NobleBlocks

Research indexJournal article

Surrogate-model-assisted optimization framework for solid oxide fuel cell stack design based on proper orthogonal decomposition

Sasaki, Mori, Kishimoto and 2 others

Transformative Energy22 Sep 2026Open access

NobleIDark:/48914/w1/20261105/C62F0BCCDOI10.1016/j.tegy.2026.100026

Abstract

Optimizing solid oxide fuel cell (SOFC) stack configurations is vital for ensuring durability and mechanical reliability, where managing thermal constraints is particularly critical to prevent structural failure. However, conventional 3D-CFD simulations are computationally prohibitive for large-scale design optimization. This study proposes a high-efficiency framework utilizing a surrogate model to rapidly predict internal stack states. The model is constructed by applying proper orthogonal decomposition (POD) to temperature and electric potential distributions obtained from 3D-CFD analyses. Validation results demonstrate high predictive accuracy, with a mean absolute temperature error of 5.5 K and a stack voltage error below 1.1%. Furthermore, computational time per case was reduced from approximately 1 h to the millisecond order. Using this framework, multi-parameter optimization was performed to maximize the stack power density under specified thermal constraints. The developed framework enables rapid exploration of the coupled effects of cell size, interconnector thickness, and operating current density under multiple thermal constraints, and quantitatively identifies feasible stack configurations and the corresponding maximum achievable power density. The results further demonstrate that the active thermal constraint can shift with stack configuration, thereby changing the optimal design. This surrogate-based approach offers a practical tool for the accelerated development of SOFC systems, enabling optimal configurations that balance power performance and thermal reliability.

Discussion

Citations

More

Suggest a correction or report