A high-fidelity reduced-order modeling framework for accurate and real-time performance simulation of lead-acid batteries considering buoyancy-driven electrolyte stratification

Abstract

The main contribution of this study is developing an efficient and high-fidelity reduced-order model (ROM) for accurate and real-time simulation of the complex electrochemical and hydrodynamic behavior of a two-dimensional flooded lead-acid battery cell during the charging process, particularly under free-convection effects. A computationally expensive full-order model based on finite-volume method (FVM) is developed to capture the system dynamics, while proper orthogonal decomposition (POD) combined with Galerkin projection is employed to derive the ROM. A detailed eigenvalue analysis determines that only 21 algebraic equations are needed compared to 16368 equations in FVM to capture 99.999 % of system energy, which significantly reduces the computational cost. A deep insights into the buoyancy-driven electrolyte stratification, velocity patterns, and dominant flow structures, especially during early charging stages are provided using comprehensive basis mode and temporal coefficient analyses. The results shows that the developed ROM accurately reproduces concentration, velocity, and voltage profiles across various operating conditions, validated against experimental data. With a computational speed-up factor of 86 compared to FVM while maintaining accuracy, the POD-based ROM proves to be a reliable and efficient tool for real-time simulation and analysis of complex battery systems influenced by natural convection.