AI Feasibility Study: Optimizing Crash Performance Using PhysicsAI

Crush zone management is an important aspect of automotive safety design. During a collision, the crush zone helps control the forces passed onto passenger sections by absorbing much of the kinetic energy involved. The idea is to break the car, not the people. This whitepaper instructs readers how to use Altair PhysicsAI to design rails — a key component of crush zone design. It outlines how AI models were trained on 450 finite element analysis (FEA) simulations of crash data and then validated using another 50 crash simulations.

It also outlines how Altair HyperStudy was used to create a workflow to optimize the rail’s design using PhysicsAI. The paper highlights that AI model results showed good consistency with FEA model results when compared via R-squared values, visual inspections, and Pareto fronts. The AI predicted results from the Pareto fronts had only a maximum deviation of 15% from FEA results. Thus, the paper suggests that PhysicsAI models can effectively perform early-stage rail design evaluations and optimizations while returning results 5 times faster than traditional FEA workflows.