High-order finite-difference nonlinear filter methods for subsonic turbulence simulation with stochastic forcing

Authors

Alexei G. Kritsuk, University of California - San DiegoFollow
Dmitry Kotov, NASA Ames Research CenterFollow
Bjorn Sjögreen, Lawrence Livermore National LaboratoryFollow
Helen C. Yee, Ames Research CenterFollow

Date of this Version

2019

Citation

Conference paper, ASTRONUM (Paris) 2019

Abstract only

Comments

U.S. government work

Abstract

Numerical stability of high-order fi lter schemes developed by Yee & Sjogreen is tested on thr~dimensional turbulence simulations with stochastic forcing and their performance is compared with that ofTVD and \VENO schemes. The bestperforming filter method employs an eighth-order central base scheme with the Kennedy & Gruber skew-symmetric splitting of the inviscid fttLx derivative, a wavelet-based local flow sensor, a nonlinear filter utilizing the dissipative portion of seventh-order \\'ENO scheme, and an explicit third- or fourth-order RungeKutta t ime integration. \Ve show that the filter scheme is more computational]y efficient and provides a wider spectral bandwidth compared to the seventh-order \VENO scheme. The method also demonstrates robust Jong-time integration for moderately compressible turbulence. In contrast, the the fifth- and seventhorder \¥ENO schemes show non-trivial evolution of the ve]ocity a.nd density power spectra. over a. few dozen dynamical times, where both TVD and filter schemes reoover a soHd statistically stationary turbulent st.ate.