CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)
-   -   FSI problem with topology of 3D beam via UDF (http://www.cfd-online.com/Forums/fluent/41687-fsi-problem-topology-3d-beam-via-udf.html)

greg July 12, 2006 10:12

FSI problem with topology of 3D beam via UDF
 
Hi,

i having some problems with my 3D beam mesh. What i am doing is trying to update my beam in fluent from displaced coordinates from my FE package. I can achieve this for small deflections using the spring analogy. however i want to model larger deflections. i have tried creating a fluid zone around the beam that i was hoping would deform as the beam deforms, but i can't get it to work!

i want the fluid zone around the beam to move in sync so that the same topology is kept. Thus when restarting the Fluent job it will continue the calculations from the previous time step with the same mesh topology.

Any ideas how i could define this region, or indeed if i actually need it?

Many thanks,

Greg

Luca Cavagna July 12, 2006 10:46

Re: FSI problem with topology of 3D beam via UDF
 
Unfortunarely Fluent's spring analogy is not so robust. Try using tetrahedra mesh and if you're doing an unsteady analisys try lowering the time step. Luca

greg July 12, 2006 10:59

Re: FSI problem with topology of 3D beam via UDF
 
Hi Luca,

i am already using tet mesh and a low time step. The problem occurs when the limitations of the spring is reached. As the beam gradually bends more and more the mesh fails due to a negative volume.

thanks for your reply,

Greg

Luca Cavagna July 13, 2006 02:29

Re: FSI problem with topology of 3D beam via UDF
 
Yes I know. Unfortunately as said this algorithm is not robust. You should try to switch on the remeshing but I've never been able to remesh my domain when I got a negative volume. As you have already reduced the time step I'd suggest raising the number of smoothing iterations (the default is 20). This should help you. Luca

greg July 14, 2006 04:16

Re: FSI problem with topology of 3D beam via UDF
 
i have already set my smoothing iterations upto 100000 and it will still fail eventually. it will work with remeshing but then i lose the topology. think i will try remeshing and see how long the next time step takes to converged. thanks again Luca

Luca Cavagna July 14, 2006 04:30

Re: FSI problem with topology of 3D beam via UDF
 
Could you explain how you succeed in remeshing? You've all tetrahedra or a hybrid grid? What kind od setting do you use for remeshing? You mean you loose topology on the beam nodes? Luca

greg July 17, 2006 05:11

Re: FSI problem with topology of 3D beam via UDF
 
i have a udf that operates after every time step that updates the mesh form the FE package. I have not ran a full 3D case yet but in 2D i have an unstructured mesh. as the exterior nodes of the beam move the mesh updates itself by means of remeshing for large deflections, or smoothing if it is a small displacement. i just use the basic setting for smoothing, k=0.05, beta=1, and smoothing its= 80000 ( i just set this as a really high number). for remeshing it is all standard stuff as per the tutorials.

what happens if remeshing is used is that the cell around the beam would change, i.e lose there topology. thus when resarting the computation since the there is a new topology the time to reach convergence has increased. it some instants it is basically like making a whole new mesh again and starting from the beginning again.

is this what you are meaning?

Greg

Luca Cavagna July 17, 2006 05:32

Re: FSI problem with topology of 3D beam via UDF
 
yes exactly! thank you so much for sharing these infos. I can suggest you 1 more thing: try creating some large elements as you come closer to the "far field", this may help you. This elements should take most of the deformation of the beam. What kind of deformation you have when you say "large"? Moreover try raising the far field dimension to have more space to adsorb the deformation. Luca


All times are GMT -4. The time now is 20:16.