CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   OpenFOAM Bugs (http://www.cfd-online.com/Forums/openfoam-bugs/)
-   -   Recent updates done to libforces (http://www.cfd-online.com/Forums/openfoam-bugs/76548-recent-updates-done-libforces.html)

podallaire May 28, 2010 08:42

Recent updates done to libforces
 
Good day,

I'm very impressed with the new updates that were done in libforces / the
ability to use constraints and restraints is very interesting.

I just have a simple suggestion regarding the springs. With the report option "on",
the spring lengths and angles are displayed. However, I understand that the
magnitude is returned, meaning that the values are always > 0. It would be a
good thing if the "sign" would be included as well / it makes it easier to track
the exact position in time of the rigid body.

Regards,

PO

gmacpherson May 31, 2010 06:47

Hello,

Quote:

I'm very impressed with the new updates that were done in libforces / the
ability to use constraints and restraints is very interesting.
Thanks.

Quote:

I just have a simple suggestion regarding the springs. With the report option "on",
the spring lengths and angles are displayed. However, I understand that the
magnitude is returned, meaning that the values are always > 0. It would be a
good thing if the "sign" would be included as well / it makes it easier to track
the exact position in time of the rigid body.
I have pushed this to 1.6.x.

What to do with the sign of the angle was unambiguous, but linear springs can start with zero length and move in any direction, so the concept of a sign for their displacement is not always clear. I added reporting of the vector from the anchor to the instantaneous attachment point on the body.

Graham

podallaire May 31, 2010 10:54

Thanks for the changes / will give it a try.

Another question -> did you try to run a case with single cpu for some time
and then, decompose the latest time step for a multi-cpu run ? I most be
missing something because I just cannot run multi-cpu if I try that. I guess I
need to move the latest time/polyMesh/points file in constant/polyMesh before
decomposing or something like that ... Not sure what is the best practice for
this.

Regards,

PO

gmacpherson June 2, 2010 05:10

There are some issues to do with decomposing moving mesh cases in 1.6.x that have been fixed in our internal development version.

For the moment, best practice would be to decompose the case at the start and do everything in parallel.

Graham

podallaire June 3, 2010 08:20

Ok, thanks Graham

podallaire June 10, 2010 10:33

Good day Graham,

I have another question related to the definition of the damping for the
restraints. Can you confirm that the damping input is in fact the damping ratio,
in percentage, meaning that an input of 0.5 means 0.5% damping ratio (or
0.005). Also, can you confirm that this works for both axial spring and vertical
spring / I'm facing some discrepancies in my decay tests.

Regards,

PO

gmacpherson June 10, 2010 10:51

It is not a damping ratio, it is a damping coefficient.

For the definition see linearSpring.H

Code:

//- Damping coefficient (Ns/m)
scalar damping_;

and .C

Code:

restraintForce = -stiffness_*(magR - restLength_)*r - damping_*(r & v)*r;
Graham

lakeat October 6, 2010 19:13

Some results
 
Hi I made some fsi videos using pimpleDyMFoam, but quite rudiment, It is a wind tunnel case. I found the new patch is very capable for some rigid body movement simulation, Good.


Video Link


All times are GMT -4. The time now is 07:37.