Experimental Operation, Modelling and Simulation of Solid Oxide Cell Reactors with Multiple Stacks for Process Systems Von der Fakultät Energie-, Verfahrens- und Biotechnik der Universität Stuttgart zur Erlangung der Würde eines Doktors der Ingenieurwissenschaften (Dr.-Ing.) genehmigte Abhandlung Vorgelegt von Marius Tomberg aus Kamp-Lintfort Hauptberichter: Prof. Dr. rer. nat. K. Andreas Friedrich Mitberichter: Prof. Dr. Massimo Santarelli Tag der mündlichen Prüfung: 28.03.2024 Universität Stuttgart Institut für Gebäudeenergetik, Thermotechnik und Energiespeicherung 2024 Acknowledgments I would like to thank Prof. K. Andreas Friedrich for giving me the opportunity to carry out my research on this important topic. I furthermore thank him for the support and advice he gave me whenever needed. I also thank Prof. Massimo Santarelli for being the co-examiner. Furthermore, I thank Dr. Marc P. Heddrich for his support, his honesty, trust and feed- back as well as our inspiring professional and private conversations. Especially, his valu- able feedback on my manuscripts, which he gave me despite lack of time, was crucial for this thesis. Next, I would like to thank the Hybrid Power Plant crew for their support and for keep- ing up humor even in difficult times. These are my (former) colleagues Christian Sch- negelberger, Dr. Matthias Metten, Dr. Mike Steilen and Dr. Moritz Henke as well as the colleagues from the DLR Institute of Combustion Technology Anna Marcellan, Jürgen Roth, Dr. Martina Hohloch, Melanie Herbst, Thomas Krummrein and Timo Lingstädt. I would like to thank the TEMPEST guys – Daniele Fortunati, Faisal Sedeqi, Hans Wiggenhauser, René Lorenz and Santiago Salas Ventura – for too long stand-up meet- ings, their motivation and their input. Special thanks also go to my former office mates for our discussions about science and everything else: Dr. Marc Riedel for his input on experimental issues and Dr. Srikanth Santhanam for laying the foundation for TEMPEST. iii Moreover, I want to thank Dirk Ullmer, Maximilian Groß and Dr. Diana Amaya Dueñas for support with experiments and validation as well as Daruska Miric Fuentes and San- chit Gupta for discussions on modelling and simulation. I would also like to thank all other DLR colleagues who supported me as well as all project partners with whom I had interesting discussions. Finally, this thesis would not have been possible without the support of the most im- portant people in my private life. I would like to express my sincere gratitude to my parents and my brother, who have always supported and motivated me, as well as to my friends, who may have been neglected by me, but kept my spirits up. Last but not least I would like to thank Tatjana – for everything. iv Abstract Solid Oxide Cell (SOC) reactors are highly efficient electrochemical energy converters. They can be operated in electrolysis mode (SOEC) to produce chemical feedstocks and in fuel cell mode (SOFC) to convert chemical energy into electricity. These character- istics enable them to meet the challenges of the energy transition and the increasing penetration of renewables, such as intermittency and integration into specific industrial sectors and processes. SOC reactors can therefore play a central role in the energy sys- tem of the future. Today’s large-scale SOC reactors are composed of multiple stacks/sub- reactors, resulting in a modular design. However, such an arrangement leads to several operational challenges for further scaling and operation. The objective of this dissertation is to establish a general understanding of the opera- tional behavior of large SOC reactors and to contribute to the deployment of SOC re- actors in the future energy system by developing generic scaling, operation and control strategies. In this work, above objective is addressed by experimental and numerical studies of SOC reactors with multiple stacks. The approach is based on the construction and operation of test rigs to study large reactors, the development and application of a simulation framework and a strong interaction between these two. The experiments demonstrated the general operational behavior and provided parameterization as well as validation data for the simulations. These, in turn, supported the experimental investigations, for example, by providing estimates of operating parameters for specific operating points. v Finally, the simulations were used to develop operation and control strategies that were iteratively improved by using feedback from the experiments. A unique test rig for the investigation of SOC reactors with multiple stacks was built, which contains a blower for off-gas recirculation at SOC reaction temperature and a pressure vessel for operation under pressure. This pressurized reactor test rig was used to study a modular 30 kW SOFC reactor with multiple stacks. In addition, a simulation framework for the study of process systems with SOC reactors was created. The simulation framework has two unique features. First, it allows modelling of complete SOC reactors consisting of mul- tiple stacks, pipes, manifolds, thermal insulation, and thermal interaction between all these components. Second, the framework provides the capability for transient simu- lation of not only the SOC reactors, but also all the required BoP components. Both the test rig and the simulation framework were used to develop generic strategies for reliable operation. In a measurement campaign with the pressurized SOC reactor test rig, fuel gas, reac- tant conversion, and pressure were varied in stationary and transient experiments. The experimental results showed that the operating conditions of the individual stacks of large SOFC reactors vary largely due to flow distribution and heat losses. Methods for the investigation of these critical characteristic parameters were derived from the ex- perimental results. Furthermore, the impact of pressurization and fuel gas recirculation on the SOFC reactor was analyzed. These experimental investigations showed the need to understand the behavior of large SOFC reactors with multiple stacks to increase the performance and robustness of complete process systems. Therefore, the simulation framework was applied to an entire SOC reactor consisting of multiple stacks, pipes, manifolds, and thermal insulation. After experimental validation on stack and reactor level, the model was used to investigate the fundamental behavior of the SOC reactor and its individual stacks in fuel cell and electrolysis mode. Subsequently, the simulation framework was applied to develop operational and control strategies. An example that also provides generic conclusions, is the model of a megawatt scale flexible electroly- sis system consisting of twelve reactors with a nominal load of 80 kW. The model was used to define crucial and efficient operation points and to establish transitions between these by comparing different strategies and control approaches. The simulation results showed that systems with SOCs can be operated more transiently than usually assumed. For instance, the start-up time from a hot standby point was reduced by 80 %, while at the same time the temperature gradients were significantly reduced. Furthermore, by vi taking advantage of the modular nature of state-of-the-art reactors, fast power modula- tion was achieved. In addition to the study of electrolysis systems, operating strategies for fuel cell operation were developed with a focus on the challenges that arose from the experiments with the pressurized SOC reactor test rig. A control strategy was devel- oped for sub-reactors with separate electrical channels but shared reactant processing units. It uses the operating parameters of each stack in each sub-reactor to improve power and efficiency. This was successfully tested on the pressurized SOC reactor test rig. In addition, a feed forward temperature control was developed and experimentally validated, resulting in a significantly improved controller and the possibility of faster power ramps. A unique test rig was operated for the first scientific investigations focusing on SOC reactors with multiple stacks, and a simulation framework was developed to study large SOC reactors in process systems. Both activities contributed to a better understanding of large reactors and to the integration and operation of SOC reactors in the future energy system. In the process, unique experimental results were obtained and operating as well as control strategies were developed. vii Zusammenfassung Reaktoren mit Festoxidzellen (englisch: Solid Oxide Cell (SOC)) sind hocheffiziente elektrochemische Energiewandler. Sie können im Elektrolysemodus (SOEC) zur Her- stellung chemischer Grundstoffe und im Brennstoffzellenmodus (SOFC) zur Umwand- lung chemischer Energie in Strom betrieben werden. Aufgrund dieser Eigenschaften sind sie in der Lage, einen Beitrag zur Lösung der Herausforderungen der Energiewende und der zunehmenden Nutzung erneuerbarer Energien zu leisten, wie z. B. dem Aus- gleich von Fluktuationen und der Integration in bestimmte Industriesektoren sowie -prozessen. SOC-Reaktoren können daher eine zentrale Rolle im Energiesystem der Zukunft spielen. Heutige großtechnische SOC-Reaktoren sind modular aus mehreren Stacks/Teilreaktoren aufgebaut. Eine solche Anordnung bringt jedoch eine Reihe von Herausforderungen für die weitere Skalierung und den Betrieb mit sich. Ziel dieser Dissertation ist es, ein allgemeines Verständnis des Betriebsverhaltens von großen Reaktoren mit Festoxidzellen zu erlangen und durch die Entwicklung von allge- meinen Skalierungs-, Betriebs- und Regelungsstrategien einen Beitrag zum Einsatz von SOC-Reaktoren im zukünftigen Energiesystem zu leisten. In dieser Arbeit wird das oben genannten Ziel durch experimentelle und numerische Un- tersuchungen von SOC-Reaktoren mit mehreren Stacks angegangen. Der Ansatz basiert auf dem Aufbau und Betrieb von Prüfständen zur Untersuchung großer Reaktoren, der Entwicklung und dem Einsatz eines Simulationsframeworks und einer starken Inter- aktion zwischen beiden. Die Experimente zeigten das allgemeine Betriebsverhalten ix und lieferten Parametrisierungs- und Validierungsdaten für die Simulationen. Diese wiederum unterstützten die experimentellen Untersuchungen, indem sie zum Beispiel Parameter für bestimmte Betriebspunkte lieferten. Schließlich wurden Simulationen zur Entwicklung von Betriebs- und Regelungsstrategien verwendet, die durch Nutzung experimenteller Daten iterativ verbessert wurden. Für die Untersuchung von SOC-Reaktoren mit mehreren Stacks wurde ein einzigar- tiger Prüfstand gebaut, der ein Heißgasgebläse für die Abgasrezirkulation und einen Druckbehälter für den Betrieb unter Druck enthält. Dieser Teststand wurde für die Un- tersuchung eines modularen 30 kW SOFC-Reaktors mit mehreren Stacks verwendet. Darüber hinaus wurde ein Simulationsframework für die Untersuchung von verfahrens- technischen Systemen mit SOC-Reaktoren geschaffen. Das Simulationsframework hat zwei besondere Merkmale. Erstens ermöglicht es die Erstellung von Modellen komplet- ter SOC-Reaktoren, die aus mehreren Stacks, Rohren, Verteilern, thermischer Isolierung und thermischer Interaktion zwischen all diesen Komponenten bestehen. Zweitens bie- tet das Framework die Möglichkeit zur instationären Simulation nicht nur der SOC- Reaktoren, sondern auch aller erforderlichen BoP-Komponenten. Sowohl der Prüfstand als auch das Simulationsframework wurden genutzt, um generische Strategien für einen zuverlässigen Betrieb zu entwickeln. In einer Messkampagne mit dem druckaufgeladenen SOC-Reaktor-Teststand wurden Brenngas, Umsatz und Druck in stationären und transienten Experimenten variiert. Die experimentellen Ergebnisse zeigten, dass die Betriebsbedingungen der einzelnen Stacks von großen SOFC-Reaktoren aufgrund von Strömungsverteilung und Wärme- verlusten stark variieren. Aus den Versuchsergebnissen wurden Methoden zur Unter- suchung dieser kritischen Kenngrößen abgeleitet. Darüber hinaus wurde der Einfluss von Druck und Brenngasrezirkulation auf den SOFC-Reaktor analysiert. Diese expe- rimentellen Untersuchungen haben gezeigt, dass es notwendig ist, das Verhalten von großen