Pressfit bearing and applycation load rbe3

Hello ccx users.
I am trying to set an analysis formed by two step:
First step : resolution interference between bearing and housing
Secondo ste: application a load at the center of the bearing througha a rbe3 (distributing coupling).
The results in terms of the dispacement if the first step are the same, but in the second step they are very different.

The input file are the following:

What could be the reason? When Do I use multiple step, I need to define a particular command line or the rbe3 does not work correctly?

What’s the other software with which you are comparing the ccx results ? Nastran ?

Can you share the whole input file (you could upload it to some hosting website and paste the link here) ?

The housing is also changing color (is moving).


¿Maybe some rigid body move?

I did a comparison with Abaqus

The constraints are very far. It is correct.

If the bearing has settle properly in the housing as it seem, both parts should move togheter in the plane of the seating surface with continuity in the displacements colour field.
¿Does the gasket have some BC or is it completely free?
¿Could you increase by lets say x10 the deformation scale to see where is each part going and why are colours discontinous between bearing and housing.?
¿Do the stresses match Abaqus?

Oh, I see. So it’s HyperMesh used for postprocessing. Was it the same input file as in CalculiX (just with small syntax adjustments) or recreated from scratch for Abaqus ?

I think you have to provide more details on the modeling of the loading in the second step so we can provide useful hints.
Hello, I copied the model ccx and abaqus. I create a smaller model. As I told you, in the secondo step the result are different in terms of displacements.

It looks that ccx does not apply the load correctly.

Hello juan,
I copied both model ccx and aba as you will read below

Why are you using C3D4 elements for the housing ? Linear tetrahedrons should be avoided in the majority of cases. Small sliding contact is risky too but I guess that you chose it to use *CLEARANCE in Abaqus.

I know that c3d4 generates stress different than C3D10, but in terms of displacements the results should be the same. However I used C3D4 becuase really we have a big model and in local Pardiso got crashed. I gave you just an extrapolation of a big model.
Yes I used small sliding becuase in abaqus it is mandatory to use clearance.
However as you can see in the first step, the correlation of the displacement are very good. There is something that doesn not work in the second step.
The results of the abaqus are fine by physical point of view.

To me it looks pretty much the same. The maximum disp’s are due to peak loads, and we know for concentrated loads CCX elements aren’t as good as abaqus, just a bit far away looks the same. Check the values

Also from the manual, this limitation:

A *DISTRIBUTING coupling is usually selected in order to distribute a
force or moment area-weighted among the nodes of a surface. For this to work
properly the surface should be plane.

maybe you should try a different loading method. In prepomax you can apply a total force on a surface directly.

If you see the colouer map the position of the maximum displacement Is very different.
So there Is something of strange

How can apply a force directly on the surface?
What Is the command line?

I dont agreee about you are saying. Because I have Always used in Abaqus this command line and It world properly. Maybe for ccx Is different?

Linear tetrahedrons may also overstiffen the results significantly.

In PrePoMax, you can use the Surface Traction load for that.

I use both C3D4 for ccx and abaqus. It is just a preliminary analysis to set the simulation and find a correlation. Afterwars I will switch the element into c3d10

Exactly. That’s what I am saying. Ccx is different software despite syntax is very similar. So don’t expect same modelisations will work the same in both softwares.

1 Like