Accuracy when using upwind for interpolationSchemes in hypersonic flow
 

Hi,

I'm now running some hypersonic problems, such as double cone, concave ramp etc. In the interpolationSchemes, I used:

reconstruct(rho) upwind;
reconstruct(U) upwind;
reconstruct(T) upwind;

I tried all others but they caused the solution to diverge rather quickly. I even tried changing the scheme after running the upwind solution for a while, but the solution still diverges.

1st of all, am I restricted to using upwind scheme because it is upwind?
Secondly, is using the upwind scheme acceptable for accuracy?

Lastly, is there any other option I can do if my current accuracy is not acceptable?

Thanks.

Hi quarkz,

I assume you are running rhoCentral? I have found that Gamma schemes work best for the interpolation schemes. With the standard vanLeer interpolation the temperature solution isn't that great.
If your solution diverges with the other schemes then there is probably an other issue, which is hard to diagnose without more info.

Hi Hillie,

Thanks for the reply. I have tried most other changes, except for y+. The current grid is still too coarse in terms of y+. However, to reach y+ of even 50 at our Re, we will need maybe a few hundred millions to a billion, which is beyond our capability. We will try to see what else can be done.