AI enhanced methods for virtual prediction of short circuit in full vehicle crash scenarios

A new artificial intelligence (model order reduction)/finite element coupled approach will be presented for the risk assessment of battery fire during a car crash event. This approach combines standard crash finite element for the main car body with a reduced order model for the battery. Simulation is today used by automotive engineering teams to design lightweight vehicle bodies fulfilling vehicle safety regulations. Legislation is rapidly evolving to accommodate the growing electrical vehicle market share and is considering additional battery safety requirements. The focus is on avoiding internal short circuit due to internal damage within a cell which may result in a fire hazard. Assessing short circuit risk in CAE at the vehicle level is complex as there involves phenomena at different scales. The vehicle deforms on a macroscale level during the impact event. The battery cells deform locally and damage extremely thin components such as the separator, generating short circuits at the meso scale level. It is therefore necessary to evaluate phenomena from macroscale to microscale. AI methods will help to bridge the various length scales. A detailed cell model is first assessed with mechanical properties of cell components based on the battery geometry. Realistic boundary conditions on the cell level are gathered from standard electrical vehicle crash cases and cascaded down to the battery cells. Cell stiffness and short circuit risk evaluation are used to build a parametric reduced order model with low computation cost. This is the first step to assessing virtually and efficiently short circuit risk in full vehicle crash scenarios based on physical-non empirical-values.