PEM Electrolyser Development Supported by Multiphysics System Simulation

EFCF 2025: Low-Temp. Fuel Cells, Electrolysers & H2 Processing 1 – 4 July 2025, Lucerne Switzerland https://doi.org/10.5281/zenodo.17476276 B0806 Page 1/11 B0806 PEM Electrolyser Development Supported by Multiphysics System Simulation Aleš Cvikl* (1), Matic Belak Vivod (1), Christoph Pötsch (2), Ivo Prah (1), Gašper Zajc (1), Rok Kaufman (1) (1) AVL-AST d.o.o., Maribor/Slovenia; (2) AVL List GmbH, Graz/Austria; *Contact corresponding authors: www.EFCF.com/ContactRequest Abstract The increasing focus on regulations and initiatives for green energy production highlights the growing significance of electrolysers in hydrogen production. Among the various existing and emerging technologies, Proton Exchange Membrane (PEM) electrolysers are distinguished by their high efficiency and flexibility, particularly in handling fast transient operations. However, addressing all critical design, integration, and operational challenges during development is complex. The application of advanced multi-physics system simulation tools is a promising measure to reduce development time and cost. These tools facilitate virtual exploration of key aspects of the system and components design, their operation and control strategy development, while frontloading tasks to early stages of the development and therefore reducing costly physical prototypes. This work presents advancements in AVL’s simulation methodology and capabilities for PEM electrolyser system development, covering system architecture, component selection, integration, and operational strategies at given boundary conditions. Key aspects of system design and optimization, such as aging, safety, and component limitations across all development stages are presented for a PEM electrolyser with hydrogen production capacity of 1 kg/h and 99.999 % purity, injected into a natural gas pipeline at 70 bar. EFCF 2025: Low-Temp. Fuel Cells, Electrolysers & H2 Processing 1 – 4 July 2025, Lucerne Switzerland https://doi.org/10.5281/zenodo.17476276 B0806 Page 2/11 1. Introduction The current global transition to renewable energy sources has led to remarkable advances in technologies aimed at reducing carbon emissions and achieving a sustainable, carbon‐ free future [1]. Governments, industries, and research institutions worldwide are investing substantial resources into the development of innovative energy solutions. In this context, hydrogen production via water electrolysis has emerged as a pivotal technology, offering a versatile and clean energy carrier capable of storing intermittent renewable energy and facilitating a low-carbon economy. Over the past year, global electrolysis capacity has grown by approximately 60%, reaching 1.75 GW of installed water-splitting power (Figure 1). Alkaline electrolysers continue to dominate the market, accounting for the majority of this capacity, especially in China. However, PEM electrolysers are gaining traction, particularly in Europe and North America, where they now represent a significant share [2]. Figure 1: Adapted from [2], global cumulative installed electrolysis capacity with electrolyser technology share in different locations. Alkaline electrolysers represent a mature, cost-effective technology, employing inexpensive, readily available catalysts (e.g., nickel) and robust materials that tolerate high impurity levels in feed water, which makes them well-suited for large-scale industrial hydrogen production [3,4]. However, compared to the PEM systems, alkaline electrolysers typically operate at lower electrical efficiencies and exhibit slower dynamic response, leading to longer startup and shutdown times that make an integration with variable renewable energy sources extremely challenging. The use of a liquid caustic electrolyte requires careful corrosion management and safety measures, increasing maintenance demands and operational risk [3,4]. Additionally, thick diaphragms, while essential for gas separation, introduce higher ohmic resistance and can exacerbate hydrogen–oxygen crossover, which limits gas purity and partial-load performance [5]. PEM electrolysers offer fast response, high power density, and compact, low-temperature operation, making them ideal for integration with renewables [6]. Their high current density, precise control, and modular design support both scalable industrial use and distributed energy systems. Despite the promising capabilities of PEM systems, the design, testing, and optimization of electrolyser configurations remain EFCF 2025: Low-Temp. Fuel Cells, Electrolysers & H2 Processing 1 – 4 July 2025, Lucerne Switzerland https://doi.org/10.5281/zenodo.17476276 B0806 Page 3/11 challenging and resource-intensive efforts, particularly when scaling up for industrial applications. Advanced simulation tools have emerged as a critical solution to these challenges, enabling researchers and engineers to model complex electrochemical and thermodynamic interactions accurately. In this work, we demonstrate how AVL is utilizing system simulation tool in the development process of a design, integration and validation of a PEM electrolyser system. The simulation framework was validated against experimental data reported in the literature, ensuring that the model reliably reflects real-world performance and captures the dynamics of electrochemical processes. This paper will focus in greater detail on two specific use cases: aging and transient dynamics, such as warm-up transitions, load-point changes and hot stand-by operation (marked red in Figure 2). Figure 2: Illustrated use cases of virtual development for PEMEL systems (highlighted in red are briefly discussed in the paper). 