4.6 Article

Fully-Discrete Analysis of High-Order Spatial Discretizations with Optimal Explicit Runge-Kutta Methods

Journal

JOURNAL OF SCIENTIFIC COMPUTING
Volume 83, Issue 3, Pages -

Publisher

SPRINGER/PLENUM PUBLISHERS
DOI: 10.1007/s10915-020-01243-8

Keywords

High-order methods; Optimal explicit Runge-Kutta methods; Fully-discrete analysis; Flux reconstruction

Funding

  1. Natural Sciences and Engineering Research Council of Canada (NSERC) [RGPAS-2017-507988, RGPIN-2017-06773]
  2. WestGrid
  3. SciNet
  4. Calcul Quebec
  5. ComputeCanada

Ask authors/readers for more resources

High-order unstructured methods have become a popular choice for the simulation of complex unsteady flows. Flux reconstruction (FR) is a high-order spatial discretization method, which has been found to be particularly accurate for scale-resolving simulations of complex phenomena. In addition, it has been shown to provide sufficient dissipation for implicit large-eddy simulation (ILES). In conjunction with an FR discretization, an appropriate temporal scheme must be chosen. A common choice is explicit schemes due to their efficiency and ease of implementation. However, these methods usually require a small time-step size to remain stable. Recently, the development of optimal explicit Runge-Kutta (OERK) schemes has enabled stable simulations with larger time-step sizes. Hence, we analyze the fully-discrete properties of the FR method with OERK temporal schemes. We show results for first, second, third, fourth and eighth-order OERK schemes. We observe that OERK schemes modify the spectral behaviour of the semidiscretization. In particular, dissipation decreases in the region of high wavenumbers. We observe that higher-order OERK schemes require a smaller time step than the low-order schemes. However, they follow the dispersion relations of the FR scheme for a larger range of wavenumbers. We validate our analysis with simple advection test cases. It was observed that first and second-degree temporal schemes introduce a relatively large amount of error in the solutions. A one-dimensional ILES test case showed that, as long as the time-step size is not in the vicinity of the stability limit, results are generally similar to classical RK schemes.

Authors

I am an author on this paper
Click your name to claim this paper and add it to your profile.

Reviews

Primary Rating

4.6
Not enough ratings

Secondary Ratings

Novelty
-
Significance
-
Scientific rigor
-
Rate this paper

Recommended

No Data Available
No Data Available