Abstract
This study presents a comprehensive computational framework that integrates physics-based modeling and data-driven approaches for analyzing and optimizing photoelectrochemical (PEC) water splitting systems. Utilizing COMSOL Multiphysics 6.1 and MATLAB, we simulate key electrochemical behaviors such as linear sweep voltammetry (LSV) and electrochemical impedance spectroscopy (EIS). COMSOL’s multiphysics environment allows for the direct incorporation of electrolyte parameters, semiconductor photophysics, and current distribution, while MATLAB enables custom modeling of impedance behavior and predictive LSV analysis using artificial neural networks (ANNs). By coupling computational fluid dynamics (CFD), machine learning, and experimental validation, the proposed methodology provides an in-depth understanding of light-driven hydrogen generation on semiconductor electrodes such as ZnO/BiVO 4 . Comparative analysis of simulation results shows that COMSOL and MATLAB produce consistent outputs, yet COMSOL demonstrates superior flexibility, accuracy, and ease of use, especially for systems influenced by variable physical and chemical conditions. The study further explores two-phase flow modeling, mesh independence testing, and the influence of gas bubbles on electrolyte conductivity. The findings contribute to the development of efficient, scalable PEC systems for clean hydrogen production and offer a foundation for future integration of hybrid simulation and AI techniques in renewable energy research.