried out to determine whether the above-mentioned substitution reactions will occur with higher wt% of transition metals and/or longer processing times. This phenomenon is not mentioned extensively in the literature due to the limited number of studies on Al MMC in the wrought alloys (7xxx, in particular). These two series are the most popular in the automotive and aerospace fields. Hence, if Al MMC are to be used in these industries, it will be imperative to understand which reactions will occur when carbide reinforcements are incorporated in the mentioned alloys and how these will ultimately affect the final properties.

Conclusions The present work studied the effect of addition of silicon carbide and aluminium oxide reinforcement particles to an aluminium alloy. The base aluminium alloy and composite materials were DC cast, homogenised, hot rolled, and heat treated. Characterisation was subsequently carried out. The study investigated the distribution of the particles in the matrix before and after hot rolling, as well as chemical reactions between the reinforcement and the alloy. The following key points can be concluded from the study: • Whilst the mechanical stirring and ultrasonication did not fully disperse the particles in the matrix, the Al2O3 particles were more dispersed than the SiC. This is most likely as a result of the difference in size between the two types of reinforcement (the SiC being smaller). The rolling contributed to a more homogenous distribution of the particles in the matrix for both types of reinforcement. • SiC reinforcement was found to produce a desirable oxidation reaction, in which an SiO2 layer is formed around the particles. This will increase the wettability of the particles with the matrix. Al2O3 particles produced a

Stability of SiC and Al2O3 Reinforcement Particles in Thermomechanical Processed Direct Chill Cast Al MMnCs

non-desirable spinel phase which can reduce the mechanical properties; however, according to existing literature, the negative effect of spinel can be reduced through a T6 heat treatment [31]. • Theoretical calculations of the free energies of formations of different carbides showed that some transition metals in the alloy should substitute the silicon in the SiC reinforcement. However, no such reactions or substitutions were observed in the physical experiments. This is likely as a result of the compositions of the transition metals in the alloy being relatively low, and the holding times of the melt being too short.

Acknowledgements The authors gratefully acknowledge the financial support of the European Union (FLAMINGo Grant Agreement No. 101007011) on this project. Constellium and MBN Nanomaterialia are also gratefully acknowledged for providing the aluminium alloys and the reinforcement nano-powder, respectively. Finally, the Experimental Techniques Centre at Brunel University is thanked for providing access to the equipment to conduct the characterisation work.

References 1. Energy.gov. 2020. Lightweight Materials for Cars and Trucks. [online] Available at: [Accessed 5 January 2020]. 2. Professional Engineering, 2022. Three final FlyZero concepts show vision of the future for aerospace. [online] Available at:

[Accessed 2 February 2022]. 3. Lazarova, R., Bojanova, N., Dimitrova, R. et al. Influence of Nanoparticles Introducing in the Melt of Aluminum Alloys on Castings Microstructure and Properties. Inter Metalcast 10, 466– 476 (2016). 4. Ramanathan, A., Krishnan, P.K., Muraliraja, R., A review on the production of metal matrix composites through stir casting – Furnace design, properties, challenges, and research opportunities, Journal of Manufacturing Processes, 2019, ISSN 1526-6125. 5. Ajith, C., M et al matrix composites: History, status, factors and future. Dissertation (2011). 6. Malaki, M., Xu, W., Kasar, A.K., Menezes, P.L., Dieringa, H., Varma, R.S., Gupta, M. Advanced Metal Matrix Nanocomposites. Metals. 2019. 7. Shinde, D.M., Sahoo, P. Fabrication of Aluminium Metal Matrix Nanocomposites: An Overview. In: Sahoo, S. (eds) Recent Advances in Layered Materials and Structures. Materials Horizons: From Nature to Nanomaterials. Springer, Singapore (2021). 8. Rasidhar, Lanka & Krishna, A. & Rao, Ch.Srinivasa. Fabrication and investigation on properties of Ilmenite (FeTiO3) based al-nanocomposite by stir casting process. International Journal of Bio-Science and Bio-Technology. 5. 193–199 (2013). 9. Ma Z.Y., Li Y.L., Liang, Y., Zheng, F., Bi, J., Tjong, S.C., Nanometric Si3N4 particulate-reinforced aluminum composite, Materials Science and Engineering: A, Volume 219, Issues 1–2, 1996, Pages 229–231, ISSN 0921-5093.

