CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   OpenFOAM (http://www.cfd-online.com/Forums/openfoam/)
-   -   pressure spikes due to addition/deletion of layers (http://www.cfd-online.com/Forums/openfoam/93740-pressure-spikes-due-addition-deletion-layers.html)

kdneroorkar October 25, 2011 08:45

pressure spikes due to addition/deletion of layers
 
Hello All
I was running icoDyMFoam on the tutorial case movingConeTopo. I added one line to icoDyMFoam to write out the maximum pressure in the domain at every time step. I found that almost every time there is a cell addition/deletion, there is a pressure spike in the domain. Has anyone else seen these spikes? and has anyone found a solution?
Thanks
Kshitij

kalle October 26, 2011 03:48

Hi,

Which version are you using? The implementation of the movingCone(or Body)TopoMesh is altered in later revisions of the 1.6-ext version. In the newer (and parallel) version, a lot of lines are commented out in Foam::movingBodyTopoFvMesh::update() . These lines affects how meshPhi (in fvc::makeAbsolute(phi, U) and fvc::makeRelative(phi, U)) is determined (if I understood correctly). If meshPhi is wrong, I would guess it could result in pressure peaks. Do you have any possibility to repeat your check on a previous, non-parallel version of the movingCone/Body case?

Kalle

kdneroorkar October 27, 2011 09:48

Hi Kalle
Thanks for your post. We updated to the latest version of 1.6-ext to see what difference it made, and we found that the spikes were considerably smaller, but still existed.
if you/ anyone else would like to test this, you can just add the following functionObject to the controlDict and run the movingConeTopo test case with icoDyMFoam.
Thanks a lot for your help
Kshitij


functions{
fieldMinMax
{
type fieldMinMax;
functionObjectLibs ("libfieldFunctionObjects.so");
log yes;
outputControl timeStep;
outputInterval 1;
mode magnitude;
fields
(
U
p
);
}
}

kalle October 28, 2011 03:35

Hi,

Thank you for bringing this up. I tried the case, and there are spikes as you say. Pressure is initially fluctuating around ~4e-4 to ~1e-4 as layerAR is working. I tried a few modifications to the movingBodyTopoFvMesh-class without success. Right now I do not have any clue how to get on. To solve this issue, I think one should find out how this adding layer functionality is handled within the finite volume method. This paper may provide some help:

http://web.univ-ubs.fr/limatb/EG2M/D...pers/01178.pdf see especially section 2.3

I also need to fix this for my own work, so I will likely get back to it when I have time.

Regards,
Kalle

kdneroorkar October 28, 2011 08:58

Hi Kalle
Thanks for looking into this and thanks for the paper. Please let me know if you find anything.
Kshitij

jimteb November 29, 2014 13:33

Hi,

I can see this thread is quite old, but I would like to know if this problem was ever resolved?

I am having the same issue but with pimpleDyMFoam with layer addition/removal and translation (foam-extend 3.1). I have tried to resolve this by mostly messing around with the following settings, but with no real success:

nCorrectors - increase
nNonOrthogonalCorrectors - increase
nOuterCorrectors - increase
relaxation factor for pressure - reduce
use of smoother & preconditioner with solver
minimum layer thickness in dynamicMeshDict - increase
Gauss upwind for div schemes and Guess linear limited for laplacian schemes zeroGradient for all pressures at boundary patches
relTol for all solvers set to 0

Any advice would be greatly appreciated

Thanks


All times are GMT -4. The time now is 01:30.