An improved multiphysics-integrated computational framework for 3D proton exchange membrane fuel cell modeling. Part II: Voltage and reference current density iteration strategies, framework validation and discussions

This study presents an improved 3D multiphysics-integrated solver for PEMFC design, developed within our in-house CFD framework as detailed in Part I. In this paper, to resolve manual iteration bottleneck, we propose automated outer iteration strategies incorporating Newton's method and similarity theory, applicable to both current density-prescribed and voltage-prescribed conditions. Subsequent numerical validation demonstrates 2.1% mean relative deviation between simulated and experimental polarization curves, with physically reasonable multiphysics field solutions. Simulations recommend 1 × 10−4% as convergence tolerance for inner iterations. Implementation of similarity theory-guided potential field updates achieves 34.6% reduction in required CFD iterations. Comparative analysis reveals comparable convergence rates between reference current density outer iteration and voltage outer iteration, while current density boundary implementation exhibits an order-of-magnitude slower convergence. Ionic potential and membrane water content fields demonstrate existence and uniqueness through intrinsic feedback mechanisms by source term, despite lacking benchmark values via constant-boundary conditions. Thermal-flow computations dominate processing time (>50%). Other physical fields exhibit near-positive correlation with domain scale. The dual species system and domain-segregated solving approaches reduce computational duration by 23.4% and 32% respectively. Quantitative evaluation shows potential 24% liquid water conservation error at CGDL-CCH interface using conventional methods, while the improved framework achieves absolute conservation.