View Issue Details

IDProjectCategoryView StatusLast Update
0003590OpenFOAMBugpublic2020-11-25 14:25
ReporterChrisK15Assigned Towill 
PrioritynormalSeverityminorReproducibilityalways
Status closedResolutionsuspended 
Product Version8 
Fixed in Version 
Summary0003590: All dirs are valid for wedge geometries mesh.nSolutionD() -> Lagranian problems
DescriptionHi,
all directions are reported as valid on wedge geometries using mesh.nSolutionD().
Check tutorial tutorials/multiphase/interFoam/LES/nozzleFlow2D with "checkMesh"

Report is:
    Mesh has 2 geometric (non-empty/wedge) directions (1 1 0)
    Mesh has 3 solution (non-empty) directions (1 1 1)

This causes problems at least with dsmcFoam which uses nSolutionD
do constrain the velocity check:
https://github.com/OpenFOAM/OpenFOAM-dev/blob/master/src/lagrangian/DSMC/parcels/Templates/DSMCParcel/DSMCParcel.C#L60
since simulations abort when the wedge face is hit.
I haven't check other lagranian solvers...
A possible solution would be to use:
https://github.com/OpenFOAM/OpenFOAM-dev/blob/master/applications/utilities/mesh/manipulation/checkMesh/checkGeometry.C#L602
to constrain the velocity.

Regards,
Christoph
TagsNo tags attached.

Activities

ChrisK15

2020-11-13 18:25

reporter   ~0011683

Well it's an edge case, since p.deviationFromMeshCentre() lets the particle move to the center...
Only noticed this because of another problem in my personal code, but still I think using geometricD() to constrain the velocity is better suited, correct me if I am wrong.

will

2020-11-18 12:17

manager   ~0011689

I agree. In a wedge case non-zero tangential velocities should be permitted, but tangential changes in position should not. So the velocity should be constrained with solutionD, but tracking displacement should be constrained with geometricD. We could trivially fix it so that is the case in both the standard Lagrangian models and in DSMC.

However, it would still be wrong. In a 3-D cyclic case, when a particle crosses a cyclic patch it is reintroduced on the other side with a transformation of it's velocity (and other vector and tensor properties). In wedges, this transformation process should also occur, but continually as a result of any out of plane velocity. This is not currently the case. No such process is implemented for wedge geometries.

So, at present, no Lagrangian model in OpenFOAM can be considered to support wedge geometries. Implementing the necessary transformation processes for Lagrangian to function correctly on wedges is feasible with a modest amount of effort. Are you able to fund such a development?

ChrisK15

2020-11-24 15:38

reporter   ~0011755

At least for DSMC there aren't any out of plane velocities, so fixing them to the center plane of the wedge should be okay. Only problem with DSMC then are particles colliding with the centerline (as part of the wedge boundary). Right now after this initial test, I myself do not have the need for a wedge geometry anymore. And I am really sorry, I am not in the position of funding an extension of OF.

will

2020-11-25 14:25

manager   ~0011759

> At least for DSMC there aren't any out of plane velocities

In DSMCParcel::move, there is explicit explanation of the fact that out-of-plane velocities need to be retained in reduced dimensionality cases. The 2D DSMC tutorials all feature non-zero out-of-plane velocities. Only the tracking is constrained.

I do not see a bug here. It is simply a limitation that the Lagrangian library simply does not support wedge geometries when there is any out of plane motion. In the case of DSMC, that is always the case.

> I am not in the position of funding an extension of OF.

Then this report is closed, pending funding for extension of Lagrangian.

Issue History

Date Modified Username Field Change
2020-11-13 15:51 ChrisK15 New Issue
2020-11-13 18:25 ChrisK15 Note Added: 0011683
2020-11-18 12:17 will Note Added: 0011689
2020-11-24 15:38 ChrisK15 Note Added: 0011755
2020-11-25 14:25 will Assigned To => will
2020-11-25 14:25 will Status new => closed
2020-11-25 14:25 will Resolution open => suspended
2020-11-25 14:25 will Note Added: 0011759