[ander]

I had a look at the SU2 tutorials and notices something that seemed strange to me. For the case "inv_wedge_HLLC" no slope limiter is used, even though this is a supersonic case involving shocks. Could someone please explain why it is set up like this? I have long had the impression that slope limiters were necessary at discontinuities to avoid spurious oscillations, but are less important in smooth regions. Does this only apply to explicit solvers?

[wallym]

Hi Ander,

Your reasoning is valid. However, these oscillations can depend on the convective numerical scheme and the flow problem itself. The wedge test case isn't a validation case, so the impact of any numerical instabilities isn't much of a concern.

Hope that helps,
Wally

[ander]

Thanks for your reply, however I don't think everything is addressed yet.

My understanding is that the wedge flow case is basically a Riemann problem with a transverse velocity component, where the analytical solution is known. Using HLLC with Muscl scheme for Riemann problems did require slope limiters based on my previous experience. If oscillations were present, I guess the solution wouldn't converge as well as it did in this case. Would be interesting to hear the reasoning behind it from the person that set up the case.

[wallym]

Correct, the analytical solution is known and the tutorial may not exactly match said solution. The tutorial problem is meant to serve as a baseline to set up supersonic flows, not necessarily serve as a validation case.

The oscillations, if large, could limit the converges or degrade solution quality. Limiters are not a necessity to simulate/converge supersonic problems but do provide better accuracy.

As for the person who wrote that tutorial, I imagine they have since moved on from working on SU2.