View Issue Details
| ID | Project | Category | View Status | Date Submitted | Last Update |
|---|---|---|---|---|---|
| 0003094 | OpenFOAM | Bug | public | 2018-10-18 15:25 | 2019-02-26 16:00 |
| Reporter | sarathkrishnan11 | Assigned To | henry | ||
| Priority | normal | Severity | minor | Reproducibility | always |
| Status | closed | Resolution | no change required | ||
| Platform | Linux | OS | Ubuntu | OS Version | 18.01 |
| Summary | 0003094: problem with createBaffle updating pointLevel | ||||
| Description | While creating cyclic patches using createBaffle, the boundaryfield of the file 0/pointLevel is not updated with the new cyclic patches. Even though it works well in serial with a single processor, it is not working while trying to decompose the case and run in parallel. | ||||
| Additional Information | --> FOAM FATAL IO ERROR: Cannot find patchField entry for cyclic porous_half0 Is your field uptodate with split cyclics? Run foamUpgradeCyclics to convert mesh and fields to split cyclics. file: /home/sarath/OpenFOAM/sarath-5.0/run/case121/0/pointLevel.boundaryField from line 28 to line 36. | ||||
| Tags | No tags attached. | ||||
|
|
Have you tried version 6 and/or dev? Could you please upload a case which reproduces the problem? |
|
|
Yes I have tried it on both version 6 and dev-8f3f2f7365c1. The problem exists. Here is the case files for version 6 or dev |
|
|
Is the casefile too big for you just to test, I could make one simple if you need? Regards |
|
|
No, sorry the case file is fine. I can reproduce the problem. There is no easy way of fixing this. As it stands, your solution is to delete the point fields before running createPatch or set them up with the final set of patches, put them in a separate directory whilst createPatch runs, and then copy them in place after. The issue is that the createBaffle utility (and fvMeshTools on which it depends) does a manual clear-out and then adds the patches to the mesh in-place. The pointMesh gets destroyed by this operation, which invalidates all the point fields. So, the point fields ca n't be included in the conversion that the utility does. Changes like this are supposed to go through the topology engine, which would cause the pointMesh and pointFields to get mapped. The patch addition needs to be rewritten in this framework. Fixing this is a big job. Are you funding this sort of maintenance work (https://openfoam.org/maintenance/)? |
|
|
Okay. I am using this temporary solution as well. At the moment, this is not a priority bug for me. And regarding funding for this maintenance work, I worry I can not since I am a student learning OpenFOAM (Not using it for business purposes). But I feel sorry to say this. Thank you for your time and hope it would be fixed sometime in future. |
|
|
Currently createBaffles does not manipulate the pointMesh or pointFields but even if it did the case provided has field handing switched-off in the createBafflesDict so either way the pointLevel field in the 0 directory would be inconsistent with the final mesh. The best general solution to this problem is to set snappyHexMesh to not write the pointLevel file, which is now the default: // Write flags writeFlags ( // scalarLevels // Do not write the pointLevel field . . . ); If the pointLevel field file is needed for diagnostics it should be deleted before createBaffles or other mesh manipulations are run. |
| Date Modified | Username | Field | Change |
|---|---|---|---|
| 2018-10-18 15:25 | sarathkrishnan11 | New Issue | |
| 2018-10-19 09:02 | will | Note Added: 0010113 | |
| 2018-10-19 09:59 | sarathkrishnan11 | File Added: BaffleNet.tar.xz | |
| 2018-10-19 09:59 | sarathkrishnan11 | Note Added: 0010114 | |
| 2018-11-07 08:42 | sarathkrishnan11 | Note Added: 0010149 | |
| 2018-11-07 10:39 | will | Note Added: 0010150 | |
| 2018-11-07 11:09 | sarathkrishnan11 | Note Added: 0010151 | |
| 2019-02-26 16:00 | henry | Assigned To | => henry |
| 2019-02-26 16:00 | henry | Status | new => closed |
| 2019-02-26 16:00 | henry | Resolution | open => no change required |
| 2019-02-26 16:00 | henry | Note Added: 0010334 |
