Optimizing High-Fidelity Crash & Safety Simulation Performance

Crash/safety simulation is one of the most widely adopted HPC applications in the automotive industry and a core technology in the vehicle development process. As the volume and complexity of these simulations increases, the compute requirements are growing exponentially. Advancing the parallel performance and efficiency of explicit structural simulations is critical, and companies that integrate their HPC hardware and software strategies for simulation-driven design are more successful in developing better vehicles faster.

In this recording, Altair and Cray experts discuss results of a crash simulation study using RADIOSS running on a Cray® XC40™ system with 16,000+ cores (as discussed in a recent benchmark paper) – and will present two new examples to illustrate RADIOSS performance for very large models: A b-pillar model with a very fine mesh of 10 million solid elements (built for a PRACE project with PSA and Ecole Polytechique to asses crash failure prediction), and a new demo model of 25 million elements.

What you will learn:

  • Best practices for crash/safety simulation using high performance systems
  • Performance requirements for RADIOSS simulations for both larger and smaller models
  • Results of crash simulation scaling at 16,000+ cores
  • Recommended HPC architectures for crash, including both cluster and supercomputer systems

Who should watch?

  • Crash engineers
  • Safety engineers
  • HPC architects and support specialists
  • CAE managers
  • IT managers

Taurus_10M_Crash_one.png

Watch the Recording