[ashkan]

Hi,
I would like to simulate an oscillatory flow at inlet. I have managed to setup and run the simulation using the "uniformFixedValue" boundary condition for "U". The simulation runs just fine. However, when the flow reverses at the inlet, it creates some unphysical velocity the lower corner where I have no slip wall. Please see the snapshot of the issue. I ran the model for few oscillation cycles and this always happen at each cycle when the flow reverses.

I tried the model with different schemes in the fvScheme dictionary but could not resolve the issue.

Attached my case and would really appreciate any help/comments/suggestions.

Cheers
Ashkan

[jherb]

You give a constant velocity over the whole inlet, but within the domain, you have some turbulent boundary layer profile. This does not work, if the fluid is moving in the direction of the inlet (because you impose a negative velocity).

You might get better results, if you create your own boundary condition, which just imposes the integral flow over the whole inlet and use the cell values of the cells attached to the inlet for the deviations.

Have a look at the fixedMean boundary condition: https://github.com/OpenFOAM/OpenFOAM...ived/fixedMean

[ashkan]

Originally Posted by jherb

You give a constant velocity over the whole inlet, but within the domain, you have some turbulent boundary layer profile. This does not work, if the fluid is moving in the direction of the inlet (because you impose a negative velocity).

You might get better results, if you create your own boundary condition, which just imposes the integral flow over the whole inlet and use the cell values of the cells attached to the inlet for the deviations.

Have a look at the fixedMean boundary condition: https://github.com/OpenFOAM/OpenFOAM...ived/fixedMean

Many thanks for your comments Joachim. I did implement "fixedMean" boundary as well. You are absolutely correct that this is due to the boundary layer and using fixedMean is more stable. However, the resulting "turbulent oscillatory boundary" would not be correct with "fixedMean" approach.

I managed to resolve my stability issue with a sort of "agricultural" way. I divided the bottom wall into two patches with the one close to inlet being free slip and the other one being no-slip. This actually resolved the problem and the boundary layer matches well with experiments. Attached my results.