2. Simulation assisted development of PEM Electrolysers The PEM electrolyser system model development and simulations were carried out in the modular dynamic system platform AVL CRUISETM M [7]. The overall system is very similar to the one reported by Stansberry and Brouwer [8]. The system is designed to produce hydrogen for injection into a natural gas pipeline at a pressure of 70 bar, with a maximum production rate of 1 kg/h and target purity of 99.999 %. The PEM electrolyser stack consists of 66 cells with an active area of 200 cm2. 2.1 PEM Electrolysis system model In Figure 3 the system architecture is illustrated with flow directions of different species in the stack. On the anode (oxygen) side, water is introduced into the system from a Water tank and directed into the O ₂ /water separator. The water inflow is precisely regulated by Valve 5, which maintains a liquid volume fraction of 0.5 within the separator. A Water pump maintains continuous flow through the system, initially directing water toward a deionization filter. The flow is dynamically split between the deionization path and a bypass line by modulating Valve 1 and Valve 2, with adjustments based on real-time conductivity sensor to achieve target water purity. During the warm-up phase, the water stream is routed through a Heater to accelerate system heat-up to the target temperature. After reaching a defined temperature threshold, electrolysis begins with electrically assisted warm-up. Heating through electrical Heater is discontinued when thermal equilibrium is reached. Further on, Valve 3 and Valve 4 are dynamically adjusted to maintain the desired inlet temperature. The Pump speed regulation controls the target temperature differential across the Stack. Oxygen EFCF 2025: Low-Temp. Fuel Cells, Electrolysers & H2 Processing 1 – 4 July 2025, Lucerne Switzerland https://doi.org/10.5281/zenodo.17476276 B0806 Page 4/11 generated during electrolysis is initially collected in the O ₂ /water separator before being released into the environment at a controlled pressure of 1.1 bar, consistent with the conditions reported in reference [9]. For high-pressure operation, Throttle 1 is controlled to adjust the oxygen release. On the cathode (hydrogen) side, hydrogen is first separated from liquid water in H ₂ /water separator I. The resulting gas mixture, comprising of hydrogen, water vapour, and traces of oxygen is passed through a Deoxo catalyst to remove residual oxygen. Before this step, it is preheated to prevent condensation and the resulting degradation of the catalyst layer. Further purification process continues in Condenser, where part of water vapor is condensed (the amount depends on the pressure and temperature conditions) and collected in H ₂ /water separator II, ensuring a hydrogen purity of approximately 99.9 %. Condensed water is recirculated back to H ₂ /water separator I through Valve 7. To reach the target hydrogen purity of 99.999 %, additional purification is carried out via Pressure swing adsorption (PSA) system. The purified hydrogen is then compressed and stored at 70 bar, making it suitable for high-efficiency energy applications. During system start-up, a nitrogen purge is conducted to eliminate residual oxygen from the cathode and hydrogen from the anode, ensuring a controlled and inert environment before electrolysis begins. Figure 3: System architecture of PEM electrolyser with flow chart of different media streams. 2.2 PEM stack and system model validation The system model is validated against the data reported by Stansberry and Brouwer [8]. In Figure 4 validation examples are presented whereby simulation results are compared to the reference data. In all cases, a strong agreement between the simulated and referenced results is observed, with minor deviations arising from uncertainties in cell and stack geometries, the ageing status of the components, and variations in operating conditions. Additional factors, such as potential measurement inaccuracies and fluctuations in catalyst activity, may have contributed to these discrepancies. EFCF 2025: Low-Temp. Fuel Cells, Electrolysers & H2 Processing 1 – 4 July 2025, Lucerne Switzerland https://doi.org/10.5281/zenodo.17476276 B0806 Page 5/11 Figure 4:Compariosn of simulation and reference data for: (a) polarization curves at different temperatures; (b) Faraday efficiency; (c) specific system consumption and (d) specific stack consumption. 