MASTER ASSIGNMENTÂ
Recent advances in electrical vehicle (EV) design have shown that including batteries as structural components of the EV can be greatly beneficial. The main goal of this is to reduce EV weight, to increase EV range, while keeping EVs safe for use.
The finite element method (FEM) can be used to predict battery behaviour under loading. However, FEM cannot include the multitude of parameters for optimization (think about: case thickness, height, shape, heterogeneous material, uncertain material parameters, etc.). A Monte Carlo simulation that takes many samples of FEM model solutions is required, while the FEM model is expensive to compute. The calculation times for such a model are unfeasible. Therefore, simpler mathematical models are used to reduce the sample size, which are called surrogate models.
Surrogates are optimized on discrete sampling points to predict their surroundings continuously. Such a surrogate model can be used to connect the predictions of finite element models to predict smoothly between design parameters. A popular choice for surrogate models in the industry and recent literature are machine learning techniques. In this case, to explore the design space of a mechanical battery model for optimal parameters based on a trade-off between weight and safety.
Contact person
