Efficient Battery Modeling with Surrogate Models

Battery modeling can be greatly sped up by building simulation apps that contain surrogate models. Whereas the detailed Newman model provides accuracy at a significant computational cost, lumped models are faster but less accurate. For cases that call for both accuracy and speed, surrogate models are worth considering.

In this webinar, we will discuss the tradeoff of complexity vs. speed of results and how battery applications may benefit from surrogate models, as well as the process of training and using surrogate models for time-dependent battery models. Through an example of a simulation app that predicts the performance of a Li-ion battery undergoing a test cycle, we will demonstrate functionality available in the COMSOL Multiphysics^® software for training deep neural network models.