Storing Energy: with Special Reference to Renewable Energy Sources [2 ed.] 0128245107, 9780128245101

Storing Energy: With Special Reference to Renewable Energy Sources, Second Edition has been fully revised and substantia

463 98 16MB

English Pages 871 [874] Year 2022

Report DMCA / Copyright


Polecaj historie

Storing Energy: with Special Reference to Renewable Energy Sources [2 ed.]
 0128245107, 9780128245101

Table of contents :
Front Cover
Storing Energy
Storing Energy: with Special Reference to Renewable Energy Sources
List of contributors
A - Introduction
1 - Global warming, greenhouse gases, renewable energy, and storing energy
1. Introduction
2. Global warming and greenhouse gases
3. Carbon dioxide in the atmosphere
4. Renewable energy
5. Our present energy situation
6. The urgent need for storing energy
7. Conclusion
2 - Energy storage options to balance renewable electricity systems
1. Introduction
2. The need for new types of storage
2.1 Impact of demands on generation imbalances
2.2 Strategies to cope with electricity system imbalances
3. Storage technologies
3.1 Gravitational/mechanical
3.2 Electrochemical and electrical
3.3 Thermal
3.4 Chemical
4. Comparing storage systems
5. Challenges for energy storage
5.1 Integrating energy storage into low-carbon energy systems
5.1.1 Generation-integrated energy storage
5.1.2 Analysing energy storage integration using models
5.2 Innovation to reduce technology costs
5.3 Adapting energy markets to realize the value of energy storage
5.4 Public acceptance
5.5 Regulatory barriers to energy storage
5.6 Finding the most appropriate roles for energy storage technologies
6. Conclusions
B - Gravitational/thermomechanical storage techniques
3 - Pumped hydro storage (PHS)
1. Introduction
2. Storage cycles duration
3. Conventional arrangement types
4. Services provided by PHS plants
5. New arrangements for PHS
5.1 Combined short and long-term cycle seasonal pumped hydro storage (CCSPHS)
5.2 Combined hydropower and pumped hydro storage (CHPHS)
5.3 Integrated pumped hydro reverse osmosis (IPHRO) system
5.4 Other arrangements
6. Pump-turbine types
7. World potential for PHS
8. Conclusion
4 - Novel hydroelectric storage concepts
1. Introduction
1.1 Scope and purpose
1.2 Constraints
1.3 How did we get here?
1.4 Novel hydroelectric storage categories
1.5 Future applications and markets
2. High-density fluid PHES
2.1 Background
2.2 Operation and performance characteristics
2.3 Development plans and economic performance
3. Piston-in-cylinder electrical energy storage
3.1 Background and operating principle
3.2 Piston versus PHES?
3.2.1 Location, location, location
3.2.2 Pressure by design
3.2.3 No energy limitations
3.2.4 No power limitations
3.2.5 Asymmetric charging
3.2.6 Other performance characteristics
3.2.7 What could possibly go wrong?
3.3 Piston storage economic performance
3.3.1 Capital cost
3.3.2 Finance cost
3.3.3 Operation and maintenance (O&M) costs
3.3.4 Energy costs
3.3.5 Overall storage costs
3.4 Markets for piston storage
3.4.1 Size matters, or does it?
3.4.2 Short-duration markets (

Citation preview

Storing Energy

This page intentionally left blank

Storing Energy with Special Reference to Renewable Energy Sources

Second Edition

Edited by

Trevor M. Letcher

Emeritus Professor, School of Chemistry, University of KwaZulu-Natal, Durban, South Africa

Elsevier Radarweg 29, PO Box 211, 1000 AE Amsterdam, Netherlands The Boulevard, Langford Lane, Kidlington, Oxford OX5 1GB, United Kingdom 50 Hampshire Street, 5th Floor, Cambridge, MA 02139, United States Copyright © 2022 Elsevier Inc. All rights reserved. No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher. Details on how to seek permission, further information about the Publisher’s permissions policies and our arrangements with organizations such as the Copyright Clearance Center and the Copyright Licensing Agency, can be found at our website: This book and the individual contributions contained in it are protected under copyright by the Publisher (other than as may be noted herein). Notices Knowledge and best practice in this field are constantly changing. As new research and experience broaden our understanding, changes in research methods, professional practices, or medical treatment may become necessary. Practitioners and researchers must always rely on their own experience and knowledge in evaluating and using any information, methods, compounds, or experiments described herein. In using such information or methods they should be mindful of their own safety and the safety of others, including parties for whom they have a professional responsibility. To the fullest extent of the law, neither the Publisher nor the authors, contributors, or editors, assume any liability for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions, or ideas contained in the material herein. Library of Congress Cataloging-in-Publication Data A catalog record for this book is available from the Library of Congress British Library Cataloguing-in-Publication Data A catalogue record for this book is available from the British Library ISBN: 978-0-12-824510-1 For information on all Elsevier publications visit our website at Publisher: Joseph P. Hayton Acquisitions Editor: Lisa Reading Editorial Project Manager: Leticia M. Lima Production Project Manager: Sojan P. Pazhayattil Cover Designer: Mark Rogers

Typeset by TNQ Technologies


List of contributors Preface

Section A 1



Global warming, greenhouse gases, renewable energy, and storing energy Trevor M. Letcher 1 Introduction 2 Global warming and greenhouse gases 3 Carbon dioxide in the atmosphere 4 Renewable energy 5 Our present energy situation 6 The urgent need for storing energy 7 Conclusion References Energy storage options to balance renewable electricity systems Paul E. Dodds and Seamus D. Garvey 1 Introduction 2 The need for new types of storage 3 Storage technologies 4 Comparing storage systems 5 Challenges for energy storage 6 Conclusions References

xv xxi


3 3 3 5 7 7 10 11 11 13 13 14 18 22 23 30 31

Section B Gravitational/thermomechanical storage techniques




Pumped hydro storage (PHS) Julian David Hunt, Behnam Zakeri, Andreas Nascimento and Roberto Brand~ ao 1 Introduction 2 Storage cycles duration 3 Conventional arrangement types

37 42 45



4 5 6 7 8





Services provided by PHS plants New arrangements for PHS Pump-turbine types World potential for PHS Conclusion Acknowledgments References

47 50 54 58 59 61 61

Novel hydroelectric storage concepts Frank Escombe 1 Introduction 2 High-density fluid PHES 3 Piston-in-cylinder electrical energy storage 4 Endpiece References


Gravity energy storage systems Miles Franklin, Peter Fraenkel, Chris Yendell and Ruth Apps 1 Introduction 2 History 3 Physics 4 The Gravitricity system 5 Technical characteristics 6 Levelized cost and comparison with other technologies 7 Market 8 Gravitricity technology development References


67 69 72 89 90

91 93 95 102 107 111 113 115 115

Compressed air energy storage (CAES) Seamus D. Garvey and Andrew Pimm 1 Introduction 2 CAES: modes of operation and basic principles 3 Air containments for CAES 4 System configurations and plant concepts 5 Thermal storage for CAES 6 Performance metrics for CAES 7 Integrating CAES with generation or consumption 8 Concluding remarks References Further reading


Compressed air energy storage Sabine Donadei and Gregor-S€ onke Schneider 1 Introduction 2 Mode of operation


117 119 125 131 135 137 138 139 139 140

141 142


3 4






Plant concept Underground storages References

Underwater compressed air energy storage Andrew Pimm and Seamus D. Garvey 1 Introduction 2 Storage vessels for UWCAES 3 Anchorage and installation 4 System configurations 5 Locations 6 Cost and efficiency 7 Contrasting UWCAES with pure gravitational storage approaches in deep water 8 State of development 9 Concluding remarks References A novel pumped hydro combined with compressed air energy storage system Erren Yao, Hansen Zou, Ruixiong Li, Huanran Wang and Guang Xi 1 Introduction 2 Basic principles of PHCA system 3 Characteristics of PHCA system 4 A novel constant-pressure PHCA system 5 Storage density analysis 6 Thermodynamic analysis 7 Results References

144 147 154 157 157 158 163 166 168 169 174 174 175 176

179 179 180 181 182 183 184 186 189

Liquid air energy storage Yulong Ding, Yongliang Li, Lige Tong and Li Wang 1 Introduction 2 Energy and exergy densities of liquid air 3 Liquid air as both a storage medium and an efficient working fluid 4 Applications of LAES through integration 5 Technical and economical comparison of LAES with other energy storage technologies References


Flywheel energy storage Keith R. Pullen 1 Introduction 2 Principles of operation 3 High-performance electric flywheel storage systems


191 192 194 196 203 205

207 209 217



4 5 6


Performance attributes in comparison with other electrical storage technologies Current and future applications Conclusion References

Rechargeable lithium-ion battery systems Matthias Vetter and Stephan Lux 1 Introduction 2 Physical fundamentals of lithium-ion batteries 3 Development of lithium-ion battery storage systems 4 System integration 5 Conclusions References

227 232 240 240 243 243 244 244 259 261 262

Section C Electrochemical and electrical energy storage techniques




The road to potassium-ion batteries Titus Masese and Godwill Mbiti Kanyolo 1 Introduction 2 The evolution of modern batteries 3 Mechanisms of lithium-ion battery operations 4 Cathode chemistries 5 Electrolytes 6 Anode materials 7 Beyond cation intercalation chemistries 8 Perspectives Acknowledgments References

265 265 266 269 277 288 294 298 300 300


Lithiumesulfur battery: Generation 5 of battery energy storage systems 309 Mahdokht Shaibani and Mainak Majumder 1 Introduction 309 2 Anatomy of LieS battery, challenges, and latest developments 310 3 Potential applications of lightweight LieS battery: existing, emerging, and new avenues 320 4 Conclusion and outlook: custom-designed LieS battery is on its way 323 References 324


Sodiumesulfur batteries Zhen Li and Jingyi Wu 1 Introduction 2 Principles of NaeS batteries

329 329 330


3 4 5 6 7 8





Technical challenges Cathode Anodes Electrolyte Cell configuration Conclusions and perspectives References

332 334 336 338 340 340 340

All-solid-state batteries Zhen Li and Yuyu Li 1 Introduction 2 Solid-state electrolytes (SSEs) 3 Interface in ASS-L/SIBs 4 Conclusion References


Vanadium redox flow batteries Christian Doetsch and Jens Burfeind 1 Introduction and historic development 2 The function of the VRFB 3 Electrolytes of VRFB 4 VRFB versus other battery types 5 Application of VRFB 6 Recycling, environment, safety, and availability 7 Other flow batteries References Further reading


Supercapacitors Narendra Kurra and Qiu Jiang 1 Introduction 2 Basics of charge storage 3 Historical evolution from capacitors to electrical double-layer capacitors 4 Models to explain electrical double layers 5 Evolution of electrode materials for supercapacitors 6 State-of-the-art energy storage technologies 7 Pseudocapacitive energy storage 8 Material requirements for achieving simultaneous high energy density at high power density 9 Electrochemical characterization techniques for supercapacitors 10 Energy storage devices 11 Applications of supercapacitors 12 Conclusions and challenges References


343 343 353 357 358

363 364 369 370 371 373 374 378 380

383 384 386 389 394 396 397 400 401 406 411 413 414




Sensible thermal energy storage: diurnal and seasonal Cynthia Ann Cruickshank and Christopher Baldwin 1 Storing thermal energy 2 Design of the thermal storage and thermal stratification 3 Modeling of sensible heat storage 4 Second law analysis of thermal energy storage 5 Solar thermal energy storage systems 6 Thermal storage integrated with heat pumps 7 Cold thermal energy storage 8 Seasonal storage 9 Concluding remarks References

419 419 420 422 426 427 428 429 432 438 438

Section D Thermal storage techniques



Storing energy using molten salts Michael Geyer and Cristina Prieto 1 Introduction to molten salt thermal energy storage systems 2 Molten salt energy storage uses 3 Molten saltsda medium for heat transfer and heat storage 4 Molten salt thermal storage system 5 Reference plant examples 6 Conclusions and outlook References


Pumped thermal energy storage Zhiwei Ma, Max Albert, Huashan Bao and Anthony Paul Roskilly 1 Introduction 2 Rankine PTES cycle 3 Brayton PTES cycle 4 Transcritical PTES cycle 5 Economics of PTES References


Phase change materials John A. Noël, Samer Kahwaji, Louis Desgrosseilliers, Dominic Groulx and Mary Anne White 1 Introduction 2 Heat storage at subambient temperatures 3 Heat storage at ambient temperature 4 Heat storage at moderate temperatures 5 Heat storage at high temperatures 6 Heat transfer in PCM-based thermal storage systems




445 452 465 471 479 482 483

487 489 493 497 500 501

503 508 511 514 518 522


7 8




Gaps in knowledge Outlook References

525 529 530

Solar ponds César Valderrama, José Luis Cortina, Aliakbar Akbarzadeh, Mohammed Bawahab, Hosam Faqeha and Abhijit Date 1 Introduction 2 Types of solar ponds 3 Investment and operational cost 4 Applications of solar ponds References


Hydrogen from water electrolysis Greig Chisholm, Tingting Zhao and Leroy Cronin 1 Introduction 2 Hydrogen as an energy vector and basic principles of water electrolysis 3 Hydrogen production via water electrolysis 4 Strategies for storing energy in hydrogen 5 Technology demonstrations utilizing hydrogen as an energy storage medium 6 Emerging technologies and outlook 7 Conclusion References


537 538 547 547 555

559 562 565 571 576 581 587 587

Section E Chemical storage techniques



Power-to-Gas Robert Tichler, Stephan Bauer and Hans B€ ohm 1 Introduction 2 Dynamic electrolyzer operation as a core part of power-to-gas plants 3 The methanation processes within power-to-gas 4 Multifunctional applications of the power-to-gas system 5 Underground gas storage in the context of power-to-gas References


Large-scale hydrogen storage Fritz Crotogino 1 Hydrogen economydfrom the original idea to the future concept 2 Why use hydrogen storage to compensate for fluctuating renewables? 3 Hydrogen in the chemical industry 4 Options for large-scale underground gas storage 5 Underground hydrogen storage in detail References



595 598 601 603 607 609

613 614 619 620 626 631






Traditional bulk energy storagedcoal and underground natural gas and oil storage Fritz Crotogino 1 Introduction 2 Coal 3 Oil 4 Natural gas storage 5 Summary References

633 633 634 636 641 649 649

Thermochemical energy storage Huashan Bao and Zhiwei Ma 1 Introduction 2 Overview of thermochemical sorption energy storage 3 Overview of thermochemical energy storage without sorption 4 Hybrid thermochemical sorption energy storage References


Energy storage integration Philip C. Taylor, Charalampos Patsios, Stalin Mu~noz Vaca, David M. Greenwood and Neal S. Wade 1 Introduction 2 Energy policy and markets 3 Energy storage planning 4 Energy storage operation 5 Demonstration projects 6 Integrated modeling approach References


651 653 667 672 676

685 686 692 699 705 715 724

Section F Integration




Off-grid energy storage Catalina Spataru and Pierrick Bouffaron 1 Introduction: the challenges of energy storage 2 Why is off-grid energy important? 3 Battery technologies and applications 4 Dealing with renewable variability 5 The emergence of mini- and microgrids 6 Energy storage in island contexts 7 Bring clean energy to the poor 8 The way forward: cost structure evolution 9 International examples 10 Conclusions References

731 732 734 740 741 742 743 744 745 749 749



Energy storage worldwide Catalina Spataru, Priscila Carvalho, Xiaojing Lv, Trevor Sweetnam and Giorgio Castagneto Gissey 1 Introduction: the global energy storage market 2 Barriers to the development and deployment 3 Case studies 4 Lessons for the development of storage 5 Conclusions References

Section G 32




International and marketing issues

Storing energy in Chinadan overview Haisheng Chen, Yujie Xu, Chang Liu, Fengjuan He and Shan Hu 1 Introduction 2 Imperativeness and applications 3 Technical and development status 4 Summary and prospects 5 Conclusions and remarks Acknowledgments References Further reading Legislation, statutory instruments and licenses for storing energy in UK Priscila Carvalho and Catalina Spataru 1 Introduction 2 Low-carbon policy in the UK for storage 3 Electricity markets and storage: legislation, statutory instruments, codes, and licenses 4 Standards applicable to storage 5 Regulatory, legal, and market constraints that impact storage 6 Conclusions References Electricity markets and regulatory developments for storage in Brazil Priscila Carvalho, Catalina Spataru and André Serr~ao 1 Introduction 2 Electricity market developments in Brazil: past, present, and future 3 Regulation of Brazilian electricity market


753 755 756 762 765 766

769 771 771 772 773 786 788 789 789 791

793 793 796 797 803 803 807 808

811 811 813 816



4 5 6 7


Distributed renewable generation: current state-of-the-art Electricity storage in Brazil Discussing challenges Conclusions References

817 818 827 828 828 831

List of contributors

Aliakbar Akbarzadeh Mechanical and Automotive Engineering, School of Engineering, RMIT University, VIC, Australia Department of Engineering, Durham University, Durham, United

Max Albert Kingdom Ruth Apps

Gravitricity, Edinburgh, Scotland

Christopher Baldwin Department of Mechanical and Aerospace Engineering, Carleton University, Ottawa, ON, Canada Department of Engineering, Durham University, Durham, United

Huashan Bao Kingdom Stephan Bauer

Green Gas Technology, RAG Austria AG, Wien, Austria

Mohammed Bawahab Department of Mechanical and Materials Engineering, Faculty of Engineering, University of Jeddah, Jeddah, Saudi Arabia Hans B€ ohm

Energieinstitut an der Johannes Kepler Universit€at Linz, Linz, Austria

Pierrick Bouffaron MINES ParisTech, PSL Research University, Centre de Mathématiques Appliquées, Paris, France; Berkeley Energy & Climate Institute, University of California, Berkeley, CA, United States Roberto Brand~ ao GESEL - Electric Sector Study Group, Rio de Janeiro, Brazil Jens Burfeind Fraunhofer Institute for Environmental, Safety, and Energy Technology UMSICHT, Oberhausen, Germany Priscila Carvalho Energy Institute, University College London, Central House, London, United Kingdom Giorgio Castagneto Gissey Energy Institute, University College London, Central House, London, United Kingdom Haisheng Chen Institute of Engineering Thermophysics, Chinese Academy of Sciences, Beijing, China Greig Chisholm Kingdom

School of Chemistry, University of Glasgow, Glasgow, United


List of contributors

José Luis Cortina Departament d’Enginyeria Química, Universitat Politecnica de Catalunya$BarcelonaTECH, Barcelona, Spain; Water Technology Center CETaqua, Barcelona, Spain Leroy Cronin Kingdom Fritz Crotogino

School of Chemistry, University of Glasgow, Glasgow, United DEEP-KBB, Hannover, Germany

Cynthia Ann Cruickshank Department of Mechanical and Aerospace Engineering, Carleton University, Ottawa, ON, Canada Abhijit Date Mechanical and Automotive Engineering, School of Engineering, RMIT University, VIC, Australia Louis Desgrosseilliers

Dalhousie University, Halifax, Nova Scotia, Canada

Yulong Ding Birmingham Centre for Energy Storage, School of Chemical Engineering, University of Birmingham, Birmingham, United Kingdom; School of Energy and Enviornmental Engineering, University of Science & Technology Beijing, Beijing, China Paul E. Dodds Kingdom

UCL Energy Institute, University College London, London, United

Christian Doetsch Fraunhofer Institute for Environmental, Safety, and Energy Technology UMSICHT, Oberhausen, Germany Sabine Donadei

DEEP.KBB GmbH, Hannover, Germany

Frank Escombe

EscoVale Consultancy Services, Reigate, United Kingdom

Hosam Faqeha Mechanical Engineering Department, College of Engineering and Islamic Architecture, Umm Al-Qura University, Makkah, Saudi Arabia Peter Fraenkel

Gravitricity, Edinburgh, Scotland

Miles Franklin

Gravitricity, Edinburgh, Scotland

Seamus D. Garvey Department of Mechanical, Materials and Manufacturing Engineering, University of Nottingham, Nottingham, United Kingdom Michael Geyer Institute of Engineering Thermodynamics, German Aerospace Center (Deutsches Zentrum f€ ur Luft- und Raumfahrt e DLR), Stuttgart, Germany David M. Greenwood School of Engineering, Newcastle University, Newcastle upon Tyne, United Kingdom Dominic Groulx

Dalhousie University, Halifax, Nova Scotia, Canada

Fengjuan He Institute of Engineering Thermophysics, Chinese Academy of Sciences, Beijing, China Shan Hu Institute of Engineering Thermophysics, Chinese Academy of Sciences, Beijing, China

List of contributors


Julian David Hunt International Institute for Applied Systems Analysis, Vienna, Austria; Federal University of Espirito Santo, Espirito Santo, Brazil Qiu Jiang School of Materials and Energy, University of Electronic Science and Technology of China, Chengdu, PR China Samer Kahwaji

Dalhousie University, Halifax, Nova Scotia, Canada

Godwill Mbiti Kanyolo Department of Engineering Science, The University of Electro-Communications, Chofu, Tokyo, Japan Narendra Kurra School of Chemistry, Indian Institute of Science Education and Research Thiruvananthapuram (IISER-TVM), Thiruvananthapuram, Kerala, India; Department of Chemistry, Indian Institute of Technology Hyderabad, Kandi, Sangareddy, Telangana, India Trevor M. Letcher South Africa

Chemistry Department, University of KwaZulu-Natal, Durban,

Ruixiong Li School of Energy and Power Engineering, Xi’an Jiaotong University, Xi’an, Shaanxi, China Yongliang Li Birmingham Centre for Energy Storage, School of Chemical Engineering, University of Birmingham, Birmingham, United Kingdom Yuyu Li Key Laboratory of Photoelectrochemical Materials and Devices, School of Optoelectronic Materials and Technology, Jianghan University, Wuhan, Hubei, China Zhen Li State Key Laboratory of Material Processing and Die & Mold Technology, School of Materials Science and Engineering, Huazhong University of Science and Technology, Wuhan, Hubei, China Chang Liu Institute of Engineering Thermophysics, Chinese Academy of Sciences, Beijing, China Stephan Lux Germany

Fraunhofer Institute for Solar Energy Systems ISE, Freiburg,

Xiaojing Lv China-UK Low Carbon College, Shanghai Jiao Tong University, Shanghai, China Zhiwei Ma Kingdom

Department of Engineering, Durham University, Durham, United

Mainak Majumder Department of Mechanical Engineering, Monash University, Melbourne, VIC, Australia Titus Masese Research Institute of Electrochemical Energy, National Institute of Advanced Industrial Science and Technology (AIST), Ikeda, Osaka, Japan; AISTe Kyoto University Chemical Energy Materials Open Innovation Laboratory (ChEMe OIL), Sakyoeku, Kyoto, Japan Stalin Mu~ noz Vaca

University of the Pacific, Guayaquil, Ecuador


List of contributors

Andreas Nascimento John A. Noël

Federal University of Espirito Santo, Espirito Santo, Brazil

Dalhousie University, Halifax, Nova Scotia, Canada

Charalampos Patsios School of Engineering, Newcastle University, Newcastle upon Tyne, United Kingdom Andrew Pimm School of Chemical and Process Engineering, University of Leeds, Leeds, United Kingdom University of Seville, Department of Energy Engineering, Seville,

Cristina Prieto Spain

Keith R. Pullen Honorary Visiting Professor, City University of London, London, United Kingdom Anthony Paul Roskilly United Kingdom Gregor-S€ onke Schneider André Serr~ ao

Department of Engineering, Durham University, Durham, DEEP.KBB GmbH, Hannover, Germany

Serr~ao Advogados, S~ao Paulo, Brazil

Mahdokht Shaibani Department of Mechanical Engineering, Monash University, Melbourne, VIC, Australia Catalina Spataru Energy Institute, University College London, Central House, London, United Kingdom Trevor Sweetnam Energy Institute, University College London, Central House, London, United Kingdom Philip C. Taylor Department of Electrical and Electronic Engineering, University of Bristol, Bristol, United Kingdom Robert Tichler Austria

Energieinstitut an der Johannes Kepler Universit€at Linz, Linz,

Lige Tong School of Energy and Enviornmental Engineering, University of Science & Technology Beijing, Beijing, China César Valderrama Departament d’Enginyeria Química, Universitat Politecnica de Catalunya$BarcelonaTECH, Barcelona, Spain Matthias Vetter Germany

Fraunhofer Institute for Solar Energy Systems ISE, Freiburg,

Neal S. Wade School of Engineering, Newcastle University, Newcastle upon Tyne, United Kingdom Huanran Wang School of Energy and Power Engineering, Xi’an Jiaotong University, Xi’an, Shaanxi, China

List of contributors


Li Wang School of Energy and Enviornmental Engineering, University of Science & Technology Beijing, Beijing, China Mary Anne White

Dalhousie University, Halifax, Nova Scotia, Canada

Jingyi Wu State Key Laboratory of Material Processing and Die & Mold Technology, School of Materials Science and Engineering, Huazhong University of Science and Technology, Wuhan, Hubei, China Guang Xi School of Energy and Power Engineering, Xi’an Jiaotong University, Xi’an, Shaanxi, China Yujie Xu Institute of Engineering Thermophysics, Chinese Academy of Sciences, Beijing, China Erren Yao School of Energy and Power Engineering, Xi’an Jiaotong University, Xi’an, Shaanxi, China Chris Yendell Gravitricity, Edinburgh, Scotland Behnam Zakeri International Institute for Applied Systems Analysis, Vienna, Austria; Sustainable Energy Planning Research Group, Aalborg University, Copenhagen, Denmark Tingting Zhao Kingdom

School of Chemistry, University of Glasgow, Glasgow, United

Hansen Zou School of Energy and Power Engineering, Xi’an Jiaotong University, Xi’an, Shaanxi, China

This page intentionally left blank


Renewable energy sources such as wind turbines and solar panels for electricity generation have become commonplace in our society. Their aim is to supply energy that is free from carbon dioxide production while sustainable and not dependent on a finite energy supply, such as coal and methane. Unfortunately, their full potential is reduced by their intermittency. In order for these and other developing renewable technologies, such as tidal current energy and wave energy, to make a real difference, we need to find effective ways to store their power. This book, an updated version of the first edition [1], showcases the current state of the different methods that are being explored to store energy; thus, making energy available not only when the sun is shining, the wind is blowing, the tides flowing, the sea currents moving, or when the waves are breaking. These storage methods are also useful in times when the demand for electricity is low and electricity can be bought cheaply and stored until demand rises and the stored energy can be used. At present, the chief way of storing larger amounts of energy is through pumped hydroelectric storage. Most countries have now exhausted the places where large reservoirs can be built so this focus on storing energy is both timely and necessary as the world moves toward sustainable carbon-free energy. Most of the chapters in the book focus on storing energy for electricity generation. However, the goal of the transition from fossil fuel to renewable energy is to decarbonize all energy sectors, not just the electricity sector. One possibility to generate energy for these other sectors is the production of hydrogen from renewable sources (green hydrogen) in order to match production and demand on the one hand and to ensure a fail-safe supply on the other hand, e.g., in the chemical industry. In recent years, there has been much discussion about sector coupling: The coupling of all energy sectors (electricity, industry, mobility, buildings). Indeed, the share of the additional sectors exceeds that of the electricity sector by far. In concrete terms, for example: • • •

in the industrial sector, the replacement of coal with green hydrogen in steel production, the replacement of natural gas with green hydrogen in ammonia and fertilizer production, in the mobility sector, the replacement of petrol, paraffin, and heavy oil with green hydrogen, green ammonia, or e-fuels.

There is still much catching up to do in decarbonization. However, the European Union, Germany, and especially the UK have made great efforts in recent years to make hydrogen available including storage in the future not only as an energy carrier but also as an energetic raw material. In England, for example, large projects are



planned on the east coast to produce blue hydrogen from natural gas, with the resulting CO2 to be stored in Norwegian depleted gas reservoirs. In Africa, Australia, and South America (Patagonia), huge wind or solar farms are planned to produce hydrogen, which can then be shipped in the form of cryogenic liquefied hydrogen to industrialized countries with limited solar or wind resources for use in the various energy sectors. There are plans to harvest solar power in Western Australia to produce hydrogen for Japan. This important aspect involving other energy sectors is described in more detail in Chapter 27. Some chapters in the book are concerned with developments in well-known energy storage techniques, others are concerned with new techniques which are being tested and researched for the first time, and a few involve techniques which have yet to leave the drawing board. This new edition contains 34 chapters as opposed to 25 chapters in the first edition and contains new topics such as super capacitors and pumped thermal energy, gravity energy storage, a number of different types of battery systems, thermochemical energy storage, and molten salts energy storage. Unfortunately, there still remains some topics that have eluded us and one is superconducting magnetic energy storage (SMES). This has recently been reviewed by Weijia Yuan and Min Zhang [2]. Facilities for SMES exist all round the world for use in power quality control and for grid stabilization and units of 1 MW h are not uncommon. Our book is a natural follow-up to our recently published book Future Energy: Improved Sustainable and Clean Options for our Planet 3rd edition [3]. In Future Energy, the case was made for developing new and sustainable energy sources in the light of climate change and increasing levels of greenhouse gases. In many ways, as discussed in chapter 1 of our book, storing energy also goes hand in hand with another book we published recently: Climate Change: Observed Impacts on Planet Earth, 3rd edition [4]. The present book is divided into six sections: Introduction; Gravitational/Mechanical Storage Techniques; Electrical Energy Storage Techniques; Thermal Storage Techniques; Chemical Storage Techniques; and International and Marketing Issues. Many Governments and people of influence throughout the world are supporting the drive to reduce our dependency on fossil fuels with interesting and innovative programs. One such program is the Global Apollo Programme, spearheaded by Sir David King, which calls for £15109 (£15 bn) a year to be spent on research, development, and demonstration of green energy and energy storage. Significantly, this amount is the same, in today’s money, that the US Apollo program spent in putting astronauts on the moon. Professor Martin Rees, former head of the Royal Society and another member of the Apollo group, explains the reason for using the name Apollo: “NASA showed how a stupendous goal could be achieved, amazingly fast, if the will and the resources are there.” This book has been produced in order to allow the reader to have an understanding and insight into a vital aspect of our future deployment of renewable energydits storage. The final decision as to which option should be developed in a country or region must take into account many factors including: topography, for example: are



there suitable sites for reservoirs to tap into PHES? are there convenient salt formations for creating caverns available for gas storage? is the amount of sunlight available sufficient for solar energy systems? is it possible to take advantage of thermal energy storage? is the chemical industry infrastructure sufficiently mature? is it possible to install electrolysis plants for hydrogen production or develop chemical reaction storage or install a sophisticated battery system? and is the density of population an important issue and should off-grid technologies be incorporated or can network integration and smart grids be used? It is also to be hoped that the book will act as a springboard for new developments. One way that this can take place is through contact between readers and authors and to this effect e-mail addresses of the authors have been included. The adherence of IUPAC to the International System of Quantities through its Interdivisional Committee for Terminology, Nomenclature and Symbols (ICTNS), is reflected in the book with the use of SI Units throughout. The index notation is used where possible in order to remove ambiguities; for example, billion and trillion are written as 109 and 1012, respectively. To further remove any ambiguities, the concept of the quantity calculus is used. It is based on the equation: physical quantity ¼ number  unit. To give an example: power ¼ 200 W and hence: 200 ¼ power/W. This is of particular importance in the headings of tables and the labeling of graph axes. This volume is unique in the genre of books of related interests in that each chapter of Storing Energy has been written by an expert scientist or engineer, working in the field. Authors have been chosen for their expertise in their respective fields and come from 12 countries: Australia, Austria, Canada, China, France, Germany, Japan, South Africa, Spain, Switzerland, United Kingdom, and the USA. Most of the authors come from developed countries as most of the research and development in this fast-moving field, presently, come from these countries. We look forward to the future when new approaches to storing energy from scientists and engineers working in developing countries will be developed which focus on their local conditions. A vital concern related to future energy and storing energy is: what is to be done when it appears that politicians misunderstand or ignore and corporations overlook, the realities of climate change, and the importance of renewable energy sources? The solution lies in sound scientific data and education. As educators, we believe that only a sustained grassroots movement to educate citizens, politicians, and corporate leaders of the world has any hope of success. This book is part of this aim. It presents options for readers to consider and we hope that not only students, teachers, professors, and researchers of renewable energy, but politicians, government decision makers, captains of industry, corporate leaders, journalists, editors, and all interested people will read the book, take heed of its contents, and absorb the underlying message that renewable energy sources are our future and storing energy is a vital part of it. I wish to thank all 84 authors and coauthors for their cooperation, help, and especially for writing their chapters. It has been a pleasure working with each and every one of the authors. I thank my wife, Valerie, for all the help she has given me



over these long months of putting the book together. I also wish to thank Elsevier and Lisa Reading and Leticia Lima in particular, for their professionalism and help in producing this well-presented volume. Trevor M. Letcher Stratton on the Fosse Somerset, United Kingdom April 2021 Fritz Crotogino Hannover Germany April, 2021

References [1] Storing Energy: with Special Reference to Renewable Energy Systems, Elsevier, Oxford, UK, 2016, ISBN 9780128034408. [2] Weijia Yuan, Min Zhang, in: Jinyue Yan (Ed.), A Handbook of Clean Energy System, John Wiley, Chichester, UK, 2015, [3] T.M. Letcher (Ed.), Future Energy: Improved Sustainable and Clean Options for our Planet, 3rd edition, Elsevier, Oxford, UK, 2020. [4] T.M. Letcher (Ed.), Climate Change: Observed Impacts on Planet Earth, 3rd edition, Elsevier, Oxford, UK, 2021.

Section A Introduction

This page intentionally left blank

Global warming, greenhouse gases, renewable energy, and storing energy


Trevor M. Letcher Chemistry Department, University of KwaZulu-Natal, Durban, South Africa

1. Introduction The impact of climate change, as a result of the build-up of carbon dioxide in the atmosphere, is slowly transforming the way we make electricity. To reduce the carbon dioxide levels it is necessary to develop and deploy renewable forms of energy. Apart from tidal energy all renewable energy forms such as hydroelectric power, wave energy, solar energy, and wind energy are the result of the sun’s rays falling on the earth. Solar cell technologies and wind turbines are proving to be the frontrunners in harnessing solar and wind energy, but unfortunately they are both intermittent. To successfully use solar and wind energy for grid electricity or even for private electricity production, storing of electricity in times of overproduction is vitally important. This book looks at the very latest storage options available to us. In this introduction we look at the evolving global heating problem and how it is related to the rise of greenhouse gases of which CO2 is of particular importance. We also look at the energy situation today and focus on the problems of replacing fossil fuel with renewables and investigate the grid energy options. This leads us into the storing of energy in the following 33 chapters.

2. Global warming and greenhouse gases The concept of the greenhouse effect goes back to 1827, when Jean-Baptiste Fourier suggested that some component of the earth’s atmosphere was responsible for the temperature at the surface of the earth. He noted the similarity between what happens under the glass of a greenhouse and how heat is absorbed in the atmosphere. This has led to the term “greenhouse effect.” Fourier was researching the origins of ancient glaciers and the ice sheets that once covered much of Europe [1]. Decades later in 1859, Tyndall followed up the Fourier’s suggestion, and using the first spectrophotometer, measured the radiant heat absorptive powers of gases such as water vapor, carbon dioxide, oxygen, nitrogen, and ozone and showed that H2O, CO2, and O3 absorb heat radiation very much more than does N2 or O2. His wonderfully innovative apparatus, using the thermopile for measuring radiant heat, as designed by Macedonio

Storing Energy. Copyright © 2022 Elsevier Inc. All rights reserved.


Storing Energy

Melloni in Italy, can be seen at the Royal Society of Chemistry’s headquarters in London [2,3]. Tyndall concluded that water vapor is the strongest absorber of radiant heat and is the most important gas in controlling the earth’s temperature; and that without water vapor the “Earth would be held fast in the iron grip of frost.” Tyndall’s results fitted in with Fourier’s concept and pointed to CO2 and H2O as the components in the atmosphere that Fourier was looking for. Tyndall’s results were published in Refs. [4,5] and as a result, Tyndall has been named as the discoverer of the CO2 and H2O greenhouse gas effect. Linking CO2 in the atmosphere to the burning of fossil fuels was to be the last link in the chain in understanding the reasons for the ice ages and also our own climate change. In the 1890s, Svante Arrhenius, an electrochemist, calculated that by reducing the amount of CO2 in the atmosphere by half the temperature of Europe would be lowered by about 4e5 C. This would bring it in line with ice age temperatures. This idea would only answer the question of why the ice age formed and then retreated, if there were large changes in atmospheric composition and in particular, changes in CO2 concentration. At much the same time, also in Sweden, a geologist, Arvid Högbom, had estimated that CO2 from volcanic eruptions, together with the ocean uptake of CO2, could explain how the CO2 concentrations in the atmosphere could change and hence provide some explanation for the ice ages. Along the way Högbom stumbled on a strange and new idea that the CO2 emitted from industrial coal burning factories might influence the atmospheric CO2 concentration. He did indeed find that human activities were contributing CO2 to the atmosphere at a rate comparable to the natural geochemical processes. The increase was small compared to the CO2 already in the atmosphere, but if continued, it would influence the climate. Arrhenius, took up this concept, and his calculations were published in 1896 in Ref. [6]. Arrhenius concluded that the emissions from human industry might someday bring on global warming. Hence, Arrhenius’s name is also forever linked to the greenhouse theory of global warming. Arrhenius’s calculations were at first dismissed as unimportant or at worst faulty. A similar fate was met by G. S. Callendar who, in 1938, made the point that CO2 levels were indeed climbing [7]. It was only in the 1960s, after C. D. Keeling measured the CO2 concentration in the atmosphere and showed that it was rising rapidly, that scientists woke up to the fact that global warming was real and that anthropogenic activity was to blame. A recent graph relating the mean global temperature to carbon dioxide concentration in the atmosphere is given in Fig. 1.1. This is a striking piece of evidence strongly supporting the idea that rising global temperatures are indeed related to the concentration of CO2 in the atmosphere. Unfortunately, the relationship cannot be proven absolutely as the definitive experiment of drastically reducing the level of CO2 in the atmosphere or even stopping anthropological production of CO2 cannot be done. Furthermore, even if we did stop producing CO2, its effects in the atmosphere would continue as the half-life of CO2 in the atmosphere is of the order of many decades. However, all the evidence points to CO2 as being the root cause and that other possible effects such as the periodic variations in the sun’s radiation, the wobbling movement of the earth, space weather, and volcanoes have been discounted [8]. But this is not the full story.

Global warming, greenhouse gases, renewable energy, and storing energy


Figure 1.1 The mean global temperature versus CO2 concentration from 1880 to 2019. Figure by kind permission of Climate Central (

As discovered by Tyndall, water vapor is an even more effective greenhouse gas than CO2. Furthermore, its concentration in the atmosphere is very much higher than that of CO2 (of the order of a 100 times higher), and as a result H2O contributes over 60% of the global warming effect. The amount of water vapor in the atmosphere is controlled by the temperature. An increase in the CO2 concentration in the atmosphere results in a relatively small increase of the global temperature but that change is enough to increase the amount of water vapor in the air, through evaporation from the oceans. It is this feedback mechanism that has the greatest influence on global temperature. In a sense, paradoxically, the concentration of CO2 acts as a regulator for the amount of water vapor in the atmosphere and is thus the determining factor in the equilibrium temperature of the earth. Without CO2 in the atmosphere the temperature of the earth would be very much cooler than it is today.

3. Carbon dioxide in the atmosphere The conclusion that CO2 is the cause of our present global warming has been supported by The United Nations organization [9] and their summarized conclusions are: • • •

The concentration of GHGs in the earth’s atmosphere is directly linked to the average global temperature on earth (see |Fig. 1.1); The concentration has been rising steadily, and mean global temperatures along with it, since the time of the Industrial Revolution (see Fig. 1.1); The most abundant GHG, apart from water vapor, is carbon dioxide (CO2) and is largely the product of burning fossil fuels [10].


Storing Energy

What follows here is a brief scientific explanation to support the UNs summary. For many thousands of years (possibly in excess of a million years) the earth’s climate has been in a state of equilibrium with the concentration of atmospheric carbon dioxide at about 280 ppm (in parts per million which refers to the ratio of number of molecules of CO2 in the atmosphere to the number of molecules of all the gases in the atmosphere) (0.028%). It is this small concentration in the upper atmosphere that has kept the earth at a relatively warm temperature. Without the CO2 the average global temperature would be about 33 C cooler than it is today [10]. This greenhouse effect is due to the adsorption by the vibrating bonds of the CO2 molecules (and other GHGs such as water), of the infrared heat leaving the earth after the earth has been heated by the sun’s rays (largely by ultraviolet short-wave radiation). The vibrating bonds have the same frequencies as sections of the infrared radiation emanating from the earth (as a result of the radiant heat coming from the sun)dhence the sympathetic adsorption of energy. Since the Industrial Revolution, the CO2 levels have increased; in 1960 it was 316 ppm and in 2020 it has reached 417 ppm [11]. This rise, largely due to human activities (such as the burning of fossil fuels), has caused an increase in global temperature. The total amount of CO2 in the atmosphere and its concentration value are the most dependable measurements we have for the progress of global warming. In the 1960, the rate of increase of CO2 (as measured at Mauna Loa, in Hawaii) was less than 1 ppm per year. It is now 2.4 ppm per year [11]. It is this rate of change that is the best indicator of any progress we are making in reducing global warming. At the moment there is no sign that this is happening, in fact the reverse is true. Even if we stopped burning fossil fuel, the CO2 levels will take a long time to decrease as the lifetime of CO2 in the upper atmosphere is of the order of hundreds of years. There are other GHGs, such as methane gas, CH4, and NOxs, and details of them are given in Ref. [10]. Before the 18th century, the earth’s climate had been in a state of equilibrium and weather patterns were more or less predictable. This equilibrium no longer holds and it is becoming impossible, with the destabilizing effect of greenhouse gases (GHGs), to predict future climate and weather patterns with any certainty. What would be good is that global warming stabilizes and the climate of the earth reaches a new equilibrium which we could learn to cope with. But with steadily increasing levels of GHGs, largely as a result the burning fossil fuels and an increasing global population, we are moving toward an ever-increasing unstable climate in which record-levels of rising global temperatures and extreme heat and heatwaves, wildfires, tornadoes, hurricanes, droughts, floods, and food shortages have become the new norm. Furthermore, this destabilizing effect creates economic, political, and social chaos, which is being felt in every part of the world. The stabilizing of a warmer climate is the very best we can hope; the levels of CO2 that we have now will be with us for a very long time. It is not only the vibrating GHG molecules and the evaporation of water feedback process that we have to worry about, there are also other feedback mechanisms at play which further exacerbates global warming. These include the melting of ice sheets and the albedo effect; the solubility of CO2 in the

Global warming, greenhouse gases, renewable energy, and storing energy


ocean decreases with increasing temperature; the warming of peatbogs and the release of methane; and the breakdown of methane clathrates as a result of the warming of the earth and the oceans.

4. Renewable energy Most of the renewable energy, as exploited on earth, comes from the sun’s energy in one form or another. If we are looking for renewable energy on earth, then that is as far as we need lookdthere are no other significant sources of energy entering the earth’s atmosphere. This renewable energy includes hydropower which is a result of the water cycle, which in turn depends on energy from the sun; wind energy which is also a secondary effect of the sun’s energy; wave energy which is largely a result of wind movement; and of course, direct solar energy. The only renewable energy not a direct function of the sun’s energy is tidal energy which is largely due to astronomical (mainly moon) gravitational effects and the shape of coastlines which serve to channel gravitationally induced ocean movements. Geothermal energy is often erroneously considered as a renewable source; it is a result of radioactive processes deep inside the earth. Solar and wind energy are at present two important renewable energy sources, and it is their intermittency that makes it necessary to store energy in times of oversupply for use in times when the sun is not shining or the wind is not blowing. Details of renewable options such as hydroelectric power; wind energy; various forms of photovoltaic cells to capture solar energy; wave energy concentrated solar energy; solar thermal energy; biomass; and geothermal energy can be found in reference [12]. The theoretical potential of solar energy reaching the earth is 89.3  1015 W. That is the energy (in units of joules) reaching the earth per second. At present, the world uses 170 PWh or 610 EJ per year (i.e., 610  1018 J) [13,14]. The total energy reaching the earth in a year will be 89.3  1015  60  60  24  365 ¼ 2816  1021 J. Thus, our total energy needs on earth amounts to only 0.022% of the energy from the sun. Thus, there is a lot more renewable energy available for exploitation. Put in another way, assuming our harvesting of the sun’s energy is only 10% efficient, we would require a land mass of about the size of Namibia for our solar farm.

5. Our present energy situation The Copernicus Climate Change Service (C3S) [15] has revealed that globally 2020 was tied with the previous warmest year 2016, making it the sixth in a series of exceptionally warm years starting in 2015, and 2011e20 the warmest decade recorded. Furthermore, Europe saw its warmest year on record, 0.4 C warmer than 2019 which was previously the warmest year. According to independent analyses by NASA and the National Oceanic and Atmospheric Administration (NOAA), earth’s average global surface temperature in 2019 was the second warmest since modern


Storing Energy

record-keeping began in 1880. Globally, 2019’s average temperature was second only to that of 2016 and continued the planet’s long-term warming trend: the past 5 years have been the warmest of the last 140 years. The year 2019 was 0.98 C warmer than the mean temperature measured between 1951 and 1980, according to scientists at NASA’s Goddard Institute for Space Studies (GISS) in New York [16]. This shows that we are not doing enough to reduce global warming and the only way to do that is to reduce the amount of CO2 we are pumping into the air, and if possible, begin removing CO2 from the atmosphere. At present about 40% of all energy sources used to produce electricity are either renewable (wind, solar, hydropower, biomass, tide, and geothermal) or nuclear (10%) [17,18]. The major electricity producing energy source is coal which still produces 37% of the world’s electricity. The changeover from fossil fuel to renewables is slow and the prediction for 2040 is that renewables would produce 45% of the world’s electricity, with coal still a significant supplier of energy [19]. Replacing fossil fuel is going to be a mammoth task. It has been estimated that in 2019 human activities contributed 36.8  109 t of CO2 through burning coal and other fossil fuels, cement production, deforestation, and other landscape changes. It has also been estimated that since the Industrial Revolution, over 2000  109 t of CO2 has been added to the atmosphere. Human activities emit 60 or more times the amount of carbon dioxide released by volcanoes each year [20]. The population of the world is increasing and so is the need for more energy with a greater demand for more electricity. The world population (it is now 7.8  109 according to the latest 2020 United Nation estimate) is expected to reach 9  109 in 2050. It is increasing at a rate of 1.05% per year at the moment (2020) down from 1.14%/yr in 2016 and down from a recent peak in 1963 of 2.2%/yr. The expected rate of growth in energy demand over the next decade is greater than the growth rate of the population; this is largely due to the increase in demand for electricity in developing countries. Electricity generation is expected to increase from 25  1012 kWh in 2017 to 31.2  1012 kWh in 2030 an increase of almost 2% per year [21]. The relative breakdown of electricity producers and future predictions is given in Table 1.1. At the moment, coal is still the largest producer of electricity worldwide, Table 1.1 Breakdown of electricity production worldwide and a prediction over the next two decades [22]. Electricity production Energy (1012 kWh)

Oil Nuclear Renewables Natural gas Coal





1.06 2.34 4.73 4.83 8.60

0.86 3.05 6.87 5.20 9.73

0.62 3.95 8.68 7.47 10.12

0.56 4.50 10.63 10.10 10.62

Global warming, greenhouse gases, renewable energy, and storing energy


and is not expected to be overtaken by renewables until 2040. This illustrates the energy dilemma of our timedthe positive and encouraging increase in the deployment of renewable forms of energy is masked by the increasing overall energy needs of the world and that increase is still being met by further increases in fossil fuel usage. The present and future world electricity generation is dominated by the burning of fossil fuels (over 60%) and the prediction for 2040 is not much better (58%), but with coal reduced to 29%. This is driven by a number of forces including the relative economics of fossil fuels versus renewable energy, the massive inertia linked to status quo investments and situations, and the fear of things new as opposed to well-tried technologies. Electricity production is not the only producer of CO2 in our atmosphere. The various sectors responsible for CO2 generated as a result of human activity are given in Table 1.2. If there is the necessary political will to do so, we can replace the fossil fuele derived electricity with renewable forms of energy, or with nuclear energy. However, we do have a problem with replacing transport fuel. We could one day have electric cars replace petrol vehicles and possibly even diesel vehicles, but replacing fossil fuel for air travel and sea travel is difficult, if not impossible. Furthermore, some industrial processes such as cement manufacture, involving the heating of CaCO3 resulting in the waste product CO2 are also problematic. Attempts at replacing petrol in transport with renewable fuel derived from biomass (sugar cane as done in Brazil or corn as done in the United States for petrol, and palm oil in Malaysia for biodiesel) has had some success, but the overall contribution has been relatively small. In 2018 the biofuels contributed 3% to the world’s transport fuels. The United States, Brazil, and Malaysia are the world leaders in biofuels [24]. All of this does indicate that the world is not on top of solving the global warming problem, in spite of the steady increase in the deployment of renewable forms of energy. The change-over from fossil fuel to renewables is just too slow. It is predicted that renewables will increase their share of electricity production from 26.7% in 2020 to 29.2% in 2040 (less than 0.3% per year) (see Table 1.1). We will have to work very much harder to replace fossil fuel as the main driving force of our energy industry. One slight glimmer on the horizon is the fact that natural gas, methane (including shale gas), is better for the planet than burning coal and in many countries, coal is being replaced by natural gas. The reason why natural gas is better than coal is that the amount of CO2 produced from burning CH4 per unit of energy (50 g/MJ) is less

Table 1.2 Worldwide source of CO2 (mostly fossil fuel) emissions, 2018 [23]. Carbon dioxide emissions (%) Electricity Transport Industrial (including cement manufacture) Residential (heating, wood fires) Agriculture other

27 28 22 12 11


Storing Energy

than it is for coal (92 g/MJ), and moreover, coal burning produces particulates. Of course, the burning of CH4 still produces CO2: CH4 þ 2O2 ¼ CO2 þ 2H2O Wind power is vitally important to the UK’s energy production and the UK has experienced a renewables revolution over the past decade with major growth rates in both solar and wind energy. For example, on December 26, 2020, more than half of the UK’s electricity came from wind turbines thanks to winds associated with storm Bella and 60% in August 2020 over a 24-hour period. Overall, wind and solar delivered almost 30% of electricity in the UK during 2020. Of the electricity generated in 2019, the renewables’ share was 36.9% and gas 40.9% with coal only 2.1% [25].


The urgent need for storing energy

One of the difficulties of grid electricity is that the incoming electricity must balance the output electricity. With the deployment of renewable but intermittent forms of energy, sometimes the output is less that the incoming production and some wind turbines and solar energy cells have to be turned off. One often sees wind turbine vanes in our countryside at a standstill while the wind is blowing. If only the excess renewable energy could be stored. This is the raison d’etre for this book. Looking back at the grid scenario, it is surprising that storing energy from renewable energy sources appears to have come rather late on the scene; in hindsight, one would have thought that the development of renewable energy should have gone hand in hand with the development of energy-storing methods. Today there are many viable options available to store energy from renewable energy sources; they provide flexibility for the grid to ensure an uninterrupted power supply to consumers. This flexibility is critical to both reliability and resilience. This is particularly important as the cost of outages continues to rise. The most important issue is that energy storage can smooth out the delivery of intermittent resources such as wind and solar, by storing excess energy when the wind is blowing and the sun is shining, and delivering it when needed. Another vitally important aspect of energy storage is that it can also support the efficient delivery of electricity for inflexible, baseload resources, such as nuclear energy. When demand changes quickly, and flexibility is required, energy storage can inject or extract electricity as needed to exactly match the load, at very short notice. Energy storage can save operational costs in powering the grid and can reduce the cost in providing frequency regulation and spinning reserve services, as well as offsetting the cost to consumers by storing low-cost energy and using it later, during peak periods at higher electricity rates. Furthermore, by using energy storage during brief outages, businesses can avoid costly disruptions and continue normal operations. From the point of view of home owners, if grid electricity is unreliable then home installed energy storage systems (e.g., batteries) can save residents from waste of food and medicines, and the inconvenience of not having electricity.

Global warming, greenhouse gases, renewable energy, and storing energy


In summary, energy storage enables electricity to be saved for a later time. By introducing more flexibility into the grid, energy storage can help integrate more solar, wind, and distributed energy resources. It can also improve the efficiency of the gridd increasing the capacity factor of existing resourcesdand offset the need for building new pollution-emitting peak power plants. Above all, storing electricity is important, as it allows us to deploy renewable but intermittent energy sources so that we can begin to reduce greenhouse gas emissions and save the planet from overheating. Energy storage is an enabling technology. When the sun isn’t shining or the wind isn’t blowing, stored energy could be used. When demand shifts and baseload resources can’t react quickly enough, energy storage can fill the breach.

7. Conclusion No book on storing energy from renewable sources should go without a discussion on why storing of energy has become important over these past two decades. This chapter details and references the background to climate change, anthropological production of CO2, and the necessary deployment of renewable forms of energy. The idea of storing energy in significant amounts to ensure a smooth-running National Grid is a relatively new concept. Over the past few years no one technique has proved to be the silver bullet to solve the problem of intermittency of solar or wind energy, in particular. The chapters in this book highlight the many energy storage options available to us, some of which are in service at the moment; others that are still on the drawing boards and some at the pilot stage while this book was being written with preliminary results are published here. Energy storage can save operational costs in powering the grid, as well as save money for electricity consumers who install energy storage in their homes and businesses. But above all, storing energy can help to reduce the anthropological production of CO2 and reduce global warming.

References [1] J. Fourier, Remarques Générales sur les Températures Du Globe Terrestre et des Espaces Planétaires, Ann. Chim. Phys. 27 (1824) 136e167. [2] L. Nobili, M. Melloni, Le Thermo-multiplicateur, Ann. Chim. Phys. 48 (1831) 198e199. [3] A. Sella, Melloni’s thermomultiplier, Chem. World 15 (2018) 70. [4] J. Tyndall, On the absorption and radiation of heat by gases and vapours, Philos. Mag. Ser. 4 22 (169e94) (1861) 273e285. [5] J. Tyndall, On radiation through the earth’s atmosphere, Philos. Mag. Ser. 4 25 (1863) 200e206. [6] S. Arrhenius, On the influence of carbonic acid in the air upon the temperature of the ground, Philos. Mag. J. Sci. 41 (1896) 237e276. [7]


Storing Energy

[8] T.M. Letcher (Ed.), Climate Change: Observed Impacts on Planet Earth, second ed., Elsevier, 2016 (Chapters 24e30). [9] [10] [a] R. Tuckett. In Chapter 2 of T.M. Letcher (Ed.), Climate Change: Observed Impacts on Planet Earth, third ed., Elsevier, 2021. [b] [11] [12] T.M. Letcher (Ed.), Future Energy: Improved, Sustainable and Clean Options for Our Planet, third ed., Elsevier, 2020. [13] [14] [15] [16]¼NASA%2C%20NOAA%20Analyses%20Reveal%202019%20Second %20Warmest%20Year%20on%20Record,-Credit%3A%20NASA’s%20Goddard&text¼ According%20to%20independent%20analyses%20by,record%2Dkeeping%20began%20in %201880. [17] [18] [19] [20] [21] [22], Energy & Environmental Services, Electricity. [23] [24] [25] data/file/877047/Press_Notice_March_2020.pdf.

Energy storage options to balance renewable electricity systems


Paul E. Dodds 1 and Seamus D. Garvey 2 1 UCL Energy Institute, University College London, London, United Kingdom; 2Department of Mechanical, Materials and Manufacturing Engineering, University of Nottingham, Nottingham, United Kingdom

1. Introduction Energy storage makes a vital contribution to energy security in existing energy systems. At present, most energy is stored as raw or processed hydrocarbons, whether in the form of coal heaps or oil and gas reservoirs. Since electricity storage is much more expensive by comparison, precursors to electricity are stored rather than electrical energy, and generation is varied to meet demand. The principal exception to this modus operandi is pumped-hydroelectric generation, which can generate a large power output for a short period, at very short notice, and is used to increase electricity system stability. Sustained innovation has reduced the cost of newly installed wind and solar generation below the cheapest fossil fuel plants in some countries [1]. As energy systems gradually evolve toward using renewables, and perhaps fixed-output nuclear generation, the role and type of energy storage is likely to change substantially. Two broad trends are likely to drive this transition. First, it will become increasingly difficult to match electricity supply to demand, with imbalances becoming both larger and more common over time. The move away from fossil generation will mean that it will no longer be possible to store most electricity precursors as hydrocarbons, with the exception perhaps of gas for flexible generation. Second, if low-carbon electricity displaces oil and gas for transport and heat provision, electricity demand patterns could change substantially, with peak demands becoming much more pronounced. Numerous energy storage technologies are under development that store electricity at times of excess supply in order to meet periods of high demand. Other storage technologies could support the energy system in other ways, for example, by storing excess electricity as heat or hydrogen for use in other sectors. The move to a low-carbon economy will cause nothing less than a revolution in how energy storage is used. This chapter examines the development of novel energy storage technologies to support low-carbon energy systems. It introduces a wide range of energy storage technologies, which are explored in this book, and identifies key characteristics with which to compare the technologies. Finally, it identifies challenges for commercializing and deploying these technologies into existing energy systems in the short to medium term.

Storing Energy. Copyright © 2022 Elsevier Inc. All rights reserved.



Storing Energy

The need for new types of storage

Most new storage technologies are designed to contribute to low-carbon electricity systems. Electricity generation patterns are relatively stable in most countries and intraday variations can often be larger than interseasonal variations. An example is shown in Fig. 2.1 for the UK. Average consumption through the year is 37 GW, with an average intraday variation over the year of 18 GW and a peak of 59 GW. Yet the difference in average daily consumption between winter and summer1 is much lower than the intraday variation, at only 11 GW. Larger intraseasonal variations would be expected in countries that use predominantly electric heating, with the peak demand in winter, or in countries with warmer climates, where demand for air conditioning would lead to a peak demand in summer. A mix of baseload and flexible electricity generation technologies is generally used to meet these demands. Baseload typically consists of coal or nuclear plants that produce a constant output and have relatively high capital costs and low fuel costs. Flexible generators tend to be gas and oil-fired power plants with relatively lower capital costs and higher fuel costs. There is a tradeoff between flexibility, efficiency, and cost between technologies; for example, open-cycle gas turbines (OCGTs) are more flexible and have lower capital costs than combined-cycle gas turbines (CCGTs),

Total UK consumpon

Daily demand/GW

250 200 150 100 50 0 Jan














Figure 2.1 Total UK electricity and natural gas consumption in 2010. For electricity, the width of the line demarks the maximum and minimum for each day. Electricity data are half-hourly averages from National Grid, Electricity Transmission Operational Data: Historical Demand Data, 2015. Available from: http://www2.nationalgrid. com/UK/Industry-information/Electricity-transmission-operational-data/Data-explorer/ and daily gas data are from National Grid, Gas Seasonal and Annual Data, 2013. Available from:


Winter here is defined as December to February, and summer as June to August.

Energy storage options to balance renewable electricity systems


but have lower fuel efficiencies. The optimal generation portfolio depends on demand patterns; in the UK, to meet the demand pattern in Fig. 2.1, there was around 45 GW baseload and 40 GW flexible supply in 2010, and most coal generation has closed and wind and solar generation have greatly increased over the subsequent decade. Integrating this increasing penetration of intermittent renewable generation while continuing to provide a secure electricity supply is a key challenge for electricity systems in the future. Low-carbon solar, wind, and wave technologies have high capital costs and negligible operating costs, but their intermittent outputs cannot be easily forecast or controlled. Small amounts of intermittent generation can be balanced through flexible electricity generation. However, as the proportion of renewables increases, two issues arise: 1. At times of high demand and low renewable generation, deficits occur that can affect the stability of the electricity system. 2. At times of low demand and high renewable generation, surplus electricity is generated that must be stored or lost.

These electricity system imbalances between generation and demand present an opportunity for new types of energy storage to have an important role in future energy systems.


Impact of demands on generation imbalances

The deficits and surpluses from renewable generation could be greatly magnified in the future if transport and heat are electrified to reduce greenhouse gas emissions. Fig. 2.1 shows that natural gas consumption has much wider intraseasonal variations than electricity consumption in the UK. This is primarily due to heat demand, as demonstrated by Fig. 2.2 for the UK residential sector. While electrification would not lead to intraseasonal electricity peaks of this magnitude, since heat pumps with a COP2 of 3 would likely replace gas boilers with a fuel conversion efficiency of up to 90%, the resulting intraseasonal variations would still be much more pronounced and overall electricity demand much higher in this scenario [4]. The sizes of the deficits and surpluses also depend to some extent on whether renewable outputs are correlated with demands. Wind generation tends to be higher in winter than summer, so is better correlated to winter peak demands in highlatitude countries, while solar generation is much higher in summer daytime and is most closely correlated to demands in lower-latitude countries, where peak demand occurs in summer daytime for air conditioning. Germany provides a good example of some of the challenges that can emerge. Renewable generation accounted for 23% of German electricity generation in 2012, and a strong expansion of solar photovoltaics in southern Germany in particular has led to some areas already producing more electricity than they consume [5]. The two-way electricity flows in these areas 2

Here COP stands for Coefficient of Performance, which is a measure of the efficiency of the heat pump; a COP of 3 means that each unit of input electricity produces three units of output heat on average.


Storing Energy

UK residenal sector consumpon

Daily demand/GW

250 200 150 100 50 0 Jan














Figure 2.2 Estimated UK electricity and natural gas consumption, in 2010, in the residential sector. For electricity, the width of the line demarks the maximum and minimum for each day. Electricity data are half-hourly from National Grid, Electricity Transmission Operational Data: Historical Demand Data, 2015. Available from: and are linearly interpolated to the average residential monthly consumption. Residential daily gas consumption is estimated from statistics of customers with low usage in National Grid, Gas Seasonal and Annual Data, 2013. Available from:

have resulted in some distribution networks operating at their technical limit, as these have historically been designed to transfer electricity in only one direction from transmission networks to end-users.


Strategies to cope with electricity system imbalances

Four principal strategies have been proposed to manage electricity deficits and surpluses [6]: 1. Dispatchable generation. Flexible generators are used to avoid deficits. The main disadvantages of this existing approach to imbalances are the high capital cost of generation capacity and the inability to benefit from electricity surpluses from renewables at times of low demand. 2. Transmission and distribution network reinforcement. By increasing network capacity, this strategy enables greater movement of electricity in space, so supply and demand are averaged over larger areas which is likely to lead to lower imbalances. The proposed European Supergrid is an example of network enhancement that would use Scandinavian hydropower to balance renewable generation across Europe [7]. 3. Demand-side management. Agreements with large electricity consumers are already used in some countries to reduce demand at peak times. In the future, demand-side response technologies could be used by electricity system operators to shift demand from peak to non-peak periods. For example refrigeration or water heating in homes could be turned off at peak

Energy storage options to balance renewable electricity systems


times. As electric vehicles become more widespread, their batteries could be used to support grid operations when they are charging. 4. Energy storage deployment. Some storage technologies, such as power-to-power and powerto-heat with storage, could manage both electricity deficits and surpluses. Others, such as producing hydrogen for use outside these sectors, would only address surpluses but could complement dispatchable generation.

A schematic of the relationships between these technologies is shown in Fig. 2.3. Most energy storage studies have examined grid-scale storage or, at most, the electricity system in isolation (e.g., Ref. [8]). Yet energy storage could be integrated much more widely across the energy system. One approach would be to reduce electricity system imbalances by integrating storage with renewable generation at the point of generation (e.g., Ref. [9]). Another strategy would be to convert excess electricity into other storage fuels that are not used to return electricity to the grid. For example, excess electricity could produce heat for storage in boilers at district heat or building scales, in a demand-side management version of the night storage heaters that are already widely used in some countries. Hydrogen could be stored for later electricity production [10,11], particularly interseasonal storage, but could also be used for transport or heat provision [12]. Although battery vehicles primarily provide transport services, vehicle-to-grid technologies could use car batteries for power-to-power storage in a smart grid [13]. Finding the most appropriate methods of integrating the many different types of energy storage into existing energy systems is a key research question for energy systems researchers.

Figure 2.3 Schematic of the potential roles of energy storage in a low-carbon energy system. The system is split into grid-scale technologies, the wider electricity system and the whole energy system. Network and storage technologies (denoted with bold text) are integrated throughout the energy system.



Storing Energy

Storage technologies

Numerous energy storage technologies are under development, with a wide range of characteristics that make them suitable for different roles in the energy system [14]. Many of these technologies are shown in Table 2.1, which lists the technologies examined in this book by category. One important characteristic for comparing systems is the round-trip efficiency, which is a measure of the overall loss of electricity from storage in power-to-power systems. Other characteristics and metrics for comparing energy storage systems are discussed in Section 4.



The principal power-to-power energy storage technology in operation around the world at present is pumped hydro, in which excess electricity is used to pump water to a high reservoir where it is stored as gravitational potential energy (see Chapter 3). Pumped hydro can produce a high-power output at short notice, so is normally used to meet intraday peak demands. A round-trip efficiency of 80% can be achieved but the number of suitable sites for building schemes is limited. A novel derivative of pumped hydro is ground-breaking energy storage (GBES), in which a large mass (e.g., a concrete disc) is raised or lowered hydraulically during electricity surpluses and deficits. The aim is to create a smaller, cheaper system than pumped hydro, which is not limited by a small number of available sites. Such novel hydro schemes are discussed in Chapter 4. Gravity energy storage systems use surplus electricity to power a heavy weight to a high elevation (see Chapters 5 and 6). At times of high demand, the weight is returned to the lower elevation and generates electricity on the way. This technology has no energy losses over time once the weight has reached high elevation. It is particularly suited for dry regions where little water is available and high evaporation would cause hydroelectric reservoir losses. Table 2.1 List of energy storage technologies that are examined in this book, by category. Gravitational/mechanical: • Pumped hydro • Gravity energy storage systems • (GBES) • Compressed air energy storage (CAES) • Pumped hydro with compressed air • Flywheels • Liquid air energy storage (LAES) Electrochemical and electrical: • Rechargeable batteries (e.g., lithium-ion, sulfur-lithium; sodium-sulfur) • Flow batteries • Supercapacitors

Thermal: • Sensible thermal energy storage (including molten salts) • Pumped thermal energy • Latent heat storage • Solar ponds Chemical: • Power-to-gas • Large-scale hydrogen storage • Thermochemical energy storage • Traditional energy storage (natural gas, oil and coal)

Energy storage options to balance renewable electricity systems


Mechanical storage converts surplus electrical energy into potential or kinetic energy. The two principal commercial technologies are compressed air energy storage (CAES) and flywheel storage. CAES stores compressed air in constant volume or constant pressure storage vessels (see Chapter 7). Storage media can include salt caverns, aquifers, or purpose-built pressure vessels. Underground storage (see Chapter 8) and undersea energy bags (see Chapter 9) could be integrated with offshore wind generation. A novel hybrid energy storage system has been proposed that combines CAES and pumped hydro (see Chapter 10). The principal challenge for CAES is to conserve heat energy produced during compression in order to maximize the round-trip energy efficiency. Flywheels use surplus electricity to accelerate a rotor to very high speeds and to maintain these speeds, maintaining the energy as rotational energy (see Chapter 12). The rotational speed is reduced as energy is extracted from the system due to electricity generation. The principal challenge is to minimize friction losses in order to maximize round-trip energy efficiency, so flywheels are most suited to short-duration applications. Novel thermomechanical storage technologies are under development with the aim of improving round-trip energy efficiency by combining mechanical and thermal approaches. Liquid air energy storage (LAES), also known as cryogenic energy storage, is an alternative to CAES in which surplus electricity is used to cool air until it liquefies (see Chapter 11). The liquid air is stored in a tank and electricity is generated when required by warming the air until it expands into a gaseous state and turns a turbine. Since capacity and energy are decoupled, LAES is particularly suited to long-duration applications.


Electrochemical and electrical

Rechargeable batteries are widely used in transport and electronic devices. They account for most new electricity storage investments at present, with deployment of both grid-scale and behind-the-meter storage [15]. Increasingly large grid-scale batteries have been deployed in recent years, particularly in California and Australia, and further scale-up is planned (Table 2.2). Lithium-ion batteries have been used at most sites. Lithium-ion battery innovation has focused on mobility applications in recent years, due to their high power density, as costs have been considered too high for stationary applications. However, sustained innovation investments have greatly reduced battery pack costs over the past decade from $1100/kWh in 2010 to $137/kWh in 2020 in real terms [16]. Larger MW-scale systems use banks of batteries arranged in racks, so cost reductions through economies of scale are lower than for gravitational and mechanical systems. This partly explains why IRENA forecast roughly the same capacity of behind-the-meter and grid-scale batteries being deployed in the year 2030 [17]. Lithium-ion batteries are discussed in more detail in Chapter 13. The 300 MW sodium-sulfur investment at Abu Dhabi (Table 2.2) is a notable exception to the trend toward lithium-ion. Sodium-sulfur batteries are examined in Chapter 16. Other battery chemistries such as potassium-ion and sulfur-lithium are


Storing Energy

Table 2.2 Examples of large grid-scale battery investments worldwide that have been deployed or are planned. Power (MW)

Storage (MWh)




Abu Dhabi (10 sites)

Sodiumsulfur Lithium-ion Lithium-ion Lithium-ion Lithium-ion Lithium-ion




Deployed Deployed Approved Approved Proposed

150 230 100 1500 900

193 230 400 6000 1800

Hornsdale, South Australia Gateway, California Alamitos, California Moss Landing, California Goyder South, South Australia

Sources: B. Decourt, R. Debarre, Electricity Storage, SBC Energy Institute, Gravenhage, Netherlands, 2013. Available from:; B. Robertson, Grid Scale Battery Costs Are Declining Faster than Wind and Solar,, 2020. Available from:; S. Hanley, Sodium Sulfur Battery in Abu Dhabi Is World's Largest Storage Device, CleanTechnica, 2019. Available from:

also being developed for grid-scale applications, and the latter is explored in Chapter 15. For all types of batteries, key issues are the tradeoffs between capital costs, material availability, loss of charge when idle, and loss of capacity over time. A flow battery is a type of rechargeable battery in which two chemical components are dissolved in liquids separated by a membrane. Ion exchange occurs across the membrane, meaning that the battery is similar to a fuel cell from a technical perspective. Flow batteries have longer durability than conventional batteries but reducing capital costs is a key challenge. Vanadium redox flow batteries are examined in Chapter 18. Supercapacitors bridge the gap between conventional capacitors and rechargeable batteries. They use electrostatic double-layer capacitance that can hold up to 10,000 times the charge of a conventional solid dielectric capacitor, but they still have a much lower energy density than batteries. Supercapacitors are most appropriate for applications requiring many rapid charge/discharge cycles, for example, regenerative braking in vehicles or improving power quality for electric grids. Supercapacitors are examined in Chapter 19.



Thermal storage involves the storage or removal of heat for later use. Sensible thermal energy storage heats or cools a liquid or solid storage medium, with water most commonly used (see Chapter 20). For example, many buildings have hot water storage that could in the future be used as a sink for surplus renewable electricity. Molten salts are being developed as an alternative thermal storage medium as they have high boiling points, low viscosity, and high volumetric heat capacities (see Chapter 21). For example, a 280 MW/1680 MWh molten salt storage plant is located at the Solana

Energy storage options to balance renewable electricity systems


concentrating solar power plant in Arizona [21]. Large-scale seasonal heat storage has also been proposed, using purpose-built plants or aquifers; on long timescales, the rate of heat loss is an important determinant of the value of the plant. Heat-pumped temperature difference uses a heat pump to store energy in the form of a temperature difference between two heat stores (see Chapter 22). A heat engine then generates electricity when required, with the cooler output medium being stored in a second tank that feeds the heat pump. For example, Isentropic have developed a system with a hot vessel storing thermal energy at high temperature and high pressure, and a cold vessel storing thermal energy at low temperature and low pressure [22]. Both vessels are filled with crushed rock or gravel, which acts as the heat storage medium. The whole system is filled with argon that is pumped between vessels if there are electricity surpluses or deficits, in a heat pump system, with a claimed round-trip efficiency of up to 80%. Latent heat storage uses phase-change materials to store heat through the reversible conversion from solid to liquid phases (see Chapter 23). The principal advantage of latent heat storage over sensible heat storage is that energy is stored at the temperature of the process application. A range of inorganic, organic, and bio-based materials have been developed with different characteristics. Solar ponds are saltwater pools that act as solar thermal energy collectors (see Chapter 24). Pond salinity prevents water from flowing from the bottom to the top of the pond, meaning that temperatures are much higher at the bottom of the pond and the trapped heat can be used for heating buildings or hot water, particularly in industry, or to drive a turbine or Stirling engine to generate electricity.



