[qtian]

Hello, all,

I don't have much experience of running the unsteady simulation. I keep getting Bounding warning of k. For the steady case, we can reduce the relaxation factors or change discretization schemes. How about for the unsteady case, if we have similar problem? BTW, checkMesh seems pretty good. Thanks for your help.

Quinn

[ceyrows]

Hello Quinn,

which interpolation scheme do you use for the convectice fluxes (the divSchemes entry in the fvSchemes dictionary)?

I think those bounding warnings occur if you get numerical oscillations that depend on the interpolation scheme and mesh resolution. I had the best experiences with "Gauss limitedLinear 1", which behaves relatively stable and gives, in spite of bounding values, relatively good results.

Regards,

Thomas

[qtian]

Thomas,

I used "Gauss upwind" for the convective fluxes. I thought "Gauss upwind" is first order and relatively stable numerically.I will try you suggestion. Have a good weekend.

Qing

[ceyrows]

Hmmm, this is what I thought, too. As far as I know the upwind scheme is bounded by nature and as you say quite stable. But depending on the mesh resolution, it may also introduce a serious error.
So try the limitedLinear scheme anyway. It is also bounded and stable but avoiding numerical diffusion. I also had bounding warnings but the solution did not diverge.

Have a nice weekend too and good luck on monday,

Thomas

[yousuf]

Hi all,
I have got a question, Can we rely on results if "k" and "epsilon" is bounding for few steps??...... (if it is running for 5000 timesteps without any error message)