High-fidelity solver on polyhedral unstructured grids for low-Mach number compressible viscous flow

In this article, we developed an unstructured fluid solver based on finite volume framework for the low-Mach number compressible flows. The present method, so-called FVMS3 (Finite Volume method based on Merged Stencil with 3rd-order reconstruction) formulates two different numerical procedures for spatial reconstructions based on the quadratic polynomial which is performed by using least-square approximations on a merged stencil. In order to improve the reconstruction for discontinuities, we propose the limiting projection approach and smoothness adaptive fitting (SAF) scheme to suppress the numerical oscillation and limit the numerical dissipation. The resulting discretization algorithm that combines FVMS3 with SAF-based limiting projection scheme has third-order accuracy and resolves both smooth and non-smooth solutions with excellent quality. Additionally, a novel numerical model has been proposed by introducing the advection upstream splitting method (AUSM) flux into the pressure projection formulation which results in a unified scheme that works uniformly up to the incompressible limit. The fluid solver that integrates all above new efforts provides high-fidelity solutions for compressible viscous flows particularly for the low Mach regime. The performance of this new solver is verified by numerous benchmark tests. Our numerical results show that the proposed scheme gives accurate and robust solutions for a wide spectrum of test problems.

Bin Xie, Xi Deng, Shijun Liao. High-fidelity solver on polyhedral unstructured grids for low-Mach number compressible viscous flow. Computer Methods in Applied Mechanics and Engineering, Elsevier, 2019, 357, pp.112584. ⟨10.1016/j.cma.2019.112584⟩. ⟨hal-02467981⟩

Journal: Computer Methods in Applied Mechanics and Engineering

Date de publication: 01-12-2019

Auteurs:
  • Bin Xie
  • Xi Deng
  • Shijun Liao

Digital object identifier (doi): http://dx.doi.org/10.1016/j.cma.2019.112584

x >