Amalie Bisgaard Møller

• Møller, Amalie Bisgaard; Holm, Thomas; Lie, Bernt; Ulleberg, Øystein; Llamas-Jansa, Isabel & Mirlekar, Gaurav (2025). Modeling the Dynamics of the Proton Exchange Membrane Water Electrolyzer Pressure. IFAC-PapersOnLine. ISSN 2405-8971. 59(29), s. 84–89. doi: 10.1016/j.ifacol.2025.12.186. Fulltekst i vitenarkiv Vis sammendrag

  To simulate the startup, shutdown, standby, and variable operation of Proton Exchange Membrane Water Electrolyzer (PEMWE) systems, it is necessary to model the dynamics of the operating pressure on the hydrogen side (cathode). A simple mathematical model of the cathode pressure has been developed, parametrized, and validated against dynamic high pressure data from the operation of a small-scale PEMWE system laboratory at the Institute for Energy Technology (IFE). The developed model described the actual system dynamics at different current inputs and pressure setpoints. Various equations of state (EoS) for hydrogen gas were compared and the results suggested that hydrogen should be modeled as a real gas at high pressure. The methodology presented in this work is applicable to other PEMWE systems, and the procedure for model parametrization is described in detail.

• Lunde, Vilde Sturtzel; Møller, Amalie Bisgaard; Eggert, Bruno Guilherme Fischer; Döring, Allan M.; Grivel, Jean-Claude & Bjørk, Rasmus [Vis alle 12 forfattere av denne artikkelen] (2025). Machine learning guided discovery and experimental validation of light rare earth Laves phases for magnetocaloric hydrogen liquefaction. Acta Materialia. ISSN 1359-6454. 297. doi: 10.1016/j.actamat.2025.121282. Fulltekst i vitenarkiv Vis sammendrag

  In the search for new compounds suited for magnetocaloric hydrogen liquefaction, we have combined machine learning (ML) predictions with experiments. Two series of cubic C15 Laves phases, and (0 ≤ x ≤ 1), were selected based on Curie temperature (Tc) predictions of three ML models. A superior model performance was obtained when the dataset was developed for a specific crystal class, despite it being limited to only containing 265 compounds. The random forest regression, gradient boosting regression, and neural network ML models yielded mean absolute errors of 14, 18, and 20 K, respectively. These errors are low compared to previously reported Tc-predictions using ML. The models are powerful tools for the initial screening of new candidate materials, and the accuracy for light rare earth based cubic Laves compounds will improve as the dataset increases. The selected compounds based on the ML screening were synthesized by arc melting and characterized for potential magnetocaloric hydrogen liquefaction applications. The compositions showed magnetic ordering between 20 and 36 K, thus in the lower temperature region relevant for magnetocaloric hydrogen liquefaction. Moreover, they exhibited magnetic entropy change ranging from 6.0 to 7.2 Jkg−1K−1 and adiabatic temperature change from 2.2 to 2.6 K, for a field of 5 T. In contrast, the Ce-containing compounds exhibited a mixture of magnetic Ce and non-magnetic Ce, yielding low magnetization, thus low magnetocaloric effect, and magnetic ordering temperatures below 20 K.

Se alle arbeider i NVA

• Lunde, Vilde Sturtzel; Møller, Amalie Bisgaard; Eggert, Bruno Guilherme Fischer; Döring, Allan; Grivel, Jean-Claude & Bjørk, Rasmus [Vis alle 12 forfattere av denne artikkelen] (2025). Machine Learning Guided Discovery of Light Rare Earth Laves Phases for Magnetocaloric Hydrogen Liquefaction. Fulltekst i vitenarkiv
• Lunde, Vilde Sturtzel; Møller, Amalie Bisgaard; Eggert, Bruno Guilherme Fischer; Döring, Allan; Straßheim, Marc & Grivel, Jean-Claude [Vis alle 15 forfattere av denne artikkelen] (2025). Machine Learning Guided Discovery Light Rare Earth Laves Phases for Magnetocaloric Hydrogen Liquefaction. Fulltekst i vitenarkiv
• Ulleberg, Øystein; Llamas-Jansa, Isabel & Møller, Amalie Bisgaard (2025). Modeling of Industrial Scale PEM Water Electrolysis Systems. doi: https:/ecs.confex.com/ecs/247/meetingapp.cgi/Paper/202692. Fulltekst i vitenarkiv Vis sammendrag

