Speaker
Description
Granular materials are prevalent in both natural and industrial contexts, presenting complex physical behavior that arises from their discrete composition. This complexity is further amplified by thermal effects, which explain the current gaps in achieving a comprehensive understanding of the thermomechanics of granular matter. This talk will present a data-driven framework to efficiently simulate the thermomechanical behavior of granular materials following a continuum-discrete hierarchical multiscale approach. In particular, the focus is on thermal expansion of densely packed, confined granular media. The proposed methodology handles the macroscale using a continuous model, while the microscale response is obtained from Representative Volume Elements (RVEs) using the Discrete Element Method (DEM). To significantly reduce the computational cost, the DEM computations are not performed simultaneously with the macroscale solver; instead, they are performed in advance to create a database of RVE solutions under different conditions. This dataset is then used to train an artificial neural network, which serves as a surrogate model for the macroscale solver. The method is validated against pure DEM solutions of granular domains with distinct thermal conditions. It is concluded that the surrogate model can predict the evolution of the microstructure, effective conductivity, and Cauchy stress tensor of the granular assembly with satisfactory accuracy and at a drastically lower computational cost than the pure DEM approach.