Efficient and Robust High-Order Methods for Fluid and Solid Mechanics
Abstract
The goal of the project was to develop new numerical schemes and solvers for high-order accurate simulations of problems in fluid and solid mechanics. Three main areas were addressed - space-time methods for domains with large deformations, implicit matrix-free solvers for sparse line-based discretizations, and applications to problems with highly turbulent flow. The project has led to significant developments in space-time mesh generation for complex flow problems, entropy stable line-based discontinuous Galerkin discretizations, low-memory Kronecker SVD factorizations applied to preconditioning, practical solvers for fully implicit Runge-Kutta methods, partitioned multiphysics solvers based on IMEX schemes, and new high-order schemes for shock tracking. The results were applied to important real-world problems, such as the high-order simulation of shock boundary-layer interaction. The findings were disseminated through a wide range of publications, presentations, and public domain software.
Document Details
- Document Type
- Technical Report
- Publication Date
- Feb 01, 2019
- Accession Number
- AD1085857
Entities
People
- Per-olof Persson
Organizations
- University of California, Berkeley