Development of a scalable parallel computing SPH framework for large-scale geotechnical applications

This thesis contributes to recent advances in parallel computing for geo-materials using SPH as well as limited study of large-scale geotechnical applications with advanced constitutive models. The SPH method is described, as well as general frameworks for implementation of soil material models. The parallel framework is described in detail and tested, achieving over 90% efficiency for strong and weak scaling tests up to 1,024 cores. The code is applied to study three applications requiring large-scale or high resolution: the axisymmetric granular column collapse, obstruction of granular flows using baffles, and a piled embankment model experiment.