Pressure varies with cell size!!
 

Hi,

I was using icoFoam and interFoam and noticed that the pressure range (icoFoam) or pressure value (interFoam) varies with cell size and does not seem to converge with refinement. The smaller the cell size, the larger the pressure. If you use the cavity icoFoam case and refine the mesh you see that pressure range continuously increases.

Does somebody know what is wrong or experienced this problem?

Is this a problem due to Rhie-Chow interpolation?

Anyone? :(

I think this is a problem with the boundary conditions on pressure - take a look at the boundary condition on 'p'. May be an issue with the cells at the intersection of the moving wall and the side walls - you can see the pressure peaks in those corners. I've had difficulties with pressure boundary conditions before in openfoam, they can be tricky at times.

As I said in 1st post, this also happens with interFoam. Even if the pressure is normalized or isn't absolute pressure whatever it should not depend on cell size. It seems to vary linearly with the inverse of cell distance. For example if I have a 1D mesh and have 20 cells the pressure is p, if I have 40 cells the pressure is 2*p and so on.

To repliclate this problem just use interFoam with 1 inlet and 1 outlet a simple plate for example 1m x 0.4m and place a alpha=1 at inlet. Probe the pressure near inlet and refine the mesh. Put a velocity of 0.1 m/s at inlet.
Start with mesh as 25x10 cells and then put 50x20 and then 75x30, etc. and see what happens to pressure.

Jugghead, can you try 100x100 and then 300x300 and report what pressure is doing at these stages compared to 75x30?

Attention that this is not turbulent flow. So it should converge pretty easily without resorting to high density meshes. Low Reynolds. You can use a very high viscosity and this happens.

Hello! I try to simulate processes in multisection cyclone unit with OpenFOAM. Previously I did an experiment and experimental pressure values are in 2-3 times lower than simulated values. Velocity values are similar to experimental.

Tell me please, is there another way to obtain normal values, consistent with experimental values, except splitting the mesh? May be it's possible to set some solver's parameters or even change settings in solver's source code?

I tried to simulate my case with this solvers/settings: simpleFoam & pimpleFoam solvers, limitedLinear, linearUpwind & QUICK schemes for div(phi,U), GAMG linear solver for pressure and smoothSolver for other fields with under-relaxation factors 0.3 for pressure and 0.7 for other fields. With all of that, velocity values are consistent with experimental, but pressure values are very big.

Thank you!