High Order Hermite Finite Difference Method for Euler/Navier-Stokes Equations in 2D Unstructured Meshes

TL;DR

This paper introduces a high order Hermite finite difference method on unstructured meshes for simulating 2D Euler and Navier-Stokes equations, achieving high accuracy and effective shock capturing.

Contribution

It develops a novel high order Hermite finite difference approach combining least-square flux divergence computation with WENO-based flux interpolation on unstructured meshes.

Findings

Achieves fifth order accuracy in smooth regions.

Effectively captures shocks and discontinuities.

Demonstrates high accuracy and efficiency on canonical tests.

Abstract

A high order finite difference method is proposed for unstructured meshes to simulate compressible inviscid/viscous flows with/without discontinuities. In this method, based on the strong form equation, the divergence of the flux on each vertex is computed directly from fluxes nearby by means of high order least-square. In order to capture discontinuities, numerical flux of high order accuracy is calculated on each edge and serves as supporting data of the least-square computation of the divergence. The high accuracy of the numerical flux depends on the high order WENO interpolation on each edge. To reduce the computing cost and complexity, a curvlinear stencil is assembled for each edge so that the economical one-dimensional WENO interpolation can be applied. With the derivatives introduced, two-dimensional Hermite interpolation on a curvilinear stencil is applied to keep the stencil compact and avoids using many supporting points. In smooth region, the Hermite least-square 2D interpolation of 5 nodes is adopted directly to achieve the fifth order accuracy. Near a discontinuity, three values obtained by means of least-square 2D interpolation of 3 nodes, are weighted to obtain one value of the second order accuracy. After obtaining the flow states on both sides of the middle point of an edge, numerical flux of high order accuracy along the edge can be calculated. For inviscid flux, analytical flux on vertices and numerical flux along edges are used to compute the divergence. While for viscous flux, only analytical viscous flux on vertices are used. The divergence of the fluxes and their derivatives on each vertex are used to update the conservative variables and their derivatives with an explicit Runger-Kutta time scheme. Several canonical numerical cases were solved to test the accuracy and the capability of shock capturing of this method.