2bd order convective solvers

Heinz Wilkening

Ciao everybody,

We have developed our own NS-solver for explosion modeling. The grid is based triangles (2D) or tetraherons (3D). Control Volumes are constructed around each vertex of the tri/tet by the medians of the faces of the tri/tet. Between neighbour CV a 1-D Rieman Problem is solved (Gudonov's method), using e.g. Roe's app. Riemansolver. For the 2nd order solver we calculate a gradient inside each CV and use this to extrapolate to the neigbours (van Leer's method for 2nd order solvers). To prevent instabilities the gradient has to be limited. We tried different limiters, all of then limited to much in the 3-D case, so our solution was only 1st order in 3-D. We think the reason might be, that we habe have much more neighbours in the 3-D case than in the 2-D case. Most limiters work fine in 2-D.

Does someone have simular experience. What limiter would you recommend? Thank you.

Heinz