  Background The first experimental activities at Institute for Energy Technology (IFE) on high-pressure (16 bar) proton exchange membrane (PEM water) electrolysis systems based on variable renewable power was performed in 2004-2008 [1]. In 2016 IFE started to develop a national research infrastructure for testing of small-scale (33 kW) pressurized (up to 200 bar) low temperature PEM water electrolyser stacks [2]. The performance of the high-pressure test rig and a prototype PEM water electrolyser stack operating at a differential pressure of 180 bar was demonstrated and documented in 2023 [3]. Modeling activities and different types of techno-economic analyses of high-pressure PEM water electrolysis have been performed in parallel to the experimental activities in several projects at IFE [4]. In 2023 there was established a new large national project in Norway on “Renewable Energy Hydrogen Systems based on PV, Wind, and Water Electrolysis” (REHSYS, 2023-2026), including a PhD at the University of Oslo (UiO) on advanced modeling and controls of industrial scale PEM water electrolyzer systems. Objectives The main objective with the REHSYS-project is to study and optimize the design and operation of industrial scale water electrolysis systems based on wind, photovoltaic (PV) solar, and PEM technology, with the aim to support technology developments required to reach the cost target of 1-2 USD per kg for renewable energy (RE) based hydrogen (H2). To achieve this the project seeks to maximize the efficiency, lifespan, and economy of RE/H2 systems by exploring system designs and operational schemes for specific locations having different weather conditions, scales, grid constraints, and prospective hydrogen off-takers. An overview of the research tasks in REHSYS is provided in the figure below; the PEM water electrolysis modeling activities are being carried out in work package 2 (WP2). Methodology To optimize the operation of RE-based PEM water electrolysis systems it is essential to develop control strategies that can mitigate the unfavorable effects of dynamic operation and safeguard the PEMWE stacks and systems. There have been performed several theoretical studies on energy management and optimization of water electrolyzer systems, but there is a lack of actual implementation and testing of model-based control strategies. Hence, there is clearly a need to better understand the technical constraints and requirements for more advanced instrumentation. The REHSYS-project will address this challenge by developing and implementing advanced system control strategies which includes PV and wind nowcasting to minimize parasitic losses, degradation, and costs. The hypothesis is that with access to precise RE-power nowcasts the system operator can make better and more qualified decisions, e.g., when to switch between hot and cold standby mode. Preliminary Results A summary of the recent PEM water electrolysis modeling efforts at IFE and some preliminary results from the modeling work in a MSc at UiO [5] and the newly started PhD of the REHSYS-project will be presented. The most recent modeling work at IFE has been performed using EES, an Engineering Equation Solver program, while the work at UiO has been performed in Matlab Simulink. The current plan in the REHSYS PhD is to develop open-source code in the Julia programming language. Preliminary results from the first modeling of a one-dimensional mechanistic PEM water electrolyser stack model and key component models for the electrolyzer balance of plants (BoP) sub-systems will be presented. References 1. Miland, H. and Ø. Ulleberg (2012) Testing of a small-scale stand-alone power system based on solar energy and hydrogen Solar Energy, 86(1): p. 666-680 2. Hancke, R., Ø. Ulleberg, R. Skattum, V. Torp, and J.E. Jensen (2019) High Differential Pressure PEMWE System Laboratory, in EFCF 2019 Low-Temperature Fuel Cells, Electrolysers & H2 Processing, Lucerne 3. Hancke, R., P. Buljo, T. Holm, and Ø. Ulleberg (2024) High-pressure PEM water electrolyser performance up to 180 bar differential pressure, Journal of Power Sources, Volume 601, 1 May, 234271, 4. Hancke, R., T. Holm, and Ø. UIleberg (2022) The case for high-pressure PEM water electrolysis, Energy Conversion and Management, 261 (2022) 115642 5. Møller, A. (2024) Modeling of the thermal and dynamic performance of a small-scale PEM water electrolyzer system, Master thesis, UiO

• Møller, Amalie Bisgaard (2024). Modelling of the Thermal and Dynamic Pperformance of a Small-Scale PEM Water Electrolyser System. Universitetet i Oslo. Fulltekst i vitenarkiv Vis sammendrag

  Hydrogen produced from water electrolysis plays an important role in the transition towards renewable energy, as it can serve as storage for abundant electrical energy. The varying nature of renewable energy is a challenge in the integration with water electrolysis systems. It is therefore important to understand the effect of dynamic operation on water electrolysis systems. Dynamic modelling enables such studies and allows for system analysis using simulations. This thesis focuses on dynamic modelling of PEM water electrolysis (PEMWE) systems. A semi-empirical model of a small-scale PEMWE system was developed in this work. The model includes a steady state electrochemical model which computes the cell potential and a dynamic thermal model to describe the system temperature. In the electrochemical model, the membrane resistance and slow reaction kinetics at the anode were found to have the largest effect on the cell potential. Parameters related to these overpotentials were found from a fit to an experimental polarization curve. The thermal model includes a submodel of a heat exchanger and a heater, including a control system that adjusts the amount of cooling and heating. Constants used in the thermal model, such as the thermal capacity and resistance were found from experimental data. The values of both the thermal and electrochemical model parameters were comparable to values reported in literature. Implementation of the model was done in Matlab/Simulink where it was used to perform simulations given different time dependent input data. The model was validated with experimental data and was able to simulate the cell potential accurately with mean absolute errors as low as 0.0032 V. Simulations of the system temperature described the behavior observed in the real system with the lowest mean absolute error of 0.022°C. The model was applied in different scenarios to evaluate the performance under start-up, shutdown and dynamic operation. Recommendations were given on how to calibrate the model to other PEMWE systems and how to scale it up to describe the behavior of larger systems.

Se alle arbeider i NVA