[Jelle Lagerweij]

Dear cfd friends,

I having trouble with getting my first real OpenFOAM simulation setup right. I am trying to model a sphere in a square duct, but somehow, I can't seem to get the parallelized version to work as it should.


I created a tar.gz zip with my full model included, and added two files with the run commands I use: Allrunserial and Allrunparallel contain the commands I use to run my model. I want to set it up like the motorbike example, where I can run 1 potentialFoam initialisation and then run simpleFoam as my main program afterwards.


The Allrunparallel already breaks during the potentialFoam command, with the patchField entry error. The Allrunserial gives an error only when I start the simpleFoam main program. I think both of these error messages come from errors in my boundary conditions, but I can't find the place that I go wrong. (I did check semi-columns though).


I would be really thankful for some help,
Jelle Lagerweij


PS: later on, I will make the sphere leak a bit of hydrogen and investigate the gas-gas mixing that will occur. I need to change it to reactingFoam for that. The final goal is a safety assessment for a hydrogen production location. Obviously, the ventilation should be good enough that the by leakage, the hydrogen concentrations stay below the safety limits. It is a fun in-depth project for a master student in his internship .

[Yann]

Hi,

With your current workflow, just remove the -copyZero option on your second decomposePar command and you should be fine.

What error do you get when running simpleFoam with Allrunserial?

Yann

[Jelle Lagerweij]

Thanks a lot, after your comment, I tweaked around a little. I immediately noticed that I do not remove the decompose folders correctly. I tried to, but didn't command it to search for the processor<> folders in the right directory. I believe that it means that when I decomposed it again the second time, nothing actually happened. I solved that now as well. However, now I get the error message when decomposing the second time. (I attached the new version).


The error for the parallel case is:

--> Number of processor faces = 12894
Max number of cells = 119705 (0.313694% above average 119331)
Max number of processor patches = 5 (25% above average 4)
Max number of faces between processors = 5013 (16.6356% above average 4298)

Time = 0


--> FOAM FATAL IO ERROR: 
Cannot find patchField entry for Wall

file: /home/xintc/OpenFOAM/xintc-9/run/sphere_in_duct/0/k/boundaryField from line 25 to line 44.

I also looked back to the serial case I tried. There I see the same error message, but at least I can see that the p and U fields seem to be accepted. (that at least makes me happy)

--> Reading field p

Reading field U

Reading/calculating face flux field phi

Selecting incompressible transport model Newtonian
Selecting turbulence model type RAS
Selecting RAS turbulence model kOmegaSST
Selecting patchDistMethod meshWave


--> FOAM FATAL IO ERROR: 
Cannot find patchField entry for Wall

file: /home/xintc/OpenFOAM/xintc-9/run/sphere_in_duct/0/k/boundaryField from line 25 to line 44.

    From function void Foam::GeometricField<Type, PatchField, GeoMesh>::Boundary::readField(const Foam::DimensionedField<TypeR, GeoMesh>&, const Foam::dictionary&) [with Type = double; PatchField = Foam::fvPatchField; GeoMesh = Foam::volMesh]
    in file /home/ubuntu/OpenFOAM/OpenFOAM-9/src/OpenFOAM/lnInclude/GeometricBoundaryField.C at line 184.

FOAM exiting

[Jelle Lagerweij]

Thank you very much,
After your tip, I fiddled around a little again. I noticed two things, first of all, I removed the old processor* folders incorrectly (forgot to make it run in the directory that is used in the terminal). So now I improved on that. Although I still can't make it till the simpleFoam command without errors. It breaks when decomposing it for the second time.

Number of processor faces = 12894
Max number of cells = 119705 (0.313694% above average 119331)
Max number of processor patches = 5 (25% above average 4)
Max number of faces between processors = 5013 (16.6356% above average 4298)

Time = 0


--> FOAM FATAL IO ERROR: 
Cannot find patchField entry for Wall

file: /home/xintc/OpenFOAM/xintc-9/run/sphere_in_duct/0/k/boundaryField from line 25 to line 44.

If I do the serial version of the run, I do get too the simpleFoam command without errors. However, then I get this back:

Create time

Create mesh for time = 0


SIMPLE: No convergence criteria found

Reading field p

Reading field U

Reading/calculating face flux field phi

Selecting incompressible transport model Newtonian
Selecting turbulence model type RAS
Selecting RAS turbulence model kOmegaSST
Selecting patchDistMethod meshWave


--> FOAM FATAL IO ERROR: 
Cannot find patchField entry for Wall

file: /home/xintc/OpenFOAM/xintc-9/run/sphere_in_duct/0/k/boundaryField from line 25 to line 44.

    From function void Foam::GeometricField<Type, PatchField, GeoMesh>::Boundary::readField(const Foam::DimensionedField<TypeR, GeoMesh>&, const Foam::dictionary&) [with Type = double; PatchField = Foam::fvPatchField; GeoMesh = Foam::volMesh]
    in file /home/ubuntu/OpenFOAM/OpenFOAM-9/src/OpenFOAM/lnInclude/GeometricBoundaryField.C at line 184.

FOAM exiting

At least, they point at the same problem right now. It also looks like my p and U fields and boundary conditions get excepted as they are read during the simpleFoam command.


With kind regards,
Jelle

[Yann]

Hi Jelle,

There is a closing bracket missing on the Inlet boundary condition on your k file. This is probably why simpleFoam is having trouble to read your BC on k.

Regards,
Yann

[Jelle Lagerweij]

Hi Yann,
Thanks a lot for your help. Somehow it seems to work now . I still had another error somewhere (I forgot to set nut as wallfunction somewhere) but I figured that out on my own. Thanks a lot for your help.
With kind regards,
Jelle Lagerweij

[Jelle Lagerweij]

I thought that I fixed everything and now it runs perfectly.


Except, the results are crap... Somehow, the flow doesn't seem to go through our system. Although I think that the boundary conditions are correct, when I run the model, the flow seems to stop at the inlet. I already checked, and I define the inlet patches as patch type (instead of wall) in the snappyHexMeshDictionary. I also tweaked around with my k_omega boundary conditions, but I think these should be ok. I included my full problem as attachment again, it should run simpleFoam without any fatal error now. Maybe you could explain to me why this behaves this way.


With kind regards,
Jelle Lagerweij

[Yann]

Hi Jelle,

At the inlet your are setting a velocity of 20m/s along the x direction, but your channel is oriented in the y direction.
Just change your inlet velocity vector and it should be fine!

Cheers,
Yann

[Jelle Lagerweij]

Lol, my IQ hit rock bottom .
Thanks a lot.


Jelle

[yousufaero]

Did you manage to run the case in parallel. I can not get that working.

Can you tell me how to do it


Thank you
Yousuf

[Yann]

Hello Yousuf,

This is not really possible to help you without knowing what you're doing. (check this post: How to give enough info to get help)
What did you try and what error do you get?

Regards,
Yann