During the past few decades the world of science has changed, moving from physical experimentation alone to an environment in which simulations are of primary importance. In the automotive industry, research and design is now so advanced that it’s following in the footsteps of science and moving away from physical testing. Supercomputing has allowed automobile manufacturers to transition to a simulation-driven world.
The innovative use of new materials in vehicle design is becoming the norm for automotive manufacturers — honeycomb structures and increased aluminum use being an example. With more advanced design elements coming into play, numerous manufacturing challenges are emerging; including the need for a greater number of simulations.
Moving into the age of simulation
Advanced simulations in the automotive industry are not new. However, the capabilities of HPC systems and the ability for CAE applications to exploit computing resources have changed substantially. Just five years ago a simulation based on MPI parallelism would use only a few dozen cores. Today, parallel simulations using 256 compute cores are routine and there is a growing demand for scaling to thousands of cores per simulation.
The result is a high-fidelity simulation in which five or ten thousand elements can be involved in any given automotive crash test. Scalability becomes critical in this environment, as it is in the design assessments required for stochastic simulations.
Understanding the fiscal value of simulation
Automotive manufacturing has traditionally been built around performance-based analysis. Organizations would commonly create and prototype a design and then test it in physical ways to figure out the nuances of how the new design functions — but the costs of this trial-and-error process can be huge. High performance computing systems enable the use of predictive assessments, eliminating uncertainty when developing new vehicles and creating manufacturing strategies.