Optimization of water management in PEM fuel cells: experimental HFR analysis under variable humidity conditions

Abstract

Reliable water management remains a critical challenge for the operation of proton exchange membrane fuel cells (PEMFCs), particularly under dynamic conditions such as those encountered in aviation applications. Conventional relative humidity (rH) sensors frequently fail due to condensation, which can lead to erroneous signals caused by water droplets forming on the sensor tips, thereby limiting their suitability for robust humidity monitoring and control. This work therefore investigates electrochemical impedance spectroscopy (EIS), and specifically the high-frequency resistance (HFR) at 1 kHz, as a potential alternative diagnostic parameter for assessing the membrane hydration state in PEMFC stacks in combination with the cell voltage. Using experimental data acquired from a PowerCell S3 stack, the membrane thickness was estimated as a function of current density based on measured HFR values. At a current density of 1 A/cm2, the calculated membrane thickness of 27.62 µm shows good agreement with literature values, thereby validating the underlying approach. Subsequently, two complementary models were developed to predict the hydration state of the stack: (i) an rH-based model estimating membrane humidity from temperature, membrane thickness, and HFR, and (ii) a parameter-based model constructed using a statistical Design of Experiments (DoE) approach, linking cell voltage and HFR behaviour to the operating conditions. Model validation demonstrated that the combined evaluation of cell voltage and HFR enables reliable differentiation between flooding, normal hydration, and dry-out conditions. Furthermore, measurements performed with a segmented measuring plate (SMP) provided qualitative verification of the model predictions through characteristic spatial current- and temperature-distribution patterns. A sensitivity analysis quantified the influence of key operating parameters on both cell voltage and HFR, identifying temperature and pressure as the dominant factors. Finally, both models were applied to a dynamic, humidifier-less flight profile, demonstrating the feasibility of sensor-less hydration-state estimation and providing a basis for automated onboard diagnostic algorithms. Overall, this thesis establishes a foundation for robust, sensor-less water-management diagnostics in PEMFC systems by combining electrochemical impedance measurements with model-based interpretation. Future work should focus on improving measurement precision, incorporating accurate membrane-thickness data, and implementing dynamic control algorithms to enable reliable real-time operation in aerospace applications.