[boshynova]

Hi, everyone:

I am validating RANS models with channel flow using OpenFOAM-2.3.1, however, it went wrong with kOmega and kOmegaSST.

The mesh is 2D, meanwhile, 1 cell in the streamwise direction. The bottom and upper boundaries are walls, which y+ =0.7. The left and right side are set to cyclic bc, and the frontAndBack is set to empty.(physical model, mesh and boundary conditions are shown in physical.jpg and mesh&bc.jpg)

kOmega, kOmegaSST, and kEpsilon are used. kqRWallFunction, omegaWallFunction, nutkWallFunction, epsilonWallFunction are used for k, omega, nut, and epsilon.

A pressure gradient term was added to simpleFoam, renaming as simpleFullyDevelopFoam, giving a pressure gradient to drive the fluid flow.
Modification in UEqn.H

    //solve(UEqn() == -fvc::grad(p));
    solve(UEqn() == -fvc::grad(p)-pGrad);

Modification in pEqn.H

    //U = HbyA - rAU*fvc::grad(p);
    U = HbyA - rAU*fvc::grad(p) - rAU*pGrad;

The case file is cavity_pGrad_1cells_H_simple.

The results, however, is very wrong for kOmega.
As can be seen from kOmega-res.jpg, there is a discontinuity of du/dy and dk/dy in the center of the channel. Also, kOmega converged to different results given different initial ux value.
On the other hand, the kEpsilon do not exhibits such misbahavior (kEpsilon-res.jpg).

Given all that shown, I am wondering if there is something wrong with kOmega and kOmegaSST model.
Or, is there something wrong with the method that I added the pressure gradient? Is there a better way to achieve this?

Hope to hear your suggestions.

Best wishes,
boshynova

[Xinze]

The turbulence library has been rewriten in the latest version, OpenFOAM-dev. you should try to use OpenFOAM-dev and check if it is a bug.

xz