SOC-Reaktoren mit mehreren Stacks zu verstehen, um die Performance und Robustheit verfahrenstechnischer Systeme zu erhöhen. Daher wurde das Simulations- framework genutzt, um einen SOC-Reaktor mit 24 Stacks, Rohren, Verteilern und ther- mischer Isolierung zu modellieren. Nach der experimentellen Validierung auf Stack- und Reaktorebene wurde dieses Modell zur Untersuchung des grundlegenden Verhal- tens des SOC-Reaktors und seiner einzelnen Stacks im Brennstoffzellen- und Elektro- lysebetrieb verwendet. Anschließend wurde das Simulationsframework zur Entwick- lung von Betriebs- und Kontrollstrategien eingesetzt. Ein Beispiel, das auch gene- x rische Schlussfolgerungen liefert, ist das Modell eines flexiblen Elektrolysesystems im Megawattbereich, das aus zwölf Reaktoren mit einer Nennleistung von 80 kW besteht. Mit dem Modell wurden kritische und effiziente Betriebspunkte definiert und Transien- ten zwischen diesen durch den Vergleich verschiedener Strategien und Steuerungsan- sätze bestimmt. Die Simulationsergebnisse zeigten, dass Anlagen mit SOCs transienter betrieben werden können als allgemein angenommen. So konnte beispielsweise die An- fahrzeit aus einem heißen Standby-Punkt um 80 % reduziert werden, während gleich- zeitig die Temperaturgradienten deutlich verringert wurden. Darüber hinaus wurde durch die Ausnutzung des modularen Charakters moderner Reaktoren eine schnelle Leistungsmodulation erreicht. Neben der Untersuchung von Elektrolysesystemen wur- den Betriebsstrategien für den Brennstoffzellenbetrieb entwickelt, wobei der Schwer- punkt auf den Herausforderungen lag, die sich aus den Experimenten mit dem druck- aufgeladenen SOFC-Teststand ergaben. Es wurde eine Regelungsstrategie für Teilreak- toren mit getrennten elektrischen Kanälen, aber gemeinsamen Edukt- und Produkt- gaskomponenten entwickelt. Sie nutzt die Betriebsparameter jedes Stacks in jedem Teilreaktor, um Leistung und Effizienz zu verbessern. Dies wurde erfolgreich auf dem druckaufgeladenen SOFC-Teststand getestet. Darüber hinaus wurde eine Temperatur- regelung mit Vorsteuerung entwickelt und experimentell validiert, was zu einem deut- lich verbesserten Regler und der Möglichkeit schnellerer Leistungsrampen führte. Es wurde ein einzigartiger Prüfstand für die ersten wissenschaftlichen Untersuchungen mit Schwerpunkt auf SOC-Reaktoren mit mehreren Stacks betrieben und ein Simula- tionsframework für die Untersuchung großer SOC-Reaktoren in verfahrenstechnischen Systemen entwickelt. Beides dient dem besseren Verständnis größerer Reaktoren und trägt zur Integration und zum Betrieb von SOC-Reaktoren im zukünftigen Energiesys- tem bei. Es wurden einzigartige experimentelle Ergebnisse erzielt und Betriebs- und Regelungsstrategien entwickelt. xi Author’s declaration I declare that I have written this dissertation with the title “Experimental Operation, Modelling and Simulation of Solid Oxide Cell Reactors with Multiple Stacks for Process Systems” independently and that it has not been submitted for any other academic degree. Except where indicated in the text, the thesis is the candidate’s own work. xiii Table of contents List of figures xvii Nomenclature xix List of scientific publications 1 1 Introduction 5 1.1 Background – energy storage, sector coupling and chemical feedstock . . 5 1.2 Background – power generation, re-electrification and powertrains . . . 7 1.3 Potential and challenges of reactors with solid oxide cells . . . . . . . . . 8 1.4 Thesis structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 2 Research motivation 13 2.1 State of the art . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 2.2 Scientific gap . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 2.3 Motivation and scientific approach . . . . . . . . . . . . . . . . . . . . . 18 3 Fundamentals of electrochemical process systems engineering 21 3.1 Balancing of electrochemical reactors . . . . . . . . . . . . . . . . . . . . 22 3.2 Electrochemical fundamentals and operating principle of solid oxide cells 23 3.3 Fundamentals of control engineering . . . . . . . . . . . . . . . . . . . . 28 4 Methodology 31 4.1 Theoretical and numerical investigations . . . . . . . . . . . . . . . . . . 31 4.1.1 Transient modelling and simulation (for operation) . . . . . . . . 32 xv 4.1.2 Representation of SOC reactors with multiple stacks (for scaling) 33 4.1.3 Cell model and electrochemical parameterization . . . . . . . . . 33 4.2 Experimental investigations . . . . . . . . . . . . . . . . . . . . . . . . . 33 4.2.1 Investigation of a reactor with multiple stacks in process system context . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 4.2.2 Provision of parameterization and validation data . . . . . . . . . 36 5 Publications 39 5.1 Differentiation of the author’s work from the co-authors or previous work 39 5.2 Article I . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40 5.3 Article II . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 5.4 Article III . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67 6 Cumulated results & discussion in the scientific context 83 6.1 Experimental investigation and analysis of an SOFC reactor . . . . . . . 84 6.1.1 Operation of SOC reactors with multiple stacks . . . . . . . . . . 85 6.1.2 Design of systems with SOC reactors with multiple stacks . . . . 86 6.1.3 Operation of SOC reactors in SOFC/GT hybrid power plants . . . 87 6.1.4 Fuel flexibility of SOFC systems . . . . . . . . . . . . . . . . . . . 88 6.1.5 Pressure control for SOC systems . . . . . . . . . . . . . . . . . . 89 6.1.6 Impact of elevated pressure on heat losses . . . . . . . . . . . . . 92 6.1.7 Comparison with other research on pressurized SOC systems . . 92 6.2 Modelling, simulation and reactor analysis . . . . . . . . . . . . . . . . . 94 6.2.1 Framework development and applicability . . . . . . . . . . . . . 94 6.2.2 Simulation results beyond measurements . . . . . . . . . . . . . 95 6.2.3 Derivation of parameters for system studies . . . . . . . . . . . . 96 6.3 Development and application of operation and control strategies . . . . 97 6.3.1 Support of experimental investigations . . . . . . . . . . . . . . . 97 6.3.2 Hot standby operation . . . . . . . . . . . . . . . . . . . . . . . . 98 6.3.3 Control of several sub-reactors . . . . . . . . . . . . . . . . . . . 99 6.3.4 Reactor temperature control . . . . . . . . . . . . . . . . . . . . . 101 7 Conclusion 107 8 Outlook & future work 111 Bibliography 113 xvi List of figures 2.1 Used sub disciplines of electrochemical process systems engineering in this work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 2.2 Main research steps and their interconnection . . . . . . . . . . . . . . . 20 3.1 Schematic of an electrochemical reactor and the currents and flows en- tering and leaving it. (adapted from [88]). . . . . . . . . . . . . . . . . . 22 3.2 Schematic of a solid oxide cell. Blue font and arrows represent the elec- trolysis mode and red ones the fuel cell mode. . . . . . . . . . . . . . . . 24 3.3 Thermodynamics of the H2-H2O, the CO-CO2 and the combining water- gas shift reaction, showing the variations of T∆Sr, ∆Gr and ∆Hr at atmo- spheric pressure and the corresponding stoichiometric reaction voltages at standard pressure. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 3.4 Depiction of an Open-loop controller (left) and a close-loop controller (right). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 3.5 Combination of feedback and feedforward controller. . . . . . . . . . . . 30 4.1 Interaction of the modelling framework TEMPEST and the test rig Galac- tica (adapted from [14]). . . . . . . . . . . . . . . . . . . . . . . . . . . 32 4.2 Pressurized reactor test rig built and operated at German Aerospace Cen- ter (DLR) in Stuttgart. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 xvii 4.3 Flow sheet of the SOFC/MGT hybrid power plant (a) and the derived pressurized reactor test rig before (b) and after (c) the modifications (adapted from [11] and [12]) . . . . . . . . . . . . . . . . . . . . . . . . 35 4.4 DLR’s test rigs Horst (a) and Galactica (b), which are described in the next subsections. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36 6.1 The main steps taken, the publication of these in scientific articles and the application of the two main tools/methods. . . . . . . . . . . . . . . 84 6.2 Impact of the CO2 content in the emulated biogas on the efficiency of the SOFC reactor (adapted from [13]). . . . . . . . . . . . . . . . . . . . . . 89 6.3 Two-stage feedforward controller for the sensor compartment. . . . . . . 91 6.4 Histogram showing the error of the pressure controller before and after the modification (a) and controller performance at an emergency stop (b, before modification, adapted from [13]). . . . . . . . . . . . . . . . . 91 6.5 Comparison of the results of the pre-simulations and the experiment. . . 98 6.6 Control strategy for operation of reactors with individual power channels for sub-reactors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100 6.7 Experimental heat-up results using the control strategy for operation of reactors with individual power channels for sub-reactors. . . . . . . . . . 102 6.8 Experimental comparison of different control approaches for the maxi- mum stack temperature. . . . . . . . . . . . . . . . . . . . . . . . . . . . 104 xviii Nomenclature Latin Symbols Symbol Description a Activity A Area ASR Area specific resistance C Heat capacity d Differential operator e error F Faraday constant f Function G Gibbs free energy g Specific Gibbs free energy h Specific enthalpy H Enthalpy Ḣ Enthalpy flow rate i Electrical current density I Electrical current k Controller gain Continued on next page xix Symbol Description kB Boltzmann constant l Mean free path m Mass M Molar mass ṁ Mass flow rate n Amount of substance ṅ Amount of substance flow rate P Power p Pressure Q̇ Heat flow rate q Electric charge R Resistance R Gas constant RC Reactant conversion S Entropy s Specific entropy T Temperature t Time U Voltage x Molar fractions z Number of electrons xx Greek Symbols Symbol Description ∆ Difference η Efficiency ν Number density ρ Density σ Cross sectional collision area Superscripts, Subscripts and Abbreviations Abbreviations Description 0 Standard state act Activation BoP Balance of Plant CHP Chemical CHP Combined heat and power d differential DC Direct current diff Diffusion DLR Deutsches Zentrum für Luft- und Raumfahrt EIS Electrochemical impedance spectroscopy el Electrical FF Feed forward i Integral id Ideal LHV Lower heating value LNG Liquefied natural gas (M)GT (Micro) gas turbine Max Maximum Continued on next page xxi Abbreviations Description Min Minimum nom Nominal ohm Ohmic OP Operation point p Proportional PI(D) Proportional-integral-(derivative) PLC Programmable logic controller PtX Power-to-X r Reaction R Reactor ref reference SGR Synthesis gas ratio SOC Solid oxide cell SOEC Solid oxide electrolysis cell SOFC Solid oxide fuel cell SR Sub-reactor sys system WGS Water-gas shift xxii List of scientific publications This is a cumulative thesis, which is mainly based on the three scientific articles listed below. The contributions as a co-author and to scientific conferences also led to this thesis. Additional information about the author’s contribution to the publications can be found in chapter 5. First author publications Article I: M. Tomberg, M. P. Heddrich, M. Metten, S. A. Ansar, and K. A. Friedrich. “Operation of a Solid Oxide Fuel Cell Reactor with Multiple Stacks in a Pressured System with Fuel Gas Recirculation”. In: Energy Technology 10(4), 2101075 (2022). DOI: 10.1002/ente.202101075 Article II: M. Tomberg, M. P. Heddrich, F. Sedeqi, D. Ullmer, S. A. Ansar, and K. A. Friedrich. “A New Approach to Modeling Solid Oxide Cell Reactors with Multiple Stacks for Process System Simulation”. In: Journal of the Elec- trochemical Society 169(5), 054530 (2022). DOI: 10.1149/1945-7111/ ac7009 Article III: M. Tomberg, M. P. Heddrich, S. A. Ansar, and K. A. Friedrich. “Operation strategies for a flexible megawatt scale electrolysis system for synthesis gas and hydrogen production with direct air capture of carbon dioxide”. 