cyclic boundary conditions for FSI

kanuk · April 16, 2013, 17:52

Hi All,

I've found out that the current 1.6-ext doesn't support cyclic boundary conditions for adaptive mesh motion (i.e. FSI). As such, I've spent a bit of time trying to figure out how to get around this issue....

At this point, I'm stuck and would really appreciate any insight into my problem. I *think* I'm fairly close, and have based my library off of what's found in src/tetDecompositionFiniteElement. Allwmake runs fine, no complaints.

However, when I include the relevant libs in my controlDict I get the following error:

Code:

...
// * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * //
Create time

--> FOAM Warning : 
    From function dlLibraryTable::open(const fileName& functionLibName)
    in file db/dlLibraryTable/dlLibraryTable.C at line 86
    could not load /home/dave/OpenFOAM/dave-1.6-ext/lib/linux64Gcc46DPOpt/libfaceDecompTetCyclic.so: undefined symbol: _ZTIN4Foam28cyclicTetPolyPatchFaceDecompE
--> FOAM Warning : 
    From function dlLibraryTable::open(const fileName& functionLibName)
    in file db/dlLibraryTable/dlLibraryTable.C at line 86
    could not load /home/dave/OpenFOAM/dave-1.6-ext/lib/linux64Gcc46DPOpt/libcellDecompTetCyclic.so: undefined symbol: _ZTVN4Foam28cyclicTetPolyPatchCellDecompE
Create dynamic mesh for time = 0

Selecting dynamicFvMesh dynamicMotionSolverFvMesh
Selecting motion solver: laplaceFaceDecomposition


--> FOAM FATAL ERROR: 
Unknown faceTetPolyPatchFaceDecomp type cyclic.  Valid faceTetPolyPatchFaceDecomp types are :

7
(
wall
processor
wedge
patch
empty
ggi
symmetryPlane
)


    From function faceTetPolyPatchFaceDecomp::New(const polyPatch&, const tetPolyBoundaryMeshFaceDecomp&)
    in file tetPolyMeshFaceDecomp/tetPolyPatches/faceTetPolyPatch/newFaceTetPolyPatchFaceDecomp.C at line 59.

FOAM exiting

Attached is what I have as of now, take a look if you want. I've been learning c++ the past while to make sense of it all, so I'm sure there's something I'm missing. Hopefully it's not too far off....

Cheers,
Dave

ngj · April 17, 2013, 03:55

Hi David,

In the beginning of your run, it tells you the following:

Quote:

--> FOAM Warning : From function dlLibraryTable:

pen(const fileName& functionLibName) in file db/dlLibraryTable/dlLibraryTable.C at line 86 could not load /home/dave/OpenFOAM/dave-1.6-ext/lib/linux64Gcc46DPOpt/libfaceDecompTetCyclic.so: undefined symbol: _ZTIN4Foam28cyclicTetPolyPatchFaceDecompE

--> FOAM Warning : From function dlLibraryTable:

pen(const fileName& functionLibName) in file db/dlLibraryTable/dlLibraryTable.C at line 86 could not load /home/dave/OpenFOAM/dave-1.6-ext/lib/linux64Gcc46DPOpt/libcellDecompTetCyclic.so: undefined symbol: _ZTVN4Foam28cyclicTetPolyPatchCellDecompE

So there seems to be an incompatibility problem with the additional libraries and the solver you are using. If you can get it working, I would be very much interested, as I am facing similar problems; that is periodic domains and mesh motion. I thought I had a robust (and explicit) approach, but it is producing really bad mesh quality

Therefore, I have decided to start looking at an implicit implementation of the cyclic boundary condition for the tetFem-framework.

Kind regards

Niels

kanuk · April 17, 2013, 16:02

Thanks Niels for your input.

After checking, I indeed did find the undefined symbols in my user libbin:

Code:

$ nm -u $FOAM_USER_LIBBIN/libCyclicBC* | grep cyclic
                 U _ZN4Foam15cyclicPolyPatch8typeNameE
                 U _ZTIN4Foam15cyclicPolyPatchE
                 U _ZTIN4Foam28cyclicTetPolyPatchCellDecompE
                 U _ZTVN4Foam28cyclicTetPolyPatchCellDecompE
                 U _ZN4Foam15cyclicPolyPatch8typeNameE
                 U _ZTIN4Foam15cyclicPolyPatchE
                 U _ZTIN4Foam28cyclicTetPolyPatchFaceDecompE
                 U _ZTVN4Foam28cyclicTetPolyPatchFaceDecompE

However, upon checking if there were any problems with the standard libfaceDecompFiniteElement.so library, I also found the following, related to the processor type BC:

Code:

$nm -u $FOAM_LIBBIN/libfaceDecompFiniteElement.so | grep processor
                 U _ZN4Foam18processorPolyPatch8typeNameE
                 U _ZTIN4Foam18processorPolyPatchE

So, maybe I've based my cyclic BC on a base class that already isn't compatible? If so, I'm not sure how to move forwards with this....

ngj · April 18, 2013, 01:30

Hi David,

I will give your code a try myself later today (if the time allows), so could you tell me, what solver you have been using?

