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

You are not logged in.

#1 2021-04-19 15:37:39

LouisB
Member
Registered: 2020-12-21
Posts: 9

Contact Problem

Hello,

I am trying to simulate a T-Flange bolted connection but I am getting a problem during the analysis.
Let me first introduce what I am trying to do:

In my model I have Two bolts, two washers, two Nuts, one Tflange, and two other plates that I am trying to tighten with the bolted connection. In the attached model I am simulating only the half of the bolted connection so the simulation time is shorter. (The symetric plane is OYZ)

At the end I want two simulation steps:  from T0 to T1 I pull on both bolts (to simulate the pretension in the bolted connection), then from T1 to T2 I release the pretension and add a Tie connection between the Nuts and the Bolts and I look at the displacement and the constraints on the nuts.

First I'm simulating the step from T0 to T1, so I only wants to pull on both bolts with a force F that I add on one side of both bolts.

But I am facing a problem while I start to simulate, and the error is telling me " Degré de liberté physique associé au noeud N*** et à la composante DY" and that brings to a singular Matrix during the factorization and so I can't reach the convergence.
I am guessing it is a Boundary conditions problem or a rigid body motion which is getting me this error but after a lot of research and after trying multiple times to follow the advices that the output message is giving me, I don't find how I can correct it because everytime I look for the node which is getting this physics degree of freedom and I try to put to zero the component that the error message is giving me, I am getting a new one...

Could someone please help me to correct my mistake or give me a tips so I can unblocked my problem ?

Unclosed the output message that I am getting after the simulation is done (of course a CPU limit error because my analysis is not converging) and the .hdf and .comm files of my model

Thank you in advance,
LB


Attachments:
output.mess, Size: 114.65 KiB, Downloads: 23

Offline

#2 2021-04-19 15:41:16

LouisB
Member
Registered: 2020-12-21
Posts: 9

Re: Contact Problem

I am sorry but the files are too big so I have to add them separetly in different messages

Here is the .comm

Regarding the .hdf file it is way too big (even if I compress it) so if you need it I can send it to you by email or I can send you a DropBox link with everything in it.

Regards,

LB

Last edited by LouisB (2021-04-19 15:55:20)


Attachments:
CurrentCase.comm, Size: 3.02 KiB, Downloads: 19

Offline

#3 2021-04-19 19:12:23

mf
Member
Registered: 2019-06-18
Posts: 204

Re: Contact Problem

Dear Louis,

1) in line 197 it says you have double nodes, they will most likely always lead to a singular matrix, even if the rest is ok. Check your mesh in the mesh module, there is an option to check for double nodes. Maybe think about remeshing in a different way.
2) it's easier to start without friction and get convergence, and if µ=0.12 (lubed connection?) I would leave it at SANS FROTTEMENT entirely, except something you want to see in your model depends on it (maybe a force-fit connection or similar..)
3) if that doesn't help you can suppress rigid body motion with weak springs attached to the moving body. There are some examples in the forum.
4) In DEFI_CONTACT the following options always help convergence, if it runs without it's fine:
      ALGO_RESO_CONT='POINT_FIXE',
      ALGO_RESO_GEOM='POINT_FIXE',
      FORMULATION='CONTINUE',
      FROTTEMENT='SANS',
      LISSAGE='OUI',
      NB_ITER_GEOM=some_value (small for testing, increase for better results),
      REAC_GEOM='CONTROLE'
also activate PENALISATION in your ZONE, COEF_PENA_CONT should be some small number times elastic modulus for starters (e.g. 5 times elastic modulus). That should help also.
5) also make sure the meshes of the contacting surfaces are in order and your choice of master and slave is correct (master must be coarser than slave).
   
Maybe read h_ttps://www.code-aster.org/V2/doc/v13/en/man_u/u2/u2.04.04.pdf ou la version francaise si vous préférez? :-)

That's all I can think of for know without knowing your model,

Mario.

EDIT: Louis sent me his files by mail, we solved it. I always thought double nodes lead to singular matrices, it does not seem like it! This example also works with double nodes and springs.

Last edited by mf (2021-04-20 18:08:13)

Offline

#4 2021-05-04 16:37:08

LouisB
Member
Registered: 2020-12-21
Posts: 9

Re: Contact Problem

Hello,

First thank you very much mf for your advises, it helps me a lot.

I still have a problem for my third and last step.
Has I said in my previous message, I am trying to simulate a T-Flange bolted connection. My final simulation should have three steps:

Step 1 (from T0 to T1): I pull on both bolts (to add the pretension on the bolts) --> this step is ok now

Step 2 (from T1 to T2): I realease the traction on the first bolt and I add a glued connection between the nuts of the bolt that I released and the outter surface of the bolt. --> this step is also working.

Step 3 (from T2 to T3):  I am realeasing the second bolt and I put a glued connection between the nuts of the second bolt and the outter surface of the second bolt. --> this step is not working and is giving me a singular matrix error ... I guess I have something wrong with my EXCIT of the steps 3 but after a long research I really don't understand my mistake...

Can somebody have a look to my .comm file unclosed and help me figured out where I am wrong ? My .hdf files is too big so if you need it you can contact me on my email and I will send you a dropbox link with my files.

Regards,
LB


Attachments:
CurrentCase.comm, Size: 10.55 KiB, Downloads: 10

Offline