From weeks to a day: Mitsubishi Chemical Group answers its customers faster with AI-driven material data

Carmakers simulate crashes with MCG’s data. Producing it took weeks.

Mitsubishi Chemical Group’s performance polymers help the automotive industry tackle electrification, comfort, connectivity and resource preservation. Beyond the materials themselves, MCG supports its partners with material characterizations: the input data for highly dynamic crash simulations. For the thermoplastic elastomers used in airbags, the accuracy of that data is critical.

Producing it was slow. Engineers combined experimental tests with optimization runs, painstakingly matching simulated material behavior to test results across a wide range of temperatures and strain rates. Each run took hours, many were needed, and a qualified answer to a customer could take several weeks.

From test results to material law, in seconds.

Starting in April 2020, MCG and Neural Concept developed, in close cooperation, an AI model able to predict simulation results across a wide range of material laws. Once trained, it reverses the process: starting from experimentally tested data, it automatically creates the optimized material law that matches the test curves.

What took hours per optimization run now takes seconds, with significantly improved accuracy. And MCG can go beyond standardized material laws, providing fully tailored data for each customer’s specific demands, within days.