Rotation Modelling

RichardS · 2011-04-15 10:15:35

Hello,
I've a problem with rotation modelling and symmetry boundary conditions. My model is a 40 degree 3D section of a cylinder which I want to rotate around the axis of symmetry. I can apply the load with AFFE_CHAR_MECA and ROTATION but I don't know how to apply the symmetry boundary condition.
I attached my .comm-file which is not working ( pivot null error).
Best regards,
Richard

Attachments:
Rotation.comm, Size: 1.53 KiB, Downloads: 456

BL · 2011-04-15 15:07:29

I am not sure but I think what you need is LIAISON_CYCL.

RichardS · 2011-04-15 15:40:12

Hello,
I think I solved my problem. I found the great tutorial of P.Carrico which I could adapt to my case.
I also thought about LIAISON_CYCL but there it is done with LIAISON_OBLIQUE.
Don't know if this is the only possible solution, but for now I have a running model.
Regards,
Richard

The tutorials are found here: http://www.caelinux.org/wiki/downloads/docs/PCarrico/

RichardS · 2011-04-28 13:47:51

Hello,
finally I dind't really solve the problem.
LIASON_OBLIQUE is only sufficient if my model is rotation-symmetric.
What I really need is LIAISON_CYCL as suggested by BL, since I have a cyclic symmetry which is not rotation-symmetric.
I read the ducumentation about it in AFFE_CHAR_MECA [U4.44.01] but didn't understand how to set it up correctly.
There is also no test case where this particular command is used...
I would really appreciate any help or explanation on this topic.
Best Regards,
Richard

RichardS · 2011-05-03 13:26:37

Hello,
I'm still struggling with this simulation.
Since I didn't really understand the documentation (at some point I think it might be wrong or at least confusing, e.g. the equation on page 32 of U4.44.01 or the parameters COEF_MAIT1 and COEF_MAIT2 which were named alpha and beta in the Syntax section and in the Opérandes section they were said to be the cosine and sine of the sector angle beta) I had to do some trial-and-error investigation. But i always get the error that I have "une relation linéaire entre ddls surabondante".
I attached the mesh and command-file. Hopefully somebody could have a look at it and guide me how to solve this issue.
All suggestions are welcome!
Best regards,
Richard

Attachments:
Rotation.comm, Size: 3.02 KiB, Downloads: 400

RichardS · 2011-05-03 13:29:55

And the mesh.

Attachments:
Sector_Mesh.zip, Size: 970.14 KiB, Downloads: 354

RichardS · 2011-05-15 20:04:48

Still no suggestions on this problem?
I hope I'm not the first person ever who used (or lets say tried to use...) this command.
Regards,
Richard

Hermit · 2011-05-16 22:15:31

Hi, not sure you need to give an axis, in 3-D modeling, think you just need to provide boundary conditions to the the faces where you cut it, i.e. if the cyclic symmetry is every 90-degrees and your parts axis of rotation is the z-axis, then one face of the part is given the constraint dy=0, the other face is given dx=0. Not sure exactly how you extend that to a smaller section of the part easily, perhaps someone with more experience on this could shed some light...

Sorry I can't provide any more help, I'm a bit new to this.

RichardS · 2011-05-16 23:25:53

Hello Hermit,
thanks for your reply, that means that at least somebody has had a look on ths topic.

At first I had the same idea as you, but unfortunately this is not sufficient. I had done this with the command LIAISON_OBLIQUE.It allows you to define a local coordinate system X*Y*Z* and there you can set dX* = 0. But the results do not have to be correct for 2 reasons:
1. there could be a deformation in X* direction (just imagine a section of a propellor where the blades are deformed due to rotation)
2. you need to couple the DOF's of the one surface in the symmetry plane to the other side (this is not done with LIAISON_OBLIQUE).

So I tried the command LIASON_CYCLIQUE,which should do what I need. But unfortunately the documentation seems to be pretty much cryptic to me and I'm still not able to understand it.
Regards,
Richard

todd_alan_martin · 2011-05-19 12:45:17

Hi Richard

You have over-constrained the problem.

My guess is that the cylindrical axis contains nodes which have been constrained by the LIAISON_CYCL command for both surfaces (FACE1 & FACE2) of the wedge. This causes a singular matrix, so the solver fails.

Remove the duplicate nodes from one group (either FACE1 or FACE2) and it should work.

Cheers,
Todd.

RichardS · 2011-05-19 14:39:13

Hello Todd,
thanks for the reply.
Unfortunately I already subtracted the axis nodes from the slave face (FACE1) which is a group of nodes. That gives me the nodal group FACE12 (see the COMM-file). The master face has to be a group of volumes, so i can't subtract the axial nodes there instead.
I also checked the nodes which were over-constrained and they don't belong to the axis.
Regards,
Richard

Last edited by RichardS (2011-05-19 14:40:28)

todd_alan_martin · 2011-05-23 03:08:21

Hi Richard

Your best bet might be to model the entire structure and forget the cyclical symmetry constraints.

Todd.

#1 2011-04-15 10:15:35

Rotation Modelling

#2 2011-04-15 15:07:29

Re: Rotation Modelling

#3 2011-04-15 15:40:12

Re: Rotation Modelling

#4 2011-04-28 13:47:51

Re: Rotation Modelling

#5 2011-05-03 13:26:37

Re: Rotation Modelling

#6 2011-05-03 13:29:55

Re: Rotation Modelling

#7 2011-05-15 20:04:48

Re: Rotation Modelling

#8 2011-05-16 22:15:31

Re: Rotation Modelling

#9 2011-05-16 23:25:53

Re: Rotation Modelling

#10 2011-05-19 12:45:17

Re: Rotation Modelling

#11 2011-05-19 14:39:13

Re: Rotation Modelling

#12 2011-05-23 03:08:21

Re: Rotation Modelling

Board footer