1013

10. McLeod, A.D., Gabryel, C.M. Kinetics of the growth of spinel, MgAl2O4, on alumina particulate in aluminum alloys containing magnesium. Metall Mater Trans A 23, 1279–1283 (1992). https:// doi.org/10.1007/BF02665059. 11. Ureña, A., Martınez, E.E., Rodrigo, P., Gil, L., Oxidation treatments for SiC particles used as reinforcement in aluminium matrix composites, Composites Science and Technology, Volume 64, Issue 12, 2004, Pages 1843–1854, ISSN 0266-3538. 12. Bhanumurthy, K., Schmid-Fetzer, R., Interface reactions between SiC/Zr and development of zirconium base composites by in-situ solid state reactions, Scripta Materialia, Volume 45, Issue 5, 2001, Pages 547–553, ISSN 1359-462. 13. Yang, H., Qian, Z., Zhang, G., Nie, J., Liu, X., The grain refinement performance of B-doped TiC on Zr-containing Al alloys, Journal of Alloys and Compounds, Volume 731, 2018, Pages 774–783, ISSN 0925-8388. 14. Wang, Y., Fang, C.M., Zhou, L., Hashimoto, T., Zhou, X., Ramasse, Q.M., Fan, Z., Mechanism for Zr poisoning of Al-Ti-B based grain refiners, Acta Materialia, Volume 164, 2019, Pages 428–439, ISSN 1359-6454. 15. Park, J.H., Seo, H.S., Kim, K.Y., Alloy Design to Prevent Intergranular Corrosion of Low-Cr Ferritic Stainless Steel with Weak Carbide Formers. Journal of The Electrochemical Society, 162 (8) C412–C418 (2015). 16. Djukanovic, G., Aluminium Alloys in the Automotive Industry: A Handy Guide. In: Aluminium Insider (2019). 17. Tempelman, E., Shercliff, H., Ninaber van Eyben, B., Chapter 5 Extrusion of Metals, Manufacturing and Design, ButterworthHeinemann, 2014, Pages 69–83, ISBN 9780080999227. 18. Sarafoglou, PI., Serafeim, A., Fanikos, IA., Aristeidakis, J.S., Haidemenopoulos, G.N., Modeling of Microsegregation and Homogenization of 6xxx Al-Alloys Including Precipitation and Strengthening During Homogenization Cooling. Materials (Basel). 2019 May 1;12(9):1421. https://doi.org/10.3390/ma12091421. 19. Rinderer, B., The Metallurgy of Homogenisation. Materials Science Forum (Volume 693). https://doi.org/10.4028/www. scientific.net/MSF.693.264. 20. Smithells, C., Brandes, E., Brook, G. and Smithells, C., 1992. [Metals reference book]. Oxford: Butterworth-Heinemann, pp. 29–19. 21. Shatynski, S.R. The thermochemistry of transition metal carbides. Oxid Met 13, 105–118 (1979). https://doi.org/10.1007/ BF00611975. 22. Webbook.nist.gov. 2022. NIST Chemistry WebBook. [online] Available at: [Accessed 20 September 2022]. 23. Lumley, R., 2010. Fundamentals of aluminium metallurgy. p. 370–374. 24. Mohanty, P., Mahapatra, R., Padhi, P., Ramana, CH.V.V., Mishra, D.K., Ultrasonic cavitation: An approach to synthesize uniformly dispersed metal matrix nanocomposites—A review, Nano-Structures & Nano-Objects, Volume 23, 2020, 100475, ISSN 2352-507X. 25. Choi, Hongseok & Cho, Woo-hyun & Hoefert, Daniel & Weiss, David & Li, Xiaochun. (2013). Scale-up Ultrasonic Processing System for Batch Production of Metallic Nanocomposites, American Foundry Society. 26. Bao, S., Tang, K., Kvithyld, A. et al. Wettability of Aluminum on Alumina. Metall Mater Trans B 42, 1358–1366 (2011). https://doi. org/10.1007/s11663-011-9544-z. 27. Bao, S., Tang, K., Kvithyld, A., ENGH, T., Tangstad, M., et al. Wetting of pure aluminium on graphite, SiC and Al2O3 in aluminium filtration, Transactions of Nonferrous Metals Society of China, Volume 22, Issue 8, 2012, Pages 1930–1938, ISSN 1003-6326.

1014 28. Song, L.; Lu, F.; Jin, F.; Nie, J.; Liu, G.; Zhao, Y. Effect of Cold and Warm Rolling on the Particle Distribution and Tensile Properties of Heterogeneous Structured AlN/Al Nanocomposites. Materials 2020, 13, 4001. https://doi.org/10.3390/ma13184001. 29. Nie J.F., Wang F., Li Y.S., Liu Y.F., Liu X.F., Zhao Y.H. Microstructure and mechanical properties of Al-TiB2/TiC in situ composites improved via hot rolling. Trans. Nonferr. Met. Soc. 2017;27:2548–2554. https://doi.org/10.1016/S1003-6326(17) 60283-8. 30. Bisht, A., Kumar, V., Li, L.H., Chen, Y., Agarwal, A., Lahiri, D., Effect of warm rolling and annealing on the mechanical properties of aluminum composite reinforced with boron nitride nanotubes, Materials Science and Engineering: A, 2018, Pages 366–373, ISSN 0921-5093. 31. Antonio Forn, M. Teresa Baile, Elisa Rupérez, Spinel effect on the mechanical properties of metal matrix composite AA6061/(Al2O3)

A. Abu Amara et al. p, Journal of Materials Processing Technology, Volumes 143–144, 2003, Pages 58–61, ISSN 0924-0136. 32. Kearns, M.A., Cooper, P., Effects of solutes on grain refinement of selected wrought aluminium alloys Mater. Sci. Technol., 13 (1997), pp. 650–654. 33. Bunn, A.M. & Schumacher, P. & Kearns, Martin & Boothroyd, C. B. & Greer, A. (1999). Grain refinement by Al-Ti-B alloys in aluminium melts: A study of the mechanisms of poisoning by zirconium. Materials Science and Technology. 15. 1115–1123. https://doi.org/10.1179/026708399101505158. 34. Elasheri, A., Elgallad, E.M., Parson, N., & X.- G. Chen (2022) Improving the dispersoid distribution and recrystallization resistance of a Zr-containing 6xxx alloy using two-step homogenization, Philosophical Magazine, https://doi.org/10.1080/14786435. 2022.2103597.

