Graph Network Simulator

Tsunamis, landslides, and other storm-surges can mobilize debris, compounding the hazards faced by built environments. Predicting the dynamics of debris flow in these events is challenging due to the varying materials and their properties. This study presents a novel approach to address this challenge using a combination of machine learning (ML) and advanced numerical simulation. For the creation of lightweight surrogate models, leveraging the highly flexible Graph Network Simulator (GNS), we introduce a wave flume simulation software for debris flows using the Material Point Method (MPM), implemented within the Taichi framework. By utilizing high-performance computing and deep learning networks, this approach enables the efficient representation of complex physics and facilitates uncertainty quantification in debris hazard events. This project allows for easy generation of debris flow data from various materials. The project contents are intended to be used by other researchers in the field.