Hydrogen is a potential storage vector that has received growing attention in recent years. As a zero-carbon energy carrier, it could have a similar role in a low-carbon energy system to electricity, but has the key advantage that it is much easier and cheaper to store (see Chapter 27). Large underground cavern storage of hydrogen is one of the few low-carbon interseasonal energy storage solutions that could support electrification of heat demand [10,12,23]. Power-to-power hydrogen storage systems have been proposed [24] with the hydrogen produced by electrolysis (see Chapter 25). Power-to-gas uses surplus electricity to produce hydrogen that is then injected into the natural gas network (see Chapter 26); while this uses surplus power, its disadvantage is that it does not contribute to meeting electricity generation deficits unless it feeds gas generation plants. The tight tolerance of gas appliances and the different characteristics of hydrogen compared with natural gas mean that it could only supply 1%e6% of the total gas by energy content [25]. This could be increased by methanating the hydrogen using waste CO2 from an industrial process, but at a cost and with an energy efficiency penalty. Otherwise, converting the gas networks to deliver hydrogen would avoid this issue [26]. Hydrogen could also provide flexible generation for peak energy demand, for example, by producing hydrogen from fossil fuels that is stored in salt caverns until required [27], but such a system would not utilize surplus power from renewables.


Storing Energy

New materials are being developed that store heat using reversible endothermic chemical reactions (see Chapter 29). For example, aluminosilicate minerals called zeolites absorb water in an endothermic reaction. The water is desorbed when the zeolite crystals are heated, and long-term losses of this stored heat are negligible as long as water is not present. Moreover, the energy density is higher than for sensible or latent heat storage. With the exception of pumped hydroelectricity, most existing energy storage is in the form of fossil fuels (see Chapter 28). Contemporary energy systems store precursors to electricity rather than building power-to-power storage. Coal can be piled while oil is cheap to store in reservoirs. Gas is generally stored in underground caverns or storage holders. These methods are generally cheaper than the other storage technologies discussed in this book.


Comparing storage systems

The numerous energy storage technologies reviewed above have a wide range of characteristics that affect their suitability for different roles in low-carbon energy systems. The IEA [14] categorizes technologies according to the range of time periods over which their charge/discharge cycles can operate. Barton and Infield [28] similarly identify storage durations for contributions to the electricity system from 20 seconds to 4 months, with different technologies able to operate over different time periods. Koohi-Fayegh and Rosen [21] focus on round-trip efficiency, lifetime cycles, energy and power density, and costs. Luo et al. [29] compare technologies using a wide range of characteristics. The key characteristics chosen by each of these studies are listed in Table 2.3. Table 2.3 Key characteristics used to compare energy storage technologies in studies by the IEA [14], Barton and Infield [28], Luo et al. [29], and Koohi-Fayegh and Rosen [21].

Storage duration Typical size Charge duration Discharge duration Cycles Response time Round-trip efficiency Discharge efficiency Daily self-discharge Energy and power density Specific energy and power Maturity Energy and power capital costs Operating and maintenance costs














Energy storage options to balance renewable electricity systems


This list of characteristics is by no means exhaustive. Numerous other metrics could be used to compare technologies, for example, minimum natural energy and power scales of a single device; optimum natural energy and power scales of a single device; nominal cost per unit energy and per unit power at optimum scale; marginal cost per unit energy and per unit power at optimum scale; lowest power slew rate at which performance degrades noticeably; and effective turnaround efficiency. Further metrics are required for specific technology types (e.g., operating temperature for latent heat storage) and for energy storage technologies that do not provide a power-to-power service. The relative importance of each characteristic depends on the technology application. It is necessary to identify uncertainties in each parameter and to consider the relative importance of these uncertainties on technology performance and value. The studies examined in Table 2.3 provide a valuable comparison of technologies, but there is still a need for a more exhaustive comparison using a wider range of metrics.

5. Challenges for energy storage Energy storage technologies are among the most complicated and least wellunderstood low-carbon technologies. They are arguably under-researched compared to other low-carbon technologies. For example, the Global Energy Assessment has little consideration of individual energy storage technologies, yet notes that, “providing integrated and affordable energy storage systems for modern energy carriers is . perhaps the largest and most perplexing part of the energy systems for a sustainable future that is needed for future economic security” [30]. As suggested by this statement, integrating storage into evolving energy systems is a key challenge that is not well understood, partly because there is currently no energy model that can fully represent the benefits of different types of energy storage across an energy system. There are a number of economic, social, and regulatory barriers to energy storage deployment, but efforts have been made to address these in the last 5 years. Until recently, the capital costs of most energy storage technologies were thought to be too high to justify their deployment, but increasingly large lithium-ion batteries are now being deployed in electricity systems. New energy market structures have been designed in some countries that better realize the value of energy storage to the energy system. Public acceptance of storage technologies has started to be investigated. Yet regulatory barriers and economic and social challenges still pose formidable barriers to many technologies in some countries, and there is still a need to find the most appropriate roles for different storage technologies. Roles are likely to vary between countries, as they depend on both the value that each technology offers and the barriers to the technology. These barriers are explored in this section.


Storing Energy


Integrating energy storage into low-carbon energy systems

The value of novel energy storage technologies for low-carbon energy systems is uncertain for several reasons. First, the optimal amount of storage depends on the amount of flexible generation in the overall electricity generation portfolio and the magnitude of the demand peaks. While it is straightforward to show that storage has positive economic benefits for a very inflexible system (e.g., Ref. [8]), it is unlikely that such a portfolio of generation technologies would be intentionally constructed. Second, there are tradeoffs between energy storage and alternatives such as network reinforcement, interconnection between national electricity systems, and demandside management, as described in Section 2.2. Third, the number and diversity of novel energy storage technologies make it challenging to identify their most appropriate roles in supporting different low-carbon electricity systems. Fourth, the role of storage could change if it were not viewed as being a separate system to generation technologies.


Generation-integrated energy storage

Most studies examining the role of grid-scale energy storage consider only power-topower storage, in which electricity is converted to some storable form and then back to electricity again. However, there are two other broad categories. GenerationIntegrated Energy Storage (GIES) systems store energy before electricity is generated. Load-Integrated Energy Storage (LIES) systems store energy (or some energy-based service) after electricity has been consumed (e.g., power-to-gas, with hydrogen stored prior to consumption for transport or another end-use). GIES systems have received little attention to date, but could have an important role in the future [31]. As mentioned in Section 1, most countries store precursors to electricity in the form of raw or processed hydrocarbons, whether in coal and biomass heaps or oil and gas reservoirs. These are examples of GIES technologies, and offer two general advantages over other storage systems: • •

There may be low (or even zero) marginal costs associated with the extra equipment or infrastructure required to enable energy to be put into storage. There may be low (or even zero) marginal losses of energy associated with passing energy through storage.

Many instances of GIES systems are already in existence. All natural hydropower plants with a dam fall into the GIES category. Such plants accumulate energy in the form of gravitational potential energy of water; their energy stores are filled up as a result of rain, but they can generate electricity when no rain is falling. It is conceivable that some new nuclear power stations could be equipped with thermal energy storage (as indicated in Ref. [32]) so that the reactors would run at a constant power rate, but electricity would be generated to match demand, and these nuclear power stations would be GIES systems. Fourth-generation high-temperature nuclear power plants could dissociate water directly into hydrogen and oxygen [33], and would also share this classification.

Energy storage options to balance renewable electricity systems


Selected concentrated solar power plants such as the Andasol and Gemasol plants [34] are equipped with thermal energy storage, enabling heat to be stored prior to its use in raising steam to drive electrical generators, and these are also GIES systems. Another arrangement for converting solar power into an immediately storable energy form involves photolysis, in which water is split using photons [35,36]. There are several mechanical renewable energy devices for harvesting wind, wave, and tidal power by directly compressing air. Some of these are discussed in Chapter 7, and all fall within the GIES aegis. Finally, there are several possible configurations of equipment that exploit the interaction between mechanical work and heat to integrate energy storage with devices that collected renewable energy directly [37,38]. Fig. 2.4 outlines a potential GIES system for wind generation.


Analysing energy storage integration using models

One method to better understand the relative benefits of different technologies is to compare them using a wide range of operational, economic, and environmental metrics, as described in Section 4, and this is undoubtedly important. Another is to

Figure 2.4 Schematic of a GIES system suitable for wind generation (copyright University of Nottingham, visualising a system described in [37]).


Storing Energy

compare a small number of technologies for a particular energy system (e.g., Ref. [39]). But the most common method to understand how these technologies might be integrated into a low-carbon energy system is to explore scenarios using models. These can give insights over the transition to a low-carbon system about which technologies are most economically deployed, how the level of deployment might change over time, and where the technologies are best deployed. Several model paradigms can give useful insights: •

Energy system models: market-based economic optimization models that represent all energy flows and GHG emissions in an economy and are used to examine how energy systems might evolve, at least cost, to meet long-term emission targets. Demands are specified for energy services and the contribution of individual fuels to meeting these is optimized, meaning that the demand for electricity evolves over time depending on the relative competitiveness of electricity generation against alternative energy vectors. This practically means that these models can compare different methods of balancing supply and demand (network reinforcement, energy storage, etc.) or can construct the energy system to minimize imbalances if all of these options are expensive. For example, one storage option would be to use excess electrical generation to produce hydrogen for transport or heat for buildings, rather than deploying power-to-power storage. Energy system models can be used to compare all types of energy storage, on different timescales, but they tend to have low spatial and temporal resolution, meaning the need for and the value of energy storage is generally underestimated. Electricity dispatch models: market-based electricity system models that calculate the merit order of electricity generation in a mixed portfolio to meet evolving demands. Electricity demands are fixed in each time period, meaning that in contrast to energy system models, dispatch models cannot consider options to change the electricity demand by using alternative energy vectors. Energy storage can be simulated by more advanced dispatch models, but principally grid-scale power-to-power storage. Demand-side response (DSR) has also been simulated, and simplistic network reinforcement can be investigated using multiregion dispatch models. The principal advantages of dispatch models over energy system models are first, that the temporal resolution is much higher, typically 30 min compared with 6 h, so supply/demand imbalances are better represented, and second, that network reliability requirements such as loss of load can be set by the modeler. Electricity network models: detailed spatial models of electricity transmission or distribution networks, to examine the performance of networks in meeting peak loads. These can provide detailed insights into the relative benefits of grid-scale energy storage and network reinforcement, and can be used to identify the most appropriate locations to deploy different energy storage technologies. Chapter 31, examines the integration of energy storage with transmission and distribution networks in smart grids.

Most energy storage integration studies have used only electricity dispatch models, considering grid-scale or very occasionally the wider electricity system shown in Fig. 2.3. Since these models do not consider the whole energy system, those studies necessarily exclude some storage options and make broad assumptions about electricity demand. Yet while energy system models do not have these drawbacks, their low temporal resolution causes them to underestimate the value of energy storage. Moreover, different energy storage technologies work at different scales and it is difficult to construct a model that can consider the relative benefits of in-house, distribution, and

Energy storage options to balance renewable electricity systems


grid-scale storage. No single model is capable of holistically assessing the value of all energy storage technologies. There is a need for studies that combine these three types of models to understand the whole system shown in Fig. 2.3, which could then produce more credible scenarios for energy storage within the context of energy system transitions.


Innovation to reduce technology costs

The Global Energy Assessment identifies affordability as a key challenge for energy storage systems [30]. Although most energy storage research is targeted at the technologies discussed in Section 3, they all currently have high costs relative to fossil energy storage and only pumped hydro and lithium-ion batteries are commercially competitive. Innovation is required to reduce capital costs and improve the performance of key technologies. Since the rate of technology cost reduction is generally related to the level of technology deployment [40], the lack of markets for storage has been an impediment to innovation in the past. Lithium-ion batteries have only overcome this trend as a result of investments in automotive rather than electricity supply applications. Even as demand increases, as energy storage technologies are so numerous and so diverse, deployment rates could continue to be low for many technologies, particularly for larger devices, which could inhibit cost reduction. The competitiveness of lithiumion batteries is a further barrier to investment in other technologies, even if those alternatives might have better prospects in the longer term. One promising niche area for early deployment of energy storage is in offgrid applications, where storage has a higher value for smaller, more constrained energy systems (Chapter 31). Various types of energy storage are now being developed and deployed globally (see Chapter 32). A particular source of innovation is likely to be China, which has a rapidly evolving electricity system and strong investment in infrastructure and industrial development (see Chapter 33).


Adapting energy markets to realize the value of energy storage

Even if the value of energy storage could be demonstrated, the technologies would not currently be viable in most countries because electricity markets are designed for incumbent systems in which supply is varied to meet demand. It would be necessary to adapt market regulation to reflect the technological, economic, and social value of energy storage to an energy system. Some countries have taken first steps in this direction, for example, through the development of ancillary services markets and the electricity capacity market in the UK. See Chapter 34, for a detailed discussion. As with many emerging technologies, there is a need for incentives to encourage energy innovation in the short-term and also investments that reflect the long-term value of energy storage, through the design of both electricity markets and business


Storing Energy

models. One approach would be to provide incentives from capacity markets and feed-in tariffs in the short term to encourage the deployment of new technologies, and subsequent innovation. Some experimentation would likely be required to identify optimal market structures. End-users could potentially reduce bills or offer balancing services to the electricity grid through storage provision or demand flexibility [41]. This means that retail as well as wholesale markets would need to be adapted to enable the integration of energy storage at different scales, in order to reward end-users for the value of the balancing services that they provide to the electricity system. An alternative method being pursued is “virtual batteries,” in which companies control and aggregate large numbers of small batteries in order to sell grid-scale services [42]. Other barriers, such as a need for the deployment of suitable smart meters, might also need to be overcome.


Public acceptance

Public engagement with energy supply and demand technologies has been identified as a critical issue for the future deployment of innovative and low-carbon energy systems [43]. Public acceptance is particularly important for in-house storage technologies (e.g., heat storage). There are difficult conceptual and methodological challenges to such research, including the need to integrate social science methods with appropriate technical descriptions of the technologies, in order to engage with people about technologies for which they may be unfamiliar. One study using deliberative workshops in the UK concluded that perceptions of storage technologies tend to be ambivalent, and that acceptance is likely to depend on whether storage technologies can be designed, regulated, and governed in ways which reduce concerns over safety, environmental impacts, and reliability, while meeting societal desires for equity and the protection of vulnerable groups [44]. Analyses of energy storage integration into the energy system would ideally account for public attitudes, but it is challenging to translate a range of qualitative outcomes into meaningful quantitative model inputs. Moreover, since network reinforcement and demand-side management are potential alternatives to energy storage, it would be useful to compare the societal views of these three approaches to balancing flows in the energy system, in both qualitative and quantitative studies.


Regulatory barriers to energy storage

Regulatory barriers are likely to vary by country, but there are common trends that can be identified. Castagneto Gissey et al. [45] identify and categorize 16 investment barriers hindering the near-term deployment of energy storage technologies in electricity markets, which are related to four regulatory and public attitudes barriers (Fig. 2.5). They identify the most important regulatory barrier as being the classification of storage as a generation asset in many countries, despite it being unable to provide a positive net flow of electricity, which is used to justify double network usage charges (for charging and discharging). Moreover, the merit order design of balancing and

Energy storage options to balance renewable electricity systems


Figure 2.5 Relationship between barriers to energy storage innovation and deployment. Exogenous barriers are colored red. From: G. Castagneto Gissey, P.E. Dodds, J. Radcliffe, Renew. Sustain. Energy Rev. 82 (2018) 781e790.

ancillary markets hampers the ability of storage technologies to recoup their relatively high capital cost, while even relatively novel capacity markets can penalize their limited discharge duration by requiring longer discharge duration than is economic to provide. Also, the electricity system operator is in the best position to realize the system value of storage, but in most countries the system operator owns the transmission network and so has a conflict of interest in the choice of balancing technology.


Finding the most appropriate roles for energy storage technologies

The most appropriate roles for each energy storage technology depend foremost on the design of the electricity system, particularly the fraction of inflexible renewable and nuclear generation and the fraction of flexible peak generation. Other important factors include how the technologies are integrated with the electricity system from an engineering perspective, the value of each technology relative to the cost of deploying it, accounting for potential cost reductions in the future, and social acceptance of preferred technologies. Identifying the most appropriate roles is therefore challenging. The integrated modeling approach in Section 5.1.2 could provide insights, if cost innovation and


Storing Energy

social barriers were incorporated into “integrated” scenarios. Yet other than pumpedhydro, most technologies are currently at the demonstration stage and there are no broad guidelines available about the suitability of different technologies for particular situations. Some technologies have been tested at a large scale (100e1000 MW), but the future performance of most technologies is not well understood. Novel approaches to integration are still under development, for example, the GIES systems discussed in Section 5.1.1. GIES systems tend to perform especially well when a high proportion of all energy passing through these systems also passes through their internal storage [31], so their utility again depends on the wider electricity system configuration. There is a need for R&D programs to further develop such novel approaches and to test a range of technologies at scale so their operating characteristics and costs can be better understood. A key challenge for finding the most appropriate roles for energy storage is how to cope with future uncertainty, both in terms of the electricity system and the technologies themselves [46]. Roadmaps for energy storage are occasionally published (e.g., Ref. [14]), based on a long-term vision, but these do not generally consider uncertainties in the costs and the value of energy storage technologies, nor whether energy storage investments can be justified now in order to keep the option open for their use in the future. A real options approach could be used to assess the option value of energy storage in an uncertain future.



Energy storage makes an important contribution to energy security. Most contemporary storage systems are based around fossil fuels, but novel energy storage technologies could make an important contribution to future low-carbon energy systems. They could be particularly important if electrical heating and transport become widespread, or if intermittent renewables and other inflexible low-carbon technologies come to dominate electricity generation. Energy storage would be in competition with existing system control strategies, notably dispatchable generation and network reinforcement, as well as newer strategies such as demand-side response. Numerous energy storage technologies have been proposed to store excess electricity, with electrical energy conversion to mechanical, thermal, gravitational, electrochemical, and chemical energy for storage. Pumped-hydro is the only widely used electricity storage technology at the moment, but increasingly large lithium-ion batteries are being built. As storage technologies have a wide range of characteristics that affect their suitability for different roles in low-carbon energy systems, a series of energy storage metrics has been proposed by different studies for comparison purposes, but these are by no means comprehensive, and there is still a need for a more exhaustive comparison using a wider range of metrics. Energy storage technologies are complicated and poorly understood relative to most low-carbon technologies. Understanding how to integrate energy storage into low-carbon energy systems is a difficult challenge for several reasons. First, the

Energy storage options to balance renewable electricity systems


proportion of inflexible generation in the electricity system affects the value of energy storage to the system; if the cost of energy storage were too high, then the proportion of inflexible renewable generation might be reduced. Second, existing energy system, dispatch and network models are either not broad enough to examine all energy storage and alternative options, or have insufficient temporal resolution to realistically portray the need for and performance of storage technologies. This means that the optimum approaches to integrating novel energy storage technologies into low-carbon energy systems cannot be fully understood using existing models. Third, the costs of most energy storage technologies are currently too high. Innovation is required to reduce costs but the rate of cost reduction generally depends on the level of technology deployment. Fourth, there are substantial regulatory barriers to energy storage that need to be overcome. Finally, even if the long-term value of energy storage could be demonstrated, existing electricity markets are designed for incumbent systems and market structures would need to be adapted to reflect the technological, economic, and social value of energy storage to an energy system. A whole energy system approach to energy storage is necessary to fully understand how different technologies might contribute to future low-carbon energy systems as innovation reduces storage costs. Studies have shown that value of energy storage in some electricity system configurations is substantial. It is likely that the economic value of the difference between good and bad policy decisions relating to the role of energy storage in the transition to low-carbon generation is in the order of billions of dollars. Further R&D and a better understanding of the integration of these technologies is vital to provide information to underpin future market design and regulation, so the value of energy storage can be realized.

References [1] IRENA, Renewable Power Generation Costs in 2019, International Renewable Energy Agency, Abu Dhabi, 2020. Available from: Renewable-Power-Costs-in-2019857. [2] National Grid, Electricity Transmission Operational Data: Historical Demand Data, 2015. Available from: [3] National Grid, Gas Seasonal and Annual Data, 2013. Available from: http://www. [4] I.A.G. Wilson, A.J.R. Rennie, Y. Ding, P.C. Eames, P.J. Hall, N.J. Kelly, Energy Pol. 61 (2013) 301e305. [5] P.E. Dodds, B. Fais, Network infrastructure and energy storage for low-carbon energy systems, in: P. Ekins, M. Bradshaw, J. Watson (Eds.), Global Energy: Issues, Potentials and Policy Implications, Oxford University Press, Oxford, UK, 2015, pp. 426e451. [6] Coalition study, Commercialisation of Energy Storage in Europe, McKinsey & Company, 2015. Available from: EnergyStorageFinal_3.pdf625.


Storing Energy

[7] Energy and Climate Change Committee, A European Supergrid, UK Parliament, London, UK, 2011. Available from: cmselect/cmenergy/1040/104002.htm442. [8] G. Strbac, M. Aunedi, D. Pudjianto, P. Djapic, F. Teng, A. Sturt, D. Jackravut, R. Sansom, V. Yufit, N. Brandon, Strategic Assessment of the Role and Value of Energy Storage Systems in the UK Low Carbon Energy Future, Imperial College London, London, UK, 2012. Available from: [9] A.J. Pimm, S.D. Garvey, M. de Jong, Energy 66 (2014) 496e508. [10] T. Lohner, A. D’Aveni, Z. Dehouche, P. Johnson, Int. J. Hydrogen Energy 38 (2013) 14638e14653. [11] D. Anderson, M. Leach, Energy Pol. 32 (2004) 1603e1614. [12] P.E. Dodds, A Whole Systems Analysis of the Benefits of Hydrogen Storage to the UK, World Hydrogen Technologies Convention 2013, Shanghai, China, 2013. [13] N.S. Wade, P.C. Taylor, P.D. Lang, P.R. Jones, Energy Pol. 38 (2010) 7180e7188. [14] IEA, Energy Storage Technology Roadmap, International Energy Agency, Paris, France, 2014, p. 626. [15] IRENA, Innovation Landscape for a Renewable-Powered Future: Solutions to Integrate Variable Renewables, International Renewable Energy Agency, Abu Dhabi, 2019, p. 810. [16] BloombergNEF, Battery Pack Prices Cited Below $100/kWh for the First Time in 2020, While Market Average Sits at $137/kWh, 2020. Available from: blog/battery-pack-prices-cited-below-100-kwh-for-the-first-time-in-2020-while-marketaverage-sits-at-137-kwh/. [17] IRENA, Utility-Scale Batteries: Innovation Landscape Brief, International Renewable Energy Agency, Abu Dhabi, 2019. Available from: 2019/Sep/Utility-scale-batteries860. [18] B. Decourt, R. Debarre, Electricity Storage, SBC Energy Institute, Gravenhage, Netherlands, 2013. Available from: ElectricityStorage.aspx631. [19] B. Robertson, Grid Scale Battery Costs Are Declining Faster than Wind and Solar,, 2020. Available from: [20] S. Hanley, Sodium Sulfur Battery in Abu Dhabi Is World’s Largest Storage Device, CleanTechnica, 2019. Available from: [21] S. Koohi-Fayegh, M.A. Rosen, J. Energy Storage 27 (2020) 101047. [22] Isentropic. How Isentropic PHES Works, 2015. Fareham, UK. Available from: http:// [23] J.-P. Maton, L. Zhao, J. Brouwer, Int. J. Hydrogen Energy 38 (2013) 7867e7880. [24] S. Carr, G.C. Premier, A.J. Guwy, R.M. Dinsdale, J. Maddy, Int. J. Hydrogen Energy 39 (2014) 10195e10207. [25] P.E. Dodds, W. McDowall, Energy Pol. 60 (2013) 305e316. [26] P.E. Dodds, S. Demoullin, Int. J. Hydrogen Energy 38 (2013) 7189e7200. [27] ETI, Hydrogen: The Role of Hydrogen Storage in a Clean Responsive Power System, Energy Technologies Institute, Loughborough, UK, 2015. Available from: http://www.eti. [28] J.P. Barton, D.G. Infield, IEEE Trans. Energy Convers. 19 (2004) 441e448. [29] X. Luo, J. Wang, M. Dooner, J. Clarke, Appl. Energy 137 (2015) 511e536.

Energy storage options to balance renewable electricity systems


[30] T.B. Johansson, N. Nakicenovic, A. Patwardhan, L. Gomez-Echeverri, Summary for Policymakers. Global Energy Assessment: Toward a Sustainable Future, International Institute for Applied Systems Analysis, Laxenburg, Austria, 2012. [31] S.D. Garvey, P.C. Eames, J.H. Wang, A.J. Pimm, M. Waterson, R.S. MacKay, M. Giulietti, L.C. Flatley, M. Thomson, J. Barton, D.J. Evans, J. Busby, J.E. Garvey, Energy Pol. 86 (2015) 544e551. [32] P. Denholm, J.C. King, C.F. Kutcher, P.P.H. Wilson, Energy Pol. 44 (2012) 301e311. [33] K. Verfondern, Nuclear Energy for Hydrogen Production, Forschungszentrum J€ ulich, Germany, 2007. Available from: [34] R.I. Dunn, P.J. Hearps, M.N. Wright, Proc. IEEE 100 (2012) 504e515. [35] M.G. Kibria, F.A. Chowdhury, S. Zhao, B. AlOtaibi, M.L. Trudeau, H. Guo, Z. Mi, Nat. Commun. 6 (2015). [36] A.A. Ismail, D.W. Bahnemann, Sol. Energy Mater. Sol. Cell. 128 (2014) 85e101. [37] S. Garvey, A. Pimm, J. Buck, S. Woolhead, K. Liew, B. Kantharaj, J. Garvey, B. Brewster, Wind Eng. 39 (2015) 149e174. [38] J.E. Lee, On-Demand Generation of Electricity From Stored Wind Energy, Patent US2012326445-A1, 2012. [39] S. Karellas, N. Tzouganatos, Renew. Sustain. Energy Rev. 29 (2014) 865e882. [40] A. Gr€ubler, N. Nakicenovic, D.G. Victor, Energy Pol. 27 (1999) 247e280. [41] G. Castagneto Gissey, B. Zakeri, P.E. Dodds, D. Subkhankulova, Energy Pol. 149 (2021) 112008. [42] G. Castagneto Gissey, D. Subkhankulova, P.E. Dodds, M. Barrett, Energy Pol. 128 (2019) 685e696. [43] L. Whitmarsh, P. Upham, W. Poortinga, C. McLachlan, A. Darnton, P. Devine-Wright, C. Demski, F. Sherry-Brennan, Public Attitudes, Understanding, and Engagement in Relation to Low-Carbon Energy: A Selective Review of Academic and Non-academic Literatures, 2011. Available from: Synthesis%20FINAL%20(24%20Jan).pdf639. [44] G. Thomas, C. Demski, N. Pidgeon, Energy Pol. 133 (2019) 110908. [45] G. Castagneto Gissey, P.E. Dodds, J. Radcliffe, Renew. Sustain. Energy Rev. 82 (2018) 781e790. [46] P.G. Taylor, R. Bolton, D. Stone, P. Upham, Energy Pol. 63 (2013) 230e243.

This page intentionally left blank

Section B Gravitational/thermomechanical storage techniques

This page intentionally left blank

Pumped hydro storage (PHS) 1,4

1, 2



Julian David Hunt , Behnam Zakeri , Andreas Nascimento and Roberto Brand~ ao 3 1 International Institute for Applied Systems Analysis, Vienna, Austria; 2Sustainable Energy Planning Research Group, Aalborg University, Copenhagen, Denmark; 3GESEL - Electric Sector Study Group, Rio de Janeiro, Brazil; 4Federal University of Espirito Santo, Espirito Santo, Brazil

1. Introduction Pumped hydro storage plants are energy storage solutions that consist of two water reservoirs, a tunnel connecting the lower and an upper reservoir and a powerhouse with a pump/turbine. When storing energy, the powerhouse consumes electricity and pumps water from the lower reservoir to the upper reservoir. During generation, the water from the upper reservoir flows through the turbine generating electricity. With the intention of lowering the costs of the system, the same turbines and generators are used in pump and generation models, as shown in Fig. 3.1. The system components involved with the generation capacity (GW) costs are the turbine, generator, tunnels, and powerhouse excavation, and the components involved with energy storage capacity (GWh) costs are the dam and land costs. Sites for PHS plants that focus on power services, such as daily and weekly pumped storage plants, for peak generation, and for storing electricity generated from variable renewable sources, have short horizontal and high vertical distances between the upper and lower reservoirs, as shown in Fig. 3.2. These plants are compared with the ratio between the head in m (H) and tunnel length in km (L), i.e., H/L. The higher the H/L, the more convenient is the PHS plant for power services. This indicator tends to start at 40 and it can reach up to 400. As shown in Fig. 3.2, the design of the tunnel system should minimize the costs of construction. The vertical section of the pipeline, penstock, must withstand pressures of up to 120 bar, depending on the height difference. Thus, the penstock is in most cases built vertically to reduce the length of the penstock. If the slope of the mountain is very high, the penstock and powerhouse can be constructed close to the slope, which reduces the costs of access tunnels to the powerhouse. There are several important equations to model PHS systems. These are the generation and storage equations (Eq. 3.1). The equation to estimate the energy storage of the system (Eq. 3.2). The water mass balance equation in the upper reservoir (Eq. 3.3). The optimization of the diameter and number of tunnels, according to the capital and operational costs [2,3].

Storing Energy. Copyright © 2022 Elsevier Inc. All rights reserved.


Figure 3.1 PHS representation in (A) storage mode and (B) generation mode.

Figure 3.2 PHS representation with main components [1].

Storing Energy

Pumped hydro storage (PHS)

Generation ¼ H  F  g  e



where generation is in Watts, H is the generation height in meters, F is the flow of water in the turbine in kg/s, g is the acceleration of gravity in m/s2, e is the efficiency of the system in generation mode. The same equation applies to the storage mode. Energy storage ¼ Hav  S  g  e


where, energy storage is in Joules, Hav is the average head difference between the upper and the lower reservoirs. S is the amount of water stored in the upper reservoir, g is the acceleration of gravity, and e is the efficiency of the system in generation mode. Water storage ¼ I  ðF þ EÞ


where water storage is in kg, I is the inflow of water to the reservoir in kg/s, F is the water turbined or pumped in kg/s, V is the spilled water (this usually only happens in pump-back PHS plants), evaporated water in kg/s. The average efficiency of PHS is presented in Fig. 3.3. These values vary with the length of the tunnels, the speed of the water in the tunnels, the quality of the turbines and generators. Other aspects that are not included in the system are the evaporation and percolation losses and natural inflow into the upper reservoir, which vary substantially with the climate of where the PHS is installed and the catchment area of the reservoir. Currently there exists around 158 GW installed capacity PHS plants [4]. The installed capacity in Europe, USA, and Japan increased significantly during the 1970s and 1980s due to the increase in nuclear and coal-power plants (Fig. 3.4). In the 1990s, after the fall of the Soviet Union, the rapid rise in installed capacity of PHS plants stopped mainly due to cheap and high availability of natural gas in the United States and European markets.

Figure 3.3 Efficiency of each component of PHS plants.


Storing Energy

Figure 3.4 Historical installed capacity of PHS plants in Europe, Japan, China, USA, and India [5].

The interactive map developed by the International Hydropower Association presents the existing, under-construction, and planned PHS plants around the world (Fig. 3.5). China is the country that has the most PHS plants under construction and planning phases. It is expected that 45.5 GW of PHS will be installed from 2018 until 2026 [7]. Currently, there is renewed interest from Western countries in developing pumped hydro storage projects, under the leadership of the US Department of Energy and the International Hydropower Association [8]. More details on existing PHS projects around the world can be seen in Reference [9]. Batteries and hydrogen are rapidly gaining the market for energy storage. Pumped hydro storage will have to reinvent itself to remain competitive. Bloomberg predicts that the use of batteries for grid storage in 2030 will be 280 GW, which will surpass the global capacity of PHS plants [10]. With a battery cost expected to fall under

Figure 3.5 World map with all operational, under-construction, and planned pumped hydro storage plants [6].

Pumped hydro storage (PHS)


100 US$/kWh in 2024 and around 60 US$/kWh in 2030 [11], batteries will soon be cheaper than PHS plants with daily storage cycles. Regarding long-term energy storage (GWh), hydrogen will be an important competitor for PHS, as the global network of production and distribution of hydrogen develops. For instance, a full liquid hydrogen tanker with a volume of 267,000 m3 (the size of a large LNG tanker) stores 415 GWh or 17 GWd of electricity, assuming that the efficiency to transform the hydrogen into electricity is 70%. This is equivalent to the energy stored on a monthly or seasonal PHS plant. This would significantly reduce the viability of seasonal PHS plants in countries that will rely on imported hydrogen in the future. Given that the overall efficiency for energy storage with batteries (90%) is higher than PHS plants (70%e85%), the efficiency of hydrogen (30%e60%) is lower than PHS plants. It is likely that batteries will become more competitive than PHS for short-term energy storage; however, PHS will be more competitive than hydrogen for long-term energy storage. Thus, in the future, monthly and seasonal PHS plants will become more common than daily and weekly PHS plants. The main benefits and challenges of PHS are presented in Table 3.1. Due to the competition of batteries and hydrogen storage alternatives, PHS plants in the future might only make sense if the benefits outweigh the challenges presented in Table 3.1. This chapter is organized in 7 sections. Section 2 presents the different types of PHS plants with respect to storage cycles. Section 3 presents the conventional arrangements of PHS. Section 4 presents the multiple different services provided by PHS. Section 5 presents new possible arrangements for PHS. Section 6 presents different pumpturbine types and their applications. Section 7 presents the world potential of PHS.

Table 3.1 Benefits and challenges of PHS. Benefits


Low maintenance costs. Store energy in hourly, daily, weekly, monthly, seasonal, and pluriannual cycles. Life cycles of 50 years or more.

High capital costs. Limited to mountainous regions or highlands with appropriate topography. Gains in scale limit the application of PHS to plants with large capacity factors, greater than 50 MW. PHS store energy far from the demand and require transmission and distribution costs. The water used to store energy might be better used for water supply during droughts. Need for upper and lower reservoirs, which can be larger than 50 km2. Energy storage efficiencies of 70%e80%.

Good stationary solution to centralized, large-scale energy storage. Mature technology with low risks of failure.

Store water for long periods with low land requirement and evaporation. Provide water management services in locations where conventional dams are not an option.



Storing Energy

Storage cycles duration

Depending on the water storage capacity of the upper reservoir, the height difference between the upper and lower reservoirs, and the availability of water in the lower reservoir, a PHS plant can have hourly, daily, weekly, monthly, seasonally, or pluriannual storage capacities. The larger the size of the upper reservoir, the more storage cycles it can perform. A daily PHS plant cannot store energy weekly, monthly, or seasonally. However, a seasonal PHS can store energy daily, weekly, and monthly. Table 3.2 presents the different PHS cycle types for meeting different energy and water storage needs. Hourly, daily, and weekly PHS plants tend to be built focusing on having high head, short distance between the lower and upper reservoirs, and low generation costs (i.e., low USD/GW). In other words, the cost of the tunnels, pump/turbine, motor/generator, excavation is usually small. The potential for this type of PHS depends a lot on the topography of the locations. Even though mountainous regions and highlands are usually environmentally protected areas, the potential for daily and weekly PHS is vast and surpasses many times the regional demand for daily and weekly storage. Table 3.2 Different PHS cycle types for meeting energy needs [12]. PHS type

Operation mode

Occasions when the PHS type operates

Pluriannual pumped storage (PAPHS)


Annual surplus in hydroelectric generation [13]. Annual fuel prices cheaper than average. Lower than average annual electricity demand [14]. Annual deficit in hydroelectric generation [13]. Annual fuel prices more expensive than average. Higher than average annual electricity demand [14]. Rainy seasons or ice melting seasons, with high hydropower generation [15]. Summer, with high solar power generation [16]. Windy seasons, with high wind power generation [17,18]. Low demand season, when electricity demand reduces. Dry period or freezing winters, with low hydropower generation [15]. Winter, with low solar power generation [16]. Not windy seasons, with low wind power generation [17,18]. High demand season, when electricity demand increases.


Seasonal pumped storage (SPHS)



Pumped hydro storage (PHS)


Table 3.2 Different PHS cycle types for meeting energy needs [12].dcont’d PHS type

Operation mode

Occasions when the PHS type operates

Weekly pumped storage (WPHS)


During the weekends, when power demand reduces [19]. Windy days, with high wind power generation [18]. Sunny days, with high solar power generation [20]. During weekdays, when power demand increases [19]. Not windy days, with low wind power generation [18]. Cloudy days, with low solar power generation [20]. Night, when electricity demand reduces [21]. Day, when there is solar power generation [22]. Day, when electricity demand increases [21]. Night, when there is no solar power generation [22]. Ancillary services: frequency control, remove harmonics in the grid, provide backup power in case of disturbances in supply.


Daily pumped storage (DPHS)



Hourly pumped storage (HPHS)

Pump & generation

On the other hand, monthly, seasonal, pluriannual PHS plants could have high USD/GW costs; however, they have low energy storage costs (USD/GWh). This is because to find sites close to a river with large water resources, with reasonable height difference between the lower and upper reservoir that can store large amounts of water is more difficult. Thus, the length of the tunnels ends up being larger and the head smaller, when compared to daily PHS. This also makes the number of good sites for seasonal PHS small. If the costs for USD/GW are too high, the plant can be built with a small capacity and operate only with seasonal cycles to increase the capacity factor of the plant. PHS plants operating only seasonally can have capacity factors of up to 40%, assuming that the plant also operates 40% of the time pumping. Pump-back arrangement is convenient because it not only stores energy, but also generates hydroelectricity. It would be ideal to have a project with low GW and GWh costs. Since it is not easy to find a PHS plant with low GW and GWh costs, there is the possibility of building two PHS plants, one seasonal PHS with low storage costs of 10 USD/MWh and high power costs 2 USD/GW and one weekly PHS with high storage costs 100 USD/MWh and low power costs 0.8 USD/GW. The seasonal PHS has a 100 MW of installed capacity and 1 TWh of storage, and would operate only with a seasonal cycle. The weekly PHS would have a capacity with 2 GW and 50 GWh of storage and would focus on daily and weekly storage cycles. The final cost of the two PHS combined for


Storing Energy

power generation would be (2  0.8 þ 0.1  2)/2.1 ¼ 0.86 USD/GW and for storage it would be (10  1000 þ 100  50)/1050 ¼ 14.2 USD/MWh. These final costs would be much cheaper than building one seasonal PHS that provides both power and storage services. SPHS consists of two reservoirs, a lower and an upper reservoir connected by a power conversion system (pump/turbine) and a tunnel (Fig. 3.6). The lower reservoir is meant for storing water and it may or may not have a large storage capacity. Typically, a month-long storage capacity in the lower reservoir is enough to store water in days with intense rainfall, allowing the water in the main river to be pumped to the upper reservoir. The upper reservoir should have a large storage capacity to take up a large part of the water from the main river during the wet period. During the dry period, the water stored is released generating electricity and regulating the flow of the river. The regulation of the river flow reduces the spillage on the hydropower plants downstream the river, optimizing its hydropower generation. The water stored can also be used to alleviate droughts. Thus, most of the water will be stored in the upper reservoir and the lower reservoir would control flow fluctuation in the main river so that water will be available to be pumped to the upper reservoir. The upper reservoir of an SPHS plant allows for a large level variation, of up to 150 m, reducing the land requirement for water and energy storage [12]. This lowflooded area and high-level variation result in a low evaporation per stored water ratio. This makes SPHS suitable for regions where water availability and energy storage potential are complementary and regions where evaporation has a large impact on water management. Locations where a 200 m high conventional dam with 150 m level variation can be constructed are not common because the major rivers are typically populated areas, with valuable infrastructure and important economic activities. SPHS increases the possibility of building large reservoirs, as there are more potential sites in small tributaries compared to conventional dams in large rivers. The water intake in an SPHS reservoir has two different origins. Firstly, water flows from the tributary river directly to the SPHS reservoir. This can be due to precipitation and/ or ice melting. The other portion of the water in the SPHS reservoir comes from pumping water from the lower reservoir.

Figure 3.6 Diagram of a seasonal pumped hydro storage plant [12].

Pumped hydro storage (PHS)


3. Conventional arrangement types The most well-known PHS arrangements are open-loop, closed-loop and pump-back storage. Open-loop consists of a PHS plant where there is a significant stream of water to the upper or the lower reservoir (Fig. 3.7A). In this setup the operation of the pumpturbine may interfere with the river flow and this should be carefully cared for. In order to minimize the impact on the river flow, open-loop PHS schemes usually make use of existing hydropower dams as the lower reservoir. In cases where the lower reservoir is an existing dam, the powerhouse can be built downstream the dam. This way, the powerhouse will not require to be excavated as the head of the dam already increases the pressure in the powerhouse to avoid cavitation, like Seneca PHS in the USA [25] as shown in Fig. 3.7A. Closed-loop PHS consists of an upper and lower reservoir far from a large water source and, thus, with a limited water input into the system (Fig. 3.7B). These systems can be implemented in small artificial lakes, filled either by the precipitation of its limited catchment area or by water brought from a different location [21,26]. The environmental impact of closed-loop PHS plants is usually smaller than open-loop plants.

Figure 3.7 Three types of PHS arrangements. (A) Open-loop PHS plant with no need for excavation [23], (B) closed-loop PHS with no considerable inflow in the upper or lower reservoir [24], (C) pump-back PHS with no need for excavation [24].


Storing Energy

However, they are usually limited to daily or weekly storage cycles. An example of a closed-loop project is the Marmora PHS in Canada [27]. Pump-back storage consists of installing pump-turbine in hydropower dams wherever there is another reservoir immediately downstream. This allows the water flow back and forward between the two reservoirs [28] (Fig. 3.7C). This arrangement increases flexibility and operational range as the pump-turbines can be used for both hydropower and energy storage. For example, in case of a drought, conventional hydropower generation will be reduced, but the plant can still be used as pumped storage. The generation head of pump-back storage plants is usually low. However, the system is viable, as long as tunnels are not required. In Japan, a number of dams were built with reversible turbines [24]. This is due to the historic dependence of Japan on nuclear energy, an inflexible source of generation, which creates the need for daily energy storage. The pump-back plants can also be used as part of a water supply solution. The precipitation downstream Japanese rivers can be pumped upstream by pump-back storage plants to be stored on the head of the river for later use. Without a pump-back solution, some of the water would be discharged to the sea. An example of such a scheme is Kannagawa in Japan [29]. Run-of-the-river SPHS plants can store water from a main river, without the need to dam the river (Fig. 3.8), thus, reducing social and environmental impacts [30,31]. Run-of-the-river SPHS are used to extract continuous amounts of water from the river during periods of high river flowrate and return continuous amounts of water to the river during periods with low river flowrate. The constant return of water intends to reduce the impact of river flow variations, which impacts the ecosystem in and around the river. The lower reservoir, which is not on the main river, is used as a standard PHS lower reservoir. In this way, the same pump-turbines can be used both to regulate the river and as an energy storage solution. The high-head pump-turbines can only move water from the lower reservoir or from the river to the upper reservoir and vice versa. With the intent of increasing the flexibility of the plant, a low-head pump-turbine can be built to pump water from the river to the lower reservoir, to keep a constant flow in the river. An example of run-of-the-river PHS is Malta in Austria [32].

Figure 3.8 Run-of-the-river seasonal pumped hydro storage with a large upper reservoir and a small lower reservoir [12].

Pumped hydro storage (PHS)


4. Services provided by PHS plants Apart from generating electricity during peak hours, PHS plants can provide several other services. Given the rise in new energy storage alternatives, such as batteries and hydrogen, PHS should provide other services other than short- and long-term energy storage. Table 3.3 presents the possible services provided by PHS plants. Regarding water, PHS plants are both a challenge and opportunity. The challenges are the dry regions might not have water for PHS storage or evaporation rates might result in large energy losses. The opportunity is that PHS not only stores energy, but it also stores water. This characteristic increases the availability of efficient water reservoirs to provide water management solutions, such as river flow regulation,

Table 3.3 Possible services provided by PHS plants. Uses for PHS


Energy services Energy storage

Highly seasonal hydropower generation [14,15,33]

Goal for CO2 emissions reduction [34e36]

Seasonal energy supply and demand variations [37]

Energy storage for peak generation, intermittent renewable energies such as wind and solar, optimize electricity transmission, among others. Increase water and energy storage in water basins to regulate the river flow and increase hydropower generation. Store excess water during periods of high hydropower generation and reduce spillage. Hydropower, solar and wind generation usually do not have the same seasonal generation profile as the demand for electricity. Natural gas is an option for flexible electricity generation; however, it is a fossil fuelebased source of energy and emits CO2. A seasonal storage option should be considered by countries that intend to considerably reduce CO2 emissions. Countries in high latitudes have a very seasonal solar power generation profile. Seasonal storage allows using the energy stored in the summer during the winter, when there is lower solar generation. Countries in the mid and high latitudes tend to have a seasonal electricity demand profile, consuming more electricity in summer for cooling and during the winter for heating purposes, respectively. Typically, the peak national grid demand can be two to three times as high as the minimum demand. With the electrification of the heating sector in countries at high latitude, the demand of electricity during the winter will increase even further. Continued


Storing Energy

Table 3.3 Possible services provided by PHS plants.dcont’d Uses for PHS


Hydrogen production

With the increase in hydrogen production in the future, PHS could be used to store electricity during periods of excess VRE generation, and generate electricity during periods with low VRE generation in order to increase the capacity factor of the hydrogen production and liquefaction facilities, which have high investment costs. Reduction in fluctuation of electricity prices with fossil fuel prices and supply. Reduction in fluctuation of electricity prices with renewable energy availability, especially hydropower. Reduction in fluctuation of electricity prices with the demand for electricity.

Energy security [38]

Water services Water storage

High storage reservoir sedimentation

Better water quality control

Flood control

Transport with waterways

PHS plants can store water on higher ground away from the river, in cases where along the river is infeasible or due to high evaporation rates. PHS projects have much smaller sedimentation rates than conventional dams due to the small catchment area. This is a significant advantage to conventional reservoir dams with high sediment flow. Conventional dams could lose their water storage capacity in less than 50 years due to sedimentation. Storing the water parallel to the river allows for a better control of the water quality in the reservoir, as it would not be directly affected by the fluctuations in water quality in the main river. When the river flow is high and pollution is diluted, the PHS stores the water. When the river flow is low and pollution concentration is high, the cleaner water stored in the PHS plant can be used to supply the water needs. PHS plants can be used in combination with conventional flood control mechanisms to improve their efficacy. For example, the lower reservoir in the main river could operate at its lowest level, as the upper reservoir of the PHS plant provides long-term water storage services. When there is a flooding event, the lower reservoir has enough storage capacity to contain the flood. PHS plant channels could be also used for transport in waterways, combining the transport of water and goods. Additionally, the improvement in water management resulted from an SPHS plant would reduce the changes when a waterway runs out of water.

Pumped hydro storage (PHS)


Table 3.3 Possible services provided by PHS plants.dcont’d Uses for PHS


Interbasin transfer

PHS projects can be combined with an interbasin transfer project to increase the water security of a region or provide balancing between watersheds. PHS plants used for interbasin transfer usually have longer tunnels or use the upper reservoir as a canal to facilitate water basin transposition, e.g., Snowy Mountain scheme in Australia [39] and the Grand Coulee dam in the USA [40,41]. PHS plants can be designed with high water level variations, which reduce the flooded area/water storage ratio, which reduces the evaporation of water in the reservoir per volume of water stored. This is particularly interesting in dry regions without appropriate locations for effective water storage reservoirs. An example of this is shown in Reference [42]. PHS plants increase the water storage capacity in regions where effective conventional storage reservoirs are not available. This greatly increases the number of water management solutions for a basin, increasing its water security.

Low evaporation

Water security

Environmental services Lower environmental and social impacts [43]

Damming a major river for energy and water storage would have higher environmental and social impacts than damming a small tributary river. PHS plants also provide energy and water storage services without fragmenting the ecosystem of a main river.

pluriannual water storage, flood control, water quality control, interbasin transfer, low evaporation storage, water security, low sedimentation water storage. PHS can provide energy and water storage combined with demand side management desalination as an effective way to store energy from variable energy sources and optimize the energy and water supply in an island or coastal water-scarce regions. During periods with excess energy some of the excess energy is used to desalinate water and the other part is used to pump the desalinated water into the PHS system and then to the upper reservoir. The PHS system not only stores energy but also stores water. When there is a lack of electricity in the system, the water stored in the PHS system is used for water supply and electricity is generated in the PHS plant. An example of this integration happens in the Soria-Chira plant in the Canary Islands [17,18].



Storing Energy

New arrangements for PHS

This section presents new PHS arrangements that have not yet been implemented.


Combined short and long-term cycle seasonal pumped hydro storage (CCSPHS)

As mentioned previously in this chapter, the need for large, long-term, seasonal reservoirs will increase in the future, and PHS will be a competitive long-term energy and water storage alternative. The CCSPHS has the main objective to allow for head variation greater than 50% in order to increase water and energy storage capacity in the main reservoir in locations where the topography is not appropriate for conventional dams and PHS plants. In these arrangements the water can be shifted around the three proposed reservoirs and fulfill short-term energy storage needs and longterm energy and water storage needs [44]. The SPHS arrangement presented in Fig. 3.9A consists of a small lower reservoir in the river, a large intermediate reservoir, and a small upper reservoir. Water flows from the lower and intermediate reservoir to the upper reservoir and vice versa. However, the large head variation does not allow water to be pumped from the lower to the intermediate reservoir at all heads. Thus, this arrangement would only work if shortand long-energy storage needs are combined. For example, water pumped from the river to the upper reservoir at night is then released during the day to the intermediate reservoir as part of a daily energy storage cycle [44]. During the day water from the upper reservoir flows to the intermediate reservoir generating electricity while at the same time storing water in the seasonal reservoir. The large intermediate reservoir can have a large head variation as the water comes from the upper reservoir. The combination of the two cycles (short- and long-term) is important because a pump-turbine system would not be able to pump water from the lower reservoir to the intermediate reservoir due to the pump-turbine limitation in head variation. This arrangement is proposed for a location where the topography does not allow the construction of large storage reservoirs. Another possibility is to build two medium-sized reservoirs, as shown in Fig. 3.9B. The operation would be similar to the presented in Fig. 3.9A. Given that the storage potential is split in two medium-sized reservoirs, the social and environmental impacts may be larger. It also has a greater operation flexibility, as the two reservoirs will have enough water for long-term storage cycles regardless of the river flow [44]. Fig. 3.9C presents the arrangement that allows the highest water level variation in flat topography regions, which in turn contributes to a smaller land requirement in relation to water storage capacity. It also significantly reduces evaporation for water storage in arid regions. In this arrangement, the intermediate reservoir would be filled up with water from the lower reservoir when the intermediate reservoir level is high, and it would be filled from the upper reservoir, when the intermediate reservoir level is low [44]. This change in operation from the lower to the upper reservoir is important because the head of the pump-turbine cannot vary with all the reservoirs level variation

Pumped hydro storage (PHS)


Figure 3.9 SPHS arrangements for combined short and long-term storage with (A) small upper reservoir and a large intermediate reservoir, (B) medium upper reservoir and medium intermediate reservoir, (C) intermediate reservoir divided in two sections [44].

as it is limited to, for example, to 50% of the maximum head. The operation in Fig. 3.9C divides the head variation of the pump-turbine in almost half, and the minimum designed pumping head capacity is higher than in Fig. 3.9A, which reduces tunnel costs.



Storing Energy

Combined hydropower and pumped hydro storage (CHPHS)

A CHPHS plant can be used for hydropower generation or for energy storage (Fig. 3.10A). The lower reservoir is built on the main river and the powerhouse is built downstream of the dam. This arrangement does not require excavation, as the water level in the river dam already maintains the required pressure on the pump-turbine to prevent cavitation [44]. This considerably reduces project costs, especially if the plant has a low generating head [3]. This arrangement is similar to the one in the Seneca PHS [25]. It offers flexibility for the operation of the system, making it possible to decide if the dam generates hydropower, e.g., during periods of large river flow, or if the pumped hydro storage is to be used to help manage the grid (energy storage) or to increase river flow during dry periods. In order for these arrangements to work properly, the height of the reservoirs must match each other as shown in Fig. 3.10, where “X” represents the height of the reservoir. Another alternative for CHPHS plant is to excavate the powerhouse and integrate a lower reservoir to the system. This would result in three or more reservoirs instead of two. These can be the upper, intermediate and lower reservoirs, as shown in Fig. 3.10B for a three-reservoir case. This arrangement consists of 2 dams built in the main river

Figure 3.10 Combined hydropower and pumped hydro storage (CHPHS) arrangement. (A) Without lower reservoir and without the need for powerhouse excavation. (B) With lower reservoir and upper reservoir divided into two sections. (C) With multiple reservoirs connected [44].

Pumped hydro storage (PHS)


and a larger reservoir dam on a tributary river. These reservoirs are connected via tunnels to the same pump/turbines, providing flexibility to operate in a variety of different modes. The upper reservoir should store large amount of water and energy, like SPHS plants. If there is only need to store short-term energy, a pump-back solution would be much more practical and cheaper [44].


Integrated pumped hydro reverse osmosis (IPHRO) system

An innovative alternative for combining PHS with reverse osmosis is named Integrated Pumped Hydro Reverse Osmosis (IPHRO) system. This integration is interesting because the constant pressure exerted from a salty upper reservoir in the reverse osmosis membranes lowers the costs and increase the efficiency of reverse osmosis systems. Another benefit is that the brine returned to the ocean from the IPHRO system is diluted due to the operation of the PHS plant. The challenges of this technology is to find limited locations close to the coast with PHS potential and water desalination demand [45]. Fig. 3.11 presents a diagram of an integrated pumped hydro reverse osmosis system.


Other arrangements

Other less common configurations of PHS include underground PHS [46e49], decommissioned open pit mines PHS [50,51], seawater PHS [52e54], gravity-based cylindrical systems [55,56], offshore water storage at sea [57], and storage of water and energy inside wind turbine towers [58].

Figure 3.11 Diagram describing an integrated pumped hydro reverse osmosis plant [45].



Storing Energy

Pump-turbine types

PHS plants can have turbines that operate with a fixed rotation speed or variable speed. The selection between fixed and variable speed depends on two main factors, the need to increase the variability of the pumping and generation power capacity of each individual plant, and the need to increase the variability for the head variation of PHS plants and project costs. Variable speed pump-turbines cost approximately 30% more than fixed-speed alternatives and are not commonly used today [59]. The need to increase the variability of the pumping and generation power capacity of each individual plant is particularly interesting in grids that have a large concentration of wind and solar power generation. Fixed speed pumpeturbines can only increase the pumping capacity in fixed steps proportional to the number of fixed pumpeturbines, as shown in Figs. 3.12 and 3.13. This fixed speed arrangement does not allow for the generation from VRS to the stored, and some of the generation has to be curtailed or stored with other energy storage alternatives. On the other hand, variable speed pumpeturbines allow more flexibility to the variation of the pumping mode power consumption. This allows for a better utilization of the generation of VRS, as more energy can be stored. If the main objective of a four-pump-turbines plant is to store energy from VRS, the PHS plant only requires two variable speed pumpe turbines to store the generation from VRS. The final choice between fixed and variable speed turbines depends on technoeconomic and demand aspects [61]. With the increase of intermittent renewables in the grid, variable speed turbines might become more common, which would reduce its price.

Figure 3.12 Power consumption and generation variation with (A) two units with fixed speed and (B) two units with variable speed [60].

Pumped hydro storage (PHS)


Figure 3.13 Flexibility for integrating surplus power from the grid with (top) two units with fixed speed and (bottom) two units with variable speed [60].

Another important benefit of variable pumpeturbines is the increase in the allowed variation in generation head in the design of a PHS. This is particularly interesting for monthly, seasonal, and pluriannual PHS plants, with large reservoirs. Instead of increasing the flooded area of the upper reservoirs, a PHS with large head variation can store large amounts of water with a reservoir with a small flooded area and a high level variation. Assuming that the gross head variation factor is equated by Hmax/Hmin (maximum head/minimum head), the following rule of thumb may help for a basic orientation as presented in Table 3.4 [60]. Table 3.5 presents existing PHS sites with pumping/generation head variations as high as 42.5%. Table 3.4 Typical head variation with different pump-turbine types [60].


Pump-turbine type

Head variation


Reversible pump-turbine units with fixed speed Ternary units with fixed speed Reversible pump-turbine units with variable speed using a DFIMa Reversible pump-turbine units with variable speed using a full-size power converter

20% 20% 30%e50%

1.2 1.2 1.5e2



Doubly fed induction machine.


Table 3.5 PHS sites with high pumping/generation head variation [31,62].

Project name


Head (m)

Head variation (m)

Variation percent (%)

Power (MW)

Speed (rpm)

Rotation speed


Nant de Drance Linthal Tehri Limberg II

6 4 4 2

250e390 560e724 127e221 273e432

140 164 94 159

35.9 22.7 42.5 36.8

157 250 255 240

428.6%  7% 500%  6% 230.8%  7.5% 428.6

Variable Variable Variable Fixed

Switzerland Switzerland India Austria

Storing Energy

Pumped hydro storage (PHS)


PHS plants can operate with single reversible turbine units or multistage pumps in ternary units. Single reversible turbines are usually applied to PHS plants with generation heads lower than 800 m. Multistage pumps in ternary units are used in PHS plants with heads higher than 800 m [60], as shown in Fig. 3.14. PHS plants with heads over 800 m are interesting due to the lower water flow requirements, shorter number of tunnels and the smaller tunnel diameter to achieve the same generation and pumping capacity. Additionally, these plants require small upper reservoirs to store large amount of energy. A ternary unit consists of a multistage pump for charging and a Pelton turbine for generation. These plants have the advantage of operating in hydraulic short circuit mode to provide ancillary services. In hydraulic short circuit mode, the pump capacity can vary from 160 to 25 MW, while the pumping capacity can vary from 15 to 150 MW. For more details on ternary pump/turbine design, please refer to Reference [60]. The quaternary PHS technology is the fastest responding pumped hydro technology available for grid services [65]. An example is the Gordon Butte facility. Its configuration consists of separate pumps and turbines, each with a dedicated 134 MW motor and 134 MW generator [66]. The equipment is also connected in a hydraulic short circuit - basically a hydraulic loop connecting the turbine and the pump utilizing the lower reservoir. This configuration allows the facility to both pump and generate at the same time and seamlessly switch from pumping to generation and back again (including cold-start) at an estimated 20þ MW/s. Another alternative to further increase the head variation of an SPHS plant is to have a multistage pump connected to one generator, in which the pumps can operate in series to achieve high heads and operate in parallel with the generation head is low [67].

Figure 3.14 Operation ranges for single-stage reversible pump turbines (charging and discharging) and multistage pumps for high heads (charging) [60].



Storing Energy

World potential for PHS

There are several researches that have explored the global potential for PHS plants worldwide. In this book chapter, we will describe three mapping efforts. The AREMIS model developed 616,000 potentially feasible offriver, closed loop PHS plants, which sum up to 23 PWh of energy storage (Fig. 3.15) [63,64,68]. It provides a virtual identification of potential reservoirs, focusing on PHS plants with 100e800 m height difference for targeted energy storage of 15 GWh. The CAPEX costs are estimated as a function of head, water-to-dam volume ratio, power, and distance between the upper and lower reservoir. It also retains and ranks good sites, removing from the database projects that overlap with higher costs. ANU is about to release new maps, one where the lower reservoir is an existing water body, such as reservoir, lake, or the ocean, and the other using pit mines as the upper or lower reservoirs. Fig. 3.16 presents a section of the model used to estimate the world potential for Integrated Pumped Hydro Reverse Osmosis (IPHRO) system [45]. The model estimates the potential for IPHRO combining the distance from the coast (tunnel length), the altitude (generation head), the slope of the topography (suitable sites).

Figure 3.15 AREMIS PHS potential mapping software, showing the representation of the upper and lower reservoirs, and the tunnel of a proposed PHS plant [63,64].

Pumped hydro storage (PHS)


Figure 3.16 Integrated pumped hydro reverse osmosis system potential in a coastal region of California, USA [45].

A recent IIASA PHS mapping potential model mapped four million open-source seasonal PHS projects developed around the world. The projects are open cycle and connected to a river with large flowrate, energy storage potential of 17.3 PWh (below 50 US$/MWh), storage costs as low as 1.8 US$/MWh in the Indus Basin, and high generation capacity factors of 30%e40%. They can operate in daily, weekly, monthly, and seasonal cycles. Interesting energy, water, and land mitigation are solution to climate change vulnerability [16]. Fig. 3.17 presents the costs for energy and water storage worldwide and the capacity curves for energy and water storage in different continents. Fig. 3.18 presents mode cost details of seasonal pumped storage plants developed by the IIASA model. An interactive map presenting the results from this work in detail can be seen in Reference [69]. An upgraded version of this mapping tool has been developed by GESEL for Brazil. The interactive map for the potential of daily PHS is available at and the seasonal PHS map potential is available at sazonais.php.

8. Conclusion This book chapter concludes that, given the rapid price reduction of batteries, PHS plants in the future should not be limited to daily and weekly storage cycles, and


Storing Energy

Figure 3.17 Seasonal pumped hydropower storage world cost and flooded area maps. (A) Water storage costs and capacity curve in km3. (B) Energy storage without considering hydropower plants in cascade costs and capacity curve in US$ MWh1. (C) Energy storage considering hydropower plants in cascade costs and capacity curve in US$ MWh1. (D) Additional generation capacity costs and capacity curve in US$ kW1. (E) Percentage of the reservoir that is filled with the river inflow into the SPHS reservoir. (F) Average land requirement for energy storage in different basins.

PHS with larger storage cycles will become more common. The main competitor for large-scale PHS plants in the future will be hydrogen, with PHS having the advantage of storing energy with higher efficiencies. Another important contribution that will increase the viability of PHS plants is their capacity for storing large amounts of water with low evaporation and land requirement.

Pumped hydro storage (PHS)


Figure 3.18 Seasonal pumped hydropower storage (SPHS) costs and description. (A) Water and energy SPHS project cost distribution shows that the most expensive components tend to be the tunnel and dam. (B) Example of energy storage cost variation with cascade according to different heights for the example project in Fig. 3.18C. The energy storage cost reduces with the increase in dam height due to economies of gains; however, it then increases because the reservoir becomes larger than the amount of water available to be sustainably stored. (C) Presentation of selected project in Tibet, China, on a topographic map, presenting its tunnel in black and reservoir in purple. (D) Zoom in the selected project.

Acknowledgments This research was funded by National Agency of Petroleum, Natural Gas and Biofuels (ANP), the Financier of Studies and Projects (FINEP) and the Ministry of Science, Technology and Innovation (MCTI) through the ANP Human Resources Program for the Oil and Gas Sector Gas PRH-ANP/MCTI, in particular PRH-ANP 53.1 UFES, for all the financial support received through the grant. This research was also funded by R&D funding from the Brazilian Agency of Electric Energy, Enercan, BAESA, Ceran, Foz do Chapec o, Paulista Lajeado Energia and CPFL.

References [1] R. Viadero, M. Rehbein, A. Singh, Hydropower on the Mississippi River. Rais. Grade up. Mississippi River - Hydro Potential Mississippi River, Moline, IL, 2017. [2] L. Rognlien, Pumped Storage Development in Øvre. Otra, Norway, MSc Thesis. Otra, 2012.


Storing Energy

[3] J. Slapgard, Cost Base for Hydropower Plants, Norwegian University of Science and Technology, Oslo, 2012. [4] International Hydropower Association, Hydropower Status Report. 2020, 2020. [5] E. Barbour, I.A.G. Wilson, J. Radcliffe, Y. Ding, Y. Li, A review of pumped hydro energy storage development in significant international electricity markets, Renew. Sustain. Energy Rev. 61 (2016) 421e432, [6] International Hydropower Association, Pumped Storage Tracking Tool, 2019. https:// [7] P. Barbosa, M. Adelaide, S. Bajay, J. Yasuoka, V. Pinheiro, Usinas hidrelétricas reversíveis combinadas com hidrelétricas em cascata e seus benefícios para a gest~ao do setor elétrico brasileiro. Rio de Janeiro, 2020. [8] International Hydropower Association, International Forum on Pumped Storage Hydropower. IHA, 2020. [9] Sandia National Laboratories, DOE OE Global Energy Storage Database. USA Dep Energy, 2020. [10] BloombergNEF. Energy Storage is a $620 Billion Investment Opportunity to 2040 2020. [11] K. Keen, As battery costs plummet, lithium-ion innovation hits limits, experts say, S&P Glob. (2020). [12] J. Hunt, E. Byers, K. Riahi, S. Langan, Comparison between seasonal pumped-storage and conventional reservoir dams from the water, energy and land nexus perspective, Energy Convers. Manag. 116 (2018) 385e401. [13] E. Solvang, J. Charmasson, J. Sauterlaute, A. Harby, Å. Killingtveit, H. Egeland, et al., Norwegian Hydropower for Large Scale Electricity Balancing Needs - Pilot Study of Technical, Environmental and Social Challenges, SINTEF Energy Research, Trondheim, 2014. [14] J.D. Hunt, M.A.V.D. Freitas, A.O. Pereira Junior, A review of seasonal pumped-storage combined with dams in cascade in Brazil, Renew. Sustain. Energy Rev. 70 (2017), [15] J.D. Hunt, M.A.V. Freitas, A.O. Pereira Junior, Enhanced-Pumped-Storage: combining pumped-storage in a yearly storage cycle with dams in cascade in Brazil, Energy 78 (2014), [16] J. Hunt, E. Byers, Y. Wada, S. Parkinson, D. Gernaat, S. Langan, et al., Global resource potential of seasonal pumped-storage for energy and water storage, Nat. Commun. 11 (2020). Article number: 947. [17] C. Bueno, J.A. Carta, Wind powered pumped hydro storage systems, a means of increasing the penetration of renewable energy in the Canary Islands, Renew. Sustain. Energy Rev. 10 (2006) 312e340, [18] U. Portero, S. Velazquez, J.A. Carta, Sizing of a wind-hydro system using a reversible hydraulic facility with seawater. A case study in the Canary Islands, Energy Convers. Manag. 106 (2015) 1251e1263, [19] D. Newbery, Shifting demand and supply over time and space to manage intermittent generation: the economics of electrical storage, Energy Pol. 113 (2018) 711e720, https:// [20] D. Huertas-Hernando, H. Farahmand, H. Holttinen, J. Kiviluoma, E. Rinne, L. Söder, et al., Hydro power flexibility for power systems with variable renewable energy sources: an IEA Task 25 collaboration, Wiley Interdiscip. Rev. Energy Environ. 6 (2017) e220, https://

Pumped hydro storage (PHS)


[21] M. Chazarra, J.I. Pérez-Díaz, J. García-Gonzalez, Deriving optimal end of day storage for pumped-storage power plants in the joint energy and reserve day-ahead scheduling, Energies 10 (2017), [22] G. Butera, S.H. Jensen, L.R. Clausen, A novel system for large-scale storage of electricity as synthetic natural gas using reversible pressurized solid oxide cells, Energy 166 (2019) 738e754, [23] J. Kerr, Usinas reversíveis e outros elementos especiais de sistemas de reservat orios. IV Semin, Nac. Produç~ao e Transm. Energ. Elétrica (1977) 1e32.  Gallachoir, E.J. McKeogh, Techno-economic review of existing and [24] J.P. Deane, B.P.O. new pumped hydro energy storage plant, Renew. Sustain. Energy Rev. 14 (2010) 1293e1302, [25] J.P. Fitzgerald, Operation of seneca pumped storage plant, IEEE Trans. Power Appar. Syst. (1973), PAS-92:1510e6. [26] A. Blakers, B. Lu, M. Stocks, 100% renewable electricity in Australia, Energy 133 (2017) 471e482, [27] Northland Power, Marmora Pumped Storage, 2018. [28] Nadler H. Hydropower pump-back projects/perspectives. Southwest. Fed. Hydropower Conf., Tulsa, Oklahoma: n.d. [29] R. Peltier, Kannagawa hydropower plant, Japan, Power 150 (2006) 54e58. [30] G. Grill, B. Lehner, A. Lumsdon, G. MacDonald, C. Zarfl, C. Liermann, An index-based framework for assessing patterns and trends in river fragmentation and flow regulation by global dams at multiple scales, Environ. Res. Lett. 10 (2015). [31] T. Datry, A. Boulton, N. Bonada, K. Fritz, C. Leigh, E. Sauquet, et al., Flow intermittence and ecosystem services in rivers of the Anthropocene, J. Appl. Ecol. 1e12 (2017). [32] B. Wagner, C. Hauer, A. Schoder, H. Habersack, A review of hydropower in Austria: Past, present and future development, Renew. Sustain. Energy Rev. 50 (2015) 304e314, https:// [33] J. Olauson, M.N. Ayob, M. Bergkvist, N. Carpman, V. Castellucci, A. Goude, et al., Net load variability in Nordic countries with a highly or fully renewable power system, Nat. Energy 1 (2016), [34] International Energy Agency, I. Shin-Ichi, Prospects for Large-Scale Energy Storage in Decarbonised Power Grids. Paris, 2009. [35] A. Converse, Seasonal energy storage in a renewable energy system, Proc. IEEE 100 (2011) 401e409. [36] M. Pehl, A. Arvesen, F. Humpenöder, A. Popp, E.G. Hertwich, G. Luderer, Understanding future emissions from low-carbon power systems by integration of life-cycle assessment and integrated energy modelling, Nat. Energy 2 (2017) 939e945, s41560-017-0032-9. [37] International Energy Agency, Energy Technology Perspectives: Scenarious & Strategies to 2050. Paris, 2008. [38] D. Conway, C. Dalin, W.A. Landman, T.J. Osborn, Hydropower plans in eastern and southern Africa increase risk of concurrent climate-related electricity supply disruption, Nat. Energy 2 (2017) 946e953, [39] Snowy Mountains Hydro-Electric Authority, The Snowy Mountains Scheme: A National Engineering Landmark, Minister for Resources, Talbingo, 1990. [40] US Army Corps of Engineers, Columbia River & Tributaies Pacific Northwest Regional Pumped-Storage Study, 1980. [41] Central Arizona Project, 2013 Annual Water Quality Report, 2014. Phoenix.


