Physics Behavior Prediction of Heterogeneous Materials with a Varying Morphology: A Parametric Deep Material Network Approach

On February 5th was held the Scientific and Technical Day (Journée Scientifique et Technique) on multiscale approaches for composite materials and structures modeling. These JSTs are regularly organized by the Association pour les Matériaux Composites. What makes this particular JST special is that it took place at Dassault Systèmes Paris Campus!​​​​​​​

As one of the Dassault Systèmes speakers of this conference, I would like to express my sincere thanks to the scientific committee for having ensured an excellent scientific quality of the presentations, and most of all to the organizing committee for having made possible everything: welcoming, coffee breaks, buffet, logistics, etc. Chapeau 👏👏👏 (hats off) to @PM, @AC, @AB, @EF and many more!

Personally, I enjoyed very much my discussions with both academic and industrial participants enthusiastic in multiscale material and structure modeling and simulation. New contacts have been made thanks to this conference.

Please find below the slides of my presentation on a micromechanics-informed machine learning surrogate for multiscale material behavior prediction. The objective is to accelerate both linear and nonlinear microscale simulations of a heterogeneous material (composite, for instance). The presentation is based on my scientific paper published recently (as well as the arXiv preprint​​​​​​​). Please don’t hesitate to contact me (tianyi.li@3ds.com) for more information.