A critical review of modeling hydrogen production using water electrolyzer

Abstract The quest for effective hydrogen production through water electrolysis depends on the performance. Yet, making good models for performance improvement is naturally difficult because operation of an electrolyzer is both a multi-physics and multi-scale problem. Interaction of such complex phenomena across disparate spatial and temporal scales makes system design and optimization an extremely difficult task that indeed calls for advanced computational approaches. This review explores the application of recently developed computational methods to address such problems. Key methods examined include the lattice Boltzmann method (LBM), computational fluid dynamics (CFD), response surface methodology (RSM), and artificial intelligence (AI) methods. Water electrolyzer simulations are dominated by two-phase liquid–gas models; the LBM is particularly effective for microscale flows and interfacial phenomena where surface effects are important, while Eulerian volume of fluid approaches are the most effective for treating bubble behavior. Briefly, optimal surrogate models for integrated systems are provided by empirical correlations and experiment design techniques (such as RSM). AI and hybrid AI-CFD techniques are making modeling and optimization easier and faster. For instance, DeepONet has predicted current density, oxygen mole fraction, and cell temperature with a root-mean-squared error of less than 1 %. This review concludes that LBM is a valuable tool for microscale multiphase dynamics and that AI-augmented CFD has proven capable of supplementing, and in certain situations, even replace conventional CFD workflows for the design and optimization of electrolyzers.