1 https://doi.org/10.1002/ente.202101075 https://doi.org/10.1149/1945-7111/ac7009 https://doi.org/10.1149/1945-7111/ac7009 List of scientific publications In: Sustainable Energy & Fuels 7(2) (2023), pp. 471–484. DOI: 10.1039/ d2se01473d Co-author publications and contributions to conferences M. Tomberg, Srikanth, S., M. Steilen, M. Riedel, and M. P. Heddrich. Effi- cient Fuel Cell Models for SOFC/GT Hybrid Power Plant System Simulations Designed to be Parameterized with Experimental Results. Presented at: The 7th International Conference on ”Fundamentals & Development of Fuel Cells”. Stuttgart, 2017. URL: https://elib.dlr.de/111217/ C. Schnegelberger, M. Henke, M. Tomberg, M. P. Heddrich, and K. A. Friedrich. Current progress in the design and setup of a SOFC/GT hybrid power plant. Presented at: WHTC 2017. Prague, 2017. URL: https: //elib.dlr.de/116577/ M. Tomberg, M. P. Heddrich, A. Ansar, and K. A. Friedrich. Experimental Setup of a Pressurized 30 kW SOFC System and Transient Large SOC Module Simulation. Presented at: 11th Progress in Fuel Cell Systems. Bruges, 2018. URL: https://elib.dlr.de/122694/ M. Heddrich, M. Steilen, M. Tomberg, and A. Friedrich. “Influence of heat transfer on operation of a solid oxide fuel cell/gas turbine hybrid demonstrator”. In: Proceedings of 13th European SOFC & SOE Forum, Chapter 05: Sessions A11, A12. Ed. by E. Ivers-Tiffée et al., pp. 82–86. ISBN: 978-3-905592-23-8. URL: https://elib.dlr.de/122611/ M. Tomberg, S. Santhanam, M. P. Heddrich, A. Ansar, and K. A. Friedrich. “Transient Modelling of Solid Oxide Cell Modules and 50 kW Experimental Validation”. In: ECS Transactions 91(1) (2019), pp. 2089–2096. DOI: 10.1149/09101.2089ecst A. Marcellan, A. Abrassi, and M. Tomberg. “Cyber-Physical System of a Solid Oxide Fuel Cell/Micro Gas Turbine Hybrid Power Plant”. In: E3S Web of Conferences 113, 02006 (2019): SUPEHR19 SUstainable PolyEn- ergy generation and HaRvesting Volume 1. DOI: 10 . 1051 / e3sconf / 201911302006 2 https://doi.org/10.1039/d2se01473d https://doi.org/10.1039/d2se01473d https://elib.dlr.de/111217/ https://elib.dlr.de/116577/ https://elib.dlr.de/116577/ https://elib.dlr.de/122694/ https://elib.dlr.de/122611/ https://doi.org/10.1149/09101.2089ecst https://doi.org/10.1051/e3sconf/201911302006 https://doi.org/10.1051/e3sconf/201911302006 S. Santhanam, A. Padinjarethil, M. Tomberg, M. P. Heddrich, and A. Ansar. “Transient Operation Strategies for MW-Scale SOC Systems”. In: ECS Transactions 91(1) (2019), pp. 2571–2578. DOI: 10.1149/09101.2571ec st M. Metten, M. Tomberg, M. Heddrich, and K. A. Friedrich. Experimen- tal Setup of a Pressurized Solid Oxide Fuel Cell System for Hybrid Power Plants. Presented at: Jahrestreffen der ProcessNet-Fachgruppe Energiev- erfahrenstechnik und des Arbeitsausschusses Thermische Energiespeicher. Frankfurt, 2019 M. Metten, M. Tomberg, M. P. Heddrich, and K. A. Friedrich. “Analysis of experimental results of a Pressurized Solid Oxide Fuel Cell System sim- ulating a Hybrid Power Plant”. In: E3S Web of Conferences 113, 02007 (2019): SUPEHR19 SUstainable PolyEnergy generation and HaRvesting Vol- ume 1. DOI: 10.1051/e3sconf/201911302007 M. Tomberg, M. Metten, C. Schnegelberger, M. P. Heddrich, A. Ansar, and K. A. Friedrich. “Experimental analysis of a pressurized 30 kW SOFC system with fuel gas recirculation at reaction temperature”. In: 14th Euro- pean SOFC & SOE Forum. Lucerne, 2020. DOI: 10.5281/zenodo.4550589 M. Tomberg, D. M. Amaya Dueñas, M. P. Heddrich, D. Ullmer, S. A. Ansar, and K. A. Friedrich. “Operation of Co-SOEC reactors for syngas production utilizing CO2 from air”. In: 15th European SOFC & SOE Forum. Ed. by O. Bucheli, G. Geisser, F. Moore, and M. Spirig. Lucerne, 2022 D. Fortunati, M. Tomberg, D. M. A. Dueñas, M. P. Heddrich, and A. S. Ansar. Simulative and experimental investigations of transient operating behaviour of SOEC reactors with multiple stacks for syngas production. Pre- sented at: 3rd Internaional Conference on Electrolysis. Golden, Colorado, 2022 S. Salas Ventura, M. Metten, M. Tomberg, D. Ullmer, M. P. Heddrich, and S. A. Ansar. “Operation Analysis of a Flexible Solid Oxide Cell Module for Power to Hydrogen and Polygeneration”. In: Chemie Ingenieur Technik 94(9) (2022), pp. 1321–1321. DOI: 10.1002/cite.202255103 3 https://doi.org/10.1149/09101.2571ecst https://doi.org/10.1149/09101.2571ecst https://doi.org/10.1051/e3sconf/201911302007 https://doi.org/10.5281/zenodo.4550589 https://doi.org/10.1002/cite.202255103 List of scientific publications D. M. Amaya Dueñas, D. Ullmer, M. Riedel, M. Tomberg, M. P. Heddrich, and S. A. Ansar. “Critical Operating Conditions for Co-SOEC Reactors for Syngas Production with Fischer-Tropsch Recirculation”. In: Chemie Ingenieur Technik 94(9) (2022), pp. 1322–1322. DOI: 10.1002/cite. 202255067 S. Salas Ventura, M. Metten, M. Tomberg, D. Ullmer, M. P. Heddrich, and S. A. Ansar. “Transient simulation and experimental validation of a solid oxide cell module in electrolysis and polygeneration mode”. In: 15th Eu- ropean SOFC & SOE Forum. Ed. by O. Bucheli, G. Geisser, F. Moore, and M. Spirig. Lucerne, 2022 R. Lorenz, M. Tomberg, F. Resink, M. P. Heddrich, and S. A. Ansar. Tran- sient operating strategies for Solar Heat Supported Solid Oxide Electrolysis systems for hydrogen production. Presented at: 21st Wind & Solar Integra- tion Workshop. The Hague, 2022 D. M. Amaya Dueñas, D. Ullmer, M. Riedel, M. Tomberg, D. Fortunati, M. P. Heddrich, and S. A. Ansar. “Operating Strategies for Fischer-Tropsch Tail Gas Recirculation on a 100 kW SOEC Reactor”. In: ECS Transactions 111(6) (2023), pp. 1941–1946. DOI: 10.1149/11106.1941ecst S. Salas Ventura, M. Metten, M. Tomberg, D. Ullmer, C. Ünlübayir, M. P. Heddrich, and S. A. Ansar. “Transient Solid Oxide Cell Reactor Model Used in rSOC Mode-Switching Analysis and Power Split Control of an SOFC-Battery Hybrid”. In: ECS Transactions 111(6) (2023), pp. 1795– 1801. DOI: 10.1149/11106.1795ecst 4 https://doi.org/10.1002/cite.202255067 https://doi.org/10.1002/cite.202255067 https://doi.org/10.1149/11106.1941ecst https://doi.org/10.1149/11106.1795ecst 1 Introduction Global warming and endangered biodiversity are considered to be the greatest chal- lenges of the present and the coming decades. To address these and to reduce human greenhouse gas emissions, governments and authorities around the world are imple- menting action plans as EU’s European Green Deal [22] and USA’s new Federal Sus- tainability Plan [23]. Furthermore, the costs of fossil fuels are rising due to dwindling reserves and the global security situation [24]. In many industries, this leads to the need to defossilize the production of goods. The issue of climate change is also very present in public opinion and investors increase pressure on companies [25]. Central to these plans and efforts is power generation from renewable sources, which raises some challenges due to their intermittent nature. In addition, some sectors and processes are hard-to-electrify, which means that direct electrification is difficult to achieve. The following sections identify pathways and technologies that are required to replace fossil fuels with renewables. 1.1 Background – energy storage, sector coupling and chemical feedstock The most important renewable energies are solar, wind, hydro, biomass and geothermal energy. Solar and wind energy not only already account for the largest share, but also 5 1 Introduction have the potential to meet humanity’s energy demand many times over [26]. How- ever, these are subject to short-term fluctuations and seasonal changes. To create a fully renewable energy system, the growing share of intermittent energy sources must be integrated into an energy grid and infrastructure that can meet the demand [27]. Short-term fluctuations and the limited controllability of electricity from wind turbines and photovoltaic plants lead to a temporary discrepancy between supply and demand of renewable energy [28] and a need for short-term energy storage. Furthermore, there are seasonal differences as well as events which are characterized by a prolonged ab- sence of wind and sun (German: Dunkelflaute). Both make long-term and large-scale storage as well as conversion necessary [27]. Thermal, electrical and mechanical en- ergy storage systems as well as chemical storage systems can be used for this purpose. Studies show that the cost of chemical liquid storage can be much lower than the cost of battery storage [27, 29]. In addition to the dedicated installation of storage infrastructure for power supply, it is possible to use storage options from other sectors. A well-known example is the use of batteries from electric cars for electricity storage. Electric cars are an example of sector coupling. Sector coupling involves the integrated use of different energy infrastructures (energy converters, storage) and vectors (especially electricity, heat and gas) [30]. One possibility is to use renewable energy instead of fossil-based energy for heating and cooling, in transport as well as in industrial processes. This can be achieved by direct electrification or by chemical energy carriers such as hydrogen. There is progress in the electrification of individual passenger transport. However, heavy-duty, long-haul, and aviation applications are hard-to-electrify because battery storage is expensive and heavy. Solutions may be power-to-X (PtX) technologies, which aims to convert elec- trical energy into a variety of energy carriers and products such as liquid fuels for the transport sector. Power-to-X technologies are also the key to the industrial sector, which is responsible for a large share (28 % in Germany [31]) of final energy consumption. Particularly energy-intensive and hard-to-electrify are the cement and steel production. However, steel production can be defossilized by using hydrogen from renewables for direct reduction of iron (DRI). Finally, the chemical industry and the production of com- modities can also be defossilized with PtX processes. Hydrogen and synthesis gas are base chemicals for the chemical industry and are today mostly produced from natural gas. Another example is the heating sector that relies heavily on fossil fuels (e.g. only 16.5 % in Germany are renewable [32]). Along with improved insulation as well as 6 1.2 Background – power generation, re-electrification and powertrains use of waste heat, geothermal heat and solar thermal energy, electrical energy can be used for heating. This can be done by heat pumps, Joule heaters or by renewable fuels (hydrogen, synthetic gases etc.), which would be produced by electrolysis plants from renewable electrical energy. Electrolysis is the core technology for all hydrogen and PtX solutions. It is the electrically driven splitting of molecules, such as the H2O electrolysis to produce hydrogen. By supplying electrolysis with renewable electricity, the production of fuels and chemicals can be decoupled from fossil resources. This can lead to an energy system based solely on renewable energy [27]. 1.2 Background – power generation, re-electrification and powertrains In the future, fuels will be more expensive compared to today’s fossil fuel prices. The future price of fossil fuels will be higher than today’s price due to lower supply and higher taxes. This also applies to the price of renewable fuels due to their generation from electricity, storage costs, and limited quantities of biomass. Therefore, conversion efficiency becomes even more important than it is today. One example is the afore- mentioned seasonal storage of renewable electricity as chemical energy. Due to the preceding conversion steps and storage costs, re-electrification should be as energy effi- cient as possible. Already today, technologies with very high electrical efficiencies exist, such as modern combined cycle power plants with over 60 % electrical efficiency. How- ever, these plants are large and suitable primarily as medium-load power plants. Small decentralized plants with a high fuel and load flexibility as well as high efficiencies are beneficial for the future energy system. Another important aspect of future power gen- eration plants is their integration into the heating sector. Either for room heating or for industrial processes. In addition to the provision of electrical energy to the grid, the transport sector is of importance. The defossilization of powertrains can be realized directly via batteries and electrical drives or via synthetic fuels. Electrification is advantageous for light com- mercial vehicles, e.g. cars for private transport. However, in aviation, maritime, and heavy-duty transport, the use of synthetic fuels, with their high energy density appears 7 1 Introduction to be advantageous [27]. For aviation applications, high power and storage density as well as low weight are very important. Synthetic kerosene could replace today’s fossil kerosene using the infrastructure already in place. High power and storage density are also required in shipping. In recent years, the global shipping industry has started to switch from crude oil to liquefied natural gas (LNG). Methanol, synthetic LNG and am- monia are being investigated for the next step. Heavy-duty transportation requires high performance, long driving ranges and reliability. 