3. Results and discussion 3.1 Stack aging PEM electrolysers degrade over time due to a combination of chemical, mechanical, and thermal stressors that reduce efficiency and shorten operational lifespan. A central factor is the degradation of the membrane electrode assembly (MEA), particularly the protonconducting membrane. Operational conditions, including high current densities and fluctuating power inputs, exacerbate these degradation processes, leading to uneven current distribution and localised hotspots that further deteriorate the PEM electrolyser assembly [9]. As PEM electrolysers age, degradation of the membrane, particularly thinning and the formation of pinholes, can lead to increased hydrogen crossover from the cathode to the anode. This phenomenon not only reduces system efficiency as demonstrated in Figure 5a but also poses significant safety risks. When hydrogen permeates into the oxygen-rich anode compartment, it can form combustible mixtures. If the hydrogen concentration approaches or exceeds low explosive limit (LEL), typically close to 4 % volume fraction of hydrogen, the mixture becomes explosive [10–12]. Such risks are especially pronounced during low-load or idle operations, where lower oxygen generation reduces dilution of the hydrogen crossover, increasing the likelihood of reaching hazardous concentrations. To mitigate these dangers, operating strategy of the electrolyser must ensure system operation below 2 % of hydrogen concentration (50 % of the LEL). This can have significant influence, especially at aged stack, where crossover of oxygen and especially hydrogen can increase significantly. In Figure 5b, the constraints on the lower load limit between the Beginning of Life (BoL) and End of Life (EoL) stack can be seen. In our study, this limitation was considered not only in the development of the control strategy, but also in the calculation of EFCF 2025: Low-Temp. Fuel Cells, Electrolysers & H2 Processing 1 – 4 July 2025, Lucerne Switzerland https://doi.org/10.5281/zenodo.17476276 B0806 Page 6/11 the Levelized Cost of Hydrogen (LCOH), as it significantly reduced operational flexibility at EoL. Figure 5: (a) System efficiency (Lower Heating Value); (b) Hydrogen concentration on the anode at BoL and EoL. 3.2 Transient start-up process Developing and calibrating control functions within a virtual environment is a crucial step in accelerating early-stage design, reducing the need for expensive test bed operations, and safeguarding the physical system from potential damage. While the control strategies for steady-state operations are relatively simple, the implementation of transient control, particularly for processes such as warm-up, remains challenging due to the extended timescales involved. Virtual simulations enable comprehensive testing and refinement of these control algorithms, facilitating a more seamless transition to real-world operation while improving system stability and overall efficiency. The warm-up process employed in this study consists of five steps (modes) as presented in Figure 6. Mode 1: both anode and cathode are purged with nitrogen gas, which ensures that there is no residual oxygen on the cathode side and hydrogen on the anode side. This is critical to ensure both safety and system performance as the presence of residual gases can lead to uncontrolled side reactions, efficiency losses, and most critically, the formation of an explosive H2/O2 mixture within the system. Mode 2 involves establishing a water flow system that includes a pump and a deioniser. Deionisation is a necessary step to remove potential ionic contaminants that may originate from sources such as pipes, tanks, or other system components. Mode 3: the heater is activated to gradually increase the water temperature on the anode side, considering temperature gradients to ensure controlled heating. In mode 4, the electrically assisted warm-up phase begins with part-load operation. As electrolysis begins, the exothermic nature of the reaction leads to a rapid increase in cathode temperature. Mode 5: once the system reaches the targeted operating temperature, the heater is deactivated, and the current is gradually increased to full load. At this stage, thermal stability is maintained through active cooling, ensuring precise temperature regulation of the water stream temperature. Full-load hydrogen production is reached almost immediately after switching to Mode 5 (approximately 7 minutes), while thermodynamic steady-state stability is achieved after 14 minutes, marking the end of the warm-up sequence. EFCF 2025: Low-Temp. Fuel Cells, Electrolysers & H2 Processing 1 – 4 July 2025, Lucerne Switzerland https://doi.org/10.5281/zenodo.17476276 B0806 Page 7/11 Figure 6: Selected monitored temperatures as a function of time during each mode of the warm-up process of the PEM electrolyser. 3.3 Load change Load change operations in PEM electrolysers are essential for assessing system stability, efficiency, and durability under transient conditions, particularly in the context of renewable energy-driven hydrogen production. As power inputs fluctuate due to the variability of wind and solar energy, understanding the electrolyser’s dynamic response is crucial for maintaining performance and preventing operational instabilities. In this study, the operational load of the PEM electrolyser was systematically reduced to lower load limit of 25 % for a duration of 250 seconds before returning to full-load conditions. Upon load reduction, the rate of electrolysis decreased proportionally, leading to a significant decline in heat generation within the stack. This reduction in thermal output resulted in a decrease in the anode outlet temperature from 55 °C to approximately 51 °C (Figure 7a). Notably, the stack inlet temperature on the anode side was consistently maintained at 50 °C through precise valve regulation, as illustrated in Figure 7b. The difference between stack inlet and outlet temperature on the anode is also indirectly controlled by the water flow through the stack (Figure 7c). At full-load operation, the water pump operates at rated speed, which maintains the stack temperature difference of 5 °C. Reducing the water pump speed decreases the flow rate through the stack, thereby increasing the residence time of water within the stack. This prolonged exposure allows the water to absorb more heat generated during operation, resulting in a higher outlet temperature. Reducing the load on PEM electrolyser decreases the current passing through the system, leading to a significant reduction in heat generation within the stack (Figure 7a). EFCF 2025: Low-Temp. Fuel Cells, Electrolysers & H2 Processing 1 – 4 July 2025, Lucerne Switzerland https://doi.org/10.5281/zenodo.17476276 B0806 Page 8/11 Figure 7: (a) Stack in and stack out temperatures; (b) valve opening ratios; (c) water pump speed with temperature difference on the anode and (d) voltage, all plotted as a function of time in the load change simulation experiment. This diminished thermal output reduces the necessity for active cooling, allowing for a more passive thermal management approach. However, maintaining a minimum water flow level specified by stack supplier results in a stack temperature difference of approximately 1 °C. Concurrently, the decrease in current correlates with an increase in cell voltage, rising from about 1.7 V under 25 % load to nearly 2.1 V at full load conditions (Figure 7d). 3.4 Hot stand-by process Hot standby is one of the crucial operating modes for PEM electrolysers, especially when trying to optimize LCOH together with variable renewable energy sources. In this mode, the electrolyser maintains its operating temperature and pressure without hydrogen production, allowing for a rapid return to full operation within seconds. This state of readiness minimizes thermal cycling, which can degrade system components and reduce lifespan. Moreover, it reduces energy consumption compared to full shutdowns by avoiding the substantial energy required for reheating, which is a valuable advantage when operation resumes after short interruptions. As presented in Figure 8, the hot standby experiment begins by bringing the PEM stack to its full‐load setpoint, with the current held constant to elevate the cell to 50 °C. At t = 10 min, the current is reduced to zero, immediately halting the electrolysis and preventing heating of the feed water. As conductive and convective heat losses then drive the stack outlet temperature downward, the auxiliary heater is turned on and the cooling valve (Valve 3 in Figure 3) is gradually closed. Heater power rapidly rises and converges to approximately 0.7 kW, exactly offsetting the system’s intrinsic heat losses and maintaining the 50 °C setpoint. EFCF 2025: Low-Temp. Fuel Cells, Electrolysers & H2 Processing 1 – 4 July 2025, Lucerne Switzerland https://doi.org/10.5281/zenodo.17476276 B0806 Page 9/11 Figure 8: Stack temperatures and heater power as a function of time during the hot standby operation. The first six minutes of the simulation were omitted for clarity. The current is plotted as negative to indicate that the stack is operating in electrolysis mode (power consumption) rather than in fuel-cell mode (power generation). By avoiding full shutdown, the hot‐standby strategy minimizes thermal cycling stresses on the membrane and catalyst layers, preserves component durability, and enables an almost instantaneous return to electrolysis once the current is reapplied. 4. Conclusions In summary, this work demonstrates that advanced dynamic modelling of a PEM electrolyser not only replicates key experimental performance metrics with high accuracy, but also offers valuable insights into operational flexibility, control strategy design, and long-term durability considerations. Through rigorous validation against literature data, the developed models prove suitable for a wide range of use cases across different stages of the development cycle. The incorporation of aging effects, such as declining stack performance and increased hydrogen crossover, underscores the critical safety and performance trade-offs between operating current density and component longevity important for LCOH consideration. By capturing the full spectrum of transient behaviours, the model provides a robust foundation for exploring dynamic phenomena that are often overlooked in steady-state analyses. This includes scenarios such as warm-up phases, load transitions, and hotstandby operation, which were demonstrated as representative transient cases. Identifying optimal operating regimes that balance efficiency, durability, and safety, the presented framework serves as a powerful methodology to accelerate the design, scale-up, and costeffective deployment of reliable PEM electrolysis systems for integration into renewablebased hydrogen infrastructure. References [1] J. A. Turner, “A Realizable Renewable Energy Future,” Science, 285, 687-689, 1999. [2] M. &. Company, “Hydrogen Insights September 2024,” Hydrogen Council, 2024. [3] https://senzahydrogen.com/pem-hydrogen-generator-vs-alkaline-hydrogengenerator/. [Accessed 23 4 2025]. [4] https://www.hydrogennewsletter.com/untitled-13/. [Accessed on 23 4 2025]. [5] C. Karacan, F. P. Lohmann-Richters, G. P. Keeley, F. Scheepers, M. Shviro, M. Müller, M. Carma and D. Stolten, “Challenges and important considerations when benchmarking single-cell alkaline electrolyzers,” Int. J. Hydrogen Energy, 47, 42944303, 2022. [6] M. Carmo, L. F. Fritz, J. Mergel and D. Stolten, “A comprehensive review on PEM water electrolysis,” Int. J. Hydrogen Energy, 38, 4901-4934, 2013.