A Marching in Space and Time (MAST) solver of the shallow water equations. Part I: The 1D model

Abstract

A new approach is presented for the numerical solution of the complete 1D Saint-Venant equations. At each time step, the governing system of Partial Differential Equations (PDEs) is split, using a fractional time step methodology, into a convective prediction system and a diffusive correction system. Convective prediction system is further split into a convective prediction and a convective correction system, according to a specified approximated potential. If a scalar exact potential of the flow field exists, correction vanishes and the solution of the convective correction system is the same solution of the prediction system. Both convective prediction and correction systems are shown to have at each x-t point a single characteristic line, and a corresponding eigenvalue equal to the local velocity. A MArching in Space and Time (MAST) technique is used for the solution of the two systems. MAST solves a system of two Ordinary Differential Equations (ODEs) in each computational cell, using for the time discretization a self-adjusting fraction of the original time step. The computational cells are ordered and solved according to the decreasing value of the potential in the convective prediction step and to the increasing value of the same potential in the convective correction step. The diffusive correction system is solved using an implicit scheme, that leads to the solution of a large linear system, with the same order of the cell number, but sparse, symmetric and well conditioned. The numerical model shows unconditional stability with regard of the Courant-Friedrichs-Levi (CFL) number, requires no special treatment of the source terms and a computational effort almost proportional to the cell number. Several tests have been carried out and results of the proposed scheme are in good agreement with analytical solutions, as well as with experimental data.