Storing Energy

[42] B. Lonnecker, Generator/motors and adjustable-speed drives for Waddell pumped-storage plant, in: Proc. Int. Conf. Hydropower, Portland, 1987. [43] K.O. Winemiller, P.B. McIntyre, L. Castello, E. Fluet-Chouinard, T. Giarrizzo, S. Nam, et al., Balancing hydropower and biodiversity in the Amazon, Congo, and Mekong, Science 351 (2016) 128e129, [44] J.D. Hunt, B. Zakeri, R. Lopes, P.S.F. Barbosa, A. Nascimento, NJ de Castro, et al., Existing and new arrangements of pumped-hydro storage plants, Renew. Sustain. Energy Rev. 129 (2020) 109914. [45] A.H. Slocum, M.N. Haji, A.Z. Trimble, M. Ferrara, S.J. Ghaemsaidi, Integrated pumped hydro reverse osmosis systems, Sustain. Energy Technol. Assess. 18 (2016) 80e99, [46] F. Winde, F. Kaiser, E. Erasmus, Exploring the use of deep level gold mines in South Africa for underground pumped hydroelectric energy storage schemes, Renew. Sustain. Energy Rev. 78 (2017) 668e682, [47] J. Menéndez, J. Loredo, M. Galdo, J.M. Fernandez-Oro, Energy storage in underground coal mines in NW Spain: assessment of an underground lower water reservoir and preliminary energy balance, Renew. Energy 134 (2019) 1381e1391, 10.1016/j.renene.2018.09.042. [48] E. Pujades, A. Jurado, P. Orban, A. Dassargues, Parametric assessment of hydrochemical changes associated to underground pumped hydropower storage, Sci. Total Environ. 659 (2019) 599e611, [49] C.R. Matos, J.F. Carneiro, P.P. Silva, Overview of large-scale underground energy storage technologies for integration of renewable energies and criteria for reservoir identification, J. Energy Storage 21 (2019) 241e258, [50] E. Pujades, P. Orban, S. Bodeux, P. Archambeau, S. Erpicum, A. Dassargues, Underground pumped storage hydropower plants using open pit mines: how do groundwater exchanges influence the efficiency? Appl. Energy 190 (2017) 135e146, 10.1016/j.apenergy.2016.12.093. [51] E. Pujades, T. Willems, S. Bodeux, P. Orban, A. Dassargues, Underground pumped storage hydroelectricity using abandoned works (deep mines or open pits) and the impact on groundwater flow [Hydroélectricité par pompage-turbinage en utilisant des excavations souterraines abandonnées (mines profondes ou carrieres) et, Hydrogeol. J. 24 (2016) 1531e1546, [52] N. Ghorbani, H. Makian, C. Breyer, A GIS-based method to identify potential sites for pumped hydro energy storage - case of Iran, Energy 169 (2019) 854e867, 10.1016/ [53] C.S. Ioakimidis, K.N. Genikomsakis, Integration of seawater pumped-storage in the energy system of the Island of S~ao Miguel (Azores), Sustain. Times 10 (2018), 10.3390/su10103438. [54] M.H. Albadi, A.S. Al-Busaidi, E.F. El-Saadany, Seawater PHES to facilitate wind power integration in dry coastal areas - Duqm case study, Int. J. Renew. Energy Res. 7 (2017) 1363e1375. [55] A. Berrada, K. Loudiyi, I. Zorkani, System design and economic performance of gravity energy storage, J. Clean. Prod. 156 (2017) 317e326, j.jclepro.2017.04.043. [56] Heindl-Energy, Gravity Storage, 2019. [57] M. Puchta, J. Bard, C. Dick, D. Hau, B. Krautkremer, F. Thalemann, et al., Development and testing of a novel offshore pumped storage concept for storing energy at sea  Stensea, J. Energy Storage 14 (2017) 271e275,

Pumped hydro storage (PHS)


[58] T. Grumet, This Unique Combo of Wind and Hydro Power Could Revolutionize Renewable Energy. GE Reports, 2016. [59] VOITH, Pumped Storage Machines: Reversible Pump Turbines, Ternary Sets and MotorGenerators, 2011. [60] A. Rimpel, K. Krueger, Z. Wang, X. Li, A. Palazzolo, J. Kavosi, et al., Mechanical energy storage, Therm. Mech. Hybrid Chem. Energy Storage Syst. (2020). London. [61] W. Yang, J. Yang, Advantage of variable-speed pumped storage plants for mitigating wind power variations: integrated modelling and performance assessment, Appl. Energy 237 (2019) 720e732, [62] J. Henry, F. Maurer, J. Drommi, T. Sautereau, Converting to Variable Speed at a PumpedStorage Plant. Orlando, 2013. [63] M. Stocks, A. Blakers, Australian renewable energy mapping infrastructure (AREMIS), Aust. Renew. Energy Agency (2018).¼soDPMo1jDBBtwBNhD. [64] M. Stocks, R. Stocks, B. Lu, C. Cheng, A. Blakers, Global atlas of closed loop pumped hydro energy storage, Joule (2020) 19. [65] Z. Dong, J. Tan, E. Muljadi, R. Nelms, A. St-Hilaire, M. Pevarnik, et al., Developing of quaternary pumped storage hydropower for dynamic studies, IEEE Trans. Sustain. Energy 1 (2020), [66] Northwestern Energy, Electricity Supply Resource Procurement Plan: Comments Before the Montana Public Service Commission, 2019. [67] M. Marriott, Nalluri and Featherstone’s Civil Engineering Hydraulics: Essential Theory with Worked Examples, Wiley-Blackwell, Oxford, 2016. [68] B. Lu, M. Stocks, A. Blakers, K. Anderson, Geographic information system algorithms to locate prospective sites for pumped hydro energy storage, Appl. Energy 222 (2018) 300e312, [69] J. Hunt, Global Resource Potential of Seasonal Pumped Hydropower Storage for Energy Storage, Google Maps, 2021.¼1O9aK_ dTL3mDOgLgY2G0BSgmlHqRNSlHA.

This page intentionally left blank

Novel hydroelectric storage concepts


Frank Escombe EscoVale Consultancy Services, Reigate, United Kingdom Contributor retains copyright and grants to us, the Publisher, the non-exclusive license, to publish.

1. Introduction 1.1

Scope and purpose

Following on from coverage of conventional pumped hydroelectric storage (PHES), this chapter examines other concepts that share the same principledusing hydroelectric equipment (pump-turbines/motor generators) to convert electrical energy to and from gravitational potential energy. It explores their ability to complement PHES by: ⁃ Extending use into regions where the terrain is unsuitable for conventional PHES or where public opposition/approval difficulties (areas of outstanding natural beauty or villages or agricultural land or conflicting priorities for water resources) make PHES projects impractical. ⁃ Opening up new applicationsdin particular, to meet future requirements in electricity networks with high concentrations of wind and solar resources that need much longer duration/ higher energy solutions than are usually feasible with conventional PHES.



Developers of these novel concepts hope to benefit from the status, technical expertise, and market dominance of PHES, which accounts for >95% of the global grid-scale installed capacity. They also have to live within its limitations. The main problem with gravitational storage is gravity, which is incredibly weak compared to the other fundamental forces. This is a good thing for the universe but rather inconvenient for energy storage developers. As an exercise in the absurd, a record-breaking weightlifter can support an astonishing mass of over 400 kg. If an electricity company could persuade him to store energy by carrying this from sea level to the top of Everest, it could afford to pay him almost $1 for his trouble. A simple time-shifting energy storage service is not worth much to the average electricity supplier [100e500 h, say). In addition to providing fairly longterm storage and quasi generator services, the objective is to harvest energy that would otherwise be lost from the electricity supply system. ➢ “Long-duration seasonal storage” (w1000 h, say). A hypothetical seasonal storage system may use summer sunshine to power winter loads. In practice, there usually comes a point where the marginal cost of adding electrical energy storage capacity cannot be justified by the additional energy throughput. The marginal cost of storing thermal energy is much lower and seasonal storage of hot water, ice, or other media can be attractive, if thermal losses can be kept to an acceptable level over a long period.

2. High-density fluid PHES 2.1


PHES systems normally use freshwater sources, but a few operate with seawater. For a given configuration, seawater should provide slightly higher output, simply because its density is 2%e3% greater than freshwater. RheEnergise [2] has taken this much


Storing Energy

further, with an innovative concept based on a fluid 2.5 times as dense as water. Benefits include: • •

It enables projects to be undertaken at relatively small hills. For example, a 300 m elevation difference between upper and lower reservoirs provides 7.5 MPa (75 bar) operating pressureda figure exceeded by only a handful of conventional PHES systems. Potential sites are commonplace (RheEnergise estimates over 100,000 in Europe and some 1.8M globally within their 75e300 m target range). This provides opportunities in many networks that could not contemplate conventional PHESdand suitable sites are likely to be much closer to transmission networks, access roads, and load centers than projects in mountainous terrain. Some could be colocated with large renewable energy projects. Storing a MWh requires 2.5 times less fluid transfer than conventional PHES, for a given difference in reservoir elevation. This reduces the physical size of hydroelectric equipment, reservoirs, and penstocks, with the potential for significant capital cost savings and short development times (perhaps 2 years or so, with a construction phase averaging 12e15 months). Projects are unobtrusive because of their reduced scale and the fact that they are easily landscaped on completion (fluid storage tanks, penstocks, etc., can be buried). They can be constructed so that they do not require or interfere with local water resources, avoiding another contentious issue. The large number of potential sites enables a business model that seeks repeatable projects, using standard designs at multiple sites with similar topology. This should build confidence among clients, simplify approval procedures, and reduce costs further through series production of major subsystems, while avoiding most of the one-off design and planning costs associated with PHES projects.


Operation and performance characteristics

The plant layout is much the same as that for conventional PHES, except that the highdensity fluid (HDF) is transferred between tanks, rather than open reservoirs. The system is based on familiar reversible pump-turbines, designed for use with HDF (there are significant differences between a 200 m head HDF turbine and a 500 m head water turbine of the same rating). Use of HDF introduces specific O&M requirements. The fluid is a water-based suspension of multiple components. It is nontoxic, environmentally benign, nonreactive, and noncorrosive. It is stable enough to remain idle for 2e3 weeks, if there are periods when plant operation is not required (after which, the clock can be “reset” by circulating w20% of the fluid volume through the HDF management subsystem). If necessary, the fluid can be dewatered and kept stable for monthsdfor example, during a major refurbishment or upgrade. HDF is more abrasive than water, which would shorten the working life of key components (e.g., the pump-turbine) if the same specification were used as for a water-based system. In practice, the specification can be altered to provide an appropriate balance between capital and operating costs. A closed-loop HDF system is highly predictable and so remedial work forms part of the planned maintenance schedule. It is expected that the interval between the replacement of wearing parts (at >95% share of the longer duration market (4 h and above). Most analyses credit PHES and/or CAES as the lowest LCOS electricity storage technologies, usually by a considerable margin. Typically, these have been the only ones capable of delivering a differential (above the cost of input charging energy) of 10 GW h worldwide. This compares with w150 medium energy (1e10) GW h plants and maybe 200þ smaller systems of 10 GW h PHES are already hard to come by (see Chapter 2). This limits the ability of PHES to serve future requirements for gigawatt-scale week-ahead and larger strategic storage markets (see Sections 1.5 and 3.4), with system ratings of tens or hundreds of gigawatt-hours. A 500 GW h PHES would need to transfer water between 500 Mt reservoirs (assuming an elevation difference of 400 m). There are few locations where such systems are feasible, requiring two large natural bodies of water (and even with 100 km2 lakes, there may be objections to 5 m changes in water levels). Naturally, 500 GW h (100þ h) piston storage would also be a huge engineering task, but of a different nature. Prospective sites are almost as common for 100 h as for 10 h systems, and are as likely (in fact, more likely) to be found in flattish terrain or fairly close to load centers, rather than in mountains. Hundred hour systems need not require much more space than 10 h schemes and marginal costs are moderate. We can illustrate this with a 3 GW GBES design (matching the current highest power PHES plant). A piston diameter of 800 m (0.5 km2 area) and 8 MPa (80 bar) operating pressure delivers a convenient 1 GW h for every 1 m of travel [6]. A 10 h, 30 GW h “day-ahead” unit (Fig. 4.2, for example) requires 30 m vertical travel. The energy capacity could be doubled at negligible cost by extending the vertical cylinder wall section to allow 60 m travel. There is a limit to the height that the piston can be raised (one reason being that the operating pressure at the power plant increases as the piston rises above the surface). It should be possible to get well into the “weekahead” sectordin our example a 40 h, 120 GW h, 120 m travel system would result in a pressure increase of about 10%. Ultrahigh-energy strategic storage requires a different approach, with a radical increase in the piston diameter, travel distance, or operating pressuredthe latter seems unlikely. A 600 GW h, 200 h system, with the same travel distance and pressure would require a massive (but not inherently more difficult) piston of w1800 m diameter. The marginal cost of the additional storage capacity would be low (perhaps 10% of the dollars per kilowatt-hour cost of the 40 h design and 1 MW h/m2 in terms of cylinder area and >100 kW h/m2 site area. This is one or two orders of magnitude better than PHESdindeed, the energy storage capacity of the entire global PHES portfolio could be matched by just two of the above 600 GW h piston storage systems.


No power limitations

Geological piston storage has no practical upper power limitdthe bigger, the better because of strong economic benefits of scale. If required, power ratings far above the present storage maximum of w3 GW are feasible. GW-class systems will remain the norm for large-scale storage, although some locations could accommodate >10 GW. The lower threshold will depend on other design features, especially the energy rating, but may be w100 MW. GP’s fabricated piston design is a special case, enabling use of relatively small diameters at powers down to tens of MW in applications that are not accessible to Heindl Energy and GBES.


Asymmetric charging

Asymmetric charging is an interesting design quirk for slow-speed piston designs. It should be one of the most valuable features for future networks with a very high proportion of variable renewables. Slow speed gives many hours of power delivery from a limited travel distance and more options in terms of seal subsystems. It also simplifies transition from discharge to charge and makes it much easier to handle extreme faultsde.g., loss of grid connection when the piston is traveling at its highest descent speed and generating maximum power. Basically, if the turbines are unable to extract energy and send it to the grid, we need to deal with the kinetic energy in the system and the large amount of potential energy that would be released, if the piston continued to descend an appreciable distance before it is brought to a halt. That is relatively easy if the highest descent speed is a brisk snail’s pace. Piston speed is not an issue during the charging phase, with energy taken from the grid and converted via pumps into potential energy as the piston rises. Gravity assists (rather than counteracts) a switch from charge to discharge mode and aids piston deceleration if grid power is lost. Also, some of the more complex components of seal systems may not be needed when the piston is ascending. Consequently, the storage system can incorporate extra pumping capacity (at low cost), increasing the input power rating during the charge cycle. The charge rating can be several times the power delivery rating, if this is beneficial.


Storing Energy

Asymmetric charging has little value at present. Large-scale storage systems usually have plenty of time to gather sufficient overnight energy to provide services during the following day. The situation will be quite different in networks that rely on variable renewables for a substantial part of their energy (see Section 3.4).


Other performance characteristics

The above piston storage features (power, energy, footprint, pressure, location) are quite different from PHES. In other respects, performance and longevity should be similar. Round-trip efficiency may be marginally higher [6] and should exceed 80%. The construction phase would be as disruptive as it is for PHES and more evidentd PHES schemes are usually in very remote locations. However, completed piston storage projects can be unobtrusive and may be aesthetically pleasing or locally beneficial (e.g., submerged piston designs and storage integrated with public water supply or flood prevention). Storage has a much better safety record than electricity production, but all energy projects have safety issues and battery fires attract a lot of attention. In practice, PHES is the main culprit as far as storage is concerned, because of its nearmonopoly of grid-scale systems and the amount of energy stored. Risks are largely confined to construction and supply industry personnel. Piston systems carry an extra element of risk because they will be somewhat closer to the general public. However, in the unlikely event that a terawatt-hour of gravitational energy escapes, it will be comforting if this happens hundreds of meters below the general public’s feet, rather than above its head. This is not just a question of up and down. As in a traffic accident, the speed at which it happens is critical. A megaton nuclear explosion can release 1 TW h almost instantaneously, destroying a city. The same energy equivalent of a Richter scale 7 earthquake can cause widespread damage in less than a minute. Catastrophic failure of a TW h hydroelectric dam might drain a 109 t (billion tonne) reservoir in tens of minutes, with serious consequences along downstream watercourses. In comparison, fragmentation of a 1 TW h GBES piston would be a slow-motion car crash as rock and water jostle to swap places in a confined space dnot a disaster, but hugely expensive and best kept a sensible distance from population centers.


What could possibly go wrong?

GW-class piston storage system will face formidable engineering issues. Those raised most frequently relate to construction difficulties; the seal subsystem; structural integrity of the piston; preventing leakage from the high-pressure chamber beneath the piston; safety considerations (see Section 3.2.6, for example); “parking” the piston for maintenance; and preventing piston tilt (GBES only [6]). These are serious challenges, but they are not obvious show-stoppers. - Construction: In essence, piston storage is much like any other very large construction project and could be undertaken by adapting equipment and techniques that are well-understood in the civil engineering, electromechanical, hydroelectric, and extractive industries. Piston

Novel hydroelectric storage concepts


storage may encourage development of advanced technologies (e.g., for rock cutting), but it does not depend on them. - Seal: The seal subsystem is absolutely critical to the success of piston storage. All the developers believe that they have plausible solutions for evaluation during the crucial technical development stages. - Piston integrity: There is never a cubic kilometer of seamless basalt around when you want one. It may seem that there is little chance of a real-world piston surviving the huge forces exerted by the high-pressure water needed to lift it. In fact, the force on the base of the piston is virtually identical to that which was provided by the underlying rock for millions of years before the piston was separated. This, of course, was just enough to support it. If the exerted force is fractionally more, the piston starts to move up: if fractionally less, it starts to move down. - Leakage: For the same reason, containment of the water beneath the piston is relatively easy, because it is surrounded by material of very similar lithostatic pressure (except for the seal, of course).

The nonengineering challenges are just as important. Given a choice, potential backers prefer charismatic technologies that can be demonstrated convincingly on a laboratory bench. They should promise short-haul development, with opportunities to gain experience (and early revenue) from niche markets. Modular systems are preferred, with economies of production scale rather than physical scale. GWclass piston storage ticks all the wrong boxes, especially for most venture capital investors. It will need to make a particularly convincing case, both technically and economically.

3.3 3.3.1

Piston storage economic performance Capital cost

Cost comparisons for new technologies are not very meaningfuldearly estimates are notoriously unreliable and the comparison point is a moving target. For piston systems, PHES is the logical benchmark and current gold standard for high power storage. PHES specific costs vary widely from project to project, but 1500 d2000 $ kW1 (in terms of power) or about 200e300 $ (kW h)1 (in terms of energy) are reasonable averages. Such figures exclude interest costs during construction, land acquisition, approvals, hook-up costs, and dedicated transmission links. Multibillion dollar annual markets suggest that these cost levels are broadly acceptable for dayahead storage. The range spans w1000 $ kW1e>5000 $ kW1. Below-average costs are more prevalent in developing economies, which account for much of the global activity. High-end costs are sometimes attributed to PHES by those advocating alternative we-can-do-better-than-that technologies. We are not aware of any significant PHES activity toward the top of the cost range. Capital cost estimates put forward by piston storage companies suggest that parity (or better) with PHES costs is plausible in key markets from hundreds of megawatt to multi-gigawatt (Table 4.1). One cannot yet put much confidence in such figures, but there is nothing to suggest that dramatic cost reduction will be needed to match PHES.


Storing Energy

Table 4.1 Piston storage capital cost estimates. Power capital cost/ $ kWL1

Energy capital cost/$ (kW h)L1

Comments (see text)

4 4 16

2000 950 3000

500 240 180




















2000 2000 2000

20 80 400

10 40 200

1300 1400 1800

130 35 9

GP GP GPdexternal estimate HEdconstruction costs only HEdconstruction costs only HEdconstruction costs only HEdconstruction costs only GBES GBES GBES

Power/ MW

Energy/ GW h

200 1600 1600

0.8 6.4 6.4



Gravity power made some quite detailed cost appraisals for its early 4 h modular system, targeting peak power markets, where high-energy batteries are probably the best comparison point. Gravity power has not yet published estimates for 16 h systems and those shown are based on external estimates. The company’s website no longer shows capital costs but does contain comparisons of the levelized cost of storage output [7]. Heindl Energy’s figures [8] are intended to underline the steep fall in specific costs as size increases, rather than to represent the all-in costs of practical storage systems. They relate to the core mechanical construction costs of the piston/cylinder (excavation, reinforcement, seal system, etc.), where the HE model assigns costs to the various construction tasks [9]. They exclude major items such as hydroelectric equipment, the hydraulic circuit, upper reservoir, and powerhouse. The full system cost will be appreciably higher in dollar per kilowatt terms (although below the PHES average). System costs in dollar per kilowatt-hour terms will remain very low for the larger HE systemsdsay 2 $ (kW h)1 for the 2000 GW h design. This is an extreme example, operating at ultrahigh pressures with variation between 15 and 20 MPa (150 and 200 bar) during the cycle. GBES features should keep costs low, but preparation of authoritative estimates is not a priority at this stage. The figures in Table 4.1 are tentative (back of envelope plus w50% contingency) and intended to illustrate the progression in moving to higher levels of stored energy discussed in Section 3.2.3. Moving from day-ahead to weekahead applications is virtually cost-free. Longer duration strategic storage incurs considerable additional cost, but the dollar per kilowatt-hour metric continues to fall sharply.

Novel hydroelectric storage concepts


Capital cost estimates require caution, but piston storage economics seem likely to be acceptable in proven day-ahead markets. Indeed, piston storage could out-compete PHES, if necessary. More importantly, piston storage is a potential frontrunner in areas where PHES cannot be used, and where competition is perceived as being much weaker. There are provisos, of course: piston storage has yet to demonstrate its technical and economic prowess; other emerging technologies may set tough new benchmarks in dayahead markets; and smaller-scale storage is becoming much more prominent. Piston storage is a promising candidate, not a shoo-in, but it has a couple of extra shots in its locker, when there is value in going beyond the energy needed for dayahead markets. (1) Even if we are mistaken and piston storage holds no cost advantage over other contenders at 10 h capacity, the marginal cost of adding energy storage is inherently very low and should be far less than for other electricity-in/out storage technologies. (2) Asymmetric charging enables networks with highly variable production to capture large surpluses (well in excess of the power delivery rating of the storage unit).

Capital costs are important, but the overall cost and value of the storage service are crucial. As well as the return required to recover the investment, additional factors include operation and maintenance costs, input energy costs (including round-trip losses), annual utilization, and the value of the output energy.


Finance cost

In its simplest form, the finance cost (Cf, in terms of dollars per unit of energy) depends on the specific capital cost (C in terms of dollars per unit of power), utilization (U which is the equivalent annual power delivery time at the rated output), and the required rate of return (r expressed as a percentage) according to: Cf ¼ C r=U As an example: assuming C ¼ 2000 $ (kW)1; r ¼ 6% a1 (where a is annum), and U ¼ 3000 h/a then Cf ¼ 40 $ (MW h)1 In practice, adjustments may result in a lower figure to take account of the tax position or specific incentives for this type of investment, or the fact that storage is said to increase the value of other assets in the network’s portfolio. Against this, it should be noted that the utilization is far higher than the norm for storage at present (w1000 h/a), relying on a projection for long-duration storage and asymmetric charging in networks with a high proportion of variable renewable resources.


Operation and maintenance (O&M) costs

O&M costs for PHES are among the lowest of all storage technologies, typically around 5 $ (MW h)1 for high power installations. Ignoring special issues for the moment, piston storage conditions bring benefits that might reduce PHES O&M costs (w3000 h utilization, higher average rating, use of standard designs and subsystems, variable speed technology).


Storing Energy

In practice, this advantage will be offset by seal subsystem maintenance and other deep level work. While 5 $ (MW h)1 is a reasonable target, it would be prudent to budget for a higher figure (say 10 $ (MW h)1).


Energy costs

For a storage system with 80% round-trip efficiency, the input energy is 1.25 times the delivered energy. Input energy costs can be calculated in several ways. ➢ One approach is to operate the storage system as a network asset in which participants can park surplus energy and share revenue from subsequent energy sales, reserve services, etc. The input energy could be treated as having zero value, since it is being parked rather than sold at that stage. Other essential participants (e.g., transmission and distribution operators, government agencies and those financing the system) would also be included as beneficiaries. An agreement would account for energy and other costs as part of the equitable distribution of revenue. ➢ Alternatively, the storage system can be regarded as an independent entity, with input costs based on the market price of electricity at the time it is used to charge the system. Storage can take advantage of low off-peak power (there are times when it is zero or negative in some networks). This is attractive initially, but is not particularly stable; e Introducing storage increases demand and narrows the window during which low cost surplus power is available. e The sale of power from storage adds to supply and reduces the price of on-peak electricity.

At some point, it becomes uneconomic to invest in further storage capacity (other than for low-energy reserve and regulation services). This will fall short of the optimum capacity in terms of overall benefit to the network (independent storage requires a production surplus to drive down the market price of input electricity and a subsequent supply deficit to drive up the price of peak power). A totally competitive storage market is unlikely to be in the best interests of the sector, or of the wider electricity supply industry. However, if this happens, piston storage could be the biggest bully on the street: it can afford an input energy price that would put some competing storage technologies out of business; asymmetric charging enables it to grab more than its fair share of input energy during shorter intervals of low prices; higher capacity systems (>20 h say) will be the only buyers left when prolonged periods of high renewables production exceed the charging capacity of day-ahead storage. ➢ A third option is to base input energy costs on an estimate of the levelized cost of electricity (LCOE). This gives stability, but introduces anomalies regarding the “correct” LCOE for input energy. The future storage market will be closely linked to the management of variable renewables, which will be the principal source of charging energy. LCOE estimates for solar and wind energy have fallen rapidly and, for our purposes, 30 $ (MW h)1 is probably a reasonable figure for input energy. This will vary widely, depending on local wind and solar conditions, policy, and any allowance for T&D delivery costs to the storage sites. LCOE costs per MW h in regions with the best PV conditions are heading toward $10; in others the renewables figure will remain above $50. Clearly, a low LCOE is a great benefit. However, it affects the economic ranking of storage technologies. In particular, it can bring inexpensive technologies with low round-trip efficiency into contention (even though this often creates appreciably higher costs elsewhere in the system).

Novel hydroelectric storage concepts



Overall storage costs

At its simplest, the cost of providing the piston storage service is the sum of the costs of finance, O&M, and input energy. This might put the levelized cost of storage (LCOS) at around 90 $ (MW h)1, comprising: Finance cost O&M Input power Energy losses

w40 $ (MW h)1 w10 $ (MW h)1 w35 $ (MW h)1 w5 $ (MW h)1

A storage service costing w60 $ (MW h)1 (on top of the input energy price) would probably be seen as viable by most networks, especially as it does not take account of other revenue opportunities, such as provision of reserve and regulation services. Other analyses are more ambitious. An estimate on Gravity Power’s website [7] gives the output LCOS as 76 $ (MW h)1 for their 4-h, GW-class design, based on 40 $ (MW h)1 charging power and 67 $ (MW h)1 for a 10-h intermediate system. Heindl Energy goes further still, estimating that it would take just 4 years to recoup the investment in a much higher energy GW-class system [10].

3.4 3.4.1

Markets for piston storage Size matters, or does it?

Piston storage will typically target power ratings >100 MW (up to multi-GW) and energy ratings >10 h at rated delivery power (up to >100 h). This puts it at the top of the power/energy spectrumdin fact, it goes well beyond the present electricity storage envelope. There isn’t much advantage to be gained from output ratings above the 3 GW already achieved by PHES, but ultrahigh-energy/long duration storage should be very important in future (and will remain difficult to achieve for many other technologies). Competition will also come from much further afield. Storage services are delivered by wire and users requiring high-power, high-energy storage can get it from a large number of low-power, high-energy storage systems or, for that matter, from an even larger number of low-power, low-energy units. Low-power technologies (100 h storage capacity)

Compared to week-ahead systems, the main additional objective is to harvest energy that would otherwise be lost from the electricity supply system. This is not much of an issue at present, but will become prevalent in networks where variable renewables are the principal energy source. There is no doubt that strategic storage would be extremely useful and there is likely to be less competition from other electricity in/out technologies at this level. As noted in Section 3.2.3, a 600 GW h strategic store equates to about half the world’s electrical energy storage capacity. It is inconceivable that half the present global PHES population would be installed in California or a large European country or a Chinese province, where 600 GW h (or more) strategic storage might be appropriate. This could be an attractive application for piston storage. The key questions are whether the cycle count and cycle value of the additional energy capacity is sufficient to justify its cost; and how this will mesh within the mix of measures available, such as long-distance interconnectors and power-to-X technologies. Low marginal cost helps, as does asymmetric charging, which can radically improve utilizationdultra-marathon runners would be at a huge disadvantage if their “recharging rates” at feeding stations were restricted to the rate at which they burn energy while running, especially if the energy gels are taken off the table every time the wind drops! Opportunities for strategic (and possibly seasonal) storage will probably be confined to large supply networks. However, piston storage should greatly expand the envelope within which it makes sense to retain electrical energy within the network. It opens prospects for electrical energy storage at much higher energy levels than are envisaged with today’s technologies, and in networks that have made little or no use of storage in the past.

3.5 3.5.1

Competition to high-energy electricity storage Competition or collaboration?

There is no golden-bullet solution to achieve low-carbon energy targets, but it is easy to devise approaches based on sensible combinations of measures. These will require a

Novel hydroelectric storage concepts


degree of collaboration and acceptance of priorities. However, meeting targets by 2050 is likely to involve investment of >$100trillion and it will be no surprise if special interests or particular technologies are promoted vigorously, without much regard to their overall efficacy. Electricity storage is assured of an important role within future networks. There is no assured place for piston storage, but it is an interesting candidate within the highenergy/long-duration sector, especially in networks that are heavily dependent on lowcost variable renewables. In general terms, this sector will help tackle problems beyond the scope of higher-priority measures such as load management, interconnectors, and fast-response/short-duration storage. Piston storage will have to interface with these; with other long-duration electricity storage technologies; and with alternatives, such as thermal storage and power-to-X.


Short-duration storage

Networks require both short- and long-duration storage. In an ideal world, technologies such as piston storage should be used in a supporting role in short-duration markets (collaboration rather than active competition, where long-duration storage would often win). As noted in Section 3.4.1, 1-h storage units will seldom be relevant in 50-h applications. There is an important exception. There will be large populations of battery electric vehicles with w100 kW h batteries and w500 km range, typically requiring the equivalent of only 20e30 full charges annually. If the batteries were managed by the electricity supply industry (ESI), the state of charge of the fleet could be adjusted to suit forecast periods of low or high renewables production. In effect, the fleet would simulate a w100-h battery. A tentative assessment [11] suggests that this might increase or decrease average demand by around 5% during periods of prolonged surplus or deficit, making inroads into issues that might otherwise be addressed by long-duration storage (load management is usually the most effective of all balancing measures). This only applies if the ESI actively manages a substantial part of EV battery charging. Private charging will tend to exacerbate the problem as users seek to fully charge their batteries at times when there are supply shortages. Vehicle-to-grid (V2G) operation, feeding energy back to the network, could double the cycle count for a typical EV battery [11]. This is more likely to interact with (and reduce the requirement for) short-duration storage, with less effect on long-duration prospects.


Hydrogen and power-to-X

There is great interest in the concept of a green hydrogen economy, using renewables or other clean electricity to produce hydrogen on a massive scale. This can be stored and used on-demand for heating, transportation, industrial processes, distributed generation/cogeneration, etc., and also in larger generating plant at times when there is insufficient renewable power to meet network requirements. It provides a clean


Storing Energy

substitute for much of the present oil and gas sector, adapting its distribution structure to serve traditional loads and solve intermittency issues, with little need for load management or electricity storage. This sounds attractive and is supported by influential energy majors and their allies. Unfortunately, it would result in a dramatic increase in the costs of a clean energy economy. The main problem is that accomplishing a task that requires 1 MW h of electrical energy usually requires about 3 MW h of fuel energy. And the overall difference is even greater, since it takes roughly 4 MW h of electricity to produce 3 MW h of green hydrogen (or another electricity derived fuel). To put this in context, it might take 100,000 TW h to run an electricity-led world in 2050, plus a margin (say 10%e20%) to allow for year-on-year fluctuations and/or to produce fuels for applications that cannot be electrified. The hydrogen economy route would require a massive increase in electricity production (>250,000 TW h), if we were to use hydrogen-fuelled alternatives in place of half the electrical loads. Some companies and governments thrive in a high-cost energy environment, but this does not seem to have much merit, even in the minority of countries that already have natural gas infrastructures that might be adapted to support widespread hydrogen use. There are some savings from lower-rated electricity networks and less need for balancing mechanisms (including long-duration electricity storage), but this does not offset much higher investment and energy costs [12].


Thermal energy storage

Thermal energy is used as a storage medium in some electricity-in/electricity-out technologies, which will be direct competitors in the long-duration sector, as covered elsewhere. As far as this chapter is concerned, thermal storage can also be extremely effective in load-leveling/time-shifting where electricity is the energy source in heating and cooling applications. Indeed, thermal storage is usually a better solution in longduration, high-energy situations. The thermal medium is usually hot water or ice. Energy densities are quite low, but containment of large volumes is inexpensive. Heat losses (or gains for ice storage) are usually acceptable, especially for the large-scale systems with energy capacity comparable to piston storage. This, together with the low capital cost of thermal storage, means that seasonal storage can be feasible, storing >1000 h of energy at rated output and making economic sense with just one or two cycles a year. Modern district heating and cooling (DHC) schemes illustrate the potential value of thermal storage in an electricity-led energy system [13]. Benefits of scale mean that the large MV (medium-voltage) heat pumps and chillers in DHC schemes have much higher efficiency than individual units in residential and commercial premises. The input electricity for these heat pumps would be about one-fifth that needed to produce clean hydrogen for boilers doing the same job. DHC also displaces heating and air conditioning loads from the LV distribution networkd advocates of gas or hydrogen heating argue that these networks would require a

Novel hydroelectric storage concepts


radical upgrade, if boilers were replaced by heat pumps. Benefits of adding thermal storage to the DHC system include: w10 h of storage gives the electricity supply industry another large load management tool, avoiding operation at peak periods. w100 h of storage takes this further, scheduling operation to run down storage before a forecast period of surplus renewables, or maximizing storage so that it can be the primary source for a week or so during a forecast period of renewables deficit. w1000 h of storage would help in mid-latitude territories with large seasonal swings. Storing a million tonnes of ice by pumping heat from cold winter water and rejecting the heat into cold winter air should require

ð p1  p0 Þ R  smax 2



Figure 6.6 Stress calculations for thin-walled tanks (A) spherical and (B) cylindrical.


Storing Energy

The minimum volume of material required for the sphere wall is then found to be proportional to the volume contained within the sphere by Vwall > V1 

ðp1  p0 Þ 3  smax 2


A similar analysis can be conducted for a long cylindrical thin-walled tank radius R and internal pressure p1 . From Fig. 6.6B, this must have a wall thickness t obeying t>

ð p1  p0 Þ R smax


and substituting this thickness back to assess the volume of material required for the wall returns Vwall > V1 

ðp1  p0 Þ 2 smax


Not surprisingly, a comparison of Eqs. (6.16) and (6.18) reveals that the spherical tank makes best use of material. Since the two equations do not differ by much, it is conservative and reasonable to apply Eq.(6.18) as a general lower bound and to remove any constraint on the shape of the tank. It could comprise, for example, a long coil of very thin-walled tube. Having knowledge of the volume of material required to construct a tank is the first step in estimating how much it might cost, and it also provides a very good estimation of the weight of the tank. A brief calculation provides some very interesting insight. Suppose that we arrange a compressed air tank to contain 1 m3 of air at 200 bar and that we will allow the internal pressure to fall to 2 bar. Consider that this tank is to be constructed from steel with maximum allowable stress of 1000 MPa and density 7800 kg/m3. From Eq. (6.13) (with T0 ¼ T1 ), the exergy stored in this tank is 86 MJdabout 24 kW h. For an ambient temperature of T0 ¼ 300 K, the mass of air in the full tank is 232 kg and the mass of the tank wall itself is 155 kg bringing the complete tank mass to 387 kg. A very good battery might presently achieve an energy density of 200 W h/kg and to store 23.9 kW h in that battery would demand a mass of around 120 kg. Evidently, compressed air stored in tanks delivers an energy density that is lower than that of present-day batteriesdbut not an order of magnitude lower. Based on a present-day (2021) rough assessment of a typical cost of battery-based energy storage at $150/(kW h), this energy store might justify spending w$3600. This would correspond to w$23/kg of steel. With an appropriate manufacturing route and tank design, such a value is easily achievable.


The case for underground or underwater storage

Section 3.3 makes the point that the cost of storage in pressurized tanks at or near the surface is almost directly proportional to the volume of the storage and to the pressure

Compressed air energy storage (CAES)


50 45 40 35 30 25 20 15 10 5 0





Cavern size / 100 000 m3



Figure 6.7 Why underground storage becomes interesting at large scale. The y-axis is either cost/£106 or cost per unit/£ (kW h)1.

difference between storage pressure and ambient pressure. This is not the case for stores developed either underground or underwater. In both cases, there may be a relatively high fixed cost for causing any such air store to exist, but the marginal costs of increasing the capacity of that store can be very small in both cases. Fig. 6.7 illustrates this point with notional values for the case of a (fictional) cavern. A similar phenomenon happens in the context of underwater stores even though the cost per unit of energy stored tends to plateau at smaller values of stored volume.

4. System configurations and plant concepts CAES actually represents a broad family of technologies embracing: • • • • • •

Numerous varieties of air compression machinery Numerous varieties of air expansion machinery Heat exchangers for cooling air between/within/after compression stages Heat exchangers for heating air between/within/before expansion stages Several options for storage of pressurized air Several options for thermal storage to complement the stored pressurized air

Obviously there is a very broad set of possible configurations that would all fall under the aegis of CAES systems. Before highlighting some of these, it is useful to discuss one of the most commonly used classifications within CAES systems. This comprises • •

adiabatic CAES systems diabatic CAES systems


Storing Energy

The two classes are distinguished by the fact that an adiabatic CAES system does not use any external source of heat while a diabatic CAES system does make use of some external heat source to extract additional work (electricity) from the stored high pressure air. Strictly, a diabatic CAES system is a combination of an energy storage system and a generation system, since it is normal that the total amount of electricity generated from such systems exceeds the total amount of electricity consumed.


Diabatic concepts

There are only two large-scale CAES plants in the world in operation at present: a 321 MW plant belonging to E.ON Kraftwerke, Huntorf, Germany, and the 110 MW plant of PowerSouth Energy Cooperative in Alabama, USA. Both plants use underground salt caverns for storing the air. Both are diabatic insofar as they do not store the heat of compression but they do use a fuel to reheat air prior to one or more stages of expansion [8]. Fig. 6.8 below presents the simplest format of a diabatic CAES plant. Here the same electrical machine (labeled M/Gen to indicate that it can act as either motor or generator) can be coupled either to the set of compressor stages or to the expander via the clutches shown in Fig. 6.8. An alternative to having these two clutches is to duplicate the electrical machine so that one machine always acts as a motor when it is in operation and the other always acts as a generator. The exhaust gas in Fig. 6.8 may exit at a temperature substantially above ambient and in this case, it is clear that exergy is being wasted. An alternative system that does not waste any exergy is shown in Fig. 6.9. Here, a recuperator absorbs heat that is left in the exhaust gas leaving the (final stage of the) expander and transfers this heat to air coming from the high pressure air store before it reaches the (first stage of) expansion. It is not essential that the expansion process should comprise only a single stage of expansion. By having more than one stage (together with a recuperator), it is possible to extract substantially more work from the stored air. Fig. 6.10 shows the schematic for such an arrangement. Fig. 6.11 shows the temperature-pressure profile for the air in the case of Fig. 6.10. In this, the recuperator raises the temperature of the air leaving the tank to 800 K by cooling the air exhausted from the second expansion stage back down to ambient temperature. A combustor then adds further heat to the air entering the first stage of Compressor Stages Clutch

Clutch M/Gen

AIR TANK Inter-coolers

Figure 6.8 A simple diabatic CAES system.

Turbine Exhaust Fuel In Combustor

Compressed air energy storage (CAES)



Compressor Stages

M/Gen Air Tank



Recuperator Exhaust air out

Fuel In

Figure 6.9 A simple diabatic CAES system with recuperator.

H.P. Turbine

Compressor Stages

L.P. Turbine


AIR TANK Inter-coolers


Recuperator Exhaust Gas

Fuel In

Figure 6.10 A diabatic CAES system with recuperator and two-stage expansion.

expansion to raise its temperature to 1585 K. After the first expansion stage, a second combustor raises the temperature again to 1585 K before the second expansion stage which again reduces the temperature to 800 K.


Adiabatic concepts

The term adiabatic suggests that no heat is drawn into a process or expelled from that process. Strictly, the term is misused in the context of Adiabatic CAES systems. The real meaning is that no net external heat source is used. Heat may be exchanged with the environment, however. Fig. 6.12 shows one of the simplest possible adiabatic CAES system structures. In this, multiple stages of compression with interstage cooling approximate an isothermal compression process. Similarly, multiple stages of expansion with interstage reheating from the environment approximate an isothermal expansion process. Fig. 6.13 depicts another possible adiabatic CAES system structure. In this, one single compression stage raises the temperature and pressure of air. Heat is drawn from the air and stored in a thermal store before the air is fed into storage. The same heat is injected back into the air when the air is withdrawn from storage prior to expansion.


Storing Energy



Figure 6.11 Temperature-pressure plot for a diabatic CAES system with recuperator and two-stage expansion.


Turbine Stages

Compressor Stages M/Gen



Figure 6.12 A straightforward Adiabatic CAES plant.

If the same size and type of air store is used for the systems of Figs. 6.12 and 6.13, the total storage capacity of the latter system will be much larger since exergy is stored in the thermal store as well as in the compressed air store. It is commonly stated that the purpose of introducing thermal storage into compressed air energy storage is to improve efficiency. This is quite incorrect. Systems such as that depicted in Fig. 6.12 can be made arbitrarily efficient by using a sufficient number of highefficiency compression and expansion stages and by demanding high effectiveness

Compressed air energy storage (CAES)

135 The Grid



Motor/ Generator Air Intake


Thermal Store


Compressed Air Store Figure 6.13 An Adiabatic CAES plant with thermal storage.

of the heat exchanger unitsdthough they would become very expensive. The main motivation for introducing thermal storage to a CAES plant is to increase the total quantity of exergy that can be stored for a given size of pressurized air store. A clever concept by G. L. Guidati [9] teaches that by using a recuperator in an adiabatic CAES system, the proportion of exergy stored in the form of heat can be increased, and this can improve the economics of the energy storage system, since it is generally less expensive to store exergy in the form of heat than to store it in the form of pressurized air in tanks. What limits the proportion of exergy that can be stored as heat is the maximum allowable storage temperature as explained in Ref. [9]. Typically, temperatures of up to 600 C may be achieved without the use of very exotic materials and in these cases approximately two-thirds of the stored exergy may be retained in the form of heat.

5. Thermal storage for CAES We have seen in the previous section that CAES installations may be either diabatic or adiabatic. In both cases, air is preheated prior to the expansion processdlargely for the purposes of extracting more work from the same quantity of compressed air but partly motivated by avoiding very low temperatures that could cause problems of water or lubricant freezing. In the case of adiabatic CAES plant the heat is stored from the air compression and used again to support expansion. Thermal energy storage is itself a very large discipline, but its importance is so great in the context of CAES systems that it demands at least some discussion here. The


Storing Energy

requirement is to be able to transfer heat in and out of a pressurized air stream. If compression and expansion are both single-stage processes, then there is only one pressure to consider for thermal storage. If they both take place in, say, three stages, then there are three separate pressures to be handled by the thermal storage. For reasons connected with minimizing the destruction of exergy by forcing heat transfer to take place across large temperature differences, a well-designed adiabatic CAES plant will always use the same number of stages for compression and expansion. It takes only a little consideration to realize that there are two major options for thermal storage in conjunction with CAES [7]: (i) Heat is stored within the pressurized system and the high pressure air circulates around it (ii) Heat is stored outside of the pressurized system and transferred across the walls of that system

Fig. 6.14 makes the contrast clear. In Fig. 6.14A, a pressurized containment contains the thermal storage mediumdshown as bricks. In such cases, the heat transfer between the pressurized air and the thermal storage medium may be extremely good and one may achieve very good thermocline behavior (as shown in Fig. 6.15) where a sharp thermal gradient exists between a portion of the thermal store that is hot and another part that remains cool. In Fig. 6.14B, by contrast, a pipe containing pressurized air is shown running through a thermal storage medium that is not pressurized. Most often this arrangement comprises a heat exchanger external to the thermal store and a heat transfer fluid (that may be air or a liquid thermal storage medium such as molten salt) carries heat between the main thermal storage volume and the pressurized air. However in some cases (as indicated schematically in Fig. 6.14B), the heat exchanger function may be integrated with the thermal storage functiondas described in Ref. [10]. The main disadvantage to the first option (Fig. 6.14A) is that one requires a large pressure vessel that may not be able to exploit any geological/geographical features for strength. One rather elegant solution is noted in Ref. [11] where the thermal energy is stored within the same cavern used for the pressurized air but this is not applicable in most situations.

Figure 6.14 (A) Heat stored within pressurized system, (B) Heat stored outside approximately.

Compressed air energy storage (CAES)


Figure 6.15 A thermoclinetype packed-bed heat store.

If the material within was some rock with specific heat 850 J/(kg K) (typical of a wide variety of rocks) and if the temperature swing in that rock was, say, 500 K, then each 1 m3 volume of space within the containment would account for w640 MJ of heat (assuming that the mass of rock in each 1 m3 volume is 1500 kg). Note that this value is independent of pressure. Comparing this quantity of energy with the values in Table 6.3 shows that the energy density is quite good relative to the energy stored in the compressed air itself for all realistic storage pressures. However, the pressure containment is still expensive and is made more complex by the fact that this containment may have to operate at quite high temperatures (further weakening the containment material). The main shortcoming of the nonpressurized thermal storage is that heat transfer is much more indirect between the pressurized air and the thermal storage medium. Work is underway on varieties of nonpressurized thermal storage wherein the pipework of the heat exchangers itself forms a substantial fraction of the thermal mass used for storing the heat e as trailed in [12].

6. Performance metrics for CAES It is commonly asserted that CAES does not deliver a high-performance energy storage solution. In fact, this position is mainly derived from an erroneous assessment of the performance of the two large-scale diabatic CAES plantsdat Huntorf in Germany and McIntosh in Alabama, USA. Published data (e.g., Ref. [7]) reveal that to achieve 1 kW h of electrical energy output, 0.8 kW h of electricity is drawn in to the plant at Huntorf together with 1.6 kW h of gas. Similarly, to achieve 1 kW h of electrical energy output from the CAES plant at McIntosh, 0.69 kW h of electricity is drawn in together with 1.17 kW h of gas. Performing a straight “output-divided-by-input” calculation suggests that


Storing Energy 1 1 hHuntorf ¼ 0:8þ1:6 ¼ 41:7% ✗ and hMcIntosh ¼ 0:69þ1:17 ¼ 53:8% ✗

These values are unrepresentative because they fail to recognize that every diabatic CAES plant is really a combination of a pure energy storage plant and a generation plant. The above calculations add electrical energy to thermal energy as though 1 kW h of heat was equivalent to 1 kW h of electricity. This is wrong. In fact, the best performing combined-cycle generation plant returns 60% of the calorific value of the fuel consumed. On this basis, we can obtain much more representative values for the performance of the two long-extant CAES plants. 1 hHuntorf ¼ 0:8þ0:61:6 ¼ 56:8% U and hMcIntosh ¼ 0:69þ

1 0.61:17

¼ 71:8% U

The reason that McIntosh performs rather better than Huntorf is connected mainly to the recuperator. Note also that both Huntorf and McIntosh presently use throttles to deliver constant pressure air to the expanders. The alternative would be to use a small high-pressure reciprocating machine or a number of dynamic machines in series to extract exergy as the air pressure is reduced. Employing Eqs. (6.13) and (6.14) and the knowledge (from Ref. [7]) that Huntorf caverns operate between 5 MPa (50 bar) and 7 MPa (70 bar) indicate that Huntorf loses 4.4% of the exergy available from the cavern in each cycle. Since that exergy comprises around 43% of the total exergy including that from fuel, the improvement that could be achieved by replacing the throttle at Huntorf is around 1.9% (i.e., raising its effective turnaround efficiency to 58.7%). The plant at McIntosh runs from 4.5 MPa (45 bar) to 7.6 MPa (76 bar), and thus it loses 6.6% of the available cavern exergy in the nozzle. Moreover, the cavern exergy there comprises around 47% of total exergy and hence 3.1% improvement in effective turnaround efficiency is possibledraising this to 74.9%. Other measures can be taken to achieve still higher performance but, of course, all have associated costs.


Integrating CAES with generation or consumption

Because CAES intrinsically involves both heat and mechanical work, there are strong opportunities for integrating this with either electricity generation or consumption. The basic arguments for doing this are straightforward. Standalone energy storage for supporting the electricity system requires that energy is converted from the form of electricity to another form compatible with storage and back again, causing two additional sets of energy losses in transformation and requiring additional machinery to effect the transformations. By integrating energy storage with generation, these additional losses and costs may be reduced or avoided [13]. Various proposals have come forward for integrating compression with wind turbines [14,15] and for integrating compression with wave energy [16] and tidal energy [17,18]. For generation from wind, wave, and tides, the conversion efficiency from input energy to electricity is virtually irrelevant and what matters is the combined capital and operational costs

Compressed air energy storage (CAES)


of the energy harvester devices per kWh of electricity produced. For sound engineering reasons outlined in Ref. [15], exploiting air compression as the primary means of carrying away power from the device can account for significant cost reduction compared with direct generation of electricity. It is to be expected that as more CAES plant will emerge in the near future, there will be an increase in interest in capturing compressed air directly from renewables. CAES also provides for integration at the consumption side. We noted at the outset that substantial fractions of the electricity supply are presently used to compress air [1,2], and it follows that if suitable receivers are in place at those locations that use compressed air, these may act as excellent energy storesdobviating the requirement for any new power conversion machinery and avoiding virtually all losses normally associated with storage. One other possibility deserves particular mention: data centers are progressively consuming more and more of all electricity generated and these have very particular requirements for both cooling and power that are especially well suited to CAES. If high-pressure air is stored at near-ambient temperature and then expanded in several stages, it can deliver both cooling and electrical power in equal measured fitting exactly the requirements of a data center.

8. Concluding remarks The application of air compression to decouple energy absorption from the grid and energy consumption is known and has been practiced for decades. Two large-scale CAES power plants are in operation today, using salt caverns as storage, and both of these burn a fuel to maximize the energy recovered from the stored air. There are many different possible configurations of CAES system and most of those presently proposed do not combust a fuel. A CAES plant can be highly cost-effective and they can deliver very respectable turnaround efficiencies. The containment for the high-pressure air is at the heart of every CAES system. Storage of air in manmade tanks has been discussed in this chapter, but this comes at a cost per unit of energy stored that may be one or even two orders of magnitude higher than what may be achieved with underground or underwater storage of the air. For this reason, separate chapters are devoted to these two possibilities.

References [1] Carbon Trust, Compressed Air: Opportunities for Businesses. https://www.carbontrust. com/media/20267/ctv050_compressed_air.pdf. [2] DECC (Department of Energy and Climate Change, UK), Energy Consumption in the UK, 2015. [3] M. Budt, D. Wolf, R. Span, J. Yan, Appl. Energy 170 (2016) 250e268. [4] H. Auinger, Power Eng. J. 13 (1) (1999) 15e23.


Storing Energy

[5] S. Carnot, Réflexions sur la puissance motrice du feu et sur les machines propres a développer cette puissance (in French), first ed., Bachelier, Paris, 1824 (Reissue of 1878). [6] E. Barbour, A Novel Concept for Isobaric CAES. Offshore Energy Storage Conference OSES2014, University of Windsor, Windsor, 2014. [7] S.D. Garvey, Compressed air energy storage: performance and affordability, in: Half-day Workshop Given at Marcus Evans Biannual Energy Storage Conference, Amsterdam, Dec 1-3, 2010. Copy of presentation available at: [8] BINI Informationsdienst, Compressed Air Energy Storage Power Plants. ProjectInfo 05/ 2007. ISSN 0937-8367. [9] G.L. Guidati, Energy Storage System and Method for Energy Storage, Patent Application EP2687702A1. [10] B. Cardenas, A.J. Pimm, M.C. Simpson, J.E. Garvey, S.D. Garvey, Propuls. Power Res. 6 (2) (2017) 126e133. [11] [12] S.D. Garvey, B. Kantharaj, B. Cardenas, J.E. Garvey, M.C. Simpson, P. Codd, T. Davenne, A.J. Kitchener, A novel approach to thermal storage for CAES systems, in: Offshore Energy Storage Conference OSES2017, Cape Cod, MA, 2017. [13] S.D. Garvey, P.C. Eames, J. Wang, A.J. Pimm, M. Waterson, R.S. MacKay, M. Giulietti, L.C. Flatley, M. Thomson, J. Barton, D.J. Evans, J. Busby, J.E. Garvey, Energy Pol. 86 (2015) 544e551. [14] E. Ingersoll, Wind Turbine System, Patent Application US20080050234-A1, 2008. [15] S.D. Garvey, Proc. IMechE, Pt. A, J. Power Energy 224 (5) (2010) 1027e1043. [16] J. Sieber, Wave Energy Accumulator, Patent Application US2009226331-A1, 2009. [17] O. Fumio, Transducer for the Conversion of Tidal Current Energy, Patent Application US4071305-A, 1978. [18] A.G. Southcombe, Wave or Tidal Power Harnessing Apparatus, Patent Application GB2267128-A, 1992.

Compressed air energy storage Sabine Donadei and Gregor-Sönke Schneider DEEP.KBB GmbH, Hannover, Germany


1. Introduction Unlike fossil fuels, renewable energy sources such as wind and solar are characterized by short-term and long-term seasonal fluctuations, and cannot deliver energy on demand. Moreover, only a very small amount of the electrical energy they generate has so far been stored. Compared to the storage of fossil fuels, the storage of electrical energy at a grid scale has so far only played a very subordinate role. In Germany, for instance, electrical storage can only maintain the power in the whole of the German grid for less than 1 h. Storage of electrical energy at a grid scale is gaining in significance because of the increased use of fluctuating renewables. In addition to pumped hydro technology, which has proven its worth over many decades, and future hydrogen systems (power-to-gas), attention is again being focused on a storage technology which was developed over 50 years ago: compressed air energy storage (CAES) [1e3]. The use of compressed air to store energy is currently deployed in applications ranging from very small outputs up to triple-figure megawatt installations. In this chapter, the focus is on energy storage at a grid scale comparable to conventional pumped hydro power plants, which means that the following chapters are restricted to the investigation of the MW class. The concept of large-scale compressed air storage was developed in the middle of the past century. The first patent for compressed air storage in artificially constructed cavities in deep underground, as a means of storing electrical energy, was issued in the USA in 1948. Frazer W. Gay, the patent holder, described his invention as follows: “In the present invention, I propose to provide equivalent storage space for gas relatively close to the earth’s surface and, furthermore, to make this storage space available for the storing of compressed air to be used for power generation purposes during periods of heavy power load, as well as for natural gas or manufactured gas, butane, propane or other fluids. The invention in general comprises the construction of huge caverns located comparatively close to the earth’s surface” [4]. In Germany, a patent for the storage of electrical energy via compressed air was issued in 1956 whereby “energy is used for the isothermal compression of air; the compressed air is stored and transmitted long distances to generate mechanical energy at remote locations by converting heat energy into mechanical energy.” [5]. The patent holder, Bozidar Djordjevitch, is sometimes quoted as the inventor of compressed air

Storing Energy. Copyright © 2022 Elsevier Inc. All rights reserved.


Storing Energy

energy storage technology. His concept was supplemented by the former Federal Institute for Geology (today’s BGR) which chose cavities in underground rock salt as the storage facilities for the air because these were considered to be the most economical and the safest [6,7]. Compressed air energy storage power plants attracted interest in the USA at the end of the 1960s with the focus in this case on air storage. This led to the discussion and patenting of compressed air energy storage systems with salt caverns and aquifer structures [8,9]. Several studies and projects on compressed air energy storage arose in Europe in the subsequent years. Salt caverns, aquifer structures, and mines were investigated and taken into consideration as potential storage spaces. The world’s first CAES power plant was constructed in Huntorf, Germany, in the middle of the 1970s, and was primarily aimed at storing the electrical energy produced by less flexible coal and nuclear power plants during low periods of demand, and to feed this energy back into the grid again during periods of high demand. Other motives for constructing the plant were the cold start capacity, its ability to regulate the grid frequency, and the phase shifter operation [10]. Another CAES power plant was constructed in McIntosh, Alabama, USA, in 1991. After the expected demand for additional CAES power plants evaporated as a result of the merger of smaller grids to form larger shared grids, interest was reawakened at the beginning of this century by the transition from fossil fuels to renewables. This was initially stimulated by the growing demand for minute and hour reserves in the power grid to balance out deviations between forecast and actual wind energy generation. The growing demand for flexibility again focused attention on CAES power plants because of their analogous properties to pumped hydro plants. Attention is now concentrated particularly on the development of new power plant components.


Mode of operation

In CAES power plants, electrical energy from the power grid drives a compressor to inject large volumes of air under high pressure into a storage facility. When electricity is required, this air can be released from the storage and passed through a turbine and generator to regenerate electrical power which can be fed back into the grid. The heat energy generated by compression is either lost to the environment or made available to other users or stored for later deployment. Before being fed back into the turbine and generator installation, the air is warmed back up again by using heat from this thermal storage, or a different source, or from combustion gases (recuperation). A CAES power plant consists of a storage space for the air, and a power plant with motor compressor and turbine generator units. Although the storage of compressed air on the surface is possible, e.g., in spherical and pipe storage systems, or in gasometers, these have much lower storage capacities compared to underground storage systems. Installation concepts at a grid scale therefore usually depend on the underground storage system. Because these underground geological storage systems must be injected

Compressed air energy storage


Figure 7.1 Diagram of a diabatic CAES power plant [11].

with cool air (400 m within the territorial waters of each country in Europe, categorized by distance from shore. Almost all of Norway’s vast areas of near-shore deep water are found within the fjords. Aside from Norway, it is typically the case that island nations have the largest areas of deep water close to shore.

6. Cost and efficiency With suitable pipework, the round-trip efficiency of an underwater CAES plant can be very similar to that of an underground CAES plant. The effective round-trip efficiency of a CAES plant was discussed in Chapter 5. It is not unusual for compressors to achieve adiabatic efficiencies of 88% and for turbines to achieve efficiencies above 92%, so the basic elements of the system point to a round-trip efficiency in excess of 80% being achievable.


Storing Energy

Table 8.1 Areas of water within European territorial waters with depth greater than 400 m, categorized by distance from shore. Area with depth >400 m/km2 Country

0e22.2 km

0e10 km

0e5 km

0e2 km

0e1 km

Norway Greece Azores Canary islands Italy Turkey France Madeira Georgia Gibraltar Portugal Russia (to 65 E) Cyprus Spain Monaco Albania Ukraine Malta Iceland Croatia Totals

4787.19 76,346.16 19,561.23 23,340.65 37,577.86 22,813.23 7934.15 9141.76 2123.36 231.24 1583.32 3173.24 10,173.09 12,645.34 54.33 585.17 1182.22 970.83 411.76 107.13 235,783.68

2692.07 22,473.24 5746.84 5746.81 8460.96 5385.08 1887.44 2444.39 512.82 75.90 159.82 269.55 2593.82 1039.71 14.07 51.44 48.80 41.95

1810.26 5289.61 1472.01 1171.44 1663.27 725.03 430.85 513.05 99.98 17.90 21.83 23.20 445.98 67.38 1.02

733.01 440.02 168.61 51.42 139.94 27.48 25.65 19.91 3.83 0.95 0.45 0.43

149.43 28.07 27.92 16.70 15.40 2.26 1.23 0.49





Assuming that compressed air is stored at a similar temperature to the surroundings (as is the case at Huntorf and at McIntosh), the additional losses introduced by underwater storage are those associated with leakage and pressure drop. With a well-manufactured vessel, leakage losses should be small. The leakage rate in the first prototype offshore flexible vessel was estimated at 1.2% per day at the end of a 3-month deployment [16], with air loss mainly found around stitching of seams and areas of repaired trauma due to handling. It is anticipated that future deployments would benefit from improved handling procedures and more heavy-duty material, and so would not be damaged on deck. A detailed exergy analysis of generic UWCAES systems was carried out in Ref. [16a]. A subsequent analysis by Maisonnave et al. [16b] shows that by using water pumping and liquid-piston approaches to achieve the air compression and expansion, there is the potential to achieve higher turnaround efficiencies than might be achievable using compressors and expanders directly. The pressure drop in the pipework is likely to be the main source of loss in a UWCAES systemdespecially if the set of underwater containments is not located very close to the power conversion. Table 8.2 shows the pipe diameter required for

Underwater compressed air energy storage


Table 8.2 Pipe diameters required to transmit 100 MW of cool air for various percentage pressure drops [17]. Required pipe diameter, D/m Pressure/ 105 Pa (bar)

Mass flow rate/kg/s

0.05% loss per kilometer, D/m

0.2% loss 1% loss per per kilometer, kilometer, D/m D/m

4% loss per kilometer, D/m

20 30 40 50 60 70

415.2 365.7 337.2 317.9 303.8 292.8

2.182 1.761 1.518 1.356 1.237 1.145

1.648 1.331 1.147 1.025 0.935 0.866

0.904 0.730 0.630 0.562 0.513 0.475

1.192 0.962 0.830 0.741 0.676 0.626

various percentage pressure drops per km of pipe in a UWCAES system taking cool compressed air at 100 MW. Two points should be noted here: (i) the mass flow rate of air is determined here by pessimistically assuming that the air will be expanded isothermally, and (ii) the actual losses in output power are much lower than the losses in pressure. As an example of the latter point, at 2.0 MPa (20 bar), a 1% drop in pressure causes only a 0.33% drop in output power, and at 7.0 MPa (70 bar), a 1% drop in pressure causes only a 0.24% drop in output power. Table 8.3 gives the pipe masses corresponding to the diameters given in Table 8.2, assuming a 10 km length of cylindrical steel pipe with density of 7800 kg/m3, working stress of 100 MPa, and a constant wall thickness. Associated costs are also given, assuming a cost of £2000 per tonne of steel (about five times the cost of the raw material to account for welding and other working). The pressure drop has not been taken

Table 8.3 Pipe mass,m, and approximate costs assuming 10 km length of steel pipe, calculated assuming constant wall thickness from surface down to storage vessels. Pipe mass for 10 km length/t Pressure/ 105 Pa (bar)

0.05% loss per kilometre m/t

0.2% loss per kilometer m/t

1% loss per kilometer m/t

4% loss per kilometer m/t

20 30 40 50 60 70 Approx. Cost/£

11,792 11,463 11,334 11,293 11,270 11,261 23  106 (23 m)

6709 6540 6465 6449 6436 6439 13  106 (13 m)

3502 3412 3383 3368 3363 3364 7  106 (7 m)

2011 1963 1948 1937 1936 1936 4  106 (4 m)


Storing Energy

into account as to fully account for its effects on air density (and hence net buoyancy) would require knowledge of the seabed profile along the length of the pipe, which varies by location. In reality, because the highest differential pressure across the pipe would be seen by the parts closest to the surface of the sea, it would be possible to reduce the total pipe mass by using a pipe whose wall thickness reduces in stages as it reaches greater depths. Interestingly, the pipe masses shown are almost independent of internal pressure, because the required pipe diameter for a given pressure drop decreases with greater internal pressure. The only reason pipe mass is not constant for all pressures is because air density (and hence differential pressure across the pipe wall) increases with pressure inside the pipe. While the pipe masses shown are large, it should be remembered that the pipe connecting the vessels to the machinery is also a store of compressed air. The underwater pipes would need to rest on the seabed, so it is necessary to take into account the pipe buoyancy and pipe masses in order to calculate the net buoyancy of the pipes, which must be counteracted with ballast. The net buoyancy of the 10 km long pipes referred to in Tables 8.2 and 8.3 is shown in Table 8.4. Clearly higher storage pressures used in deeper water allow narrower pipes to be used for a given pressure drop, hence reducing the net buoyancy of the pipe. The existing underground CAES plant at Huntorf has the machinery house located almost directly above the caverns, so for each cavern there is only 300 m of surfacebased pipework and w500 m of downpipe. UWCAES plants with offshore machinery could have the machinery located on a platform directly above the storage vessels, and so would only require a pipe directly down to the seabed. Near-shore plants would clearly need to have a longer pipe run than offshore plants and as such would have greater transmission losses. UWCAES has an advantage over underground CAES in that the pressure characteristic in an underwater CAES vessel is roughly isobaric. This contrasts with a fixed

Table 8.4 Net buoyancy, m*, in tonnes for 10 km of pipe, using pipe diameters from Table 8.2 and pipe masses from Table 8.3. Net buoyancy, m*, for 10 km length/t Pressure/ 105 Pa (bar)

0.05% loss per kilometer m*/t

0.2% loss per kilometer m*/t

1% loss per kilometer m*/t

4% loss per kilometer m*/t

20 30 40 50 60 70

25,590 12,576 6300 2595 135 1620

14,615 7193 3602 1487 80 924

7654 3762 1889 779 43 482

4405 2168 1089 449 25 277

Underwater compressed air energy storage


Absolute internal pressure/105 Pa




40 Underwater Fixed volume








50 60 Charge/%





Figure 8.11 Pressure against percentage charge for a 10 m high underwater compressed air store at 400 m depth and an equivalent underground store with maximum and minimum cavern pressures of 7 MPa (70 bar) and 2.5 MPa (25 bar), respectively.

volume containment (such as a cavern), in which the air pressure varies strongly with the amount of stored energy. As such, CAES plants that use underground caverns must incorporate pressure regulators to ensure that the turbine inlet pressure is below a certain level. Pressure regulators have energy losses associated with them, losses which are not seen by an underwater system (which needs not use pressure regulation, or certainly doesn’t need such a severe pressure drop). Fig. 8.11 shows a curve of pressure against stored energy for the air in an underwater CAES plant at 400 m depth, and the curve for an equivalent plant using a fixed-volume container (e.g., an underground salt cavern). The isobaric characteristic of UWCAES is clear. The equivalent underground cavern (of the Huntorf/McIntosh type, with 4 MPa (40 bar) maximum turbine inlet pressure) requires the air to be throttled down to 4 MPa (40 bar) during discharge until roughly two-thirds of its energy content has been removed, and then at lower states of charge the turbine inlet pressure reduces below 4 MPa (40 bar). Both of these characteristics of fixed volume containment serve to reduce efficiency. Using reasonable costs for materials, it was shown in Ref. [3] that total costs of flexible vessels and ballast for UWCAES can be less than £10 (kW/h) (assuming storage at 500 m depth). Installation and machinery will increase the total cost. Through discussions with a major lift bag manufacturer, it has been found that currently the costs of flexible vessels could be less than £15 (kW/h) at 500 m depth. The design life of a UWCAES system will be at least 20 years. Concrete structures underwater have lasted much longer than this (concrete reacts well with seawater), and inflatable vessels have been used underwater for significant periods of time without malfunction.



Storing Energy

Contrasting UWCAES with pure gravitational storage approaches in deep water

A concept that appears at the surface to be very similar to UWCAES could be described as inverse pumped hydro. A number of publications have emerged from different sources in relation to this concept including [17a] which describes it as STENSEA [17b], which refers to it as ORES and [17c] which attaches the name DOGES to it. The core concept is the same in all cases: a rigid volume mounted at the seabed has water pumped out of it to charge the system and the pressure within the containment falls to a very low value (below atmospheric pressure in some cases). Then to discharge the system, water is allowed to flow back in to the containment and does work against a turbine to release the energy. The term inverse pumped hydro is applicable because in normal pumped hydro, the “reservoir” that is deliberately built is high relative to the main repository of water whereas in the cases being considered here, that manmade reservoir is low relative to the main body of water. A first difference between UWCAES systems and the inverse pumped hydro relates to the stress state of the containment when the system is fully charged. Both system types are at maximum stress when fully charged but the UWCAES systems have positive internal pressure differential while the inverse pumped hydro systems have significantly higher pressure externally. The pressure differences are much larger for the inverse pumped hydro systems, so these require significantly higher quantities of structural material for the same subsea volume. Moreover, the energy density for UWCAES systems are significantly higher. Denoting atmospheric pressure by pa , the reservoir volume by V, and the local seawater pressure by psw , the ratio between the energy densities is approximately lnðpsw =pa Þ din favor of UWCAES [17d]. Thus, each 1 m3 of reservoir volume in a UWCAES system at a seabed depth of 500m would store approximately four times more energy than 1 m3 of reservoir volume for inverse pumped hydro at the same depth. The inverse pumped hydro systems have the advantage that they use water pumps and turbines rather than air compression/expansion machines; so they have generally higher turnaround efficiencies and are simpler to operate since there are no significant thermal aspects.


State of development

After early development at The University of Nottingham, Hydrostor Inc. of Toronto pushed forward commercialization of UWCAES technology, working with researchers at the University of Windsor. A 750 kW h/1.5 MW h pilot plant was installed at a depth of 80 m in Lake Ontario in 2014. The pilot plant combines flexible and rigid storage vessels, and the first commercial plant was intended, in 2015, to be installed off the coast of Aruba in 2015/2016 but development paused and did not resume. Hydrostor Inc. continues (in 2021) to work on CAES systems, but its focus has shifted to isobarically compensated caverns for the air storage.

Underwater compressed air energy storage


Other companies working on or considering UWCAES include Bright Energy Storage, Brayton Energy, Exquadrum, AGNES, DNV, and Moffatt-Nichol. One company working on a novel method of moving the air down to the seabed is OC Energy, who propose the use of an inclined Archimedes screw coupled to a surface-mounted electrical machine [18]. The areas that are in particular need of development are: installation procedures and the securing of vessels to the seabed. Rigid vessels are designed such that they have ballast mass built into the structure and, to date, flexible vessels have been held to the seabed using gravity base anchors (i.e., large weights, such as concrete blocks on a steel tray) that were transported to site from shore. However, use of dead weights could become prohibitively time-consuming and costly if carried out for commercial-scale installations of large amounts of storage capacity. By way of example, let us consider a 1 GW h store at 500 m depth. This would require up to 179,000 m3 of storage capacity, and hence it would be necessary to react 172,000 tonnes of buoyancy force. With a 2:1 ratio of holding-down force to buoyancy force, 344,000 t of holding-down force would be required. Using concrete as a dead weight, approximately 250,000 m3 would be required to provide this ballast force underwater. To put this in perspective, the largest concrete dam in the UK (at Clywedog reservoir) has height of 72 m and length of 230 m, and contains about 200,000 m3 of concrete [19]. Other than using dead weights, alternative methods of securing vessels to the seabed include driven piles, suction piles, suction embedded anchors, torpedo piles, and screw (or helical) piles. Of these, screw piles are of particular interest because they provide some amount of positive drive. To minimize cost per unit of anchorage capacity, anchorage which can be installed without use of remotely operated underwater vehicles (ROVs) is of particular interest.

9. Concluding remarks Underwater compressed air energy storage is a developing storage technology which is a natural extension of compressed air energy storage for coastal environments. It is very similar to underground CAES in all aspects but the energy store. Compared with a fixed volume underground store, an underwater store brings the benefit of isobaric containment, raising the system’s round-trip efficiency. Around the world there are many coastal locations suitable for UWCAES, particularly around islands, and UWCAES plants could either have the machinery based on shore or on an offshore platform; the decisions over the number of compression/expansion stages and whether to use heat storage or burn gas are largely based on the same factors as underground CAES. The main challenges currently facing UWCAES are cost-effective access (including deployment, maintenance and recovery) and anchorage, and a number of companies and organizations are working to develop solutions to both.


Storing Energy

References [1] P.R. Cowie, Biofouling patterns with depth, in: Biofouling, Wiley, 2009. [2] C. VanWalleghem, Concept to construction: the world’s first grid connected UWCAES facility, in: Offshore Energy and Storage Symposium (OSES) 2014, Windsor, ON, 2014. [3] A.J. Pimm, S.D. Garvey, Analysis of flexible fabric structures for large-scale subsea compressed air energy storage, in: 7th International Conference on Modern Practice in Stress and Vibration Analysis, Cambridge, 2009. [4] A. Vasel-Be-Hagh, R. Carriveau, D.-K. Ting, Ocean Eng. 95 (2015) 59e77. [5] G. Brading, Private Communication from Seaflex Ltd. To A. J. Pimm Relating to Performance of Lift Bags Still in Use after 7 Years to Support a Pontoon at Brighton, July 2015. [6] PhD ThesisA.J. Pimm (Ed.), Analysis of Flexible Fabric Structures, The University of Nottingham, 2011; [6a] J. Mas, J.M. Tubular, Design for underwater compressed air energy storage, J. Energy Storage 8 (2016) 27e34. [7] C.M. Davies, “Energy Storage in Offshore Windfarms,” Report for Corus Group, Millbank, London, 2005. [8] R.J. Seymour, Ocean energy on-demand using underocean compressed air storage, in: 26th International Conference on Offshore Mechanics and Arctic Engineering, San Diego, 2007; [8a] S.D. Garvey, A.J. Pimm, M. Chee, Deployment methods for flexible air containments in deep water far offshore, in: Offshore Energy and Storage Conference 2016 (OSES2016), Malta, 2016. [9] S.D. Garvey, Leveraging energy bags as a cost effective energy storage solution, in: Dufresne Energy Storage Forum, Rome, 2012. [10] J. Ruer, Energy storage for offshore wind power, in: Conference Proceedings REM2012, Ravenna, 2012. [11] C. Tsuha, N. Aoki, G. Rault, L. Thorel, J. Garnier, Can. Geotech. J. 49 (2012) 1102e1114. [12] RWE Power, ADELE - Adiabatic compressed air energy storage for electricity supply, [Online]. [13] A.J. Pimm, S.D. Garvey, Potential locations for underwater compressed air energy storage in Europe and North America, in: Offshore Energy & Storage Symposium 2015 (OSES2015), Edinburgh, 2015. [14] F. Crotogino, K.-U. Mohmeyer, R. Scharf, Huntorf CAES: more than 20 Years of successful operation, in: Solution Mining Research Institute Spring 2001 Meeting, Orlando, 2001. [15] F. Meyer, “Compressed Air Energy Storage Power Plants,” BINE Projekt-Info 05/07, Karlsruhe, 2007. [16] A.J. Pimm, S.D. Garvey, M. de Jong, Design and testing of energy bags for underwater compressed air energy storage, Energy 66 (2014) 496e508; [16a] Z. Wang, W. Xiong, D.S.K. Ting, R. Carriveau, Z. Wang, Conventional and advanced exergy analyses of an underwater compressed air energy storage system, Appl. Energy 15 (2016) 810e822; [16b] O. Maisonnave, L. Moreau, R. Aubrée, M.F. Benkhoris, T. Neu, D. Guyomarc’h, Optimal energy management of an underwater compressed air energy storage station using pumping systems, Energy Convers. Manag. 165 (2018) 771e780.

