Fusion power promises to be one of the most transformative technologies of our time, however, many fundamental questions about plasma physics and reactor operation remain. Answering these questions requires exascale multiscale and multiphysics simulations and a broad range of domain and computational expertise. Over the past 50 years, large teams have invested thousands of person-hours into the development of software that can efficiently simulate the physics in specific portions of the reactor volume that use fundamentally different discretizations, time-stepping methods, mathematical models, and so forth. Given the need for specialized numerics in each part of the reactor volume, and the expertise required to do so, it is not feasible to develop a new multiphysics code that encompasses the entire reactor volume using a homogenized framework. Currently, coupling methods exist for specific applications (i.e., core-edge), but do not scale to the needs of full-device, or full-plant modeling. This Project develops a new framework for tight-coupling of distinct codes at-scale on exascale supercomputers. It will describe the approach for data-transfer, parallel control, and interrogation-based field transfer operations.

Hi-fidelity multiscale simulations using frameworks such as MuMFiM are necessary to develop an understanding of fibrous materials on an engineering or biological scale. However, they require significant computational resources and HPC expertise that are out of reach for most biomechanicians. This project seeks to develop constitutive models for fibrous materials making use of machine learning methodologies. This has led to the development of a framework for using neural-network hyperelastic materials in MuMFiM and a set of physics constrained neural-networks that can estimate the constitutive response of fibrous materials. In an example test case, a model of a FCL that took 432 GPU-hours on 72-NVIDIA V100 GPUs could be solved in less than 30 minutes on a m1-macbook pro laptop. By continuing to incorporate the complex fiber network physics into these machine learned models, more of the biomechanics community can access the benefits of high-fidelity multiscale simulations.

This project is developing models of Vertical Shear Fracture in the Pelvis on patient specific geometry. It is a collaborative effort with Dr. Regis Renard (UAMS), and Dr. Brian Salazar (U.C. Berkeley).