1.3 Potential and challenges of reactors with solid oxide cells Reactors with solid oxide cells (SOCs) can efficiently convert chemical energy into elec- trical energy in fuel cell mode (SOFC) and electrical energy into chemical energy as well as high-value molecules in electrolysis mode (SOEC) [28]. These possibilities en- able various applications. In fuel cell mode, SOFCs have very high electrical efficiencies due to the direct conver- sion of chemical energy to electrical energy without combustion [33, 34]. Of all fuel cells, SOFCs can achieve the highest electrical and overall efficiencies due to fast kinetics at higher operating temperatures as well as the possibility of combined heat and power (CHP) operation. The favorable thermodynamics and kinetics result in the achieve- ment of unparalleled efficiencies in electrolysis operation [27], which can be further increased by heat integration. In comparison to low-temperature electrolyzes, SOECs can be supplied with steam instead of water. This allows waste heat to be used for evap- oration, which can save about 16 % of the total DC power consumption at a reaction temperature of 800 °C. Another feature of SOC reactors is their high fuel flexibility in fuel cell mode. The high operating temperature and the stable catalysts, allow the cells to cope with various fuels based on hydrogen and reformed hydrocarbons [33, 35]. Moreover, in electrolysis mode, SOCs can split not only H2O into hydrogen and oxygen, but also CO2 into carbon monoxide and oxygen. Consequently, co-electrolysis, which is the parallel reduction of H2O and CO2 directly to synthesis gas (syngas; CO+H2), can be achieved using SOCs. The synthesis gas can be used in chemical syntheses reactors to 8 1.3 Potential and challenges of reactors with solid oxide cells produce synthetic natural gas, gasoline, methanol, and ammonia. These synthesis pro- cesses are typically exothermic and allow thermal integration with the SOEC process [27] allowing the boost via electrolysis of steam. In addition, SOCs can be operated under pressure. Pressurization of SOCs can offer several advantages to further increase the impact of this technology. First, it increases the ideal voltage, which is positive in fuel cell mode and negative in electrolysis mode. It has a positive impact on reaction kinetics, which in combination can lead to a higher reactor efficiency also in electrolysis mode [36]. Second, the overall efficiency of a process system can be increased. For ex- ample, in fuel cell mode in combination with a gas turbine cycle [37] or in electrolysis mode when liquid water is compressed instead of the product gases. Reactors with SOC cells have potentially low costs. Compared to the low temperature technologies, they do not contain precious metals but more abundant materials such as nickel, zirconia, and steel [27, 33, 38]. This enables the large-scale deployment of fuel cell and electrolyzer systems using SOCs. However, even these materials may become scarce in the course of the energy transition [39]. Referring to the previous sections, it can be concluded that reactors with SOCs can play a significant role in the future energy system. They can be used to efficiently convert renewable energy into chemical energy carriers, both directly into hydrogen or synthesis gas and via synthesis routes into gaseous or liquid fuels. Furthermore, these products play an important role in sector coupling applications and as chemical feedstock. The chemical energy carriers produced can be re-electrified in fuel cell operation and used in powertrains with SOFCs. Unfortunately, there are still some challenges that need to be resolved. These can be divided into challenges related to materials and manufacturing, and those related to process system technology. At material and manufacturing level, challenges arise from the high operating temper- atures and brittle materials. The different ceramic layers have different sintering tem- peratures, resulting in a time-consuming and expensive manufacturing process. These different layers also have different coefficients of thermal expansion, which can lead to interface problems between materials [33, 40]. Although the catalysts used are rela- tively cheap, the costs of metallic compounds and internal manifolds are high because high operating temperatures require expensive alloys. These high temperatures, ther- mal cycling as well as the oxidizing and reducing atmospheres result in difficulties to 9 1 Introduction find reliable sealants [41]. However, material research is directed towards lower oper- ating temperatures of SOCs. Another potential is to extend the lifetime of SOCs. During operation, degradation of cell components occurs, such as corrosion and creep of in- terconnects, crystallization of glass sealings, coarsening of cermet electrodes, Nickel depletion and poisoning due to impurities (S, Si etc.) [27]. In addition, SOCs are highly susceptible to pressure differences across the electrolyte. Temporal and local temperature gradients can lead to cell damage due to different thermal expansion of the materials. Both pressure and temperature effects can be mitigated by a sophisticated process system control. The large amount of required electrolysis capacity discussed can be achieved by significantly increasing the active cell area and current density. How- ever, beyond a certain point this is limited by technological and manufacturing barriers, since a larger cell area also increases the probability of fractures in the electrolyte [42]. An alternative to cell scaling is the arrangement of several cells in one layer, the so- called window design [43]. However, additional expensive steel is required for the gaps between the cells. Furthermore, there are challenges related to reactor scaling, system integration, and op- eration that can be addressed by process systems engineering. The integration of SOCs into large reactors with multiple stacks and eventually systems is still in its early stages, and the required gigawatt-scale electrolysis power may not be achievable at reasonable cost with current reactors. Thus, strategies for the integration of solid oxide cells into larger reactors are required. In addition, the high operating temperature of SOCs leads to material issues for balance of plant (BoP) components. At the system level, BoP com- ponents must be designed to operate reliably at high temperatures and in aggressive atmospheres without excessive cost increase. Furthermore, the alloys of BoP compo- nents contain chromium and silica that may lead to cell degradation. In addition, given the changing energy system and the absence of baseload power, electrolyzer plants may have a low capacity factor. Intermittent renewables request fast transients and cur- rent SOC systems are limited in their ability to perform fast load changes due to the high thermal inertia of SOC and the fragile materials. Therefore, operation and control strategies must be developed to allow fast transients without exceeding the operating limits. Cells and systems are still expensive because of low economies of scale. However, prices will fall as demand and production capacity increase, and SOECs could be cost- competitive not only with other electrolysis technologies, but with fossil-based processes 10 1.4 Thesis structure due to rising fossil fuel prices and high taxes on CO2 emissions [27]. Beyond that, this work will address the challenges in scaling and integration as well as operation of large reactors. 1.4 Thesis structure This thesis is divided into seven chapters. The following is a brief outline of each chap- ter: Chapter 1 Introduction: Describes the overarching questions addressed in this dissertation. Chapter 2 Research motivation: Explains the current scientific state, scientific gaps and the scientific re- search approach. Chapter 3 Process systems with reactors consisting of solid oxide cells: Gives a brief presentation of the fundamentals of SOCs and process sys- tems. Chapter 4 Methodology: Explains how the research approach is implemented and which method- ology is used. Chapter 5 Publications: The published articles are presented. Chapter 6 Results and discussion: Discussion of all results and explanation of the context. Chapter 7 Conclusion: Concludes all of the presented work. Chapter 8 Outlook & future work: Describes already started follow-up activities as well as possible future activities. 11 2 Research motivation A large amount of electrolysis and fuel cell power is needed to fulfill the potentials men- tioned in the introduction. However, large commercial systems in the multi-megawatt and gigawatt range are not yet available. In this chapter, the scientific motivation of this thesis is explained. The current state of the technology is listed in section 2.1. The scientific gap from which the motivation of this work is derived is explained in section 2.2. The approach taken to fulfill this gap is described in section 2.3. 2.1 State of the art Nearly 200 years have passed since Christian Friedrich Schönbein discovered the work- ing principle of fuel cells in 1829 and Walther Nernst ceramic material consisting of 85 % ZrO2 and 15 % Y2O3 in 1899 [44, 45]. The first major commercial entity that re- searched solid oxide fuel cells was the Westinghouse Electric Corporation (later Siemens Westinghouse Power Corporation) [46] resulting in several demonstrators and systems up to 250 kW between the 1960s and 2000s [44]. In the 1980s, Dornier System GmbH developed and reported on solid oxide electrolyzers [47]. Today, SOFCs and SOECs are being developed worldwide [33, 48]. Development con- tinues on the material and cell levels, but increasingly development efforts shift to larger reactors with multiple stacks and SOC systems. These can be divided into two groups: 13 2 Research motivation First, standalone systems for CHP applications and second, industrial scale reactors for integration into process systems. There are mostly standalone systems for fuel cell applications. Several manufacturers have launched products or are on the way to the market. These systems are targeted at small or medium combined heat and power CHP plants for households, districts or small businesses. Often, fuel processing (natural gas, propane, hydrogen) and off- gas treatment are integrated into each plant (Sunfire-Home 750 ][49] (Sunfire Fuel Cells GmbH1), SolydEra2 BlueGen BG-15 [50], Osaka Gas3/AISIN4 ENE-FARM Type S [51]). Another application are datacenters and systems for power generation in remote locations (e.g. Sunfire-Remote [52]). Data centers require a reliable power supply, which is provided by modular SOFC systems [53, 54]., Additionally, cooling power is needed, which can be provided by absorption cooling. Industrial scale stationary SOFC systems in the 100 kW to MW range are relatively rare. The leading manufacturer is Bloom Energy5 with its 300 kW SOFC energy servers [55]. These are composed of several SOFC sub-reactors and several energy servers can be combined to form large multi-megawatt plants. Even though Bloom is scal- ing the sub-reactors slowly, an