View Issue Details
| ID | Project | Category | View Status | Date Submitted | Last Update |
|---|---|---|---|---|---|
| 0001783 | OpenFOAM | Bug | public | 2015-07-12 20:04 | 2015-10-29 15:46 |
| Reporter | dhora | Assigned To | henry | ||
| Priority | normal | Severity | minor | Reproducibility | always |
| Status | closed | Resolution | no change required | ||
| Summary | 0001783: interPhaseChangeFoam with MULESCorr: parallel simulations | ||||
| Description | I observed differences between parallel and serial interPhaseChangeFoam simulations. I was able to construct a case where one can see a stuttering in the transport of the phase interface at processor boundaries. I didn't investigate further but it seems as if information is not being updated properly. | ||||
| Steps To Reproduce | Case and solver are attached. Compare the serial and parallel run (16 cores, scotch decomposition). The solutions in front of the leading edge look identical until t=0.055 and different in the next time step. | ||||
| Additional Information | There seems to be no difference between serial and parallel simulations with MULESCorr=no in fvSolution. | ||||
| Tags | No tags attached. | ||||
|
|
|
|
|
|
|
|
Do you see the same problems with OpenFOAM-dev? |
|
|
I can test it with the latest version but need a bit of time. Quite busy at the moment... |
|
|
Made just a quick test and it seems to be even worse. Didn't retest with 24x. |
|
|
What is the purpose of the specialized solver attached? Does the problem only occur with this solver or also the standard interPhaseChangeFoam we provide with OpenFOAM? |
|
|
The case you supplied is rather odd, I am not sure what it is supposed to demonstrate. I have run it in serial and parallel with the solver running without flow and the solutions look very similar, certainly as similar as I would expect. From these results it is not clear that there is any problem with the solver. |
|
|
In order to ensure that running in parallel provides the same results as serial you need to use sufficiently tight tolerances on the solvers. I ran your case with solver smoothSolver; smoother symGaussSeidel; tolerance 1e-10; and get the same results in serial and parallel. |
| Date Modified | Username | Field | Change |
|---|---|---|---|
| 2015-07-12 20:04 | dhora | New Issue | |
| 2015-07-12 20:04 | dhora | File Added: airFoil2D.tar.gz | |
| 2015-07-12 20:05 | dhora | File Added: interPhaseChangeFoamNoFlow.tar.gz | |
| 2015-10-22 10:30 | henry | Note Added: 0005471 | |
| 2015-10-27 07:55 | dhora | Note Added: 0005497 | |
| 2015-10-28 21:39 | dhora | Note Added: 0005510 | |
| 2015-10-29 11:01 | henry | Note Added: 0005512 | |
| 2015-10-29 14:33 | henry | Note Added: 0005514 | |
| 2015-10-29 15:46 | henry | Note Added: 0005515 | |
| 2015-10-29 15:46 | henry | Status | new => closed |
| 2015-10-29 15:46 | henry | Assigned To | => henry |
| 2015-10-29 15:46 | henry | Resolution | open => no change required |