TRC Combi Box: A Compact Inline Melt Treatment Unit for Continuous Casting M. Gorsunova-Balkenhol, M. Badowski, M. Betzing, J. Stotz, and Ø. Pedersen

Abstract

Background and Scope

An efficient inline melt treatment process is a key prerequisite for aluminum producers to ensure compliance with today’s product requirements. Modern DC casting lines use an equipment chain including Argon degassing, grain refiner addition, and mechanical filtration to reduce hydrogen and non-metallic inclusions to low levels. Individual units of this process chain have been adapted to twin roll casting (TRC) conditions with low melt flow rates, long casting campaigns, and high temperature sensitivity, while a holistic concept is pending. Speira and Drache initiated a project to develop a TRC degassing and filtration unit based on joint experience combining Argon degassing, CFF filtration and immersion heater temperature management into a user-friendly single box concept with small footprint and high metallurgical performance. This paper describes the technological approach for degassing, filtration and electrical heating, the integration into a single-unit concept and initial results of the pilot installation at Speira Karmøy. Keywords

Combi box




Melt treatment




Continuous casting

M. Gorsunova-Balkenhol (&)
M. Badowski Speira GmbH, Research & Development, Georg-Von-Boeselager Str. 21, 53117 Bonn, Germany e-mail: [email protected] M. Betzing
J. Stotz Drache Umwelttechnik GmbH, Werner-Von-Siemens-Straße 24-26, 65582 Diez, Germany Ø. Pedersen Speira Karmoy, Hydrovegen 160, 4265 Håvik, Norway

Production of aluminum rolled products is a multistage process, where the quality of a final product is strongly affected by the cleanliness of the cast aluminum melt. Even minor impurities in the melt can cause formation of harmful defects in the solid product, which lead to loss of material in scrap. Expenses of such a failure may rise significantly, depending on which production stage it was detected. Today’s quality requirements for aluminum products are very high, therefore an efficient melt treatment process has become a vital production step, as well as an important competing factor for aluminum manufactures. The main goal of the melt treatment is the elimination of harmful impurities, such as hydrogen, alkali metals, and non-metallic impurities. Whereas treatment of alkali metals is conducted in crucibles and furnaces, hydrogen and non-metallic inclusions can be most effectively removed during the inline treatment process chain. Negative impact of hydrogen is well known in the aluminum industry. Excess of hydrogen in liquid aluminum leads to porosity in solidified ingots and formation of defects in plates or strips such as blister [1]. Various degassing aggregates for the removal of hydrogen have been developed in recent decades. Today the most common practice in Direct Chill (DC) casting is to remove hydrogen by injecting inert gas (mainly Argon) with a spinning rotor. In this process, dissolved monoatomic hydrogen from the melt, driven through the difference in partial pressure in the melt and in the gas bubbles, diffuses into the rising Argon gas bubbles. In the bubbles H2 molecules are formed, which are then removed together with the inert gas bubbles on the melt surface. One of the most widely used apparatus of this kind is a box-degasser, such as Spinning Nozzle Inert Flotation (SNIF) or Aluminium Purification (Alpur) [2, 3]. It consists of a deep reactor, which is divided into several chambers—1 to 4 depending on the required capacity—and a lid, where rotors for each chamber are installed. The box-degasser

© The Minerals, Metals & Materials Society 2023 S. Broek (ed.), Light Metals 2023, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-031-22532-1_135

1015

1016

provides a high hydrogen reduction efficiency and is quite simple in handling and maintenance. The main disadvantages of this reactor type are a large footprint and a relatively high melt volume remaining in the box between the drops, which does not allow fast alloy changes and requires additional heating. Another concept developed to compensate disadvantages of the box-degasser is a launder degasser, such as Alcan Compact Degasser (ACD) [4]. It does not require heating and demands less space. A launder degasser consists of a wide launder segment and a lid, where rotors are installed. Due to a low reactive volume and smaller rotors, a higher number of rotors, typically 6–12, and higher inert gas flow rates are applied. Initially, the ACD is operated with addition of chlorine. When chlorine is eliminated a higher dross generation is reported. Hydrogen reduction rates are reported to be comparable with the box-degasser technology. The latest models of ACD are supplied with sealed lids, which is supposed to reduce dross generation and allow chlorine-free degassing. The siphon inert reactor (SIR) represents another drainfree type of degasser. It was developed to combine the large reaction volume of the box-degasser with the flexibility of the launder degasser. Unique features of the SIR unit are the application of underpressure to rise the metal level in the box and a use of bottom-installed rotors for gas purging, which makes maintenance and handling more demanding. The hydrogen reduction efficiency of the SIR technology is quite good and is comparable with one of the convenient box-degassers [5]. Depending on the boundary conditions all mentioned equipment can reach in average a 50–70% reduction rate of hydrogen. Non-metallic inclusions are another important aspect determining ingot quality. They vary strongly in form and origin, but all cause unwanted defects during processing of the cast ingot, such as tears and pin holes [1]. Although degassing is able to reduce the number of particles, mechanical filtration is more efficient and is essential for the assurance of the melt cleanliness. In DC casting the most commonly used technologies are ceramic foam and deep bed filtration. The principle of both systems is similar. They consist of a box divided into two chambers. In a bigger chamber a filter medium is positioned. Metal enters the box from the top and passes through the filter medium. Through the underflow in the lower part of the chamber, metal passes to the rising chamber and exits the box. In ceramic foam filtration (CFF) technology, the filtration medium consists of a thin ceramic foam plate made of alumina and some additives. The filtration efficiency of CFF depends on the porosity grade of the filter, a higher grade number corresponds to a finer pore size. The CFF technology is the most widely used due to a number of benefits such as easy maintenance and handling of the filter plates and low cost.