extreme numbering up strategy is used to deliver large power outputs. Accelerated by state policies as California’s Self-Generation Incentive Program (SGIP) and Fuel Cell Net Energy Metering Program, Bloom has been able to sell many plants from 300 kW to multi-megawatt scale [56, 57]. In 2021 Bloom sold 1897 units at an aggregate capacity of 190 MW [56]. In Europe, activities to develop large SOFC systems have slowed down as the focus has shifted to SOEC technology. Convion6 presented its C60 SOFC system with Elcogen7 stacks [58, 59, 60], SolydEra (former SolidPower) is testing a 32 kW reactor called large stack module (LSM) [61], Ceres power licenses their stack technology to several engineering companies around the world, which could result in a scale-up in the next years [56, 62]. Pressurized hy- brid systems have been intensively researched by industry and science for many years to increase of the power output by pressurization and combination with gas turbines 1Sunfire Fuel Cells GmbH, Germany, https://sunfire-fuel-cells.de/en/ 2SolydEra SpA, Italy, https://solydera.com/ 3Osaka Gas Co., Ltd., Japan, https://www.osakagas.co.jp/en/ 4Aisin Seiki Co. Ltd., Japan, https://www.aisin.com/en/ 5Bloom Energy, USA, https://www.bloomenergy.com/ 6Convion Ltd., Finland, https://convion.fi/ 7Elcogen AS, Estonia, https://elcogen.com/ 14 2.2 Scientific gap (GT) or internal combustion engines. Of the industrial actors, only Mitsubishi Power8 continues to work on its Megamie pressurized SOFC/GT hybrid system [56]. Currently, a 250 kW system is being offered, which is to be complemented by a 1 MW system from 2023 [63]. Electrolyzer systems with are already moving into the multi-megawatt range due to higher power densities, lower heat generation and lower flows on the oxygen electrode side. Sunfire9 offers their HyLink (steam electrolysis [64]) and SynLink (co-electrolysis [65]) systems in the megawatt range based on several sub-reactors. Topsoe A/S10 offers their technology for steam [66, 67] as well as CO2 electrolysis [67, 68] and Bloom announced a steam electrolysis system [69] based on their numbering-up strat- egy. SolydEra’s LSM is rated for 100 kW in electrolysis mode [61] and the Convion announced field demonstration of their first electrolyser system C250E in late 2022 [60]. 2.2 Scientific gap The technological readiness of the SOC technology has been increasing in recent years and commercially available reactors and systems grew in size as presented in section 2.1. However, the size of the systems with SOC reactors is still too small to play a significant role in the future energy system. To satisfy the demand in system power a numbering-up strategy can be applied. However, for further cost reduction, a scale-up SOC reactors into the megawatt range is beneficial. Additionally, operation strategies for upcoming larger reactors need to be investigated to ensure efficient and robust op- eration (e.g. for grid support). Scale-up can be achieved by increasing the power of individual cells. However, area increase of cells is limited due to mechanical stability [42] of the electrolyte and an increase in the current density is also limited by techno- logical and material barriers. Furthermore, integration strategies need to be developed to transfer SOC technology into reactors for process systems. Many of these issues have not even been addressed much less solved on a generic level. Consequently, the chal- lenges of industrial manufacturers can be solved by scientific research. 8Mitsubishi Heavy Industries, Ltd., Japan, https://power.mhi.com/ 9Sunfire GmbH, Germany, https://www.sunfire.de/en/ 10Topsoe A/S, Denmark, https://www.topsoe.com/ 15 2 Research motivation There was a huge amount of research in the field of SOFCs since the 1990s [33, 70]. Through political will and initiatives such as the "Clean Hydrogen Partnership", the focus has shifted towards electrolysis and reversible operation in recent years, particularly in Europe. The research focuses on the one hand on material, component and cell development resulting in a large amount of publications and progress on state-of-the- art technologies and next generation devices (e.g. proton conducting ceramics). On the other hand, there are many publications on system studies and basic system design, which often do not consider how to operate these systems and how the electrochemical reactor will behave in detail. In the following, the gap between these two sides will be explained. First, in the field of experimental studies, publications exist from material up to stack level. Stack level investigations are important as the performance and degradation can highly differ com- pared to that of single cells. Investigations on stack level were intensified only in the last years. In Europe, R&D activities at the French Alternative Energies and Atomic Energy Commission (CEA), the European Institute for Energy Research (EIFER), the Forschungszentrum Jülich (FZJ), the German Aerospace Center (DLR) and the Danish Technical University (DTU) are leading the research efforts on stack level. Some experimental system investigations were conducted but mostly at a rather small size [71, 72] or without focusing on the SOC reactor and its stacks [73, 74] As men- tioned before, stacks behave differently compared to cells. But also stacks will behave differently in a larger reactor with multiple stacks in a system. Furthermore, scaling ap- proaches will be modular in the near future [48, 75]. Thus, experimental investigations are necessary with a focus on reactors and their individual stacks. Until now, these have only been conducted by the manufactures themselves and a small number of research projects. Peters et al. built a 20 kW SOFC system with 4 stacks that was operated for 760 hours until one stack was damaged. Subsequently, the system with 2 stacks was op- erated for 4000 hours with a rather high degradation [76]. Srikanth et al. investigated a 4-stack SOFC reactor and published heat-up and pressure loss results [77]. Aicart. et al. did a performance evaluation of a similar 4-stack solid oxide reactor mostly in electrolysis mode [78]. However, these two publications only show first results without an in-depth analysis. Additionally, the mutual impact of a large SOC reactor and the surrounding BoP components has not been investigated experimentally. 16 2.2 Scientific gap Regarding simulations, a similar distribution can be observed. Detailed simulations on material, cell and recently stack level (e.g. [79, 80, 81]) are available. These are often complex and thus computationally intensive, making it impractical to use them in process engineering system studies including heat losses and interactions with BoP components. On the other hand, a large number of publications about systems studies is available [82, 83]. However, these often treat the SOC reactors as zero-dimensional, which results in two major inaccuracies from cell perspective. 1. Temperature gradients along the cell cause degradation [84]. In some system studies, this is ignored, leading to unrealistic conditions, resulting in a too wide operating range. In other system studies, a maximum temperature gradient for the reactor is assumed based on cell studies. This limits the operating range too much, because reactor-internal preheating is ignored. 2. The temperature gradients along the cell also lead to a distribution of the electrical resistance and thus the current density along the cell. Finding a characteristic temperature for the reactor is challenging. Often the outlet temperature is used for the electrochemical calculation, leading especially in fuel cell operation to a strong overestimation of the efficiency. This shows that even for system studies dimensional cell models are beneficial, which was also demonstrated by Li et al., Oryshchyn et al. [85, 86] and Magistri et al. [87]. However, these models must be fast enough for system calculations and in the best case also be able to represent both fuel cell and electrolysis mode. In addition, the individual cells are exposed to different operating conditions in large reactors, due to heat and pressure losses. This is not considered in any model for large reactors with multiple stacks. Moreover, experiments as well as numerical investigations mostly focus on design and characterization of SOC reactors and systems. The operation of these reactors and systems is neglected. Neither is it checked whether all cells are actually operational at individual system operating points, nor are reactor-internal effects considered when determining operating and control strategies. Therefore, two main scientific gaps can be identified. 1. Scaling of reactors and systems 2. Definition of operating strategies 17 2 Research motivation On top of that, these problems result in a technological gap for the industrial entities, mainly the manufactures. As discussed, reactor scaling is one of the main development topics of the industry. However, missing experience and missing tools result in slow upscaling. 2.3 Motivation and scientific approach The described groups of publications do not address the identified main challenges. Thus, the motivation of the thesis is to bridge the scientific gap and support the devel- opment of SOCs and their integration into process systems of the future energy system. As discussed, scaling of reactors with solid oxide cells by significantly increasing the active cell area is challenging and future reactors will be modular. This leads to several questions for reactors and process systems, whose answers can differ for fuel cell and electrolyzer systems. To develop large reactors, it is necessary to investigate how to arrange several sub-reactors/stacks in a large reactor and how to integrate them with balance of plant components. It is also urgent to analyze how to cool or heat such a reactor and to find out which operating points are beneficial for long lifetime, high efficiency but also low cost. Furthermore, it needs to be investigated how to operate these modular systems. In this work, the challenges were addressed using the established (scientific) methods of electrochemical process systems engineering. Different sub disciplines of electro- chemical process systems engineering were used to investigate the integration of elec- trochemical reactors with SOCs into process systems. These were electrochemistry, re- action engineering, simulation technology, plant engineering and process engineering. Figure 2.1 depicts the interaction between these disciplines. A particularly important focus of this work was the combination of experimental investigations with simulations. Figure 2.2 shows the three main steps of research and the strong interconnection be- tween these. In detail the areas can be described as follows: • Experiments on SOC reactors with multiple stacks: As identified, an increase of the cell area is only possible up to a certain extent. Thus, many smaller reactors must be combined to increase power of SOC reactors. 