Problem with scalarTransportFoam illistrated using pitzDaily tutorial
 

Hi All,

I have been using scalarTransportFoam to simulate chemical transport in a very complex geometry. While the value of my scalar should remain between 0 and 1, I see values as high as 2.5.

To trouble-shoot this problem I went to the scalarTransportFoam tutorial pitzDaily. When the diffusivity (DT) is set to 0.01 in transportProperities the maximum value of the scalar should remain between 0 and 1, but goes as high as ~1.007 - this is not a big problem.

However, if the diffusivity (DT) is changed to 1e-6 and the tutorial is re-run, the scalar value goes to ~1.43, with the maximum value along the wall. This level of error is unacceptable.

Additionally, I tried changing the fvSchemes from the default tutorial settings to:
div(phi,T) Gauss upwind;
laplacian(DT,T) Gauss linear uncorrected;
This did not help the problem.

Does anyone have any suggestion on settings for the fvSchemes dictionary that would fix this problem?

Thanks in advance,

-Mike

Hi again,

The problem appears to be with "createPhi.H".

If you run the simpleFoam tutorial pitzDaily and copy U and phi at the final timestep over to the scalarTransportFoam pitzDaily tutorial, then rerun scalarTransportFoam the solution remails bounded between 0 and 1. If you then move phi to phi.hold and rerun scalarTransportFoam, the solution is no longer bounded between 0 and 1.

Additionally, the phi that is created by scalarTransportFoam is significantly different from the phi.hold which was written by the simpleFoam solver. In many cells, the phi value differs by 10% or more.

I'm at a loss for what the problem could be here... Anyone have any ideas?

-Mike

Im not sure if you are still interested but I had similar bounding issues with turbulent passive scalar transport. If you look at the scalar transport equations for say the Spallart-almaras or k-epsilon models then that might help, since they use an implicit/explicit source term. Also, look a the tread:

http://www.cfd-online.com/Forums/ope...lations-2.html

I hope this helps.

Dan