Keynote: Towards manufacturing digital twins using GPU based physics simulations: a particles perspective - Nicolin Govender
2024-06-13 , Olav Tryggvason 2+3

Digital twin, GPU, particle flow


GPU computing is enabling groundbreaking innovation and enhanced efficiencies for a number of manufacturing industries. From digital twins that enable production automation and optimization, to finding unique patterns in material combinations that can lead to new formulations, simulation is transforming what’s possible. Although significant success has been achieved with data-driven digital twins, inferences are limited when the underlying physics is complicated. Physics-based digital twins on the other hand offer far richer inferences but require access to verified and validated solvers and models that are computationally tractable. While there has been progress in applying AI to accelerate numerical methods themselves, industrial materials are complex with a mixture of fluid and solids for which there are no closed form solutions. In this talk progress towards physics based simulations of particulate materials using GPU computing will be presented along with a technical discussion on the art of utilizing GPUs for other solution methods ending with an outlook of the potential of AI for industrial simulations.