A Mesh Free Framework for Mechanical Simulations of Microstructure Data Files

The Exascale Computing Project is tasked to develop the next generation of high performance computing systems capable of computing at 50 to 100 times faster than current HPC systems. Computing at this extreme-scale will significantly enhance the value of materials modeling and simulation to basic materials research and engineering In particular, computing at the extreme scale will enable a higher fidelity consideration of material microstructure in multiscale modeling frameworks. This project seeks to develop and test a new continuum mechanics code that is optimized to operate at the exascale using microstructure datasets as the input structure. Mechanical simulation of microstructure datasets is limited by the extreme file sizes and the difficulty in converting complicated geometries to digital meshes for analysis using classical continuum mechanics methods. Peridynamics, a non-local continuum mechanics theory, offers several unique capabilities beyond classical continuum mechanics when applied to the analysis of microstructure data. Peridynamics is most often implemented in a meshfree framework which allows for efficient conversion of data in voxel file formats to grid files suitable for simulation. At the microscale, void growth and fracture dynamics are key constituents in the material response. Peridynamics specifically addresses mechanics in the presence of discontinuities. The peridynamic equations can be solved as a discrete sum of pair interactions which is readily adapted to parallel processing using many of the code structures developed to accelerate molecular dynamics, including acceleration using GPU’s. This project will demonstrate a peridynamics code developed for execution in hybrid CPU/GPU architectures that is able to covert and refine a digital microstructure for mechanical analysis. Assessment criteria and procedures will be established to measure code performance. The performance of the newly developed code will be compared to currently available implementations of non-local methods. There is a commercial market for materials modeling and simulation services that take advantage of capabilities of extreme scale computing. Particular to this project, the ability to model fracture nucleation and growth is well suited to support the use of ceramic materials in harsh service conditions. Designing for fracture resistance is a challenge that is not well met using classical finite element analysis. Meshfree tools directly support the development and incorporation of advanced ceramics for use in high temperature applications in energy, defense and aerospace industries such as next generation gas turbine engines.