Development of a calibration methodology for fitting the response of a lithium-ion cell P2D model using real driving cycles

Pseudo-two-dimensional models based on physical processes are of significant relevance in this field, especially now that computational cost is getting more affordable with new technological advancements. Their biggest demerit is the difficulty in selecting a reduced number of parameters to consider during the optimization process to maintain the coherence of the physical processes and a good compromise in complexity. The current work proposes a methodology in which a selection of 14 critical constructive and performance parameters are itera-tively fitted with an affordable computing cost using a genetic algorithm. The objective is to represent with high fidelity the experimental response of real 18,650 lithium-ion cells based on different cathode chemistries (NMC 811 and NCA). The results show that the proposed methodology can deliver better results if the calibration process is performed with a single dynamic driving cycle test instead of a series of constant C-rate curves, maintaining high reliability when simulating dynamic conditions such as driving cycles representative of real transport applications. The maximum voltage Root Mean Square Error (RMSE) of the validation profiles is not exceeding 0.0315 V and 0.0357 V for the NMC 811 and NCA cells, respectively.

Automated summary

Pseudo-two-dimensional models based on physical processes are relevant and affordable with new technological advancements. However, selecting a reduced number of parameters is a challenge. This study proposes a methodology using a genetic algorithm to fit 14 critical parameters of lithium-ion cells and achieves high fidelity in simulating real transport applications.