NHERI-SimCenter/HydroUQ: Version 4.0.0

This open-source research application provides an application that can be used to predict the response of communities subjected to water-borne hazard events like tsunami and storm surge.

Major Changes

• Digital wave-flume of the Waseda University’s Tsunami Wave Basin, including scaled debris and built-environments, added to ClaymoreUW MPM. Validated against experiments  by Goseberg at al 2016, DualSPHysics SPH, and Siemens STAR-CCM+ FVM, pending publication. Includes a GUI and example case. (WBS1.1.2.5.1)
• Hydrodynamic PBE-demands at the building level for OpenFOAM-OpenSees and ClaymoreUW MPM. (WBS1.1.2.1.2)
• Initial surrogate capabilities from EE-UQ introduced, including Gaussian Process surrogates. Includes an example that trains a surrogate model on forward-sampled simulations of OpenSees structures during EVTs. Initially validated against a Monte-Carlo approach. (WBS1.1.2.2.2)
• Surrogate model approach which takes text-prompts as intensity measures (IMs), e.g. “a large, heavy bus” and generates an EVT’s initial conditions, e.g. debris geometry. Present in MPM as a point-e interface (https://github.com/OpenAI/point-e). (WBS1.1.2.2.2, 1.1.3.1.2)
• Surrogate modeling approach for CFD in EVTs. Graph-networks (GNS, https://github.com/geoelements/GNS) train using high-fidelity simulation data, creating a surrogate for the local dynamics of fluids, debris, and structures at graph nodes. Produces a time series of 3D field quantities (e.g. position, velocity, acceleration, material properties) at 100-1000x the speed of classical numerical simulations. Example workflow coupling digital twin simulations of the OSU LWF into a GNS surrogate model available at https://github.com/JustinBonus/Taichi-to-GNS-Wave-Flume. (WBS1.1.2.2.3, WBS1.1.2.5.1)
• Local and remote high-performance computation support for brand agnostic GPU and CPU accelerated numerical simulations. Use the Taichi Lang to speed-up high-fidelity EVTs, with examples provided for MPM, SPH, LBM, and FEM. (WBS1.1.2.5.5)
• Migration to tapisV3 for remote runs. (WBS1.1.2.5.5)

This work is based on material supported by the National Science Foundation under grants CMMI 1612843 and 2131111