M. Gorsunova-Balkenhol et al.

CFF is also very flexible regarding the change of an alloy because it is drained after each cast. Deep bed filtration (DBF) technology is known for higher filtration efficiency as CFF especially for the finer inclusion range. Filtration medium fills the whole chamber and consists of many layers of alumina balls and grits of different grain size, which can be stacked in different order. This construction provides a better efficiency but is also more demanding in maintenance. Due to a large metal volume remaining in the box between the drops, it also requires additional heating and is not flexible. Another filtration technology mostly used for high purity products is bonded particle filtration (BPF). BPF is also produced in the shape of a plate-like CFF but is made of silicon carbide grits. It is much denser and heavier than CFF, which makes handling more difficult. However, the filtration efficiency is very high. The cost of BPF is also much higher than CFF. All the above-mentioned aggregates and technologies were mainly designed for the DC casting lines, where high flow rates and a batchwise casting mode are the main boundary conditions determining the process. In continuous casting, such as twin roll casting (TRC), opposite conditions with much lower flow rates and long casting campaigns are present. Besides, a much precise temperature management due to high sensitivity of the casting process has to be considered. So far, the main approach was to adapt existing DC casting equipment to continuous casting requirements, by reduction of the reaction volume. Several solutions are available on the market: • Smaller SIR unit consists of one chamber with one rotor head, instead of double version of a standard SIR unit [6]. • There is also a version of ACD for the flow rates of 1,7 t/h available [4]. • Novelis PAE distributes smaller versions of the Alpur degasser and the Pechiney Deep Bed Filtration (PDBF) unit specially for continuous casting lines. Alpur model G3 uses a single or double rotor concept and a sealed lid [7]. The smallest version of PDBF is designed for flowrates of 5 t/h by reduction of bed dimensions [8]. The main disadvantages of all existing concepts are still a quite oversized capacity and throughput, together with a large footprint and high operational and capital costs. The Pyrotek Multicast filtration system represents an integrated solution, where mechanical filtration by means of a bonded particle filter and degassing with porous plugs are combined [9]. It was initially used in a continuous casting line in 80 s; however, for today’s product requirements further advanced solutions are required. The main disadvantages identified by Speira Karmøy, where several units are installed, are the large footprint, inefficient temperature

TRC Combi Box: A Compact Inline Melt Treatment Unit for Continuous Casting

control, low hydrogen reduction efficiency of porous plugs, and high consumable cost for bonded particle filters. Moreover, a spare parts supply is very limited. The Speira production site at Karmøy in Norway operates several TRC units which require a better technical solution to keep up with the increasing product demands and provide better operating conditions in compliance with modern safety requirements. In absence of a fitted solution, a project in cooperation with the company Drache Umwelttechnik GmbH has been initiated to develop a filtration unit specially tailored for the TRC line conditions and requirements. High particle and hydrogen removal efficiency as well as low temperature loss are the most critical requirements to be fulfilled under the following boundary conditions: • Low metal flow rate of 1.5–2.5 t/h. • Long casting campaign of 6–12 weeks and >2500 mt. • Footprint of less than 3 m2. This paper summarizes basic considerations about the equipment concept and presents initial results of a pilot installation at the Speira Karmøy plant.

Concept Development A concept of combining several melt treatment steps in one unit was inspired by previous work done at R&D Department of Speira (earlier Hydro Aluminium Rolled Products). In early 2000s, a longterm study on the filtration efficiency was conducted at Speira’s R&D Department in Bonn. The results lead to a development of the XC-Filter concept, where ceramic foam filtration, deep bed filtration, grain refinement and heating elements were combined in one unit [10, 11]. The performance of the XC-Filter was investigated under different casting conditions on the dedicated production-scale casting center at Speira Rheinwerk plant. The key learnings from this development were that a ceramic foam filter can provide very high filtration efficiency and its capacity can substantially exceed the manufacturer specification. Furthermore, it was demonstrated that a combined filter box reduces the footprint of inline treatment, which also minimizes the heat loss. Additionally, it was learned that the application of immersion heaters can provide efficient temperature control for small metal volumes and throughputs. Based on this experience, an integrated design concept combining degassing, mechanical filtration and temperature control was considered for the TRC lines in Karmøy. The design concept and functionality of each step are discussed in the following chapters.

