[philippose]

Hello,

Good Morning :-)!

There seems to be a problem in the Hydrofoil tutorial case for interTrackFoam in OpenFOAM-1.5-dev.

When the case is executed, it aborts with an error stating that the specified motion solver does not exist.

On changing the motion solver to either of the two solvers (laplaceFaceDecomposition, pseudoSolidFaceDecomposition) given in the error message as possible candidates, the simulation aborts after the first time step (0.005) with the following error:

Problem with edge counting in lduAddressing.

From function tetPolyMeshLduAddressingFaceDecomp::tetPolyMeshLdu AddressingFaceDecomp
(
const tetPolyMeshFaceDecomp& mesh
)
in file tetPolyMeshFaceDecomp/tetPolyMeshLduAddressingFaceDecomp.C at line 156.

FOAM aborting

Have a great Sunday!

Philippose

[virginie_e]

Hello,

I am also trying to make the hydrofoil tutorial run in the OpenFoam-1.5-dev version (02-02-2009) and I get some errors.
I first run makeFaMesh from the hydrofoil directory and everything seems to run fine:

Exec : makeFaMesh
Date : Feb 13 2009
Time : 15:07:43
Host : fire
PID : 30209
Case : /users/V1117324/OpenFOAM/v1117324-1.5-dev/run/tutorials/interTrackFoam/hydrofoil
nProcs : 1

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

Create mesh for time = 0

Create faMesh ... Done
Add faPatches ... Done
Write finite area mesh ... Done


(NB: I had to add FoamFile{ ... } in the faMeshDefinition file

But then, when I run interTrackFoam, here is the error message I get:

Create time

Create mesh, no clear-out for time = 0

Reading field p

Reading field U

Reading/calculating face flux field phi

Selecting motion solver: laplaceTetDecomposition


Unknown solver type laplaceTetDecomposition

Valid solver types are:

3
(
pseudoSolidFaceDecomposition
RBFMotionSolver
laplaceFaceDecomposition
)


From function motionSolver::New(const polyMesh& mesh)
in file meshMotion/motionSolver/motionSolver.C at line 107.

FOAM exiting


Can someone help me to understand what is wrong? Thank you a lot.

Virginie

[hjasak]

I have updated the case + added ldu solvers to the executables - see svn. The problem is that the case runs, but it gets disturbed on the free surface - we need Zeljko to have a look.

(he is out of action at the moment)

Hrv

[virginie_e]

Thank you Hrv for your advise. I would have another question. I am trying to simulate with interTrackFoam the flow of a liquid along a slope. Everything runs fine until a certain time where I have a floating point exception.
By looking at the results closely, I think the problem comes from the fact that some cells of the mesh become very flat. (I have a kind of wave at the bottom of the slope and there are the "flat" triangles).
Are there any remeshing option already implemented in interTrackFoam? Otherwise, are there any remeshing tools which are already implemented in OpenFOAM? Otherwise, how would you advise me to proceed if I want to build one?

Thank you a lot for your help and your patience.

Virginie

[deepsterblue]

Virginie,

OpenFOAM mesh-modifiers don't currently implement any triangular/tetrahedral re-meshing options. That's been the bulk of my effort in the past year. Please let me know if this is what you're looking for:

www.ecs.umass.edu/~smenon

Cheers,
Sandeep

[philippose]

Hello Sandeep,

A Good Day to you!

I was just having a look at your videos... specially the one with the adaptive re-meshing of tetrahedral domains using Shewchuk's approach... and I find it very interesting..!

This is something I have been toying around with as an idea which needed to be implemented in OpenFOAM, but havent had the chance to sit with it yet...!

Have you tested your code with complex geometry? As in.... say something where gap between the fixed part and the moving parts of the mesh are not so large? About 3 to 4 cells?

And... how computationally expensive is the re-meshing code? For example, how many cells did your test case (the one with the sphere) have, and what kind of time frames did you achieve for that simulation?

Also, have you been able to successfully map all the fields after the re-meshing step, and maintain continuity of the fields, and conservation of mass, etc..etc.. ?

Finally... is the code you are working on also eventually targeted at being open-source?

Have a nice weekend!

Philippose

[deepsterblue]

Philippose,

I'm quite confident about the code handling complex geometry - I'm also looking for test cases. If you have a tet/tri-mesh case that looks demanding, I'd like to have a look at it. Please specify the nature of the boundary motion along with it as well.

I have not performed rigorous tests for code efficiency yet, but I'm willing to bet that the re-meshing costs are on par with the flow-solver. The translation case had about 8500 cells in it. On my Core-2 Extreme, re-meshing took about 1 to 1.5 secs per time-step. The code is also being made multi-threaded, so that should work well on today's multi-core processors.

Field-mapping, I will admit - is quite challenging. Interpolation tends to introduce errors, and minimizing that is quite a task. I often see pressure spikes after re-meshing, so that still needs a lot of work.

I'd like to have this work as open-source, but I'd also like to have it working reasonably well before releasing it. Besides, all this stuff is hard work, and I'd also like to be given credit, if you know what I mean!

Cheers,
Sandeep

[philippose]

Hey Sandeep :-)!

A Good afternoon to you....! Thanks a lot for the quick reply!!

I totally understand it when you say that you would like to be given credit for the hard work that you have put into the system...

However, correct me if I am wrong.... Open Source does in no way imply "no credit for the creator" or? I think its the other way around... each time someone in the community uses your code and your concepts... or each time when someone looks into your code.... they would say.... "wow... that guy is cool.... its great that he actually let something like this out as open source" :-)!

Look at the kind of respect the brains behind OpenFOAM get... and personally, every time I use OpenFOAM, I keep saying "wow... and this is Open Source :-O !".. and it actually inspires me to try and give something back to the community...

Of course...there is always the success story of Linux.... Even though half the world uses Linux, everyone who uses it knows that it basically stemmed from the "musings" and "tinkering" of one guy... Linux Torvalds... So he is not forgotten yet!

And... finally, I think the satisfaction one gets at contributing quality stuff to the open source community is in a way.... unparalleled... :-)! (I can see a hundred people trying to stomp me into the ground now :-)!)


Now... getting to the serious stuff.... I shall see if I can send you a simulation which my line of work (hydraulic valve design) often sees, and you can see if your code works on it....

And yes... I guess the whole field mapping issue is a big issue that would have to be handled in this case....

Well... 8500 => 1.5 s / 500000 => 88 s ... hmmm... :-) but I guess thats something that really cannot be avoided....

Have a nice weekend!

Philippose