Surface traction as rbe3. Momentum?

to simulate a rb3r on a curved surface, you suggested a surface traction and it works correctly.
However on prepomax I saw that surface traction can be defined only only the translation components.
How can I apply some momentum since the c3d10 cant transfer rotation?
I think that for the momentum I cant use surface traction as well.
Do you have any advice?
Thank you

As I said here: Pressfit bearing and applycation load rbe3 - #23 by Calc_em you need to use a rigid body constraint to apply moments to solid parts.

it seems PrePoMax implementation of surface traction loads more similar to coupling type distributing instead of rigid body, it did not over stiffening the models.

Internally, the surface traction load in PrePoMax is just *CLOAD.

right, coupling type distributing also similar, except in limitation of surface traction in distributed moment loads by coupling forces. Did not why not any further enhancement which actually seems to be possible.

Maybe it is not clear. I have to apply both forces and moments on a bearing through rbe3 (no rbe2).
Since I have a not flat surface, you adviced me to use surface traction because coupling distribuiting does not work. It is of couse worked verywell when I apply only the force.
What about the moments? I know that I cant apply directly moments on the nodes
In the abaqus model, I applied forces and moments on the master of rbe3.

The Set Up for distributing moments on solids is similar to the one for distributing forces but in my opinion is conceptually more difficult to apply, read and anticipate the result.
Give it a try with some simple case like a cantilever solid beam for example. Linear elements and small displacements. If you push it or increase element order your will notice the *distributing limitations.


Nastran’s RBE3 is equivalent to distributing coupling in Abaqus and CalculiX. This constraint is available in 2 forms: *DISTRIBUTING COUPLING and *COUPLING followed by *DISTRIBUTING. Unfortunately, in CalculiX, it doesn’t work as well as in Abaqus (even after the improvements in the recent versions - check the previous forum discussions on this topic). I would still give it a try though. *DISTRIBUTING can be used to transfer moments. There’s also kinematic coupling (*KINEMATIC), equivalent to RBE2, and rigid body but if you don’t want to introduce rigidity to the model, you should stick go distributing coupling.

If the surface is cylindrical, you could try *TRANSFORM, TYPE=C so that BCs and concentrated forces applied to the selected nodes are defined in a cylindrical coordinate system.

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.
This is by ccx manual. I tried to use at the beginning the rb3r (distribuiting coupling) to apply a force, but I got different results than results.
I worked correcly with surface traction.

If you don’t mind local stresses you can consider using a couple of tractions, the resultant being a moment. Maybe we can request to prepomax developers to create a “moment traction”. Rbe3 equations are pretty straightforward.

For me it is important local stress.
Yes It could be nice to add the Surface traction for moment in PrePromax

The question is how it would be supposed to work internally. If via *DISTRIBUTING then exactly how the constraints would be applied to the surfaces ? IMO general support for this constraint should be added first.

I think the issue is how it behaves in a non linear analysis, probably that’s the reason it (*distributing) does not work properly in ccx. For a linear analysis you can find the formulation here in paragraph C.1. If you do not consider follower forces, the distribution of forces/moment from a point to nodes in a surface is straightforward and can be carried out by the preprocessor transforming it to *cload. This would be useful as well.

Sorry I’m confused due to the fact that both cards have the same wording.
I guess you mean *Coupling + *Distributing doesn’t work on curved surfaces?

Juan, yes if you see on the manual coupling+distribuiting cant be used on curved surfaces (no flaft surfaces). I tried to simulate it but the reason looks quite different. The right approachI to have a correlation with abaqus is to use the surface traction but it allows only to transfer forces and not moments.
I dont know how transfer moments without using a rigid system sincerly.

so, says the manual:


however I meant that in the past people reported *distributing and *distributing coupling not working very well in all situations and I think (my hypothesis) is due to the difficulties in formulating it for non linear analysis as should be expected for a software like ccx.

That note on the manual made me fail when comparing with different source. *Coupling + *Distributing result is also position dependent of the REF node when distributing forces. I check it myself and *Coupling + *Distributing + Forces agreed well if that is taken into consideration. In the case of *Coupling + *Distributing + moment it is still unclear to me. I mean the meaning itself of what distributing a moment across a surface is. I did not pay too much attention anyway as it’s been reported as working improperly for a long time. Victor deprecated *Coupling in Mecway some versions ago. (By the way, I still think Kinematic is perfectly usable)

*Distributing Coupling + forces do not dependent on the Dcoup’s position and it’s not comparable with *Coupling + *Distributing + forces. I see it more like a traction on which one can adjust the weights of each node. That makes it difficult to set up as in order to obtain a uniform distribution and avoid spiky results in the corners or midside nodes the weights need to be written one by one (I did it once).

That could be interesting in Prepomax as a workaround.

So the only way to apply the moments is to try an equivalent surface traction?

*Distribuiting coupling is not suitable to apply moments.
I have apply *Coupling + *Distributing + moment +force on a curved surface (Linear Solid elements) and at least all the reaction forces and moments are in balance and as expected.

Are you sure your refs nodes are both at the same place when comparing?
Could you share the inp or a simplified model showing your discrepancy.?

This picture is the model in Abaqus where you can see an housing in blu, in green a bearing in interference with the housing and in red the rb3e (master at the center of the bearing).
I simulated at the beginning this rb3e with coupling+ distribuiting but I got different displacements.
After this, I used surface traction and I got good correlation with abaqus (the same results). Now I would like to trasfer moments though rb3r ( in Abaqus on the master of rb3r I applied forces and moments)