Underwater compressed air energy storage


[17] S.D. Garvey, Compressed air energy storage: performance and affordability, in: Half-day Workshop Given at Marcus Evans Biannual Energy Storage Conference, Amsterdam, 2010; [17a] M. Puchta, J. Bard, C. Dick, D. Hau, B. Krautkremer, F. Thaleman, M. Hahn, Development and testing of novel offshore pumped storage concept for storing energy at sea, Energy Storage 14 (2) (2017) 271e275; [17b] A.H. Slocum, G.E. Fennell, G. Dundar, B.G. Hodder, J.D.C. Meredith, M.A. Sager, Ocean renewable energy storage (ORES) system: analysis of an undersea energy storage concept, Proc. IEEE 101 (4) (2013) 906e924; [17c] R. Cazzaniga, M. Cicu, T. Marrani, M. Rosa-Clot, G.M. Tina, Deep ocean gravitational energy storage. Energy Storage 14 (2) (2017) 264e270; [17d] S.D. Garvey, R. Carriveau, Offshore Energy Storage. (Chapter 9 in Renewable Energy from the Oceans: From Wave, Tidal and Gradient Systems to Wind and Solar, IET, in: 2019. [18] P. Agrawal, A. Nourai, L. Markel, R. Fioravanti, P. Gordon, N. Tong, G. Huff, Characterization and Assessment of Novel Bulk Storage Technologies: A Study for the DOE Energy Storage Systems Program, Sandia National Laboratories, 2011. [19] Severn Trent Water, Llyn Clywedog, [Online].*/tab/about/. (Accessed 25 June 2015).

This page intentionally left blank

A novel pumped hydro combined with compressed air energy storage system


Erren Yao, Hansen Zou, Ruixiong Li, Huanran Wang and Guang Xi School of Energy and Power Engineering, Xi’an Jiaotong University, Xi’an, Shaanxi, China

1. Introduction With the increasing depletion of traditional fossil energy sources, which accounts for a large proportion of total energy demands, it is becoming more and more urgent to find alternative types of energy [1]. Many countries are researching renewable energy, such as wind energy, solar energy, bioenergy, etc. However, the intermittent nature of renewable energy is a large barrier to its development. Only by solving this problem can we embrace the future of renewable energy [2,3]. Unfortunately, a considerable proportion of wind and solar power fails to reach the power grid due to base level overloading. It can be expected that this loss will increase with further expansion of wind and solar power, see Chapter 1. Two solutions for this problem are to build new pumped hydro storage (PHS) facilities (see Chapter 2) and also to build compressed air energy storage (CAES) facilities (see Chapters 6 and 7) [4,5]. In this way, excess energy can be stored when power demand is low and released when required during power peaks. PHS is currently the most practical and mature energy storage technology available [6e8]. According to the data from the Electric Power Research Institute (EPRI), PHS ranks first in the global energy storage market and accounts for more than 99% of the total stored energy. However, owing to the shortcomings of this technology, including large investment requirements, long construction periods, dependence on topography, and its influence on regional ecology and geology, the development of new PHS facilities is slow and limited to only certain areas [9e11]. On the other hand, CAES technology also has its problems which include large water flows, the stability and sealing problem of the underground air storage caverns, and auxiliary heating problems. In this chapter, the novel pumped hydro combined with compressed air system (PHCA) is introduced. The proposed system is a combination of PHS and CAES, which could not only integrate the advantages but also overcome the disadvantages of both the PHS and the CAES systems. Therefore, it is an attractive solution for the large-scale storage of intermittent renewable energy.

Storing Energy. Copyright © 2022 Elsevier Inc. All rights reserved.



Storing Energy

Basic principles of PHCA system

Consider a pressure vessel containing high pressured air and water connected to a pump by a pipeline and valve (see left-hand side of Fig. 9.1). During the offpeak electricity times, the pump starts operating and delivers water to the vessel, and the potential energy of water is increasing while the pressure of contained air is raised, thus building a virtual dam between the lower water tank and the upper air vessel. Furthermore, when the pressure in the vessel is up to 5 MPa, the water head of virtual dam h is equivalent to 500 m height of gravitational potential energy (see right-hand side of Fig. 9.1). During peak times, high-pressure water enters a hydroturbine which generates electricity. The scale of the constant-pressure PHCA mainly depends on the model selection of water pump and hydroturbine, as well as system’s operating time. Based on this principle, Wang et al. [12] proposed an improved model for PHS by using compressed air to build a virtual dam artificially. The system consists of a storage vessel with a spray system, an air compressor connected to the vessel, an open water tank, a set pump driven by electricity, and a hydroturbine connected to a generator (see Fig. 9.2). By adjusting the vessel’s air pressure, the fall (upstream and downstream) of the virtual dam is changed. Initially valves 7, 8, and 9 (see Fig. 9.2) are closed while valve 6 is open; this starts the compressor and pumps the air into the vessel, thereby setting up a preset pressure in the storage vessel. The compressor and valve 6 are then closed once the preset pressure has been reached. When storing energy, valve 8 is opened, and the water pump is switched on, and water is injected into the storage vessel. The storage vessel is further pressurized when the water level rises during this injection period. As the compressed air temperature in the storage vessel increases, the pumping process becomes more and more difficult. Furthermore, valve 7 will be opened, and a small amount of water will be flowed and sprayed into the vessel to decrease compressed air temperature. When air pressure in the storage vessel reaches its target value, valves 7 and 8 are closed and the energy storage process is completed. When electricity generation is required, valve

Figure 9.1 The physical model of PHCA.

A novel pumped hydro combined with compressed air energy storage system


Figure 9.2 Schematic of the PHCA. 1: Compressor; 2: Water pump; 3: Hydroturbine; 4: Storage vessel; 5: Water tank; 6e9: Valves.

9 is opened, and high-pressure water drives the hydroturbine which powers the electricity generator. The exhausting water then goes back to the water tank, which is connected to the atmosphere. Neither gas turbine nor gas combustion chamber is required in this process, although they are often used in CAES systems.

3. Characteristics of PHCA system According to the second law of thermodynamics, thermal energy cannot be converted entirely to mechanical energy, and some energy would be transferred to a lower temperature reservoir. However, mechanical energy can be converted entirely to thermal energy. It implies that the quality of mechanical energy is higher than that of thermal energy [13]. Experience shows that mechanical energy is one of the highest quality energies of all forms with an extensive range of applications. In essence, the thermal machinery (compressor/expander) of the CAES system is substituted by hydraulic machinery (water pump/hydro-turbine) of the PHCA system in energy storage and power generation. In the PHCA system, the stored energy is converted from mechanical energy (pressure energy) to electrical energy through the water turbine. Due to hydraulic machinery’s high power density, an excessively high energy head is not required, and hydraulic machinery does not need to be multistaged. Hydroturbines are singlestage units and thus more convenient to operate than multistage units. The total efficiency of large modern hydroturbines is of the order of 95%, the highest efficiency of all the prime movers. As a result, the PHCA system’s energy efficiency is theoretically higher than that of the CAES system.


Storing Energy

Compared with CAES [14e16], the PHCA system has the following advantages: (1) The physical structure is more straightforward, the electrical cooling and auxiliary heating system is not required; (2) The efficiency is higher; the efficiencies of the water pump and hydroturbine are both higher than that of the compressor and expander, and thus, the potential for improving the efficiency of a PHCA system is more significant than that for a CAES system; (3) The cost is lower; for the same scale of energy storage, the price of a high-pressure water pump and hydroturbine are less than those of the compressor and expander and the PHCA system does not involve a cooler or a heater.

Comparing with the PHS system [14], the PHCA system does not need very stringent geographical and geological requirements. Moreover, a PHCA installation does not require a significant water supply; it avoids evaporation problems and is more flexible in that the water-energy storage density is adjustable. However, the PHCA system does have some drawbacks. Firstly, high-pressure ratio and low mass flow through both the water pump and the hydroturbine would provide the system with high work density and low water requirement. Still, it is challenging to implement with existing industrial technology. Secondly, since the vessel’s pressure fluctuates with changing water level, neither the power generation process nor the energy storage process is stable. It has a deleterious effect on the energy storage efficiency and power generation quality and results from the design specifications of the hydroturbine and water pump. The next section will introduce a novel constant-pressure PHCA that will ensure a more stable and efficient operation.


A novel constant-pressure PHCA system

The system discussed here is a constant pressure PHCA system [17]. The figure below presents the basic idea behind its operation (Fig. 9.3). The process consists of the following three phases. (1) The initial compression process. By controlling the compressor 1, both the air in the storage vessel and high-pressure vessel can be pressurized. This process improves the working capacity per unit working medium in the same way as water should be pumped to a higher level in a traditional pumped hydro energy storage station to increase the upstream water’s gravitational potential energy. (2) The water injection process for energy storage. The water in the water tank is pumped into a storage vessel, and at the same time, the air in the storage vessel will be transferred to a high-pressure vessel. This step ensures the pressure in the storage vessel remaining the same, and the air in the high-pressure vessel is compressed to a predetermined level. (3) The power generation process. While the throttle valve 9 is opened, the hydroturbine generates power, driven by the high-pressure water. As the pressure in the high-pressure vessel is higher than that in the storage vessel, air will flow into storage vessel when water level decreases, thus keeping the pressure level in the storage vessel constant. Water will flow into a water tank to complete the cycle.

A novel pumped hydro combined with compressed air energy storage system


Figure 9.3 A schematic diagram of the constant-pressure PHCA. 1: Compressor 1; 2: Compressor 2; 3: Water pump; 4: Hydroturbine; 5: Storage vessel; 6: Water tank; 7: High-pressure vessel; 8: Moisture separator; 9: Throttle valve.

In discussing the thermodynamic performance of the system, the following assumptions are made: (1) The gas consists only of nitrogen which is scarcely soluble in water and is considered as an ideal gas. (2) In the thermodynamic calculations, the hydraulic water is considered incompressible. (3) The high-pressure vessel is an adiabatic container. (4) It is assumed that there are negligible potential and kinetic energy effects in the gas and liquid flow, and no phase change or chemical reaction takes place. (5) There is no pressure drop or loss along the gas and liquid pipelines.

5. Storage density analysis In the constant-pressure PHCA system, the storage vessel is one of the key components: it ensures the regular operation of the whole system, and its cost is related to the overall cost of the whole system. The input and output capacities of the pressure vessel per unit volume form an essential basis for optimizing of the system. The work density is defined by: Ep ¼

Wp ðp1  p0 Þε ¼ ð1 þ εÞhp Vh



Storing Energy

where p0, p1 is the initial pressure and the terminal pressure of storage vessel, respectively, Vh is the volume of the storage vessel, Vw is the volume of water in the storage vessel, ε ¼ Vw/(VhVw) is the water-air volume ratio, hp is the efficiency of the water pump, Wp is total work done by the pump, Ep is the work done by the water per unit volume flowing through the water pump. Focusing on the hydroturbine: Et ¼

Wt ð p1  p0 Þεht ¼ Vh ð1 þ εÞ


Moreover, ht is the efficiency of the hydroturbine, Wt is the total work output of the hydroturbine, Et is the work done by the water per unit volume flowing through the hydroturbine. According to the two equations Eqs. (9.1 and 9.2), the relationship between work density, preset pressure, and hydrosphere ratio can be obtained. As shown in Figs. 9.4 and 9.5, with increased preset pressure and water-air volume ratio, the storage system’s work density increased.


Thermodynamic analysis

The first law of thermodynamics states that energy in a closed system is conserved and the second law of thermodynamics establishes the difference in the quality of different

Figure 9.4 The variation of work density with preset pressure.

A novel pumped hydro combined with compressed air energy storage system


Figure 9.5 The variation of work density with water-air volume ratio.

forms of energy and explains why some processes can spontaneously occur while others cannot. It is essential to perform a thorough analysis of the quantity and quality of the energy in the constant-pressure PHCA system [18e20].


Energy analysis

The energy analysis is based on the first law of thermodynamics. As suggested working process of the constant-pressure PHCA, when the air in the storage vessel reaches the preset pressure, compressor 1 stops working in subsequent energy storage and power generation processes unless air leakage occurs in the high-pressure vessel. The system efficiency he can be expressed as: he ¼

Wht Wp þ Wc2


where Wc2 is the total power consumption of compressor 2 to maintain the pressure of storage vessel as constant.


Exergy analysis

Exergy is the maximum theoretical work obtainable from an overall system (consisting of a system and the environment) as it reaches equilibrium with the environment. Exergy analysis is a method that uses the conservation of mass and conservation of energy principles together with the second law of thermodynamics for the analysis.


Storing Energy

In contrast, exergy destruction is the measure of irreversibility, which is the source of performance loss [21]. The exergy analysis assessing the magnitude of exergy destruction identifies the equipment, the magnitude and the source of thermodynamic inefficiencies in an energy storage system. It indicates that an exergy analysis would help produce an efficient system that minimizes the exergy destruction in the system [22]. The exergy efficiency of the system hex is given by: hex ¼

Ep;tot Ef;tot


where Ep,tot is the exergy product of the whole system (i.e., the sum of the output power during discharging) and Ef,tot is the exergy fuel of the whole system (i.e., the sum of input power during both charging and discharging process). Considering electricity can be treated as 100% exergy, the energy analysis and exergy analysis have the same meaning for the proposed system [23].



When changing the preset pressure in the storage vessel and maintaining pressure P1 to a level below pressure P2 in the high-pressure vessel, the system thermodynamic efficiency variation with the preset pressure is obtained. Fig. 9.6 shows that the thermodynamic efficiency increases with increase in the preset pressure. While the system thermodynamic efficiency is not sensitive to the water-air volume ratio, the system’s

Figure 9.6 The variations of system efficiency and exergy efficiency with preset pressure.

A novel pumped hydro combined with compressed air energy storage system


energy intensity could be improved by increasing the water-air volume without influencing the system thermodynamic efficiency. It must be done within the allowable volume range for the storage vessel. When changing the pressure in the high-pressure vessel while maintaining pressure P2 at a level above pressure P1 in the storage vessel, the variations in system thermodynamic efficiency with pressure in the high-pressure vessel can be obtained. It can be seen from Fig. 9.7 that the thermodynamic efficiency of the system decreases with increased pressure in the high-pressure vessel. When P1 is constant, the smaller the pressure difference between storage vessel and high-pressure vessel, and the higher the proportion of the energy utilization in the constant-pressure PHCA. Fig. 9.8 shows the system thermodynamic efficiency versus its isentropic efficiency: with increased isentropic efficiency the system thermodynamic efficiency increases continuously, but the rate of increasing extent with different factors is different. The system thermodynamic efficiency increases slowly with increasing compressor efficiency, while the system thermodynamic efficiency shows an apparent upward trend as the water pump and hydroturbine efficiency increases. As far as the system thermodynamic efficiency is concerned, the compressor only increases the air pressure; the water pump does the work, and hydroturbine is much larger than that done by the compressor. Because the work done by compressor 2 accounts for only a small proportion of the total work, and as a result of the irreversible nature of the process, the isentropic efficiency variation of compressor 2 had little influence on the whole system’s thermodynamic efficiency. Moreover, compared with the hydroturbine isentropic efficiency, the water pump isentropic efficiency exerts a more significant influence on the thermodynamic efficiency of the whole system. As the low destruction exergy of the hydroturbine and the water pump are the critical issues needed to

Figure 9.7 The variation of system efficiency with pressure in the high-pressure vessel.


Storing Energy

Figure 9.8 The variation of system efficiency with isentropic efficiencies: (A) compressor 2; (B) water pump; and (C) hydro-turbine.

A novel pumped hydro combined with compressed air energy storage system


improve both the quantity and quality of energy consumption, the key to improving the system performance is to improve the isentropic efficiency of the water pump and hydroturbine. Figs. 9.6e9.8 confirm that the thermodynamic efficiency of the proposed system is much higher than that of the CAES systems. This finding is mainly attributed to the following reasons. Firstly, the isentropic efficiencies of the hydroturbine and water pump are both higher than that of the compressor and the expander. Secondly, auxiliary heating systems are not required in the proposed system while the exergy destruction of the intercooler and external combustion heater is high. Finally, the constant-pressure PHCA ensures both the hydroturbine and water pump can operate at their rated working conditions with high efficiency, which could improve the energy utilization.

References [1] A.A. Bazmi, G. Zahedi, Sustainable energy systems: role of optimization modeling techniques in power generation and supplyda review, Renew. Sustain. Energy Rev. 15 (2011) 3480e3500. [2] W. Liu, H. Lund, B.V. Mathiesen, Large-scale integration of wind power into the existing Chinese energy system, Energy 36 (2011) 4753e4760. [3] M.B. Blarke, H. Lund, The effectiveness of storage and relocation options in renewable energy systems, Renew. Energy 33 (2008) 1499e1507. [4] M. Beaudin, H. Zareipour, A. Schellenberglabe, W. Rosehart, Energy storage for mitigating the variability of renewable electricity sources: an updated review, Energy Sustain. Dev. 14 (2010) 302e314. [5] C. Yang, R.B. Jackson, Opportunities and barriers to pumped-hydro energy storage in the United States, Renew. Sustain. Energy Rev. 15 (2011) 839e844. [6] D. Connolly, H. Lund, P. Finn, B.V. Mathiesen, M. Leahy, Practical operation strategies for pumped hydroelectric energy storage (PHES) utilising electricity price arbitrage, Energy Pol. 39 (2011) 4189e4196. [7] C.K. Ekman, S.H. Jensen, Prospects for large scale electricity storage in Denmark, Energy Convers. Manag. 51 (2010) 1140e1147.  Gallachoir, E.J. McKeogh, Techno-economic review of existing and [8] J.P. Deane, B.P. O new pumped hydro energy storage plant, Renew. Sustain. Energy Rev. 14 (2010) 1293e1302. [9] M. Kapsali, J.K. Kaldellis, Combining hydro and variable wind power generation by means of pumped-storage under economically viable terms, Appl. Energy 87 (2010) 3475e3485. [10] J. Kondoh, I. Ishii, H. Yamaguchi, A. Murata, K. Otani, K. Sakuta, N. Higuchi, S. Sekine, M. Kamimoto, Electrical energy storage systems for energy networks, Energy Convers. Manag. 41 (2000) 1863e1874. [11] H. Ibrahim, A. Ilinca, J. Perron, Energy storage systemsdcharacteristics and comparisons, Renew. Sustain. Energy Rev. 12 (2008) 1221e1250. [12] H.R. Wang, L.Q. Wang, X.B. Wang, E.R. Yao, A novel pumped hydro combined with compressed air energy storage system, Energies 6 (2013) 1554e1567.


Storing Energy

[13] Z. Wang, Thermal and Power Mechanical Basis (in Chinese), China Machine Press, Beijing, China, 2000. [14] G. Grazzini, A. Milazzo, A thermodynamic analysis of multistage adiabatic CAES, Proc. IEEE 100 (2012) 461e472. [15] Y.M. Kim, D. Favrat, Energy and exergy analysis of a micro-compressed air energy storage and air cycle heating and cooling system, Energy 35 (2010) 213e220. [16] Y.M. Kim, D.G. Shin, D. Favrat, Operating characteristics of constant-pressure compressed air energy storage (CAES) system combined with pumped hydro storage based on energy and exergy analysis, Energy 36 (2011) 6220e6233. [17] E. Yao, et al., A novel constant-pressure pumped hydro combined with compressed air energy storage system, Energies 8 (1) (2015) 154e171. [18] M. Shekarchian, F. Zarifi, M. Moghavvemi, F. Motasemi, T.M.I. Mahlia, Energy, exergy, environmental and economic analysis of industrial fired heaters based on heat recovery and preheating techniques, Energy Convers. Manag. 71 (2013) 51e61. [19] M.J. Moran, H.N. Shapiro, D.D. Boettner, M.B. Bailey, Fundamentals of Engineering Thermodynamics, seventh ed., John Wiley & Sons, New York, NJ, USA, 2010. [20] I. Dincer, M.A. Rosen, Exergy: Energy, Environment and Sustainable Development, second ed., Elsevier, Oxford, UK, 2013. [21] K.R. Ranjan, S.C. Kaushik, Energy, exergy and thermo-economic analysis of solar distillation systems: a review, Renew. Sustain. Energy Rev. 27 (2013) 709e723. [22] E. Yao, et al., Thermo-economic optimization of a combined cooling, heating and power system based on small-scale compressed air energy storage, Energy Convers. Manag. 118 (2016) 377e386. [23] E. Yao, et al., Multi-objective optimization and exergoeconomic analysis of a combined cooling, heating and power based compressed air energy storage system, Energy Convers. Manag. 138 (2017) 199e209.

Liquid air energy storage 1, 2





Yulong Ding , Yongliang Li , Lige Tong and Li Wang 1 Birmingham Centre for Energy Storage, School of Chemical Engineering, University of Birmingham, Birmingham, United Kingdom; 2School of Energy and Enviornmental Engineering, University of Science & Technology Beijing, Beijing, China

1. Introduction Liquid air energy storage (LAES) refers to a technology that uses liquefied air or nitrogen as a storage medium [1]. LAES belongs to the technological category of cryogenic energy storage. The principle of the technology is illustrated schematically in Fig. 10.1. A typical LAES system operates in three steps. Step 1 is the charging process whereby excess (off-peak and cheap) electrical energy is used to clean, compress, and liquefy air. Step 2 is the storing process through which the liquefied air produced in Step 1 is stored in an insulated tank at w196 C and approximately ambient pressure. Step 3 is the discharging process that recovers the energy through pumping, reheating, and expanding to regenerate electricity during peak hours when electrical energy is in high demand and expensive. The Step 2 also includes the storage of heat from the air compression process in the Step 1 and high-grade cold energy during reheating process in the Step 3. The stored heat and cold energy can be used, respectively, in the Step 3 and Step 1 for increasing power output and reducing energy consumption of the liquefaction process. Clearly, the LAES involves electricalmechanical-thermal energy conversion and energy is stored mainly in the thermal form. As a result, although many regard LEAS as a thermal-mechanical energy storage method, it should be mainly in thermal energy storage category. The concept of the LAES technology was first proposed by researchers at the University of Newcastle upon Tyne in the UK in 1977 for peak shaving of electricity grids [2]. Although the work involved mainly theoretical analyses, it led to subsequent development particularly by Mitsubishi Heavy Industries [3] and Hitachi [4,5] of Japan, and Highview Power Storage (branded as Highview Power today) in collaboration with University of Leeds (UK) [1,6e9]. The work by Mitsubishi Heavy Industries led to a 2.6 MW pilot plant with air liquefaction and power recovery processes operated independently, leading to a low roundtrip efficiency [3]. The work by Hitachi looked at the integration of the air liquefaction and power recovery processes through a regenerator [4,5]. Such a regenerator stores cold energy released during the power recovery process and reuses the stored cold for reducing energy consumption of the air liquefaction process. Both simulation and experiments were carried out on the regenerator using solid materials and fluids as cold carriers. Based on their results, Hitachi claimed that the round-trip efficiency of cryogenic energy storage system could exceed 70% so long as the regenerator was efficient. Storing Energy. Copyright © 2022 Elsevier Inc. All rights reserved.


Storing Energy

Compression heat storage

Excess electricity


Air cleaning


Liquid air storage


Electricity out

High-grade cold storage

Air in



Air out



Figure 10.1 A schematic illustration of the liquid air energy storage technology.

However no fully integrated system demonstration was done. Working with the University of Leeds, Highview Power started to design and build the world’s first integrated LAES pilot plant (350 kW/2.5 MWh) from around 2009. The pilot plant was colocated with a Scottish and Southern Energy (SSE) biomass power plant in Slough (UK) and the whole plant was operational from around 2011. The pilot plant has now been relocated to the University of Birmingham, UK, for further research and development. In collaboration with Virido and under the support of the UK Department of Energy and Climate Change (now part of UK Department for Business, Energy and Industrial Strategy), Highview Power built and tested a 5 MW/15 MWh commercial demonstration plant in Virido’s Manchester (UK) landfill gas power plant. The work on the 5 MW/15 MWh plant was completed around 2019. Highview Power is now building a large-scale (50 MW/250 MWh) LAES plant with Carlton Power, a UK independent power station developer. The grid-scale plant is located in Carrington Village of Great Manchester, UK. It is expected to be operational in 2023.


Energy and exergy densities of liquid air

Fig. 10.2 shows the exergy density of liquid air as a function of pressure. For comparison, the results for compressed air are also included. In the calculation, the ambient pressure and temperature are assumed to be 100 kPa (1.0 bar) and 25 C, respectively. The exergy density of liquid air is independent of the storage pressure because the compressibility of liquid is extremely small. Fig. 10.2 indicates that although the mass exergy density of liquid air is only 1.5 to 3 times that of compressed air, the volumetric exergy density of liquid air is at least 10 times that of compressed air if the storage pressure of compressed air is lower than 10 MPa (100 bar). Such a high volumetric exergy density of liquid air makes it highly competitive even compared with current battery technologies [10].

Liquid air energy storage


Figure 10.2 Exergy density of liquid air and compression with compressed air.

Assuming the specific heat Cp of a sensible thermal storage material is a constant, an increase or a decrease in its temperature by DT from the ambient temperature, Ta , will lead to an amount of thermal energy, dQ, being charged into or discharged from the material: dQ ¼ Cp DT


Considering a reversibly infinitesimal heat transfer process, the exergy change, dE, of the material can be calculated by: dQ dE ¼ dH  Ta $dS ¼ dH  Ta $ T


where dH is the enthalpy change, dS is the entropy change. The exergy, DE, stored in the material is therefore obtained by integrating Eqs. (10.1)e(10.2) from Ta to (Ta þ DT):   Ta þ DT DE ¼ Cp ðDT  Ta $ ln Ta



Storing Energy

Combining Eqs. (10.1) and (10.3) gives the percentage of the available energy stored in the material, h, as: 

Ta þ DT DT  Ta $ln DE Ta h¼ ¼ jDQj jDTj


Eq. (10.4) is illustrated in Fig. 10.3 where the ambient temperature is assumed to be 25 C. It can be seen from Fig. 10.3 that, for heat storage, only a significant temperature difference can give a reasonable percentage of available energy. For cold storage, however, the available energy increases far quick with the increasing temperature difference compared with heat storage, suggesting that cold storage be a very attractive option. The physics behind this conclusion rests mainly with role played by the entropy, which increases with increasing temperature.


Liquid air as both a storage medium and an efficient working fluid

Currently low-to-medium grade heat is often recovered by steam cycles with water/ steam as a working fluid [11,12]. However, water/steam is not an ideal working fluid for efficient use of low-grade heat due to its high critical temperature of 374 C compared with the ambient temperature and its extremely high critical pressure of 22.1 MPa (221 bar). It is because of this that a large proportion of heat is consumed for vaporizing the water during phase change in subcritical or even transcritical

Both kinetic energy and entropy decrease with decreasing temperature

Fraction of available energy η

1.2 1.0

Heat storage Cold storage Both kinetic energy and entropy increase with increasing temperature

0.8 0.6 0.4 0.2 0.0 0






Temperature difference ΔT/K

Figure 10.3 Exergy percentage as a function of temperature difference for heat and cold storage.

Liquid air energy storage


cycles. This leads to the loss of a large portion of exergy in the heat transfer processes due to temperature glide mismatch between the heat source and the working fluidd the so-called pinch limitation [13,14]. The use of liquid air as a storage medium as well as a working fluid in the power recovery step of the LAES technology is thermodynamically more efficient than water in terms of recovering low-grade heat as demonstrated in the next paragraph. Consider a heat transfer process between a heat source and a working fluid with the working fluid heated from the ambient temperature (Ta) to TH ¼ 400 C, and define a normalized heat, Q, as the ratio of heat load at a certain temperature, T, to the total heat exchange amount during the whole process, one has: QðTÞ ¼

HðTÞ  HðTa Þ HðTH Þ  HðTa Þ


where H is the enthalpy. Fig. 10.4 shows the results of Eq. (10.5) for liquid nitrogen (the main component of air) and water. For comparison, the results for liquid methane and hydrogen are also included. One can see that, given a working pressure, the specific heat (the slope of the lines) is approximately the same for liquid nitrogen, hydrogen, and methane. However, water exhibits very different behavior. If the working pressure is lower than its critical value, the specific heat of water changes greatly due to phase change, leading to ineffective use of the heat source considering that the heat sources are mostly provided by fluids with a constant specific heat (e.g., flue gases or hot air). Although water behaves in a similar manner as methane under supercritical conditions (e.g., the case with pressure of 30 MPa (300 bar) in Fig. 10.4), the high working pressure increases the technical difficulties in realizing the process.


Normalized heat


0.6 H2 (1 bar) H2 (100 bar) N2 (1 bar)


N2 (100 bar) CH4 (1 bar) CH4 (100 bar) H2O (1 bar)


H2O (50 bar) H2O (100 bar) H2O (300 bar)

0.0 0


100 150 200 250 300 350 Temperature of cold-side working fluid/(°C)


Figure 10.4 Normalized heat as a function of cold side working fluid temperature.


Storing Energy


Applications of LAES through integration

Like other thermal-mechanical-based energy storage technologies, issues such as capital cost, round trip efficiency, and annual operating hours remain key challenges in the industrial take-up of the LAES technology. Integration of LAES with other processes/ systems provides a way to address the challenges. Examples are given in the following subsections as illustrations.


Integration of LAES with gas turbine peaking plants

Integration of LAES with a gas turbine peaking plant provides an opportunity to use the waste heat from the power generation process leading to high peak shaving capacity and increased overall efficiency [6]. The integration can also capture CO2 to give dry ice at no additional efficiency penalty. Fig. 10.5 shows the process diagram of the cycle, which works in the following manner: at off-peak hours, excess electricity generated by the base-load is used to power an air separation and liquefaction (ASU) plant to produce oxygen and liquid nitrogen while the rest of the system is powered off. The produced oxygen and liquid nitrogen are stored in a pressurized vessel

CH4 26

1 C1





HT 23








12 15



ASU 17 Air












WS 7

CS 10





20 Helium


ASU - Air separation (liquefaction) unit GT - Gas turbine

Ar, … HT/LT - High/Low pressure turbines

B - Combustor

CH4 O2

HE - Heat exchangers

C - Compressors


WS - Water separator

P - Cryogenic pump


CS - CO2 separator

G - Generator


Figure 10.5 Process diagram of an integrated LAES and gas turbineebased peaking power system, consisting of a closed Brayton cycle and an open nitrogen cycle [6].

Liquid air energy storage


and a cryogenic tank, respectively, for generating power via the high pressure turbine (HT) and low pressure turbine (LT), and assisting combustion in the combustor (B) at peak hours. The produced liquid nitrogen also serves as energy storage medium. At peak hours, natural gas is compressed in the compressor C1 to the working pressure. The working fluid then mixes with the oxygen in the combustor (B) where combustion takes place to give high temperature and high pressure flue gas consisting of CO2 and H2O. Combustion of the natural gas in an oxygen environment can produce a temperature that is too high for the gas turbine (GT). To control such a temperature, an appropriate amount of helium gas is mixed with the flue gas before entering the GT for power generation through a generator (G). Note that the helium gas is not consumed but circulates in the system; see Fig. 10.5. The flue gas containing helium from the GT then goes through a series of heat exchange processes via heat exchangers 1 (HE1), 2 (HE2), and 3 (HE3) to recover the waste heat by passing the heat to a nitrogen stream from liquid nitrogen storage tank; see Fig. 10.5. During the heat recovery processes, steam in the flue gas is removed via a condenser (WS), whereas CO2 is removed in the form of dry ice through a solidification process in CS [the triple point of CO2 is 571.8 kPa (5.718 bar) and 56.6 C]. As a result, the flue gas stream after CO2 removal contains only helium. The helium stream is then cooled down further in HE3 and compressed in compressor C2 to the working pressure, and finally goes through further heat exchange in HE2 and HE1 before flowing back to the combustor. The nitrogen stream starts from the cryogenic storage tank where liquid nitrogen is pumped to the working pressure by a cryogenic pump (P). The high-pressure nitrogen is then heated in heat exchangers HE3, HE2, and HE1 in turn, and expands in two stages via, respectively, a high-pressure turbine (HT) and a low-pressure turbine (LT) to generate electricity. Heat exchanger 1 (HE1) serves as an interheater between the two-stage expansion. After expansion, the pure nitrogen can be used to purge the sorbent bed of the ASU dryer. From the above, it can be seen that the integrated system consists of a closed-loop topping Brayton cycle [6] with He/CO2/H2O as the working fluid and an open-loop bottoming nitrogen direct expansion cycle. The topping Brayton cycle can be identified as 4 / 5 / 6 / 8 / 9 / 11 / 12 / 13 / 14 / 15 / 16 / 4, whereas the bottoming cycle is 18 / 20 / 21 / 22 / 23 / 24 / 25 / 26. It is the combination of the two cycles that produces electricity at the peak hours. The Brayton cycle uses natural gas, which is burned in the pure oxygen produced by the ASU during off-peak hours. Helium is only used to control the turbine inlet temperature (TIT) and is recirculated. The working fluid of the open cycle, nitrogen, is the actual energy carrier of the off-peak electricity. As CO2 is captured, only water and nitrogen are given out from the process. The optimal energy storage efficiency of such an integrated system is nearly 70% whereas the CO2 in the flue gas is fully captured. Economic analyses also show that if the integrated system is used for energy arbitrage and peak power generation both the capital and the peak electricity costs are comparable with Natural Gas Combined Cycle (NGCC), which are much lower than the oxy-NGCC if the operation period is relatively short [6]. Note that not only helium but also oxygen could be used as the recirculating fluid and similar conclusions could be obtained.



Storing Energy

Integration of LAES with concentrated solar power plants

Additional heat sources can enhance the round-trip efficiency and reduce the capital cost of the LAES system. Such heat sources can come from industrial processes and renewable solar radiation. This subsection explores the use of solar heat in largescale Concentrated Solar Power (CSP) plants. Fig. 10.6 shows an integrated LAES and CSP hybrid power system [15]. As can be seen, liquefaction process is not included in the system so external supply of liquid air/nitrogen is needed. This is practicable if there is a large-scale centralized liquefaction plant within a reasonable distance. The system shown in Fig. 10.6 consists of a direct expansion (open cycle) of liquid air/nitrogen from an elevated pressure and a closed-loop Brayton cycle operated at a medium to low pressure. The use of the Brayton cycle in place of conventional Rankine cycle gives a more efficient heat transfer process and a much lower working pressure. The expansion occurs sequentially in three stages (high-, medium-, and low-pressure turbines) and solar heat is used to superheat the working fluid. Simulation results show that such a system provides over 30% more power than the summation of a solar thermal power-only system and an LAES-only powered system.


Integration of LAES with nuclear power plants

To balance demand and supply at off-peak hours, nuclear power plants often have to be downregulated particularly when the installations exceed the base-load requirements.

Figure 10.6 Integration of LAES with a solar power system.

Liquid air energy storage


Part-load operations not only increase the electricity cost but also impose a detrimental effect on the safety and lifetime of nuclear power plants. Integration of nuclear power generation with LAES provides a promising solution to effective time shift of the electrical power output. Fig. 10.7 shows a flow sheet of such integration [16]. The integrated system consists of a nuclear power plant subsystem and the LAES subsystem. The nuclear power plant subsystem in the integrated system is similar to the conventional pressurized water power station. The only difference lies in that there are 2 three-way valves in the secondary loop, which enables the working fluid to feed into either the steam turbine to produce electricity or the heat exchanger 4 to superheat high-pressure air in LAES subsystem. The LAES subsystem consists of an air liquefaction unit in the left part and an energy extraction unit in the right-bottom part of Fig. 10.7. The integrated system could have three operational modes depending on the end-users’ demands as described briefly in the following: •

• •

Energy storage mode: At off-peak hours when the demand is much lower than the rated power of the power plant, the power plant operates in a traditional way to drive the steam turbine to produce electricity and the excessive power is used to drive the air liquefaction unit to produce liquid air. Energy release mode: At peak hours when the demand is higher than the rated power of the plant, the energy extraction unit is turned on to produce additional power. Conventional mode: when the supply is approximately balanced by the demand, both the air liquefaction unit and energy extraction unit are switched off so that the nuclear power operates in a conventional way to drive the steam turbine to produce electricity.

The air liquefaction subsystem works in a similar way as the simplest LindeHampson liquefier except for the use of external cold energy in the heat exchanger 6 (Fig. 10.7). It should be noted that in the air liquefaction unit, a cryoturbine is used to generate liquid product instead of a throttling device in a conventional setup. The working fluid expands in a near-isentropic manner in the cryoturbine with both temperature and enthalpy decreasing and hence generating more liquid product while producing additional shaft power. The cold storage and recovery steps act as a bridge between the air liquefaction unit and the cryogenic energy extraction unit. Such an arrangement enables the recovery of the cold energy released in the liquid air preheating process. In this process air is under the supercritical state, and as a result the cold is produced in the form of sensible thermal energy. Fig. 10.8 shows the heat capacity of air as a function of temperature at different pressures. One can see that the heat capacity of air changes only slightly in the heating process, particularly at very high pressures. This is similar to the use of liquids as sensible heat storage materials. In fact, the cold energy can also be stored in thermal fluids and such fluids can give good temperature gradient match during heat exchange and hence an efficient cold recovery. In this process the thermal fluids are used not only as a working fluid but also as a cold storage medium. Fig. 10.9 shows the heat capacities of some commonly used fluids that may be used as the storage media. Clearly, no single fluid can fully cover the working temperature region of liquid air preheating process. However, the combination of propane and methanol could work both as cold storage liquids and working fluids for heat transfer. Such


Exhaust air


Input air 10 1 2

Reactor vessel

Compressor 2

Heat 3exchanger 5


Pump 1


Pump 2

Heat exchanger 6



30 12


11 32 8


34 Heat exchanger 4 33


Heat exchanger 2


Containment structure


Generator 1 Cooling tower

Compressor 1

Steam turbine

Heat exchanger 1




Liquid air tank


19 18

21 20

23 22

25 24

Thermal fluid tanks 15

Cryogenic pump

Heat exchanger 3 27

Air turbines

Generator 2 26

Storing Energy

Figure 10.7 Integration of LAES with a nuclear power plant.


Liquid air energy storage


Isobaric heat capacity/(kJ kg–1 K–1)

4.4 4.0 Heat capacity of air at different pressures

3.6 3.2

p = 7 MPa p = 10 MPa


p = 15 MPa p = 20 MPa


p = 35 MPa p = 30 MPa

2.0 1.6 1.2 0.8 60


100 120 140 160 180 200 220 240 260 280 300 320

Temperature/K Figure 10.8 Heat capacity of air at different pressures.

Isobaric heat capacity/(kJ kg–1 K–1)

2.8 2.6 2.4 2.2 2.0 Propane R218 Propylene Ethane R12 Butane R11 Methanol

1.8 1.6 1.4 1.2 1.0

Comparison of heat capacity of different heat transfer fluids

0.8 80

100 120 140 160 180 200 220 240 260 280 300 320 340

Temperature/K Figure 10.9 Heat capacity of different cold storage fluids.


Storing Energy

as combination covers the required temperature range and has a high heat capacity. For each of the two fluids, a two-tank configuration is proposed for cold recovery and storage (Fig. 10.7). The two thermal fluids are pumped from the hot tanks to the cold tanks during the cold storage process (the energy storage mode), and flow back during cold release process (the energy release mode). The use of thermal fluids for both transferring and storing thermal energy can greatly simplify the design of the system in that no additional heat exchangers will be needed. Moreover, the operating strategy can be much more straightforwarddthe amount of cold energy and the objective temperature can be easily adjusted by controlling the flow rate of the fluids. This is extremely difficult to achieve using the conventional way of storing cold in a packed pebble bed. The cryogenic energy extraction unit is coupled with the nuclear power plant through the thermal energy utilization process via heat exchanger 4. One can see that hardly any thermal energy is wasted in the cooling process and hence the power output is expected to increase significantly. By integrating with LAES technology the reactor core and the primary loop of nuclear power plants can operate steadily at full load at all times while the net output power is adjusted only by the LAES unit. As the energy extraction process in the LAES subsystem is similar to power generation using a gas turbine, a much faster rate of power change could be achieved in comparison with the conventional downregulation of nuclear power plants. The combination of nuclear power generation and LAES enables the use of cryogen instead of steam as the working fluid in power generation process. As a result this provides an efficient way to use the thermal energy of nuclear power plants, delivering around three times the rated electrical power of the nuclear power plant at peak hours, effectively shaving the peak. Simulations have been carried out on this process, which show that the round-trip efficiency of the LAES is higher than 70% due to the elevated topping temperature in the superheating process.


Integration of LAES with liquefied natural gas regasification process

LAES can also be integrated with Liquefied Natural Gas (LNG) regasification plants to make use of the waste cold in the air liquefaction process [17,18]. The waste cold in LNG import terminals is significant due to the large volume of LNG storage. The LNG is normally regasified by heating with seawater and burning some natural gas. This leads to wasting of cold contained in the LNG and the burned natural gas. If LAES were colocated at the LNG terminal, and air rather than seawater was used to provide heat for the LNG regasification process; the resulting cold air could then be fed into the air liquefier, potentially reducing its electricity consumption by as much as two-thirds. Currently there are a number of nitrogen liquefiers in operation at LNG import terminals in Japan and Korea, which take advantage of this refrigeration to reduce the power consumption. The only challenge to be overcome is to reduce the capital cost of such an integrated system.

Liquid air energy storage



Integration of LAES with decentralized energy systems

Most LAES-related research so far has been on a system configuration similar to the patented one [1] (termed as a baseline LAES) with a similar operating pressure range provided in the patent. A recent study examines an operating pressure over a far wider range of charging pressure of 1e21 MPa [19]. The analyses show that the baseline LAES could achieve an electrical round-trip efficiency (eRTE) above 60% at a high charging pressure of 19 MPa. The baseline LAES, however, produces a large amount of excess heat particularly at low charging pressures with the maximum occurring at w1 MPa. The performance of the baseline LAES using the excess heat is also evaluated which gives a high eRTE even at lower charging pressures with a local maximum of 62% achieved at w4 MPa. As a result of the above, a hybrid LAES system is proposed to provide cooling, heating, hot water, and power. To evaluate the performance of such a hybrid LAES system, three performance indicators can be considered: nominal-electrical round-trip efficiency (neRTE), primary energy savings, and avoided carbon dioxide emissions [19]. The results show that the hybrid LAES can achieve a high neRTE between 52% and 76%, with the maximum at w5 MPa. For a given size of hybrid LAES (1 MW  8 h), the primary energy savings and avoided carbon dioxide emissions are up to 12.1 MWh and 2.3 tonne, respectively. These new findings also suggest that small-scale LAES systems could be best operated at lower charging pressures, and the hybrid system offers great potential for applications in local decentralized micro energy networks.

5. Technical and economical comparison of LAES with other energy storage technologies In this section, a brief comparison is made between the LAES and other energy storage technologies. The comparison will be from both technical and economic aspects as detailed in the following two subsections.


Technical comparison

Only pumped hydro storage can be currently regarded as a mature technology, which has been practiced for over 100 years. Although the compressed air energy storage technology has been developed and is commercially available, actual applications have not been widespread. LAES, together with flow batteries, hydrogen storage, and a number of other energy storage technologies [10], is still under development. Like pumped hydro and compressed air energy storage technologies, LAES offers a long discharge time (hours) compared with coupled energy storage technologies. However the power discharge rate depends on the scalability of the power regenerating unit of the energy storage technologies. Pumped hydro storage uses hydraulic turbines for power regeneration and as a result offers the largest power discharge rate (up to


Storing Energy

several gigawatts). Compressed air energy storage uses traditional gas turbines for power regeneration so the power rate is of the order of hundreds of megawatts. An LAES turbine is somewhat like a gas turbine but with a lower expansion temperature so the power rate is expected to be a little lower than compressed air energy storage, but can still reach hundreds of megawatts. However, the scalability of flow batteries and hydrogen storage is a big challenge and hence their power rates are expected to be lower than the LAES. Flow batteries and pumped hydro storage have a high (system-level) round-trip efficiency of 65%e85%. The round-trip efficiency of compressed air energy storage ranges from about 40% (commercially realized) to about 70% (still at the theoretical stage, although a recent 10 MW scale system designed and built by Institute of Engineering Thermophysics of Chinese Academy of Sciences and partners, have indicated a round trip efficiency of w60%). LAES has a low round trip efficiency of about 50%e60% mainly due to the low efficiency of the air liquefaction process. However, it should be noted that the round-trip efficiency of the LAES can be significantly enhanced if it is integrated properly and used for multivector services. In terms of the energy density, hydrogen storage has the highest volumetric energy density of 500e3000 W h/L depending on the storage methods (e.g., compressed gas, liquid, physical/chemical adsorption, etc.). As an extremely flammable gas, however, the technical requirements for hydrogen storage are high. The energy storage density of the LAES is an order of magnitude lower at 120e 00 W h/L, but the energy carrier can be stored at ambient pressure. Pumped hydro storage has the lowest energy density of (0.5e1.5) W h/L while compressed air energy storage and flow batteries are at 5e30 W h/L.


Economic comparison

Economic comparison can be based on the costs per unit amount of charging/discharging power that storage can deliver (dollars per kilowatt) and costs per unit amount of energy capacity (dollars per kilowatt-hour) that is stored in the system. It is difficult to evaluate a specific technology since the costs are influenced by many factors such as system size, location, local labor rate, market variability, local climate, environmental considerations, and transport/access issues. The capital costs provided in this section are intended to provide a high level understanding of the issues and are not intended as cost inputs into a design. The capital costs per unit amount of power relate mainly and directly to the cost of charging/discharging devices. Hydrogen storage is characterized by high capital costs for power (>$10 000 k/W). Pumped hydro storage, compressed air energy storage and flow batteries, and LAES have a more or less similar level of capital cost for power [about $(400e2000) k/W]. The capital costs per unit amount of energy cannot be used accurately to assess the economic performance of energy storage technologies mainly because of the effect of the discharging durations. An alternative measure is the capital cost of storage devices such as a dam for pumped hydro storage and a storage tank for the LAES. It is expected that hydrogen storage and compressed energy

Liquid air energy storage


storage have the highest storage costs, as the energy carriers are either combustible or at a high pressure. Pumped hydro storage has a low cost due to low energy density. LAES and flow batteries have the lowest cost even though insulation is required. In terms of the cycle life, thermal-mechanical-based technologies, including pumped hydro storage, compressed air energy storage, and LAES, have a long cycle life of 20e60 years as these technologies are based on conventional mechanical engineering, and the life time is mainly determined by life time of the mechanical components. By contrast, the life times of hydrogen storage and flow batteries are currently expected to be about 5e15 years.

References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19]

H. Chen, Y. Ding, T. Peters, F. Berger, Energy Storage and Generation, US Patent, 2009. E.M. Smith, Proc. Inst. Mech. Eng. 1847-1982 191 (1977) 289e298. K. Kenji, H. Keiichi, A. Takahisa, Mitsubishi Heavy Industries, Ltd, Tech. Rev. 35 (1998) 4. K. Chino, H. Araki, Heat Tran. Asian Res. 29 (2000) 347e357, 1523-1496(200007);2-a (2000). H. Wakana, K. Chino, O. Yokomizo, Cold Heat Reused Air Liquefaction/vaporization and Storage Gas Turbine Electric Power System, US Patent, 2005. Y. Li, Y. Jin, H. Chen, C. Tan, Y. Ding, Int. J. Energy Res. 35 (2011) 1158e1167, https:// Y. Li, X. Wang, Y. Ding, Int. J. Energy Res. (2013) 547e557, er.1942. Y. Li, X. Chen, Y. Ding, Int. J. Therm. Sci. 49 (2010) 941e949, j.ijthermalsci.2009.12.012(2010). Y. Li, H. Chen, X. Zhang, C. Tan, Y. Ding, Appl. Therm. Eng. 30 (2010) 1985e1990, H. Chen, T.N. Cong, W. Wang, C. Tan, Y. Li, Y. Ding, Prog. Nat. Sci. 19 (2009) 291e312, S. Vaivudh, W. Rakwichian, S. Chindaruksa, Energy Convers. Manag. 49 (2008) 3311e3317. J.Y. Shin, Y.J. Jeon, D.J. Maeng, J.S. Kim, S.T. Ro, Energy 27 (2002) 1085e1098. B. Saleh, G. Koglbauer, M. Wendland, J. Fischer, Energy 32 (2007) 1210e1221. Y. Chen, P. Lundqvist, A. Johansson, A.P. Platell, Appl. Therm. Eng. 26 (2006) 2142e2147. Y. Li, X. Wang, Y. Jin, Y. Ding, Renew. Energy 37 (2012) 76e81, 10.1016/j.renene.2011.05.038. Y. Li, H. Cao, S. Wang, Y. Jin, D. Li, X. Wang, Y. Ding, Appl. Energy 113 (2014) 1710e1716, X. Peng, X. She, B. Nie, C. Li, Y. Li, Y. Ding, Energy Proc. 158 (2019) 4759e4764. X. She, T. Zhang, L. Cong, X. Peng, C. Li, Y. Luo, Y. Ding, Appl. Energy 251 (2019) 113355. X. She, T. Zhang, X. Peng, L. Wang, L. Tong, Y. Luo, X. Zhang, Y. Ding, J. Therm. Sci. (2020) 1e17,

This page intentionally left blank

Flywheel energy storage


Keith R. Pullen Honorary Visiting Professor, City University of London, London, United Kingdom

1. Introduction A flywheel stores kinetic energy when a mass is rotated about a fixed axis, such mass being known as the rotor. Energy stored in the flywheel rises when the angular speed of the rotor is increased and reduces when it is slowed down. The maximum energy is usually limited by the maximum angular speed, itself limited by structural considerations. In order to speed up the rotor, a torque must be applied in the direction of rotation, to slow it down; the torque acts in the reverse direction. On one level, flywheel storage is very simple to implement and understand in comparison with many other energy storage methods and can store and release energy for potentially unlimited cycles. It has been used in practical applications for millennia with early applications ranging from smoothing out the pulsating torque on a potter’s wheel, lathes, and mills in ancient times to smoothing out the intermittent torque of steam piston engines. Indeed, as the industrial revolution gathered pace, the need for flywheels grew dramatically as more rotary machinery was produced. A comprehensive historical account of flywheel applications is given in Ref. [1] and in a more succinct treatise Ref. [2]. There is a class distinction between flywheels used for smoothing the intermittent output of an engine or load on a machine to those designed to store energy for a distinct period of time to be retrieved later. The latter are defined as flywheel energy storage systems (FESS). For the first category, the flywheel rotor is attached directly to another machine with the most common example being the type of flywheel directly mounted on every internal combustion engine. To illustrate one important difference, for a car engine equipped with a 10 kg flywheel, the energy stored is around 15 kJ (4 Wh) at maximum speed whereas an FESS rotor of the same mass might store upwards of 100 times this amount. In spite of this difference, the internal combustion engine flywheel application experiences around a billion charge and discharge cycles during engine life, absorbing significant engine torque pulsations although energy flows during this smoothing process are small relative to mean stored energy. This illustrates one very important characteristic of flywheels, the ability to accept almost infinite cycles without degradation. Flywheels on thousands of machines in museums around the world, sometimes hundreds of years old can still store the same energy as they did when made, demonstrating the second attribute of no calendar life degradation.

Storing Energy. Copyright © 2022 Elsevier Inc. All rights reserved.


Storing Energy

A second class of distinction is the means by which energy is transmitted to and from the flywheel rotor. In a FESS, this is more commonly done by means of an electrical machine directly coupled to the flywheel rotor. This configuration, shown in Fig. 11.1, is particularly attractive due to its simplicity if electrical energy storage is needed. It allows the rotor to be placed inside a hermetically sealed casing facilitating vacuum pressure needed to mitigate gas friction losses. However, there are many examples of FESS in which energy transmission is entirely mechanical or hybrid electromechanical and these are more common for transport applications in which mechanical shaft power is ultimately needed to propel or regenerate energy in a vehicle. With the move to electrification of all energy systems including transport, interest in mechanical transmission for FESS has reduced considerably in the past few years. DC Link

Airght safety containment casing

Radial bearing

Flywheel rotor Motorgenerator rotor

Motorgenerator stator Radial and thrust bearing

Figure 11.1 A flywheel system configured for electrical storage.

The flywheel schematic shown in Fig. 11.1 can be considered as a system in which the flywheel rotor, defining storage, and the motor generator, defining power, are effectively separate machines that can be designed accordingly and matched to the application. This is not unlike pumped hydro or compressed air storage whereas for electrochemical storage, the power-to-energy ratios, or C ratings (kW/kWh), are intrinsically intertwined in overall design and chemistry. A small motor-generator can be paired with a large flywheel to give a low C rating or the opposite can be decided. FESS typically operates best with C ratings in the range C100-C5, sitting in the zone between ultracapacitors and electrochemical batteries. More is given on how FESS is placed in the energy storage landscape in Section 4.

Flywheel energy storage


2. Principles of operation 2.1

Fundamentals of kinetic energy storage

It is interesting to start with the question of how much energy can be stored in a particular mass of material. From fundamentals, if a mass m shown in Fig. 11.2A rotates around a fixed axis with velocity U, the kinetic energy stored E is given by: 1 1 E ¼ mU 2 ¼ mr 2 u2 2 2


since U ¼ ru with u being the angular velocity in rad/s. The same mass m can now be distributed in a ring, Fig. 11.2B without changing the velocity of the mass or the energy stored. By knowing the moment of inertia for such a geometry; I ¼ mr2, the energy stored can be expressed as: 1 E ¼ Iu2 2


Now if the same mass m has the shape of a thin disc of outer radius r, Fig. 11.2C, then the moment of inertia reduces to I ¼ 1/2 mr2 and the energy stored is halved. This can be explained in physical terms by realizing that the velocity of any part of the disc is proportional to its radial position so any part of the mass at a radius lower than r will store less energy, given its velocity is lower. As an extreme example, if all the mass were at the center, Fig. 11.2D, no energy would be stored even if rotating meaning I ¼ 0. Therefore, for any shape, the value of I must sit between zero and mr2. It would appear at first glance at Eq. (11.2) that a ring or a close practical shape such as a thin-walled hollow cylinder would be the best shape, since it is clearly beneficial to maximize I by attempting to get this as close to mr2 as possible. However, the maximum value of U is limited by the strength of the material resisting the stress, commonly known as centrifugal stress, needed to provide centripetal acceleration of

Figure 11.2 Energy stored in rotating elementary shapes.


Storing Energy

Figure 11.3 Force balance diagram to obtain the stress in a rotating ring.

the material. Examining the balance of forces on a section of a thin-walled cylinder of cross-sectional area A, density r, and stress s as shown in Fig. 11.3 yields Eq. (11.3). s ¼ rU 2 ¼ rr 2 u2


For the thin disc, Fig. 11.2C, the stress analysis, now 2D, is a little more complex, Ref. [3] also requiring consideration of the material Poisson’s ratio v. Such analysis shows that the maximum stress occurs at the center of the disc and has a value of: sdisc ðcentreÞ ¼

3þv 2 2 rr u 8


Given that for most metallic materials, v is 0.3, the stress is a factor of 0.4125 lower than the ring. However, the disc loses a factor of two by nature of lower inertia but, in spite of this, still offers a net benefit of 0.5/0.4125 ¼ 1.212 times more energy than the ring. The disc therefore has 1.21 times more specific energy than a ring for the same material and maximum stress. The question must be raised whether there is a better shape than a simple ring or disc, so this is addressed in the next section.


Flywheel rotorespecific energy and shape factors

Energy storage for different rotor shapes has been considered, for example, in Ref. [1] with use of nondimensional shape factors KS, defined as: E smax ¼ Ks m r


Flywheel energy storage


The closer the value of KS is to unity, the more energy can be stored in a particular mass of material, material strength, and density. KS can be derived for the thin ring by rearranging Eq. (11.5) and substituting for E from Eq. (11.2) and stress from Eq. (11.3), hence: 1 2 2 E r 2 mr u r Ks ¼ ¼ ¼ 0:5 ms mru2 r 2


It is important to note that the value of KS depends on both moment of inertia and stress, this point being rarely explained. In order to unravel this, these two factors have considered in turn which is vital when considering other bodies of revolution shapes. Firstly, the well-known radius of gyration kG can be used to characterize the moment of inertia where; 2 I ¼ mkG


This can be nondimensionalized relative to the outer diameter of the rotor ro by defining radius of gyration factor KG. kG 1 KG ¼ ¼ ro ro

rffiffiffiffi I m


This factor presents how good the shape is in maximizing moment of inertia for a given mass and cannot exceed unity. In terms of the limiting maximum stress in the rotor shape, a stress factor Ks can be defined as a multiplier of the stress in a thin hollow cylinder, thus: smax ¼ Ks rr 2 u2


This stress factor represents how good the shape is in minimizing the value of maximum stress found at any point in the rotating body. The value of KS can now be calculated by substituting for E and s in Eq. (11.6) for any shape with knowledge of moment of inertia, determined by integration, and stress factor determined by classical theory of elasticity or finite element analysis: 1 2 2 2 E r 2 mKG ro u r 1 KG2 ¼ Ks ¼ ¼ ms 2 Ks mKs ru2 r 2


As an example, for the thin ring, both KG and Ks are 1 so KS is 0.5 as before. For the thin disc with radius of gyration proffiffi, KG equals p1ffiffi. From Eq. (11.4), Ks equals 0.4125; 2

so this gives a value of KS equal to 0.606.



Storing Energy

Figure 11.4 Shape factors for common flywheel shapes.

For other common shapes, values of relative radius of gyration, stress factors Ks and KS are shown in Fig. 11.4. Note that this diagram only applies for isotropic materials, for example metals with similar properties in all directions. A number of observations can be made moving left after the thin ring and thin disc, such shapes having already been discussed. Starting then with the third shape, as soon as a disc has even the tiniest hole, the stress doubles due to creation of a stress concentration from elastic analysis available in Ref. [3]. As the hole grows in size, both the stress factor and radius of gyration factor increase but rise in KG is greater, so KS increases as the hole size is increased until it reaches a maximum of 0.5 for the thin ring. The next shape is based on a Laval type disc in which the cross-section of the disc reduces with radius in order to keep every part of the disc at the same stress utilizing all the material fully rather than being limited to a small portion of the disc being at maximum stress. This design is attributed to the engineer De Laval and forms the basis of most high speed turbomachinery discs for holding axial blades. Due to three-dimensional effects, a disc with perfect constant stress is not possible and, for reasons of limiting outer diameter, a more practical design would have the outer extreme portion removed; hence, the description of the “Practical” Laval shape. It is common practice also to create a rim so combining the benefits of higher radius of gyration factor with lower stress with the rim is supported by the disc. The final two shapes are the solid cylinder which some texts erroneously consider to have the same stresses as a thin disc, but this is incorrect since Eq. (11.4) and similar derivations, Ref. [3] are only valid for plane stress. The correct stress can be calculated using finite element analysis and the difference is significant with maximum stress rising by nearly 10% for the geometry shown as compared to a thin disc of the same diameter and speed. A thick hollow cylinder is similarly affected by threedimensional effects but much less so, since the region of greatest stress is the free surface at the inner hole diameter.

Flywheel energy storage


Having looked at different geometries, the next question is which material is best and intuition might lead one to think that a highly dense material is good. Eq. (11.5) shows this is incorrect but rather it is the ratio of material strength to density that is important to maximize, the explanation being that a lower density material will allow a higher speed from Eq. (11.9), which leads to higher energy storage, see Eq. (11.1). Material densities and strengths can vary considerably for engineering materials, and it is not ultimate strength that is important but rather the maximum design stresses at which they can operate under cyclic conditions as needed for flywheel application without failing in fatigue or compromising safety. The cost of materials also varies considerably, and material cost must include raw material, treatments, and manufacture. The exact materials used in particular commercial flywheels and their maximum operating stresses are proprietary information, but it is possible to estimate the maximum operating stress by using dimensions from publicly available information and back calculating against published data. The critical number is the rotor peripheral speed, since if the shape and density are known the stress can be calculated from Eq. (11.9). Potential flywheel materials are listed in Table 11.1 with densities and typical practical design stresses as found in commercial systems. The table also indicates whether materials are isotropic and this is important given values in Fig. 11.4 are only valid for isotropic materials with exception of the thin ring. Such materials have the same strength in all directions whereas fibrous composite materials are orthotropic, with strength being in one direction for a uniaxial fiber orientation. All the design stresses are considerably lower than ultimate strengths, but this is to give sufficient fatigue life and account for variations in material quality and properties which tend to be greater in composites inherent to their nature. From Table 11.1, it can be seen that composite materials offer a considerable advantage in terms of strength-to-density ratio which translates into lower rotor mass. However, consideration of specific strength alone can create an incomplete comparison and this is further explored in the next section on rotors. It should be noted that glass fiber is not suitable as a rotor material on its own due to low creep resistance but it has been used in conjunction with carbon fibre composite, the latter forming the outer part of the rotor, Ref. [4].


Flywheel rotor-specific volume

Another important consideration is the volume energy density of the rotor which may be as important as the mass energy density particularly in transport applications. Table 11.1 Comparison of candidate flywheel materials. Material

r (kg/m3)

smax (MPa)

smax/r (kJ/kg)


Alloy steel Aluminum 7075-T6 Titanium Ti-6Al-4V Glass fiber composite Carbon fiber composite

7850 2810 4430 2580 1650

600 240 510 735 1000

76 85 115 284 606

Isotropic Isotropic Isotropic Orthotropic Orthotropic


Storing Energy

There are two volumes to consider, the first is the volume that the rotor material takes up, VR, and the second is the external volume of the rotor, VRE, which is rarely considered but arguably more important. Taking the first, as reported in Ref. [5], E ¼ Ks smax VR


Here, the shape factor KS is the same as that used for specific energy density but it can be seen that only the strength of the material is important not the specific strength. To give an example, steel and titanium cylindrical rotors each of 1 kg mass would store 46.32 kJ and 69.77 kJ, respectively, based on values given in Table 11.1. The volumes VR of each rotor are 0.127 L and 0.225 L, respectively, so by dividing the energy stored by volume for each, steel has the greater value of E/VR in proportion to material strengths as indeed shown by Eq. (11.11). Now for specific energy to external volume VRE, this will follow Eq. (11.11) with VRE ¼ VR if the body is a solid cylinder and with no hole or undercut. For any other shape, VRE is the external cylindrical volume enclosing the shape and will be greater than VR. VRE dictates the overall volume of the machine including the size and weight of the casing, hence it is important. It is convenient to use a modified form of Eq. (11.11) adding a correction shape factor KSVE given as Eq. (11.12): E ¼ KSVE Ks smax VRE


The value of KSVE is plotted for a hollow cylinder as a function of inner-to-outer radius ratio in Fig. 11.5. As expected, KSVE goes to unity as the radius ratio goes to zero and VRE ¼ VR as before. 1 0.9


0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0











radius rao ri /ro Figure 11.5 Correction shape factor KSVE plotted as a function of hollow cylinder radius ratio.

Flywheel energy storage


The shape of the curve is a parabola as given by Eq. (11.13): KSVE ¼ 1 

 2 ri ro


For other shapes, for example, the practical Laval disc shown in Fig. 11.4, KSVE is calculated by dividing the material volume by the external enclosing volume, so KSVE is 0.284. To illustrate this by way of another example, a hollow cylinder with thin wall looks to be an attractive design given its shape factor of KS ¼ 0.5. However, this shape has less practical value, since external volume would be very high due to KSVE being so low, making the size and weight of the casing substantial.


Flywheel rotor maximum operating speeds

Another important issue to be addressed is the flywheel operating speed range. The mass of the flywheel can be determined from Eq. (11.5) and the product of rotor radius r and angular velocity u, the rotor peripheral speed, from Eq. (11.6). There is hence a series of designs which satisfies a given energy storage requirement with r being inversely proportional to u, one of which must be specified to define the geometry. Typically, the rotor speed will be determined by either the limits of the motor generator or a mechanical shaft input. Alternatively, a value of u can be chosen as a starting point and varied in an iterative process and, once chosen, yields the value of r. The rotor length can now be obtained since the volume of material is known from Eq. (11.11) and V ¼ pr2L. As an example, taking a solid cylinder as the rotor shape, a steel rotor with properties in Table 11.1 is required to store 10 MJ (2.77 kWh) of energy. The maximum operating speed first tried is 15 krpm. From Eq. (11.9), the rotor radius is found to be 262 mm and the material volume calculated from Eq. (11.11) is 0.03127 m3 taking KS ¼ 0.553 from Fig. 11.4. The rotor length needed is 145 mm for this material volume which has a mass of 246 kg. This aspect ratio may not be desirable since the shape turns out to be more of a disc than a cylinder, so an option is to increase the shaft speed. Doubling this to 30 krpm will half the rotor radius to 131 mm but increase the length four times to 580 mm, calculated since conservation of volume means that halving the radius leads to a quadrupling of rotor length. This leads to a more compact design with length around twice the diameter. However, trying to increase the speed further may lead to impractically long rotor. This simple calculation shows there is limited freedom in the choice of practical speeds that are workable for a given storage level since the choice of speed strongly affects rotor radius-to-length aspect ratio.


Flywheel rotor operating speed range

Up to this point, all the analysis has been based on the energy stored at maximum speed but in order to store energy, mechanical torque T must be applied to the shaft


Storing Energy

in the direction of rotation and applied in the opposite direction to extract energy. It is usual that any storage system must provide or absorb a specified value of power regardless of its state of charge and since power is the product of T and u: T¼

P u


Eq. (11.14) dictates that for any given power, the torque tends to infinity as the speed tends to zero. In order to limit the torque at an acceptable level, the rotor must therefore operate between a defined minimum speed umin and its maximum speed umax and this means the useful energy which can be stored is less than that given in Eq. (11.2) as:    u2 1  1 Eusable ¼ I u2max  u2min ¼ Iu2max 1  2min 2 2 umax


If an electric motor generator is used to provide the charge and discharge torque then umin is typically set at between one-third to half of umax. The usable energy 1 will be 8/ths 9 of the actual energy at maximum speed, Emax, for umin ¼ /3 umax, and 3 1 /4 Emax for umin ¼ /2 umax. Given the size and cost of the electric machine is roughly proportional to torque, choosing umin ¼ 1/3 umax would lead to an electric machine that is 50% larger than if umin ¼ 1/2 umax but more energy could be extracted from the rotor. The choice of speed range is therefore a tradeoff but as a guide, the system with lower umin would be preferred if the C rating was low, since the most expensive part of the system would be flywheel rotor and it is therefore beneficial to increase usable capacity. For the high power, high C rating, the opposite would be true. It is also useful to relate the speed of the flywheel to the state of charge (SoC) which can be set at 0% when u ¼ umin and 100% when u ¼ umax. SoC can be found for any speed u as: 1 2 1 2 Iu  Iumin 2 SoC ¼ 2 1 2 1 Iu  Iu2 2 max 2 min


and by letting f ¼ umax/umax u2 f u2max SoC ¼ 1f


Rearranging: u ¼ umax

pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi SoCð1  fÞ  f


Flywheel energy storage


As an example, for a flywheel with umax ¼ 20 krpm, and umin ¼ 10 krpm, the speed of the flywheel when its SoC is 50% is 20,000 rpm x (O5/8) ¼ 15,811 rpm.

3. High-performance electric flywheel storage systems As explained in Section 1, the most common flywheel energy storage configuration is the hermetically sealed system incorporating a motor generator shown in Fig. 11.1. Since it is a system of components, a number of design choices exist for each of the components and many of these components have been developed for other, different applications. For example, high-speed brushless electrical machines and power conversion electronics have also been developed for generating power from microturbine engines and turbo expanders and similar technology in motor form was developed to drive turbo compressors avoiding gearboxes. More recently, the electrification of automobiles has led to considerable investment in development of brushless traction motors which operate around 10 krpm. High speed bearings were required for the same technologies, so availability of these components has aided development of modern high-performance flywheels. Given that the flywheel is a collection of components, it is useful to map out the design choices and these are shown in the form of a morphological chart in Fig. 11.6. The options shown in Fig. 11.6 are those that have been seen either in commercially available flywheels or in precommercial development over the past decade but are by no means exhaustive. Each design choice area is briefly discussed in turn.


Inertia rotor design approaches

The fundamentals of the design of the inertia rotor were discussed in Section 2 and along with their effect on two key performance indicators, specific energy with respect to mass and volume. It might be asked why the Laval disc, with or without rim, Fig. 11.6-1(A)e(B), would not be commonly used given its superior shape factor. This shape was previously in commercial use mid-20th century on buses [1] but is no longer generally used. The answer lies in the need to keep rotor diameter to an acceptable overall diameter yet provide a good level of storage. For a solid cylindrical design, Fig. 11.6-1(C) the diameter and length can be tailored to any ratio within reason but this is not so for the Laval disc whose diameter-to-length ratio can only be varied slightly, it must remain a disc not a drum. Taking the same example given in Section 2.4 in which 10 MJ storage is required, for the Practical Laval disc of the geometry shown in Fig. 11.4, this has a shape factor of 0.72 and radius-to-axial length ratio of 2.3. From Eq. (11.9), the product of radius and speed is 714 m/s and given Ks is 0.15 and from Eq. (11.5), the volume of the rotor can be determined to be 0.0188 m3. For this Laval shape, with KSVE is 0.284, the enclosed volume of the rotor is 0.0188/ 0.284 ¼ 0.0662 m3. Since the ratio of length to radius is known, the radius can be calculated to be 357 mm and axial length 155 mm. The speed then drops out of the calculation rather than being an input as 714/0.357 ¼ 2000 rad/s or 19,100 rpm. The


Storing Energy

1. Ineral rotor (a) Praccal Laval metallic

(b) Rimmed Laval metallic

(c) Monolithic metallic cylinder

(d) Laminated metallic

(e) Wire wound metallic

(f) Sheet wound (g) Filament wound composite metallic

2. Motorgenerator topology (a) Convenonal

(b) External rotor

(c) Axial flux

(d) Magnecally acve rotor

3. Motor-generator electromagnec (a) PM surface mount

(b) PM (c) Inducon (d) Switched (e) Synchronous embedded reluctance reluctance

(f) Homopolar

4. Bearing type (a) Rolling element

(b)Passive magnec axial repulsive

(c) Passive magnec axial aracve

(f) Acve (d) Passive (e) Acve magnec magnec magnec axial radial radial

(g) Super (h) Super conducng conducng magnec magnec radial axial

5. Installaon

(a) On board vehicle

(b) Free standing stac

(c) Bunkered stac

Figure 11.6 Morphological chart of the design options for electrical energy storage flywheels.

result is a larger diameter and longer rotor than is the case for an equivalent simple cylinder of the same energy and the resulting larger casing will add additional cost and weight. Also, since it operates at higher speed, this would dramatically increase aerodynamic friction losses. Correlations in Ref. [6] show that, for a given pressure, the losses increase approximately in proportion with the fourth power of diameter and cube of speed leading to a factor seven increase. Given its mass is 62% of the mass of a simple cylinder for the same energy, the Laval disc would have an advantage in terms of bearing losses although some of the benefit would be lost due to its higher speed operation. Using a rimmed Laval might improve the situation but will not compensate for the larger external volume, higher losses, and the higher cost of machining the shape with much wasted of material. The simple solid cylinder wins out and a number of commercial organizations vouch for this choice [7,8]. Having a shape similar to the solid cylinder, the laminated design (Fig. 11.6-1(D)) comprises of a stack of thin sheet discs held together by various means [9e11]. This approach offers a reduction in hoop stress by elimination of 3D effects leading to a higher shape factor, always at 0.606 and for the 10 MJ storage example given earlier, reduces mass from 246 kg to 216 kg. In the process, axial stresses are removed, the

Flywheel energy storage


presence of which creates the triaxial stress field found in a monolithic cylindrical rotor which promotes fracture. It is also easier to detect defects in thin sheet rather than solid material but the design and assembly process must be simple in order to be cost effective as compared to the monolithic cylindrical design. An alternative approach to rotor construction using steel is to wind the material tangentially either as a wire or sheet, Fig. 11.6-1(E)e(F), as described in Refs. [1,9] for the reason that strong steel can be sourced as wire and sheet drawn in one direction. However, bonding this together, particularly at the ends is challenging and the shape factor will be lower than a solid or axially laminated design since the strength of the material cannot be utilized in the radial direction. Moving to the filament composite rotor, Fig. 11.6-1(G), many publications conclude that the specific strength advantage offered by carbon fiber over steel of eight to one (see in Table 11.1) is so overwhelming that steel designs were made obsolete. However, since many recent commercial providers have chosen steel over composites, clearly the comparison deserves greater enquiry. A composite material typically comprises of a filament wound fiber bonded together with polymer matrix material, a type of orthotropic material. Such a construction is excellent at resisting loads in the direction of the fiber but not perpendicular to it with strength in that direction being two orders of magnitude lower. In the case of the shapes shown in Fig. 11.4, only hollow cylinders can be utilized since a rotating solid cylinder generates radial stresses as high as the tangential values at its center. The analysis of the stresses within different constructions of composite flywheel rotors is given in Ref. [12]. A thin ring geometry is ideal for composites since radial stresses are negligible and results in a shape factor of 0.5. However, the external volumetric shape factor is extremely low as explained previously. This can be resolved by increasing the wall thickness but then significant tensile radial stresses start to appear. This issue can be partly mitigated by dividing the thick-walled rotor up into layers, essentially thinner wall cylinders nested within thin cylinders. If the inner cylinders have lower stiffness-to-density ratios, the radial growth of the inner tubes can be matched to avoid separation from the outer tubes. This may be accomplished by choosing different fibers types for each layer or having carbon fiber for the outer layers and E-glass for the inner layer. In this way the inner radius can be as low as 2/3rds of the outer radius, but the maximum speed is still dominated by the stresses in the ring at the outer radius. To a first approximation, the value of Ks ¼ 1 from the thin ring can be used and combined with KG for a hollow cylinder with radius ratio of 2/3rds which is 0.85. The overall shape factor KS is then 0.36 and now a comparison can be made with a monolithic steel rotor. For a storage of 10 MJ and material values from Table 11.1, the mass of the rotor can be calculated to be 44 kg using Eq. (11.5). If the same rotor radius of 131 mm is chosen as for the earlier steel rotor, the speed can be determined as using Eq. (11.9) as 6092 rad/s or 58,170 rpm. Finally, the rotor length can be determined based on material volume of 0.0270 m3 converted to an enclosed cylinder volume of 0.0476 m3 by applying correction shape factor KSVE of 0.566 from Eq. (11.13). The length needed is 0.833 m which is over three times the rotor diameter. Such a design is feasible given there are machines commercially available [15] but the option exists for reducing speed and increasing diameter.


