Welcome to the forums. Please post in English or French.

You are not logged in.

#1 2011-05-03 12:47:35

pierre_j
Member
Registered: 2010-01-19
Posts: 773

[?] most stable way to model surfacic rigid body in Aster?

Dear all,

Please if anyone can share some tips and tricks about surfacic rigid body modelling in Code_Aster, we would be glad to know about them.

Here is most notably some items that make me raise the demand:

1- moving rigid body modelling based on AFFE_CHAR_CINE
I have seen a stamping simulation failing because some nodal stiffness was applied to the nodes of a rigid body modelled simply as a group to which was applied a Z displacement (through AFFE_CHAR_CINE), while the same simulation, without the nodal stiffness (which had anyway no influence on the result because of AFFE_CHAR_CINE loading) was behaving a bit better, and only failing some timestep later...
Is it known that nodal stiffness would be "unadviced" when considering rigid body modelling based on AFFE_CHAR_CINE?

2- differences between high Young modulus and AFFE_CHAR_MECA \ LIAISON_UNIF
I wonder if one solution is known to allow more stable simulations than the other?

3- moving surfacic rigid body modelling
Is there a need to set to a moving surfacic body (modelled with triangles or quadrangles) a DKT formulation and a thickness while at the same time imposing a AFFE_CHAR_MECA / LIAISON_UNIF?
I have seen such a model and I wonder if we could instead set the formulation to 3D and skip the thickness definition?
Is one solution known to allow more stable computation than the other?

I thank you in advance for any hints about that topic.
Bests,

Pierre

Last edited by pierre_j (2011-07-19 12:21:25)

Offline

#2 2011-05-03 15:43:39

Thomas DE SOZA
Guru
From: EDF
Registered: 2007-11-23
Posts: 3,066

Re: [?] most stable way to model surfacic rigid body in Aster?

Hi,

In the U2.04.04 manual (offline for the moment but should hopefully be back soon), you'll find advices on how to model rigid tools. In a nutshell it is usually done this way :

- model the tool in the mesher. Two possibilities arise : either you model only the border or the whole body. The advantage of modelling the whole body is that it will be easier to orient its normal
- orient the normal either with ORIE_PEAU_* or ORIE_NORM_COQUE depending what you chose at the previous step
- add only the border of the tool to the model (and apply a 3D modelization)
- constrain all the dofs of the tool with AFFE_CHAR_CINE so as not to add any additional dofs in the model

TdS

Offline

#3 2011-05-03 16:10:13

pierre_j
Member
Registered: 2010-01-19
Posts: 773

Re: [?] most stable way to model surfacic rigid body in Aster?

Dear Thomas,

Thank you very much, but I seem to have a quite incomplete knowledge in AFFE_CHAR_CINE:
Is there a LIAISON_UNIF or LIAISON_SOLIDE in AFFE_CHAR_CINE to constrain the nodes of the rigid body together?

Indeed, let say that you want your rigid body to press with an imposed force a deformable body (and not an imposed displacement). How can it be managed thanks to AFFE_CHAR_CINE?

I would use AFFE_CHAR_MECA \ LIAISON_UNIF and AFFE_CHAR_MECA \ FORCE_NODALE.

I thank you in advance for your help.
Bests,

Pierre

Offline

#4 2011-05-03 17:25:13

Thomas DE SOZA
Guru
From: EDF
Registered: 2007-11-23
Posts: 3,066

Re: [?] most stable way to model surfacic rigid body in Aster?

pierre_j wrote:

Indeed, let say that you want your rigid body to press with an imposed force a deformable body (and not an imposed displacement). How can it be managed thanks to AFFE_CHAR_CINE?

This is no longer the moving of a rigid body and therefore you'll still have a null matrix in this case if you model only the border of the tool.
In this case springs or slow-dynamics are mandatory since the force you apply must face a stiffness.

TdS

Offline

#5 2011-05-04 03:12:54

todd_alan_martin
Member
Registered: 2008-03-06
Posts: 131

Re: [?] most stable way to model surfacic rigid body in Aster?

Hi Pierre

Have a look at the attached COMM file, to see how I modelled a rigid curved surface for contact.
Particularly the keyword AFFE_CHAR_CINE.

Todd.


Attachments:
contact_example.tar.gz, Size: 89.67 KiB, Downloads: 319

Offline

#6 2011-05-04 06:05:38

pierre_j
Member
Registered: 2010-01-19
Posts: 773

Re: [?] most stable way to model surfacic rigid body in Aster?

Thank you both.

Thomas, as always, your explanations are perfectly clear and I understand what you point out.

Todd, thank you also for your model. I think to understand your modelling but the use of AFFE_CHAR_CINE is here for a still rigid body (not a moving one through which is applied a force).

Todd, this becomes off-topic, but there is 1 thing I do not understand in your model:
- you consider a single node "pin" (POI1) to which you give a dicrete mass), and constrain rotations and translations but Z
- you link it to a group of nodes called "bulkhead" to which you apply FORCE_NODALE and linke to "pin" through a LIAISON_SOLIDE

=> why don't you skip the "pin" node, discrete mass, and LIAISON_SOLIDE and directly apply rotation and translation constraints on "bulkhead" group?

Bests,

Pierre

Offline

#7 2011-05-04 06:44:04

todd_alan_martin
Member
Registered: 2008-03-06
Posts: 131

Re: [?] most stable way to model surfacic rigid body in Aster?

Hi Pierre

=> why don't you skip the "pin" node, discrete mass, and LIAISON_SOLIDE and directly apply rotation and translation constraints on "bulkhead" group?

In this case the pin and LIAISON_SOLID is unnecessary, but it is a technique I use for convenience, so that the bulkhead nodes are constrained to rotate around a particular point.

I usually know the applied force at the centre of the cross-section (extracted from a beam model). In some cases I need to apply loads and moments to the top of the tube. In other cases I want to constrain the pin, so that it moves vertically, but not horizontally, for example, and extract the reactions. The discrete mass is fictitious, but is there to satisfy code-aster.

I think to understand your modelling but the use of AFFE_CHAR_CINE is here for a still rigid body (not a moving one through which is applied a force).

Thomas should be able to answer this better than me. A motion can be applied to the rigid body by specifying a non-zero displacement, possibly using a ramp function.

Todd.

Last edited by todd_alan_martin (2011-05-04 06:47:11)

Offline