18 2.3 Motivation and scientific approach Electrochemical process systems engineering Process engineering Electrochemistry Plant engineering Reaction engineering Experiments & simulations Figure 2.1: Used sub disciplines of electrochemical process systems engineering in this work In this work experiments on such reactors were performed and the possibilities as well as challenges of these reactors were investigated. To not only investigate the reactor itself but also the impact of a process system on such reactors, a system- near testing facility was built and experiments were conducted. The testing facility includes features as pressurization, fuel gas recirculation at reaction temperature and emulation of gas turbine components, as these experiments were also aimed at investigating combined SOFC/GT systems. Besides results from this pressur- ized reactor test rig, experimental results from the other sources were used as reference. • Modelling and simulation of SOC reactors: As large system experiments are time-consuming and SOC reactors in a relevant size are scarce and expensive, a transient simulation framework was set-up that focuses on two aspects: First, based on the learnings from the experiments, an in-depth analysis of the operability of large reactors with multiple stacks was re- quired. Therefore, the framework was designed to represent such reactors. Sec- ond, as it was necessary to investigate the operation and control of such reactors, the framework was set-up for transient simulations. With the framework various simulation studies were performed and an understanding of SOC reactors with multiple stacks was achieved. • Operation and control as well as scaling strategies: Finally, experiment and simulation were combined to formulate, numerically in- vestigate and experimentally test operation and control strategies. Basic operating 19 2 Research motivation modes of SOC reactors were formulated. From these, complex operating and con- trol strategies were derived, some of which were validated by experiments. With these strategies it was shown that fast and robust transitions between operating points are possible. By deploying an iterative approach (s. Figure 2.2) on these three main steps, theoretical understanding and simulation models were continuously improved. Modelling and simulation of SOC reactors Experiments on SOC reactors with multiple stacks Operation and control as well as scaling strategies Figure 2.2: Main research steps and their interconnection To summarize, this work was motivated by the intention to contribute to the integra- tion and operation of electrochemical reactors with SOCs into industrial process sys- tems. The results were achieved by combining and linking system-near simulations and experiments. 20 3 Fundamentals of electrochemical process systems engineering Electrochemical process systems engineering is an interdisciplinary approach that can be used to develop and investigate the solutions targeted in this work. In contrast to general process engineering and the calculation of chemical reactors, in the case of electrochemical reactors the balance of the electrical charge must be considered in addition to the mass and energy balance. As usual in process and power engineering, in this manuscript flows are expressed as dx/dt = ẋ. Inlet values are identified with a single stroke (x′) and outlet values with a double stroke (x′′). Figure 3.1 shows a schematic of an electrochemical reactor and the currents and flows entering and leaving it. First of all, there are the substance flows of the reactants and products. These also include the electrolyte flow in reactors with a liquid electrolyte. Second, there is the electric current supplied to or taken from the reactor via electronic components and an AC/DC converter for the grid connection. And third, there are the heat flows that pass through heat exchangers. These can be heat flows that are supplied or extracted from different locations in the reactor or from external sources or sinks. 21 3 Fundamentals of electrochemical process systems engineering 6 Electrolyte Heat flow Electric current Substance flowSubstance flow Electrochemical reactor Pre-treatmentReactants Post-treatment Products Heat exchangers Electronics AC/DC converter Grid Heat source/sink Figure 3.1: Schematic of an electrochemical reactor and the currents and flows entering and leaving it. (adapted from [88]). 3.1 Balancing of electrochemical reactors The total mass balance of an electrochemical reactor with a solid electrolyte can be expressed by equation (3.1) considering all flows (indexed by i): dm dt = ∑ i ṁ ′ i − ṁ ′′ i (3.1) The balances of all species in the reactor (indexed by j) can be expressed by the amount of substance n. In addition to the inlet and outlet flows, the consumption or production of substances by chemical and/or electrochemical reactions (indexed by k) must also be considered. dnj dt = ∑ i ṅ ′ i,j − ∑ i ṅ ′′ i,j + ∑ k ṅrk,j (3.2) 22 3.2 Electrochemical fundamentals and operating principle of solid oxide cells In electrochemical reactions, the mass and charge balance are related by Faraday’s law with the Faraday constant F and the number of transferred electrons z: q = nzF d/dt−−→ Ik = ṅzkF = ṁ M zkF (3.3) The energy balance of an electrochemical reactor is equivalent to that of a chemical reactor. It contains the electric power Pel entering or leaving the reactor, the sum of all heat flows and the enthalpy flows (Ḣ = ṁh). Neglecting the flow velocities and geodetic heights, equation (3.4) is obtained. Pel + ∑ i Q̇i = ∑ i Ḣ ′ i − ∑ i Ḣ ′′ i (3.4) The influence of the supplied or extracted electrical power and how it is coupled with equation (3.3) will be explained in the next section after discussing the fundamentals of solid oxide cells. 3.2 Electrochemical fundamentals and operating principle of solid oxide cells Solid oxide cells comprise a porous oxygen electrode and a porous fuel electrode sep- arated by a dense and gas impermeable ion-conducting (O2−) ceramic electrolyte. In addition, other components such as interconnects and sealants are needed, especially when individual cells are joined together to form a cell stack [33, 34]. As shown in Fig- ure 3.2 for a SOC with hydrogen as product or fuel, electrochemical reduction of H2O or CO2 or electrochemical oxidation of H2 or CO occurs at the fuel electrode. Oxide ions are conducted through the electrolyte and to or from the oxygen electrode, while electrons flow through an external circuit [27]. Depending on the process, electrical power must be supplied or can be extracted from the process. This process is only possible at elevated temperatures to reach sufficient ionic conductiv- ity. Most ceramic SOCs typically operate at 650 to 850 °C, leading to high material and integration costs. Research is aimed at lowering the operating temperature [40] in order to simplify process design and reduce material costs for BoP components. Solid oxide 23 3 Fundamentals of electrochemical process systems engineering 4 O x y g e n e le c tr o d e F u e l e le c tr o d e E le k tr o ly te Fuel cell mode Electrolysis mode H2 H2O H2O H2A n o d e C a th o d e C a th o d e A n o d e O2- ½ O2 Load Supply 2e-2e- Power Figure 3.2: Schematic of a solid oxide cell. Blue font and arrows represent the electrolysis mode and red ones the fuel cell mode. cells are available in different mechanical designs (electrolyte-supported (ESC), metal- supported (MSC), fuel electrode–supported (ASC/CSC)), geometric shapes (tubular, planar) and layer designs (single cell per layer, window design). Regardless of the design, however, the general characteristics of the technology remain the same. Ap- proximately 30 to 100 cells are combined electrically in series to a SOC stack using metallic interconnects, for electrical contact and gas separation between the cells. In the next step, these stacks are built into reactors [27]. For this step, different arrange- ments of stacks in the reactor are possible (number of stacks per tower, number of towers, co- and counterflow). Additionally, individual stacks or cell towers can be ther- mally insulated or a large number of stacks can be arranged in one thermal envelope. The flow distribution and electrical interconnections can also be different. All these design choices have an impact on the electrochemical operation of individual cells. For each cell, the operation can be characterized by determining the ideal work and the electrochemical losses, which are described in the next section. The electrical power that must be supplied to or can be extracted from the process depends on the applied current and the resulting potential difference of the cell, which drives the oxide ion through the electrolyte. Two primary electrochemical reactions (equations (3.5) and (3.6)) occur at the triple phase boundary (TPB) of fuel electrode and electrolyte. At the TBP of the of oxygen electrode, the oxidation/dissociation of the 24 3.2 Electrochemical fundamentals and operating principle of solid oxide cells oxide/oxygen occurs (equation (3.7)) to complete the set of redox reaction between the two electrodes. H2O + 2e− ⇌ H2 + O2− (3.5) CO2 + 2e− ⇌ CO + O2− (3.6) O2− ⇌ 1 2O2 + 2e− (3.7) Due to the high temperatures and catalyzing materials at the fuel electrode, in addition to electrochemical reactions, chemical reactions occur at the fuel electrode and in the fuel channel. For operation near typical design points, the (reverse) water-gas shift re- action and the methane reforming/methanation reaction are the most important. These occur during co-electrolysis, and fuel cell operation using hydrocarbons as fuel. H2 + CO2 ⇌ H2O + CO ∆rh 0 = + 41 kJ mol−1 (3.8) CO + 3H2 ⇌ CH4 + H2O ∆rh 0 = −206 kJ mol−1 (3.9) Furthermore, depending on the conditions, other reactions, mostly unwanted, may oc- cur. The Boudouard reaction and CO electrolysis, which can lead to solid carbon depo- sition, should be avoided because they reduce the activity and performance of the cells. In addition, reactions involving the cell materials (e.g. oxidation of the catalyst, inter- action with contaminants) should be avoided. These may be reversible reactions that temporarily reduce performance or irreversible reactions that damage the cell materials or structure. All these reactions determine the electrical power that must be supplied to the reac- tor or that can be extracted from it. The total energy demand/surplus of the reactions corresponds to the reaction enthalpy ∆Hr. This energy is supplied/extracted by a com- bination of electrical work (∆Gr) and thermal energy (T∆Sr). In electrolysis mode, of course, as much thermal energy as possible should be used, and in fuel cell mode, as much electrical energy as possible should be generated. Figure 3.3 shows the tempera- ture dependence of these values. While the advantage of high temperatures is obvious in electrolysis mode, one could conclude that lower temperatures would be preferable in fuel cell mode. However, this only applies to an ideal process, which will be explained in the following sections, including the loss mechanisms that occur. Furthermore, the zero crossing of ∆Gr,WGS does not decide whether H2O or CO2 electrolysis takes place. 25 3 Fundamentals of electrochemical process systems engineering This depends on the exact operating conditions and the reaction kinetics and is still the subject of research. 0 250 500 750 1000 Temperature / °C 0 100 200 300 Sp ez . e ne rg y / k J/m ol Reaction enthalpy hr Thermoneutral voltage Gibbs free reaction energy gr Nernst voltage Heat demand/generation T sr H2O = H2 + 1 2O2 CO2 = CO + 1 2O2 H2 + CO2 = H2O + CO 0.0 0.5 1.0 1.5 Vo lta ge / V Figure 3.3: Thermodynamics of the H2-H2O, the CO-CO2 and the combining water-gas shift reaction, showing the variations of T∆Sr, ∆Gr and ∆Hr at atmospheric pressure and the corresponding stoichiometric reaction voltages at standard pressure. When the electrical current is zero, there is no electrochemical mass and charge transfer at the electrodes and the system is in equilibrium. The resulting cell voltage is called equilibrium or open-circuit voltage [88]. The voltage depends on pressure and temper- ature as well as the activities of species at the electrodes. It can be described by the Nernst equation given in equation (3.10) for a hydrogen oxidation/reduction system assuming ideal gases and an equal pressure at both electrodes (ai = pi/p0) by using mole fractions. ∆G0 is the Gibbs reaction energy at standard state. UNernst = −∆G0 zF + RT zF ln  x̄H2 x̄0.5 O2 x̄H2O ( p p0 )0.5  (3.10) When the electrical circuit is closed and an electrochemical reaction takes place, elec- trical current flows through cell and external circuit. An ideal reactor would operate at the ideal voltage Uid, which can be calculated by using the Nernst equation with conversion-dependent composition or determined by the time derivative of the ideal work (Uid = Ẇideal/I = −(∑i Ġ ′′ i −∑ i Ġ ′ i)/I) [68]. Of course, electrochemical reactors have losses that depend on the operating condition. Thus, the cell voltage is then de- pendent on the current, which affects not only the ideal voltages through the conversion dependence, but also the losses leading to a voltage difference ∆Uloss. Besides the cur- rent influence, the losses also depend on temperature, pressure and gas composition, 26 3.2 