Storing Energy

Table 11.2 Comparison of rotors of different constructions for 10 MJ storage, all 262 mm diameter. Value

Monolithic steel

Laminated steel

Carbon composite

Rotor mass (kg) Rotor external volume (m3) Rotor length (mm) Rotor speed (rpm)

246 0.0313 580 30,000

216 0.0275 509 32,016

44 0.0476 833 58,170

The comparison of steel and composite rotors is summarized in Table 11.2 for 10 MJ storage but it must be noted that these values are for the inertia element only. For the composite rotor, metallic parts must usually be added to connect to bearings and the motor generator whereas steel rotors, these parts are small given the rotor extends to the centreline. The substantially lower weight of the composite rotor does bring great benefits in weight-sensitive applications, so most flywheels developed for motorsport were of composite design and often true for space applications. For most other applications, cost is usually the dominant factor and composites have much higher material costs than steel, not just raw materials but in terms of the high cost of filament wound manufacture. Other issues include the cost and weight of the containment which may be much greater than that of the rotor as explained later.


Motor generator topologies

The motor generator (MG) is an electric machine generates a shear stress on the magnetically active surface of the rotor that creates torque on the rotor. This surface is illustrated in Fig. 11.7. The most common topology in Fig. 11.6 denoted “conventional” has a cylindrical active magnetic surface and calculation of its key dimensions is an important first step in electric flywheel design.

rMG LMG Figure 11.7 Dimensions of the motor generator magnetically active surface.

Flywheel energy storage


If the maximum magnetic shear stress J is known for a particular design of MG and the flywheel rotor speed is known, the rotor dimensions can be determined for a given maximum MG rotor peripheral speed UMGmax. Firstly, rMG ¼ UMGmax/umax and then the length of the MG rotor can be found using Eq. (11.16) by knowing the maximum torque Tmax which will occur at the lowest operating speed of the flywheel as explained in Section 2.5. LMG ¼

1 Tmax 2p JrMG


Typical values for J are between 10 kN/m2 and 20 kN/m2 and maximum peripheral speeds can be up to 200 m/s. The most common topology shown in Fig. 11.6-2(A) denoted “conventional” has an external active magnetic surface and is the topology of choice since the vast majority of electric machines are designed this way. It may be attractive to adapt an existing machine for use in a flywheel and it will operate in a similar way apart from its vacuum atmosphere which creates challenges for rotor cooling. On the other hand, if such a machine is already in volume production, costs should be lower. The external rotor design, Fig. 11.6-2(B), has been used particularly for composite rotors since it is convenient to place such a machine inside the hollow rotor. If the MG has permanent magnets, these can be held in compression without need of a containment ring and the magnetic airgap can be reduced. There are, however, two drawbacks. The first is the unusual end winding on the stator of bespoke design which requires different winding techniques as compared to the conventional design. The second is rotor cooling when the rotor is composite since this material has low thermal capacity and conductivity. Another option is to use an axial flux topology, Fig. 11.6-2(C), and this would be attractive if the speed of the flywheel is relatively low, meaning relatively higher torque. What can happen with a conventional machine is its axial length becomes very short and the design becomes less efficient due to relatively large end windings as compared to the magnetically active rotor length. The rotor of a lower speed flywheel will be more of a disc than a cylinder and so have a “pancake” MG which helps with compactness. The last topology, Fig. 11.6-2(D), is one in which the MG is totally integrated into the rotor and this can be done in at least two ways. A steel flywheel rotor can have features such as lobes at the outer surface to turn into a variable reluctance machine, typically switched reluctance [16]. Although this eliminates needing to add a separate MG, the rotor steel must be electrically soft and such steels will always be mechanically weaker than steels which do not have this requirement. For composite rotors, a special layer can be created in the inside in which the permanent magnetic material is embedded in the matrix. Although this will not create a particularly high magnetic shear stress, the high peripheral speed of the magnetically active surface mitigates this so the product J rMG is still high.



Storing Energy

Motor generator electromagnetic options

Six types of MG have been shown as the most common types encountered. They all share in common the need to be brushless since speeds are normally too high to allow any physical rubbing of electrical brush contacts. Even if brushes were feasible, the frictional loss would almost certainly be unacceptably high. The magnetic field on the rotor must therefore either be made by mounting permanent magnets on the rotor or by inducing them using stationary coils. One version of the first method is to mount permanent magnets on the surface, Fig. 11.6-3(A), and these must vary in polarity such that there is a minimum of two poles. Each pole can be made up of many magnet segments, however, in order to make the mechanical design easier. Since high strength permanent magnets are brittle, they must be kept in compression and so require a containment of metallic or filament wound composite material. The magnets can also be embedded into the rotor, Fig. 11.6-3(B), and the magnetic field flows through laminated electrically soft steel which itself can generate additional torque, acting partially as a synchronous reluctance machine. Using permanent magnets avoids the need for external excitation, so efficiency is generally high and little heat is generated in the rotor, such heat being difficult to remove due to the vacuum. However, a major downside is the continual electromagnetic drag when the flywheel is operating in standby; so energy losses can be significant if the flywheel needs to be on standby for long periods. The induction machine, Fig. 11.6-3(C), has conductors on the rotor in what is often described as a “squirrel cage.” Here, currents are induced by the coils in the stator and magnetic poles are then generated on the rotor. The rotor must rotate slower than the field generated by the stationary coils; hence, the machine is termed asynchronous. The currents in the rotor are a source of significant loss which reduces efficiency and generates heat which must be conducted away into the rotor or removed by other means to avoid overheating of the rotor. Despite this, standby losses are eliminated once the rotor is deenergised. The switched reluctance machine, Fig. 11.6-3(D), operated on the principle of magnetic attraction of its lobbed poles. The presence of lobes which, in atmospheric air, generate significant windage loss, is less of a problem given the vacuum environment. As with the induction machine, standby losses are very low when the machine is deenergised as long as electrically soft steel is used for the rotor. Efficiencies are typically in between that of permanent magnet and induction machines. The synchronous reluctance machine, Fig. 11.6-3(E), operates on the principle of the reluctance of the rotor varying relative to the stator. This is similar to switched reluctance although the rotor construction is considerably different requiring sandwiches of magnetic and nonmagnetic metallic sheets. Efficiencies are again typically in between that of permanent magnet and induction machines and standby losses will be very low. The last type of machine shown is the homopolar machine, Fig. 11.6-3(F), not to be confused with a brushed homopolar machine which shares the same name but whose principle is entirely different and of no use for high-speed flywheels. The machine in Fig. 11.6-3(F) has two magnetically active rotor drums (or discs) which are energized

Flywheel energy storage


using a simple coil such that one rotor is south, the other north. By having lobes on the magnetically active part, this has the effect of changing the field in the tangential direction so stationary coils are then able to generate torque on the rotor. As well as the ability to eliminate standby losses, the rotor field strength can be varied easily such that when the flywheel is at higher speed, the back emf can be reduced making the design of the power electrics easier. However, the machine is relatively bulky and requires a bespoke design. More information on electric machine choices for flywheels is given in Ref [5].


Bearing systems

The bearings on a flywheel system perform a critical duty of supporting the rotor which is subjected to load from its weight and, in the case of nonstationary systems, many other loads including inertial from accelerations, linear, angular, and gyroscopic loads. The bearings must also operate in a vacuum which means any lubricants, if required, must tolerate this environment. Fluid bearings using vacuum tolerant oils are feasible in principle, but the losses may be too great given that flywheels typically have to operate much of the time in standby. Gas bearings are not an option due to the vacuum requirement. The first type of low friction bearing shown as Fig. 11.6-4(A) is the rolling element type, and it is common to choose the angular contact topology which can provide both radial and axial load. An axial preload is needed to prevent skidding of the balls in the races, but this has to be kept low as possible. The balls used are typically ceramic, which reduces friction and centrifugal loads on the balls and leads to greater bearing life. Although rolling element bearing friction is low relative to other mechanical bearing types, supporting a flywheel rotor only on such bearings is likely to lead to losses which are too high; hence rolling element bearings are usually used in conjunction with magnetic bearings are described later. For nonstationary systems, rolling element bearings are almost always necessary in order to withstand high but intermittent loads resulting from movement of the flywheel system, for instance, on a vehicle. Rolling element bearings have a considerable ability for withstanding peak loads relative to their mean load. This contrasts with magnetic bearings which can support the weight of the flywheel rotor but have low stiffness, so have limited ability to withstand dynamic loads. Rolling element bearings inherently have low damping so it is usually the case that this must be introduced at their mounting, an effective design solution being shown in Ref. [5]. Magnetic bearings work on the principle of either magnetic attraction or repulsion, Fig. 11.6-4(B)e(H), and can work in the axial direction or to create a radial bearing forcing the shaft axis onto the bearing center. The field can be created either by permanent magnets or by means of coils but only the latter are directly controllable by means of varying the current in the coils. Permanent magnetic bearings are called passive type and are particularly attractive for the flywheel application since they require no external services, have negligible loss, and are not affected by vacuum. Supporting a flywheel only on passive type magnetic bearings would appear to be an ideal solution for stationary systems but this is not stable as explained by Ernshaw’s theorem [17].


Storing Energy

The simplest type of passive magnetic bearing, Fig. 11.6-4(B) consists of two annular magnets magnetized on a coaxial direction oriented such that the same poles face each other. This configuration is stable in that reducing the gap between them increases force. However, the bearing is unstable radially and the shaft requires radial bearings with radial stiffness greater that the antistiffness of the axial bearing. One popular complete bearing solution is to use two rolling element bearings to maintain radial location of a vertical shaft and use a passive repulsive bearing to support the majority of the weight. An alternative is to use an attractive passive magnetic bearing, Fig. 11.6-4(C), which has a substantial advantage of elimination of the second magnet, the magnet that would have to be mounted on the rotor as in the case of the repulsive type. The radial antistiffness issue is also eliminated but this bearing is not stable given the axial attractive force increases with reducing gap. This solution tends to work best in combination with rolling element bearings which can also take a substantial share of the axial load leaving the passive attractive bearing to reduce the axial load rather than support all of it. The passive radial type, Fig. 11.6-4(C), requires stacks of annular magnets on both the shaft and the stator with the fields for all magnets aligned radially. The annular magnets in the gap have like poles so the forces act to keep the shaft centrally aligned. However, if not constrained, the shaft will move axially, so, without independent axial constraint, is unstable. Moving to the active type, an active radial magnetic bearing, Fig. 11.6-4(D), uses coils to attract a ferromagnetic shaft and would normally be unstable but for use of a fast control system with position sensors which attracts the side of the shaft in which the gap is becoming larger. The bearing restoration force can be controlled by the charging the current in the stationary coils. In spite of the complexities of control, use of such bearings is very attractive given bearing maintenance can be eliminated which is not possible with rolling element bearings whose lubricants degrade, and elements will eventually suffer from wear and fatigue. The active axial type can also be used, Fig. 11.6-4(E), but requires coils on either side of a disc which can only attract the disc not repel it, so it works in a similar way to the radial active bearing, attracting the side that is moving away rather than creating a repulsion force. The last type of bearings is based on use of superconductivity, Fig. 11.6-4 (E), for radial and Fig. 11.6-4 (F) for axial. Here, magnetic fields created currents in superconducting coils in the stator which repel permanent magnets mounted on the rotor. Such bearings are inherently stable and low loss but have a substantial disadvantage in the need to generate cryogenic temperatures. If higher-temperature superconducting materials are developed, this type of bearing could be very useful as a solution for future flywheel systems [18].


Flywheel system installation

As reported earlier, flywheels have been used extensively in transport applications, mounted on vehicles, Fig. 11.6-5(A). There are a number of additional challenges for this type of installation and these include the imperative to minimize weight, mitigation of specific risks, and accommodation of additional loads on the bearings. The need for low weight naturally favors the use of composite rotors although the weight

Flywheel energy storage


of the whole system inclusive of containment must be taken into account. Composite rotors suffering a major failure can release very high pressures inside the containment, and in a transport application there is no option but to contain this. Relatively lightweight solutions have been developed to avoid unacceptably high-weight metallic containment but at additional cost. Containment of the rotor under self-induced failure is not the only issue, since the vehicle itself could put the flywheel under high loads in the event of a vehicle crash. This challenge is faced by engineers designing any on board vehicle energy storage system, be in electrochemical batteries or fuel, and requires identification and assessment of the risks by means of failure modes and effects analysis (FMEA). Fortunately, the energy stored in vehicle mounted flywheel systems is typically low being of similar magnitude to the kinetic energy of the vehicle operating at a moderate speed. This is orders of magnitude lower than the energy stored, which would be needed to power the vehicle for long distances, i.e., the battery of an electric vehicle or chemical fuel for a bus, truck, or train. The additional loads on the flywheel are due to acceleration of the vehicle in any of the six degrees of freedom and could be several times the gravitational acceleration, g. These loads typically have short durations and can be reduced by attaching the flywheel in compliant mountings. Much concern is often made of gyroscopic torque which occurs when the axis of the flywheel rotor is subjected to angular velocity in any direction other than around its own axis of rotation. The maximum gyroscopic torque generated, Mgyro, is given by Eq. (11.20); Mgyro ¼ Iumax uprecession


The gyroscopic torque creates loads on the two radial bearings and this is also transmitted to the vehicle. Generally, the magnitude of the gyroscopic torque is low relative to other torques on the vehicle. For example, a 500 kJ useable energy flywheel is designed to store the kinetic energy of a 1500 kg car up to speeds of 66 kph. With maximum speed of 5000 rad/s (47,747 rpm) it has a moment of inertia of 0.05 kgm2. If it has a horizontal shaft, a severe vehicle yaw angular velocity of 1 rad/s from a skid in cornering will generate a gyroscopic torque of 250 Nm. This compares to rollover moment for the vehicle in the order of 10 kNm. Of more concern is the load on the bearings which will be 1250 N on each if the bearing separation is 0.2 m and this will normally dictate a need for rolling element bearings. It is possible to mount the whole flywheel in a gimbal such that the gyroscopic torque is not generated but this adds cost and a weakness in constraining the flywheel system in the event that the flywheel bearings or shaft fails, and the flywheel rotates loose in its casing with potential to generate large forces. The stationary installation shown in Fig. 11.6-5(B) is less challenging and still can be used for some transport applications as explained in Section 5. A flywheel system which can operate safely above ground is attractive, since it can be placed in any location and moved as required forming a containerized solution. Here, the bearings only need to take load of the weight of the flywheel and out of balance except during earthquakes which only applies to certain regions of the world


Storing Energy

on rare occasions. Still this situation must be mitigated if the installation is in such regions. Since the installation is above ground, the containment must withstand the worst kind of rotor failure and/or prevent significant movement of the system after a major bearing failure or shaft breakage. The latter is less catastrophic but more likely and if the flywheel is rotating loose within the casing, the rotor with casing must be allowed to vibrate to prevent large forces being generated which could allow the whole system to break free from its mountings. This can be solved by supporting the main containment casing in a strong but compliant mounting. The question is how heavy and bulky the containment has to be in order to contain the rotor in the event of burst? One estimate for containment of full burst is around 10 times the mass of the rotor [19]. If the rotor is one-piece monolithic steel then growth of a crack could lead to full burst and release of many large chunks, this type of failure demonstrated in Ref. [7]. Composite rotors do not fail in this way but break into very small fragments which is more benign. However, under certain conditions an explosive failure mode may occur releasing very high pressures in the order of thousands of atmospheres. This can then fail the containment which then also allows release of rotor debris. More dangerous perhaps than the rotor debris is the shrapnel of the containment that is ejected. It is possible to design a casing capable of containing this type of failure as long as the composite rotor has a relatively low diameter so the wall of the containment, still thick, is feasible. Alternatively, a substantially strong but lighter containment is possible but also employing composite for the containment as well as the rotor. If only a small fraction of the rotor is allowed to fail at any one time then the casing strength can be greatly reduced hence the benefit of a laminated rotor solution or modular flywheel rotor. An alternative is to adopt a design strategy similar to that used for gas turbine engine turbine wheels in which high quality steel is used, carefully inspected and monitored such that growth of cracks large enough to cause fracture can be avoided. However, this incurs a high rotor material cost and need for rotor replacement or re-inspection at particular periods. Yet another approach is to have sufficiently high safety margin that likelihood of burst is highly improbable according to the rules of conservative engineering design codes. However, this leads to a high rotor weight, higher bearing losses, and low performance. The preferred solution for high performance solid steel cylindrical rotors and composites is shown in Fig. 11.6-5(C). Here, the large containment mass is provided in the form of concrete and soil. In the case of a rotor failure, the rotor is allowed to burst through its containment and the fragments, along with the remains of the rotor, are safely contained underground [8]. Clearly a strong lid is also needed to prevent any debris shooting upwards [19] but the installations, typically large, can be fenced off from the public. Each flywheel has to be placed at a distance from its neighbor to avoid a cascade failure whereby one burst triggers another if fragments passed through soil so bunkered designs typically take up large areas and each flywheel is large. Examples of such installations can be seen in Refs. [7,8,19]. The disadvantages of this approach are the large real estate needed and the permanence of the installation, but in some locations and applications these may not be an issue.

Flywheel energy storage


4. Performance attributes in comparison with other electrical storage technologies The purpose of comparing electrical energy storage systems with each other is to identify which technology will meet the requirements of the application and do this at the lowest cost. These two are dealt with in turn and set the context for describing where FESS sits within the energy storage landscape.


Performance requirements of electrical storage technologies

In any electrical power network, supply and demand must be balanced at any instant. In most cases, loads are rarely controllable by the grid whereas nonrenewable supply technologies could be controlled and could react very quickly to maintain stability. For example, in grid networks, large steam plant turbo-generators can provide additional power for short periods or reduce output since the inertia of the turbo-generator provides or absorbs energy acting as a large flywheel. However, now that this steam plant is replaced by renewables, the supply has also become uncontrollable, adding to uncontrollable demand, so there is a growing need for electrical storage including fast acting capability as well a capacity. In islanded small networks, a diesel engine operating in load following mode can also react quickly to balance changing loads but introduction of renewables such as solar PV or wind to such systems means fast response energy storage is essential in order to maintain system stability. At longer timescales, if supply and demand do not match, energy must be stored and retrieved for longer periods. Here, fast response is not necessarily needed if this requirement has already been met by other technologies. For example, a pumped hydro storage system can provide a few hours of storage but requires time to reach full output. It must therefore be coupled with a fast response technology in the grid system, be it the inertia of steam plant, an electrochemical battery system, or FESS. Even pumped hydro cannot cope with the durations of days and even weeks that will be needed in a system dominated by renewables and the only solution is an engine converting chemical energy in fuel into electrical power. Excess renewable power that cannot be stored can always be dumped or generation curtailed for long periods despite this being wasteful in the shorter term. Eventually will be used to generate renewable (green) hydrogen by electrolysis. With renewable fuels, an engine plus electrolysis can now be considered to be a storage system as opposed to being way generation as is the case for fossil fuels but the engine and fuel generator are not the same. The two critical characteristics of the main energy storage system technologies of response time and duration are shown graphically in Fig. 11.8. To meet the fast response requirement, FESS competes with ultracapacitors and electrochemical batteries. The response time for all three is dictated by the power electronics but this can achieve full power ramp up within mS. The most battery types are lead-acid and lithiumeion, but the latter is, of course, a family of chemistries each with


Storing Energy



Lead Acid (VRLA)


Lithium Ion baeries Idling (incurs losses)

Flow baeries

Flywheel (FESS) Ramp up

Compressed/liquid air

Electromechanical On standby (incurs losses)

Ramp up


Ramp up

Ramp up

Fuel cell with renewable or clean fuels – ie green or blue H2

Idling (incurs losses)


Pumped hydro


Recip or gas turbine engines with renewable or clean fuels – ie green or blue H2




1week 1month

Chemical to electrical (engines unidireconal)

1 year

Figure 11.8 Energy storage technologiesdresponse and duration.

different characteristics. The duration shown is typical for lithium phosphate (C0.5 rating). There is an inbuilt assumption in Fig. 11.8 which sets the duration for each technology at its most effective point according to its specific energy and specific power costs. A low running loss FESS could, in theory, physically operate for 8 h at a C0.125 rating but since its specific energy cost is much greater and Li-ion, this would not make sense. Equally a Li-ion battery can produce very high power for less duration but most high power/short duration applications have many daily cycles, so this would rapidly degrade the cells. Li-ion can be operated for short durations and many cycles if depth of discharge (DoD) is low but requires overcapacity as explained in Section 4.2. The other six storage technologies shown require anywhere between seconds or minutes to ramp up so cannot meet the ms level fast response requirement. Some technologies can have shortened response times by being on standby, for instance, the pumps are turned on for a flow battery, a pumped hydro turbine kept spinning in compressed air, an engine could be left on idle. However, the cost is significant in idling energy losses which will mount up when idling occurs over long periods. This leads onto another issue of standby losses and in many publications, FESS is often listed as a high standby loss technology which can be true for certain designs with high electromagnetic drag in the motor generator or poor bearing systems. Although VRLA and Li-ion cells have negligible self-discharge, a well-designed FESS can have a similar standby loss as a Li-ion based system in many installations, since the latter requires the operating temperature of the cells to be carefully controlled, with heating and cooling auxiliary systems drawing power depending on the ambient temperature. FESS is unaffected by ambient temperature with motor generator cooling carried out using a simple water-cooling system with heat exchanger. Ultracapacitors have a similar self-discharge to FESS but are affected by temperature as with batteries and, unlike FESS, suffer calendar degradation

Flywheel energy storage


FESS has another benefit over Li-ion and ultracapacitors in accurate determination of state of health and SoC. As long as the rotor is spinning without issue and the speed is measured, both are known with complete certainty this reducing availablity risk. In summary, FESS essentially competes with ultracapacitors and electrochemical batteries and all these technologies must be combined with other long-duration technologies to provide a solution across the entire duration spectrum. A solution across the duration spectrum can only be found by what is essentially hybrid storage by combining a number of the technologies shown in Fig. 11.8. These technologies do not need to be of the same rating or in the same physical location but where they are placed, either small-scale distributed or large centralized, has a significant effect on the cost of the whole grid supply system. This issue in general is addressed for the whole grid in Ref. [20]. With their higher durations, it would seem that batteries can totally outclass ultracapacitors and FESS. For many applications, this is the case with Li-ion able to provide both rapid response and sufficient duration to perform many other functions. For example, in a grid, it can provide rapid response services for grid stability but at the same time can earn additional revenue through secondary response, arbitrage, black start for example. However, the one characteristic in which FESS can outperform Li-ion is high cycle and calendar life and if this characteristic is needed, FESS can be a superior option. Could FESS provide the rapid response function leaving technologies such as pumped hydro, compressed air to provide the long-duration storage? This depends on the availability of lower-cost FESS and this issue is dealt with in the next section under costs.


Cost comparison of flywheel energy storage systems with other technologies

From Section 4, FESS is identified an alternative to ultracapacitors and Li-ion batteries so it only makes sense for it to be compared in cost with these technologies. Cost should not necessarily be measured only in financial terms but should also take into account the cost to the greater environment. It would be quite absurd to replace fossil fuelebased systems with alternative technologies only to find the earth’s ecology is damaged or ethical principles are ignored in its manufacture. Of course, this might be a price worth paying in the short term if negative effects of particular technologies can be resolved in the near future. The concern and debate about difficulty in recycling Li-ion batteries and use of supply chains for particular materials with questionable ethics will continue but perhaps for now is considered a lesser evil than the urgency of climate change. With its lack of degradation and use of sustainable materials, FESS offers a much better alternative for some applications currently met using Li-ion. In spite of these issues, financial cost is really the dominant driver of the choice of which technology is adopted and unless this is changed, FESS only succeeds when it offers the lowest overall cost for a particular application or there are unique circumstances which prohibit alternatives. Financial cost is addressed first followed by some


Storing Energy

further consideration of sustainability and the main comparison is with Li-ion, the technology which has mainly driven FESS and ultracapacitors from the market in fast response storage to date. Many publications compare the costs of storage technologies in both specific energy cost ($/kWh) and specific power cost ($/kW) and typically cite what is commercially available as a basis for this. Although this is useful as a starting point, this misses two important issues. The first is degradation of the storage asset, linked closely with the application. These issues have been dealt with using the concept of Levelized Cost of Storage and Schmidt et al. [21], as an example, capture the costs of degradation as well as capital, operating cost, and several other important issues. This analysis shows that for high cycle applications, FESS is the lower cost alternative to Li-ion and ultracapacitors. The second issue is that cost is a strong function of manufacturing volume. Of course, if a technology cannot reach sufficient volumes to achieve costs below the alternative, the technology will remain a niche solution. FESS previously had cost advantage in specific power cost ($/kW), but this has been nullified by recent substantial cost reductions for Li-ion just at the time when the market for electrical energy storage has expanded exponentially. Schmidt et al. [21] capture further cost reductions for Li-ion based on continued investment but base costs for FESS on development of current commercial offerings which are in low volume manufacture. Cost reductions in all technologies have been estimated based on the amount of investment in each. This approach does not capture innovations in new design approaches with potential for stepwise cost reduction in FESS or any other technology and this is more or less impossible to predict. Given costs of commercially available FESS are high, it is unlikely that FESS will be competitive in many potential applications unless some other factors come into play such as lithium and cobalt prices are driven up by supply issues. This cannot be assumed and it is worthwhile exploring how FESS and Li-ion can be properly compared for high cycle storage applications in which FESS may have the advantage. Even if an application only requires 1 or 2 minutes of storage duration, the power cost of Li-ion is now lower than FESS, something which has been driven by the “Giga factory” volumes of Li-ion cells needed to satisfy the electrification of transport and grid storage. High volumes have also driven down the costs of battery management power electronics and conditioning systems which are essential for reliable and safe systems. Li-ion cells have a full depth of discharge (DoD) life of around 1000e2000 cycles although some chemistries have higher life, albeit at much greater specific energy cost and lower mass specific energy. This life is completely adequate for say a battery electric vehicle which will do 250 miles per charge so 1000 cycles is 250,000 miles. However, what if the application requires 50 cycles per day and storage system is expected to last 10 years? This is 182,500 cycles, two orders of magnitude higher than full DoD life of Li-ion cells. Other, higher cycle life Li-ion technologies do exist but these are much more expensive than those in mass manufacture. It is possible to satisfy this demand using Li-ion of a moderate life chemistry such as lithium phosphate by having substantial over capacity. This approach takes advantage of the characteristic of Li-ion cells in which degradation is considerably less when DoD range is kept small and

Flywheel energy storage


Cycles to 80% capacity (Log scale)


C1 rate, ideal temperature Curves for higher temperatures or C rates


1000 0











Depth of Discharge (DoD)

Figure 11.9 The effect of depth of discharge range of Li-ion cycle degradation.

mean state of charge (SoC) kept within a narrow band away from both high and low SoC. This effect is illustrated in Fig. 11.9 for a typical cell and an empirical equation can be fitted to any manufacturer’s data of the form: Cell cycle life N ¼ k1 DoD k2


where k1 and k2 are constants. It is, however, known that the cycle life will vary considerably according to the mean DoD. For example, a battery cycled between 80% DoD and 100% DoD or 0% DoD and 20% DoD will have considerably less life than one cycled between 50% DoD and 70% DoD. Batteries held at low or high DoD suffer additional life reducing effects, so the use of Eq. (11.21) is simplistic, likely optimistic, but is useful in understanding how Li-ion can be used for high cycle applications. Cycling with a DoD range of 80% capacity will degrade the cell in around 2000 cycles, red line, whereas cells of 4 times the capacity will yield around 10 times the life, green line since only 20% of the full capacity needs to be used. Now to reach something like 186,000 cycles, the DoD range must be very small, beyond the graph, but would be perhaps 0.05 DoD range. This would mean the battery capacity must be increased another 4 times to 16 times the battery giving 2000 cycles over an 80% DoD range. So, for this particular application in which Li-ion meets the cycle life by having an overcapacity factor X, the cost of FESS must be less than X times Li-ion to be competitive. In order to do a comparison, the C rating must also be defined for an application and let it be assumed to be 6 min (C10) which is a little lower than ideal for FESS. A stationary Li-ion system power cost is currently around $400/kWh (C1 rating) for industrial systems [22], so in order for FESS to be competitive, it must be lower than 16  $400 ¼ $6400/kWh [21] indicates that costs for current commercial FESS are predicted to fall below this but given current investment in FESS is so much lower than in Li-ion, costs for Li-ion are likely to fall more quickly than FESS so the current advantage will be eroded. The conclusion can only be that a step cost reduction is needed for FESS by virtue of fundamentally lower cost designs in terms of materials, assembly, and safe containment. Interestingly, this simple analysis can also be


Storing Energy

used to estimate the costs FESS would need to reach in order to compete head on with Li-ion for applications with 1 cycle a day/2e3 h such as backing up solar. In 10 years, this would require 3650 cycles and a Li-Ion battery could meet this DoD range of 70% meaning little over capacity will be needed. In reality, given the variation is daily solar output means many cycles will not be full, the overcapacity needed would be even less. Given it requires around 200 kg of steel for 1 kWh of storage assuming a steel type rotor, FESS storage including rotor and containment will have raw material costs of $70/kWh assuming $350/tonne for steel. Added to this are all the other costs of the system so achieving costs of less than $400/kWh for Li-Ion is unlikely. A better alternative is to hybridize FESS with very low $/kW technologies such as PHES and CAES and use the FESS only for fast response. A cost estimate was carried in Ref. [23] for FESS based on steel laminated rotors, avoidance of bunkers, and showed the costs of $1400e2200/kWh and $2300e4000 kWh could be achieved for C30 and C12 ratings, respectively. Given that many high cycle, short-duration applications are emerging, FESS has a future as an alternative to Li-ion when the latter must have substantial overcapacity in order to give the necessary cycle life. Returning to issues of sustainability, a useful quantitative measure is Energy Stored on Investment (ESOI), the ratio of energy throughput in an energy storage device to its embedded energy required in manufacture [24]. This is described in equation form as; ESOI ¼

Energy stored ðcapacityÞlhD lhD ¼ ¼ Embodied energy ðcapacityÞεgate εgate


where l is the number of cycles during the storage system lifetime, h is the round-trip efficiency, D the depth of discharge for the given cycle life and εgate is the embedded energy in manufacture of the system. Analysis in Ref. [24] shows that mechanical storage systems such as pumped hydro and compressed air have ESOI values of around 200e250 whereas electrochemical batteries are around 2e10. Analysis of flywheels is not included in Ref. [24] but Eq. (11.22) yields an ESOI value of 264, assuming a conservative value of 400 k full discharge cycles (actual or equivalent), round trip efficiency of 88% and embedded energy of 24 MJ/kg. The latter estimation is 20% greater than the value of steel on the basis that a small proportion of the FESS mass is made from other materials. Any other values could be used but are unlikely to significantly reduce what is a two ordes of magnitude advantage over batteries.


Current and future applications

FESS has been proposed and used for many applications and these can be divided into two major areas, stationary and on-board vehicle as per Figs. 11.6 and 11.10 below. Within this division, several applications are listed and described in turn and avoiding replication of many other publications giving more details on applications such as

Flywheel energy storage


Flywheel energy storage (electrical transmission)

5.1.9 Rail (Trackside)

5.1.1 Electric vehicle fast charging

5.1.8 Islanded grid storage

5.1.2 Dockside cranes stac

5.1.7 Mining stac


5.1.3 Grid storage fast response services

5.2.1 Vehicles for construcon and mining

5.2.8 Marine 5.2.7 Rail (on vehicle) 5.2.6 Hybrid vehicles

5.2.2.Space Mobile 5.2.3 Dockside cranes mobile

5.1.6 Research facilies 5.1.4 UPS - DRUPS 5.1.5 UPS - hybrid

5.2.5 Motor racing

5.2.4 Electromagnec aircra launch

Figure 11.10 Applications of flywheel energy storage with electrical transmission.

[2,5,25] to cite a few. It is possible that some have been missed and it is likely than new ones will emerge given the revolution currently underway in the power and transport industries.

5.1 5.1.1

Stationary applications for FESS Electric vehicle fast (ultra) charging

The uptake of battery electric vehicles (BEV) is rising exponentially due to BEV total costs reaching parity with internal combustion engine vehicles (ICEV) combined with a spate of government policies targetting cleaning up of urban air with the ICEV being identified as the main culprit. BEVs themselves are superior to ICEVs in performance and driver experience, the low-cost Li-ion battery providing the breakthrough for this. However, charging is still a major issue, less with the Li-ion battery but with the supply of high powers to the charger. For BEVs to be a practical solution for all motorists, not just “two car” families with off street parking, ultra high power chargers with 100350 kW must be available over a wide coverage area. The next generation of BEVs will be capable of accepting very high or “ultra” charging levels but for the grid provide what is a very high and peaky power demand in many locations will be challenging. To put this into perspective, a 350 kW charge power is the same as the average power of several hundred dwellings. FESS has been suggested as a solution given the majority of drivers have short journeys so are looking for “top-up” charges Ref. [24] of 50e100 miles which can be done in 2e4 min at ultracharging rates. The C rating is between 30e15 and if the charging is accessed only 30 times a day, cycles over 10 years are over 100,000. Technology for this application is being developed by Refs. [26] and [27].


Dockside cranes and container handling equipment

In ports, dockside cranes, typically rubber tyred gantry cranes (RTG) continually lift containers from ships to dock and vice versa. Containers are also handled by other


Storing Energy

machines within the dock, and so there is potential to recover energy and resupply it using energy storage. Not only is this good for energy efficiency but it also allows reduction in size of the prime mover. Even as diesel engines are being replaced by electrified supplies, on board or local storage can reduce the size of cabling. To give an idea of the requirement, a 40 tonne container lifted 6 m in 30 s equates to 0.65 kWh and if the operation is repeated 6 min on average, this is 240 cycles per day so would require a storage system with a C120 rating. Container handling vehicles will require storage but this is lower due to lower heights. As well as FESS, ultracapacitors and Liion have been offered for these applications but it is noteworthy that Li-ion-based solutions have storage capacities nearly two orders of magnitude greater than needed in order to give sufficient cycle life limiting DoD range to micro cycles. FESS applications are described in Refs. [2,26].


Grid storage applications

The exponential demand for grid storage is currently being satisfied by large scale Liion installations, grid storage being previously dominated by PHES. The main reason for this is because Li-ion is able to provide both rapid response services and those which require a durations of up to 2 hours. Revenue from these services can be stacked so the cost of the installation is covered by more than one income stream. Although mechanical solutions such as PHES have lower specific energy costs, they cannot provide fast response services. PHES suffers from the issue of having limited places in which it can be installed whereas CAES suffers from higher costs due to low volumes relative to Li-ion, and it has to be built on a large scale. In spite of this, there is great interest and investment in mechanical storage technologies, since these will ultimately offer the lowest specific energy, have greater durations than Li-ion, and also avoid the materials supply and sustainability issues of Li-ion. Now, if there is considerable penetration of slow response mechanical technologies, FESS could provide the fast response services with large numbers of smaller flywheels arranged in arrays. Two mechanical technologies, for instance, FESS with CAES may squeeze out Li-ion but only if the levelized cost of both together is lower which requires investment and further technological development in mechanical storage. An example of a grid-scale FESS is described in Ref. [29] and it is still operating. An alternative proposed by Ref. [30] is to use a very large flywheel driving a large grid-scale generator via a continuously variable transmission. This avoids the need for power electronics so the generator can react completely instantly in the same way as a directly connected large generator of a power station as described in Ref. [30].


Uninterruptible Power SuppliesdDRUPS

An Uninterruptible Power Supply (UPS) is required for any building in which the electrical supply must be maintained at all times. Examples include hospitals and data centers, but there are many more. The cycle requirement here is not high at all, it may be never used so one may ask why FESS could possibly be appropriate. The answer lies in the uncertainty in how long the outage will continue and the risk the customer is

Flywheel energy storage


willing to take. If the outage was guaranteed to be less than 1 h then VRLA batteries offer the lowest cost solution, but if there is a chance power may not be restored and the situation would go on for hours or even days, the only option is an engine and fuel tank. A hybrid system is therefore needed and is typically diesel plus VRLA. The problem is that diesels require a few seconds of time to start so one solution is diesel rotary uninterruptible power supply devices (DRUPS). Here, energy is stored in a flywheel which is kept spinning using power from the supply which has sufficient energy to turn the generator until the diesel takes over. Although this system is low cost and effective, energy is wasted spinning a flywheel which is not in vacuum and whose bearings support the entire weight of the rotordi.e., has no magnetic levitation. Another issue is that if the engine fails to start first time there is no backup unless a second or even third system is installed albeit at higher cost and space. It is likely that diesel will be phased out for UPS in urban areas due to issues of toxic emissions and due to risk of DRUPS engines not starting, FESS in its electrical form may be better as explained in the next section.


Uninterruptible Power Suppliesdhybrid

To give unlimited duration for UPS, an engine is needed to convert chemical fuel to electrical power; such chemical fuel can be refreshed every few days by delivery. The current engine solution is a diesel using fossil fuel, but this can be replaced by renewable fuels, liquid, or gas. If legislation starts to prohibit use of internal combustion engines, then the only other option is a fuel cell or perhaps a microturbine, given toxic emissions are very low. In all cases, a rapid response technology is needed, so again the options are ultracapacitors, batteries, or FESS. Cycle life is not an issue unless the storage technology is also being used for maintaining high power quality as would be the case in a data center but still the DoD for this service is very small. The main issues are cost and reliability. Given that power is only needed for bridging until the engine comes on stream, durations of only 10’s of seconds to perhaps minutes are needed and specific power cost is the important measure. State of charge and state of health are also issues since it is vital that the system delivers immediately when there is a break in supply. FESS can provide a lower specific power cost than Li-ion, but it would be more challenging to beat the power cost of VLRA batteries. However, both battery solutions require greater space and need to be placed in climate controlled space. There may are also safety issues of placing large Li-ion systems in buildings and VRLA has reliability issues associated with unknown state of health.


Research facilities

Many research facilities require substantial power levels which exceed the capacity of the local grid, but this is needed intermittently so it is more cost-effective to install local electrical storage. Although a niche application, the power demand can be very high so the value of one installation is large. The most well-known example of FESS used for research is described in Ref. [2] in which a very large flywheel was designed and constructed, and placed in a bunker for safety. Such bespoke solutions


Storing Energy

are very expensive and once beyond their service life could potentially be replaced by arrays of smaller FESS systems or other rapid response storage systems, which would be the best technology depending on the duration of power needed, i.e., the C rating.



Mining can take place in open cast sites or underground mines and often this is done at remote sites away from grid infrastructure. In all cases, power demand will be intermittent giving lifting of discrete large packets of material. Typically, an array of diesel engines would provide the power in load following mode but now with concern over climate change this is being scrutinized. There is potential for reducing energy generated from fossil fuels by adding renewables such as solar PV, but this creates intermittency in the power supply as well as the demand. There is hence a need for local energy storage but which technology is best much depends on the details of the system, particularly the ratio of renewable to controllable power. The problem of fossil fuels can be mitigated by using renewable fuels in a diesel engine and the issue of toxic emissions is less of an issue due to the location away from urban centers. Whether FESS offers a lower-cost solution compared to Li-ion will depend on the details of the system and other factors such as placement of storage. If this must be inside a mine, the potential fire risk of Li-ion would give advantage to FESS.


Islanded grid storage

In places with sparse populations, power is typically provided using diesel engines which are able to follow the load and react to balance supply and demand. Use of fossil fuels can be displaced by generating energy using solar PV or wind but substantial savings only occur when the engine can be switched off. The problem is that if the demand exceeds the renewable supply, the system will fail. Even a cloud passing over a solar array in a fairly sunny day could trigger a problem so rapid response storage is essential. Given the need to provide reliable power when solar and PV are not available, the diesel engine system needs to be available at all times but needs 10’s of seconds to start. The question then is what duration the storage needs to have, should it only bridge the time needed for the engine to start rather like a UPS system or should the storage last perhaps an hour such that it can bridge the gap to avoid the engine starting again since the renewable source has come back on stream. FESS has been explored for this application [31] and may be very suitable for locations which are remote, suffer extreme ambient conditions, or if replacement parts such as Li-ion cells are hard to obtain.


Trackside rail storage

Rail electrification has been increasing worldwide as the best solution for decarbonizing this form of transport. With connection to the grid, one might think that storage would not be necessary, and this is the case particularly for intercity and crosscontinental journeys. However, when the distances between stations is less than a few km/miles, the amount of energy needed for acceleration and regenerated in braking becomes significant and results in significant spikes in power demand or

Flywheel energy storage


reinjection into the grid in the order of low MW power. This is less welcome particularly as the grid struggles to balance supply and demand given the other changes taking place. If the operating voltage is low, for example, electrified DC rail of hundreds of volts, transmitting this power is inefficient and energy is often dumped into resistors. If this occurs in tunnels, then this leads to another problem of heating of the air in the tunnels leading to passenger discomfort and/or higher air conditioning load. One solution is to have energy storage placed at the stations and given the duration of acceleration and braking is around 40e60 s and cycles around 70e100 a day (a 10e15 min service level over 18 h), this puts the application firmly as FESS applicability. A further application of trackside storage is the ability to additionally power main city stations, allowing trains to accelerate faster and improve station capacity. A number of systems have been piloted as described in Ref. [32], but the idea has not yet been deployed in large numbers and perhaps requires a low-cost FESS solution to make it more attractive.


Mobile applications for FESS

As with stationary, known mobile applications are shown in Fig. 11.10 and are described in turn.


Vehicles for construction and mining

Vehicles in construction and mining are currently powered by diesel engines and hydraulic actuators are used to perform functions of lifting and applying large forces. Flywheels have been investigated for energy storage with mechanical connection via hydraulic or continuously variable transmissions [4,31] but this did not go beyond demonstrator stage. As vehicles are electrified in order to eliminate fossil fuels there will be a need for energy storage. If the vehicle has a relatively large battery then it is likely to be able to cope with high power demand without damaging life, but given that the range and speed of these vehicles are low, it is less likely to be the case and short-term added power will be needed. FESS can provide this but will need to compete with ultracapacitors. FESS has the advantage of being more rugged, and so may be a good choice for these applications given the harsh environments.



FESS has been used in space for energy storage as described in Ref. [2] and since low weight is paramount and accommodates higher costs, designs with composite rotors are most suitable. In space the presence of vacuum is an advantage and bearing lubricants for operation of high-speed rotors in space were developed especially for this application. FESS can also be used for angular positing of space vehicles since controlled movement of the flywheel relative to the craft will impart a gyroscopic torque which can be used to change the angular attitude of the craft.



Storing Energy

Dockside cranes mobile

This application has already been mentioned in Section 5.1.2 and there is perhaps a blurred distinction in that an RTG may be stationary while performing a lift yet it is a vehicle. Its own kinetic energy, when moving as a vehicle, could be captured in braking into an FESS system.


Electromagnetic aircraft launch

In order to assist the launch of military aircraft from an aircraft carrier, steam catapults are normally used. This takes advantage of the stored energy in the steam boiler which has not yet been passed into the steam turbines. However, this can be replaced by an electromagnetic launch system and given the energy is required over a few seconds and numbers of cycles could be high, this represents a good application for FESS. Li-ion might be better but it is likely that a large Li-ion battery could be a fire risk on a warship. More details of electromagnetic aircraft launch (EMALS) are given in [2].


Motor racing

Given the need for “green” motorsport, kinetic energy recovery or KERS became a subject of great interest leading up to the 2009 Formula 1 season and continued, while in parallel Formula E grew alongside although it has not gained the same public interest as Formula 1. FESS, ultracapacitors, and Li-ion high C rate technologies were all investigated with, for FESS, a focus on composite rotor-based system for reasons of lower rotor weight. The power system duty cycle for motor racing consists of essentially three conditions, maximum acceleration, maximum braking, and then coasting around corners. Although high power is needed to maintain speed at the end of a longer straight, this condition is less frequent than accelerating and braking with circuits designed with many corners. What would be best from a standpoint of physics is quite different from the technology chosen, since all motorsports, particularly F1, are highly regulated by a set of rules governing what is allowed in terms of the power system. Often the rules are set to satisfy other demands than create a powertrain matched to what is needed to propel the vehicle around the track at the fastest possible speed. As one example, energy recovery power was limited to 100 kW in the 2014 rules which tipped the scales in favor of high-power Li-ion systems. The battery could be replaced in every race unlike the internal combustion engine so there was little benefit in having a longer lasting FESS system. FESS was successfully used in the 24 h Le Mans race and the requirement here for cycle longevity was a likely factor in its success. For Formula E, the battery is able to provide the powers necessary, since it has a relatively high capacity and does not undergo too many cycles for this to be of concern. Depending on the rules, FESS could still have an application if a motorsport class develops for the slow response and relatively heavy fuel cell and hydrogen storage technology. The alternatives would be Li-ion or ultracapacitors, but the large energy and high power that could be harvested under heavy braking would favor FESS if this is not restricted by the rules.

Flywheel energy storage



Hybrid vehicles

FESS, both with mechanical and electrical transmission, has been investigated for use in hybrid vehicles over many years, but more recently interest in mechanical transmission has diminished. There was a flurry of interest in the past decade spurred on by the developments in KERS for F1 and Le Mans, but this appears to have petered out with BEVs becoming the clean technology of choice for passenger cars and city buses. However, there is an opportunity for FESS as a means of mitigating the slow response of fuel cells and for energy recovery. Currently this application is being met by high C rate batteries, Li-ion, or nickel metal hydride. Fuel cell powered vehicles must be hybrids in order to have sufficient acceleration response, but this is rarely advertised. FESS would be particularly good for higher power applications such as trucks and long-distance buses, which require robust solutions and high cycle life. The commercial vehicle market has yet to be affected by decarbonization, but this will be next and fuel cell hybrids powered by either blue or green hydrogen offer a more practical solution than battery electric given the mileage of these vehicles. Most long journeys are made on trunk roads and relatively few filling stations will be needed on these routes given the vehicle range will be hundreds of miles. An alternative will be fuel internal combustion engine trucks and long-distance buses with renewable fuels, most likely blue or green hydrogen. When such vehicles operate in urban environments it is important to keep NOx emissions down, and this can be done by limiting engine power. There may be a good application here for FESS to provide short bursts of power such that acceleration is not compromised.


Rail (on vehicle)

FESS has been examined for use in rail on board the vehicle as has other storage systems. Since rail vehicles are by nature heavy and rolling resistance and aerodynamic drag are relatively low, having on board energy storage makes a great deal of sense. One study [13] showed that for a diesel electric train, fuel and emissions were reduced by up to 30% even for a duty cycle in rural UK. The savings were due to a combination of engine downsizing and braking energy recovery. For urban routes with more regular stops, the savings will be greater. If the diesel was replaced by a fuel cell, the FESS system could allow the fuel cell to be reduced in size and energy recovered in braking. Of more concern is the saving in hydrogen fuel which is both expensive and bulky to store on the train.



Electrification is becoming increasingly important in marine applications given the pressure to decarbonize the sector, and this can be anything from use of energy recovery systems on diesel engines, addition of renewable energy harvesting or full battery electric. These electric systems are, in effect, islanded micro- or minigrids and storage is required in order to stabilize these grids. The technology of choice is more typically Li-ion but depending on the system, flywheels could play a role if short durations are


Storing Energy

all that are needed, for instance, in conjunction with an internal combustion engine or fuel cell. For some marine applications there can be peak demands with many cycles, for instance, for heave compensation and dredgers. Another interesting flywheel application is boat roll stabilization. Here the gyroscopic effect is utilized to provide opposing torque to counter disturbances from waves. This can be very effective as demonstrated in Ref. [14].



Flywheel energy storage systems offer a simple, robust, and sustainable storage for high-power, high-cycle applications. Apart from use on the shaft of every internal combustion engine in the world they have not made it past satisfying niche applications. Each time a market opens up, the other technologies move forward in cost reduction denying openings and the volume needed to obtain low product costs. However, given the substantial growth in the overall energy storage market, the niches themselves are becoming larger and have potential to allow low-cost FESS to be developed and be commercially available. FESS must compete with the benchmark set by Li-ion which is able to perform as well as FESS in all but cycle life apart from certain applications in which fire risk is an issue. Li-ion is able to cope with high cycles if oversized sufficiently, but it is here that FESS can be competitive since this oversizing of the incumbent adds to its cost. A number of applications for FESS have been explained, all of which require fast response, short duration, and high cycle capability. Many are hybrid systems in which another storage technology or prime mover requires another system which mitigates its slow response. There are sufficient applications to build a commercial market for flywheels and a few companies do exist which manage on relatively small volumes. The market could be expanded dramatically given a steep reduction in costs brought about by innovative design which is then produced in sufficiently high volume to further drive down costs. Supply issues for lithium and cobalt driven by ethical legislation, market distortions, or geopolitics could increase the cost of these materials; so, it would be useful to have FESS as an alternative to fill the gap. If technologies such as CAES/liquid air grow to displace Li-ion, then there will be need for short-duration, fast-response technologies such as FESS as long as the specific power cost is competitive.

References [1] G. Genta, Flywheel Energy Storage, first ed., Butterworth-Heinemann, London United Kingdom, 1985, ISBN 9781483101590. [2] D. Bender, Flywheels, Sandia National Laboratories, 2015. SAND2015-3976, https:// (Accessed 4 January 2021).

Flywheel energy storage


[3] P.P. Benham, F.V. Warnock, Mechanics of Solids and Structures, Pitman Publishing, 1973, ISBN 9780273361862. [4] F.J.M. Thoolen, Development of an Advanced High Speed Energy Storage System, PhD thesis, Eindhoven University of Technology, 1993, [5] M. Amiryar, K.R. Pullen, A review of flywheel energy storage system technologies and their applications, Appl. Sci. 7 (2017) 286. [6] M.E. Amiryar, K.R. Pullen, Analysis of standby losses and charging cycles in flywheel energy storage systems, Energies Vol 13 (2020). [7] Flywheels Support Energy Grids of the Future PC Control 02, 2015. https://www.ethercat. org/download/documents/Canada.pdf. (Accessed 4 January 2020). [8] W. Sutherland, M. Senesky, W. Kwok, M. Stout, S. Sanders, E. Chiao, R. Bhat, Flywheel Systems for Utility Scale Energy Storage Prepared for California Energy Commission 2019 CEC-500-2019-012. [9] T.J. Reinhart, Laminated Rotary Structures, 1967. US Patent 3 296 886. [10] W.C. Gabrys, Stacked Disc Flywheel, 2007. US Patent 7,267,028. [11] Pullen K. R. Flywheel Assembly US Patent App 20160178031A1, China CN104025429B. [12] S. Kitade, K. R. Pullen, Chapter 3.20 flywheel, In, Comprehensive Composite Materials II, Elsevier, 545-555, doi:10.1016/b978-0-12-803581-8.03974-6. [13] K.R. Pullen, M. Read, R. Sellick, A. Fenocchi, S. Etemad, Efficiency benefits from the deployment of a novel flywheel solution for non-electrified lines, Int. J. Real. Ther. 5 (4) (2016) 79e100, [14] Seekeeper (accessed on 5 1 2021). [15] (accessed on 5 1 2021). [16] R. Rocca, S. Papadopoulos, M. Rashed, G. Prassinos, F. Giulii Capponi, M. Galea, Design trade-offs and feasibility assessment of a novel one-body, laminated-rotor flywheel switched reluctance machine, Energies 13 (2020) 5857. [17] E. Earnshaw, On the nature of the molecular forces which regulate the constitution of the luminiferous ether, Trans. Camb. Phil. Soc. 7 (1842) 97e112. [18] M. Strasik, A. Day, P. Johnson, J. Hull, Energy Storage Systems with Superconducting Bearings for Utility Applications Final, Report Cooperative Agreement No. DE-FC3699GO10285 for Department of Energy, USA, 2007. [19] D. Bender, Recommended Practices for the Safe Design and Operation of Flywheels, Presentation, 2017 accessed 5 Jan 2021, [20] D. Pudjianto, M. Aunedi, P. Djapic, G. Strbac, Whole-systems assessment of the value of energy storage in low-carbon electricity systems, IEEE Trans. Smart Grid. 5 (2014) 1098e1109. [21] O. Schmidt, S. Melchior, A. Hawkes, I. Staffell, Projecting the future levelized cost of electricity storage technologies, Joule 3 (2018) 81e100, j.joule.2018.12.008. [22] K. Mongird, V. Fotedar, V. Viswanathan, V. Koritarov, P. Balducci, B. Hadjerioua, J Alam Energy Storage Technology and Cost Characterization Report, July 2019. Pacific North West Laboratory No: PNNL-28866. [23] K. Pullen, The status and future of flywheel energy storage, Joule 3 (6) (2019) 1394e1399, [24] C. Barnhart, S. Benson, On the importance of reducing the energetic and material demands of electrical energy storage, Energy Environ. Sci. 6 (2013) 1083.


Storing Energy

[25] M. Lafoz, J. Santiago, I. Etxaniz, Kinetic Energy Storage Based on Flywheels: Basic Concepts, State of the Art and Analysis of Applications, European Energy Research Alliance JP Energy Storage Mechanical Storage Sub-program Ref 2013-05 CIEMAT., 22. 28040. Madrid, 2013. [26] Gyrotricity Ltd, (accessed on 5 1 2021). [27] Chakratec, (accessed on 5 1 2021). [28] F. Alasali, S. Haben, V. Becerra, W. Holderbaum, A Peak Shaving Solution for Electrified RTG Cranes, 19the20th August 2017. Proceedings of 70th IASTEM International Conference, Oxford, UK. [29] Beacon Power LCC, Beacon Power’s Operating Plant in Stephentown, New York. https:// on 5 1 2021). [30] J.P. Rouse, S.D. Garvey, B. Cardenas, T.R. Davenne, A series hybrid “real inertia” energy storage system, J. Ener. Stor. 20 (2018) 1e15. [31] M.E. Amiryar, K.R. Pullen, Assessment of the carbon and cost savings of a combined diesel generator, solar photovoltaic, and flywheel energy storage islanded grid system, Energies 12 (17) (2019), [32] A.M. Gee, R.W. Dunn, Analysis of Trackside Flywheel Energy Storage in Light Rail Systems, IEEE Transac. Vehic. Technol. 64 (9) (2015). [33] J. Li, J. Jiyun Zhao, X. Xiaochun, Zhang, A novel energy recovery system integrating flywheel and flow regeneration for a hydraulic excavator boom system, Energies 13 (2020) 315,

Rechargeable lithium-ion battery systems


Matthias Vetter and Stephan Lux Fraunhofer Institute for Solar Energy Systems ISE, Freiburg, Germany

1. Introduction There are many different types of rechargeable batteries; some are available on the market, some are close to market entry, and many are still under research and development. The list of different technologies is impressive and includes lead-acid, nickel-cadmium, nickel metal hydride, the huge family of different lithium-ion types, sodium-ion, zinc-ion, sodium sulfur, sodium nickel chloride, and redox-flow such as the vanadium battery. Furthermore, research efforts focus on post lithium-ion. As one of the most prominent new development to be highlighted involves an all solidstate approach with metallic lithium as anode material, which allows highest energy densities, seen as crucial for certain electromobility applications. Within recent years various types of lithium-ion batteries have been developed and are available on the market. The cells can be classified by their cathode materials such as LCO (lithium cobalt oxide), NMC (nickel manganese cobalt), NCA (nickel cobalt aluminum), and LiFePO4 (lithium iron phosphate). Today typically graphite is used as anode material in lithium-ion batteries, for certain specific applications LTO (lithium titanate oxide) is also used. New approaches are targeting the so-called silicon-rich materials for higher specific capacities and energy densities. The impressive developments in these lithium-ion technologies are driven by many kinds of electromobility. Besides these large and fast growing markets, lithium-ion batteries also play an increasingly dominant role in stationary applications. The main advantages of lithium-ion batteries are high energy density, high cycle and calendar life times, fast and efficient charging capability, with comparably little energy wasted, low self-discharge rates, no need to be held upright, fairly maintenance-free, and little voltage sag. Main disadvantages are safety issuesdthe possibility of thermal runaway and propagationdand the resulting additional efforts in the field of appropriate safety concepts. Besides careful selection of high-quality cells, safety tasks are being solved by elaborate protective measures such as battery management circuitry which is discussed in this chapter. Lithium-ion batteries with its various types are still relatively new and have only been commercially available since the late 1980s. Even though the chemistry and the technology paths are now reasonably well proven, there is still a huge potential for improvement and optimization potential according to the application specific needs. Within recent years this battery has edged out older rechargeable batteries in the consumer sector such as the nickel metal hydride (NiMH) battery. Storing Energy. Copyright © 2022 Elsevier Inc. All rights reserved.


Storing Energy

In the meanwhile, the exponentially growing market of electromobility is dominated by lithium-ion batteries as well and this will remain at least in the current decade [1]. Furthermore, this battery technology is also rapidly increasing its market shares in the stationary energy storage sectorde.g., in Germany [2]dwhich is achieving more and more relevance in the context of energy transition with increasing shares of fluctuating renewables such as photovoltaics and wind turbines [3].


Physical fundamentals of lithium-ion batteries

In the market segment of stationary electrical energy storage various battery technologies have been developed for the specific needs of different applications and corresponding suitability. However, in this chapter the focus is on lithium-ion batteries as their market shares in most of the stationary applicationsde.g., utility scale, commercial, and industrialdare remarkable and in some cases they have achieved much over 90% market share, as in the residential solar home storage sector in Germany [2]. As mentioned, there is a variety of different lithium-ion battery technologies suitable for these stationary applications, which are on the market or close to market entry. Table 12.1 shows an extract of selected lithium-ion technologies and provides a comparison in terms of key parameters. The term lithium-ion is used as there is no elemental lithium in the battery. As the lithium ions move from one host to another, the process has been likened to a rocking chair. The basic half-cell reactions at each electrode during discharging (using the LiCoO2 battery as an example) are at the anode: LixC6 / C6 þ xLiþ þ xe and at the cathode: Li1-xCoO2 þ xLiþ þ xe / LiCoO2 with the overall reaction: LixC6 þ Li1-xCoO2 / C6 þ LiCoO2


Development of lithium-ion battery storage systems

Lithium-ion battery systems are typically assembled from several modules interconnected in parallel and/or serial, whereas a module consists of several cells which may be switched in parallel and/or in serial. The principal design of lithium-ion battery systems will be discussed in the following section using as an example the battery used

Rechargeable lithium-ion battery systems


Table 12.1 Comparison of different selected lithium-ion battery technologies [4]. The values can be interpreted as a current review and it must be noted that progress in technology developments is still ongoing and furthermore parameters are dependent on various manufacturer-specific variables.

Gravimetric energy density [Wh/ kg] Power density [W/kg] Nominal cell voltage [V] Cycle life time (equivalent full cycles) Calendar life time [a] Efficiency [%] Temperature range charging [ C] Temperature range discharging [ C] Monthly selfdischarge [%]

LCOa/ Graphite

NMCb/ Graphite

LFPc/ Graphite

NCAd/ Graphite

LCO,a NMC,b LMOe/ Titanate






300e4000 (dependent on design and current rating of cell) 3.6










8e20 years 90%e98% 0e45 (typically)f

20 to 55


2e10 (typically 3)


LCO, Lithium cobalt oxide. NMC, Lithium nickel cobalt manganese oxide. c LFP, Lithium iron phosphate. d NCA, Lithium nickel cobalt aluminum oxide. e LMO, Lithium manganese oxide. f Charging below 0 C possible for some types, especially automotive. g Titanate possesses lowest capacity losses at lower temperatures. b

for photovoltaic (PV) home storage applications. This fulfills the task of increasing self-consumption of generated PV electricity and there is, for example, in Germany at present a huge market with strong growth rates [2]. In general, the development steps for a battery system are as follows: • •

Cell characterization and selection of appropriate cell technology Module and system design


• • • •

Storing Energy

Cell interconnection • Electrical • Mechanical • Thermal Cooling system Safety concept Battery management Interfaces and integration in energy systems

Typical storage sizes for residential PV applications are in the range of 2 kWh to 10 kWh. An example of such a storage battery system based on lithium-ion pouch bag cells is shown in Fig. 12.1; the corresponding parameters are listed in Table 12.2. The system was designed originally, for a battery inverter with a nominal input voltage of 48 V. Today, battery invertersdwith transformerless technologydare on the market with higher input voltage levels.


Design of battery modules and systems for stationary applications

For the construction of a battery module several aspects must be considered. Among others, safety, reliability, efficiency, long life time, and reduced maintenance efforts play a key role. As temperatures have a huge impact not only on the safe operation but also on the aging mechanisms, the design of an appropriate cooling system is one of the key aspects. As a result, certain temperature levels should not be exceeded and furthermore a homogeneous temperature distribution within one cell, within a module, and within the whole battery system is important. In the following section the simulation-based design of the battery module is described. For the design and the construction of the modules and the battery system thermal simulations were carried out using the simulation tool Dymola [6] and the model description language Modelica [7]. The approach is based on the partition of the cells

Figure 12.1 Concept of a 5.33 kW h lithium-ion battery system for residential photovoltaic (PV) applications consisting of three modules switched in parallel to be connected to a 48 V battery inverter. One module consists of 12 cells, switched in serial [5]. The list of parameters is given in Table 12.1.

Rechargeable lithium-ion battery systems


Table 12.2 Parameters of the lithium-ion battery system, shown in Fig. 12.2. NMC refers to nickel manganese cobalt (the use of the abbreviation 0.5C and 0.2C will be described in the next section). Parameter battery module

Value and unit

Cell chemistry Number of cells per module Rated voltage/V (L  W  H)/cm Gravimetric energy density/Wh kg1 Volumetric energy density/Wh L1 Energy content/kW h Efficiency/%

Graphite/NMC 12 (switched in serial) 44.4 30  24  15 105.3 164.8 L 1.78 95 (0.5C), 97 (0.2C)

Parameter battery system Number of modules per system (L  W  H)/cm Gravimetric energy density/Wh kg1 Volumetric energy density/Wh L1 Energy content/kWh Cooling Rated current/A Battery management

3 (switched in parallel) 82  25.5  57.5 82.2 44.4 5.34 2 fans 100 1 central unit and 3 balancing boards

Figure 12.2 Segmentation of lithium-ion cells and aluminum cooling plates for the simulation of the thermal behavior [8].


Storing Energy

and cooling plates into segments. Each cell was divided into 49 segments and the cooling plates between the cells were divided into 81 segments (Fig. 12.2). For every segment an electrical and thermal model was developed. The electric model is based on an inner resistance model of the used NMC cell, describing the dependence on SOC, temperature, and C-rate. The C-rate is a measure of the charge and discharge current of the battery and a discharge of 1C draws a current equal to the rated capacity. For example, a battery rated at 1000 mAh provides 1000 mA for 1 h if discharged at 1C rate. Based on the results of the thermal simulation the cooling area and the thickness of the cooling plates can be calculated. The challenge was to reduce the temperature difference between the single battery cells to a minimum. The low temperature difference should avoid too much cell balancing and allows a homogeneous aging of the single cells. To assure a homogeneous air flow over each cell and the three modules, the system was simulated within a computational fluid dynamics (CFD) simulation program. By using this CFD simulation an optimized angle of skewness could be identified (Fig. 12.3). The skewness enables a uniform airflow over every module of the system. To verify the goal of a maximum temperature difference of 2 K, tests in a climate chamber were carried out with one battery module. Using the flow channel, the air distribution over the cells could be investigated. Tests in the climate chamber showed that the temperature difference between the cells of one module is nearly constant and almost below 1 K for a charge/discharge C-rate of 1 (Fig. 12.4). For the analyses of the critical temperature segments module tests were carried out and a thermal imaging camera was used to identify the hotspots. The results (Fig. 12.5) show that the module was warmed-up mostly at the cells and the cooling plates between the cells. Within these tests no critical sectors could be detected.


Consideration of efficiencies

In comparison to all other battery technologies, lithium-ion batteries offerdin principledvery high efficiencies. In Fig. 12.6 achieved efficiencies of the developed

Figure 12.3 Principle of the air-cooling system, showing the built-in skewness [9].

Rechargeable lithium-ion battery systems


Figure 12.4 Temperature profile of three lithium-ion cells of a battery module for a 1 C charge/ discharge test in the laboratory [10].

Figure 12.5 Analysis of a thermo camera with 1C in a climate chamber [10]. Here “min” refers to minute.

battery module (Fig. 12.2) are shown. Due to internal losses within the cells and at the cell interconnectors, the efficiencies are decreasing with higher C-rates. But even at a C-rate of 1, which is not a common operation mode for storage systems in residential PV applications, the results show values clearly above 90%.


Battery management systems

Battery systems, for example, those using the lithium-ion technology, need to be managed. The battery cells must be monitored and controlled. Challenges in terms of safety, electrical isolation, and energy efficiency have to be considered [12]. Within


Storing Energy

Figure 12.6 Efficiencies of a lithium-ion battery module, consisting of 12 cells, switched in serial, for different C-rates [5].

this section the principles of battery management systems are explained, and different concepts are introduced. There are several functions to be handled by a battery management system (BMS). Fig. 12.7 provides an overview to these functions. Every BMS needs a safety layer to prevent the batteries from being overcharged or deep discharged. Furthermore, the temperature of the cells must be controlled. Therefore, a monitoring of the system temperaturedideally each cell temperature, the load current, and the cell voltages is required. Functions such as switches and coolers or fluid pumps must be triggered by the BMS. Advanced BMS possess sophisticated and precise state estimation algorithms for the state of charge and state of heath as well as end-of-life predictions and a

Figure 12.7 Overview of the management functions of a battery management system.

Rechargeable lithium-ion battery systems


model-based thermal control. To secure high overall system efficiencies, optimization algorithms for smart cell balancing, for load and thermal management are necessary. Furthermore, the battery management system needs a communication interface to internal and external components like the power electronics or supervisory energy management device to secure safe and reliable system integration. To handle these functions, there exist several types of BMS with their specific advantages and disadvantages. One may classify the types into modular, central, and single-cell BMS approaches [12].


Modular concept

In a modular approach the battery management contains a central control unit and module management systems (MMSs). The latter are responsible for the measurement and control of each lithium-ion battery module, which contains several battery cells, mostly connected in series for reaching higher voltages. The number of cells connected in series strongly depends on the application [11]. Especially for storage applications with higher capacities a higher number of battery cells per module helps to reduce costs for the electronics. Typically, the MMS electronic system carries out the measurement and control on battery cell level and communicates with the central management system, which collects the measurement and the control data of the single modules. Furthermore, it controls switches and a cooling system and establishes a communication to external components. A schematic representation of the modular approach is shown in Fig. 12.8. The particular description of an MMS and a central management system (CMS) is based on a system developed at Fraunhofer ISE. The concept is an example for a modular approach. The main parts of the MMS are the battery front-end unit and a controller unit. The front-end unit is responsible for measuring the voltages of the battery cells and for controlling the temperature of the printed circuit board (PCB). Furthermore, it provides a hardware interface for cell balancing, which is needed to balance the state of charge of the cells within a battery module. The controller unit consists of a low-power reduced instruction set computer (RISC) microcontroller. The microcontroller controls the battery front-end unit by communicating via a bus system. In addition, there is a bus interface to send the measured data and the commands to a central management system or a host PC for

Figure 12.8 Concept of a modular battery management system with an MMS and a CMS system with internal and external communication [12].


Storing Energy

further handling. By means of an integrated analog-to-digital (AD) converter, a temperature measurement on cell level [13] is obtained. Fig. 12.9 shows an example of such a module management system. It is a prototype developed at Fraunhofer ISE and has been tested in both laboratory and prototype applications. The module management system is scalable up to 1000 V of battery voltage. This is ensured by the integration of a high isolation barrier [12]. Especially in electric vehicle applications, it is important to minimize the selfdischarge rate of the battery in order to prevent a deep discharge after extended times without operation. To ensure a very low self-discharge rate the MMS system has a low current consumption. The module management control unit contains flexible software modules. Filter algorithms based on Kalman and Particle filter approaches for state of charge and state of health estimations on cell level with high accuracy are implemented. Error management of events like overcharging, deep discharging, or too high temperatures ensures a reliable operation of the batteries. The central management unit retrieves all management and measurement data from the modules. Furthermore, it communicates with the modules via a bus system. For communication with external components a second bus interface is implemented. The central management unit controls the battery switches and the integrated cooling system. Signals are integrated for the control of external switches, cooling pumps and other components. Air cooling, using fluid cooling or a combination of both is possible. Additionally, temperature, high current, and high voltage measurement are integrated to measure the parameters on system level via a controller area network (CAN) bus. Important parameters for system safety are monitored. These comprise the measurement of the insulation resistance for an electrically isolated system and a redundant control of the battery switches.

Figure 12.9 Photo of a module management system unit with a passive resistive cell balancing, voltage and temperature control, and a controller unit [12].

Rechargeable lithium-ion battery systems


The central controller consists of an embedded system involving a thermal management, a central error handling system, and a charge and discharge management system. With thermal and electric battery models a precise central battery management is needed. Furthermore, with the integrated battery models and filter algorithms a life time prediction of the battery system can be made available. Fig. 12.10 shows a photo of the central management unit.


Single central concept

The single central BMS approach (Fig. 12.11) is based on only one printed circuit board, which controls all battery cells of a storage system, and which can be switched in serial or parallel. The advantage in comparison to the modular system is the lower cost of the electronics. All functions regarding safety, control, and measurement are integrated. Furthermore, the energy efficiency is higher, because there is only one central controller. Battery systems based on such a BMS are not easily expandable, but this approach enables significant cost reductions in many applications. All functions regarding safety, control, and measurement are integrated in this central BMS. Furthermore, the energy efficiency is higher, because there is only one central controller. The decreased system complexity of this solution is often an advantage, too. A disadvantage is the decreased flexibility in comparison to the modular approach. In the modular approach another module is easily integrated on the bus system, which enables also easy integration of modules in a spatially distributed system, since only a bus cable must be connected and no cable for cell voltage measurement and cell balancing for each single cell are needed. The battery front-end unit and the controller unit are integrated on one PCB (see modular concept). Several front-end units control and measure all cells of a battery pack. The controller unit could contain an advanced RISC machine (ARM) controller with a low current consumption.

Figure 12.10 Photo of a central management system with an embedded system as central controller and several interfaces [12].


Storing Energy

Figure 12.11 Single central BMS concept of a battery management system with one PCB [12].


Single-cell concept

A third approach is placing a single-cell control unit (SCU) on each single-battery cell. The SCU is a rather simple integrated circuit, measuring the cell temperature and voltage. It is also possible to integrate a cell balancing. A communication with a central management system is established through a bus interface. Normally, the management functions of the CMS are more complex than for the modular approach. The state of charge (SOC) and the state of health (SOH) estimation plus the error management must be executed on this unit. There are only basic safety and measurement functions on the SCU. Fig. 12.12 shows the architecture of such a system.


State of charge estimation

Since an accurate knowledge of the batteries state of charge (SOC) is unavoidable for a proper usage of a battery system a lot of different approaches have been investigated. Today, in many sophisticated battery management systems it is state-of-the-art to use Kalman filters to estimate the actual SOC. But since the Kalman filter uses some particular assumptions (like Gaussian distributions), its correctness and applicability are limited. A new approach is the so-called particle filter which is derived from the same family (the Bayesian filters) as the Kalman filter [14]. By employing Monte Carlo sampling methods, the particle filter offers the possibility to deal with any distribution. As depicted in Fig. 12.13 for the particle filter a Markov chain is assumed: • • •

ut: input which will change the system’s state over time. It can be measured. xt: state of the system at time t. zt: the output is in some correlation with the system’s state enabling a rough estimation. The quantity can be measured.