I have a vague feeling of the reason for your problems, but it is too vague for me to put in words

Kind regards

Niels

ngj · April 18, 2013, 04:55

Hi David,

I can reproduce the error, if I am using moveDynamicMesh.

On the other hand, if I use a small tool for mesh motion, which I made myself, then it does recognise your cyclic boundary condition and subsequently crashes with another error

I have tried looking around and changing the include-statements and the linking properties of moveDynamicMesh.C and Make/options, though, I have not been able to isolate the problem. Though, if I were you, I would start to add a lot of include-statements related to the tetFem-framework to the solver, which you are using and test, whether this resolves your problem.

I can unfortunately not share my successful code with you, as it hinges on a large internal development project.

Kind regards

Niels

ngj · April 19, 2013, 03:36

Hi David,

It occurred to me that my include statements in the project development could be helpful. Below are those related to the 1.6-ext release; all includes related to the project have been removed:

Code:

#include "addToRunTimeSelectionTable.H"
#include <algorithm>
#include "autoPtr.H"
# include <cassert>
#include "cellSet.H"
#include "cellZone.H"
#include "coordinateSystem.H"
#include "cyclicFvPatch.H"
#include "displacementFvMotionSolver.H"
#include "dynamicFvMesh.H"
#include "edgeInterpolationScheme.H"
#include "elementFields.H"
#include "faceTetPolyPatchFaceDecomp.H"
#include "faCFD.H"
#include "faMesh.H"
#include "fieldTypes.H"
#include "fixedValueFvPatchFields.H"
#include "fixedValuePointPatchFields.H"
#include "fixedValueTetPolyPatchFields.H"
#include "fvCFD.H"
#include "fvcMeshPhi.H"
#include "fvMesh.H"
#include "fvMeshSubset.H"
#include "fvPatchField.H"
#include "fvPatchFieldMapper.H"
#include "fvPatchFields.H"
#include <gsl/gsl_matrix.h>
#include <gsl/gsl_multiroots.h>
#include <gsl/gsl_vector.h>
#include "hydrodynamicProperties.H"
#include "incompressible/RAS/RASModel/RASModel.H"
#include "interpolateXY.H"
#include "IOdictionary.H"
#include "IOList.H"
#include "lduMatrix.H"
#include "mapPolyMesh.H"
#include "mathematicalConstants.H"
#include "mixedFvPatchFields.H"
#include "motionDiff.H"
#include "objectRegistry.H"
#include "pointPatchFields.H"
#include "polyMesh.H"
#include "regIOobject.H"
#include "runTimeSelectionTables.H"
#include "scalarSquareMatrix.H"
#include "SortableList.H"
#include "subCycle.H"
#include "surfaceFields.H"
#include "Switch.H"
#include "tetDecompositionMotionSolver.H"
#include "tetFec.H"
#include "tetFem.H"
#include "tetFemMatrices.H"
#include "tetFemMatrix.H"
#include "tetPointFields.H"
#include "tetPolyMesh.H"
#include "tetPolyMeshMapper.H"
#include "tetPolyPatchFields.H"
#include "Time.H"
#include "Tuple2.H"
#include "uniformDimensionedFields.H"
#include "volFields.H"
#include "wallFvPatch.H"
#include "wallPolyPatch.H"

Have a nice weekend

Niels

kanuk · April 19, 2013, 14:54

Hi Niels,

Thanks for providing your included header files!

Unfortunately though, after including them in my solver (icoFsiFoam) I still end up with the same error as stated previously.

And, without a more verbose error message (and my limited c++ experience) it may take awhile to find where the problem is...

Regardless, I'm still looking into it. Will let you know if I come up with anything.
Cheers
Dave

kanuk · April 25, 2013, 14:29

It's a shot in the dark, but here's a newer version of the cyclic BC. Basically I replaced a few member functions from the processor type BC with the ones needed for the cyclic type BC, and with some changing of class names, it still compiles fine.

Unfortunately though, I still get the same error message about undefined symbols. I was hoping it worked for your case though, Niels...

Like I said, likely this is a no. But maybe worth a try.

cyclicBC_V2.tar.gz

ngj · May 3, 2013, 03:28

Hi David,

I am sorry that I have not answered before, though, I have been busy and will also be so in the coming time, thus I do not think that I will have time to take a look on the code.

Good luck,

Niels

hxaxtma · March 7, 2014, 03:44

I am struggling with the same case, need to use cyclicGgi BCs, you got any solution on that?

Jiricbeng · April 25, 2014, 02:52

Do you have an example for cyclic BC of a rotating segment of rotor domain, just an example..