Electrochemical fundamentals and operating principle of solid oxide cells which can vary over the length of each cell. Furthermore, the individual cells in one reactor may have different operation parameters. Ucell = f(I) = Uid(I) − ∆Uloss(I) (3.11) Assuming an electrochemical reactor with a single cell, the electrical power Pel of the reactor is determined by the product of cell voltage Ucell and current I: Pel = Ucell I. As for any electric circuit, the relationship between current and voltage difference can be described by Ohm’s law ∆U = R I or in its area-specific form ∆U = ASR i with the area-specific resistance ASR and the current density i = I/A. The total voltage difference ∆Uloss or the total ASR is a combination of the individual loss mechanisms occurring in the of the reactor: • Ohmic voltage losses arise from the resistance to the movement of charge carriers and can be calculated using Ohm’s law (∆Uohm = ASRohm i). The ohmic resistance (ASRohm) is the sum of the resistance in the electrolyte to oxide ion transport, the resistance in the electrodes to electron transport and additionally the resistance in the other functional layers such as contacting layers, interconnects and protective barriers. The ohmic resistance of the electrolyte contributes significantly to the total ohmic resistance and depends on the oxide ion conductivity of the electrolyte. Electrolyte-supported SOCs therefore have a very high ohmic loss due to their thick electrolyte [89]. • Activation losses (∆Uact) result from the charge transfer reactions at the electrode- electrolyte interfaces and depend on the reaction kinetics and reaction mecha- nisms. Activation losses are more significant at lower current densities and de- crease at higher temperatures and pressures. They are usually described by the Butler-Volmer equation [89]. • Diffusion and concentration losses (∆Udiff) lead to lower reactant and higher prod- uct concentrations at the reaction sites. The reactants in the flow channel must diffuse through the electrode to the reaction site and the products must diffuse out. This mass transport limits conversion and leads to a voltage loss especially at high current densities. The voltage loss increases with increasing electrode thick- ness and decreasing porosity. In addition, the conversion leads to a decreasing share of reactants in the course of the flow channel. 27 3 Fundamentals of electrochemical process systems engineering Regarding the reactions (3.5) to (3.7) SOFC operation is always exothermic and heat must be dissipated while in electrolysis operation this depends on T∆Sr, temperature and current density, as the electrical current provides heat through joule heating. At low current densities, the electrochemical losses are small, therefore thermal behavior is dominated by the endothermic electrolysis reaction and electrolysis process is en- dothermic. At high current densities, the higher losses cause the process to become exothermic. In between, if the electricity input is equal to the enthalpy of reaction, the reaction is thermoneutral. This thesis defines another operating point. If the supplied electrical energy is equal to the enthalpy of reaction plus the heat losses to the environ- ment, the operating point is called isothermal because the inlet and outlet flows have roughly the same temperature. 3.3 Fundamentals of control engineering Control engineering uses algorithms and feedback to achieve a desired behavior of equipment and systems [90]. As with any process engineering system, the use of con- trol technology is also necessary in systems with electrochemical reactors. A distinction can be made between open-loop and closed-loop controllers as shown in Figure 3.4. Controller Process iset I U Controller Process Pel,set e I − Pel,act Figure 3.4: Depiction of an Open-loop controller (left) and a close-loop controller (right). An open-loop controller is used when no feedback occurs. An example is the control of the electrical current density i in a SOC system. The controller calculates the actuated variable (the electrical current). In this example, this is done with the only parameter, the cell area. The response of the process is the cell voltage. If the electrical power is to be controlled, a closed-loop controller is required because a feedback exists. Referring to the example, it is not possible to derive the necessary electric current for a certain power from the momentary operating parameters, because if the current changes, the voltage will change as well. Even if the U(I) curve is known, this leads to errors due to disturbance variables (in this example: temperature, reactant 28 3.3 Fundamentals of control engineering conversion, degradation etc.). Thus, a feedback loop is necessary. Instead of trying to guess or calculate the correct actuated variable, the controller increases or decreases the current based on the controller error e = Pact − Pset. Such feedback to correct control errors can be performed in a number of ways [90]. The simplest case is a so-called on-off controller. The actuated variable is set to its maximum or minimum value depending on the sign of the error: u = umax if e > 0, umin if e < 0. (3.12) This leads to large errors and the actuated variable will change over its full range if the error changes its sign, resulting in strong oscillations. Such a controller can be used, for example, to fill a storage tank with a pump, but is unsuitable for power control of an SOC system. The behavior can be improved by using a proportional relation between error and actuated variable [90] with the gain kp: u = kp e (3.13) Proportional control is a strong improvement over on-off control but it has the disad- vantage that the process variable usually deviates from its setpoint, since u = 0 if there is no error. This can be avoided by introducing an integral control with the integral gain ki, which has no steady state error but can lead to oscillations [90]: u(t) = ki ∫ t 0 e(τ)dτ (3.14) The integral control uses the past control error to calculate the actuated variable. By using a linear extrapolation, also the future error can be predicted Td time units ahead [90]. e(t + Td) ≈ e(t) + Td de(t) dt (3.15) Combining this derivative control with the proportional and integral control leads to a so called PID controller [90], which can be expressed as u(t) = kp e + ki ∫ t 0 e(τ)dτ + e(t) + Td de(t) dt . (3.16) 29 3 Fundamentals of electrochemical process systems engineering Often the derivative part is omitted leading to a PI controller. PI(D) controllers have three major drawbacks. First, the three parameters must be determined, second, the controllers are always reactive and third, the process should behave approximately lin- ear. The parameters can be determined using tuning rules. Ziegler and Nichols [91] de- veloped the frequency response and the step response method based on simple charac- terization experiments. In this thesis, the step response method, sometimes also called time domain method, is used to parametrize PID controllers. The response of the pro- cess to a unit step input is measured. The response is characterized by the intersections of the steepest tangent of the process response with the coordinate axes as well as the ratio of the amplitudes of unit step and process response. This information can be used to calculate the parameters of the PID controller applying a variety of tuning rules [90, 91]. PID controllers are reactive. There must be an error before corrective actions can be taken. Having knowledge on the system makes it possible to take corrective actions before an error occurs. This way of controlling a system is called feedforward. Returning to the power control example, a U(I) curve can be used to estimate the current and fine tune it with a PID controller. Additionally, a disturbance compensation can be included as shown in Figure 3.5. Feedback controller Process Feedforward controller IPel,set T , RC, Xi, ... e Pel,act − Figure 3.5: Combination of feedback and feedforward controller. Finally, it should be mentioned that actuators and sensors in the process are never exact. Therefore, additional errors affect the control system. Furthermore, noise on signals must be considered and if necessary filtered. More sophisticated controls, such as adaptive controls, are being researched and increasingly used, but are not a focus of this work. 30 4 Methodology To achieve the goal of large SOC plants, larger reactors must be developed and op- eration strategies must be established. This was addressed in this work through the construction and operation of test rigs to investigate state-of-the-art reactors and the development and use of a simulation framework. Both contributed to the understand- ing and improvement of scaling and operational strategies. The numerical and experi- mental approaches are explained in detail in the following sections. The benefits of this methodology were summarized in Figure 4.1 for the developed simulation framework and one of the test rigs used. The figure shows on the left side the capabilities of the framework (see also section 4.1) and on the right side the test rig Galactica (see sec- tion 4.2.2). The most important aspect was the strong interaction between these two. The experiments provided parameterization and validation data for the simulations. These, in turn, supported the experimental investigations, for example, by providing es- timations of operating parameters for specific operating points. Finally, the simulations were used to develop operation and control strategies that can be iteratively improved by using feedback from the experiments. 4.1 Theoretical and numerical investigations The experimental investigation of these reactors is expensive due to high investment and operation costs. Furthermore, these experiments are time-consuming due to high 31 4 Methodology Modelling Framework TEMPEST GALACTICA Test Rig for SOC Reactors Data for parameterization and validation Simulation results to improve experimental investigations Operation and control strategies Experimental results Iterative improvement M as s f lo w CO2 H2O El . p ow er Sy ng as flow ratio SO C re ac to r T U Figure 4.1: Interaction of the modelling framework TEMPEST and the test rig Galactica (adapted from [14]). thermal inertia (the solids’ heat capacities). On top of that, experimental facilities have a limited operation range and cannot answer questions about reactor scaling. These problems can be mitigated by combining experiments with simulations. To achieve the goals stated in the motivation section, the simulations must meet two criteria. First, the models used to study the operation of process systems with SOC reactors must be transient in order to observe the processes over time. Second, the models used to study scaling must represent all the individual stacks and sub-reactors in a reactor. In combination, these two features allow the study of the operability of all cells and stacks in SOC reactors during steady-state and transient operation. 4.1.1 Transient modelling and simulation (for operation) In this work, the operation of process systems that are operated transiently was in- vestigated. Therefore, the mathematical models used to simulate these systems must represent the time-dependent laws of conservation of mass, energy, charge, and mo- mentum. The modelling framework TEMPEST was built, developed and used, which is designed for transient simulations of systems with electrochemical reactors. The models of the framework are implemented in Modelica® [92], which is an object oriented and acausal programming language. In addition to the electrochemical reactors, all neces- sary BoP components were modeled or taken from existing open source libraries (e.g. ThermoPower [93]). The modelling and simulation work leading the framework was performed by the author, except for the modelling of the cell models and the 10-cell short stacks, which were the result of previous research (see section 4.1.3 for details). 32 4.2 Experimental investigations The framework and its development are discussed in Article II and section 6. 