[hg2lf]

Hello all,
I have used rhoCentralFoam to calculate a supersonic nozzle. When I use the upwind scheme, it's all right. But when I change the interpolation scheme to the second order scheme, such as quadraticUpwind, there is a zone next to the wall boundary in which the velocity is lager than the main steam.See the attachment, the two figures show the mach number distribute in the nozzle.
Could someone tell me that why does this happen?

The following is the fvSchem file:

fluxScheme Kurganov;

ddtSchemes
{
default Euler;
}

gradSchemes
{
default Gauss linear;
}

divSchemes
{
default none;
div(tauMC) Gauss linear;
div(phi,k) Gauss Minmod phi 1;
}

laplacianSchemes
{
default Gauss linear corrected;
}

interpolationSchemes
{
default linear;

interpolate(rho) quadraticUpwindFit phi 1;

reconstruct(rho) quadraticUpwindFit 1;
reconstruct(U) quadraticUpwindFit 1;
reconstruct(T) quadraticUpwindFit 1;
reconstruct(K) quadraticUpwindFit 1;

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

reconstruct(rho) Minmod;
reconstruct(U) MinmodV;
reconstruct(T) Minmod;
reconstruct(Yi) Minmod;


reconstruct(rho) linearUpwind phi;
reconstruct(U) linearUpwindV phi;
reconstruct(T) linearUpwind phi;
reconstruct(K) linearUpwind phi;
*/
}

snGradSchemes
{
default corrected;
}

[immortality]

hi
Is you mesh fine enough near wall?
Does it occur in linearUpwind too?
Or Have you found out the reason?

[hg2lf]

Thanks,immortality,
The case file is too large to upload. I'm not sure whether the mesh is fine enough. It occur in any other scheme except upwind. But the un-physical solution is not in the boundary area, and it develops from the throat. I'll go on to find out the reason. Best regards.

[hg2lf]

It seems as it linearUpwind scheme works well.

[immortality]

how did you make the mesh?
then the problem is resolved?

[hg2lf]

Sorry, there comes the latest news, the problem haven't been solved. The finest mesh is the one associated with the characteristc line.

[hg2lf]

If I using an apreciate discrete schem to solve the subsonic part of the flow well, then it will be fine, Because the high value region develops from this part.