1017

Degassing As mentioned before, in DC casting argon injection by several spinning rotors and a large reactor volume are used to ensure an efficient removal of Hydrogen for high flow rates of 20–50 t/h. The melt throughput in TRC is significantly lower, but an efficient reduction of hydrogen up to 75% is still required. Besides hydrogen reduction efficiency, low gas consumption, low surface turbulence, and impact of argon bubbles transport on heating and mechanical filtration had to be pursued. A water model study was performed at R&D Facility in Bonn to understand the degassing step for TRC and to determine the optimal process parameters. A cross section of 600  500 mm was selected as degasser chamber based on the footprint requirements and single-chamber hydrogen reduction rates at Speira sites. It was assumed that this cross section is still close to a square shape and supports a good argon bubble distribution. These dimensions were transferred into the model in 1:1 ratio. Different rotor head types, rotor speed, position of rotor head, and argon flow rates were tested in the model. The rotor head model Carré 140 from Gerken SA has performed the best regarding the bubble distribution and surface turbulence and was chosen for the pilot unit. Based on the reactor volume and the highest flow rate, the minimum residence time of the melt in the degassing chamber was calculated to be about 8 min, from which a half-time period of approx. 4 min required to reduce the hydrogen concertation to 50% could be derived. A diagram on Fig. 1 demonstrates the monitored half-time period for oxygen reduction in the water model, depending on the rotor speed for different argon flow rates. To reach a desired hydrogen reduction of 75%, a double half-time period is required, which could be achieved by most of the parameter settings. Based on these results and the aim to minimize the energy and argon consumption, the lowest rotation speed of 200–250 rpm and a 10 l/min argon flow rate were set as a starting point for the pilot unit. Finally, even lower parameters were then defined in the commissioning phase. The position of the rotor head was chosen to be 100 mm above the bottom with 75 mm underflow to ensure very low transfer of the gas bubbles into the heater chamber (which will be discussed in more detail below).

Filtration Ceramic foam filters offer a significant advantage regarding cost and handling in comparison with other technologies such as bonded particle filters or bed filtration. As learned from XC-filter development, a high filtration performance

1018

M. Gorsunova-Balkenhol et al.

Temperature Management

Fig. 1 Half-time period for oxygen reduction in the water model in dependence of rotor speed and argon flow rate for rotor head model Carré 140

can be achieved by CFF, and therefore it was decided to use this technology for the mechanical filtration step. The main parameters defining the filter layout are the flow rate and the filter capacity. A filter porosity of 30–40 ppi was chosen based on the quality of the melt and on the quality requirements for the final product. For the required filter porosity and TRC throughput of 1.5–2.5 t/h, a 7–9 inch CFF plate would be required. Based on the nominal filtration capacity more than forty 26-inch filter plates would be required for a standard casting campaign of 2500 mt. Even though XC-filter results showed that nominal capacity was significantly exceeded, the use of one filter plate for the whole casting campaign was not feasible. A vertical gate CFF was chosen for the pilot unit. This solution provides more flexibility, since it can be exchanged during the cast if needed. The highest possible cross section of the plate close to 26 inch was chosen, which contributes to the extension of the filter plate life. Moreover, the vertical installation ensures a suitable priming height without increased launder depth by the height of the filter plate. Priming will likely start in the lower area of the filter plate and will move up during the use of the filter when particle deposition increases the local flow resistance. The sequential use of the filter increases the local melt velocity significantly above the average throughput and therefore closer to the recommended flow rates ensuring improved filtration performance. Nevertheless, the approach to exchange a CFF plate during running production is a new process and bears several risks, which need to be carefully assessed. A gripping tool was designed to ensure safe and ergonomic replacement.

The TRC process demands a very precise temperature control, and therefore the temperature of the melt coming from the filter box has to be stable and easily adjustable. For this purpose, an immersion heating technology offers the best solution, which was already proven by the application in the XC-filter. An immersion heater consists of two main components. In the center, an electrical resistance heater embedded in a high-performance insulating material, which ensures electrical insulation and high heat transfer. This heating element is protected with a silicon nitride sheath. This material is insensitive to thermal shock, poses high mechanical resistance, excellent thermal conductivity and optimum resistance to corrosion. Immersion heaters offer a whole range of benefits. The heating efficiency of immersion heaters is above 99%, which provides significant reduction of energy consumption in comparison with conventional heating systems. Temperature control accuracy lies in a range of ±2 °C. Heating up is fast and smooth. Immersion heaters are also easy to install and maintain. The main condition to assure all the benefits of the immersion heater technology is an unhindered heat transfer from the heater to the melt. Therefore, a perfect contact of the melt with the surface of the heater is essential for proper functionality and long life of the heater. In case gas bubbles or dross stick to the surface of the heater a local overheating occurs, which can cause damage of the immersion heater. To avoid the bubble transfer from the degasser chamber, the immersion heaters are positioned in a separate chamber. Both chambers are connected by an underflow the height of which was determined in the water model study for the degasser. Considering the requirements of the TRC process, the heating system is supposed to provide a heating rate of 50 K/h in idle mode and an online temperature adjustment of 15 K at highest flow rate. A heat loss model was calculated to determine the amount of heat needed to maintain the required melt temperature. According to the model thermal inertia, the degassing process, radiation, and convection contribute to the total heat loss of approx. 15 kW (Fig. 2). An additional energy input of 12 kW would be required to fulfill the requirements of the plant. So, in total 27 kW should be generated by the heaters to ensure the proper heating conditions. The heating power of the immersion heater can be calculated based on the heating zone length for a given diameter of the heater. Based on the lowest metal level in the box and required bottom clearance height a heating length of 310 mm was derived. Considering a specific heat flux of 20 W/cm2 an

