[mathieu]

Hi people !

I used to run 2D fluid-structure interaction simulation with "closed" fluid-solid interface (see case A in the attachment). My current installation is OpenFOAM 1.4.1-dev revision 784.

Now, if I update my OpenFOAM installation to the latest 1.4.1-dev or even to the latest 1.5-dev version, those kind of simulation don't work anymore. However, if the fluid-solid interface is "open" (see case B in the attachment), then there is no problem at all (in all version).

The following package contains both cases A and B that can be run with the standard icoFsiFoam solver. More details are also provided in a pdf file.

Any help would be appreciated ! Kind regards,

Mathieu

[mathieu]

package.tgz

[hjasak]

Sorry, I do not get it. I have just run caseA and I can see forces being transferred, a flow field, a structural deformation field and a moving mesh for both the fluid and the solid.

When you say "it does not work any more", what precisely does not work:
- the case does not run
- you get a floating point exception
- you do not get the flow field
- you do not get structural deformation
- you do not get mesh motion
- force transfer is not correct

I am missing something, but don't know what...

More info please,

Hrv

[hjasak]

Got it: this is to do with a search algorithm for a multiply projected boundary (thank Gavin and his FSI artery simulation).

Whta happens is that for a closed surface with escapes, you cannot guarantee the projection will be correct using the fast algorithm. I have changed the switch in

~/OpenFOAM/OpenFOAM-1.5-dev/etc/controlDict

to use

nSquaredProjection 1;

in the OptimisationSwitches section and all works well. The deformation on the solut surface looks a bit stupid (you may want to use mesh motion to deform that side as well), but at least we know what the problem (and solution) are.

Enjoy,

Hrv

[mathieu]

Dear Hrv,

Thanks for the reply ! I changed nSquaredProjection to 1 and the problem seems to be gone. Thanks again,

Mathieu

[mathieu]

Dear Hrv,

I investigated this issue further and I found that there is still a problem with the mesh when using OF-1.5-dev with nSquaredProjection set to 1 (no more problems in the latest revision of 1.4.1-dev though!). See the right tip of the plate in the following pictures.

1.4.1-dev:


1.5-dev:


Is there something else I am missing or is this a bug ?

By the way, what did you mean by "solut surface" in your last message ?

Regards,

Mathieu

[hjasak]

This is the kind of work I do under a support contract. Alternatively, you have a dig through the code yourself.

Hrv

[maruthamuthu_venkatraman]

Originally Posted by mathieu

Dear Hrv,

I investigated this issue further and I found that there is still a problem with the mesh when using OF-1.5-dev with nSquaredProjection set to 1 (no more problems in the latest revision of 1.4.1-dev though!). See the right tip of the plate in the following pictures.

1.4.1-dev:


1.5-dev:


Is there something else I am missing or is this a bug ?

By the way, what did you mean by "solut surface" in your last message ?

Regards,

Mathieu

Hello Mathiew,
Have you fixed the error in 1.5-dev version for icoFSIFoam solver. As u stated 1.4-dev vrsion has no problems, Do I lack something if i use that version.

Regards
Muthu