Figure 12.12 Single-cell BMS concept of a battery management system with several single-cell control units (SCU) and a central management system (CMS) [12].

Rechargeable lithium-ion battery systems


Figure 12.13 Assumption of a Markov chain for SOC determination.

So, one assumes only the quantities of input ut1 and output zt and not the former values influence the calculated probability of the current state. From Bayes’ theorem one gets the following equation: Pðxt Þ ¼ h


Z ðPðzt jxt Þ

Pðxt jxt1 ; ut1 ÞPðxt1 Þdxt1

The particle filter algorithm (see Fig. 12.14) runs in three steps:

Figure 12.14 Illustration of the particle filter algorithm. For initialization all samples (or particles) are distributed uniformly over the possible value range of state x. In the first step of the algorithm the influence of input u on every sample is calculated. To the value of u noise ε is added which is taken from a suitable probability distribution. The second step gives every sample a weight according to the probability taken from the measurement value z and the measurement model. The last step resamples the weighted particles to gain an unweighted particle set. In the picture the low variance resampling method is depicted [15].


Storing Energy

1. State Transition: The influence of input ut1 on each sample (particle) skt is calculated. By adding a different random value to the input value for each sample measurement error, the uncertainty of this step is considered. 2. Weighting: The samples are weighted according to the observed measurement zt and a prob ability density function Pðzt jxt Þ. Then the sum of all weights (wkt is normalized to one. 3. Resampling: In this step the samples are resampled according to their weights. After that all samples have the same weight again. The introduced approach uses the low variance resampling method for low computational afford.

In step one the so-called process model is used, which describes the influence of the input ut1 on the state xt . For state of charge estimation following model equation is used:   Ibatt þ εk Dt k k sSOC;t ¼ sSOC;t1 þ SOH$Cn where εk represents the random value, which is added to the input value. It can be sampled from any distribution which is suitable for the application. For this application a Cauchy-Lorentz distribution is used. By adding this noise to the particles, the diffusion of the particles is increased. This diffusion modelsdin case of state of charge estimationdthe increasing uncertainty of ampere hour counting. The second step decreases this diffusion by weighting the particles due to their probability. Therefor the so-called measurement model is used, which calculates the estimated terminal voltage at the actual state of charge of the particle. For phosphate-based lithium-ion batteries like lithium iron phosphate (LFP) cells, the open circuit voltage (OCV) versus SOC curve is very flat (see Fig. 12.15) and shows a hysteresis between charging and discharging. For that, two voltages are calculated, one for charging and one for discharging. The following equations show the measurement model for an LFP battery:     k Vdischarge;t ¼ OCVdischarge skSOC;t þ Ri skSOC;t ; T; Ibatt $Ibatt

Figure 12.15 Open circuit voltage versus state of charge curve of a lithium iron phosphate battery with graphite anode. One can see the hysteresis between charging and discharging voltage and the very flat voltage in medium SOC range [14].

Rechargeable lithium-ion battery systems


    k Vcharge;t ¼ OCVcharge skSOC;t þ Ri skSOC;t ; T; Ibatt $Ibatt These two voltages are compared to the measured terminal voltage. So, two probabilities are determined with the following formula (using a Cauchy-Lorentz distribution): 0 1 wkSOC;t ¼ $@ p

1 g


  2 þ 2 A k k g2 þ Vdischarge;t g2 þ Vcharge;t  Vmeas  Vmeas

After that, in step three, these weighted particles are resampled to regain an unweighted set of particles. In this approach the low variance resampling method is used like depicted in Fig. 12.15. Low variance resampling puts all particles in a row, each particle has a length according to its weight. After that this row is sampled with a fixed sampling width as many times as the number of particles. The sampled particles represent now an unweighted set of particles and the filter goes back to step one. The estimation value for the state of charge is the mean value of all particles’ values.


State of health estimation

As an example, for estimation of state of health (SOH) is introduced the particle filter approach as well. Both filters run in parallel and share their results. As the particle filter for SOC estimation needs the SOH as parameter, the filter for SOH estimation needs the SOC difference between two calculation steps as parameter. Therefore, the approach is called parallel particle filter for SOC and SOH estimation. SOH in this section is defined as the ratio of the actual battery capacity to the nominal battery capacity: SOH ¼

Cact Cn

In general, the particle filter for SOH estimation works similar to that for SOC estimation as described in the section above. Due to the very slow change of SOH, in the process model it is assumed that no input value u takes influence on the state. Just a noise value ε is added to every particle: skSOH;t ¼ skSOH;t1 þ εk


Storing Energy

The weighting step is done by using the SOC change of the SOC filter during the last step (DSOCmeas ) as measurement value z and following equation as measurement model: DSOCkt ¼

Qstep skSOH;t $Cn

Variable Qstep represents the integrated charge flown into and out of the battery. If using the Cauchy-Lorentz distribution the weight is calculated as follows: wkSOH ¼

1 g $   p g2 þ DSOCk  DSOCmeas 2 t

Thereafter the low variance resampling step is performed as described above. The state of health estimation value is gained by taking the mean value of all particles’ values. Validation

The dual particle filter approach is validated for different types of lithium-ion batteries (LFP and NMC, both with graphite anode) and different types of current profiles including electric vehicle (EV) cycles and photovoltaic (PV) applications [15]. The EV profile is characterized by large currents, high dynamics, but also relatively long pauses with no current. The PV profile on the other hand shows lower currents, but therefore there are virtually no phases without any currents [15]. In Fig. 12.16 a validation sequence for lithium iron phosphate (LFP)-graphite lithium-ion cells is shown by using a PV profile [15]. Even the estimation with LFP is much more complicated, due to flat open circuit voltage (OCV) and hysteresis, exact SOC is found fast and estimated reliably. SOH estimation is very accurate as well, as depicted in Fig. 12.17.

Figure 12.16 Depicted is the state of charge during a PV current profile. The used battery is a lithium iron phosphate battery with a graphite anode [15].

Rechargeable lithium-ion battery systems


Figure 12.17 Shown is the state of health of a lithium iron phosphate/graphite battery during a PV current profile. The reference value was determined with a standard charge discharge regime [15].

4. System integration 4.1


Battery systems and direct current (DC) power sources like photovoltaic generators can be coupled via power electronics on a DC bus bar or on the alternating current (AC) side. Exemplarily an AC coupled system is introduced in the Fig. 12.18, which allows the integration of lithium-ion battery systems in PV systems by using a market available battery inverter.

Figure 12.18 Integration of the developed lithium-ion battery system (Fig. 12.2) in a residential PV system by using a market available battery inverter [5].


Storing Energy

In these AC coupled system configurations the PV generator and the battery system are connected to the AC grid via two separate inverters. The conventional PV system, consisting of PV modules and a PV inverter, is in principle not affected by the integration of a battery. Therefore, installed PV systems can easily be complemented with battery storage at a later point of time without any adaptation. Due to the modular concept, sizing of the battery system is almost independent of the size of PV system components like the PV inverter. Disadvantages of this topology are the limited cost reduction potentials, as two full inverters are needed, as well as the voltage levels of the market available battery inverters for residential applications, which have beeen in the past in the range 24 V to 48 V. Therefore, the inverters possess transformers and offer only relatively low efficiencies (w94%) at nominal operating point and efficiencies much below this value at a wide range of the typical operating window. Meanwhile transfomerless battery inverters are on the market as well and are used more and more in residential PV battery storage systems. In principle, such inverters offer much better efficiencies at partial load.


Communication infrastructure

Lithium-ion batteries are a very promising storage technology especially for decentralized grid-connected PV battery systems. Due to several reasons, for example, safety aspects, the battery management is part of the lithium-ion battery system itself and is not integrated into the battery inverter or the charge controller as it is usual for lead-acid and nickel-based batteries. This battery management system must control the battery system itself and the connected power electronics. Furthermore, it must exchange all relevant data with the supervisory energy management system. For both tasks a field bus communication is necessary (Fig. 12.19), but market available products offer only proprietary solutions. Therefore, system integrators are not free in choosing different system components for specific solutions. Furthermore, it is predefined which battery systems can be operated with which inverters or charge controllers without having a huge adaptation effort for the communication system. To enable higher degrees of freedom in system assembly, a standard for the communication on a field bus level among battery systems, power electronics, and energy management systems is necessary. Such an approach represents currently the socalled EnergyBus [17], which was initially developed for a simplified connection of the system components of light electric vehicles. This standard defines the communication protocol as well as the power connectors. As communication protocol the CANopen user profile CiA 454 “energy management systems” is used [16]. This protocol specifies the data exchange between the single components such as storage systems, generators, loads, and energy management systems and enables the implementation of optimized operating control strategies. Based on this standard field bus communication, components of different manufacturers can be assembled by system integrators. The power connectors are defined especially for light electric vehicles, whereas the

Rechargeable lithium-ion battery systems


Figure 12.19 Field bus communication based on the so-called EnergyBus/CiA 454 protocol [16e18] for setting up a network of different generators, storage systems, power electronics, and energy management systems.

communication protocol was extended for stationary applications like PV battery systems. New specification parts for generators and loads have been designed in such an abstract way that a general manageability by a supervisory control system is enabled. For example, a PV generator and a CHP (combined heat and power) unit can be described by the same specification part. Furthermore, smart metering components can be easily integrated.

5. Conclusions In conclusion, in the relatively short time of its development and application, the main advantages of the lithium-ion battery such as high energy density, long life cycle, fast and efficient charging capability (in comparison to other battery technologies), fairly maintenance-free, and little voltage sag have been undercut by the possibility of thermal runaway and propagation. This has resulted in much research and development enabling improved battery system designsdbesides various measures on cell level, including integration of safety improving materials and battery management protective circuits and programs, which have been the focus of this chapter. These crucial developments have opened the door for lithium-ion technologies for almost all applications, in which batteries are already used or will be used in the futuredportable, electromobile, and stationary.


Storing Energy

References [1] M. Sanders, The rechargeable battery market and main trends 2019-2030, in: AABC Virtual Conference, 2020. [2] J. Figgener, et al., The development od stationary battery storage systems in Germany - a market overview, J. Energy Storage (2020). [3] H.-M. Henning, et al., Wege zu einem klimaneutralen Energiesystem, 2020. www.ise. [4] M. Vetter, S. Lux, J. W€ullner, The use of batteries in storing electricity, in: T. Letcher (Ed.), Chapter of Future Energy, 2020. [5] M. Vetter, Development of an Optimized Battery System for Residential PV Applications, Intersolar North America, 2014. [6] [7] Accessed 28 September 2015. [8] S. Lux, M. Dennenmoser, M. Becker, N. Lang, M. Jung, M. Vetter, Entwicklung und thermische Modellierung eines innovativen Lithium-Ionen-Speichers f€ ur den station€aren dezentralen Einsatz, Entwicklerforum (2013). Aschaffenburg, 25.August 2013. [9] Project Urban Hybrid Energy Storage, 2014. [10] J.B. Goodenough, J. Power Sources 174 (2007) 996e1000. [11] T.A. Stuart, W. Zhou, J. Power Sources 196 (2011) 458e464. [12] M. Jung, S. Schwunk, Green 3 (2013) 19e26. [13] L. Rao, J. Newman, J. Electrochem. Soc. 144 (1997) 2697e2704. [14] S. Schwunk, N. Ambruster, S. Straub, J. Kehl, M. Vetter, Particle filter for state of charge and state of health estimation for lithiumeiron phosphate batteries, J. Power Sources 239 (2013) 705e710, [15] N. Armbruster, et al., Particle filter for state of charge and health estimation for both NMC and LFP based batteries, in: Poster, Advanced Automotive Battery Conference, Strasbourg, 24e28 June, 2013. [16] [17] [18] M. Vetter, Dezentrale netzgekoppelte PV-Batteriesysteme. Intersolar e PV Energy World, M€unchen 8th of June 2011.

Section C Electrochemical and electrical energy storage techniques

This page intentionally left blank

The road to potassium-ion batteries


Titus Masese 1, 2 and Godwill Mbiti Kanyolo 3 1 Research Institute of Electrochemical Energy, National Institute of Advanced Industrial Science and Technology (AIST), Ikeda, Osaka, Japan; 2AISTeKyoto University Chemical Energy Materials Open Innovation Laboratory (ChEMeOIL), Sakyoeku, Kyoto, Japan; 3 Department of Engineering Science, The University of Electro-Communications, Chofu, Tokyo, Japan

1. Introduction Massive advancements in technology that continue to redefine the worldview of critical sectors such as telecommunications, energy, transport, and healthcare are quickly pushing the global energy resources to their limits. Against this backdrop is an ongoing campaign for energy sustainability that has exacerbated the deployment of renewable energy technologies such as solar, wind, and tidal energy sources. However, the intermittent nature of these technologies still remains an inhibition to their advancement, spurring the growth of energy storage systems particularly secondary batteries. Secondary batteries (also known as rechargeable batteries) are electrochemical energy storage devices that facilitate multiple cycles of electrical energy storage and dispensation (charge and discharge) through a series of reversible redox reactions. Although the history and development of rechargeable batteries spans over centuries, their basic operation principles have not changed significantly. Nonetheless, the past few decades have witnessed significant improvements in rechargeable battery components and chemistries in a bid to meet the requirements of the rapidly evolving electronic and automotive sectors.

2. The evolution of modern batteries Perhaps one of the earliest forms of commercially available secondary batteries, leadacid batteries, have gained immense utility in the automotive industry and as uninterrupted power supply devices due to their low cost and scalability. Although lead-acid batteries are still relevant in some applications today, their dependency on hazardous materials such as lead, short cycle life, and low energy densities (30e50 Wh kg1) render them impractical for most modern applications [1]. Consequently, in order to meet the growing demand for small portable electronics, the smaller sized nickel-cadmium (Ni-Cd) batteries were favored as suitable replacements for primary (nonrechargeable) batteries. Utilizing nickel hydroxide and Storing Energy. Copyright © 2022 Elsevier Inc. All rights reserved.


Storing Energy

cadmium as the cathode and anode, respectively, along with an alkaline electrolyte (mainly potassium hydroxide), these battery systems delivered higher current discharge characteristics and slightly improved energy densities (45e50 Wh kg1) [1]. Moreover, they exhibited fast charge/discharge capabilities with stable performance even at considerably low temperatures (20 C). However, a major stumbling block to their performance occurs during the discharge process, when hydrogen and oxygen gases are released from the negative and positive electrodes, respectively, unfortunately posing safety and reliability issues. In addition, Ni-Cd batteries produced low voltages (w1.2 V) and high self-discharge rates, necessarily rendering them unsuitable for high-performance electronics. In a bid to improve the safety and performance of Ni-Cd batteries, the cadmium electrode was eventually replaced with a hydrogen absorbing metal alloy to create the nickel-metal hydride (Ni-MH) battery. Although the output voltage of these batteries stagnated at 1.2 V, while their cycle life slightly declined, significant improvements in the energy densities of these batteries (around 60e120 Wh kg1) were noted[1], making them propitious for portable electronics such as cellular phones, digital cameras, and laptop computers. In addition, because Ni-MH batteries do not use cadmium, a highly hazardous material, their ecological feasibility extended their utility into the early hybrid vehicles. Despite these momentous developments in the evolution of secondary batteries, lithium-ion batteries (LIBs) have been credited for the exponential proliferation of high-performance electronics such as smartphones and wearable devices, as well as the paradigm shift within the automotive industry which ushered in the practical deployment of market-viable electric cars. First commercialized by Sony in 1991, LIBs unveiled a new realm of lightweight, high-voltage batteries (3.7 V) with very high energy densities (110e265 Wh kg1) and low self-discharge rates[1]. However, this technology is heavily inhibited by several factors; among them are its high production cost, the scarcity of lithium reserves in the planet coupled with moderate safety concerns. This chapter aims to delineate the milestones that have defined the advancements of rechargeable batteries. The chapter begins by paying homage to the illustrious efforts and discoveries by esteemed researchers such as John Goodenough, Stanley Whittingham, and Akira Yoshino for their crucial development of LIBs which culminated in their being awarded the 2019 Nobel Prize in Chemistry, but eventually goes beyond their efforts into the future direction of rechargeable batteries in the form of sodium-ion and potassium-ion batteries while touching on compatible electrodes and electrolytes.


Mechanisms of lithium-ion battery operations



Like most rechargeable batteries, the LIB is comprised of two electrodes with different chemical potentials, a separator, two current collectors (positive and negative), and an

The road to potassium-ion batteries


ionically conductive electrolyte (lithium-based salt). The positive electrode is typically based on a layered oxide (such as LiCoO2), a polyanion compound (such as LiFePO4), or a spinel oxide (such as LiMn2O4) [2e4]. For the negative electrode, graphite is a common choice due to its low reactivity with the electrolyte. Conventional electrolytes usually entail mixtures of organic carbonate solvents such as ethyl carbonate or diethyl carbonate and noncoordinating anion salts such as LiPF6, LiAsF6, LiClO4, LiBF4, and LiCF3SO3 [5]. A thin separator film made from insulation material separates the two electrodes to prevent an internal short circuit form occurring. Depending on the choice of materials, the resultant battery parameters can vary to suit different applications.



During charging, an oxidation half reaction (Eq. 13.1) occurs on the cathode side where a positively charged external power source removes the electrons from the lithium atoms in the cathode (for instance, LiCoO2) and channels them to the anode side (for instance, graphite) through an external circuit, as shown in Fig. 13.1. This creates an electrical imbalance within the cell, prompting the flow of lithium ions toward the negatively charged graphite through the electrolyte. Graphite, which has a layered structure of loosely bonded graphene sheets, facilitates a reduction half reaction (Eq. 13.2) by allowing the insertion of lithium ions for storage within the layers to form lithiated graphite (LiC6) in a process known as lithiation (lithium intercalation). The process continues until the graphite is fully lithiated, achieving maximum

Figure 13.1 Schematic illustration of the operating mechanism of lithium-ion batteries. Akin to other rechargeable battery systems such as sodium- and potassium-ion batteries, lithium-ions shuttle back and forth through the electrolytes to the electrodes. A layered cathode and graphite as anode are shown for brevity.


Storing Energy

electrical potential difference between the electrodes, also known as fully charged state. During discharge, in a reverse process, electrons extracted from the graphite anode are channeled to the cathode side, driving the positively charged lithium ions out of lithiated graphite in a process known as delithiation (lithium deintercalation). Lithium ions flowing through the electrolyte are recombined with the cathode material. This process continues until the graphite is fully delithiated and the potential difference of the electrodes is at the minimum voltage. Reaction at the positive electrode (cathode): For instance, in a cathode such as LiCoO2 LiCoO2 #Li1x CoO2 þ xe þ xLiþ


Reaction at the negative electrode (anode): For example, in an anode such as graphite 6C þ xe þ xLiþ #Lix C6


This intercalation process involving the continued insertion and extraction of guest ions into cathodes and anodes has become the bedrock of modern rechargeable batteries [2]. It is important to note that graphite does not participate in the redox reactions but rather serves as a storage medium for the lithium ions. In fact, it is important to mitigate the occurrence of side reactions between the electrolyte and the anode, as they may substantially degrade the components, causing diminished battery performance and in some extreme cases ignite hazardous fires. Fortunately, in an accidental discovery during the initial charge, electrons in the highly charged graphite electrodes were found to react with lithium ions covered in solvent molecules, which caused a partial degradation of the electrolyte resulting in a minimal consumption of lithium ions. The degraded material was deposited on the graphite electrode forming a thin protective film now known as the solid electrolyte interface (SEI) that protects the electrolyte from any further contact with the electrons, effectively increasing performance and reducing the risk of igniting fires [6,7].


Improvements on battery chemistry

As rechargeable battery applications have become increasingly diverse over the years, the optimization of battery components has become crucial to the advancement of sectors such as smart grid systems, electric vehicles, and portable electronics. Although lithium-ion batteries have already proven to be substantially better than their earlier counterparts, their practicality has come into question due to the high cost of development, safety issues, and scarcity of lithium resources [8]. These issues have galvanized exploration efforts into more practical lithium battery components as well as alternative chemistries.

The road to potassium-ion batteries


This chapter proceeds to explicate the advancements in rechargeable battery technologies, focusing on the chemistries within their components, namely: 1. Cathode chemistries 2. Electrolyte chemistries 3. Anode chemistries

4. Cathode chemistries 4.1

Cathode materials for rechargeable lithium-ion batteries

Advancements in rechargeable battery systems have to a large extent been ascribed to the development of cathode materials. In most batteries, positive electrodes (cathodes) not only determine crucial battery parameters such as the voltage, energy and power densities, capacity, and cyclability of the battery but also determine practical aspects of development such as the cost, scalability, and even the size of the energy storage devices. Thus, it is imperative to understand the cathode mechanisms involved during battery operations. Fast ion kinetics, good electronic conductivity, and high voltage capabilities are key prerequisites of cathode materials for rechargeable batteries. Accordingly, early cathodes utilized some transition metal disulfides (such as TiS2) or metal oxides (such as WO3) not only to host guest ions but also for their ability to change their inherent optical and electronic properties upon the intercalation of guest ions. For instance, the intercalation of monovalent ions, Aþ (where A ¼ Liþ, Naþ or Hþ) into WO3 to form AxWO3, was found to alter the electronic properties from insulator to semiconductor to metallic with varied values of x [2]. A fundamental concept in the selection of cathode materials is that the cell voltage is determined by the redox energy differences between the cathode and the anode. Therefore, monovalent ions with low redox potentials such as Liþ (3.04 V vs. standard hydrogen electrode) promised very high theoretical voltages (over 4 V), marking the onset of the new generation rechargeable batteries. This chemistry was investigated by Whittingham who used a TiS2 cathode, a lithium-metal anode, and a LiClO4 liquid electrolyte to develop a rechargeable battery with a 3.5 V) [60] bringing them to the fore as high-voltage and low-cost cathode contenders for high energy density and sustainable potassium-ion batteries. A drawback for most of the potassium-based polyanionic compounds is the inherently low electronic conductivity, which mandates the utilization of conductive agents such as carbon, which debilitates their volumetric energy density. Nanosizing and carbon coating have also been employed in a bid to increase the electronic conductivity of polyanionic compounds [18,60]. As can be seen in Fig. 13.2, polyanionic compounds display lower theoretical gravimetric capacities than Prussian analogs, and thus appear to compromise their energy density. Although their high voltages may offset the energy density deficiency, material screening of polyanionic compounds exhibiting large gravimetric capacities (>130 mAh g1) as well as high voltages (>4.5 V) is highly propitious.


Layered oxides

As earlier explicated (see layered oxides for sodium-ion batteries section), the large ionic size of Kþ ions substantially diminish the performance of layered oxide materials with the KxMO2 compositions. Fig. 13.3 shows the capacity-voltage curves of layered oxide material KxCoO2 alongside polyanionic compound (KFeSO4F) and Prussian analogs (K2Fe2(CN)6). It is apparent that KxCoO2 displays low average voltage relative to the other class of cathodes. A similar trend has also been observed in layered chalcogenides such as KCrS2. However, with the partial substitution of the transition metal species with other transition metal species or highly valent species (such as chalcogens or pnictogens) to form honeycomb-layered oxides the performance of potassium-ion cathode materials has been significantly elevated. Honeycomb-layered oxides Honeycomb-layered oxides typically adopt the chemical compositions A2M2DO6, A3M2DO6, and A4MDO6 (where M can be divalent or trivalent transition metal atoms

The road to potassium-ion batteries


Figure 13.3 Capacity-voltage plots of cathode materials for rechargeable potassium-ion batteries, displaying low average voltages attained by layered oxides such as KxCoO2. Reproduced with permission from Refs. [51,60,61].

such as Mn, Fe, Cr, Ni, Co, Cu, or some combination thereof; D represents pentavalent or hexavalent metal atoms such as Te, Bi, Nb, Mo, Sb, Ru, Ta, W, Ir, Bi; and A can be alkali atoms such as Li, Na, K, etc., or coinage-metal atoms such as Cu, Ag, etc.) [31]. The resulting heterostructures entail a layered framework with A alkali atoms sandwiched between parallel transition metal oxide slabs (MO6 and DO6 octahedra). Oxygen atoms from the transition metal oxides in turn coordinate with Aþ cations forming interlayer bonds whose strength is significantly weaker than the covalent in-plane bonds within the transition metal oxide slabs. In a concept known as the “inductive effect” [2], the introduction of the highly valent D atoms increases the ionization of MeO bonds, thus resulting in higher voltages to effectively oxidize the M metal cations. In honeycomb-layered frameworks, a unique correlation between the interlayer distance between the transition metal slabs, where alkali metal atoms reside, and the ion kinetics has been found to be a critical attribute to unlocking their capabilities [31,62]. Alkali atoms with large Shannon-Prewitt radii such as Na and K have been found to produce larger interlayer distances which allow a faster ionic diffusion during the charge-discharge cycles compared to Li [62]. As a result, potassium-based honeycomb-layered oxide compositions such as K2Ni2TeO6 and cobalt-based derivatives such as K2Ni2-xCoxTeO6 have been found to produce cathode materials that not only entail high voltages (beyond 4 V vs Kþ/K) but also allow facile Kþ insertion and desertion that supersede their lithium-ion counterpart during battery operations [31,63e66]. Fig. 13.4 summarizes the various classes of cathode materials for rechargeable potassium-ion batteries along with the challenges to overcome.

5. Electrolytes The electrolyte plays a vital role in determining some key battery performance parameters such as current (power) density, time stability, and the overall safety of the battery [67]. Due to the close interaction with other battery components, namely the cathode, anode, and the separator, the physicochemical and electrochemical characteristics of electrolytes such as compatibility, stability, viscosity, ionic conductivity, electrochemical window, cation coordination, and solvation are important determinants of a suitable choice of electrolyte.


Storing Energy

Figure 13.4 Main classes of cathode materials for rechargeable potassium-ion batteries. The merits and demerits are highlighted along with the strategies to overcome the demerits [44].

During battery operations, the electrolyte should ideally undergo no net chemical changes as all faradaic processes occur within the electrodes. However, it is important to note that slight degradations do occur during the formation of solid electrolyte interfaces on the electrodes to ensure chemical compatibility. As a general rule, an ideal electrolyte should meet the following criteria[67]: 1. Low electronic conductivity to minimize self-discharge. 2. Low reactivity with other battery components 3. Wide electrochemical window to prevent electrolyte degradation within the potential range of the cathodes and anodes. 4. High ionic conductivity to allow for the facile mobility of cations during charge and discharge processes. 5. Good thermal stability with melting and boiling points outside the operation temperature range. 6. Low cost and scalable development processes. 7. Low toxicity for safe handling and to limit environmental hazards. 8. Sustainable chemistries using abundant materials.

Considering the metrics mentioned above, most electrolyte explorations so far focus on the following types of electrolytes: a. Nonaqueous electrolytes comprising of salts solubilized in an organic solvent or solvent mixture. b. Aqueous electrolytes comprising salts solubilized in water. c. Ionic liquids comprising suitable salts in their liquid phase. d. Polymer electrolytes made from gel or solid polymers. e. Hybrid electrolytes made from a mixture of more than one of the above-mentioned electrolytes.

The road to potassium-ion batteries

5.1 5.1.1


Lithium-ion battery electrolytes Nonaqueous (organic) electrolytes

Nonaqueous electrolytes comprising lithium-based salts account for most of the electrolytes used in commercialized lithium-ion batteries so far. Typically, nonaqueous electrolytes are formulated by dissolving lithium salts such as lithium hexafluorophosphate (LiPF6) in organic carbonate solvents such as mixtures of ethylene carbonate (EC) with dimethyl carbonate (DMC), propylene carbonate (PC), diethyl carbonate (DEC), and/or ethyl methyl carbonate (EMC) [68]. Despite the widespread utilization of these electrolytes, the commercialized organic solvents employed (e.g., DEC, EC, DMC, and their mixtures) have been found to suffer low voltage stability, posing safety issues that have mired this class of batteries. Attempts to improve the safety standards and voltage tolerance of the electrolytes have stimulated explorations into organic fluoro compounds such as fluorinated cyclic carbonate, fluorinated linear carbonate, and fluorinated ether which thus far have been found to facilitate high voltage operations due to the high oxidation potentials of fluorine molecules [68]. In fact by using such organic fluoro compounds, the electrolyte decomposition thresholds were found to increase from w4.5 to w5.7 V. Besides the solvent choice, another key parameter for the evaluation of electrolyte performance is the solid electrolyte interface (SEI) formed on the electrode surface. The primary function of an SEI layer is to prevent unfavorable side reactions between the electrode and the electrolyte as a necessary measure against irreversible charge loss, reduced rate capability, poor cyclability and safety issues. Thus, a robust passivation layer is expected to be porous and ionic conductive in order to facilitate Li-ion transport while adequately restricting interactions between the electrolyte and the electrons. Thus, in order to improve the film-forming capabilities of the electrolytes, additives such as sulfone, vinylene carbonate (VC), cyclic sulfates, and fluoroethylene carbonate (FEC) have been found to improve the quality of the SEI, thus improving battery performance parameters [67,68].


Aqueous electrolytes

Although nonaqueous electrolytes have been widely investigated for Li-ion batteries, their low thermal stability and high flammability have been a limiting factor for the adoption of commercialized LIBs particularly on large-scale applications [67,68]. As a solution, aqueous electrolytes consisting of salts such as lithium sulfate (Li2SO4), lithium nitrate (LiNO3), and lithium perchlorate (LiClO4) dissolved in deionized water have been investigated for their limited flammability, high ionic conductivity, and low development costs [67e70]. These propitious characteristics enable them to not only yield better cyclability and higher rate capability but also improved capacity retention.


Storing Energy

However, aqueous electrolytes tend to have a very narrow voltage window with decomposition observed at low voltages of 1.23 V [67]. Moreover, during the formation of SEI passivation layers, they decompose to produce hydrogen and oxygen instead of a protective film, increasing the risk of cell venting, fire and rapid disintegration. As a result, one way to improve this class of electrolytes has been to adjust the pH values by eliminating oxygen to influence hydrogen evolution, effectively suppressing decomposition and in effect widening the voltage window. In addition, the use of lithium-ion superconductor (LISICON) and gel polymer electrolyte (GPE) film to protect electrodes from electrolyte side reactions has been reported to yield a high voltage output (4.0 V) and high energy density (446 Wh kg1) using Li2SO4 aqueous solution [67].


Ionic liquids

Molten salts at room temperatures or below, also known as ionic liquids (ILs), have been gaining immense attention as electrolytes for next-generation LIBs due to their unique set of properties such as low vapor pressure; nonflammability; high chemical, thermal, and electrochemical stability windows; and low heat capacities [71,72]. Moreover, they exhibit very high solubilities with organic, inorganic, and polymer materials allowing for the combination of a plethora of cations and anions which gives the flexibility to design LIBs for numerous applications. However, this class of electrolytes hasve some drawbacks such as higher viscosities than organic solvents, which result in reduced ionic conductivity and generally higher costs. Common lithium-ion battery ILs commonly utilize imidazolium, quaternary ammo nium, pyrrolidinium and piperidinium cations along with anions such as PF 6 , BF4 and  bis(trifluoromethanesulphonyl)imide (abbreviated as TFSI ) [73]. The low melting point of ILs has been accredited to the asymmetry of IL ions which hinders the crystal packing of the ions thus providing them with high thermal stability. Fig. 13.5 below shows the combination of salts, solvents and ionic liquids used to design electrolytes along with solid-based electrolytes. Amongst the cations used, imidazolium-based ILs exhibit low viscosity as well as relatively high conductivities at room temperatures. However, these electrolyte systems have poor electrochemical stabilities with narrow voltage windows of about 4 V (vs Li/Liþ) [67]. Furthermore, they suffer cathodic instability owing to their high reduction potential (viz., 1 V vs Li/Liþ) which increases the electrode deposition of Li, limiting the use of some cathode materials. Despite their demerits as electrolytes, imidazolium-based ILs have also been explored as electrolyte solvents and additives showing massive improvements in the formation of robust passivation films on the electrodes. On the other hand, quaternary ammonium-based ILs have shown better electrochemical stability than imidazolium-based ILs, providing an electrochemical stability window of more than 5 V (vs Li/Liþ) [67]. As a result, they can better sustain lithium deposition and dissolution throughout subsequent charging and discharging (cycling) without decomposition, making them ideal for high-voltage operations. However, these ILs have larger size cations, high viscosity, and low ionic conductivities, which

The road to potassium-ion batteries


Figure 13.5 Combination of salts, solvents, ionic liquids and additives used to design electrolytes for rechargeable alkali-ion batteries.

greatly inhibit their applications. In a bid to improve these electrolytes, the introduction of other functional groups such as nitrile- and ester-functionalized ILs as well as ether groups have been explored as means to reduce the viscosities and melting points of quaternary ammonium-based ILs while maintaining their electrochemical stability. Similarly, the use of organic solvent additives such as EC helps reduce the viscosity of these electrolytes and improve the electrochemical stability window. Pyrrolidinium and piperidinium-based ILs represent a burgeoning class of ionic liquid electrolytes due to their similar physicochemical properties as ammoniumbased ILs which endow them with relatively low viscosity, good electrochemical stability, and higher ionic conductivity [74e76]. Furthermore, the smaller ion sizes of pyrrolidinium cations contribute to their low viscosity making them exemplar electrolyte contenders. Reports on these ILs have revealed that adjustment of Li salt concentrations in these electrolytes help enhance their capacity, rate capability, and the cycle life of the electrodes. Additionally, the use of additives such as propylene carbonate (PC), vinylene carbonate (VC), and gamma-butyrolactone (g-BL) further improve the performance of pyrrolidinium- and piperidinium-based ILs by reducing their viscosity and improving conductivities [77].


Polymer electrolytes

As a solution to the low mechanical strength and safety issues of liquid electrolytes, ion conducting membranes, also known as polymer electrolytes, have been adopted in rechargeable batteries [67]. Polymer electrolytes are further classified as (i) solid polymer and (ii) gel polymer electrolytes, based on their material states.


Solid (dry) polymer electrolytes

This class of polymer electrolytes consists of solvent-free polymer salt systems comprising lithium salts dissolved in a polymer matrix, most commonly poly(ethylene


Storing Energy

oxide) (PEO). Solid polymers have gained momentous growth in the recent past, as they represent a breakthrough in rechargeable batteries in terms of design flexibility, robustness, and safety, opening new avenues for the application of electrochemical devices [78]. On its own, PEO polymer has low ionic conductivity, requiring the addition of lithium salts such as LiTFSI, lithium bis(pentafluoroethanesulfonyl)imide (LiN(SO2C2F5)2 (LiBETI)), LiClO4, and lithium bis(oxalato)borate (LiBOB) to improve performance [79]. However, the addition of such salts does not necessarily make significant improvements in conductivities at room temperatures. This is because the PEO-lithium salt electrolytes are composed of crystalline and amorphous phases that only allow ionic motion through ion hopping from site to site in the crystal region and micro-Brownian segmental motion in the amorphous region; both of which are slow processes that fail to yield facile ion migration. However, the use of supramolecular self-assembly synthesis techniques has allowed the enhancement of cavity sizes within the channels, thus improving the ionic conductivities [67]. Further improvements of solid polymer electrolytes can be done through the addition of ceramic particles such as ZrO2, Al2O3, SiO2, TiO2, and Fe3O4 in order to enhance the conductivity, mechanical, and electrochemical properties of these electrolytes [78]. In addition, carbon nanotubes used as additives have been reported to boost the electrochemical and mechanical performance of this group of electrolytes [79].


Gel polymer electrolytes

Compared to solid polymers, gel polymers entail enhanced chemical and thermal stability and low volatility, rendering them as more practical electrolytes. The most commonly used polymer gel electrolytes are poly(vinylidenefluoride) (PVdF) due to their strong electron-withdrawing functional group (eCeF) and high anodic stability [80]. They have also been found to have a high dielectric constant (ε ¼ 8.4) which assists in the ionization of lithium salts, thus, facilitating the development of electrolytes with a high concentration of charge carriers. The superior performance of gel polymers is attributed to their multisize porous structure which acts as a spongelike membrane loaded with liquid electrolyte. Despite the advantages presented by gel polymer electrolytes, their poor mechanical properties encumber the range of their applications. As a solution, PVdF-glass fiber mats (GFM) composite membrane has been utilized to improve their tolerance to mechanical stress and strain during battery development and operations [81]. For LIBs, gel polymer electrolytes also exhibit low interfacial stability toward lithium metal therefore resulting in high cell impedance. In resolution, PVdFhexafluoropropylene (HFP) with high oligomeric ionic liquids have been implemented to promote ion transport at the electrode interfaces [82]. However, interfacial instability during storage still remains a critical issue to be solved for gel polymer electrolyte LIBs. Other improvements include the addition of poly(methyl methacrylate) (PMMA) and poly(acrylonitrile) (PAN) to increase the pore size and porosity of the PVdF membrane, effectively increasing their electrolyte intake [83,84].

The road to potassium-ion batteries

283 Beyond lithium-ion battery electrolytes In the pursuit of energy sustainability, sodium and potassium battery electrolyte materials have gained massive traction in the last few decades owing to their abundance in nature, low cost, and improved safety [18,19]. Although Naþ (2.71 V vs SHE for Naþ/Na) and Kþ (2.93 V vs SHE for Kþ/K) entails slightly higher redox potentials than Liþ (3.04 V vs SHE for Naþ/Na) [18], their larger ionic radii (1.02 Å for Naþ and 1.38 Å for Kþ) compared to Liþ (0.76 Å) deliver smaller Stokes radii in propylene carbonate (PC) than Liþ (Fig. 13.6), [18] endowing them with faster ionic mobility and low activation energies in electrolytes. Moreover, their similarities in chemistries allow the use of well-established lithium analogues which have played a vital role in the exploration of these materials.


Sodium-ion battery electrolytes

Similar to their lithium counterparts, ideal sodium electrolytes are expected to display high ionic conductivity, low viscosity, good thermal stability, and wide potential windows for their adoption into rechargeable batteries. Based on their physical and chemical properties, sodium electrolyte materials are adopted for various purposes as highlighted below.


Nonaqueous electrolytes

This class of materials has gained widespread traction as a result of their favorable ionic conductivities, excellent solubilities, and compatibility, as well as stable electrochemical performance. Most electrolytes in this class utilize sodium salts such as NaClO4, NaPF6, NaBH4, NaFSI, NaTFSI, sodium trifluoromethanesulfonate (NaOTf), and NaCF3SO3 dissolved in ester-based or ether-based organic solvents. Ester-based

Figure 13.6 Comparison of Shannon’s ionic radii and Stokes radii in water and nonaqueous solvents (such as propylene carbonate [PC]) amongst Kþ, Naþ, and Liþ ions. Reproduced with permission from Ref. [18].


Storing Energy

solvents such as EC, EMC, PC, DMC, and DEC have been widely adopted for most nonaqueous sodium-ion electrolytes, typically used alone or in combination to form binary or ternary mixtures [85]. Previous works on these electrolytes have shown mixtures of PC and EC solvents to have the best film-forming capability when used alongside Na-salts such as NaPF6 and NaClO4 with hard carbon electrodes [19]. In fact, capacity values of over 200 mAh g1 at high current densities have been reported after 180 cycles. A major disadvantage in the implementation of sodium-ion batteries lies in the high reactivity of sodium electrolytes with common electrodes. As a means to mitigate these undesirable side reactions, additives such as fluoroethylene carbonate (FEC), vinylene carbonate (VC), and ethylene sulfite (ES) are often added to improve their SEI formation capabilities. In sodium-ion batteries, the use of FEC additives has been found to form the most robust SEI layers with homogenous Naþ-conducting structures on the anodes.


Aqueous electrolytes

In sodium-ion batteries (SIBs), aqueous solutions have been pursued for their safety, sustainability, low costs, and excellent corrosion resistance [86]. These electrolytes utilize Na salts such as Na2SO4, NaNO3, and NaClO4 dissolved in deionized water. The most commonly used salt is Na2SO4 due to its nonflammability and high ionic conductivity. However, this class of electrolytes, like their lithium analogs, has inherently narrow voltage windows and low energy densities, making them a less popular choice among high-performance battery materials.


Ionic liquids

Much like their lithium analogs, ionic liquids have very wide electrochemical windows, with negligible volatility and nonflammability [72]. Among them imidazolium-based ionic liquids have low viscosities and high-ionic conductivity. [72]. In fact, these materials have been found to have conductivities of up to 5.5 mS cm1 at room temperature with a wide thermal stability window of up to 150 C. On the other hand, pyrrolidinium-based electrolytes for SIBs have been reported to have relatively low conductivities at room temperatures but with significant improvements noted when the temperatures were increased [72]. Reports have suggested that high temperatures unlock the performance of some ILs, implying possible prospects of waste heat repurposing applications. Nonetheless, this class of electrolytes is usually costlier to implement compared with their organic counterparts.


Polymer electrolytes

Although polymer electrolytes have been considered by many as the next-generation batteries, their advancements are still inhibited by inferior interfacial properties and poor compatibility with electrolytes and electrodes [67]. However, these issues could be solved through the introduction of metal alloys or intermediate layers.

The road to potassium-ion batteries


In this class of electrolytes, we focus on solid polymer electrolytes and gel ceramic electrolytes as they are the most commonly used categories today. Solid polymers have been widely studied in sodium systems, as they show high design flexibility as well as excellent mechanical properties that accommodate volume changes during charge and discharge [87]. Most SIB solid polymer electrolytes are synthesized from polyacrylonitrile (PAN), poly-methyl-methacrylate (PMMA), polyvinyl alcohol (PVA), and polyethylene oxide (PEO). A remarkable milestone for this class of SIB components has been the use of nanosized TiO2 alongside PEO which has been found to display remarkable improvements in ion transport due to the improved mobility of the PEO chain [87]. At room temperatures, gel ceramic electrolytes generally exhibit better ionic conductivities than the solid polymer electrolytes, and thus have gained more utility among SIBs. Among them, sulfide-based gel ceramic electrolytes have gained most attention due to their favorable mechanical properties and low-grain boundary resistance. By utilizing salts with low activation energies, such as Na3PS4, extremely high ion conductivities exceeding 4.6  104 S cm1 can be achieved.


Potassium-ion battery electrolytes

As discussed earlier, the large potassium ionic radius entails small Stokes radius that allow them easier mobility than the aforementioned counterparts [18]. Besides the possible combination of high-voltage cathode materials such as Prussian blue analogs [53], polyanionic compounds [60], or honeycomb-layered oxide materials [31,63e65], potassium-ion chemistries not only avail a sustainable solution to the current energy issues but also promise to cut down production costs. Particularly, during the synthesis on electrolytes, sodium and lithium salts tend to have complex synthesis process, which require the use of potassium precursor materials; thus, the use of potassium not only reduces production turn-around times but also allows the use of locally available resources. Despite these prospects, potassium-based electrolytes still face some major obstacles in their implementation. Below is a list of issues hindering the advancement of these materials as reported [88]. 1. The poor solubility of potassium salts in ester electrolyte solvents. Although ether solvents typically have favorable solubilities with potassium salts, there is a high potential platform when they collocate with carbon-based materials and poor antioxidative at high potential, which may limit their application. 2. The high reactivity of K metal tends to create severe side reactions between the electrode and the electrolyte, creating safety issues in the use of potassium-ion batteries. 3. Potassium suffers cathodic dissolution in organic solvents, limiting their practical performance. 4. Different SEI chemistry from that of Li-ion and Na-ion. K is a very soft cation and it prefers to precipitate with large anions. So far, literature on these mechanisms is scarce.



Storing Energy

Solvents for electrolytes

Particularly for nonaqueous solutions, organic solvents form a crucial part of electrolyte, as they closely relate to the performance of electrolyte. An ideal electrolyte solvent must meet the following requirements [18]: 1) Have low viscosity, so as to facilitate the transmission of K-ions; 2) Dissolve metal salts in sufficient concentration, i.e., have a high dielectric constant; 3) Have a wide temperature stability range, the solvent remains liquid, that is, the solvent itself has a higher boiling point and a lower melting point; 4) Have low reactivity with other components of the battery. Ideally, the solvent should be inert with charged cathode and anode especially during the operation process; 5) Must be safe with a high flash point, low price, and low toxicity.

As mentioned earlier, ester-based solvents have poor solubility with potassium salts; thus common solvents tend to be limited to organic ether families (dimethyl ether [DME], dioxane [DOL], and glycol dimethyl ether) [18]. Nonetheless, explorations into organic esters such as carbonic acid (ethylene carbonate [EC], diethyl carbonate [DEC], propylene carbonate [PC], and dimethyl carbonate [DMC]) have also been made [88].


Potassium salts for electrolytes

The commonly used potassium salts are KClO4, KBF4, KPF6, KTFSI, and KFSI. The poor solubility of KClO4 and KBF4 have limited their wide application in K-ion batteries [89]. Meanwhile, KPF6, KFSI, and KTFSI have sufficient solubility (>0.5 mol dm3) in most solvents. Thus, they tend to be more popular for nonaqueous electrolytes. Due to the moderate solubility of KPF6 in esters and the passivation effect of PF 6 anions on aluminum current collector, KPF6-based electrolyte has been extensively studied and applied in K-ion batteries [18]. However, they usually have a low coulombic efficiency and large irreversible capacity in KPF6-based electrolyte because it is inclined to form an unstable interfacial film on the electrode surface [18]. Imides (KFSI, KTFSI, etc.) as new-type potassium salts have been proposed to replace traditional ones. KFSI-based electrolytes have been utilized to lower the polarization and form a stable SEI layer on K metal anode. Moreover, it is also reported that KFSI-based electrolyte can improve electrochemical performance of other electrode materials, such as carbonaceous materials, metal oxides, metal sulfides, etc. Nevertheless, imide-based electrolytes always exhibit corrosion reaction with the aluminum current collector at high potential (>4 V); it has become a tricky issue to be solved in this electrolyte. Stability against aluminum current collector corrosion has been confirmed in the use of highly concentrated electrolytes (such as concentrated KFSIeDME electrolyte [7 mol kg1], KTFSI-DME, etc.) [90]. It has been pointed out that with salt concentration increase, the interaction between Kþ and TFSI (or FSI) will be enhanced and the free solvent molecules reduced, so a high concentrated electrolyte usually has a good compatibility with the aluminum current collector [18].

The road to potassium-ion batteries


Binary salt electrolytes (such as KPF6eKFSI) have also been explored, as they display better stability and improved electrochemical performance when coupled to electrodes compared with single salt KPF6 electrolyte [91]. There is no doubt that the salts have significant effect on the electrochemical performance of K-ion batteries. Overall, KFSI-based electrolyte is favorable to form a more uniform and robust SEI and exhibits a more excellent cycling performance than other potassium salt electrolytes. Nevertheless, it is necessary to avoid the aluminum corrosion in KFSI-based electrolytes at high potentials (high voltage regimes). Due to the solvation effect of concentrated electrolytes, a stable interface can be formed, and the corrosion of collector can be retarded [18]. Moreover, the binary salt electrolyte has also attracted the attention of researchers due to low viscosity and cost [91].


Electrolyte additives

As known, electrolyte additives can effectively improve the performance of Li-ion and Na-ion batteries, especially fluoroethylene carbonate (FEC) has been extensively reported. Therefore, the influence of FEC on K-ion electrolyte has also been studied in recent years [92e94].


Novel electrolytes

Owing to the excellent mechanical strength, high ionic conductivity, and low grainboundary resistance, polymer electrolytes have also been considered as electrolyte candidates for K-ion batteries [95e97]. Ionic liquids, which are composed of cations and anions of salts without any solvents, have very low vapor pressure and good ionic conductivity [65]. Due to the lower Lewis acidity of K-ion, K-based ionic liquids display lower viscosity and higher ionic conductivity than Li and Na. Furthermore, electrochemical measurements illustrate that the redox potential of K metal deposition/dissolution is lower than for lithium in nonaqueous electrolytes, resulting in a wider electrochemical stable window [65,98]. Recently, aqueous-based electrolytes have also attracted attention of researchers due to their high safety and low cost. Side reactions and electrode dissolution are common problems for such batteries. However, highly concentrated salts of aqueous electrolytes such as 30 mol dm3 potassium acetate (KAc) aqueous electrolyte and 22 mol dm3 KCF3SO3-aqueous electrolyte provide wide voltage window. Moreover, due to the strong Kþ-solvation in the concentrated electrolyte, the reduction of free water was effectively suppressed; thus, the electrolyte exhibits excellent dissolutioninhibiting property for electrode materials. To date, K-ion battery is still in its infancy, and there are many challenges, especially the electrolyte. For ester-based electrolytes, they always have severe side reactions due to the strong reducibility of the K metal, resulting in low coulombic efficiency and short cycle life. In comparison, ether-based electrolytes display great stability, which facilitates the formation of stable SEI and minor side reactions. Nonetheless, ether-based electrolytes readily oxidize at high voltages, which limits the


Storing Energy

energy density of batteries. KFSI-based electrolyte is favorable, as it forms a more uniform and robust SEI, but its compatibility with current collector needs to be considered in the future [18]. Usually, concentrated electrolytes facilitate a stable interphase and inhibit corrosion reaction due to the solvation effect. In addition, the binary salt electrolytes display excellent performance in stabilizing interface and inhibiting current collector corrosion, and have a price advantage compared with concentrated electrolyte. Solid electrolytes with wide electrochemical window and excellent thermal stability are expected to achieve reversible charge/discharge reactions at high voltages [99e101]. In particular, owing to their good mechanical properties, solid electrolytes can inhibit dendrite growth and improve interface stability. Solid electrolytes can be considered as competitive candidates for K-ion batteries if their ionic conductivity and manufacturing process are improved. Ionic liquid electrolytes can avail negligible volatility and electrochemical stability, but there are still few studies in this area [65,102]. In addition, owing to the low-cost and high-safety, aqueous batteries have attracted attention as well. But the issues of low energy density (10e50 Wh L1) and severe dissolution of electrode materials need to be solved. There are still many challenges in the development of an ideal electrolyte with good reliability. Future research should address the following aspects: 1) Reduce or control side reactions between potassium metal and electrolyte. Improving the solubility of potassium in ester-based electrolytes may help design a stable interphase and reduce side reactions. Although concentrated electrolytes can regulate the polarization and reduce the side reactions by solvation effect, their resistance to oxidation at high voltages should be considered. KFSI-based electrolytes have been widely studied for their advantages in constructing stable interphases, but their corrosion to the current collector at high voltages should be addressed. 2) The formation mechanism and charge transfer of SEI layer for K-ion batteries are still unclear, and a lot of investigations need to be done in the future. Due to the high reactivity of potassium, advanced in situ characterization methods (such as cryo-electron microscopy and in situ transmission electron microscopy [TEM]) in conjunction with theoretical simulations can be used to explore the reaction mechanism, microstructures, and charge transfer kinetics of electrode/electrolyte interphase.

Another pursuit that needs to be done is the search for more suitable new types of potassium salt systems that can confer a stable interphase and avoid corrosion to the collector simultaneously, such as the aforementioned binary salt electrolytes, etc. With the further development of the chemistry of electrolytes and their associated electrode/electrolyte interphase, we believe that K-ion batteries with a high energy density, low cost, and multiscale usage will better serve our daily life.


Anode materials


Anode materials for rechargeable lithium-ion batteries

A key component that has paved the way for the success story of lithium-ion batteries is graphite, which has served for over decades as a lithium-ion host structure for the anode (negative electrode).

The road to potassium-ion batteries


There are three processes via which anode materials for Li-ion batteries and other rechargeable batteries operate in: (i) (de)intercalation, (ii) conversion reaction, and (iii) (de)alloying reaction [103]. For (de)intercalation process, Li ions, for instance, are electrochemically inserted (intercalated) into the space between the layers of materials and vice versa. This is the process by which carbonaceous anode materials such as graphite operate in. Intercalation processes possess a lot of advantages with high discharge/charge efficiency. The major drawback of intercalation process is the occurrence of some irreversible reactions during the lithium-ion insertion (lithiation) process, which causes the decomposition of a number of electrolyte constituents. Another disadvantage is the low capacities of intercalation electrodes such as graphite. Therefore, the hunt for other carbonaceous materials has been pursued in order to obtain better performance. As for conversion reactions, the theoretical capacity of anode materials operating through the conversion reaction is remarkably high. The major shortcomings are pulverization and “electric isolation.” This is because large volume changes in the electrode materials are known to lead to “electrical isolation” of the active materials, resulting in capacity attenuation during successive (dis)charge (also referred to as cycling). To ameliorate the huge volume changes accompanying conversion-type reactions, anode materials with high surface area structures have been employed (such as mesoporous materials). Alloying reactionebased processes entail the formation of alloys of tin, silicon, and germanium (and so forth) with lithium, for instance. These alloying reactionebased materials are well-known to exhibit high specific capacity. Reducing the huge volume expansion is also a challenge for (de)alloying processes, and to circumvent this nanosizing has been deemed an effective strategy. The major anode materials used for rechargeable Li-ion batteries are: (1) (2) (3) (4) (5) (6) (7)

graphite and nanostructured carbonaceous materials metal oxides metal nitrides metal sulfides metal phosphides elements such as silicon, germanium, tin, phosphorous, antimony, indium, etc. organic compounds such as dilithium rhodizonate (LiC6O6)

Despite extensive research efforts to find suitable alternatives with enhanced power and/or energy density, while maintaining excellent cycling stability, graphite is still used in the great majority of presently available commercial lithium-ion batteries [104]. The intercalation process of not only Li ions but also other alkali ions such as K into graphite (or what is referred to as the “staging” process), based on various models such as R€udorff-Hofmann and Daumas-Herold model, has generated great attention from a fundamental point of view [18]. It is notable that the theoretical specific capacity of graphite is 372 mAh g1, which is higher than the capacity of most commonly utilized cathode materials, but lower than the capacity obtained from anode materials that undergo alloying- or conversion-type reactions. Nonetheless, the final energy density attained when


Storing Energy

designing a full battery is contingent not only on the capacity attained but also on the battery voltage. In this respect, graphite has a salient advantage when compared to the other anode materials, owing to its lowest average lithiation and delithiation potential (i.e., 0.2 V versus Li/Liþ) apart from metallic lithium. Moreover, a comparative study of different anode materials and their effect on the practical full-battery energy densities reveal that most anode materials other than graphite would provide lower energy densities, explaining why graphite is still the numero uno anode material in most commercially used lithium-ion batteries. In addition, the ratio of energy that is stored (during charging process) relative to the amount of energy that is released upon discharging process (i.e., energy efficiency) is directly linked to the difference between the charge and discharge potential (taking into account a constant coulombic efficiency). This difference (also referred to as voltage hysteresis) is relatively low in graphite, rendering it additionally favorable compared to most alternatives investigated so far, although some progress has been reported very recently for (prelithiated) alloying/conversion-type materials when limiting the lithiation and delithiation processes to a rather narrow potential range. Therefore, carbonaceous materials such as graphite can be envisaged to be among the opposite anode contenders for the next-generation sustainable batteries such as potassium-ion batteries. Among the major challenges for graphite anodes in LIBs that also have to be tackled when utilized in other battery systems are the limited rate capability (which is contingent on the operating temperature, particle morphology, electrode architecture, etc.), especially for the lithiation process, i.e., the charging of the full battery, and the associated risk of lithium metal plating on the electrode surface, potentially resulting in a short circuiting of the cell or at least rapid aging and accelerated cell fading. Apart from the rather poor rate capability and safety concerns due to low delithiation potential (voltage), another drawback of graphite anode in LIBs is the irreversible capacity at the initial cycle due to the reductive electrolyte decomposition and, thus, the consumption of Liþ as the charge carrier. To address this, a number of strategies such as surface modification and utilization of electrolyte additives have been employed.


Anode materials for rechargeable sodium-ion batteries (SIBs)

Research on anode materials for sodium-ion batteries has been developed on four different categories, which are basically the same as those for LIB [19]: (1) carbonaceous materials (i.e., hard and soft carbon). (2) oxides and polyanionic compounds (such as phosphates) as topotactic insertion materials for sodium (oxides include Na2Ti3O7, Na2Ti6O13, Na0.66[Li0.22Ti0.78]O2, Na3LiTi5O12; polyanion-based compounds include sodium superionic conductor (NASICON)-type NaTi2(PO4)3).

The road to potassium-ion batteries


(3) p-block elements (metals, alloys, phosphorus/phosphides [such as MnP4]) showing reversible sodiation/desodiation (Group 14 and 15 elements, including metals [Sn, Pb, Bi], metalloids [Si, Ge, As, Sb], metal alloys [such as SnSb, NaK], and polyatomic nonmetals [P] are known to form binary compounds with Na. These electrode materials, alloying with Na or forming Na binary compounds, have been studied as potential anode materials for rechargeable SIBs and have the particularity to interact with a larger number of Na (compared to insertion (intercalation)-type materials), leading to a much higher capacity than hard carbon and titanium oxides). (4) metal oxides, sulphides and selenides with conversion reaction. (oxides include TiO2, CuO, SnO2, Sb2O4, Fe3O4, etc.; sulphides include TiS2, Cu2S, FeS, FeS2, MoS2, Ni3S2, etc.; selenides include Cu2Se.) and(5) organic compounds such as disodium terephthalate Na2C8H4O4, disodium rhodizonate (Na2C6O6), disodium croconate (Na2C5O5), 2,5dihydroxyterephthalic acid (Na4C8H2O6) bifunctional electrode, etc.

For LIBs, graphite is widely used as anode material in comparison with other carbonaceous materials because of its high gravimetric and volumetric capacity. Graphite electrodes deliver a reversible practical capacity of more than 360 mAh g1, comparable to the theoretical capacity of 372 mAh g1. By electrochemical reduction, Liþ ions are inserted between graphene layers and Liegraphite intercalation compounds are formed with stage transformations. At the first stage, all graphite layers are completely filled by Li, forming LiC6 at the end of the electrochemical reduction process. However, graphite is electrochemically less active in Na cells. Although a small amount of Na atoms seems to be inserted into the graphite by heating with Na metal under vacuum or helium atmosphere and by electrochemical reduction, resulting in formation of NaC64 [105]. NaC64 possesses a Na insertion amount that is far much smaller than that for Li and K insertion into graphite (i.e., LiC6 and KC8). In 2014, reversible Na intercalation into graphite in diglyme-based electrolyte was demonstrated [106]. Although Naþ ions probably intercalate together with diglyme molecules into graphite (a process referred also as cointercalation) and the Na/graphite cell exhibits a reversible capacity of ca. 100 mAh g1 with excellent capacity retention, the reversible capacity attained is still lower than that of Li/graphite and K/graphite cells. Although graphite is not a suitable anode material for sodium-ion batteries, relatively larger reversible capacity has been attained using carbonaceous anode materials such as soft carbon (graphitisable carbon) and hard carbon (referred also as nongraphitisable carbon) [19]. Hard and soft carbons have intensively been studied as anode materials for LIBs, as they exhibit high capacity and good rate performance that are ideal for high-power applications. It is worthy to mention that hard carbon had been utilized as an anode material in the first commercial LIBs, and the hard carbon electrodes delivered comparable reversible capacity without a “staging” transition owing to the inherent structural disorder. The reversible capacity attained by hard carbon exceeds the theoretical capacity of graphite, even though the volumetric capacity is lower than that of graphite. Hard-carbon electrodes can deliver ca. 300 mAh g1 of reversible capacity in Na cells [107]. In contrast to graphite, a much higher amount of Na


Storing Energy

ions can reversibly be inserted into hard carbon by electrochemical reduction in Na cells; thus, hard carbon materials have been intensively pursued as suitable anode materials for sodium-ion batteries [19]. Extensive research efforts have also been devoted to studying the reaction mechanisms for not only Li and K, but also Na (de)insertion into the disordered carbons to explain the origin of the anomalously high capacities attained [19].


Anode materials for rechargeable potassium-ion batteries (KIBs)

On the basis of the developments of carbonaceous materials particularly for LIBs, graphite is a promising anode candidate for KIBs because potassium graphite intercalation compounds (GICs) were synthesized by potassium vapor synthetic techniques between the 1920 and 1960s [108e111]. In a Li cell, graphite electrode delivers the largest reversible capacity of approximately 370 mAh g1, corresponding to the formation of LiC6. On the other hand, in a Na cell, negligible capacity is observed as electrochemical Na intercalation hardly occurs in the defect-free graphite. In contrast to Na, K is electrochemically and reversibly inserted into graphite in a K cell (as shown in Fig. 13.7). Reversible capacities in the range 240e260 mAh g1 are typically attained in K cells, values that are consistent with the theoretical capacity of 279 mAh g1, by assuming reversible formation of KC8 [112]. Owing to the stable graphite intercalation compounds of K- and Li-containing structures, graphite has been utilized as anode material for both KIBs and LIBs. This indicates that the industrial knowledge and technologies (supply chain, etc.) accumulated for LIB graphite electrodes in their processing could be applied directly to KIBs. Thus, graphite can be envisaged to be broadly

Figure 13.7 Electrochemical properties of graphite when evaluated in Li-, Na-, and K-cells. Reproduced with permission from Ref. [89].

The road to potassium-ion batteries


utilized as anode material in KIBs as well, as a result of its high gravimetric and volumetric capacities, low operation potential, and high electronic conductivity. The low operation potential of graphite in K-cell (which is slightly higher than that in Li-cell [see also Fig. 13.7]) is advantageous in bringing out the full cell voltage of KIBs when coupled to suitable high-voltage cathode materials. It also obviates the likely formation of K metal dendrites that usually occurs at lower potentials (close to 0 V vs Kþ/K). Furthermore, rate capability observed in the K/graphite cell has been reported to be superior to that of the Li/graphite cells [18]. This means that one can design a potassium-ion battery system with high power density than the current LIBs using the same material configuration. Nevertheless, the electrochemical properties of the K/graphite cells are influenced by the binder and electrolyte, the structural properties of graphite, etc.; thus, a number of intensive researches are being done to understand these aspects. Apart from graphite, carbonaceous materials such as hard and soft carbon have been intensively pursued as anode materials for KIBs, owing to their increased reversible capacity. For instance, K-insertion properties have been examined for hard carbon materials derived from sustainable materials such as skimmed cotton, cellulose, waste-tire rubber, oak, rice-starch, maple leaves, loofahs, potatoes, and so forth [18]. A compendium of the anode materials investigated so far for KIBs is shown in Fig. 13.8 below. Basically, the same type of material compositions employed in both LIBs and SIBs are being pursued as anode materials for KIBs. Fig. 13.9 displays

Figure 13.8 Materials investigated as anode materials for potassium-ion batteries.


Storing Energy

Figure 13.9 Anode materials for potassium-ion batteries. Reproduced with permission from Refs. [113,114].

the capacity-voltage profiles of representative anode materials for rechargeable potassium-ion batteries along with the merits, demerits, and the challenges to be overcome.


Beyond cation intercalation chemistries

So far, we have discussed in detail the components and key mechanisms that shape the modern rechargeable batteries. However, it must be noted that rechargeable batteries are not only limited to intercalation chemistries. In fact, innumerable energy storage mechanisms have been pursued in the search of sustainable electrochemical energy devices. Below is a brief summary of rechargeable batteries that operate beyond the realms of cation intercalation.

7.1 7.1.1

Rechargeable metal-chalcogenide batteries Rechargeable metal-sulfur batteries

Rechargeable metal-sulfur batteries represent one of the most attractive electrochemical systems in terms of cost and energy density [115]. Most of the proposed systems entail metallic (for instance, alkali metals such as Li, Na, K and alkaline-earth metals such as Mg and Ca) as the anode side while elemental sulfur (impregnated in a porous matrix) is utilized as the cathode. Despite the relatively low voltage of these systems, they have garnered attention as promising next-generation batteries owing to the following aspects: (1) sulfur as a cathode exhibits high theoretical specific capacity (1675 mAh g1), (2) utilization of metallic anodes enables a leap in the energy density due to the high capacity of metal anodes in comparison to intercalation compounds, and (3) system components make rechargeable metal sulfur batteries less toxic and

The road to potassium-ion batteries


low-cost batteries. Nevertheless, the high reactivity of metallic anodes (e.g., Li, Na, Mg, and Al) and the solubility of sulfur species (polysulfides) in the electrolyte render these batteries unstable and encumber their practical realization. Currently, intensive research efforts are being done in these systems to alleviate the aforementioned challenges.


Rechargeable metal-selenium batteries

Selenium (Se)-based materials present a tantalizing opportunity for rechargeable battery systems. Se cathode has attracted great interest due to its high volumetric capacity (3253 Ah L1) and good electrical conductivity (1.0  103 S m1) than sulfur [116]. Se is more compatible with low-cost carbonate-based electrolytes, and there is no “shuttle effect” in these electrolytesda problem that plagues the performance of metal-sulfur batteries. Nonetheless, metal-selenium batteries also suffer from dissolution of polyselenides into the electrolyte. Effective physical and chemical immobilization of Se can inhibit the dissolution of polyselenides and the side reaction between Se anions and carbonate-based electrolytes. These strategies are being pursued to improve the electrochemical performance of metal-selenium batteries.


Rechargeable metal-tellurium batteries

Tellurium (Te) is a member of the chalcogen family. Metal-Te batteries have been explored with rising enthusiasm [117]. Compared to selenium (Se) and sulfur (S), Te shows remarkable advantages, such as better stability and higher electrical conductivity. Te displays a suitable redox potential, high volumetric capacity, good conductivity, and excellent stability, making it a potent anode material candidate. The first report of metal-Te battery was in 2014, and it has since then been intensively investigated as a potential next-generation energy storage system. However, metal-Te batteries suffer from similar problems as metal-sulfur batteries, such as large electrode volume change and dissolution of intermediate telluride compositions.


Rechargeable metal-gas batteries

These types of batteries can be further classified into:


Rechargeable metal-air (oxygen) batteries [118]

In metal-air batteries, the cathode is a carbon-based material covered with some precious metals to react with oxygen. The anode is composed of a metal such as potassium, aluminum, zinc, lithium, magnesium, and so forth. Since in these batteries, air is flowing through the cell, they are sometimes categorized as fuel cells.


Rechargeable metal-CO2 (carbon dioxide) batteries [119]

Most reported metal-air batteries are actually operated in a pure oxygen atmosphere; however, metal-CO2 batteries utilize carbon dioxide thus, providing a promising


Storing Energy

“green and clean” strategy for reducing fossil fuel consumption and consequently, lessening global warming. In addition, metal-CO2 batteries are potential energy sources for scientific exploration and future immigration to Mars, for the air there contains 95% of CO2.

7.3 7.3.1

Rechargeable metal-halogen batteries Rechargeable metal-iodine batteries [120]

Metal-iodine batteries hold practical promise as next-generation electrochemical energy storage systems owing to their low cost and high electrochemical reversibility. Iodine (I2) is a promising cathode candidate due to its fast electrochemical kinetics, highly reversible redox reactions, and appreciable redox potential (1.2 V vs Zn2þ/Zn, 3.6 V vs Liþ/Li). Distinct from the complex electrochemical processes occurring in O2 and S cathode-based batteries, metal-iodine batteries have relatively simple cathodic reactions and less parasitic disruption. Furthermore, iodine also has relatively high chemical stability in the majority of commonly available solvents, even water. These advantages avail new opportunities for various electrolyteeelectrode designs geared toward practical applications that demand both high energy density and safety. Beyond that, profiting from its electrochemical reversibility, high solubility, and the suitable redox potential, iodine-involved redox has been also widely used as an electrolyte additive to make the kinetics-limited systems more sensitive and controllable.


Redox-flow batteries [121,122]

A redox flow battery is an electrochemical energy storage device that converts chemical energy into electrical energy via reversible reduction and oxidation of working fluids. The concept was initially conceived in 1970s. Sustainable and clean energy supplied from renewable sources in future necessitates reliable, cost-effective, and efficient energy storage systems. Due to the flexibility in system design and competence in scaling cost, redox flow batteries are promising in stationary storage of energy from intermittent sources such as wind and solar. Several systems, for example, zinc-cerium, zinc-bromine, vanadium-vanadium, bromine-polysulfide, iron-chromium, and soluble lead, are under development worldwide.

7.5 7.5.1

Rechargeable alkaline-earth metal batteries and related systems Rechargeable magnesium batteries [123,124]

Magnesium metal is an attractive anode due to the high abundance of magnesium and its volumetric capacity of 3833 mAh cm3 and gravimetric capacity of 2205 mAh g1 combined with a low redox potential (2.37 V vs SHE). The divalent character of Mg2þ avails two electrons per metal (Mg) compared to only one electron in the case of alkali metals, such as Li, Na, and K. Thus, with a properly designed

The road to potassium-ion batteries


electrolyte-cathode system, one can expect to obtain practical rechargeable (secondary) magnesium batteries with an energy density of at least 320 Wh kg1 or 600 Wh L1. The realization of a rechargeable Mg battery technology primarily requires identification and development of suitable electrodes and electrolytes.


Rechargeable calcium batteries [125,126]

Most Ca battery cells are analogous with LIBs on the cathode side; during the discharge, the charge carrier ion (Ca2þ) migrates from the anode to the cathode through the electrolyte, while the electrons flow across an external circuit. These processes are reversed upon charging.


Rechargeable aluminum batteries [127,128]

A rechargeable battery based on aluminum chemistry is envisioned to be a low-cost energy storage platform, considering that aluminum is the most abundant metal in the earth’s crust. The high volumetric capacity of aluminum, which is four and seven times larger than that of lithium and sodium, respectively, unarguably has the potential to boost the energy density of aluminium-batteries on a per unit volume basis. Efforts to develop rechargeable aluminum batteries can be traced to as early as the 1970s; however, this area of research has seen a surge in activity since 2010, when the possibility of achieving an ambient temperature aluminum system was convincingly demonstrated.


Rechargeable halogen batteries

Unlike reversible migration (or shuttling) of metal cations in traditional batteries, new battery-based energy storage systems are emerging that operate in a process whereby anions shuttle between two electrodes, providing more options and opportunities for electrochemical energy storage.


Rechargeable fluoride-ion batteries [129e132]

In fluoride-ion batteries, fluoride anion acts as a charge carrier ion between a metal/metal fluoride pair where it will react with metal or evolve from metal fluoride depending on the flow of current. Since fluoride is the most stable anion with a high mobility, fluoride-ion batteries can theoretically provide a very wide potential window. Apart from some limited reports on primary batteries based on fluoride ions in the 1970s, the first report on electrochemical rechargeable fluoride ion battery was reported in 2011, based on an all-solid-state battery with a solid electrolyte, Ce, Bi, etc. as metal anodes, and conversion-based metal fluorides (CuF2, CeF3, BiF3, SnF2, etc.) as cathode materials. Since then, the efforts on investigating such systems have further increased, addressing the development of solid and liquid electrolytes and improved cell designs as well as the screening of both intercalation-type and conversion-based electrode materials. Further, since fluoride ion has essentially the same size as that of potassium ion, a comparative study on fluoride ion versus potassium ion intercalation electrochemistry is an attractive research avenue.


Storing Energy


Rechargeable chloride-ion batteries [133,134]

Chloride-ion batteries are another example of a promising new emerging rechargeable anion battery technology that exhibits large theoretical volumetric energy density performance and good safety. In particular, chloride-ion batteries exhibit high theoretical volumetric energy density (w2500 W h L1), comparable with that of metal-sulfur batteries such as Li-S. To date, only a few cathode materials have been reported for chloride-ion batteries and most of them are metal chlorides/oxychlorides, such as CoCl2, CuCl2, FeOCl, BiOCl, and their hybrids. However, unsatisfactory electrochemical performance and poor structural stability of the cathode currently hinder the development of chloride-ion batteries. Therefore, the key challenge in chlorideion batteries is to develop alternative cathode materials with long cycle life, competitive capacity, and low-cost characteristics.


Rechargeable dual-ion batteries [135]

Dual-ion batteries utilize the simultaneous insertion (intercalation) of anions into the cathode and cations into the anode during the charging process and conversely deintercalation of ions into the electrolytes during the discharging process. Both cations and anions are taken from and released to the electrolyte. The anode can be a metal (such as Li, Na, K, etc.) or any material that can facilitate reversible insertion of cations. The cathode comprises materials that can facilitate reversible insertion of anions such as graphite or organic compounds. Dual-ion batteries avail high voltages and high-energy densities; thus, promising energy storage systems. However, the concentrations of both anions and cations in the electrolyte vary during the (dis) charging operation of dual-ion batteries. Considering that the electrolyte provides charge carriers in dual-ion batteries, both the power and energy density can be restricted by the electrolyte. A holistic improvement of the design of the dual-ion battery along with the optimization of all constituent components (such as electrodes, separators, and electrolytes) are among the approaches undertaken to attain dualion batteries with high energy density.



