Michelin - France
Using modern model reduction and machine learning techniques, virtual Michelin tires are now used for in-house design, in driving simulators, and to define tire-oriented services for our customers. Beyond tires, simulation and data science play a key role across the product lifecycle. During the talk, we will introduce the global context of simulation and data science usage at Michelin R&D. Then, we will focus on three example applications combining physics-based models, reduced-order modeling, physics-informed machine learning, and online update techniques: (1) predicting rubber mix behavior during the manufacturing phase, (2) accelerating convergence of fluid-structure interaction simulations, and (3) reducing the cost of expensive 3D tire rolling models. These examples will illustrate how hybrid approaches leveraging physics and data are paving the way toward robust, efficient, and versatile digital twins.