Case Study: PSA Peugeot Citroën achieves improved failure prediction for large-scale crash simulations with Altair RADIOSS

Toward Better Failure Prediction for Crash Simulations using HPC

PSA Peugeot Citroen, Europe’s second-largest carmaker, faced the challenge of predicting material failure and cracks in crash simulations as it introduced higher strength-to-weight materials. To evaluate advanced rupture criteria at an unprecedented scale, PSA partnered with Altair and used Altair’s RADIOSS solver (part of the HyperWorks suite) on PRACE’s Curie supercomputer, since the simulations required supercomputing performance beyond standard clusters.

Altair ported and optimized RADIOSS to Bullxmpi on Curie, tuned MPI/OpenMP parallelization, I/O and contact treatments, and implemented a validated new material law and rupture model developed with PSA and Ecole Polytechnique. The work enabled 15M‑element runs (scaling efficiently to 8,192 cores and tested to 16,384), validated RADIOSS for multi‑million solid‑element crash models, and produced a 10M‑element 100 ms simulation plan expected to run in under 60 hours on 2,048 cores—demonstrating measurable scalability and readiness for large‑scale rupture studies.

PSA Peugeot Citroen

Laurent Caplain

CAE Crash Specialist for Car Body & Equipments