A new high-order discontinuous galerkin solver for dns and les of turbulent incompressible flow
New Results in Numerical and Experimental Fluid Mechanics XI: Contributions to …, 2017•Springer
We present recent developments within a high-performance discontinuous Galerkin solver
for turbulent incompressible flow. The scheme is based on a high-order semi-explicit
temporal approach and high-order spatial discretizations. The implementation is entirely
matrix-free, including the global Poisson equation, which makes the solution time per time
step essentially independent of the spatial polynomial degree. The algorithm is designed to
yield high algorithmic intensities, which enables high efficiency on current and future CPU …
for turbulent incompressible flow. The scheme is based on a high-order semi-explicit
temporal approach and high-order spatial discretizations. The implementation is entirely
matrix-free, including the global Poisson equation, which makes the solution time per time
step essentially independent of the spatial polynomial degree. The algorithm is designed to
yield high algorithmic intensities, which enables high efficiency on current and future CPU …
Abstract
We present recent developments within a high-performance discontinuous Galerkin solver for turbulent incompressible flow. The scheme is based on a high-order semi-explicit temporal approach and high-order spatial discretizations. The implementation is entirely matrix-free, including the global Poisson equation, which makes the solution time per time step essentially independent of the spatial polynomial degree. The algorithm is designed to yield high algorithmic intensities, which enables high efficiency on current and future CPU architectures. The method has previously been applied to DNS and ILES of turbulent channel flow and is in the present work used to compute flow past periodic hills at a hill Reynolds number of . We also outline our on-going work regarding wall modeling via function enrichment within this framework.
Springer