shock capture in 3D case

[hshnpu]

I use hifiles to compute a Ma=2.8 flow, I found it is very unstable when strong shock occurs. I read the code, found that the artificial viscosity is not added in 3D case(in function shock_capture_concentration_cpu on x,y are coded, see eles.cpp), how to overcome this problem?

[CottonTensor]

3D is not implemented. However you can extend it to 3D without too much extra effort because the shock sensing is done along 1D lines (slices) in x and y directions. So you will need to do it in the z-direction as well.

[hshnpu]

thanks, I am a fresher for this code, I need to know more about this kind of method before implementing, could you share some reference with more detail or simple description about concentration method with me? I also find some papers according to AIAA 2014-2688, but they are too mathmatical to understand.

[eeethon]

Hi, hshnpu. Recently I want to run a 2D inviscid supersonic cylinder case, but the computation always blows up. I set the option of shock-capturing as follows:
==== Shock-Capturing =====
artif_only 1
artif_type 0
ArtifOn 1
epsilon0 0.005
s0 0.0123457
kappa 0.0

the value of s0 is computed with the formula in Persson's paper in form of 1/p^4, epsilon0 is computed with h/p， where h is the minimum size near wall and p is the polynomial order 3. I run the code with a small timestep but the computation blows up before the shock is appeared. Can you help me out of the trouble? Thanks!

[weiqishen]

Hi CottonTensor, I'm running a case which is a supersonic flow of M 1.25 over a flate plate to form a free shear layer. The simulation becomes very unstable and I tried to use shock capturing(concentration method) to stablize it. From the sensor plot I found that the sensor values in the supersonic vortex and shock wave are from 0.0006 to 0.003.However when I set s0 to this value, the simulation crashed in the first step. Please help with this problem, thanks