View Issue Details
ID | Project | Category | View Status | Date Submitted | Last Update |
---|---|---|---|---|---|
0001398 | OpenFOAM | Bug | public | 2014-09-12 17:21 | 2015-03-02 19:15 |
Reporter | Assigned To | henry | |||
Priority | high | Severity | major | Reproducibility | always |
Status | resolved | Resolution | fixed | ||
Platform | Linux | OS | Ubuntu | OS Version | 13.10 |
Summary | 0001398: compressibleInterFoam: Different results between in parallel and in serial computing | ||||
Description | Hello, I'm using compressibleInterFoam for my project of supersonic jet and liquid flow (compressible 2-phase flow). Before using this solver for my project, I made simple test case. Test case is "Supersonic flow over a forward-face step". I ran simulation for rhoPimpleFoam and compressilbeInterFoam both in parallel(4 processors) and serial computing. So, I turn on transonic in the FVsolution to capture supersonic flow. For rhoPimplFoam, simulation results are the same both in parallel and serial. However, for compressibleInterFoam, simulation results are different between in serial and parallel. Discontinuities of pressure distribution are observed at the interface between sub-domains. I tested for openfoam 2.2.2 and openfoam 2.3.0, both of them have same problem. And this problem happens the computational domain of my project. I've attached case files and results at 2(s) for rhoPimpleFoam and compressibleInterFoam both in parallel and serial computing. | ||||
Tags | compressibleInterFoam, parallel, Parallel computing | ||||
2014-09-12 17:21
|
|
|
In the case of simulation using compressibleInterFoam, I turn off second fluid flow feature to make simulation conditions for test case. In that case, I think that this solver is the same as compressible 1-phase flow solver. |
|
Using fvc::div(phid1, p_rgh) instead of fvm::div(phid1, p_rgh) and fvc::div(phid2, p_rgh) instead of fvm::div(phid2, p_rgh) seems to fix the problem in 2.3.x |
|
I tried to use fvc::div(phid1, p_rgh) instead of fvm::div(phid1, p_rgh) and fvc::div(phid2, p_rgh) instead of fvm::div(phid2, p_rgh). This seems to fix the problem of pressure discontinuity at the interface between sub-domains. However, the results is not in agreement with other test cases(rhoPimpleFoam or sonicFoam). |
|
Using fvc::div(phid?, p_rgh) rather than fvm::div(phid?, p_rgh) is equivalent to using the sub-sonic algorithm, have you tried this? The solution will indeed be different to using rhoPimpleFoam or sonicFoam because it is based on a compressible correction to div(U) to support the very large phase density differences typical of multiphase systems. |
|
I just tested the case with OpenFOAM-2.3.x and OpenFOAM-dev and see on features associated with the decomposition -- the solutions look identical to those from serial running. |
Date Modified | Username | Field | Change |
---|---|---|---|
2014-09-12 17:21 |
|
New Issue | |
2014-09-12 17:21 |
|
File Added: forwardStepCIFparallel.zip | |
2014-09-12 17:25 |
|
Tag Attached: compressibleInterFoam | |
2014-09-12 17:25 |
|
Tag Attached: Parallel computing | |
2014-09-12 17:25 |
|
Tag Detached: Parallel computing | |
2014-09-12 17:26 |
|
Tag Attached: Parallel computing | |
2014-09-12 17:26 |
|
Tag Attached: parallel | |
2014-09-12 17:34 |
|
Note Added: 0003233 | |
2014-11-03 20:48 | dhora | Note Added: 0003277 | |
2015-03-02 17:13 |
|
Note Added: 0003944 | |
2015-03-02 18:18 | henry | Note Added: 0003945 | |
2015-03-02 19:15 | henry | Note Added: 0003946 | |
2015-03-02 19:15 | henry | Status | new => resolved |
2015-03-02 19:15 | henry | Resolution | open => fixed |
2015-03-02 19:15 | henry | Assigned To | => henry |