Model Order Reduction of the Doyle-Fuller-Newman Model via Proper Orthogonal Decomposition and Optimal Collocation

Adapting physics-based models for digital twin frameworks, estimation and control of Lithium ion batteries requires a significant reduction of the computational complexity of electrochemical models without sacrificing accuracy. In this study, we propose a model order reduction technique for the Doyle-Fuller-Newman model of a battery cell that integrates Proper Orthogonal Decomposition with an optimized orthogonal collocation method. Simulation results indicate that the reduced-order model is faster than the full-order model, reducing computation time by up to 99% while maintaining negligible voltage discrepancy with a root mean square error (RMSE) of less than 0.005V. The proposed approach provides a powerful framework for improving the computation time of complex, physics-based models based on coupled partial differential equations, leading to more accurate predictions and better design and optimization of complex components and processes.