TRC Combi Box: A Compact Inline Melt Treatment Unit for Continuous Casting

1019

Fig. 3 A side view of the filter box showing the sequence of the chambers along the metal flow

Fig. 2 Heat loss distribution in combi box calculated in a heat loss model

immersion heater of 55 mm diameter can provide approx. 11 kW. Based on this evaluation two immersion heaters with the nominal capacity of 11 kW would be sufficient to compensate the heat loss of the box and a third heater would be necessary for heating up. Considering theoretical calculations three vertical immersion heaters with 12 kW heating power were chosen for the pilot unit. The heaters are arranged in a triangle to fit to the set dimensions and provide homogeneous heat distribution in the heating chamber. Two thermocouples are used for controlling of the heaters. One is for temperature control and another is a safety thermocouple for the melt level control, which prevents the heaters from overheating when the melt level is too low. Additionally, a thermocouple is positioned behind the CFF plate to track the temperature development in this area due to unknown heat conductivity of the CFF plate.

System Integration After definition of a basic design of each process step, the chambers were set in the following sequence into a single unit:

chamber are not physically separated. Afterwards the grain refiner is added in the trough. This layout was chosen based on the following considerations: • The addition of grain refiner after the box allows high filtration performance of the CFF independent of the position within the sequence. • The CFF lifetime will be extended as the incoming inclusion load is already reduced by the degassing step. • The immersion heaters are centrally collated with good heat transfer to degasser and CFF area. Lowest temperature expected downstream CFF. • Low underflow between degasser and heater chamber inhibits a bubble transfer to the heater surface. • The immersion heaters are located at the highest metal depth, which increases the heating length. The final dimensions of the box are as follows 2000  1000  900 mm which fits to the space requirements on the casting line and provides enough space for operators to work. The metal volume in the box during the cast is approx. 1000 kg, and 700 kg has to be drained when the casting campaign is done.

Operational Concept • a first chamber for degassing, • a second chamber for heating of the melt • a third chamber for mechanical filtration A cross section of a filter box along the metal flow is shown in Fig. 3. First metal enters the degassing chamber, through the underflow it comes to the next chamber, where it is heated up. After that metal passes through the vertical CFF plate and exits the box. The heater chamber and the CFF

Besides the metallurgical performance, an improvement of working conditions and safety aspects was a very important aim for the new box. During the design phase various working situations were systematically analyzed to determine required conditions. The final design is presented on Fig. 4. The layout is organized in a way that all operations can be done from one side of the box (“operator side”). On the other side only a drainage bin is positioned.

1020

Fig. 4 An overview of the box design with the main components necessary for operation

The lid of the box is divided into three parts, one for each chamber. The lids were designed individually, and each lid can be lifted by crane. Dedrossing is one of the most frequent operations to be done on the box during the cast. In order to minimize dross formation, removable dams can be installed in the inlet and the outlet of the box. The dams are cut at the bottom which allows the melt flow in and out of the box but prevents entrance of air. An easy access for skimming is provided through the opening on the side of the degasser lid on the “operator side”. A bottom part of the opening is lined with an inclined fused silica inlay which allows comfortable skimming and prevents sticking of the dross to the steel surface. Immersion heaters are mounted in the lid of the heater chamber and can be lifted together with the lid. Skimming of the heater chamber is supposed to be done through the opening of the filter chamber. The filter chamber has to be opened for skimming of the filter and heater chambers as well as for eventual changing of the filter plate. The lid of the filter chamber can be moved manually, for safe closing a counterweight is built in. Preheating of the box is done with hot air blowers from two positions. One is for the degasser chamber and another one is for the heater and filter chamber. The filter plate has to be pre-heated separately.

M. Gorsunova-Balkenhol et al.

functioning as expected. In April 2022, a first measurement campaign was conducted to evaluate a performance of the new filter concept. The results of the evaluation are presented below. The hydrogen content in the melt before and after the combi box was measured by means of AlSCAN, with different measurement probes from Bomem and Accurity. Figure 5 demonstrates average values of measured hydrogen concentration and reduction efficiency in comparison with historical data. A relatively low hydrogen content of incoming melt under 0.3 ml/100 g was measured with both probes. The hydrogen concentration after the box was in average 0.087 ml/100 g measured with AlSCAN probe and 0.102 ml/100 g according to Accurity probe. The reduction efficiency was above 60%. These values mean a significant improvement of the degassing efficiency, since in previous years an average reduction efficiency of 37% was reached. The filtration efficiency of the new box was evaluated by means of PoDFA. Sampling was done before and after the combi box and after addition of grain refiner. Figure 6 shows an overview of particle loading for measurements done in 2022 and in previous years from 2018 to 2020, when an old filter box was utilized. The particle concentration at the outlet of the combi box was very low in both samples (0.023 and 0.008 mm2/kg) and a very high reduction efficiency up to 99% could be reached. However, the concentration of inclusions in the melt entering the box was higher than in years before, which could have promoted a better filtration efficiency through the cake formation. Based on the initial results it can be concluded that the reduction efficiency of the new filter concept is at least comparable with previously used bonded particle filters, which supports the statement about the high filtration efficiency of CFF technology.

