SOL 106 Non Linear Elements

[josemanuellodoso]

I have a joint between a fitting (modelled by tetra10 elements) and a spar (modelled by cquad elements) in which each fastener is modelled by:
-RBE2 attaching 4 nodes of the fitting to central node 1 (DoF 123456)
-CBUSH between central node 1 and central node 2 (the same node duplicated). (PBUSH properties according to the fasteners size, material... in the 6 DoF)
-RBE2 attaching 4 nodes of the spar to central node 2 (DoF 123456)

When a run a linear analysis everything looks like Ok, loads in CBUSHs have an adequate direction and value. At the same way, stresses at fitting and spar elements have reasonable values.

However, the model has large deformations and then I have decided to run a sol 106 to obtain more accurate results. The problem is that when I do this, nothing has sense: the loads in the CBUSH have random direction and induce stress peaks in the elements caught by the RBE2s.

I have though that this could be because RBE2 and CBUSHs don't have "non-linear capacity", and then rotations induced by them in the elements of the fittings and spar are wrong, producing these peaks of stress. Any idea?

Thank you

 

[nlgyro]

Quoting:

> the loads in the CBUSH have random direction and induce stress peaks in the elements caught by the RBE2s.

Not sure what you mean by CBUSH loads have random directions... Are you reading the loads w.r.t to the local co-ordinate system of the CBUSH?

Local stress peaks at RBE2 is common. They are numerical singularities on the FEM. Usually one uses the induced fastener loads and does a hand check to clear these area's.

 

[josemanuellodoso]

Hi,

I want to say that when I plot CBUSH_Forces as vectors they don't follow a logic pattern (independently of the co-ordinate system). Due to this illogic forces, the stresses that appear in the spar and in the fitting elements is illogit too, even taking into account the presence of the RBEs.

My main question is if the use of CBUSHs and RBEs elements can be misleading, in models with large deformations? Because, I have performed similar models and I have never had similar problems. I am worried about how these elements update their co-ordenate system in non-linear solutions, since they can be introducing wrong loads into the model…

Thanks

 

[nlgyro]

Without seeing your modeling & results, its not possible to say whether the combination of RBE's and/or cbush could be the source of your problem.

BTW, you can try using RIGID=LAGR for a full non-linear behavior for the RBE's, and use CBEAM's instead of CBUSH's to model your fasteners in your SOL106.

 

[josemanuellodoso]

Thanks, but RIGID = LAGR isn't avaliable for SOL 106.

 

[jbrackin]

I don't see any mention of what program you are using to post-process the results. It would be good to start by trying to determine if the issue is with Nastran or with post-processing.Do the f06 and vector plots match?

 

[nlgyro]

Quoting:

> Thanks, but RIGID = LAGR isn't avaliable for SOL 106.

U'll have to use a SOL400 for it then.