4.1.2 Representation of SOC reactors with multiple stacks (for scaling) The SOC reactor models in the framework should not only be time-depended but also encompass all operation-related sub-components. In this work, these components were identified and modelled (Article II). They include thermal insulation, piping, manifolds, gas, electric power and signal feedthroughs as well as the stacks, which in turn are divided into the electrochemically active cells and the passive parts around them, as well as the stack-internal manifolds. Due to the large number of cells and stacks in large reactors, a novel simplification approach is used on stack level, which is explained in detail Article II and section 6.2.1. 4.1.3 Cell model and electrochemical parameterization In both cases, operating and scaling, a one-dimensional discretization in flow direction is required. Such a cell model was developed by Srikanth S. [94, 95], extended to a short stack in the master’s thesis of R. Lin [96] and parametrized by Srikanth S. and M. Riedel [97]. For a description of the experimental setup for the parameterization see section 4.2.2. Key information is summarized in the "Detailed SOC stack model" section of Article II and in the publications cited above. 4.2 Experimental investigations Although experimental studies are time-consuming and expensive, it was not possible to conduct the intended research without experiments. Compared to simulations, exper- iments represent the real world, including random effects such as differences between cells and non-uniform flow distributions that would not have been considered in mod- elling. In addition, only experimentally validated reactor models should be used in this work to ensure a realistic representation of the reactors and systems studied. Parame- terization exclusively with results from cell measurements is not suitable, because the 33 4 Methodology closer studies are to the system level, the further cell measurements and simple models deviate from reality. A combination of experimental facilities was used. A SOC system test rig was utilized, which provides insight into the real behavior of the reactor in a system, but has a rel- atively small number of sensors and the BoP components cause limitations for reactor investigations. Therefore, for flexible studies with more sensors, a near-system test rig was also used. Although the behavior of stacks in a laboratory environment is different from the behavior of cell stacks in a process system, a pressurized stack test rig was used for model parameterization. 4.2.1 Investigation of a reactor with multiple stacks in process system context A large part of the work leading to this manuscript was the construction of a 30 kW pressurized reactor test rig with fuel gas recirculation at reaction temperature (Article I), which was the outcome of several research projects carried out by the several co- members of authors’ research group. The test rig (figure 4.2) was designed and built with the goal of understanding the operation of a large reactor in a process system. In particular, it was engineered to study a hybrid power plant based on an SOFC reactor and a micro gas turbine (MGT) by emulating the components of the MGT. This led to boundary conditions of the SOFC reactor that complicated the investigation of the SOFC reactor itself, but also revealed many challenges that were addressed in this work. The flow sheet of the SOFC/MGT hybrid power plant is shown in Figure 4.3 a). It consists of a Brayton cycle with recuperation, in which the combustion chamber is par- tially replaced by the SOFC reactor. Therefore, the SOFC reactor must be operated in a pressure vessel. The desulfurized fuel is mixed with a part of the SOFC reactor off- gas using a high-temperature recirculation blower in order to provide heat and steam for the reforming of hydrocarbons. To reduce the risk of damage to the SOFC reactor due to pressure fluctuations by the MGT and to obtain more degrees of freedom for the SOFC reactor investigation, an SOFC system with components for the emulation of the gas turbine was designed and built (Figure 4.3 b)). The compressor was replaced by an electrical heater (COMP SIM) and the off-gases of the gas turbine were emulated by a 34 4.2 Experimental investigations Figure 4.2: Pressurized reactor test rig built and operated at German Aerospace Center (DLR) in Stuttgart. gas burner (BURN). The combustion chamber and turbine were replaced by a heat ex- changer, pressure control valve and oxidation catalyst reactor. Due to problems caused by the pressure difference between the vessel and the air chamber of the SOC, as well as challenges with gas safety, further modifications were made after commissioning. An additional air flow was introduced to purge the vessel and a second chimney was built (Figure 4.3 c)). Pressure vessel REC SOFC SENSORS RECUAir Fuel COMP SIM CH II Fuel Air REF DESUL OXI CAT BURN CH I AC DC AC DC Pressure vessel REC SOFC SENSORS RECU Fuel Fuel REF DESUL BURN Air CH I COMP GEN TURBINE AC DC AC DC Air b) before modification c) after modificationa) Figure 4.3: Flow sheet of the SOFC/MGT hybrid power plant (a) and the derived pressurized reactor test rig before (b) and after (c) the modifications (adapted from [11] and [12]) 35 4 Methodology During the experiments, the system was controlled by a programmable logic controller (PLC) with the controllers presented in Article I. 4.2.2 Provision of parameterization and validation data In addition to the test facility described above, experimental results of other sources were used to parameterize and validate the numerical models. The test rigs Horst and Galactica operated by co-members of the author’s research group provided important data and insights. In addition, experimental data were available from project partners. Figure 4.4: DLR’s test rigs Horst (a) and Galactica (b), which are described in the next subsec- tions. Horst tests rig and impedance spectroscopy Even though larger reactors with multiple stacks were investigated in this work, the experimental investigation of single stacks supported the model parameterization. In fact, it was necessary for the parameterization of the loss mechanisms, since commercial reactors with multiple stacks have few sensors. Thus, results of the pressurized DLR test rig Horst were used (figure 4.4 a). 36 4.2 Experimental investigations The test rig includes a pressure vessel with an integrated furnace that can be operated at up to 8 bar and 950 °C. SOC stacks can be tested in the power range around 1000 W in electrolysis and 250 W in fuel cell operation. Besides electrical preheaters and the furnace heaters, the vessel provides space for a SOC stacks. The test rig has a gas dosing station with carbon monoxide, carbon dioxide, methane, hydrogen, nitrogen and a pulsation-free evaporator system. The pressure is regulated by a valves and pressure differences are avoided by equalizing tanks [97]. For parameterization and validation of the models, the test rig offers temperature and cell voltage sensors as well as a gas analyzer and an electrochemical impedance spec- trometer (EIS). The infrared gas analyzer provides the ability to parameterize the cell internal chemical reactions ((reverse-)water-gas shift and methanation/reforming). Us- ing the EIS investigations, the ohmic, activation, diffusion resistances were determined by varying the experimental operating parameters [97] (see also section 4.1.3). Galactica For the validation of the reactor models with multiple stacks, another test facility is operated by DLR. The Galactica test rig (figure 4.4 b) offers the possibility to flexibly test SOC reactor with multiple stacks in fuel cell (up to 40 kW) and in electrolysis mode (up to 120 kW) [77]. The modelling methodology in this thesis uses experiments for two reasons. First, ex- periments are conducted to identify the challenges and the unknown effects and mecha- nisms. Then, the modelling depth is determined to answer the open questions. Second, the experiments are conducted to parameterize and validate the models. Both reasons can be addressed by the Galactica test rig, making the input from the test rig necessary for this work. Due to the flexible experimental setup, it is possible to operate the SOC reactors under various operation conditions. The test rig consists of a fuel dosing unit, providing carbon monoxide, carbon dioxide, methane, hydrogen, nitrogen and H2O from a steam generator. The oxygen electrode side can be purged with air via two blowers. Different sets of power electronics are available covering a wide range of currents and voltages. An infrared gas analyzer, an electrochemical impedance spectrometer and a wide range of temperature as well as pressure measurements complete the equipment. 37 4 Methodology Data from project partners In addition to DLR’s test facility, experimental data were also available from partners with whom scientific research projects were carried out. However, the data from project partners was often only used as a reference, since test rig specifications and other infor- mation were missing. Concerning data from project partners, most important were results of the cell stack test, which were used for the validation of the simplified stack model (Article II). These were provided by the stack manufacturer Sunfire GmbH. The insulation of the Sunfire test rig is similar to that of reactor with multiple stacks, making the test rig and results positively distinctive from stack test rigs with furnaces. Therefore, the effects of heat losses were considered in the validation. Unfortunately, the test rig had an air leakage upstream of the stack inlet, which required an estimation of the air mass flow rate through the stack. A temperature controller as used in the experiment was implemented within the validation model to match the experiments (Article II). 38 5 Publications 5.1 Differentiation of the author’s work from the co-authors or previous work The author of this thesis was the main researcher and main author of the following three publications. This includes conceptualization, methodology, visualization and writing. Article I is based on experiments performed using the pressurized reactor test rig, which was built by the author’s research group. The author was partly involved in the procure- ment of the components and construction of the facility. Furthermore, the author was significantly involved in the development of the control system. The author led and supervised the operation of the facility as well as the planning and execution of the experiments. The scientific evaluation and analysis of the experimental results leading to this article were carried out exclusively by the author. The modelling and simulation work leading to Article II and Article III was performed by the author, except for the modelling of the cell models and the 10-cell short stacks, which were the result of previous research (see section 4.1.3 for details). The develop- ment of the control and operating strategies and the analysis of the simulation results was carried out exclusively by the author. Validation experiments on the Galactica test rig were performed by colleagues from the research group after consultation with the author. 39 5 Publications 5.2 Article I Article I: M. Tomberg, M. P. Heddrich, M. Metten, S. A. Ansar, and K. A. Friedrich. “Operation of a Solid Oxide Fuel Cell Reactor with Multiple Stacks in a Pressured System with Fuel Gas Recirculation”. In: Energy Technology 10(4), 2101075 (2022). DOI: 10.1002/ente.202101075 40 https://doi.org/10.1002/ente.202101075 Operation of a Solid Oxide Fuel Cell Reactor with Multiple Stacks in a Pressured System with Fuel Gas Recirculation Marius Tomberg,* Marc Philipp Heddrich, Matthias Metten, Syed Asif Ansar, and Kaspar Andreas Friedrich 1. Introduction Solid oxide cells (SOCs) are flexible and efficient energy conver- sion devices. In solid oxide fuel cell (SOFC) mode, they can effi- ciently convert chemical into electrical energy. Due to the high operating temperature and the stable catalysts, the cells can cope with various fuels based on hydrogen and reformed hydrocarbons. However, to play a signifi- cant role in the future energy system, SOFCs need t