As the search for a more potent replacement of the lithium-ion battery intensifies, exploration of alternative chemistries, particularly sodium- and potassium-ion energy devices for the next-generation battery systems is on the rise. This chapter takes a three-pronged approach to elaborate the ongoing pursuit of practical rechargeable batteries. For the cathode perspective, we explore the various chemical structures that have been reported as intercalation hosts in lithium, sodium, and potassium chemistries. Across the chemistries (Li, Na, and K), Prussian analogs and polyanion-based frameworks show excellent performance, displaying high voltages of up to 4 V (and even exceeding 4 V) in nascent potassium-ion chemistries. However, they have inherently

The road to potassium-ion batteries


low electronic conductivities necessitating the incorporation of a great deal of conductive materials (such as carbon) exacerbated further by the intrinsically low material density (tap density) that reduces particularly their volumetric energy densities. On the other hand, we delve into the new class of layered oxide materialsd honeycomb-layered oxides, which exhibit remarkable capabilities such as high voltage, high material densities, remarkable ion kinetics, and exquisite crystal diversity. However, it is their exceptional performance with potassium chemistries that draws interest in this category of materials. The large ion size of Kþ, a primary disadvantage for most cathode configurations, aids in enhancing its ion mobility without compromising its voltage characteristics. A challenge for the honeycomb-layered oxides is the search for Te-free compositions and handling as some of them are fairly hygroscopic. Although sodium also shows great propensity for high-performance materials, the high electrode potential of Na (2.71 V vs SHE) relative to Li (3.05 V vs SHE) renders the design of a high-voltage Na-ion battery system a big hurdle. However, the electrode potential of K (2.95 V vs SHE) which, depending on the electrolyte is sometimes lower than that of Li, guarantees the design of a high-voltage battery system with comparable performance or even surpassing that of the current Li-ion technology. Moreover, due to its lower Lewis acidity, potassium delivers electrolytes with higher ionic conductivities than sodium and lithium; meaning that we can anticipate fast K-ion kinetics which translates to a high-power battery. Further, the possibility of integrating anode materials such as graphite, conventionally used in lithium-ion batteries without the mandatory need to alter the current supply chain, particularly makes potassium-based options more appealing as the next-generation battery chemistry. Fig. 13.10 shows the combination of electrode materials to design

Figure 13.10 Electrode combination to design high-voltage potassium-ion batteries.


Storing Energy

Figure 13.11 Capacity-voltage plots for a 4V-class potassium-ion battery designed using a Prussian analog (K2MnFe(CN)6) cathode with graphite as anode and pyrrolidinium-based ionic liquid. Reproduced with permission from Ref. [102]

a high-voltage potassium-ion battery. Fig. 13.11 shows a proof-of-concept design of a high-voltage potassium-ion battery using a combination of electrode materials shown in Fig. 13.10. However, it should be noted that despite the great prospects of potassium, the road to their commercialization is long and full of challenges. Their success will require insight into their cathode intercalation mechanisms and the development of a robust electrolyte that can cater for high-voltage performance without degradation. Certainly, achieving this vision will require concerted efforts from a range of fields.

Acknowledgments This work was supported by the TEPCO Memorial Foundation. In addition, this work was also conducted in part under the auspices of the Japan Society for the Promotion of Science (JSPS KAKENHI Grant Numbers 19K15685 and 21K14730), National Institute of Advanced Industrial Science and Technology (AIST) and Japan Prize Foundation. Both authors acknowledge the rigorous proofreading work done by Edfluent, and are extremely grateful for the unwavering support from their family members (T. M.: Ishii Family, Sakaguchi Family and Masese Family; G. M. K.: Ngumbi Family).

References [1] M. Armand, J.M. Tarascon, Building better batteries, Nature 451 (2008) 652e657. [2] A. Manthiram, A reflection on lithium-ion battery cathode chemistry, Nat. Commun. 11 (2020) 1550.

The road to potassium-ion batteries


[3] N. Nitta, F. Wu, J.T. Lee, G. Yushin, Li-ion battery materials: present and future, Mater. Today 18 (2015) 252e264. [4] M.S. Whittingham, Lithium batteries and cathode materials, Chem. Rev. 104 (2004) 4271e4301. [5] M. Amereller, T. Schedlbauer, D. Moosbauer, C. Schreiner, C. Stock, F. Wudy, S. Zugmann, H. Hammer, A. Maurer, R.M. Gschwind, H.D. Wiemhçfer, M. Winter, H.J. Gores, Lithium batteries and cathode materials, Prog. Solid State Chem. 42 (2014) 39e56. [6] E. Peled, S. Menkin, ReviewdSEI: past, present and future, J. Electrochem. Soc. 164 (2017) A1703. [7] S.J. An, J. Li, C. Daniel, D. Mohanty, S. Nagpure, D.L. Wood III, The state of understanding of the lithium-ion-battery graphite solid electrolyte interphase (SEI) and its relationship to formation cycling, Carbon 105 (2016) 52e76. [8] J.-M. Tarascon, Is lithium the new gold? Nat. Chem. 2 (2010) 510. [9] K. Mizushima, P.C. Jones, P.J. Wiseman, J.B. Goodenough, LixCoO2 (0 < x < 1): a new cathode material for batteries of high energy density, Mater. Res. Bull. 15 (1980) 783e798. [10] M.M. Thackeray, W.I.F. David, J.B. Goodenough, Structural characterization of the lithiated iron oxides LixFe3O4 and LixFe2O3 (0 < x < 2), Mater. Res. Bull. 17 (1982) 785e793. [11] M.M. Thackeray, W.I.F. David, P.G. Bruce, J.B. Goodenough, Lithium insertion into manganese spinels, Mater. Res. Bull. 18 (1983) 461e472. [12] M.M. Thackeray, Structural considerations of layered and spinel lithiated oxides for lithium ion batteries, J. Electrochem. Soc. 142 (1995) 2558e2563. [13] A. Manthiram, J.B. Goodenough, Lithium insertion into Fe2(SO4)3 framework, J. Power Sourc. 26 (1989) 403e406. [14] A. Manthiram, J.B. Goodenough, Lithium insertion into Fe2(MO4)3 frameworks: comparison of M ¼ W with M ¼ Mo, J. Solid State Chem. 71 (1987) 349e360. [15] C. Masquelier, L. Croguennec, Polyanionic (phosphates, silicates, sulfates) frameworks as electrode materials for rechargeable Li (or Na) batteries, Chem. Rev. 113 (2013) 6552e6591. [16] H. Huang, S.C. Yin, T. Kerr, N. Taylor, L.F. Nazar, Nanostructured composites: a high capacity, fast rate Li3V2(PO4)3/carbon cathode for rechargeable lithium batteries, Adv. Mater. 14 (2002) 1525e1528. [17] J.-M. Tarascon, Key challenges in future Li-battery research, Phil. Trans. R. Soc. A. 368 (2010) 3227e3241. [18] T. Hosaka, K. Kubota, A.S. Hameed, S. Komaba, Research development on K-ion batteries, Chem. Rev. 120 (2020) 6358e6466. [19] N. Yabuuchi, K. Kubota, M. Dahbi, S. Komaba, Research development on sodium-ion batteries, Chem. Rev. 114 (2014) 11636e11682. [20] K. Hurlbutt, S. Wheeler, I. Capone, M. Pasta, Removal of interstitial H2O in hexacyanometallates for a superior cathode of a sodium-ion battery, J. Am. Chem. Soc. 137 (2015) 2658e2664. [21] J. Song, L. Wang, Y. Lu, J. Liu, B. Guo, P. Xiao, J.J. Lee, X. Yang, G. Henkelman, J.B. Goodenough, Prussian blue analogs as battery materials, Joule 2 (2018) 1950e1960. [22] Y. Xu, M. Zhou, Y. Lei, Organic materials for rechargeable sodium-ion batteries, Mater. Today Off. 21 (2018) 60e78. [23] V.-A. Oltean, S. Renault, M. Valvo, D. Brandell, Sustainable materials for sustainable energy storage: organic Na electrodes, Materials 9 (2016) 142.


Storing Energy

[24] T. Jin, H. Li, K. Zhu, P.-F. Wang, P. Liu, L. Jiao, Polyanion-type cathode materials for sodium-ion batteries, Chem. Soc. Rev. 49 (2020) 2342e2377. [25] Z. Lv, M. Ling, M. Yue, X. Li, M. Song, Q. Zheng, H. Zhang, Vanadium-based polyanionic compounds as cathode materials for sodium-ion batteries: toward high-energy and high-power applications, J. Ener. Chem. 55 (2021) 361e390. [26] P. Barpanda, L. Lander, S.-i. Nishimura, A. Yamada, Polyanionic insertion materials for sodium-ion batteries, Adv. Ener. Mater. 8 (2018) 1703055. [27] P. Barpanda, G. Oyama, S.-i. Nishimura, S.-C. Chung, A. Yamada, A 3.8-V earthabundant sodium battery electrode, Nat. Commun. 5 (2014) 4358. [28] P. Barpanda, G.D. Liu, C.D. Ling, M. Tamaru, M. Avdeev, S.C. Chung, Y. Yamada, A. Yamada, Na2FeP2O7: a safe cathode for rechargeable sodium-ion batteries, Chem. Mater. 25 (2013) 3480e3487. [29] Y. Kawabe, N. Yabuuchi, M. Kajiyama, N. Fukuhara, T. Inamasu, R. Okuyama, I. Nakai, S. Komaba, Synthesis and electrode performance of carbon coated Na2FePO4F for rechargeable Na batteries, Electrochem. Commun. 13 (2011) 1225e1228. [30] Y. Hironaka, K. Kubota, S. Komaba, P2- and P3-KxCoO2 as an electrochemical potassium intercalation host, Chem. Commun. 53 (2017) 3693e3696. [31] G.M. Kanyolo, et al., Honeycomb layered oxides: structure, energy storage, transport, topology and relevant insights, Chem. Soc. Rev. 50 (2021) 3990e4030. [32] Z. Yang, et al., A high-voltage honeycomb-layered Na4NiTeO6 as cathode material for Na-ion batteries, J. Power Sourc. 360 (2017) 319e323. [33] D. Yuan, et al., A honeycomb-layered Na3Ni2SbO6: a high-rate and cycle-stable cathode for sodium-ion batteries, Adv. Mater. 26 (2014) 6301e6306. [34] C.-Y. Chen, et al., High-voltage honeycomb layered oxide positive electrodes for rechargeable sodium batteries, Chem. Commun. 56 (2020) 9272e9275. [35] D.S. Bhange, et al., Honeycomb-layer structured Na3Ni2BiO6 as a high voltage and long life cathode material for sodium-ion batteries, J. Mater. Chem. 5 (2017) 1300e1310. [36] L. Zheng, M.N. Obrovac, Honeycomb compound Na3Ni2BiO6 as positive electrode material in Na cells, J. Electrochem. Soc. 163 (2016) A2362eA2367. [37] J. Ma, et al., Ordered and disordered polymorphs of Na(Ni2/3Sb1/3)O2: honeycombordered cathodes for Na-ion batteries, Chem. Mater. 27 (2015) 2387e2399. [38] A. Gupta, G.B. Mullins, J.B. Goodenough, Na2Ni2TeO6: Evaluation as a cathode for sodium battery, J. Power Sourc. 243 (2013) 817e821. [39] M.A. Evstigneeva, V.B. Nalbandyan, A.A. Petrenko, B.S. Medvedev, A.A. Kataev, A new family of fast sodium ion conductors: Na2M2TeO6 (M ¼ Ni, Co, Zn, Mg), Chem. Mater. 23 (2011) 1174e1181. [40] C. Deng, P. Skinner, Y. Liu, M. Sun, W. Tong, C. Ma, M.L. Lau, R. Hunt, P. Barnes, J. Xu, H. Xiong, Li-substituted layered spinel cathode material for sodium ion batteries, Chem. Mater. 23 (2018) 8145e8154. [41] X. Liu, X. Wang, A. Iyo, H. Yu, D. Li, H. Zhou, High stable post-spinel NaMn2O4 cathode of sodium ion battery, J. Mater. Chem. 2 (2014) 14822e14826. [42] R.R. Kapaev, P.A. Troshin, Organic-based active electrode materials for potassium batteries: status and perspectives, J. Mater. Chem. 8 (2020) 17296e17325. [43] H. Kim, H. Ji, J. Wang, G. Ceder, Next-generation cathode materials for non-aqueous potassium-ion batteries, Trends Chem. 1 (2019) 682e692. [44] R. Rajagopalan, Y. Tang, X. Ji, C. Jia, H. Wang, Advancements and challenges in potassium ion batteries: a comprehensive review, Adv. Funct. Mater. 30 (2020) 1909486.

The road to potassium-ion batteries


[45] M. Chen, Q. Liu, Y. Zhang, G. Xing, S.-L. Chou, Y. Tang, Emerging polyanionic and organic compounds for high energy density, non-aqueous potassium-ion batteries, J. Mater. Chem. 8 (2020) 16061e16080. [46] Y.-S. Xu, S.-Y. Duan, Y.-G. Sun, D.-S. Bin, X.-S. Tao, D. Zhang, Y. Liu, A.-M. Cao, L.J. Wan, Recent developments in electrode materials for potassium-ion batteries, J. Mater. Chem. 7 (2019) 4334e4352. [47] Z. Jian, Y. Liang, I.A. Rodríguez-Pérez, Y. Yao, X. Ji, Poly(Anthraquinonyl sulfide) cathode for potassium-ion batteries, Electrochem. Commun. 71 (2016) 5e8. [48] Y. Chen, W. Luo, M. Carter, L. Zhou, J. Dai, K. Fu, S. Lacey, T. Li, J. Wan, X. Han, Organic electrode for non-aqueous potassium-ion batteries, Nanomater. Ener. 18 (2015) 205e211. [49] Q. Zhao, J. Wang, Y. Lu, Y. Li, G. Liang, J. Chen, Oxocarbon salts for fast rechargeable batteries, Angew. Chem. Int. Ed. 55 (2016) 12528e12532. [50] A.S. Hameed, A. Katogi, K. Kubota, S. Komaba, A layered inorganiceorganic open framework material as a 4 V positive electrode with high-rate performance for K-ion batteries, Adv. Ener. Mater. 9 (2019) 1902528. [51] X. Wu, Z. Jian, Z. Li, X. Ji, Prussian white analogues as promising cathode for nonaqueous potassium-ion batteries, Electrochem. Commun. 77 (2017) 54e57. [52] S. Zhao, Z. Guo, K. Yan, X. Guo, S. Wan, F. He, B. Sun, G. Wang, The rise of prussian blue analogs: challenges and opportunities for high-performance cathode materials in potassium-ion batteries, Small 2 (2021) 2000054. [53] A. Zhou, W. Cheng, W. Wang, Q. Zhao, J. Xie, W. Zhang, H. Gao, L. Xue, J. Li, Hexacyanoferrate-type prussian blue analogs: principles and advances toward high-performance sodium and potassium ion batteries, Adv. Ener. Mater. 11 (2021) 2000943. [54] P. Padigi, J. Thiebes, M. Swan, G. Goncher, D. Evans, R. Solanki, Prussian green: a high rate capacity cathode for potassium ion batteries, Electrochim. Acta 166 (2015) 32e39. [55] Z. Gong, Y. Yang, Recent advances in the research of polyanion-type cathode materials for Li-ion batteries, Ener. Environ. Sci. 4 (2011) 3223e3242. [56] J. Han, et al., Investigation of K3V2(PO4)3/C nanocomposites as high-potential cathode materials for potassium-ion batteries, Chem. Commun. 53 (2017) 1805e1808. [57] K. Chihara, A. Katogi, K. Kubota, S. Komaba, KVPO4F and KVOPO4 toward 4 voltclass potassium-ion batteries, Chem. Commun. 53 (2017) 5208e5211. [58] W.B. Park, et al., KVP2O7 as a robust high-energy cathode for potassium-ion batteries: pinpointed by a full screening of the inorganic registry under specific search conditions, Adv. Ener. Mater. 8 (2017) 1703099. [59] I.V. Tereshchenko, D.A. Aksyonov, A. Zhugayevych, E.V. Antipov, A.M. Abakumov, Reversible electrochemical potassium deintercalation from >4 V positive electrode material K6(VO)2(V2O3)2(PO4)4(P2O7), Solid State Ion. 357 (2020) 115468. [60] T. Hosaka, T. Shimamura, K. Kubota, S. Komaba, Polyanionic compounds for potassium-ion batteries, Chem. Rec. 19 (2019) 735e745. [61] H. Kim, J.C. Kim, S.eH. Bo, T. Shi, D.eH. Kwon, G. Ceder, K-Ion batteries based on a P2-type K0.6CoO2 cathode, Adv. Mater. 7 (2017) 1700098. [62] K. Sau, T. Ikeshoji, G.M. Kanyolo, T. Masese, Correlation between the Interlayer Distance and Cationic Diffusion in Nickel-Based Honeycomb Layered Tellurates, 2021. [63] T. Masese, et al., Rechargeable potassium-ion batteries with honeycomb-layered tellurates as high voltage cathodes and fast potassium-ion conductors, Nat. Commun. 9 (2018) 3823.


Storing Energy

[64] T. Masese, et al., A high voltage honeycomb layered cathode framework for rechargeable potassium-ion battery: P2-type K2/3Ni1/3Co1/3Te1/3O2, Chem. Commun. 55 (2019) 985e988. [65] K. Yoshii, et al., Sulfonylamide-based ionic liquids for high-voltage potassium-ion batteries with honeycomb layered cathode oxides, ChemElectroChem 6 (2019) 3901e3910. [66] N. Matsubara, et al., Magnetism and ion diffusion in honeycomb layered oxide K2Ni2TeO6, Sci. Rep. 10 (2020) 18305. [67] Q. Li, J. Chen, L. Fan, X. Kong, Y. Lu, Progress in electrolytes for rechargeable Li-based batteries and beyond, Green Ener. Environ. 1 (2016) 18e42. [68] S. Tan, Y.J. Ji, Z.R. Zhang, Y. Yang, Recent progress in research on high-voltage electrolytes for lithium-ion batteries, ChemPhysChem 15 (2014) 1956e1969. [69] W. Li, J.R. Dahn, D.S. Wainwright, Rechargeable lithium batteries with aqueous electrolytes, Science 264 (1994) 1115e1118. [70] C. Yang, J. Chen, T. Qing, X. Fan, W. Sun, A. von Cresce, M.S. Ding, O. Borodin, J. Vatamanu, M.A. Schroeder, N. Eidson, C. Wang, K. Xu, 4.0 V aqueous Li-ion batteries, Joule 15 (2017) 122e132. [71] M. Watanabe, T.L. Thomas, S. Zhang, K. Ueno, T. Yasuda, K. Dokko, Application of ionic liquids to energy storage and conversion materials and devices, Chem. Rev. 117 (2017) 7190e7239. [72] K. Matsumoto, J. Hwang, S. Kaushik, C.-Y. Chen, R. Hagiwara, Advances in sodium secondary batteries utilizing ionic liquid electrolytes, Ener. Environ. Sci. 12 (2019) 3247e3287. [73] H. Qi, Y. Ren, S. Guo, Y. Wang, S. Li, Y. Hu, F. Yan, High-voltage resistant ionic liquids for lithium-ion batteries, ACS Appl. Mater. Interfaces 12 (2020) 591e600. [74] A. Lewandowski, A. Swiderska-Mocek, Lithium-metal potential in Liþ containing ionic liquids, J. Appl. Electrochem. 40 (2010) 515e524. [75] S. Fang, Z. Zhang, Y. Jin, L. Yang, S.-i. Hirano, K. Tachibana, S. Katayama, New functionalized ionic liquids based on pyrrolidinium and piperidinium cations with two ether groups as electrolytes for lithium battery, J. Power Sourc. 196 (2011) 5637e5644. [76] G.B. Appetecchi, M. Montanino, D. Zane, M. Carewska, F. Alessandrini, S. Passerini, Effect of the alkyl group on the synthesis and the electrochemical properties of N-alkyl-Nmethyl-pyrrolidinium bis(trifluoromethanesulfonyl)imide ionic liquids, Electrochim. Acta 54 (2009) 1325e1332. [77] N. Wongittharom, C.-H. Wang, Y.-C. Wang, G.T.-K. Fey, H.-Y. Li, T.-Y. Wu, T.-C. Lee, J.-K. Chang, Charge-storage performance of Li/LiFePO4 cells with additive-incorporated ionic liquid electrolytes at various temperatures, J. Power Sourc. 260 (2014) 268e275. [78] J.L. Schaefer, Y. Lu, S.S. Moganty, P. Agarwal, N. Jayaprakash, L.A. Archer, Electrolytes for high-energy lithium batteries, Appl. Nanosci. 2 (2011) 91e109. [79] J.W. Fergus, Electrolytes for high-energy lithium batteries, J. Power Sourc. 195 (2010) 4554e4569. [80] C. Tang, K. Hackenberg, Q. Fu, P.M. Ajayan, H. Ardebili, High ion conducting polymer nanocomposite electrolytes using hybrid nanofillers, Nano Lett. 12 (2012) 1152e1156. [81] K. Xu, Electrolytes and interphases in Li-ion batteries and beyond, Chem. Rev. 114 (2014) 11503e11618. [82] N.H. Idris, M.M. Rahman, J.-Z. Wang, H.-K. Liu, Honeycomb-like porous gel polymer electrolyte membrane for lithium ion batteries with enhanced safety, J. Power Sourc. 201 (2012) 294e300.

The road to potassium-ion batteries


[83] J. Zhang, B. Sun, X. Huang, S. Chen, G. Wang, Honeycomb-like porous gel polymer electrolyte membrane for lithium ion batteries with enhanced safety, Sci. Rep. 4 (2014) 6007. [84] C.L. Yang, Z.H. Li, W.J. Li, H.Y. Liu, Q.Z. Xiao, G.T. Lei, Y.H. Ding, Batwing-like polymer membrane consisting of PMMA-grafted electrospun PVdFeSiO2 nanocomposite fibers for lithium-ion batteries, J. Membr. Sci. 495 (2015) 341e350. [85] A. Ponrouch, D. Monti, A. Boschin, B. Steen, P. Johansson, M.R. Palacin, Non-aqueous electrolytes for sodium-ion batteries, J. Mater. Chem. 3 (2015) 22e42. [86] R.-S. Kuhnel, D. Reber, C. Battaglia, A high-voltage aqueous electrolyte for sodium-ion batteries, ACS Ener. Lett. 2 (2017) 2005e2006. [87] L. Qiao, X. Judez, T. Rojo, M. Armand, H. Zhang, Reviewdpolymer electrolytes for sodium batteries, J. Electrochem. Soc. 167 (2020) 070534. [88] Y. Lu, C. Gao, Q. Deng, L. Wang, J. Luo, S. Liu, N. Hu, The features and progress of electrolyte for potassium ion batteries, Small 16 (2020) 2004096. [89] K. Kubota, M. Dahbi, T. Hosaka, S. Kumakura, S. Komaba, Towards K-ion and Na-ion batteries as “beyond Li-ion, Chem. Rec. 18 (2018) 459e479. [90] T. Hosaka, K. Kubota, H. Kojima, S. Komaba, Highly concentrated electrolyte solutions for 4 V class potassium-ion batteries, Chem. Commun. 54 (2018) 8387e8390. [91] T. Hosaka, T. Matsuyama, K. Kubota, S. Yasuno, S. Komaba, Development of KPF6/KFSA binary-salt solutions for long-life and high-voltage K-ion batteries, ACS Appl. Mater. Interfaces 31 (2020) 34873e34881. [92] X. Bie, K. Kubota, T. Hosaka, K. Chihara, S. Komaba, A novel K-ion battery: hexacyanoferrate(II)/Graphite cell, J. Mater. Chem. 5 (2017) 4325e4330. [93] J. Liao, Q. Hu, Y. Yu, H. Wang, Z. Tang, Z. Wen, C. Chen, A potassium-rich iron hexacyanoferrate/dipotassium Terephthalate@Carbon nanotube composite used for Kion full-cells with an optimized electrolyte, J. Mater. Chem. 5 (2017) 19017e19024. [94] G. He, L.F. Nazar, Crystallite size control of prussian white analogues for nonaqueous potassium-ion batteries, ACS Ener. Lett. 2 (2017) 1122e1127. [95] H. Gao, L. Xue, S. Xin, J.B. Goodenough, A high-energy-density potassium battery with a polymer-gel electrolyte and a polyaniline cathode, Angew. Chem. Int. Ed. 57 (2018) 5449e5453. [96] H. Fei, Y. Liu, Y. An, X. Xu, G. Zeng, Y. Tian, L. Ci, B. Xi, S. Xiong, J. Feng, Stable Allsolid-state potassium battery operating at room temperature with a composite polymer electrolyte and a sustainable organic cathode, J. Power Sourc. 399 (2018) 294e298. [97] H. Fei, Y. Liu, Y. An, X. Xu, J. Zhang, B. Xi, S. Xiong, J. Feng, Safe all-solid-state potassium batteries with three dimensional, flexible and binder-free metal sulfide array electrode, J. Power Sourc. 433 (2019) 226697. [98] S. Komaba, T. Hasegawa, M. Dahbi, K. Kubota, Potassium intercalation into graphite to realize high-voltage/high-power potassium-ion batteries and potassium-ion capacitors, Electrochem. Commun. 60 (2015) 172e175. [99] X. Lu, M.E. Bowden, V.L. Sprenkle, J. Liu, A low cost, high energy density, and long cycle life potassium-sulfur battery for grid-scale energy storage, Adv. Mater. 27 (2015) 5915e5922. [100] H. Yuan, H. Li, T. Zhang, G. Li, T. He, F. Du, S. Feng, A K2Fe4O7 superionic conductor for all-solid-state potassium metal batteries, J. Mater. Chem. 6 (2018) 8413e8418. [101] N. Yoshinari, S. Yamashita, Y. Fukuda, Y. Nakazawa, T. Konno, Mobility of hydrated alkali metal ions in metallosupramolecular ionic crystals, Chem. Sci. 10 (2019) 587e593.


Storing Energy

[102] H. Onuma, K. Kubota, S. Muratsubaki, T. Hosaka, R. Tatara, T. Yamamoto, K. Matsumoto, T. Nohira, R. Hagiwara, H. Oji, S. Yasuno, S. Komaba, Application of ionic liquid as K-ion electrolyte of graphite//K2Mn[Fe(CN)6] cell, ACS Ener. Lett. 5 (2020) 2849e2857. [103] N. Pradeep, E. Sivasenthil, B. Janarthanan, S. Sharmila, A review of anode material for lithium ion batteries, J. Phys. 1362 (2019) 012026. [104] J. Asenbauer, T. Eisenmann, M. Kuenzel, A. Kazzazi, Z. Chen, D. Bresser, The success story of graphite as a lithium-ion anode material e fundamentals, remaining challenges, and recent developments including silicon (oxide) composites, Sust. Ener. Fuels 4 (2020) 5387e5416. [105] P. Ge, M. Fouletier, Electrochemical intercalation of sodium in graphite, Solid State Ionics 1172 (1988) 28e30. [106] B. Jache, P. Adelhelm, Use of graphite as a highly reversible electrode with superior cycle life for sodium-ion batteries by making use of co-intercalation phenomena, Angew. Chem. 126 (2014) 10333. [107] M. Dahbi, N. Yabuuchi, K. Kubota, K. Tokiwa, S. Komaba, Negative electrodes for Naion batteries, Phys. Chem. Chem. Phys. 16 (2014) 15007. [108] K. Fredenhagen, G. Cadenbach, the formation of kalium through carbon, Z. Anorg. Allg. Chem. 158 (1926) 249e265. € Schulze, Alkaligraphitverbindungen, Z. Anorg. Allg. Chem. 277 (1954) [109] W. R€udorff, E.U. 156e171. [110] M.S. Dresselhaus, G. Dresselhaus, Intercalation compounds of graphite, Adv. Phys. 30 (1981) 139e326. [111] A. Herold, Recherches sur les composes dinsertion du graphite, Bull. Soc. Chim. Fr. (1955) 999e1012. [112] P. Lagrange, D. Guerard, A. Herold, Sur La structure du compose KC8, Ann. Chim. 3 (1978) 143e159. [113] W.D. McCulloch, X. Ren, M. Yu, Z. Huang, Y. Wu, Potassium-ion oxygen battery based on a high capacity antimony anode, ACS Appl. Mater. Interfaces 7 (2015) 26158e26166. [114] M. Shimizu, R. Yatsuzuka, T. Koya, T. Yamakami, S. Arai, Tin oxides as a negative electrode material for potassium-ion batteries, ACS Appl. Ener. Mater. 1 (2018) 6865e6870. [115] X. Yu, A. Manthiram, A progress report on metalesulfur batteries, Adv. Funct. Mater. 30 (2020) 2004084. [116] X. Zhao, L. Yin, Z. Yang, G. Chen, H. Yue, D. Zhang, Z. Sun, F. Li, An alkali metaleselenium battery with a wide temperature range and low self-discharge, J. Mater. Chem. 7 (2019) 21774. [117] Z. Chen, et al., Metal-tellurium batteries: a rising energy storage system, Small 1 (2020) 2000005. [118] H.-F. Wang, Q. Xu, Materials design for rechargeable metal-air batteries, Matter 1 (2019) 565e595. [119] Z. Xie, X. Zhang, Z. Zhang, Z. Zhou, MetaleCO2 batteries on the road: CO2 from contamination gas to energy source, Adv. Mater. 29 (2017) 1605891. [120] M. Xing, Z.Z. Zhao, Y.J. Zhang, J.W. Zhao, G.L. Cui, J.H. Dai, Advances and issues in developing metal-iodine batteries, Mater. Today Ener. 18 (2020) 100534. [121] M. Bartolozzi, Development of redox flow batteries. A historical bibliography, J. Power Sourc. 27 (1989) 219e234. [122] M.L. Perry, A.Z. Weber, Advanced redox-flow batteries: a perspective, J. Electrochem. Soc. 163 (2015) A5064.

The road to potassium-ion batteries


[123] D. Aurbach, Z. Lu, A. Schechter, Y. Gofer, H. Gizbar, R. Turgeman, Y. Cohen, M. Moshkovich, E. Levi, Prototype systems for rechargeable magnesium batteries, Nature 407 (2000) 724e727. [124] Y. Orikasa, T. Masese, Y. Koyama, T. Mori, M. Hattori, K. Yamamoto, T. Okado, Z.D. Huang, T. Minato, C. Tassel, J. Kim, Y. Kobayashi, T. Abe, H. Kageyama, Y. Uchimoto, High energy density rechargeable magnesium battery using earth-abundant and non-toxic elements, Sci. Rep. 4 (2014) 5622. [125] M. Hayashi, H. Arai, H. Ohtsuka, Y. Sakurai, Electrochemical insertion/extraction of calcium ions using crystalline vanadium oxide, Electrochem. Solid State Lett. 7 (2004) A119. [126] M.E. Arroyo-de Dompablo, A. Ponrouch, P. Johansson, M.R. Palacin, Achievements, challenges, and prospects of calcium batteries, Chem. Rev. 120 (2020) 6331e6357. [127] S.K. Das, S. Mahapatra, H. Lahan, Aluminium-ion batteries: developments and challenges, J. Mater. Chem. 5 (2017) 6347e6367. [128] N. Jayaprakash, S.K. Das, L.A. Archer, The rechargeable aluminum-ion battery, Chem. Commun. 47 (2011) 12610e12612. [129] F. Gschwind, G. Rodriguez-Garcia, D.J.S. Sandbeck, A. Gross, M. Weil, M. Fichtner, N. Hörmann, Fluoride ion batteries: theoretical performance, safety, toxicity, and a combinatorial screening of new electrodes, J. Fluor. Chem. 182 (2016) 76e90. [130] M.A. Reddy, M. Fichtner, Batteries based on fluoride shuttle, J. Mater. Chem. 21 (2017) 17059e17062. [131] H. Nakano, T. Matsunaga, T. Mori, K. Nakanishi, Y. Morita, K. Ide, K. Okazaki, Y. Orikasa, T. Minato, K. Yamamoto, Z. Ogumi, Y. Uchimoto, Fluoride-ion shuttle battery with high volumetric energy density, Chem. Mater. 33 (2020) 459e466. [132] D. Zhang, K. Yamamoto, A. Ochi, Y. Wang, T. Yoshinari, K. Nakanishi, H. Nakano, H. Miki, S. Nakanishi, H. Iba, T. Uchiyama, T. Watanabe, K. Amezawa, Y. Uchimoto, Understanding the reaction mechanism and performances of 3d transition metal cathodes for all-solid-state fluoride ion batteries, J. Mater. Chem. 9 (2021) 406e412. [133] X. Zhao, S. Ren, M. Bruns, M. Fichtner, Chloride ion battery: a new member in the rechargeable battery family, J. Power Sourc. 245 (2014) 706e711. [134] X. Zhao, Z. Zhao-Karger, D. Wang, M. Fichtner, Metal oxychlorides as cathode materials for chloride ion batteries, Angew. Chem. Int. Ed. 125 (2013) 13866e13869. [135] Y. Sui, C. Liu, R.C. Masse, Z.G. Neale, M. Atif, M. AlSalhi, G. Cao, Dual-ion batteries: the emerging alternative rechargeable batteries, Ener. Storage Mater. 25 (2020) 1e32.

This page intentionally left blank

Lithiumesulfur battery: Generation 5 of battery energy storage systems


Mahdokht Shaibani and Mainak Majumder Department of Mechanical Engineering, Monash University, Melbourne, VIC, Australia

1. Introduction Since 1991, the year of commercialization of lithium-ion (Li-ion) battery, scientists have continued trying to improve the performance of these widely used batteries or seeking alternative battery technologies. In the beginning, the urge was the unforeseen usage of Li-ion in personal electronics such as mobile phones and music players. Then suddenly, the reconsidered thoughts of electrified transportation and smart grids made it substantial. Marked improvements in the performance metrics, cost-cutting, and packaging of Li-ion gradually happened, but at some point, the progress couldn’t keep up with the advancement in the transistor technology and the demand for lighter or smaller or faster batteries. This is a call for, at the very least, introducing other commercially viable batteries that can outperform Li-ion in at least one index of performance. Furthermore, for years now, there have been question marks over the supply of Li-ion battery’s key cathode ingredients, nickel and cobalt. Supply of these heavy metals is limited, prices are rising, and their mining often has high social and environmental costs [1]. Moreover, and as highlighted by the recent coronavirus pandemic, the near-monopolistic structure in the Li-ion battery supply chain is another cause for exploring alternative battery chemistries. Rather than being spread across several countries, the industrial activities around the lithium battery supply chain is concentrated in a few countries, giving them an outsized influence in the market place, and presents price and supply risks to buyers at a time when demand for lithium batteries is expanding rapidly (Fig. 14.1). The lithium-sulfur (LieS) battery, which uses extremely cheap and abundant sulfur as the positive electrode and the ultrahigh capacity lithium metal as the negative electrode, is at the forefront of competing battery technologies by offering a realizable twofold increase in specific energy, at a lower price and considerably lowered concerns around resource availability. Lightweight LieS batteries have recently made significant progress toward application, and specific energies in the range of 350e450 Wh kg1 have already been achieved by a few companies [2]. Such high specific energies will unlock several application avenues, in particular, in aviation industry where lighter batteries are essential. Storing Energy. Copyright © 2022 Elsevier Inc. All rights reserved.


Storing Energy

Figure 14.1 The implications in the lithium battery supply chain. The near-monopolistic structure in the battery supply chain gives one or few countries an outsized influence in the market place, and presents price and supply risks to buyers, at a time when demand for lithium batteries is expanding rapidly. Source: Assoc. of Mining and Exploration Companies.


Anatomy of LieS battery, challenges, and latest developments

A typical LieS battery, as we see today, benefits from two high-capacity elements as the electrode materials, lithium as the anode and sulfur as the cathode. The separator, sitting between the sulfur cathode and the lithium anode, is soaked in an electrolyte that contains lithium salts. The electrolyte facilitates the cell’s electrochemical reactions by allowing the movement of ions between the cathode and the anode [3] (Fig. 14.2).

Figure 14.2 Schematic illustration of a typical LieS battery. LieS cells typically comprise of a sulfur/carbon composite as the positive electrode and a lithium metal foil as the negative electrode, isolated from each other via a separator which is soaked in an ether-based electrolyte.

Lithiumesulfur battery: Generation 5 of battery energy storage systems


The redox reactionebased storage mechanism in the LieS system is fundamentally different from the intercalation process of lithium-ion batteries. In a Li-ion battery, Liþ ions shuttle between the positive electrode intercalation host (theoretical capacity as high as 280 mAh/g), where they are stored upon discharge; and the graphitic carbon negative electrode, where they are stored on charging to a maximum content of Li0.16C. The 1 to 6 ratio means the capacity for holding lithium is not that greatdthe theoretical capacity of graphite is 370 mAh/g. Cell voltages are in the range of 3.4 Ve3.8 V versus Li/Liþ. Theoretical specific energies are around 500 Wh/kg based on electrode materials. The theoretical capacity of the sulfur cathode, the positive electrode of the LieS system, is 1675 mAh/g, thanks to the formation of Li2S when sulfur combines with lithium in a series of complex multistep reactions. The 2 to 1 ratio clears the “holds-a-lot-of-lithium” hurdle and promises a wonderful match for the ultrahigh capacity lithium anode (3860 mAh/g), the negative electrode of the LieS system. The theoretical specific energy of the LieS system, based on electrode materials, is then determined by the theoretical capacity of sulfur (1675 mAh/g) and its potential of 2.15 V versus Li/Liþ to be around 2500 Wh/kg. The realizable specific energy of the future LieS battery is expected to fall in the range of 400e600 Wh/kg, considerably larger than that of today’s Li-ion batteries (100e265 Wh/kg). Given that the LieS battery’s solution chemistry is fundamentally different from Li-ion battery, every component of this complex system requires special attention and different sets of design rules. Despite extensive research over the past 10 years and marked progress, every single component is still problematic and needs further improvement. The sulfur cathode suffers from instability and low electrical conductivity, the lithium anode is highly reactive, and the separator is yet to fulfill the LieS system’s design criteria. While there is little doubt that LieS outperforms Li-ion in terms of specific energy, before the chemistry can be called revolutionary, all these issues need to be addressed.


Sulfur cathode: cheap and abundant with a large capacity to hold lithium but unstable

Cheap, abundant, with an ultrahigh capacity to hold lithium (1670 mAh/g), sulfur has been an ideal candidate to explore as a new cathode. While the “large capacity to store lithium” and “sustainability” boxes are addressed, the instability of sulfur cathode long intrigued researchers. In stark contrast with the energy delivery mechanism in Li-ion, the liquid electrolyte and the sulfur cathode almost act like a couple to form the highly soluble “lithium polysulfides” (LiPS) species, causing the well-known “shuttle phenomenon” [4,5] (Fig. 14.3). The constant shuttle of the corrosive LiPS from the cathode through the separator to the lithium anode not only depletes the cathode from the active material but severely damages the Solid Electrolyte Interphase (SEI) on the lithium metal anode, resulting in rapid capacity decay. Further, the high storage capacity means that the sulfur electrode swells up to almost double its size (80%) when fully charged.


Storing Energy

Figure 14.3 Schematic illustration of the parasitic polysulfide shuttle effect in a liquid electrolyte based Li/S8 cell. Long chain polysulfides originating from the high discharge and charge plateaus diffuse to the negative electrode to be further reduced to shorter chain length polysulfides that diffuse back to the positive electrode and are reoxidized to higher chain length polysulfides [4].

Such a volume change leads to a progressive loss of cohesion of particles and permanent distortion of the polymer binder and carbon matrix, contributing to the loss of capacity. Extensive research and considerable progress over the past 10 years have solved the instability issues of the sulfur cathode to a large extent. In 2010, Nazar et al. [6] carried out pioneering research in composite sulfur cathodes through the melt-diffusion strategy where sulfur was melted into the porous structure of a highly ordered, high surface area, conductive carbon. The physical confinement approach proved somewhat successful in retarding the liberation of polysulfides from the cathode to the electrolyte and led to a boost in the cycle life of the LieS test cells from only a few cycles to tens of cycles. Furthermore, the conductive nature of carbon hosts helped significantly overcome the insulating nature of sulfur and its reaction products and buffered the volume change during cycling to some extent. Ever since, several carbon materials such as mesoporous carbon [7], microporous carbon [8], carbon nanotubes [9], carbon nanofibers [10], and graphene [11] have been used as a host for sulfur (Fig. 14.4A). Nonetheless, the weak nature of the physical interaction between the polar LiPS and nonpolar carbon materials cannot retard the shuttling effect over long-term cycling. Furthermore, the use of a large fraction of porous carbons with intrinsic low tap density as the host for sulfur seriously jeopardizes the energy density of the LieS battery [12,14]. In attempts to overcome the above challenges with carbon hosts, there has been a relatively recent trend of using noncarbon polar host materials based on the strong affinity of their rich polar sites for binding with LiPS and their lower tap densities (Fig. 14.4B) [12,14]. To this end, several metal oxides [15,16], metal sulfides [17], metal-organic frameworks (MOFs) [18], metal carbides [19], metal nitrides [20], as well as polymers [21] have proved to have strong chemisorption capacity toward LiPS. The chemisorption between host materials and polysulfide can be classified into the following categories: (1) via chemical bonding (e.g., forming chemical bonding within polymers chains); (2) via polarepolar chemical interaction with

Lithiumesulfur battery: Generation 5 of battery energy storage systems


Figure 14.4 Strategies to improve the stability of the sulfur cathode. Schematic illustrations of (A) the fabrication process of the activated-carbon/sulfur composite via melt diffusion to benefit from the physical confinement of LiPS in the nanofibrous carbon [6], (B) Photographs that visualize the tap density of different host materials (top) and sulfur composites (bottom), demonstrating one key advantage of mediators over high surface area carbon materials to retard the LiPS shuttling [12], (C) an Expansion-tolerant architecture to accommodate the volume expansion of sulfur upon cycling [13].

LiPS (e.g., metal oxides/sulfides providing rich polar sites); (3) via Lewis acidebase interaction with polysulfide (e.g., forming Lewis acidebase interaction between MOFs and polysulfide) [14]. In a comprehensive article, Huang et al., [14] review the design criteria, progress, and prospects of various types of mediators for high energy density LieS batteries. Despite the great potential for mitigating the dissolution of LiPS, the use of the noncarbon hosts cannot assist with the large volume change of sulfur and its low electrical conductivity. The high-capacity sulfur cathode (1670 mAh/g) in the LieS system suffers from a severe volume change (around 78%) upon cyclingdtypically about eight times higher than that of the electrodes in Li-ion batteries. Such changes lead to a progressive loss of cohesion of particles and permanent distortion of the polymer binder and carbon matrix, contributing to loss of capacity. The adverse effect of


Storing Energy

electrode disintegration becomes dramatically more pronounced with any increase in the cathode thicknessdthe key parameter for achieving practical areal capacities (>6 mAh/cm2). The internal stress of cycling is harder to manage in the thick electrode, resulting in severe stress accumulation and impedance growth, hence rapid capacity loss [13]. As for the low electrical conductivity issue, while the addition of enough of a conductive agent assists reasonably with overcoming this in low-to-medium loading cathodes, few have sought to address it in thick cathodes. The conventional way of fabricating electrodes, which uses dissolved binder, tends to create a continuous network across the electrode bulk. This works for Li-ion battery cathodes given the small amount of binder used (less than 5%) [22] and the fact that porosity is not as critical in Li-ion electrodes. In silicon anode or sulfur cathode, however, considerably higher fractions of the binder are required to hold the electrode together (10%e30%), which via the commonly used methodology of using a dissolved binder system results in a number of critical issues [13]. First, full/partial particle coverage with the binder and carbon pore filling limit the electrolyte accessibility and sulfur utilization in the thick cathodesdadverse effects that are never as detrimental in thin cathodes. Second, during lithiation, the active material particles expand toward the existing pores between the particles, which ideally should offer space to accommodate the associated volume changes [23]. The lack of sufficient space observed in thick dense electrodes progressively causes macro/microstructural cracks and inevitably mechanical failure during long-term cycling [23]. Third, the fraction of binder absorbed within the pores of carbon [24] limits carbon’s capacity for physical confinement of the polysulfides. Among the issues mentioned above, observed in the thick cathodes fabricated via dissolved binder systems, particle isolation due to structural fragmentation is the main reason for the dramatic capacity loss in thick sulfur cathodes. Evidently, to achieve optimum electronic and electrochemical performance in thick sulfur cathodes, the design rules for their fabrication should be revisited such that the number of electrochemically available reaction sites would be maximized, and the mechanical failure of the thick cathode upon cycling is prevented. A new design, however, should be able to find its way from the laboratory to industry. For the LieS chemistry, which uses the too cheap sulfur as the active material to shine in the space beyond the Li-ion battery, the other two major electrode components, binder and conductive agent, cannot be out of the typical materials used in the fabrication of Li-ion electrodes, unless perhaps for more specialized applications. In 2020, Shaibani [13] addressed the stability challenge of high loading cathodes by introducing the expansion-tolerant (ET) architecture. Inspired by a bridging architecture first recorded in processing detergent powders in the 1970s, the Monash University researchers found an approach that places minimum amounts of a high-modulus binder between neighboring particles, leaving increased space for material expansion (Fig. 14.4C). The ET sulfur cathodes demonstrated to offer minimum interference with electrochemical reactions and ion movement, and accommodate natural volume expansion during discharging, thereby resulting in dramatically higher performance metrics. The thick cathodes demonstrated very high sulfur utilization (z85%) at sulfur

Lithiumesulfur battery: Generation 5 of battery energy storage systems


loadings as high as 13 mg/cm2, delivering unprecedentedly high areal capacities (z19 mAh/cm2). Complete cells also exhibited very high coulombic efficiency (>99%) as well, a crucial performance index that is often absent in the literature of high loading sulfur cathodes. While inducing porosity has proved very efficient in overcoming the volume change issue of sulfur even at high loadings [13,25], the strategies utilized so far to confine the LiPS in the cathode have not been efficient at practical loadings of sulfur and over long-term cycling. The polysulfides will inevitably leach from the cathode and reach the lithium anode to undesirably react with it unless they face an obstacle on their way to the anode side! This is where the separator of the LieS system can play a more significant role or, in fact, several vital functions other than merely isolating the two electrodes from each other to improve the cycling performance of the cell.


Separator: the solution chemistry of LieS demands far more functionalities for the separator

One key component of the LieS battery, which has been the subject of substantial research and development, is the membrane separator, where the focus is on either adding functionalities to commercial separators, with a view to making them more suited to LieS solution chemistry, or developing entirely new separators. In a Liion battery, the separator’s essential function is to physically and electrically separate the two electrodes while enabling free ionic flow between them. The separator is not involved directly in any cell reactions. Still, its structure and properties play an essential role in determining the battery performance, including cycle life, safety, energy density, and power density, through influencing the cell kinetics [26]. The LieS battery’s solution chemistry is much more complicated, though, and this demands that considerably more functionality be incorporated into the separator. The early state of affairs in separator research focused on interlayersdan inserted freestanding film between the existing separator and the sulfur cathode such as various activated carbons [27], CNT [28], rGO [29], graphene papers [30], nickel foam [31], as well as composite films with more functionalities [32] (Fig. 14.5A). However, in many cases, the mass and volume contribution from the interlayer and high electrolyte consumptions can adversely affect the energy density at the cell level. Polyolefin separators, such as Celgard, are the most commonly used separators in Li-ion battery but they cannot perform the additional and critical functions required for the solution chemistry of the Li-S battery. The ability to apply thin coatings or modifications on Celgard is a promising route for a high-energy density battery (Fig. 14.5B). Therefore a rational design for an advanced Li-S separator is to take the merits of the well-established Celgard as a platform for coatings and modification [34]. Examples of reported materials for coating polyolefins include reduced graphene oxide [19], porous carbons [35], Nafion [36], polymer of intrinsic nanoporosity (PIN) [37], MOFs and zeolites [33,38], and MXenes [39]. Some of these coatings such as graphene oxide and Nafion perform as a permselective membrane to reject polysulfides while enabling the free travel of Liþ ions, and


Storing Energy

Figure 14.5 Strategies to improve the cyclability of the LieS cell via interlayers and coated separators. (A) a schematic illustration of the polysulfide dissolution, adsorption, and conversion by FS-SiO2/C-CNFM interlayer during the dischargeecharge processes of LieS cells [32], (B) the design principle of Li-S battery employing LiX zeolite membrane-coated Celgard as separator [33].

some such as MXenes function by interacting/binding with polysulfides. Others like porous carbons are electrically conductive and have high surface areas to physically trap polysulfides and act as an upper current collector. Discourse on the separator design for Li-S batteries in many of the mentioned studies emphasized functionalities aimed at suppressing LiPS shuttling. Unfortunately, other key functionalities such as ionic conductivity, mass transport considerations, and the ability to act as a separator/lithium anode interface modulator, while equally as important, have been mainly overlooked except for a few recent studies [34].

Lithiumesulfur battery: Generation 5 of battery energy storage systems


Hence, similar to the sulfur cathode, designing an advanced separator is an ongoing research and development subject while it has seen marked progress. With the introduction of considerably improved designs of stable sulfur cathodes [13] and functional separators [34], the focus is now on the most challenging lithium metal anode component. Addressing the instability issues of lithium metal is predicted to ensure a rapid transition to commercial level life-spans [40]. It is well accepted that the introduction of this revolutionary energy storage technology to the marketplace without significant development on stabilizing the lithium anode is unlikely.


Lithium anode: achieving the practical use of lithium metal in batteries will be revolutionary

Having the highest theoretical capacity (3860 mAh/g) and lowest electrochemical potential make lithium metal the ultimate choice for the anode in a Li battery. “Lithium-metal” battery is the immediate predecessor to today’s “lithium-ion” battery. The concept was developed by the 2019 Nobel Prize winner, Stanley Whittingham, in the 1970s while working at the ExxonMobil research laboratories. However, it was soon found to be too unstable to attract a market. Lithium is highly reactive and tends to deposit in dendritic or mossy forms, and its relative volume change is virtually infinite due to its hostless nature, leading to formidable challenges: safety and cyclability [40]. In the late 1980s and after years of research, Moli Energy commercialized lithium metal batteries using a MoS2 cathode paired with excess lithium; this device could be cycled hundreds of times, and millions of cylindrical-type cells were sold to the market [40]. However, recurrent accidents brought safety concerns to public attention, ultimately leading to all the cells’ recall. A very limited number of other companies continued the production of lithium metal rechargeable batteries until Whittingham and other researchers removed the metallic lithium from the configuration of the battery and used materials in the lithiated form as the cathode and graphite (372 mAh/g) as the anode, allowing the lithium to intercalate and reducing its reactivity. That’s how the reliable, commercially viable Li-ion was born, and the commercialization of high-energy density lithium metal batteries was halted. The story of lithium metal anode deserves a better ending, though. Lithium metal is indispensable for the promising LieS system (as well as the not very mature Li-air system), and a relentless drive for improved longevity is powering research on lithium metal protection and novel designs of lithium-based anodes. Several strategies are being investigated to improve the performance of lithium anodes. The highlights are the use of solid-state electrolyte [41], electrolyte engineering (concentrated electrolyte [42] or use of electrolyte additives [43e46]), interface engineering (use of artificially induced SEI [47] or applying thin coatings on the lithium [48]), and more recently using stable hosts for lithium (electro deposition into graphene [49] or melt diffusion into 3D porous carbon matrix with “lithiophilic” coating [50]). The use of solid-state electrolyte, while the ultimate dream of any lithium metale based battery, has been largely unsuccessful. The main reasons for the lack of success


Storing Energy

with solid electrolytes are their low ionic conductivity, large thickness, and poor interfacial contact to lithium metal, which cause large cell overpotential during the cycling process [51]. Electrolyte engineering has been the most explored strategy to date by both academia and industry. In 2004, at Sion Power, Mikhaylik [52] patented the incorporation of 0.2e0.4M lithium nitrate (LiNO3) in the electrolyte solution [53]. Thus LiNO3 has proved to be the single most effective additive in the LieS battery chemistry that assists with suppressing the lithium dendrites formation. The initial belief was that LiNO3 reacts with the electrolyte on the surface of the lithium to form insoluble LixNOy products, which constitute an SEI layer on the anode, protecting it from further reacting with the dissolved PS. It was only later that researchers found that the presence of PS, although detrimental over long-term cycling, is actually critical for the formation of the protection layer (Fig. 14.6A) [45]. Unfortunately, while the SEI layer is easily formed, it also easily cracks as a result of the ion fluctuation and stress evolution in the cell, leaving the freshly formed lithium surface in dynamic exchange with the PS containing electrolyte. The continuous reformation of the SEI is accompanied by the continuous consumption of the electrolyte and LiNO3.

Figure 14.6 Strategies to improve the stability of the lithium metal anode. (A) schematic illustration of the possible mechanisms of SEI formation on the lithium anode via the synergetic interaction between LiNO3 and LiPS [45], (B) schematic of the dendrite penetration and serious side reactions (left) and an artificial protective layer function for Li-S batteries (right) [54].

Lithiumesulfur battery: Generation 5 of battery energy storage systems


Eventually, the cell dries up and failsdpresenting the biggest challenge of the LieS battery chemistry and calling for additional protection solutions [55]. One other popular strategy to protect the lithium metal is the use of thin-layer coatings such as a molybdenum disulfide (MOS2) [51]. alumina (Al2O3) [56], and carbon nitride nanosheets (g-C3N4) [54] (Fig. 14.6B). Choi et al. [51] demonstrated one of the largest cycle lives achieved to date in the literature of LieS battery by applying a 10 nm thick 2D MOS2 layer as a protective layer directly on the lithium metal. The MoS2 layer exhibited tight adhesion to the lithium’s surface and facilitates the uniform flow of Liþ into and out of bulk metal. Thus, a stable Li electrodeposition was claimed to be realized with dendrite formations effectively suppressed. Other phenomena that enabled the anode’s electrochemical stability, such as atomically layered structure and phase-transformation behavior of the MoS2 layer, enhanced the Liþ transport and conductivity between the electrolyte and lithium metal. In a Li-S full-cell configuration, using the MoS2-coated Li as anode and a 3D carbon nanotube-sulfur cathode, they obtained a specific capacity of w1000 mAh g1 and a coulombic efficiency of w98% for over 1200 cycles at 0.5 C [51]. Nonetheless, the low loading of the sulfur in the cathode (33%, 3.5 mg/cm2) and the unreasonably large E/S ratio (w30) [51], which is far from the required lean electrolyte condition (E/S < 5) of high specific energy LieS batteries, make it questionable whether this performance could be translated to a practical pouch cell. Moreover, similar to the SEI formation approach, the use of protective coatings such as MOS2 is unlikely to assist with the relative volume change of the lithium anode which is virtually infinite due to its hostless nature [40]. It is likely that in the large pouch cell where the lithium metal experiences almost 100% change in thickness upon cycling [57], the coating will crack, and the fresh surface of the lithium will be exposed to the LiPS containing electrolyte. Minimizing the volume change and managing the stress evolution in the cell by using stable hosts for lithium may be the one strategy to facilitate the revolutionary return of lithium. To this end, graphene has proved to be a main enabler of a number of these approaches on account of its mechanical robustness, chemical stability, strong capacity for functionalization, and excellent processability [49,58]. While some of these graphene-protected lithium anodes have demonstrated some cycling success in Liion systems (based on Liþ insertion/extraction chemistry), it has become clear that the solution chemistry of the high-capacity LieS battery (based on phasetransformation chemistry) is much more complicated. It is well accepted that in a liquid electrolyte system, the shuttling of the LiPS to the lithium anode cannot (and should not) be prevented entirely. Therefore, a graphene coating or host (or any other carbon material), which is also permeable to LiPS, won’t stop the parasitic reactions between these corrosive compounds and the highly reactive lithium metal. Thus, it can be concluded that the solution chemistry of the LieS system demands far more functionality to be incorporated into the protective coating or the host. It is also noteworthy that the successful translation of the coin cell performance to pouch cell demands the fabrication of extremely uniform large-size modified lithium metal electrodes. It is important to note that the adverse effects of any inhomogeneity in the thickness, morphology, and composition will be amplified in a large-size pouch cell and dramatically aggravate the anode degradation.


Storing Energy

LieS scientists’ known challenge is the transition from high performing lab-scale test cells (coin cells) to a real-size pouch cell batterydthe main reason why the breakthrough innovations around lithium anode struggle to find their way from laboratory discovery to the market [13,55]. In saying this, it is realized that many laboratories are severely limited in their capacity to evaluate the performance of a new component in practically sized pouch cells under realistic cycling conditions. Assembly lines for producing quality pouch cells and the equipment for accurately testing them at high currents are far from commonplace in university laboratories [5]. Examples of Li-S pouch data can be found in the following references: [13,34,60,63,64]. In stark contrast with the sulfur cathode and the separator, the progress on stabilizing the lithium metal has been incremental, and no one has managed it quite well. However, the success of practical size Li-S batteries mainly depends on the utilization of metallic Li as the negative electrode. It is essential to test new materials, designs, and concepts around lithium stabilization in pouch cell prototypes and under lean electrolyte condition to ensure early identification of the full range of present technical challenges [13].


Potential applications of lightweight LieS battery: existing, emerging, and new avenues

On account of being potentially light weight, the future LieS battery will unlock several application avenues, and over anything else in aviation industry where lighter batteries are key, provided that solutions to the low cycle life and poor power delivery can be devised. The applications discussed below require cycle life as low as 60 cycles for Maritime to as high as 1000 cycles for eBuses. The C rating requirement, the rate at which the battery is discharged or charged relative to its capacity, also covers a vast range from as slow as 0.1C for cruising to as fast as 1C to 2C for peak discharge, where 1C rate is charge or discharge in 1 h and 0.1C rate is charge or discharge in 10 h (Fig. 14.7).

Figure 14.7 Potential applications of LieS batteries. Illustrative schematic demonstrating main markets suitable for Li-S technology now and in the future [2].

Lithiumesulfur battery: Generation 5 of battery energy storage systems



Aviation: extended cruising time

The development of the high-performance LieS batteries expects to largely contribute to the commercial viability of new applications in aviation, such as uncrewed aerial vehicles, as well as the multimillion-dollar industries of drone delivery and agricultural drone spraying. For example, agricultural spray drones have numerous advantages over traditional tractor boom sprays but have limited their ability to cover large areas. The commonly used DJI T16 will have a maximum total flight time of 15 min before the battery needs to be replaced and recharged. Lightweight LieS batteries could potentially boost this industry by offering 300e400 Wh/kg. The major limitation of the current LieS batteries for drones’ potential use is their inadequate power delivery performance. While relatively low-to-moderate power is required for cruising, takeoff and hovering require high power density. A possible solution is using hybrid concepts where high-power lithium-polymer battery takes care of takeoff and the hover mode, and high specific energy Li-S battery is used for extending the flight time [2]. In a recent trial, in September 2020, LG Chem’s LieS battery-powered aerial vehicle EAV-3 flew at an altitude of 12e22 km for 7 h enduring difficult atmospheric conditions, including temperatures near 70 C and low pressure, where a general aircraft cannot fly.


Heavy electric vehicles: extended range and increased payload for remote areas

Li-ion battery has done an excellent job in powering light-duty electric vehicles such as the Tesla cars. While lightweight LieS batteries could offer a great range extension to EVs, this system’s low volumetric energy density will be a significant hurdle for the car manufacturers. Heavy-duty vehicles such as trucks and buses, on the other hand, do not have such a space limitation and could greatly benefit from the use of high specific energy and potentially cheaper LieS batteries. The use of LieS batteries could unlock the hurdles of carrying a heavy payload for extended ranges, with lowered downtime while charging, and at a competitive cost. Enabling extended range and the increased payload is quite beneficial for remote areas where charging stations’ installation is not viable. In 2020, the Cranfield University researchers [59] showed that the LieS battery pack can fulfill an electric city bus’s requirements in terms of power while achieving a considerable increase in vehicle’s rangedprovided that the battery pack could demonstrate a cycle life of above 1000.


Maritime: the deeper level exploration of the deep blue

The exploration of the deep blue has always been one of our dreams, and for obvious reasons, the investigation by the human being is restricted to only short periods and limited areas. Use of Autonomous Underwater Vehicles (AUVs), unmanned selfpropelled vehicles, for scientific research such as survey platforms to map the seafloor


Storing Energy

and observe oceanographic fields, commercial exploration, and military surveys has become widespread. The use of high specific energy LieS batteries (>400 Wh/kg) could offer a huge potential advantage over Li-ion batteries for this application. The urge for deeper level operations has been the primary driving force for exploring alternative energy storage systems for AUVs. As the depth of operation increases, the weight of the battery pressure vessel that houses the battery and its electronics also increases, restricting the battery’s weight, i.e., the amount of energy that can be carried. Moreover, the battery should withstand a harsh operating condition, the combination of high pressure (45 MPa eq. to 6000 m in depth) and low temperature (4 C). Finally, AUVs should be close to neutrally buoyant, which means the battery system also needs to be neutrally buoyant. A 2016 collaborative study between UK’s Oxis Energy and National Oceanography Center at Southampton showed that on account of having a high gravimetric energy density and a low specific gravity, as well as the capacity to operate in harsh conditions, LieS system could be an ideal choice of energy storage system for AUVs. The low to moderate power and cycle life (60e200 cycles or even less for military applications) requirements of this application suggests that the use of LieS in AUV’s maybe the closest to reality, provided that the effect of hydrostatics pressure on cycle life and safety will be precisely determined.


Aerospace: HAPS, the closest to commercial reality

High-Altitude Pseudo-Satellite (HAPS) is an uncrewed airship, plane, or balloon watching over earth from the stratosphere. Operating like satellites but from closer to earth, HAPS is the “missing link” between drones flying close to earth’s surface and satellites orbiting in space. HAPSs are positioned at an altitude of 20 km to support telecommunications or remote sensing applications and require light weight, i.e., high specific energy batteries (>400 Wh/kg) to enable high-altitude maintenance flights at mid-high latitudes. High specific energy LieS batteries, even at their current state of power and cycle life performance, could offer great potential for HAPS, where moderate cycle life (60e400 cycles) and low-to-moderate charge and discharge rates (5 mg/cm2; carbon content 103 S/cm), good moisture stability, low cost, and can be made using a simple synthesis route. Unfortunately, their narrow electrochemical window, large grainboundary impedance, and weak compatibility with metallic anode affect their application in ASS-L/SIBs [29]. Garnet-type ISSE is the most popular type of oxide ISSEs. Their general chemical formula is A3B2C3O12. The most representative garnet-type ISSE is Li7La2Zr3O12 (LLZO). In the LLZO system, the Liþ occupancy of tetrahedral position is decreased, while that of octahedral position is increased. LLZO has high ionic conductivity at room temperature (>103 S/cm), a wide electrochemical window (>6 V, vs. Liþ/Li), and good compatibility with lithium metal anode. These outstanding advantages make LLZO one of the ideal SSEs for the construction of ASS-LIBs. However, the electrical conductivity of LLZO is slightly high (w107 S/cm), which promotes lithium dendrite growth along the grain-boundary in LLZO ceramics [30].


Sulfide ISSEs

Compared with the oxygen atom, the sulfide atom has a larger radii and weaker polarity, resulting in lower force between metallic ion and sulfide atom. Therefore, the migration rate of metallic ion in sulfide ISSEs is higher than in oxide ISSEs, leading to larger ionic conductivity. Meanwhile, sulfide ISSEs have good plasticity, high thermostability, low cost, and good contact with metallic anode, which makes them promising SSEs utilized in ASS-L/SIBs. However, weak stability with moisture, narrow electrochemical window, and poor compatibility with electrode cause the limitation to its applicability [31]. There are many kinds of sulfide SSEs; the common kinds

All-solid-state batteries


are thio-LISICON (Liþ superionic conductor)-type, A11xB2xC1þxS12-type (0 < x < 1) (A ¼ Li, Na. B ¼ Ge, Sn, Si .. C¼P, Sb) (LGPS-type), and argyrodite-type. Thio-LISICON-type sulfide ISSEs are derived from a LISICON-type g-Li3PO4 solid electrolyte by replacing oxygen with sulfur, including (100x)A2SxC2S5, (100  x)A2SxBS2 and the ternary Li4xB1xCxS4 (0 < x < 1) (A ¼ Li, Na. B ¼ Ge, Sn, Si .. C¼P, Sb) electrolytes. Li7P3S11 is a representative of thioLISICON-type sulfide ISSEs, which has high Li-ion conductivity after W doped (w102 S/cm). However, the chemical stability with moisture and Li metal anode of Li7P3S11 is very poor [32]. By comparison, Li3PS4, another kind of thioLISICON-type sulfide ISSEs, have better stability toward moisture and Li metal 4 anode with PS3 S/cm [33]. 4 group in it, though its Li-ion conductivity is only w10 LGPS-type sulfide ISSEs were firstly reported by Kamaya at 2011 with a new 3D framework structure. LGPS-type sulfide ISSEs exhibit high metal-ion conductivity at room temperature (w102 S/cm), especially Li9.54Si1.74P1.44S11.7Cl0.3. Its Li-ion conductivity at room temperature is 2.5  102 S/cm, which is the highest reported to date. However, the electrochemical window and compatibility with metallic anode of LGPS-type sulfide ISSEs is still poor; it does need anion doping or surface treatment to optimize is usefulness [34]. Argyrodite-type sulfide ISSEs are a kind of hot SSEs because of high Ag-ion conductivity in Ag8GeS6. The general chemical formula of argyrodite-type sulfide ISSEs is Li6PS5X (X ¼ Cl, Br, I). The Li-ion conductivity at room temperature of Li6PS5X is w102 S/cm, which is due to the increase of site disorder and concentration of lithium vacancy as a result of substitution of S2 with Cl ions. Besides ionic conductivity, the cost and electrochemical window of Li6PS5X is also excellent, which consequently makes Li6PS5X a promising candidate among all the argyrodite-type electrolytes [35].


Other types of ISSEs

Layered-type ISSEs are commonly found in Na-ion SSEs; the representative compounds are Na- b00 -Al2O3 and Na2M2TeO6 (M ¼ Divalent elements such as Zn, Mg). Layered-type ISSEs have high Na-ion conductivity at room temperature. However, the narrow electrochemical window (w4 V) and poor compatibility of metallic Na anode are obstacles to layered-type ISSE applications in ASS-SIBs [36]. Anti-perovskite-type ISSEs was firstly reported by Li et al. Its chemical formular is A3OX or A2HOX (A ¼ Li, Na. X ¼ F, Cl, Br, I). The ionic conductivity of A3OX is w103 S/cm, much higher than of A2HOX (w106 S/cm) due to its larger ionic concentration and wider ionic migration pathway. Although anti-perovskite ISSEs have high ionic conductivity, nevertheless, it is still not widely used due to its low chemical stability with moisture and the complex process used in their preparation [37]. Halide-type ISSEs are new research hotspots in ionic conductors, which have high ionic conductivity at room temperature (w103 S/cm) and a simple preparation process. Halide-type ISSEs also have a wide electrochemical window (w6 V) and


Storing Energy

good chemical stability with moisture. The chemical formula of halide-type ISSEs is Li3MX6 (M ¼ Y, In, Sc, Ga ., X ¼ F, Cl, Br, I). However, halide-type ISSEs are unstable with metallic lithium anode. This problem could be solved with a surface treatment [38]. Complex borate-hydride-type ISSEs such as LiBH4 have attracted extensive research in recent years due to its excellent properties such as light weight, low grain boundary impedance, good ion selectivity, high stability with metallic anode, good mechanical properties, easy device integration, and low processing cost. Usual complex borate-hydride-type ISSEs contain M(BH4)m-based, M2(BH)n-based, and MCBnHnþ1-based (M ¼ Li, Na) metallic ion conductors. The metallic ionic conductivity of complex borate-hydride-type ISSEs is w103 S/cm. Unfortunately, poor stability with moisture and narrow electrochemical window (w3 V) hinder the application of complex borate-hydride-type ISSEs. Furthermore, a reduction of ionic conductivity due to phase transitions at low temperatures is also an important problem of complex borate-hydride-type ISSEs, which needs to be solved [39].


Amorphous ISSEs

Amorphous ISSEs are constructed by network forming materials (P2O5, B2O3, SiO2, P2S5, SiS2 ..) and network modified materials (Li2O, Na2S, Li2S ..). In amorphous ISSEs, the network modified materials enter the long disordered giant molecular chains formed by the interconnections of the network forming materials, which break the bridging oxygen bond, reduce the length of the giant molecular chains, and enable the metallic ions to move freely in the network structure As a result amorphous ISSEs have a metallic ion like conductivity [40]. Amorphous ISSEs are mainly divided into oxide amorphous ISSEs and sulfide amorphous ISSEs. Compared with crystalline ISSEs, amorphous ISSEs have better mechanical properties, so they have better filming property and more stable electrode/electrolyte interface in ASS-L/SIBs because amorphous ISSEs have certain degree of flexibility to adapt to the change in the volume of the electrode during the cycle. In addition, different from crystalline ISSEs, the composition of amorphous ISSEs can vary continuously. Moreover, amorphous ISSEs have better compatibility with metal anode and lower cost than crystalline ISSEs. However, low ionic conductivity restricts the application of amorphous ISSEs; therefore, it is very important to design and prepare amorphous ISSEs with high ionic conductivity and good chemical stability [41]. LiPON is a nitrogen-doped Li3PO4 oxide amorphous ISSE, which was first prepared by Bates et al. at with high electrochemical window (w5.5 V), high mechanical stability, and good stability of metallic lithium anode. Nitrogen insertion replaces some bridged oxygen bonds (dOd) and nonbridged oxygen bonds (¼O) in the structure of Li3PO4, then forms Nd(¼Nd) nitrogen double coordination bonds or Nt(dN104 S/cm) [42]. Therefore, Li-ion conductivity enhancement is an important aspect of LiPON SSE research.

All-solid-state batteries


Besides LiPON, there are many other oxide amorphous ISSEs, such as LieLaeZreO, LieLaeTieO, LieAleTiePeO, LieAleTiePeO, 90Li3BO3$7Li2SO4$3Li2CO3, and Na2O-P2O5. However, the metallic ionic conductivities of these oxide amorphous ISSEs still fail to meet the requirements of ASS-L/SIBs [43]. The radius of S2 is larger than that of O2, and the introduction of S2 instead of O2 enlarges the metallic ion transport channel in the structure. At the same time, sulfur is less electronegative and less bound to metallic ions than oxygen. These advantages accelerate the migration of metallic ions in crystal lattice of sulfide amorphous ISSEs, and thus enhance the metallic ion conductivity. The sulfide amorphous ISSEs Li2SeP2S5, Li2SeSiS2, Na2SeP2S5, and Na2SeSiS2 systems have been studied in detail. P2S5-based sulfide amorphous ISSEs have low electronic conductivity and wide electrochemical window, but their metallic ion conductivity does not meet the requirements of ASS-L/SIBs. This could be improved by doping and hybrid network structure utilization [44]. SiS2-based sulfide amorphous ISSEs have high metallic ion conductivity, low electronic conductivity, and wide electrochemical window [45]. However, their disadvantages of poor moisture stability and electrochemical stability limit the application of SiS2-based sulfide amorphous ISSEs in ASS-L/SIBs. Substituting a portion of the amorphous phase with a crystalline ceramic phase in amorphous ISSEs could combine the advantage of amorphous phase with no grain boundary resistance and of crystalline ceramic with high metallic ionic conductivity, which enlarge the metallic ion conductivity of glass-ceramics ISSEs [46]. Compared with crystalline-ceramic ISSEs, glass-ceramics ISSEs have better machinability, more compact structure, lower grain boundary resistance, and higher electrochemical stability, which are more suitable for ASS-L/SIBs.

3. Interface in ASS-L/SIBs Besides SSEs, electrode and interface are the other two important components in ASSL/SIBs. The electrode part in ASS-L/SIBs and traditional L/SIBs is uniform and has been described exhaustively in the previous chapter, so it will not be repeated here. Unlike in traditional L/SIBs, interface in ASS-L/SIBs has many unique problems such as poor contact, weak stability between SSEs and electrode, large grainboundary resistance because of SSEs introduction [47], as shown in Fig. 16.5. These problems cause ultrahigh interface impedance, low capacity, and poor cycle performance and rate performance, which are the bottlenecks restricting the development of ASS-L/SIBs.


