COBRAPRO: An Open-Source Software for the Doyle-Fuller-Newman Model with Co-Simulation Parameter Optimization Framework

This paper presents COBRAPRO, a new open-source Doyle-Fuller-Newman (DFN) model software package with an integrated closed-loop parameter optimization routine. A key challenge in DFN model parameterization is that parameters measured from cell tear-down experiments cannot be directly used in simulations, and parameter identification is required to accurately reflect real-world battery dynamics However, existing open-source DFN codes lack the capability to perform parameter identification and operate in open-loop mode. COBRAPRO addresses this gap by implementing a systematic parameterization pipeline to accurately determine parameters using battery current and voltage data. Concepts from structural and practical identifiability are utilized to determine parameters that can be fixed to their experimental values and parameters that are suitable for optimization. In the parameter identification process, particle swarm optimization is used to minimize the error between experimental data and simulation results. Additionally, COBRAPRO incorporates a robust method to determine consistent initial conditions and utilizes a fast numerical solver for improved performance. We demonstrate COBRAPRO's parameter identification framework on reference performance test data obtained from LG INR21700-M50T cells. The parameterized model is validated against driving cycle data, showing good agreement between the experimental and simulation results.