Hi,

I have implemented the WENO FV scheme. When I try to simulate the Forward facing step (Mach 3), I found that the code blows after certain number of iterations.

I checked that the pressure goes to negative values at some time near the corner of the step.

The paper by Shu mentions about applying an extra boundary condition at the singularity point (corner of the step). But, I am not sure if that is the problem with my current simulation.

I apply the wall boundary condition by reflecting the values on all the ghost cells and also at only one ghost cell, while maintaining others at a near infinite value. Both gives similar solutions. But, which one is best?

If anyone had tried this problem, kindly share your experience.

Thank you very much for your help.

I suggest you to try to fix the singularity, then run again. If problem persists, there has to be a bug somewhere in the code. That is because, I implemented that case with a multiblock grid and it worked well. There shouldn't be a need for an order reduction for Mach 3.

Fixing the singularity which is at the corner of the step is another issue. Try to reset the values according to the the paper of Woodward and Colella. Like, resetting density variable in the first four cells in the first row beginning from the corner, and resetting magnitudes of velocities not their directions in the first two cells, which are located in the second row just above the step.