Anodic interface in ASS-L/SIBs

Metallic lithium/sodium is considered as the best anode material for the next generation of high specific energy batteries due to its high theoretical specific capacity and low redox potential. In general, metallic anode is more suitable for ASS-L/SIBs than for traditional L/SIBs due to SSEs with good mechanical properties. However,


Storing Energy

Figure 16.5 Schematic illustration of interfacial phenomena experienced in ASS-L/SIBs [47]. Copyright 2020, American Chemical Society.

interface problems containing weak interface wettability, poor interface stability, interfacial space charge layer, and metallic dendrite shown in Fig. 16.6 greatly affect the electrochemical performance of ASS-L/SIBs [48]. Therefore, how to solve the problem of anodic interface to avoid the degradation of electrochemical performance is the focus of ASS-L/SIBs research. The solid-solid interface problem will greatly increase the interface impedance in ASS-L/SIBs, which is caused by the poor fluidity of SSEs compared with the liquid electrolytes, resulting in poor wettability between SSEs and metallic anode. In anodic interface, the optimization of SSE components can improve the wettability of the interface and make the cathode/electrolyte interface change from lithium repellency to

All-solid-state batteries


Figure 16.6 Interface problems in anodic interface of ASS-LIBs [48]. Copyright 2018, Elsevier.

attraction to lithium to reduce the interface impedance. In addition, proper treatment of SSEs or metallic anode surface can also increase the contact area between SSEs and metallic anode, which could improve interface wettability, thus accelerating metallic ion transfer rate and reducing interface impedance [49]. Metallic anode has strong reducibility, so it is easy to react with high valence ions or EO group in SSEs, resulting in poor interface stability leading to reduction of the electrochemical performance of ASS-L/SIBs [50]. Therefore, how to isolate the direct contact between the metallic anode and SSEs is the key to stabilize the anodic interface in ASS-L/SIBs. Introducing a stable SEI between SSEs and metallic anode could prevent them from having direct contact, which obstructs the reactions between the high valence ions or EO group in SSEs and metallic anode to improve the stability of anodic interface in ASS-L/SIBs effectively [51]. Moreover, adopting an alloy anode instead of metallic anode could also prevent the reactions that occur on anodic interface in ASSL/SIBs by raising the valence of element Li or Na in anode [52]. At the interface of two phases, if the electrochemical potential of two phases is different, the matching and equilibrium of the electrochemical potential at the interface of two phases requires the redistribution of boundary charge carriers. For this case the area where charge carriers are rearranged is called the space charge layer. In L/SIBs, the larger gap between the electric potential of the electrode material or its reaction product with the solid electrolyte and the solid electrolyte, the space charge layer phenomenon is more obvious. Therefore, a space charge layer exists at anodic interface of ASS-L/SIBs because there is electrochemical potential difference between SSEs and metallic anode. In general, the space charge layer on the interface of ASS-L/SIBs metallic ions preferentially moves away from SSEs, so a space charge layer with low metal ion concentration will be formed near the interface of SSEs, which causes metallic ions to move too slowly in this area thus increasing the impedance of anodic interface [53]. At present, there are very few researchers working on this space charge layer at anodic interface of ASS-L/SIBs because of the difficulty in characterizing it. A metallic dendrite will reduce the electrochemical performance, and increase the risk of short-circuit of ASS-L/SIBs. Earlier studies suggest that SSEs can inhibit the growth of metallic dendrite in ASS-L/SIBs with their high mechanical strength. However, many experimental results show that SPEs or ISSEs far from inhibiting the


Storing Energy

growth of lithium dendrite only make dendrite growth quicker. Recent studies have shown that in addition to interface wettability, electronic conductivity is also one of the important factors affecting dendrite growth. The higher the electronic conductivity the faster the growth rate of metallic dendrite [54]. Therefore, it is particularly important to construct the interface layer on anodic interface of ASS-L/SIBs with good wettability, low electronic conductivity, and excellent mechanical strength to inhibit the growth of metallic dendrites. The construction of artificial SEI on anodic interface can inhibit metallic dendrites. Artificial SEI could change the electrodeposition behavior of metal anode and reduce the surface energy of lithium metal thus decelerating the generation rate of metallic dendrites. In addition, introducing a suitable filler with low electronic conductivity, low melting point, and good ability of interstitial filling into SSEs is another effective method to prevent metallic dendrite growth along the grain boundary [55]. The filler can hinder the reduction of metallic ions when metallic ions migrate at grain boundary of SSEs, preventing deposition of metal at grain boundaries at the source, and consequently restraining metallic dendrite growth along the grain boundary of ASS-L/SIBs [56].


Cathodic interface

In ASS-L/SIBs, the cathodic part contains the active material, conductive agent, and the SSEs. The interface stability between different components of cathode is one of the bottlenecks that affect the electrochemical performance involving the capacity and cycle life of the ASS-L/SIBs. Specifically, poor contact between active materials and other components of cathodic part causes high interface impedance, thus increasing the polarization phenomenon in the charge-discharge process of ASS-L/SIBs [57]. Poor chemical and electrochemical stability lead to continuous chemical and electrochemical reactions at the solidesolid interface of the cathode, which gradually depletes metallic ions during the reaction process then resulting in capacity attenuation of ASS-L/SIBs. The poor mechanical stability of a cathodic interface results in the peeling of the cathodic interface between active materials and other component, which reduces the contact area of connection between active materials and conducting agent or collecting fluid, thus greatly increasing the interface impedance and reducing the capacity and cycle life of ASS-L/SIBs. High impedance of solidesolid cathodic interface is one of the major factors affecting the electrochemical performance of ASS-L/SIBs. One of the main reasons is poor contact performance between active materials and other components of the cathode region. Poor contact performance can reduce the contact area of active materials and SSEs or conductive agent, thus decreasing the metallic ion migration pathway and decelerating the speed of metallic ion transportation, which increases the cathodic interface impedance. This phenomenon is common in oxides SSE-based ASS-L/SIBs due to extremely poor flexibility of oxides SSEs. Mixing low melting point ionic conductor into the cathodic part to evenly distribute the ionic conductor into the space is an effective method to mitigate the effect of poor contact performance in ASS-L/SIBs [58].

All-solid-state batteries


The other main reason for high impedance in ASS-L/SIBs is the poor chemical and electrochemical stability between active materials and SSEs in cathodic part. Electrochemical inert interfacial products and space charge layers are generated by poor chemical and electrochemical stability of the interface between the active materials and SSEs in contact with each other or under charging and discharging conditions. Electrochemical inert interfacial products are generated by element interdiffusion or decomposition products of active materials and SSEs in the cathode region, which not only increase the impedance but also decrease the stability of cathodic interface resulting in poor capacity and cycle life of ASS-L/SIBs [59]. Compared with in anodic interface of ASS-L/SIBs, the phenomenon of space charge layer in cathodic interface is more obvious due to the larger electric potential difference between cathodic active materials and SSEs. Therefore, the impedance produced by a space charge layer with a low metallic ion migration rate in cathodic interface is greater than in the anodic interface of ASS-L/SIBs. Meanwhile, because of the larger potential difference between cathodic active materials and sulfide SSEs than oxide SEEs or SPEs, the influence of space charge layer in sulfide SSEs is more serious than in oxide SSEs or SPE-based ASS-L/SIBs. Surface modification of cathode active materials could intuitively avoid immediate contact between cathode active materials and SSEs, which is the most effective method to raise chemical and electrochemical stability of cathodic interfaces in ASS-L/SIBs [60]. Poor mechanical stability of cathodic interfaces can greatly increase the polarization of ASS-L/SIBs. The main reason for this phenomenon is that the lattice expansion or contraction of the cathode active materials occurs, when metallic ions are interlaced or not interlaced, so that the lattice size of the cathode active materials is changed during charging and discharging process. This volume effect could cause the interface between the cathode active materials and SSEs to break continuously during charging and discharging process, consuming the transportable metallic ions and reducing the capacity of ASS-L/SIBs [61]. At the same time, the volume change of cathode active materials during charging and discharging process will cause it to peel off with SSEs or collecting fluid, which greatly increases the impedance of the ASS-L/SIBs [62]. These phenomena caused by poor mechanical stability of the cathodic interface could be effectively restrained by surface coating on cathode active materials and applying pressure on the cathodic side of the ASS-L/SIBs.

4. Conclusion ASS-L/SIBs are the most promising next generation of secondary batteries due to their high energy density, high power density, and high safety. However, many serious problems such as high impedance, low capacity, low cycle life, and slow charge/ discharge rate hinder the application of ASS-L/SIBs. Therefore, further research is needed. The understanding of the ionic conduction mechanism in ASS-L/SIBs is one of the important scientific and fundamental problems, especially in CPEs and at the cathodic


Storing Energy

or anodic interface. These regions contain various materials with different components, making the research more challenging than in single components. Furthermore, combination of multiscale and multidimensional characterizations and simulations, particularly in situ and operando characterizations, which could elaborate the complicated interface behaviors in ASS-L/SIBs, is very important in the research of ASS-L/ SIBs. Also, high throughput technologies, big data mining, and novel battery chemistries (metal-S batteries, metal-air batteries, and batteries with no-metal ion cathodes) call for tremendous research efforts. Among the engineering and technical challenges of ASS-L/SIBs, technical challenges with metal anodes and interface engineering and in situ solidification could promote capacity and cycle stability of ASS-L/SIBs greatly, which are crucial to the electrochemical performance of ASS-L/SIBs. In addition, novel battery design, large-scale production, and cost control of key materials in ASS-L/SIBs are also important challenges related to ASS-L/SIBs deployment.

References [1] Z.H. Zhang, S.J. Chen, X.Y. Yao, P. Cui, J. Duan, W. Luo, Y.H. Huang, X.X. Xu, Enabling high-areal-capacity all-solid-state lithium-metal batteries by tri-layer electrolyte architectures, Energy Storage Mater. 24 (2020) 714e718. [2] L. Xu, S. Tang, Y. Cheng, K. Wang, J. Liang, C. Liu, Y.-C. Cao, F. Wei, L. Mai, Interfaces in solid-state lithium batteries, Joule 2 (2018) 1991e2015. [3] Z. Deng, J. Gu, Y. Li, S. Li, J. Peng, X. Li, J. Luo, Y. Huang, C. Fang, Q. Li, J. Han, Y. Huang, Y. Zhao, Ca-doped Na2Zn2TeO6 layered sodium conductor for all-solid-state sodium-ion batteries, Electrochim. Acta 298 (2019) 121e126. [4] X. Feng, D. Ren, X. He, M. Ouyang, Mitigating thermal runaway of lithium-ion batteries, Joule 4 (2020) 743e770. [5] J. Duan, X. Tang, H. Dai, Y. Yang, W. Wu, X. Wei, Y. Huang, Building safe lithium-ion batteries for electric vehicles: a review, Electrochem. Energy Rev. 3 (2020) 1e42. [6] C. Sun, J. Liu, Y. Gong, D.P. Wilkinson, J. Zhang, Recent advances in all-solid-state rechargeable lithium batteries, Nanomater. Energy 33 (2017) 363e386. [7] J. Wu, L. Yuan, W. Zhang, Z. Li, X. Xie, Y. Huang, Reducing the thickness of solid-state electrolyte membranes for high-energy lithium batteries, Energy Environ. Sci. 14 (2021) 12e36. [8] Q. Zhao, S. Stalin, C.Z. Zhao, L.A. Archer, Designing solid-state electrolytes for safe, energy-dense batteries, Nat. Rev. Mater. 5 (2020) 229e252. [9] A. Manthiram, X. Yu, S. Wang, Lithium battery chemistries enabled by solid-state electrolytes, Nat. Rev. Mater. 2 (2017) 16103. [10] Y. Zheng, Y.Z. Yao, J.H. Ou, M. Li, D. Luo, H.Z. Dou, Z.Q. Li, K. Amine, A.P. Yu, Z.W. Chen, A review of composite solid-state electrolytes for lithium batteries: fundamentals, key materials and advanced structures, Chem. Soc. Rev. 49 (2020) 8790e8839. [11] J.Y. Wu, Z.X. Rao, Z.X. Cheng, L.X. Yuan, Z. Li, Y.H. Huang, Ultrathin, flexible polymer electrolyte for cost-effective fabrication of all-solid-state lithium metal batteries, Adv. Energy Mater. 9 (2019) 1902767.

All-solid-state batteries


[12] H. Xu, P.H. Chien, J. Shi, Y. Li, N. Wu, Y. Liu, Y.Y. Hu, J.B. Goodenough, Highperformance all-solid-state batteries enabled by salt bonding to perovskite in poly(ethylene oxide), Proc. Natl. Acad. Sci. U.S.A. 116 (2019) 18815e18821. [13] H.C. Wang, L. Sheng, G. Yasin, L. Wang, H. Xu, X.M. He, Reviewing the current status and development of polymer electrolytes for solid-state lithium batteries, Energy Storage Mater. 33 (2020) 188e215. [14] Z. Xue, D. He, X. Xie, Poly(ethylene oxide)-based electrolytes for lithium-ion batteries, J. Mater. Chem. 3 (2015) 19218e19253. [15] N. Meng, F. Lian, G.L. Cui, Macromolecular design of lithium conductive polymer as electrolyte for solid-state lithium batteries, Small 24 (2020), 2005762. [16] S.L. Gao, F.Y. Sun, N. Liu, H.B. Yang, P.F. Cao, Ionic conductive polymers as artificial solid electrolyte interphase films in Li metal batteries - a review, Mater. Today 40 (2020) 140e159. [17] S.S. Zhao, Q.X. Wu, W.Q. Ma, L.S. Yang, Polyethylene oxide-based composites as solidstate polymer electrolytes for lithium metal batteries: a mini review, Front. Chem. 8 (2020) 7. [18] J.J. Zhang, J.F. Yang, T.T. Dong, M. Zhang, J.C. Chai, S.M. Dong, T.Y. Wu, X.H. Zhou, G.L. Cui, Aliphatic polycarbonate-based solid-state polymer electrolytes for advanced lithium batteries: advances and perspective, Small 14 (2018) 16. [19] T.T. Dong, J.J. Zhang, J.C. Chai, Q.M. Jia, G.L. Cui, Research progress on polycarbonatebased solid-state polymer electrolytes, Acta Polym. Sin. (2017) 906e921. [20] Q.L. Wang, H.R. Zhang, Z.L. Cui, Q. Zhou, X.H. Shangguan, S.W. Tian, X.H. Zhou, G.L. Cui, Siloxane-based polymer electrolytes for solid-state lithium batteries, Energy Storage Mater. 23 (2019) 466e490. [21] J.V. Vaghasiya, C.C. Mayorga-Martinez, M. Pumera, Flexible energy generation and storage devices: focus on key role of heterocyclic solid-state organic ionic conductors, Chem. Soc. Rev. 49 (2020) 7819e7844. [22] X.W. Yu, A. Manthiram, A review of composite polymer-ceramic electrolytes for lithium batteries, Energy Storage Mater. 34 (2021) 282e300. [23] Z.W. Cheng, T. Liu, B. Zhao, F. Shen, H.Y. Jin, X.G. Han, Recent advances in organicinorganic composite solid electrolytes for all-solid-state lithium batteries, Energy Storage Mater. 34 (2021) 388e416. [24] J.F. Wu, R. Zhang, Q.F. Fu, J.S. Zhang, X.Y. Zhou, P. Gao, C.H. Xu, J.L. Liu, X. Guo, Inorganic solid electrolytes for all-solid-state sodium batteries: fundamentals and strategies for battery optimization, Adv. Funct. Mater. (2020), 2008165. [25] Z.Z. Zhang, Y.J. Shao, B. Lotsch, Y.S. Hu, H. Li, J. Janek, L.F. Nazar, C.W. Nan, J. Maier, M. Armand, L.Q. Chen, New horizons for inorganic solid state ion conductors, Energy Environ. Sci. 11 (2018) 1945e1976. [26] C.L. Zhao, L.L. Liu, X.G. Qi, Y.X. Lu, F.X. Wu, J.M. Zhao, Y. Yu, Y.S. Hu, L.Q. Chen, Solid-state sodium batteries, Adv. Energy Mater. 8 (2018) 20. [27] A. Gurung, J. Pokharel, A. Baniya, R. Pathak, K. Chen, B.S. Lamsal, N. Ghimire, W.H. Zhang, Y. Zhou, Q.Q. Qiao, A review on strategies addressing interface incompatibilities in inorganic all-solid-state lithium batteries, Sustain. Energy Fuels 3 (2019) 3279e3309. [28] Z.L. Jian, Y.S. Hu, X.L. Ji, W. Chen, NASICON-structured materials for energy storage, Adv. Mater. 29 (2017) 1601925. [29] F. Zheng, M. Kotobuki, S.F. Song, M.O. Lai, L. Lu, Review on solid electrolytes for allsolid-state lithium-ion batteries, J. Power Sources 389 (2018) 198e213.


Storing Energy

[30] N. Zhao, W. Khokhar, Z.J. Bi, C. Shi, X.X. Guo, L.Z. Fan, C.W. Nan, Solid garnet batteries, Joule 3 (2019) 1190e1199. [31] S.J. Chen, D.J. Xie, G.Z. Liu, J.P. Mwizerwa, Q. Zhang, Y.R. Zhao, X.X. Xu, X.Y. Yao, Sulfide solid electrolytes for all-solid-state lithium batteries: structure, conductivity, stability and application, Energy Storage Mater. 14 (2018) 58e74. [32] Z. Ma, H.G. Xue, S.P. Guo, Recent achievements on sulfide-type solid electrolytes: crystal structures and electrochemical performance, J. Mater. Sci. 53 (2018) 3927e3938. [33] J.H. Wu, S.F. Liu, F.D. Han, X.Y. Yao, C.S. Wang, Lithium/sulfide all-solid-state batteries using sulfide electrolytes, Adv. Mater. (2018) 2000751. [34] J. Xu, L. Liu, N. Yao, F. Wu, H. Li, L. Chen, Liquid-involved synthesis and processing of sulfide-based solid electrolytes, electrodes, and all-solid-state batteries, Mater. Today Nano 8 (2019) 100048. [35] M. Xuan, W. Xiao, H. Xu, Y. Shen, Z. Li, S. Zhang, Z. Wang, G. Shao, Ultrafast solid-state lithium ion conductor through alloying induced lattice softening of Li6PS5Cl, J. Mater. Chem. 6 (2018) 19231e19240. [36] Y. Li, Z. Deng, J. Peng, E. Chen, Y. Yu, X. Li, J. Luo, Y. Huang, J. Zhu, C. Fang, Q. Li, J. Han, Y. Huang, A P2-type layered superionic conductor Ga-doped Na2Zn2TeO6 for allsolid-state sodium-ion batteries, Chem. European J. 24 (2018) 1057e1061. [37] Y. Li, W. Zhou, S. Xin, S. Li, J. Zhu, X. L€u, Z. Cui, Q. Jia, J. Zhou, Y. Zhao, J.B. Goodenough, Fluorine-doped antiperovskite electrolyte for all-solid-state lithium-ion batteries, Angew. Chem. 128 (2016) 10119e10122. [38] X.N. Li, J.W. Liang, J. Luo, M.N. Banis, C.H. Wang, W.H. Li, S.X. Deng, C. Yu, F.P. Zhao, Y.F. Hu, T.K. Sham, L. Zhang, S.Q. Zhao, S.G. Lu, H. Huang, R.Y. Li, K.R. Adair, X.L. Sun, Air-stable Li3InCl6 electrolyte with high voltage compatibility for all-solid-state batteries, Energy Environ. Sci. 12 (2019) 2665e2671. [39] L. Duchene, A. Remhof, H. Hagemann, C. Battaglia, Status and prospects of hydroborate electrolytes for all-solid-state batteries, Energy Storage Mater. 25 (2020) 782e794. [40] Z.A. Grady, C.J. Wilkinson, C.A. Randall, J.C. Mauro, Emerging role of non-crystalline electrolytes in solid-state battery research, Front. Energy Res. 8 (2020) 218. [41] Z. Gao, H. Sun, L. Fu, F. Ye, Y. Zhang, W. Luo, Y. Huang, Promises, challenges, and recent progress of inorganic solid-state electrolytes for all-solid-state lithium batteries, Adv. Mater. 30 (2018) 1705702. [42] J.H. Kim, C.F. Xiao, J. Han, Y.J. Kim, S. Yagi, H.S. Kim, Interface control for highperformance all-solid-state Li thin-film batteries, Ceram. Int. 46 (2020) 19960e19965. [43] T. Okumura, S. Taminato, Y. Miyazaki, M. Kitamura, T. Saito, T. Takeuchi, H. Kobayashi, LISICON-based amorphous oxide for bulk-type All-solid-state lithium-ion battery, ACS Appl. Energy Mater. 3 (2020) 3220e3229. [44] Y. Ito, A. Sakuda, T. Ohtomo, A. Hayashi, M. Tatsumisago, Preparation of Li2S-GeS2 solid electrolyte thin films using pulsed laser deposition, Solid State Ion. 236 (2013) 1e4. [45] K. Mori, K. Iwase, Y. Oba, K. Ikeda, T. Otomo, T. Fukunaga, Structural and electrochemical features of (Li2S)(x)(SiS2)(100-x) superionic glasses, Solid State Ion. 344 (2020) 115141. [46] K. Nagao, A. Sakuda, A. Hayashi, M. Tatsumisago, Formation of interfacial contact with ductile Li3BO3-based electrolytes for improving cyclability in all-solid-state batteries, J. Power Sources 424 (2019) 215e219. [47] A. Banerjee, X.F. Wang, C.C. Fang, E.A. Wu, Y.S. Meng, Interfaces and interphases in allsolid-state batteries with inorganic solid electrolytes, Chem. Rev. 120 (2020) 6878e6933.

All-solid-state batteries


[48] Y.B. Shen, Y.T. Zhang, S.J. Han, J.W. Wang, Z.Q. Peng, L.W. Chen, Unlocking the energy capabilities of lithium metal electrode with solid-state electrolytes, Joule 2 (2018) 1674e1689. [49] C.Z. Sun, Y.D. Ruan, W.P. Zha, W.W. Li, M.L. Cai, Z.Y. Wen, Recent advances in anodic interface engineering for solid-state lithium-metal batteries, Mater. Horiz. 7 (2020) 1667e1696. [50] T. Krauskopf, H. Hartmann, W.G. Zeier, J. Janek, Toward a fundamental understanding of the lithium metal anode in solid-state batteries-an electrochemo-mechanical study on the garnet-type solid electrolyte Li6.25Al0.25La3Zr3O12, ACS Appl. Mater. Interfaces 11 (2019) 14463e14477. [51] F. Hu, Y.Y. Li, Y. Wei, J.Y. Yang, P. Hu, Z.X. Rao, X. Chen, L.X. Yuan, Z. Li, Construct an ultrathin bismuth buffer for stable solid-state lithium metal batteries, ACS Appl. Mater. Interfaces 12 (2020) 12793e12800. [52] T. Krauskopf, B. Mogwitz, C. Rosenbach, W.C. Zeier, J. Janek, Diffusion limitation of lithium metal and Li-Mg alloy anodes on LLZO type solid electrolytes as a function of temperature and pressure, Adv. Energy Mater. 9 (2019) 1902568. [53] J. Maier, Ionic conduction in space charge regions, Prog. Solid State Chem. 23 (1995) 171e263. [54] F.D. Han, A.S. Westover, J. Yue, X.L. Fan, F. Wang, M.F. Chi, D.N. Leonard, N. Dudney, H. Wang, C.S. Wang, High electronic conductivity as the origin of lithium dendrite formation within solid electrolytes, Nat. Energy 4 (2019) 187e196. [55] X. Han, Y. Gong, K. Fu, X. He, G.T. Hitz, J. Dai, A. Pearse, B. Liu, H. Wang, G. Rubloff, Y. Mo, V. Thangadurai, E.D. Wachsman, L. Hu, Negating interfacial impedance in garnetbased solid-state Li metal batteries, Nat. Mater. 16 (2016) 572. [56] W. Lee, C.K. Lyon, J.H. Seo, R. Lopez-Hallman, Y.J. Leng, C.Y. Wang, M.A. Hickner, C.A. Randall, E.D. Gomez, Ceramic-salt composite electrolytes from cold sintering, Adv. Funct. Mater. 29 (2019). [57] Y. Xiao, Y. Wang, S.-H. Bo, J.C. Kim, L.J. Miara, G. Ceder, Understanding interface stability in solid-state batteries, Nat. Rev. Mater. 5 (2020) 105e126. [58] F.D. Han, J. Yue, C. Chen, N. Zhao, X.L. Fan, Z.H. Ma, T. Gao, F. Wang, X.X. Guo, C.S. Wang, Interphase engineering enabled all-ceramic lithium battery, Joule 2 (2018) 497e508. [59] C.H. Wang, J.W. Liang, M. Jiang, X.N. Li, S. Mukherjee, K. Adair, M. Zheng, Y. Zhao, F.P. Zhao, S.M. Zhang, R.Y. Li, H. Huang, S.Q. Zhao, L. Zhang, S.G. Lu, C.V. Singh, X.L. Sun, Interface-assisted in-situ growth of halide electrolytes eliminating interfacial challenges of all-inorganic solid-state batteries, Nanomater. Energy 76 (2020) 105015. [60] Z. Cheng, M. Liu, S. Ganapathy, C. Li, Z.L. Li, X.Y. Zhang, P. He, H.S. Zhou, M. Wagemaker, Revealing the impact of space-charge layers on the Li-ion transport in allsolid-state batteries, Joule 4 (2020) 1311e1323. [61] R. Koerver, I. Ayg€un, T. Leichtweiß, C. Dietrich, W. Zhang, J.O. Binder, P. Hartmann, W.G. Zeier, J. Janek, Capacity fade in solid-state batteries: interphase formation and chemomechanical processes in nickel-rich layered oxide cathodes and lithium thiophosphate solid electrolytes, Chem. Mater. 29 (2017) 5574e5582. [62] G. Bucci, T. Swamy, Y.M. Chiang, W.C. Carter, Modeling of internal mechanical failure of all-solidstate batteries during electrochemical cycling, and implications for battery design, J. Mater. Chem. 5 (2017) 19422e19430.

This page intentionally left blank

Vanadium redox flow batteries


Christian Doetsch and Jens Burfeind Fraunhofer Institute for Environmental, Safety, and Energy Technology UMSICHT, Oberhausen, Germany

1. Introduction and historic development The redox flow battery was first developed in 1971 by Ashimura and Miyake in Japan [1]. In 1973 the National Aeronautics and Space Administration (NASA) founded the Lewis Research Center at Cleveland, Ohio (USA) with the object of researching electrically rechargeable redox flow cells. The Exxon Company (USA), Giner Ind. (USA), and Gel Inc. (USA) were awarded contracts to develop a hybrid redox flow battery [2] and in the following 6 years research was done on different redox couples, membrane development, electrodes, etc. [3] Iron chloride (FeCl3) and titanium chloride (TiCl2) were proposed as electrolytes. In 1975 a patent (US Patent 3996064) was filed by Lawrence H. Thaller. The description of present-day systems is identical to that described in the original patent which involved a two-tank system and a cell with a separator and two graphite electrodes. With the present-day application of flow batteries to store sustainable electricity, they appear to have foreseen future problems: “Because of the energy crisis . and due to economic factors within the electric utility industry, there is a need for storing bulk quantities of electrical power [.] be produced intermittently [.] by devices such as wind-driven generators, solar cells or the like” (US Patent 3996064). Later Thaller [4] replaced titanium with chromium. In 1978 an all-vanadium system was proposed for the first time. In this special case, both electrolytes consist of the same metal, but at different stages of oxidation. Therefore, there is no problem with cross-contamination through the separating membrane. Maria Skyllas-Kazakos at the University of New South Wales (Australia) further developed this technology in the 1980s. This vanadium-based redox flow battery is today the most developed and popular flow battery and its sales exceed those of other flow batteries. Also, in the 1980s the Japanese company, Sumitomo, was very active in filing patents and developing new membranes and electrolytes. This activity stopped at the end of the 1990s and was restarted 5 years ago. The Canadian company, VRB Power (CA), was another very active company from 2000 to 2008. At the end of 2008 it filed for bankruptcy and was bought out by Prudent Energy VRB Systems. In Austria, Martha Schreiber developed electrochemical storage systems, with special emphasis on redox flow; she founded the company Cellstrom GmbH (Austria), which today is part of CellCube (Toronto, CA) after a colorful firm history as being part of Younicos (DE), later Gildemeister (DE) and Mori Seiki (JP). This is an example of the fact that even if technical problems have been solved and products are developed, the “valley of death” has not yet been completely overcome.

Storing Energy. Copyright © 2022 Elsevier Inc. All rights reserved.


Storing Energy

The development of scientific interest in redox flow technology can be seen in Fig. 17.1. The 1970s witnessed the start of a small number of publications each year (5e12). Since 2011 the number of publications has risen significantly and has almost doubled in number every 7 years. Currently, there are more than 1200 publications per year. The current strong interest in flow batteries can also be seen in the overview of patents involving redox flow batteries (RFBs): over the past decades (2000e20) the number of patents filed have increased rapidly with a peak in 2017 (Fig. 17.2). The inventors responsible for these patents come mainly from China, Japan, South Korea, United States, and Europe (see Fig. 17.3). In the past 5 years China has overtaken Japan, South Korea, and the United States. Today, the companies working with RFBs include large multiindustry companies such as Sumitomo (Japan) and many specialized companies like Cellcube Energy Storage Inc. (Canada), Prudent Energy VRB Systems (USA and Canada), UET-Uni Energy Technologies (USA) in cooperation with Dalian Rongke Power (China), Volterion (DE), Avalon (CA), and a few new or startup companies such as Pinflow (CZ) and others.


The function of the VRFB

The electrochemical redox flow cell consists of two half-cells which are separated by a separator which can be an anionic exchange membrane, a cationic exchange membrane, or a porous membrane. The liquid electrolyte stores electrical energy in the form of chemical ions which are soluble in liquid aqueous or nonaqueous electrolytes. The electrolytes of the negative half-cell (anolyte) and the positive half-cell (catholyte) are each circulated by a pump in separate circuits. Both electrolyte recirculation circuits are separated by a separator. The function of the separator is to prevent electrical short circuits, prevent cross-mixing of the electrolyte, and to ensure the ion exchange across the separator balances the electrical charge of the anolyte and catholyte (Fig. 17.4).

Figure 17.1 Number of publications (research with Scopus; title, key words, and abstract “redox* flow” January 2021).

Vanadium redox flow batteries


Figure 17.2 Development of filed patents regarding redox flow in the world [research with software patbase; (redox* flow) in claims, title, abstract; January 2021].

Figure 17.3 Filed patents regarding redox flow in different countries [research with software Patbase; (redox* flow) in claims, title, abstract; January 2021]. Others are: CO, 4; Si, 4; NL. 3; AR, 2; FI, 2; LT, 2; MA, 2; NO, 2; VN, 2; MK, 2; UA, 2;1; SM, 1; RS, 1; RO, 1; CY, 1.


Storing Energy

Figure 17.4 Working principle of vanadium redox flow batteries.

The ions that are exchanged depend on the kind of redox flow battery; the most common types are cationic exchange membranes such as NAFION. These perfluorinated and sulfonated membranes have been used for decades and are very stable against chemical attack and oxidative corrosion caused by high potentials. These membranes are mostly used in acid electrolyte systems for vanadium redox flow cells or iron chromium cells. Charge balancing is easily done by the transport of hydrated protons (hydronium-ion) through the membrane. These membranes are available worldwide through several commercial suppliers and because they are fluorinated membranes the price is quite high. A second type of membrane is the anionic exchange membrane. In this case the counter-ion of the active species is responsible for the charge balance. These anionic-type membranes are in general cheaper, but chemical stability has to be carefully checked. A third class of membranes is the so-called micro- or nanofiltration membranes. The working principle of these membranes is quite different because the function is based on ion exclusion. Small ions such as hydronium (H3Oþ) or sulfate (SO2) ions can freely cross the membrane for charge balancing, while much larger ions such as vanadyl ions are too bulky to cross the membrane so cross-mixing of the electrolyte is prevented. A number of variations and modifications to such membranes have been made during the past few years. One notable development involves a class of nonfluorinated membranes known as SPEEK membranesdsulfonated poly(tetramethyldiphenyl) etheretherketonedwhich show promising chemical stability and proton conductivity.

Vanadium redox flow batteries


Some attempts have been made to implement inorganic fillers such as SiO2 and ZrO2 into the membrane, with the purpose of preventing vanadium ions from crossover by inducing a charge in the membrane which excludes positive charge vanadium ions. The advantage of ionic conducting membranes is that they are effective at preventing the crossover of electrolytes, and hence make for an effective and efficient cell. Microporous ion exclusion membranes are an order of magnitude cheaper than cationic ion conducting membranes, but because of their relatively poor efficiency they have not been produced commercially. The membrane separates only the anolyte and catholyte and allows for charge balance. The chemical oxidation and reduction that result from electron transfer take place at the electrode. In most cases the reaction takes place at a graphitic carbon felt. In the VRFB no extra catalyst is necessary, but activation of the graphite felt does help to accelerate the reaction. This is done by adding carboxylic groups to the graphite surface using either heat treatment or chemical and electrochemical oxidation. Graphite is the material of choice because of its chemical resistance and low cost. The graphite felt must have a high surface area and good electrical conductivity. High surface area is necessary to provide enough reaction sites. Due to its operation in flowthrough mode the graphite felt must have a very open structure (95% void volume) with a thickness of 3e5 mm. The open structure of the felt is necessary to achieve a lowpressure drop between the electrodes. On the other hand, the open structure leads to a relatively high inner resistance of the overall stack in comparison with cells which are constructed in flow-by mode as seen in fuel cells and electrolyzers. As shown in Fig. 17.5 (left and right), respectively, compound-based bipolar plates separate the anolyte and the catholyte from each other. This arrangement gives the opportunity for electrical connection in series. The bipolar plates are often made of a compound-based material, which could be processed by hot pressing or injection molding. Hot pressing is often done with duroplastic resins; injection molding uses thermoplastic material such as polyethylene and polypropylene which are also lowcost materials. To achieve sufficient electrical conductivity a very high amount of

Figure 17.5 Left, schematic view of a single redox flow cell; right, schematic view of a redox flow stack.


Storing Energy

graphite and carbon black up to 85% by mass is necessary. This, however, limits the ability to process and limits mechanical stability. In a flowthrough design no flow field is necessary. The contact resistance between graphite felt and compound-based bipolar plate dominates the overall inner resistance of the stack; as a result, much work has been done to lower contact resistance by gluing the graphite felt to the bipolar plates with conducting glue. The role sealing in a stack is often underestimated and several gaskets are usually needed to properly seal the stack. Gaskets on each side of the sepa-rator are used to ensure a good seal toward the outside; furthermore, additional gaskets are needed to seal the bipolar half-plates and internal manifolds. The thickness of the graphite felt, frames, and gaskets must be properly adjusted to ensure good compression of the felt and sufficient strength given to the gaskets. The gaskets are made from elastomeric materials such as ethylene propylene diene monomer (EPDM) rubber or fluoroelastomeric materials. Single cells are stacked together to achieve higher voltages. The stack itself is compressed by strong end-plates and tension rods. One stack producer (Volterion, DE) works without gaskets and without tension rods. In this case all components of the polyethylene-based materials of the stack are welded together for a complete tight and force-fit connection (see Fig. 17.6). The vanadium electrolyte consists of vanadium salts which are dissolved in aqueous sulfuric acid. The liquid electrolyte corresponds to the active mass in a conventional battery. The amount of liquid electrolyte which is stored in tanks determines the capacity of the RFB. The big advantage of RFBs is that power and capacity can be scaled independently.

Figure 17.6 Innovative redox flow stack without gaskets and tension rods. Abandoning seals, complex end plates and screws also leads to a much higher power density (0.5 kW/L). Specs: 310  230  270 mm, 10 kg, and 1.8/2.5 kW power (nom./peak). Photo from Volterion.

Vanadium redox flow batteries


To operate an RFB additional pumps, piping, valves, and storage tanks are necessary. All such equipment must be able to withstand the harsh conditions of sulfuric acid and strong oxidizing power of vanadium ions in the valence state of 5. To ensure an efficient system, each vanadium redox flow system has a simple battery management program, which controls the flow rate of pumps with respect to load requirements and state of charge. The nominal charge voltage of each single cell is usually limited to 1.6 V to avoid the potential at which water is decomposed into oxygen and hydrogen. This is known as the oxygen evolution reaction (OER) which causes carbon corrosion that rapidly destroys the graphite electrodes [5e12]. The end of discharge voltage is kept to 0.9e1.0 V to reach a reasonable efficiency of the cell. It is worth mentioning that an electrochemical cell could be discharged down to 0 V without destroying the cell.

3. Electrolytes of VRFB A vanadium-based electrolyte is widely used in flow batteries. This is due to the simplicity and stability of the electrolyte system in the aqueous phase. In an aqueous solution, four different but stable valence states of vanadium exists (V2þ, V3þ, V4þ, and V5þ). In anolyte vanadium (þ2 and þ 3) ions exist as V2þ, V3, while the þ4 and þ5 valence states of vanadium exist only as oxocomplexes (VO2þ, VOþ). By changing the valence states of vanadium species, energy could be stored electrochemically. These basic redox reactions are: V2þ #V3þ þ e

Eo ¼ 0.255 V

 þ VO2þ þ H2 O # VOþ 2 þ e þ 2H

(17.1) Eo ¼ þ1.004 V


The oxidation of V2þ releases one electron and V3þ is formed. This creates a standard potential of 0.255 V. The oxidation of V4þ to V5þ by simultaneous splitting of a water molecule releases a proton and one oxygen atom which form the oxo complex. This delivers a standard potential of þ1.004 V [13]. The overall standard potential for the reaction is 1.259 V. In an aqueous electrolyte the vanadium salts in all the four different valence statesdVþ2, Vþ3, Vþ4, and Vþ5dmust be soluble in concentrations which should be as high as possible. The more vanadium salts held in a stable solution without precipitation, the higher the volumetric energy density of the electrolyte. The vanadium salt in valence state 5 has the lowest solubility. The following equation describes the reaction equilibrium between solid vanadium pentoxide and vanadium þ5 in solution [14]: 1 1 V2 O5 ðsÞ þ Hþ #VOþ 2 þ H2 O 2 2



Storing Energy

The higher the proton concentration (acid concentration), the more the equilibrium is shifted to the right side (principle of Le Chatelier) and the more the Vþ5 vanadium (in the form of VOþ) can be kept in the solution. Furthermore, the high proton concentration of the electrolyte results in high electrolyte conductivity which in turn leads to good cell performance. The sulfuric acid is usually at a concentration of between 2 mol/L and 6 mol/L [15]. Very recently mixed electrolytes of sulfuric and hydrochloric acids have been used as electrolytes. The addition of hydrochloric acid and therefore chlorine ions allows for very high vanadium concentrations in the solution at even higher temperatures. The higher stability of the electrolyte is caused by the formation of a chloroeoxo complex, and this stable complex prevents the condensation reaction and precipitation of vanadium pentoxide [16]. This mixed electrolyte has the advantage of higher energy density and temperature stability but, on the other hand, can result in possible release of poisonous chlorine gas at the positive electrode during the charge process. Furthermore, the presence of chlorine could compromise material stability in the cell.


VRFB versus other battery types

VRFBs like all other flow batteries are in competition with batteries such as lead-acid batteries. This competition is driven by techno economic needs for different applications. These storage device needs could be for: an uninterruptible power supply (UPS) for which total efficiency and cycle life time are not crucial, but capital expenditure (CAPEX), life time, and response time are the most important criteria home applications coupled with photovoltaics for which efficiency and cycle time are very important issues grid operators for which CAPEX, cycle life time, and efficiency are important criteria.

In Table 17.1 a list of the main specifications of major electrochemical storage systems is shown. Table 17.1 Cycle life time, energy efficiency and current development status of major electrochemical storage systems.

Battery type

Cycle life time (cycles)

Energy efficiency/(%)




Lead-acid NiMh Li-ion NaS (high temperature) Zn-air

300e2000 1400 500e15,000 3000e7000

75e90 70 90e95 70e85

Under development; near to market launch Commercial Commercial Commercial Commercial



Under development


Vanadium redox flow batteries


So far, we have discussed only a few applications and their specifications. To ensure the best battery type for a particular application, all specifications must be considered. In the following section the VRFB is compared with other batteries for particular applications. From an energy density (gravimetrical and volumetrical) point of view, the VRFB is low compared with zinc-air and lead-acid batteries. As a result, the VRFB is more suitable for stationary applications. Mobile applications of VRFB could only be possible in niche areas such as on ferries. Vanadium flow batteries have the highest cycle life time of all presently available batteries including lithium-ion batteries. One big advantage of VRFBs is that they have a long life, because the liquid electrolyte does not degenerate to any great extent and can be used for decades without replacement. Electrodes made of graphite felt are also very stable, and furthermore membrane failure occurrences are extremely rare. Another advantage of VRFBs is that self-discharging is extremely low due to the fact that self-discharging could only occur in the reaction chamber (cell) and not in separated storage tanks. Two unique characteristics of flow batteries are: capacity is dependent on the size of storage tanks; and power is dependent on stack size. As a result, any existing flow battery system can be improved and extended by adding additional electrolyte tanks. These characteristics lead to flow batteries being used for stationary applications (low energy density) with high cycling rates (up to 365 full cycles per year) with a long-lasting life time and the capacity for long storage times. In short, flow batteries have high storage capacities in relation to power. Since the costs for energy storage always depend on the specific application, here is an example for the levelized cost of storage ($/MWhstored) of a large-scale application, called “Wholesale” large-scale energy storage system designed to replace peaking gas turbine facilities; brought online quickly to meet rapidly increasing demand for power at peak; can be quickly taken offline as power demand diminishes [30]): 100 MW, 400 MWh, 20 years running time, 350 cycles/year, 100% depth of discharge. In this case RFB systems are about 20%e30% more expensive than lithium-ion battery systems. If the flexible scale-up is needed, e.g., another application called “Transmission and Distribution” with a ratio of 6 between storage capacity (MWh) and power output (MW) the cost figures per $/MWh are comparable to lithium-ion systems [30].

5. Application of VRFB The most promising applications for redox flow batteries are: 1. Time shift applications a. Economics-driven systems which charge the storage plant with inexpensive electric energy purchased from the grid during low-price periods and discharge the electricity back to the grid during periods of high price. b. Technology-driven systems, which charge the storage plant with surplus energy from the grid during low demand periods and discharge the electricity back to the grid (or island grid) during periods of high demand.


Storing Energy

2. Renewable integration systems which assist in wind- and solar-generation integration by reducing output volatility and variability, reducing congestion problems, providing backup for unexpected generation shortfalls, and reducing minimum load violation. 3. Network investment deferral systems, which postpone or avoid the need to upgrade the transmission and/or distribution infrastructure.

With increasing market share the size of each stack will increase as demand for larger units grows. Most present applications involve large-scale systems because VRFBs are getting cheaper as a result of upscaling than other systems. Moreover, they have in most cases a high capacity so that the time for charging or discharging takes a few hours (3e10 h). For VRFBs this power-to-capacity ratio is therefore between 1:3 and 1:10. A typical example of this: Japan, Hokkaido: 17 MW/51 MWh VRFB connected to a wind farm (renewable shifting/T&D deferral). Sumitomo Electric is going to install a 17 MW/51 MWh allvanadium redox flow battery system for the distribution and transmission system operator Hokkaido Electric Power on the island of Hokkaido from 2020 to 2022. The flow battery is going to be connected to a local wind farm and will be capable of storing energy for 3 h. The overarching aim of the project is the integration of larger shares of renewable energy by renewable shifting, to avoid or at least postpone grid extensions [32]. A map of current installations in Western Europe is given in Fig. 17.7. Worldwide there are more than 200 RFB projects realized mainly in Europe, Southeast Asia, and North America [31].

Figure 17.7 Redox flow battery projects worldwide. Here x kW, y refers to power output in kW and loading capacity in hours, respectively. Data from IFBF, assessed 2021-01-10.

Vanadium redox flow batteries


Figure 17.8 Stable price for vanadium pentoxide (98% min) Europe USD/lb based on data from Ref. [29].

While market increase and cost figures for RFB are going down due to mass production, scale-up and technological progress the price for vanadium is an additional key to success or failure. While the vanadium pentoxide price was stable for decades, there was a dramatic cost spike in 2018/2019dsee Fig. 17.8. Nevertheless, in 2020 the price returns to the overall average and the RFB market “recovers.” The development of the RFB technology over the past years was summarized by Jens Noack et al. [28]: “VRFBs have progressed beyond the prototype and demonstration stage in recent years. Due to the extremely high vanadium price in 2018, commercialization efforts of VRFBs were severely curbed but are currently experiencing a renewed upswing. Today, more and more systems in the megawatt hour range are being installed worldwide, as are smaller container-based VRFBs. [.] Alongside lithium-ion batteries, they are now one of the most important stationary energy storage technologies, especially for grids with renewable energies and with average storage times of a few hours. The costs for VRFBs have fallen significantly in recent years and a further reduction in costs with a simultaneous increase in service life can be expected in the next few years as alternative production technologies are used and economies of scale gain influence.”

6. Recycling, environment, safety, and availability One of the most important advantages of RFBs is that the electrolyte could be regenerated while operating and recycled after the lifetime of storage systems. While the stack consists mostly of uncritical material like graphite and plastic, which does not need recycling, the electrolyte (anolyte and catholyte) consists of vanadium and sulfuric acid and does need recycling. Vanadium is a high-priced material that can be almost 100% reclaimed, as can sulfuric acid [17]; from the economic point of view vanadium is the important component. Reclaimed vanadium can be used to produce new electrolyte for RFBs or for other purposes (steel industry).


Storing Energy

Figure 17.9 Criticality and vulnerability of some metals in Germany. Data from L. Erdmann, S. Behrend, M. Feil, Kritische Rohstoffe f€ ur Deutschland. Berlin, Germany: Insti- tut f€ur Zukunftsstudien und Technologiebewertung (IZT) and Adelphi (2011).

From the environmental protection point of view, only VRFB electrolyte has to be taken into account. This is because sulfuric acid is corrosive and vanadium is a heavy metal. As a result, double-wall storage vessels/catch basins and splash guards have to be provided for the whole system. In this respect the electrolytes of VRFBs can be compared with the electrolytes of lead-acid batteries. From a safety point of view, VRFBs are safer than many other types of batteries and there is almost no risk of fire because of the larger amount of water present in the system. Furthermore, in case of a short circuit [18] or mixing of anolyte and catholyte (comparable with the “nail test” for lithium-ion batteries) there is only a minor exothermic reaction with less than a 1 C temperature rise. In light of being a crucial component the availability of vanadium has often been discussed. Vanadium is an important byproduct of a number of mining operations and is used almost exclusively in ferrous and nonferrous alloys. Vanadium consumption in the iron and steel industry represents about 85% of the vanadium-bearing products produced worldwide [19]. The global supply of vanadium originates from primary sources such as ore feedstock, concentrates, metallurgical slags, and petroleum residues. The main supplier countries are South Africa, China, Russia, but supplies are also exported from Canada, United States, Argentina, etc. From the German/European point of view, vanadium has been assessed as a medium critical metal by Erdmann et al. [20] (Fig. 17.9).


Other flow batteries

During the past 40 years nearly every chemically possible electrolyte combination has been evaluated as a suitable electrolyte for flow batteries. Due to limitations in chemical stability, energy density, poisoning, or radioactivity, only a few electrolyte systems can be considered for practical flow battery application. These applications include hybrid flow batteries; in these batteries the anode is in a fully charged state and is usually a solid metal which dissolves during discharge to form the corresponding salt. Frequently used anode materials are zinc, iron, and possibly copper. In the following section the most important variations are briefly described, and in Table 17.2 the most investigated flow batteries are listed.

Vanadium redox flow batteries


Table 17.2 Most investigated flow batteries. System type/active material

Cell voltage/V

Redox flow Vanadium VRB Vanadium-bromine Polysulfide-bromine PSB Iron-chromium Hydrogen-bromine Hybrid flow

1.4 1.3 1.5 1.2 1.1

Zinc-bromine Zinc-cerium

1.8 2.4




Anode/ cathode V2þ/VOþ2 V2þ/1/2Br2 2S22 e/Br2 Fe2þ/Cr3þ H2/Br2 Anode/ cathode Zn/Br2 Zn/2Ce4þ

Anode/cathode H2SO4/H2SO4 VCl3/NaBr (HCl) NaS2/NaBr (NaOH) HCl/HCl NaBr (NaOH) Anode/cathode Zn/ZnBr2 (NaOH) CH3SO3H/ CH3SO3H

Iron-chromium flow battery

One of the first flow battery electrolyte chemistries studied was the iron-chromium flow battery (ICB). It has been extensively studied by NASA (USA) and Mitsui (Japan). The iron-chromium battery is a real RFB with energy stored in Fe2þ/Fe3þ and Cr2þ/Cr3þ couples, which are dissolved in hydrochloric acid. During discharge Fe2þ is oxidized to Fe3þ and simultaneously Cr3þ is reduced to Cr2þ. To keep the overall charge in balance, a proton is exchanged through the separator which separates the anolyte and catholyte. In the original iron-chromium system, cross-mixing of the electrolyte was a serious problem. Over time the iron and chromium ions diffuse through the membrane, so an irreversible capacity loss occurs. To avoid this cross-mixing effect, expensive ion exchange membranes were used. Modern iron-chromium batteries work with a mixed electrolyte, which uses iron and chromium on both sides. This allows the use of inexpensive porous separators. The optimal working temperature of the iron-chromium flow battery is 40e60 C, which is quite high for a battery and thus makes this battery suitable for hot climates. The electrolyte is cheap and nonflammable. One disadvantage is the possibility of hydrogen evolution, which causes a loss in efficiency.


Polysulfide bromine flow battery

The polysulfide bromine (PSB) redox flow battery is a well-investigated battery type. The great advantages of this type of battery are the very cheap and abundant electrolyte and the high voltage of 1.5 V. The electrolyte is an alkaline solution of sodium polysulfide and sodium bromide as anolyte and catholyte, respectively. During charging and discharging, sodium ions exchange through an ionic exchange membrane to


Storing Energy

keep the charge in balance. The efficiency of the battery is about 75%. As in every bromine-based electrolyte, the bromine must be dissolved in the electrolyte with the aid of a complexing agent. The reactions are: Positive electrode: NaBr þ 2Naþ þ 2e / 3NaBr (discharge) Negative electrode: 2N2a2 S / Na2 S4 þ 2Naþ þ 2e (discharge) The active species are highly soluble in aqueous electrolyte and therefore the electrolyte has a relative high energy density at low cost. In 2002 Regenesys built a 15 MW, 120 MW h PSB flow battery system at Little Barford in the United Kingdom, but the project was never fully commissioned. The business was owned by RWE Power and it left the project before final commissioning.


All-organic redox flow battery

Very recently a new type of flow battery has been under development which involves organic molecules that are soluble in aqueous phase and could easily be oxidized and reduced. A benefit of these organic flow batteries is that the electrolyte can be very cheap and not based on limited resources like vanadium. A promising candidate is the sulfonated anthraquinone redox couple. In 2013 researchers suggested the use of 9,10anthraquinone-2,7-disulfonic acid (AQDS), a quinone, as an organic redox molecule in metal-free flow batteries [21]. AQDS easily undergoes rapid and reversible twoelectron two-proton reduction at a carbon electrode in sulfuric acid. Each of the carbon-based molecules holds two functional groups which can be oxidized and reduced. This is a promising research area, as it has the potential to offer a low-cost flow battery electrolyte. By modifying the chemical structure of the basic anthraquinone molecule, solubility in the aqueous phase could be increased. Moreover, the potential could be shifted even higher.

7.4 7.4.1

Hybrid flow batteries Zinc-bromine flow battery

The zinc-bromine flow battery is a so-called hybrid flow battery because only the catholyte is a liquid and the anode is plated zinc. The zinc-bromine flow battery was developed by Exxon in the early 1970s. The zinc is plated during the charge process. The electrochemical cell is also constructed as a stack. Storage capacity is determined by the size and thickness of the plated zinc plate and of the catholyte storage reservoir, and as a result the power rating and capacity correspond to each other. The catholyte contains an organic complexing agent to keep the generated bromine in solution during the charging process. A microporous separator is used in most cases. During charging, zinc is plated on a carbon composite plate. The morphology of the plated zinc is strongly related to current density, temperature, and flow velocity.

Vanadium redox flow batteries


At high current densities, zinc tends to dendritic growth which might cause short circuits through the separator. During charging, bromine is generated. Bromine is highly oxidative and is a poison. Its solubility in water is limited, so to increase the solubility an organic complexing amine is added; this interacts with the bromine to keep it in solution. The organic dense phase behaves like oil and forms a separate phase; this has to be considered for a system layout. An important issue is the toxicity of bromine. Its high oxidative power necessitates the use of chemically resistant parts for the flow battery, which are expensive. Temperature stability of the complexed bromine is also an issue, since temperature must be kept below 50 C.


Zinc-cerium flow battery

The zinc-cerium battery is a nonaqueous battery. It is an important battery because of its high potential. The electrolyte used is methanesulfonic acid. The high potential of the catholyte cerium requires the use of very expensive electrode materials (titan electrodes and precious metal coatings) for the cathode. Graphite felt cannot be used for the reaction as the cathode side. It would be oxidized because of the high potential. At the anode, zinc is electroplated on and stripped off the carbon polymer electrode during charge and discharge, respectively [22e24]: Zn2þ(aq) þ 2e4Zn(s)(0.76 V vs. SHE) At the positive electrode (cathode), Ce(III) oxidation and Ce(IV) reduction take place during charge and discharge, respectively: Ce3þ(aq)  e4Ce4þ(aq) (ca. þ 1.44 V vs. SHE) Because of the large cell voltage, hydrogen (0 V vs. SHE) and oxygen (þ1.23 V vs. SHE) could evolve theoretically as side reactions during battery operation (especially on charging). The positive electrolyte is a solution of cerium(III) methanesulfonate. Due to the high standard electrode potentials of both zinc and cerium redox reactions, the open-circuit cell voltage is as high as 2.43 V. Methanesulfonic acid is used as electrolyte, as it allows high concentrations of both zinc and cerium; the solubility of the corresponding methanesulfonates is 2.1 mol for Zn, 2.4 mol for Ce(III), and up to 1.0 mol for Ce(IV). Methanesulfonic acid isparticularly well suited for industrial electrochemical applications and is considered to be a green alternative to other support electrolytes.


Iron/iron flow battery

One simple approach is the all-iron hybrid flow battery which uses iron as a very cheap electrolyte. Iron is plated at the anode and the Fe2þ/Fe3þ is in the form of a complex in alkaline solution. The iron electrode is well known and has been used for decades in the nickel-iron battery; the reaction is highly reversible and very stable. As the catholyte, ferro/ferricyanide can be used; this is also a well-investigated reaction and


Storing Energy

known as a stable redox system. The kinetics of the reaction are fast, so high current densities up to 200 mA cme2 could be achieved. The reactions involved are: Fe2þ / Fe3þ þ e þ 0.77 V Fe2þ þ 2e / Fe0  0.41V A challenge is hydrogen evolution as a sidereaction; this reduces the efficiency of the system [25].


Copper/copper flow battery

A relatively simple approach is the use of an all-copper hybrid flow battery. The idea is to stabilize Cu(I) as a chloro complex in solution using suitable anions such as chloride or amine. There are three different valence states of copper available for use with this battery. One interesting effect is that this battery could in principle be recharged by applying higher temperatures in which case the Cu(I) complex becomes unstable and disproportionate to metallic copper and Cu(II), which is the starting material of the charged battery. The energy density (20 W h/L1) achieved is comparable with traditional VRFBs. This is due to the high solubility of copper (3M), which offsets the relatively low cell potential (0.6 V). The electrolyte is cheap, simple to prepare, and easy to recycle, since no additives or catalysts are used. The system can be operated at 60 C eliminating the need for a heat exchanger and delivers energy efficiencies of 93%, 86% and 74% at 5, 10, and 20 mA/cm2, respectively [26].


Hydrogen-bromine battery

The hydrogen-bromine battery works with sodium-bromine in alkaline solution, which is a low-cost and well-known electrolyte. The combination with the hydrogen evolution reaction (HER) and hydrogen reduction reaction (HRR) has the advantage of being very fast with low overpotential in combination with the high oxidative power of bromine. These advantages are tempered by disadvantages such as on both sides catalysts are needed to enhance the reaction. Newdevelopments include working with nonnoble catalyst systems, but the state-of-the-art does involve precious metal catalysts at the anode as well as at the cathode. To ensure reasonable storage capacity the hydrogen must be stored under pressure. This could partly be achieved by hydrogen evolution in the stack of up to 1 MPa (10 bar) to 2 MPa (20 bar). For higher pressures a compressor is needed, making the overall system both complex and expensive [27]. Very recently, the Israeli company EnStorage has developed such a system with a target capacity of 150 and 900 kWh.

References [1] S. Ashimura, Y. Miyake, Polarization characteristics of redox type fuel cell cathode at a flow-through porous carbon electrode, Denki, Kagaku 39 (12) (1971) 977e978 (in Japanese).

Vanadium redox flow batteries


[2] M. Bartolozzi, Development of redox flow batteries. A historical bibliography, J. Power Sources 27 (1989) 219e234. [3] L.H. Thaller, in: Ninth Intersociety Energy Conversion Engineering Conference, San Francisco, CA; August 26e30, 1974, pp. 924e928. NASA TM X-71540. [4] L.H. Thaller, Electrically rechargeable redox flow cell, US Patent 3996064 (1976). [5] C.L. Chen, H.K. Yeoh, M.H. Chakrabarti, An enhancement to Vynnycky’s model for the all-vanadium redox flow battery, Electrochim. Acta 120 (2014) 167e179, 10.1016/j.electacta.2013.12.074. [6] K. Oh, H. Yoo, J. Ko, S. Won, H. Ju, Three-dimensional, transient, nonisothermal model of all- vanadium redox flow batteries, Energy 81 (2015) 3e14, [7] B. Xiong, J. Zhao, L. Jinbin, Modeling of an all-vanadium redox flow battery and optimization of flow rates, in: IEEE Power and Energy Society Meeting, Vancouver, Canada, 2013. [8] Q. Xu, T. Zhao, P. Leung, Numerical investigations of flow field designs for vanadium redox flow batteries, J. Appl. Energy 105 (2013), pages 47-5. [9] G. Qiu, C.R. Dennison, K.W. Knehr, E.C. Kumbur, Y. Sun, J Pore-scale analysis of effects of electrode morphology and electrolyte flow conditions on performance of vanadium redox flow batteries, J. Power Sources 219 (2012) 223e234, j.jpowsour.2012.07.042. [10] G. Qiu, A.S. Joshi, C.R. Dennison, K.W. Knehr, E.C. Kumbur, Y. Sun, 3-D pore-scale resolved model for coupled species/charge/fluid transport in a vanadium redox flow battery, Electrochim. Acta 64 (2012) 46e64, [11] X. Ma, H. Zhang, F. Xing, A three-dimensional model for negative half cell of the vanadium redox flow battery, Electrochim. Acta 58 (2011) 238e246, 10.1016/j.electacta.2011.09.042. [12] A. Shah, M. Watt-Smith, F. Walsh, A dynamic performance model for redox-flow batteries involving soluble species, Electrochim. Acta 53 (2008) 8087e8100, 10.1016/j.electacta.2008.05.067. [13] C. Fabjan, et al., The vanadium redox-battery: an efficient storage unit for photovoltaic systems, Electrochim. Acta 47 (2001) 825e831, 00763-0. [14] M. Hashimoto, M. Kubata, A. Yagasaki, (NH4)3[VO2(SO4)2(H2O)2]$1.5H2O, Crystal structure communications, Acta Crystallogr. C Cryst. Struct. Commun. C56 (2000) 1411e1412, [15] M. Kazacos, M. Cheng, M. Skyllas-Kazacos, Vanadium redox cell electrolyte optimization studies, J. Appl. Electrochem. 20 (1990) 463e467. [16] S. Kim, E. Thomsen, G. Xia, Z. Nie, 1 kW/1 kW h advanced vanadium redox flow battery utiliz- ing mixed acid electrolytes, J. Power Sources 237 (2013) 300e3009. [17] M. Sterner, I. Stadler, Energiespeicher, Springer Vieweg Verlag, Berlin, 2014, p. 292. [18] A. Whitehead, N. Trampert, P. Pokorny, P. Binder, T. Rabbow, Critical safety features of the vanadium redox flow battery, in: International Flow Battery Forum, Glasgow, June 2015, p. 62. [19] R.R. Moskalyk, A.M. Alfantazi, Miner. Eng. 16 (2003) 793e805. [20] L. Erdmann, S. Behrend, M. Feil, Kritische Rohstoffe f€ ur Deutschland, Insti- tut f€ ur Zukunftsstudien und Technologiebewertung (IZT) and Adelphi, Berlin, Germany, 2011. [21] B. Huskinson, M.P. Marshak, C. Suh, S. Er, M.R. Gerhardt, C.J. Galvin, X. Chen, A metalfree organiceinorganic aqueous flow battery, Nature 505 (2014) 195e198.


Storing Energy

[22] G. Nikiforidis, L. Berlouis, D. Hall, D. Hodgson, Evaluation of carbon composite materials for the negative electrode in the zincecerium redox flow cell, J. Power Sources 206 (2012) 497e503. [23] G. Nikiforidis, L. Berlouis, D. Hall, D. Hodgson, A study of different carbon composite materi- als for the negative half-cell reaction of the zinc cerium hybrid redox flow cell, Electrochim. Acta 113 (2013) 412e423. [24] P.K. Leung, C. Ponce de Leon, C.T.J. Low, F.C. Walsh, Zinc deposition and dissolution in meth- anesulfonic acid onto a carbon composite electrode as the negative electrode reactions in a hybrid redox flow battery, Electrochim. Acta 56 (2011) 6536e6546. [25] J. Sassen, J. Goldstein, L. Eliad, N. Baram, A novel/iron/iron flow battery for grid storage, in: Inter- National Flow Battery Forum, Glasgow, June 2015, p. 48. [26] P. Leung, E. Garcia-Quismondo, L. Sanz, J. Palma, M. Anderson, Evaluation of electrode materi- als towards extended cycle-life of all copper redox flow batteries, in: International Flow Battery Forum, Glasgow, June 2015, p. 36. [27] M. Tucker, A. Weber, R. Wycisk, P. Pintauro, Improving the durability, performance, and cost of the Br2eH2 redox flow cell, in: International Flow Battery Forum, Glasgow, June 2015, p. 56. [28] J. Noack, N. Roznyatovskaya, C. Menictas, M. Skyllas-Kazacos, Understanding vanadium redox flow batteries, Storage and Smart Power (May 2020) 76e83. [29]; accessed 2021-01-10. [30] “Lazard’s Levelized Cost of Storage Analysis e Version 4.0”, 2018 available via, https:// accessed 2020-01-10. [31], assessed 2021-01-10. [32] (Accessed on November 16, 2020).

Further reading [1] Ausfelder, et al., Energiespeicherung als Element einer sicheren Energieversorgung, Chem. Ing. Tech. 87 (1e2) (2015) 17e89, [2] Sandia National Laboratories on behalf of the US Department of Energy’s (DOE) Office of Electricity Delivery and Energy Reliability and the DOE’s Office of Energy Efficiency and Renewable Energy Solar Technologies Program Electric Power Industry Needs for Grid Scale Storage Applications. Industry_ Needs_2010.pdf. [3] European Association for Storage of Energy (EASE). Energy Storage Roadmap. p. 40. ES%20Roadmap%202030/EASEEERA%20ES%20Tech%20Dev%20Roadmap%202030%20Final%202013.03.11.pdf. [4] [5] [6] [7]

Vanadium redox flow batteries


[8] Sandia Corporation. Global energy storage data base. http://www.energystorageexchange. org/projects?utf8¼%E2%9C%93&technology_type_sort_eqs¼VanadiumþRedoxþFlowþ Battery&technology_type_sort_eqs_category¼Electro-chemical&technology_typesort_ eqs_ subcategory¼Electro-chemical%3AFlowþBattery&technology_type_sort_eqs_child¼Electrochemical%3AFlowþBattery%3AVanadiumþRedoxþFlowþBattery&country_sort_eq¼&state_sort_eq¼&kW¼&kWh¼&service_use_case_inf¼&ownership_model_eq¼&status_ eq¼Operational&siting_eq¼&order_by¼&sort_order¼&search_page¼1&size_kw_ll¼&size_ kw_ul¼&size_kwh_ll¼&size_kwh_ul¼&show_unapproved¼%7B%7D. [9]

This page intentionally left blank

Supercapacitors 1, 2



Narendra Kurra and Qiu Jiang 1 School of Chemistry, Indian Institute of Science Education and Research Thiruvananthapuram (IISER-TVM), Thiruvananthapuram, Kerala, India; 2Department of Chemistry, Indian Institute of Technology Hyderabad, Kandi, Sangareddy, Telangana, India; 3 School of Materials and Energy, University of Electronic Science and Technology of China, Chengdu, PR China

1. Introduction The rapid depletion of fossil fuels with adverse environmental impacts led to the deployment of pollution-free renewable energy resources. Though renewables such as solar, wind, and thermal do provide infinite energy, their intermittency poses a major concern. In this scenario, efficient electrochemical energy storage technologies offer sustainable solution toward development of fossil fuelefree society. For instance, lithium-ion batteries offer high energy density up to 270 Wh/kg, but often suffer from poor power density and cycle life. Sluggish mass diffusion kinetics of ions across the electrode material with subsequent volume expansion/contraction during charge/discharge cycles make batteries operate slowly for a limited number of cycles. Alternatively, dielectric capacitors store charge electrostatically and can be operated in milliseconds time scales but at a limited energy density of 102 Wh/kg. Though supercapacitor is a trade name for commercial electrical double-layer capacitor, it is being used in literature to refer to EDLCs or electrochemical capacitors (ECs). At operating time scales of a few seconds, supercapacitors do store more energy than batteries, and they can last longer than batteries. High efficiency, reversible storage of charges for millions of cycles at fast charge-discharge rates make supercapacitors potential power devices for many applications. A few applications with low energy requirements (5e10 Wh/kg), supercapacitors may even replace batteries. Batteries can also be operated in high power mode without losing their cycling stability by operating them in tandem with supercapacitors [1]. Electrochemical energy storage (EES) technologies offer optimal gravimetric (volumetric) energy/power performance metrics in comparison to other storage technologies including pumped hydro, mechanical, thermal, and magnetic. Moreover, ease of integration with renewable sources that are intermittent in supply of energy, EES offers an advantage in constant supply of energy at demand. Supercapacitors do not need constant voltage for charging unlike batteries, and as a result open up opportunities for the development of self-powered autonomous energy storage devices. For instance, regenerative braking energy has been exploited in charging supercapacitors.

Storing Energy. Copyright © 2022 Elsevier Inc. All rights reserved.


Storing Energy

The amount of charge that can be stored and delivered from an energy storage device is primarily dependent on the nature of charge storage mechanism of electrode materials, architecture, and device configuration. The shift from double-layer charging in porous carbon materials to redox charging in transition metal compounds is to bridge the gap between supercapacitors and batteries with the aim of achieving simultaneous high-energy at high-power performance.


Basics of charge storage


Electric charge

An electric charge can be generated by rubbing two objects against each other; this is the well-known triboelectric effect. Thales of Miletus rubbed amber with a cloth which caused attraction of lightweight particles, the basis for discovery of electric charges. Significant experimental efforts led by Michael Faraday, Thomson, and Millikan on the nature of electron provided pathways for the molecular electronic level understanding of electricity [2]. Fundamentally, attraction and repulsion are two main characteristics of charges; repulsive forces operate between like charges and attractive forces operate between opposite charges. The smallest unit of “free” charge is considered to be the charge on an electron or a proton that has a magnitude of e ¼ 1:602  1019 C


Quantized unit of a negative charge (e) is carried by an electron while proton is designated to carry one unit of positive charge (þe). After initial understanding about the nature of electric charge, development of Leyden jar has led to the new phenomenon of charge separation and storage. On the two surfaces of the Leyden jar two metal foils are separated by a layer of glass (Fig. 18.1). Thus, such capacitor separates and stores charges in an electrostatic manner, considered as a basis for the further developments in capacitor technologies over centuries.


Energy storage in a capacitor

A capacitor is often described as a passive circuit element in an electronic circuitry. It consists of two metallic conductors that are separated by a dielectric material. On application of a potential difference across the conducting plates, an electric field drives accumulation of positive Qþ and negative charges Q- on the plates, which are physically separated by a dielectric layer. Thus, transient charge storage is possible in the form of electrostatic field between conducting plates. Ideally, there is no current flow across the capacitor during the charging process. The linear relationship between amount of charge stored (Q) in a capacitor and electric potential difference between the plates (V) is given by Q ¼ CV




Figure 18.1 (A) Schematic representation of a capacitor design. (B) Cross-sectional representation of Leyden jar (glass jar is loaded with water, metal foils at the inner and outer surfaces are denoted A and B).

where C is a positive proportionality constant called capacitance; it signifies the amount of charge needed to induce a specified potential difference between the conducting plates. The physical meaning of capacitance is expressed in terms of charge storage capacity with respect to potential difference (V) between the plates. Dielectric capacitors can exhibit capacitance over a wide range from pico (1012F) to millifarad (103 F), as is primarily governed by the nature of the dielectric medium. Consider a parallel plate capacitor in vacuum, the electric field outside of the planar conductive surface is given by Gauss’ law E¼

Q ε0 A


Electric potential difference between the plates is given by V ¼ Ed


Qd ε0 A


Q ¼ CV


Q Qd ¼ C ε0 A


V¼ Since

Q ε0 A ¼ V d



Storing Energy

Thus, in a medium, capacitance of a capacitor depends on the geometric area (A) of the plates, dielectric constant (ε) of the medium, and distance between the plates (d), given by the following equation C¼

εε0 A d


where ε0 is the permittivity of the vacuum ¼ 8.8  1012 F/m, and ε is the relative permittivity of the medium. Thus, charge storage capacity is limited by the dielectric constant, area of the electrodes, and the distance between the electrodes.


Historical evolution from capacitors to electrical double-layer capacitors

Historically, a Leyden jar was the first charge storing device, which dates back to 1745 [3]. Due to nature of electrostatic charging, capacitors can store charges rapidly and also discharge very fast. They are considered as power sources which can release pulse power for small periods of time (seconds to milli seconds). Thus, capacitors are suitable power devices in backup circuits of microcomputers and timer circuits for providing periodic charge/discharge cycles. Additionally, the capacitors are being employed in filter units where they can reduce the ripples of dc voltage to meet specific requirements. Though conventional dielectric capacitors are limited by their capacitance values, they are used in frequency filtering applications where the time constant (s ¼ RC, R is the resistance and C is the capacitance) becomes the governing factor. Electrolytic capacitors are considered as second generation of capacitors. Typically, Al or Ta metals with their insulating oxide layers (through anodization process) serve as a positive or anode terminal. Liquid or gel electrolyte covering on the thin oxide dielectric serves as cathode or negative terminal with metal electrode contacts. Increasing the anode surface area and reducing the thickness of the dielectric layer result in electrolytic capacitors storing more charge (Q ¼ CV) per unit volume than dielectric capacitors. Currently, aluminum electrolytic capacitors (AECs) are being employed in filtering out ripples in an electronic circuitry, commonly known as ac-line filtering. Fig. 18.2 outlines the timeline for development of capacitor technologies. H. Becker at General Electrics gets the credit for his pioneering work on electrical double-layer capacitor (EDLC) in 1957. Initial crude EDLC device consisted of porous carbon electrodes dipped in a container of liquid electrolyte; however, this posed a challenge from the commercialization point of view. Further, Robert A. Rightmire at Standard Oil Company of Ohio (SOHIO) optimized EDLC layout by introducing a separator (only allows passage of ions while prevents the electrical shorting of electrodes) between two porous carbon electrodes in sulfuric acid electrolyte. Such kind of optimized format further led interest in researchers in exploring



Figure 18.2 Chronological developments from capacitors to supercapacitors. Schematic of pseudocapacitive charge storage of RuO2 was reproduced from Ref. [4] Copyright 2014 Royal Society of Chemistry.

EDLCs which has resulted in numerous patents and publications. The Japanese company, Nippon Electric Company (NEC), introduced EDLCs into electronics market, as backup power devices for volatile clock chips and complementary metal oxide semiconductor (CMOS) computer memories. During the past few decades, EDLCs expand their usage from portable to transportation sectors, including wireless communication and hybrid electric vehicles (HEVs). Moving from dielectric capacitors to electrochemical capacitors (Fig. 18.3), there are major differences in electrode materials, design perspective, and operating charge storage mechanisms (Table 18.1). In case of ECs, high-surface-area porous carbon/ electrolyte interface replaces the metal/dielectric interface of capacitors. Interfacial capacitance of 10e20 mF/cm2 (based on electrolyte) is recorded for electrical double layers [1]. Thus, typical layout of EDLC includes porous activated carbon electrodes interfacing with electrolyte solution. The electric charge is stored across the

Figure 18.3 Schematic representations for (A) electrolytic capacitor, (B) supercapacitor, and (C) lithium-ion battery. Reproduced from K. Jost, G. Dion, Y. Gogotsi, Y., Textile energy storage in perspective, J. Mater. Chem. 2 (2014) 10776e10787 Copyright 2014 RSC Publishing.


Storing Energy

Table 18.1 Various charge storage features of capacitors, supercapacitors and batteries.



Storage mechanism


Charge time Life time E (Wh/kg) P (kW/kg) Charge/Discharge time Coulombic efficiency Vmax limitation Charge storage limited by Discharge profile Self-discharge

Electrochemical capacitor


103e106s >10 years 10 106-103 s

Electrostatic or chemical 1e10s >10 years 5e10 0.5e20 Seconds to minutes

Chemical >10 min