View Issue Details

IDProjectCategoryView StatusLast Update
0002020OpenFOAMBugpublic2016-03-08 15:28
ReporterShorty Assigned Tohenry  
PriorityhighSeveritycrashReproducibilityalways
Status resolvedResolutionfixed 
PlatformGNU/LinuxOSUbuntuOS Version14.04
Summary0002020: Field access somehow wrong
DescriptionDear all,

I am sure this is a critical bug. Using fixedShearStress BC will fail always due to division by zero here:

operator==(tauHat*(tauHat & (tau0_*(1.0/(ry*nuEff)) + Uc)));

The problem is that nuEff is initialized wrong (field is 0());

    const label patchI = patch().index();

    const surfaceScalarField& phi =
        db().lookupObject<surfaceScalarField>(phiName_);

    scalarField nuEff;
    if (phi.dimensions() == dimVelocity*dimArea)
    {
        const incompressible::turbulenceModel& turbModel =
            db().lookupObject<incompressible::turbulenceModel>
            (
                "turbulenceModel"
            );

        //- This is somehow wrong
        nuEff = turbModel.nuEff()()[patchI];

        //- Output it Info<< nuEff -> 0()
    }

Interesting that Info<< nuEff << is different to Info<< turbModel.nuEff()()[patchI]

Maybe its correct but in my opinion nuEff should have the values of nu of each face at the patch patchI.


Steps To ReproduceFor incompressible solvers, it does not matter. I always used simpleFoam... division by zero.
TagsNo tags attached.

Activities

Shorty

2016-03-08 12:26

reporter  

ABLz0FSS.zip (1,045,205 bytes)

henry

2016-03-08 12:48

manager   ~0006008

It is not clear from your description how your case is setup such that both the laminar and turbulent components of the viscosity are both zero.

To study the problem further we will need a test-case for OpenFOAM-3.0.x or OpenFOAM-dev which reproduces the problem.

Shorty

2016-03-08 13:26

reporter   ~0006009

Last edited: 2016-03-08 13:43

Dear Henry,

I am sorry for bad explanation. I will prepare a FOAM3.0 case but at least it should be the same. Just for clearness... the output of the field turbModel.nuEff() will show me all boundaries with non-zero values (everything is correct there).

For example: The patch name is wall1, therefore I got the output:

    wall1
    {
        type calculated;
        value nonuniform List<scalar>
500
(
0.285685
0.285685
0.285685
0.285685


Please correct me if I am wrong: turbModel.nuEff()()[patchI] should return the face values of the patch (patchI), or am I wrong?

But as far as I understand operator()()[patchI], it will return a scalar and not a scalar field.

henry

2016-03-08 13:48

manager   ~0006010

It is not clear to me what is wrong with your case setup and I will look into it whe you provide a version for OpenFOAM-3.0.x or dev.

MattijsJ

2016-03-08 13:55

reporter   ~0006013

The code seems to use the patch index (patchi) to index into the internalField of a (vol)Field. Instead index into the boundaryField(), something like:

const scalarField& bla = turbModel.nuEff()().boundaryField()[patchI];

henry

2016-03-08 14:02

manager   ~0006014

scalarField nuEff(turbModel.nuEff(patch().index()));

is cleaner and more efficient.

henry

2016-03-08 14:19

manager   ~0006017

I have made this change to

OpenFOAM-dev: commit 9ef051dd18a345f96ea5e9380f82c5f5f20e96ab
OpenFOAM-3.0.x: commit 09fa0b46eb4394f632f046b6c41fe57e8a140c7c

Shorty

2016-03-08 14:57

reporter   ~0006021

Thanks henry,

I changed it locally and now it is working. I think there is no need for a test case anymore now, or?

henry

2016-03-08 15:27

manager   ~0006023

No need for the test-case, I will close this report.

henry

2016-03-08 15:28

manager   ~0006024

Resolved in OpenFOAM-dev by commit 9ef051dd18a345f96ea5e9380f82c5f5f20e96ab
Resolved in OpenFOAM-3.0.x by commit 09fa0b46eb4394f632f046b6c41fe57e8a140c7c

Issue History

Date Modified Username Field Change
2016-03-08 12:26 Shorty New Issue
2016-03-08 12:26 Shorty File Added: ABLz0FSS.zip
2016-03-08 12:48 henry Note Added: 0006008
2016-03-08 13:26 Shorty Note Added: 0006009
2016-03-08 13:27 Shorty Note Edited: 0006009
2016-03-08 13:43 Shorty Note Edited: 0006009
2016-03-08 13:48 henry Note Added: 0006010
2016-03-08 13:55 MattijsJ Note Added: 0006013
2016-03-08 14:02 henry Note Added: 0006014
2016-03-08 14:19 henry Note Added: 0006017
2016-03-08 14:57 Shorty Note Added: 0006021
2016-03-08 15:27 henry Note Added: 0006023
2016-03-08 15:28 henry Note Added: 0006024
2016-03-08 15:28 henry Status new => resolved
2016-03-08 15:28 henry Fixed in Version => 3.0.x
2016-03-08 15:28 henry Resolution open => fixed
2016-03-08 15:28 henry Assigned To => henry