Modeling gas crossover in alkaline water electrolysers

Abstract

Climate change due to the extensive use of fossil fuels has led to the deployment of alternative green ones, such as hydrogen. Green hydrogen is produced by renewable electricity and is CO2-free. This thesis focuses on the production of hydrogen by implementing alkaline water electrolysis as the core technology. 
Due to the intermittency of renewable sources, alkaline water electrolysers are forced to operate in their part-load range. The cathodic hydrogen species that remains dissolved in the liquid electrolyte can end up to the anodic compartment, and hence lower the purity of the produced gaseous oxygen. This phenomenon is prominent in the part-load range and is called gas crossover. When the concentration of hydrogen in oxygen reaches the Lower Explosive Limit which is 4 vol%, spontaneous combustion can occur. Therefore, the electrolyser is forced to shut down for safety reasons.
This thesis focuses on understanding the mass transfer mechanisms of gas crossover in alkaline water electrolysis, in the part-load range. A literature study has been conducted in which the gas crossover mechanisms are thoroughly analyzed. The mitigation of gas crossover can lead the operation to lower current density ranges. From the mitigation strategies, a focus is given on the “dynamic switching of the electrolyte cycles". The dynamic switching of the electrolyte cycles is based on the periodic changeover of the operative electrolyte cycles between the partly-separated and the mixed mode. The anodic hydrogen content acquires a sinusoidal response, where the average value is less than the impurity in traditional operation. 
The gas crossover steady-state and dynamic models are mathematically derived and developed in Python. The models consider the mechanisms of gas crossover through the diaphragm and the electrolyte mixing. Therefore, the anodic hydrogen and cathodic oxygen content are calculated in the steady state and dynamically. The dynamic switching of electrolyte cycles can be simulated with the dynamic model.
The experiments are conducted to define the anodic hydrogen and cathodic oxygen content in a single cell configuration. The first experiment outputs the steady-state impurity as a function of the current density. The steady-state impurities show a descending tendency with an increasing current density. Next, the dynamic switching is performed and the anodic hydrogen content is recorded as a function of time. The average impurity in the dynamic switching is smaller than the result in the steady-state experiment.
The steady-state model sufficiently validates the literature data and verifies the experimental results. The dynamic switching model validates the literature data. Furthermore, it verifies the experimental results, when a correction factor is applied to the total volume of the separator tanks. The correction factor is required because the experimental impurities were measured at the exit of the single cell, resulting in faster system response. Finally, a sensitivity analysis is conducted to test the robustness of the dynamic model. The sensitivity analysis shows that the dynamic model can successfully simulate the operation of an alkaline water electrolyser.