April 17, 2013, 16:02		#3
kanuk New Member Kanuk Join Date: Oct 2010 Posts: 28 Rep Power: 15	Thanks Niels for your input. After checking, I indeed did find the undefined symbols in my user libbin: Code: $ nm -u $FOAM_USER_LIBBIN/libCyclicBC* \| grep cyclic U _ZN4Foam15cyclicPolyPatch8typeNameE U _ZTIN4Foam15cyclicPolyPatchE U _ZTIN4Foam28cyclicTetPolyPatchCellDecompE U _ZTVN4Foam28cyclicTetPolyPatchCellDecompE U _ZN4Foam15cyclicPolyPatch8typeNameE U _ZTIN4Foam15cyclicPolyPatchE U _ZTIN4Foam28cyclicTetPolyPatchFaceDecompE U _ZTVN4Foam28cyclicTetPolyPatchFaceDecompE However, upon checking if there were any problems with the standard libfaceDecompFiniteElement.so library, I also found the following, related to the processor type BC: Code: $nm -u $FOAM_LIBBIN/libfaceDecompFiniteElement.so \| grep processor U _ZN4Foam18processorPolyPatch8typeNameE U _ZTIN4Foam18processorPolyPatchE So, maybe I've based my cyclic BC on a base class that already isn't compatible? If so, I'm not sure how to move forwards with this....

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Error finding variable "THERMX"	sunilpatil	CFX	8	April 26, 2013 07:00
Water subcooled boiling	Attesz	CFX	7	January 5, 2013 03:32
snnappyHexMesh with cyclic boundary conditions	longamon	OpenFOAM	2	May 9, 2011 12:28
periodic and cyclic boundary conditions	Hadi	Main CFD Forum	2	June 29, 2007 07:19
A problem about setting boundary conditions	lyang	Main CFD Forum	0	September 19, 1999 18:29

April 18, 2013, 01:30		#4
ngj Senior Member Niels Gjoel Jacobsen Join Date: Mar 2009 Location: Copenhagen, Denmark Posts: 1,900 Rep Power: 37	Hi David, I will give your code a try myself later today (if the time allows), so could you tell me, what solver you have been using? I have a vague feeling of the reason for your problems, but it is too vague for me to put in words Kind regards Niels

April 18, 2013, 04:55		#5
ngj Senior Member Niels Gjoel Jacobsen Join Date: Mar 2009 Location: Copenhagen, Denmark Posts: 1,900 Rep Power: 37	Hi David, I can reproduce the error, if I am using moveDynamicMesh. On the other hand, if I use a small tool for mesh motion, which I made myself, then it does recognise your cyclic boundary condition and subsequently crashes with another error I have tried looking around and changing the include-statements and the linking properties of moveDynamicMesh.C and Make/options, though, I have not been able to isolate the problem. Though, if I were you, I would start to add a lot of include-statements related to the tetFem-framework to the solver, which you are using and test, whether this resolves your problem. I can unfortunately not share my successful code with you, as it hinges on a large internal development project. Kind regards Niels

April 19, 2013, 14:54		#7
kanuk New Member Kanuk Join Date: Oct 2010 Posts: 28 Rep Power: 15	Hi Niels, Thanks for providing your included header files! Unfortunately though, after including them in my solver (icoFsiFoam) I still end up with the same error as stated previously. And, without a more verbose error message (and my limited c++ experience) it may take awhile to find where the problem is... Regardless, I'm still looking into it. Will let you know if I come up with anything. Cheers Dave

April 25, 2013, 14:29		#8
kanuk New Member Kanuk Join Date: Oct 2010 Posts: 28 Rep Power: 15	It's a shot in the dark, but here's a newer version of the cyclic BC. Basically I replaced a few member functions from the processor type BC with the ones needed for the cyclic type BC, and with some changing of class names, it still compiles fine. Unfortunately though, I still get the same error message about undefined symbols. I was hoping it worked for your case though, Niels... Like I said, likely this is a no. But maybe worth a try. cyclicBC_V2.tar.gz

May 3, 2013, 03:28		#9
ngj Senior Member Niels Gjoel Jacobsen Join Date: Mar 2009 Location: Copenhagen, Denmark Posts: 1,900 Rep Power: 37	Hi David, I am sorry that I have not answered before, though, I have been busy and will also be so in the coming time, thus I do not think that I will have time to take a look on the code. Good luck, Niels

March 7, 2014, 03:44		#10
hxaxtma Senior Member Join Date: Jan 2014 Posts: 179 Rep Power: 12	I am struggling with the same case, need to use cyclicGgi BCs, you got any solution on that?

April 25, 2014, 02:52		#11
Jiricbeng Senior Member Jiri Join Date: Mar 2014 Posts: 218 Rep Power: 13	Do you have an example for cyclic BC of a rotating segment of rotor domain, just an example..