Initial Operational Results The combi box was successfully installed and commissioned at Karmøy in January 2022. After the installation, several minor improvements had to be done, but the total concept is

Fig. 5 Average hydrogen content in the melt before and after the combi box measured by AlSCAN and Accurity and corresponding reduction efficiency in comparison with historical data

TRC Combi Box: A Compact Inline Melt Treatment Unit for Continuous Casting

1021

Summary and Conclusion The new filter unit based on the specific requirement of the TRC process has been developed in cooperation with Speira GmbH and Drache Umwelttechnik GmbH. The base concept is a combination of three dedicated chambers for degassing, heating, and mechanical filtration in one unit. A pilot unit was successfully installed at Speira plant in Karmøy and has been running in a full-scale production mode since February 2022. No major changes in concept are considered. First results on the metallurgical performance of the box are very promising, although further long-term evaluations have to be done. The Karmøy casthouse plans to revamp another casting line with the Combi Box in 2023. Fig. 6 PoDFA evaluation of particle concentration and reduction efficiency of the Combi Box measured in 2022 in comparison with previously used filtration unit from 2018 to 2020

References

Due to the ongoing adjusting of the process parameters, the CFF plate has been exchanged quite often and a real capacity has not been determined yet. At the time of PoDFA sampling, the CFF plate was 3 weeks in use and approx. 170 t of melt had gone through it, which is three times higher than the nominal capacity claimed by the manufacturer. The evaluation of the temperature management is ongoing. The first results have demonstrated that the chosen configuration of the immersion heaters generates enough energy to maintain the melt temperature on the required level. In the steady state during the cast, the heaters use approx. 50% of their power capacity, which correlates with the data from the heat loss model quite well. The start of the new cast is the most demanding phase where the heaters are switched on full power. However, it is still in evaluation if the lower power input is also enough. Very positive feedback is received from the casthouse personnel. The new box is reported to be more ergonomic and provides better working conditions. Much less lifting and manual work is required and the posture during the work is straight. The time required for the exchange of the filter plate has been significantly reduced from several hours to several minutes.

1. Schneider, W., Stranggießen von Al-Werkstoffen, VAW Aluminium AG 1991, 17–18 2. Frank, R.A., Mattocks, M.F., Recent developments in SNIF technology to provide improved refining capacity. Proceedings of the second international workshop on aluminium melt processing technologies, 1998, 119–128 3. Bildstein, J., Ventre, I., Alpur Technology – Present and Future, Light Metals 1990, TMS 1990, 755–763 4. Taylor, M.B., Maltais, B., Experience with the sealed Alcan compact degasser (ACD), Aluminium Cast House Technology 2005 5. Maland, G., Myrbostad, E., Venas, K., Hycast I-60 SIR – a new generation inline melt refining system., Light Metals 2002, TMS, 855–859 6. Hakonsen, A., Haugen, T., Fagerlie, J.O., Inline melt treatment for low to medium metal flow rates., Light Metals 2016, TMS 2016, 817–820 7. ALPUR® G3 cc Line degasser for continuous casting, http://www. novelispae.com/alpur-degasser/ Accessed 29 August 2022 8. Le Roy, G., Chaleau, J.-M., Charlier, P.h., PDBF: Proven filtration for high-end applications., Light Metals 2007, TMS 2007, 651– 655 9. Bopp, J.T., Neff, D.V., Stankewicz, E.P., Degassing MulticastTM filtration system (DMC) – new technology for producing high quality molten metal., Light Metals 1987, TMS 1987, 729–736 10. Instone, S., Badowski, M., Schneider, W., Development of molten Metal filtration technology for aluminium., Light Metals 2005, TMS 2005 11. Counrtenay, J.H., Reusch, F., Instone, S., A review of the development of new filter technologies based on the principle of multistage filtration with grain refiner added in the intermediate stage., Light Metals 2011, TMS 2011, 769–774

CFD Modeling of Thin Sheet Product Using the Horizontal Single Belt Casting Method Daniel R. Gonzalez-Morales, Mihaiela M. Isac, and Roderick I. L. Guthrie

Abstract

The Horizontal Single Belt Casting (HSBC) process involves casting a thin “river” of molten metal directly onto a fast-moving, continuously cooled, endless belt. By casting at higher cooling rates than is common, uniform solidification rates across unconstrained “thin sheets” of melt are maintained, yielding excellent cast properties and microstructures. A non-isothermal analysis of heat flows in the critical region where the melt first pours onto the water-cooled belt is made and compared with results from an experimental caster. For this, the CFD ANSYS-Fluent 19.1 code was run, using various combinations of melt processing parameters. It provided accurate predictions of the “windows” available for stable strips of AA6111, AA5182, and AA2024 alloys, to be cast. The HSBC’s significant reductions in capital and operating costs, due to significantly reduced processing steps, lower energy consumption, and lower greenhouse gas emissions, make it a valuable alternative to current-day practices. Keywords








