The world is witnessing unprecedented climate changes, which have worsened over the past couple of years. Despite significant progress in renewable energy, achieving a reduction of global warming remains challenging. Renewable energy sources like solar, wind, and hydroelectric technologies offer promising solutions, but challenges persist due to their intermittency and geographical dependency. Green hydrogen, generated through electrolysis using renewable energy, has come up as a solution for storing electric energy and providing a constant energy supply.

Alkaline water electrolysis has emerged as one of the most promising electrolysis methods due to its large-scale operation, long durability and low costs. This does not come without challenges, one of them being leakage currents, which reduces the efficiency of the electrolyser. Leakage currents occur when not all current is used for hydrogen production, but some of it leaks into, for example, the produced hydrogen stream. To reduce these leakage currents, more research into the origins is needed. The first step is modelling the electrolysis while considering as much as possible of the physics happening inside the system. Current models of alkaline water electrolysers are either modelled using only mathematical equations or neglect the operating parameters, which makes the results highly variable per system. Other models do use analytical methods with experimental results, but these models only comprise one electrolysis cell rather than a full stack.

This work consists of developing two three-dimensional models, validating them using experiments, and using them to predict the effect of changes in geometry. The first model was made using COMSOL Multiphysics software and used to research the water electrolysis stack, which comprised one or eight cells using electrochemical relations and physical data. It was found that a model could be made that fitted the experiments within the error margin of the experiments (<2.5%). It lacked flexibility but overall showed good results for an electrolyser stack of one or eight cells. The second model was made using an equivalent electrical circuit (EEC) of the electrolyser in Python via the PySpice module. A steady-state model, including the leakage currents, could be developed by calculating all system resistances, namely the cell, inlet/outlet, and manifold resistances. This model overestimated the performance of the electrolyser by 10-15% for low current densities and 2-4% for high current densities. Nevertheless, it was highly adaptable for different scenarios, making it valuable for research into optimising the electrolysis stack. Both models were used to predict the effect of changes in geometry; the effect of the length of the inlets, and the number of cells. This showed that the EEC model was better suited for this research.