Why the MUSCL-Hancock scheme is \(\text{L}^{1}\)-stable (Q2502238): Difference between revisions

The numerical approximation for solving hyperbolic systems of conservation laws such as the well-known Euler equations is discussed. Finite volume methods with time and space second--order of accuracy are considered. During the 20 last years, the problem of increasing the order of accuracy for solving hyperbolic system has been a very active research topic. Actually, one of the most popular second-order scheme has been introduced by \textit{B. van Leer} [J. Comput. Phys. 32, No. 2, 101--136 (1979; Zbl 0939.76063)]; namely the MUSCL scheme. It is a finite volume method where the flux approximation is second-order accurate. This scheme is considered and extended in a lot of applications. Since the scheme is used to perform very difficult numerical flow simulations, the robustness turns out to be central. Several works are devoted to such an analysis. The author [Commun. Math. Sci. 3, No. 2, 133--157 (2005; Zbl 1161.65344)], shows how to preserve the \(L^1\)-stability but without any conservation assumption in the gradient reconstruction. Moreover, second-order discrete entropy inequalities are given there. \textit{B. van Leer} [SIAM J. Sci. Stat. Comput. 5, 1--20 (1984; Zbl 0547.65065)] introduced a variant of the MUSCL scheme; namely the MUSCL-Hancock method. This scheme is full time and space second-order accurate. It differs from the other MUSCL variants in its second-order time of accuracy. Indeed, the time accuracy rises when considering a step based on a central-like scheme. The MUSCL-Hancock scheme is considered in the present paper for a stability and robustness analysis. The paper is organized as follows. In the following section, the main notations are introduced and the MUSCL-Hancock method is described. Next, in the third section, the robustness of the scheme under consideration is analyzed. Arguing a relevant limitation procedure to define the gradient reconstruction, the author establishes the \(L^1\)-stability of the scheme. The stability result involves a more restrictive CFL condition than usually met. This may imply several difficulties from a practical point of view. In the next section, a time variant of the MUSCL-Hancock scheme is proposed. It makes easy (more standard) the time CFL restriction. At this level of the paper, the piecewise-linear reconstruction is based on the conservative variables. It is proposed to extend the above analysis within the framework of a gradient reconstruction based on relevant non-conservative variables. To conclude the paper, the last section is devoted to numerical experiments. They are performed within classical Euler equations. The numerical tests show the interest and the robustness of the method.