This white paper presents a proof-of-concept study evaluating the feasibility of using Altair PhysicsAI to optimize rail geometry for automotive crash performance. Crush zone rails were modeled using simple rectangular cross sections and subjected to high-speed impact conditions. A design of experiments workflow generated 500 finite element crash simulations using Altair Radioss, with 450 cases used to train PhysicsAI models and 50 reserved for validation. Three transformer neural simulator models predicted displacement, impact force, and internal energy over time. Validation showed excellent agreement with FEA results, with R-squared values above 0.98 and a median of 0.9949. Visual comparisons across multiple crash stages confirmed accurate prediction of large deformations and buckling be

Join for free to read