Best Practice Ball Bearing

General Questions
1

Hi, first of all a big thank you to all the supporters, it’s an amazing piece of software.

I wanted to know what is the best practice for simulating ball bearings in an assembly. I’m not interested in the stress or strain of the inner, outer ring or ball, just the behavior as an assembly.

In the old days of Nastran 2001, I modeled ball bearings with RBE3 (outer or inner ring as slave and the center as master). Both masters were connected with a spring of length 0 in between. The stiffness of the bearing was defined by the spring. The reaction force was compared with the data sheet of the bearings.

image
image1450×1107 70.9 KB

How would you do this?

2

In place of RBE3 constraint, in CalculiX, you could use distributing coupling constraints - not yet supported in PrePoMax but can be added via the keyword editor: RBE3 integration in PrepoMax (Kinematic & Distributing Coupling)

Two-node springs may also be added this way: Spring between two nodes

CalculiX offers also kinematic coupling constraints (closest equivalent to RBE2) and rigid body constraints (those are available in PrePoMax but they make the surfaces completely rigid).

I have a few papers about simplified modeling of bearings in Abaqus but they discuss roller bearings.

2 Likes
3

I have applied different options trying to obtain the described coupling between the two bodies and unless I’m doing something wrong this are my results:

*COUPLING,REFNODE= ,SURFACE= ,ORIENTATION= ,CONSTRAINTNAME=

*DISTRIBUTING

Can’t transfer loads between the different bodies. It can distribute actions only if applied directly to the REFNODEs.

Might I be doing something wrong with the first option.? I have tried a spring and equation to couple both REF nodes and none of them seem to work.

*COUPLING,REFNODE= ,SURFACE= ,CONSTRAINTNAME=

*KINEMATIC

Can transfer displacements between the two bodies.

*ELEMENT,TYPE=DCOUP3D
*ELSET,ELSET=
*DISTRIBUTING COUPLING,ELSET=

Can transfer actions between the different bodies but elements may become distorted unless nodal weights are properly defined.

I haven’t try Rigid body but I have succed previously on other models.

4

Well, considering the limitations of coupling (especially distributing) constraints in CalculiX, it might indeed be best to just use rigid bodies in such cases. That’s also how pinned joints are typically modeled based on several threads on the CalculiX forum.