Modern alkaline water electrolysers for hydrogen production often use a zero-gap configuration in which electrodes are pressed directly against the separator. Counterintuitively, the inclusion of a small electrode-diaphragm gap was previously shown to reduce the cell potential significantly. This work aims to understand, quantify, and model this effect, and take first steps towards optimisation of the gap width. We present experimental measurements and simulations of the cell potential and kinetic and ohmic losses for expanded metal electrodes. We find that the configuration with our smallest used gap, created using a 60µm spacer, yields the lowest cell potential, while the zero-gap configuration incurs additional voltage losses of approximately 80mV at a current density of 10⁴A/m². This can be explained by bubbles and gas films in between the electrode and diaphragm, which block part of the diaphragm and electrode area. We introduce an analytical model that predicts the vertical gas fraction and current density distribution in the electrode-diaphragm gap, which is in good agreement with experimental and simulation results. For a maximum gas fraction below 0.7, the model can explain why there is no optimal gap width based solely on the gap resistance. Instead, the gap allows gas to escape, which mitigates the additional zero-gap overpotential. Our findings confirm, explain, and quantify that intentionally adding a small gap can be an effective way to improve the performance of alkaline electrolysers with perforated plate-like electrodes.

As current densities in alkaline water electrolysers increase, the resistive losses become increasingly important due to the locally high gas fraction around the electrodes, even in zero-gap configurations. Nonetheless, quantitative measurement of the distribution of these high gas fractions is difficult. Consequently, a numerical approach is useful to assess the impact of bubbles on electrolysis. However, models that couple current density and gas fraction distributions in a non-trivial geometry are currently lacking. We show that typically used models in the literature predict unrealistically high gas fractions in electrode-resolved simulations. To improve this, we added to the mixture model equations a solid pressure model similar to that used in simulations of dense granular flows. With the addition of this model, two-dimensional simulations of a lab-scale electrolysis cell accurately reproduce previously reported experimental results. This allows, for the first time, to predict local overpotentials influenced by the bubble distribution, opening the way towards computational optimisation of the electrode geometry.