A simple hybrid finite volume solver for compressible turbulence

Abstract

SummaryA simple, explicit, hybrid finite volume method for simulating compressible turbulence is developed by combining a fourth‐order central scheme and a shock‐capturing simple low‐dissipation advection upstream splitting method. The total flux on each of the cell faces is computed as a weighted average of central/nondissipative and upwind/dissipative fluxes. The weights are determined using an unphysical oscillation sensor in addition to a more traditional discontinuity sensor used in earlier studies. Shocks are well captured, but overshoots in density are predicted around contact discontinuities that are normal to the flow. The use of the latter sensor effectively prevents these overshoots from generating spurious oscillations that travel away from the contact lines. The efficacy of the proposed method for direct or large‐eddy simulations of supersonic turbulence is established using several canonical test problems. Copyright © 2015 John Wiley & Sons, Ltd.

High-order magnetohydrodynamics for astrophysics with an adaptive mesh refinement discontinuous Galerkin scheme

“…This test problem demonstrates the ability of higher-order methods to capture finer features in the flow with lower numerical diffusion. In particular, our third-order scheme is very competitive for this test problem: Chakravarthy et al (2015) show that they can capture the oscillations with 400 grid points (about 14 points per oscillation period, see their Figure 9), Derigs et al (2016) demonstrate good capturing with 256 points (about 9 per period, see their Figure 6), and with the DG-3 scheme, we obtain results comparable to this latter work with only 128 total grid points (i.e. 4 to 5 points per period, see the corresponding DG-3 = 7 line in Fig.…”

A novel high-order, entropy stable, 3D AMR MHD solver with guaranteed positive pressure

“…According to previous investigations, e.g. [73], a scheme is considered "acceptable" for capturing supersonic turbulence if the dynamics can be captured well with 400 cells. However, the entropy stable scheme is also able to resolve the dynamics of the flow with a much lower spatial resolution (the result used in Fig.…”

Damping numerical oscillations in hybrid solvers through detection of Gibbs phenomenon

“…This is also a feature of hyperviscosity approaches (as seen in solutions for the Shu–Osher test). Note that the kink is reduced considerably when the RHLL scheme is used instead of the SLAU2 scheme .…”

“…Only pure WENO schemes are immune to this problem, but they have been deemed [9,10] too dissipative to capture turbulence accurately. 700 V. KALYANA CHAKRAVARTHY AND D. CHAKRABORTY A simple and effective way of dealing with numerical oscillations in hybrid, explicit finite volume schemes by basing the weights on a Gibbs phenomenon detector in addition to a more traditional discontinuity sensor was proposed earlier and studied using canonical problems [11]. The present work differs from an earlier study in two ways.…”

“…A simple and effective way of dealing with numerical oscillations in hybrid, explicit finite volume schemes by basing the weights on a Gibbs phenomenon detector in addition to a more traditional discontinuity sensor was proposed earlier and studied using canonical problems . The present work differs from an earlier study in two ways.…”