View Issue Details

Jump to NotesJump to History
IDProjectCategoryView StatusDate SubmittedLast Update
0002228OpenFOAMBugpublic2016-09-01 22:322016-09-02 22:10
Reporterrlferreira Assigned Towyldckat  
PrioritynormalSeveritymajorReproducibilityalways
Status closedResolutionno change required 
PlatformGNU/LinuxOSUbuntuOS Version14.04
Summary0002228: Mesh quality decrease with renumberMesh functionality
DescriptionThe checkMesh after running "renumberMesh -overwrite" points bigger Non-Orthogonality on the mesh. In my case, maximum non-orthogonality increases of 66 to 89° after execute renumberMesh. I did a test with pitzDaily case.
Before running renumberMesh:
Non-Orthogonality max=5.94745 average=1.41505
Max skewness=0.260443
Max Aspect Ratio=7.74923

After running renumberMesh:
Non-Orthogonality max=5.97551 average=1.41506
Max skewness=0.260406

It is clear that pitzDaily case have a very good mesh quality. So, the effect is really small. However, if the mesh have high Non-orthogonality, as it is my case, I think that bad mesh quality becomes.
Max Aspect Ratio=7.74998
Steps To ReproduceExecute:

1)blockMesh
2)checkMesh
3)renumberMesh -overwrite
4)checkMesh
Additional InformationI expected the numbers of mesh quality remaining the same after execute renumberMesh.
TagsNo tags attached.

Activities

rlferreira

2016-09-01 22:32

reporter  

fixedReed.tar.gz (10,830 bytes)

MattijsJ

2016-09-02 11:01

reporter   ~0006799

You're seeing the effect of limited precision in writing meshes. Use binary writing or a higher writePrecision.

wyldckat

2016-09-02 22:08

updater   ~0006802

Confirmed Mattijs' assessment, many thanks!

To add a few more details: blockMesh will write with precision of at least 10 digits, as shown here: https://github.com/OpenFOAM/OpenFOAM-4.x/blob/master/applications/utilities/mesh/generation/blockMesh/blockMesh.C#L329

A few additional deductions/conclusions:
  - Having renumberMesh do the same can be construed as counter-intuitive, given that this kind of control is left to the user.

  - blockMesh only does this change of precision out of convenience.

  - writing in ASCII format should only be used as a debugging measure. Writing and reading in binary format is much more efficient (much faster to load/save to/from files, as well as save disk space in many cases) and does not loose precision.


I'm therefore closing this report as "no change required".

Issue History

Date Modified Username Field Change
2016-09-01 22:32 rlferreira New Issue
2016-09-01 22:32 rlferreira File Added: fixedReed.tar.gz
2016-09-02 11:01 MattijsJ Note Added: 0006799
2016-09-02 22:08 wyldckat Note Added: 0006802
2016-09-02 22:10 wyldckat Status new => closed
2016-09-02 22:10 wyldckat Assigned To => wyldckat
2016-09-02 22:10 wyldckat Resolution open => no change required