CFD modelling Horizontal Single Belt Casting (HSBC) Meniscus Thin strips

Introduction Horizontal Single Belt Casting (HSBC) is one of the surviving Near Net Shape Casting Processes (NSSC), which were being developed since the 1980s, as an alternative to traditional casting technologies like Continuous Casting (CC) and Direct-Chill (DC) casting [5, 6]. The advantages of the HSBC process include very significant energy and cost D. R. Gonzalez-Morales
M. M. Isac (&)
R. I. L. Guthrie Montreal, Canada e-mail: [email protected]

savings with associated environmental advantages, since thin metal sheets are directly obtained from molten metal poured from a tundish or launder. The final sheets have shown excellent cast microstructures for final thermomechanical processing for a wide range of copper, aluminium, and steel alloys [2, 3, 13]. The HSBC was independently conceived and patented by R. Guthrie and J. Herbertson in 1988, in Canada [9]. The operational principle of the HSBC is that molten metal, contained within a tundish or launder, is poured with the help of gravity through a slot nozzle (*3 mm), directly onto a moving, water-cooled, horizontally moving belt. This belt provides for the progressive solidification of the liquid metal along it. A schematic of the pilot-scale caster used, along with its component’s parts, is presented in Fig. 1. Extensive research of the HSBC process has been carried out, using both experimental (HSBC simulator and the HSBC pilot-scale system) and mathematical (computational fluid dynamics) routes, to determine the process parameters needed to obtain the desired properties of the final metal sheets. Some of the most important process parameters are the slot nozzle’s liquid metal inlet velocity, the speed of the moving belt, the vertical clearance of the refractory backwall from the belt, and its inclination. The McGill Metals Processing Center (MMPC) has done extensive experimental work using the pilot-scale caster, processing a large variety of thin strips of steel (high-strength high-ductility steels, electrical transformer steels, TRIP and TWIP steels), various grades of aluminum (AA6111, AA2024, AA5182, etc.), and Cu–Sn–Ni alloys. The HSBC process has also been successfully implemented industrially in North America, where it is in commercial use for producing thin strips of copper alloys, by Materion Brush, USA. In Europe, a similar HSBC process has been in operation by Salzgitter Gmbh, at the steel plant in Peine, Germany, casting various grades of steel including high Mn-steel alloys [15]. At McGill/McGill Metals Processing Centre, the HSBC technology has been successfully used for processing thin

© The Minerals, Metals & Materials Society 2023 S. Broek (ed.), Light Metals 2023, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-031-22532-1_136

1022

CFD Modeling of Thin Sheet Product Using the Horizontal Single Belt Casting Method

1023

Fig. 1 Schematic diagram of a HSBC pilot-scale machine

strips of the aluminum AA2024, AA5182, and AA6111 alloys. The chemical composition and produced sheets dimensions are presented in Table 1. Figure 2 presents the pilot-scale HSBC system processing thin strips of AA6111 aluminum alloy, providing a view towards the feeding system (a) and a view towards the in-line pinch rolls-mini mill (b) in operation. Optical microscopy of the three different aluminum alloys was then performed in order to evaluate the strips’ microstructures. The microstructures thereby obtained are shown in Fig. 3 for the three cases, at 50  magnification. For the AA2024 aluminum alloy [12], the final strip presented similar average roughness values on both top and bottom strip’s surfaces. A fine, uniform equiaxed microstructure, Fig. 3a, was obtained across the thickness of as-cast thin strip product. For the AA5182 aluminum alloy [11], the resulting 5.0-mm-thick strip microstructure, Fig. 3 b, presented acceptable levels of porosity and an average grain size of 63.1 µm, practically half the grain size of the 123 µm found for a DC (Direct Chill) cast sample. Additionally, desired secondary phases were found within the grains, rather than at grain boundaries, due to the uniformly

high cooling rates across the strip, for the HSBC process. This translated into an improved strength and hardness compared to a DC cast product. Regarding the use of the inclined feeding system, previous research has commented on the recommended practice for an isokinetic delivery of liquid metal onto the belt (inlet velocity = belt velocity) for reducing upper surface disturbances during casting. For the AA6111 alloy [13], HSBC pilot-scale experimental and CFD work was performed to compare the thin strip product quality when casting the AA6111 alloy (used in Ford F150 trucks), employing either single or double impingement feeding systems, with no side dams. The resulting strips again presented similar roughness values on both surfaces, with no surface indentations on the bottom surface. The strip microstructure, Fig. 3c, presented an average grain size of 85 µm with intermetallic phases distributed within the grains. Respective micrographs showing fine equiaxed grains, and some porosities (dark spots) were also detected, similarly to a DC cast product [13]. The resultant microstructures were suitable for rolling down directly into final sheet, without the need for performing additional long heat treatment annealing times

Table 1 Chemical composition and dimensions of aluminium alloys casted via HSBC pilot scale Al

Cu

Mg

Mn

Si

Zn

Thickness (mm)

Width (mm)

AA2024

Balance

3.8–4.9

1.2–1.8

0.3–0.9