9 Replies Latest reply on Feb 16, 2017 3:47 PM by John Willett

    Simplified Simulation of a Pivot (or Ball Joint) Between Parts?

    John Willett

      I have seen several questions about ball joints, but none seems to address my need:  I do not want to simulate a ball joint, per se.  I just want to define a contact mechanism between two solid parts that can pivot frictionlessly in any direction (only over a tiny range of angles, because this is a static simulation) but cannot separate or interfere.  (Imagine that the parts are bonded only at a point so are free to rotate but not to translate relative to one another.)

       

      Although I can imagine modeling concentric semi-spherical faces with no-penetration contact conditions between them, this is way too complicated for my application.  I am interested only in static simulation of the rest of a structure when some parts can pivot slightly relative to others.  I am not interested in the stresses or strains at these pivoting joints themselves.

       

      Is there any simple way to do this? -- John Willett

        • Re: Simplified Simulation of a Pivot (or Ball Joint) Between Parts?
          Ryan Dark

          Hi John,

          I think you can accomplish what you described here (a simplified ball joint) through the use of beam elements.  I worked up an example as a proof of concept (attached) in which I edited the definition of two beams so that they were defined to be hinges.  I edited the common joint shared by these two beams to both be joints so that this joint would take moment load and act as a ball joint.  The results below would indicate that the joint took no moment and achieved that goal.

          So, in our own setup you could create two beam bodies that you bond to your existing structure and apply hinges to their common joint and make a simplified ball joint.

            • Re: Simplified Simulation of a Pivot (or Ball Joint) Between Parts?
              John Willett

              Thanks Ryan! -- This is taking me into new territory.  I will try it out, but it may take a few days... -- John Willett

              • Re: Simplified Simulation of a Pivot (or Ball Joint) Between Parts?
                John Willett

                Ryan -- Still ignorant and beginning to dig into beams, but I already have two questions:

                 

                1) Does your "end connection/hinge" at the common beam ends allow any kind of relative rotation between the beams, or just "hinging" about two axes?  If the latter, which axis is prohibited?

                 

                2) How does your setup differ from simply inserting a "connector/link" between the two solid bodies?  (I guess this option would also allow displacements perpendicular to the link, which would not be desirable.)

                 

                Thanks again for starting me down this road... -- John Willett

                  • Re: Simplified Simulation of a Pivot (or Ball Joint) Between Parts?
                    Ryan Dark

                    Hi John,

                    1. It should be allowing unrestricted rotation through all rotational degrees of freedom.  I am just speculating but hinge is probably just the easiest word that describes that rotations are not locked down.
                    2. A rigid connection prohibits rotations on both ends of the connection.  A link connection hinges at both ends of the connection.  This setup allows one end to be set to be rigid (rotations locked down) and one end to hinged (rotations unrestricted) so you can control where the rotation is taking place.
                      • Re: Simplified Simulation of a Pivot (or Ball Joint) Between Parts?
                        John Willett

                        Ryan (or anyone) -- A modification of your elegant solution to my ball-joint problem may work in my application.  I had failed to mention a crucial requirement in my OP, namely that the pivot must be essentially in the plane of a pivoted plate.  It appears that I can achieve this by using only a single beam section that is hinged (to a solid plate, perhaps modeled as a shell) at one end and fixed (to another solid plate) at the other.  This appears to give reasonable results -- see attached image -- but I'm wondering if it is actually a valid use of your idea.  (If beam sections can be hinged only to other beam sections, one of two beams sections must be made much shorter than its diameter, since they apparently cannot be made rigid.)  Can you comment?

                        Hinged at Top.png

                        More Background on my Application:  The pivot must equalize the normal forces at the three vertices of a triangular plate (a "whiffle-tree" for amateur astronomy), in spite of small tilts, while not introducing transverse displacements.  Transverse displacements cannot be allowed because there will be three such pivots (three wiffle-trees) supporting the back of a mirror; small tilts in one plate cannot introduce transverse forces among them.  I hope this brief description is clear...

                          • Re: Simplified Simulation of a Pivot (or Ball Joint) Between Parts?
                            Ryan Dark

                            What you have described here is not clear to me.  Do you have an image of the real life object that I can reference against?

                              • Re: Simplified Simulation of a Pivot (or Ball Joint) Between Parts?
                                John Willett

                                >>What you have described here is not clear to me.  Do you have an image of the real life object that I can reference against?<<

                                 

                                Ryan -- Below is the best photo I can find quickly of a simple mirror "flotation" cell.

                                tryan.19a.jpg

                                The right-hand object is a 9-point mirror cell.  The back of the mirror rests on the small studs at the vertices of each triangular part.  Each triangle pivots on a point at what would be its COG, assuming a uniform flat plate, equalizing the normal forces on its three vertices.  For obvious reasons there are exactly three equally spaced pivots.  As long as the triangles are oriented with circular symmetry, as shown, the result (if the pivots are nearly frictionless) is that all 9 pads carry equal weight when the mirror is pointed vertically.  More importantly, no distortions of the metal parts are transmitted to the mirror because the triangles can tilt to maintain equal force distribution.  Does this answer your question?

                                 

                                I fear I am now wasting your time.  My SW question is how best to implement a pivot point that lies essentially in the plane of a flat plate.  I think I've already got this working -- see attached model (in SW 2016 -- I hope you can run it) using rectangular plates instead of triangles only for testing purposes -- but I wanted to confirm there is no reason why your ball-joint simulation has to connect two beams.  One beam can connect two other objects, either solids or shells, with one end of the beam set as a hinge (the pivot between the beam and one object) and the other set as bonded to the other object.  Right?

                                 

                                Another thing I just discovered:  It may not be possible to get accurate free-body forces that involve the ends of a beam because it won't let me define a contact set including a beam end.  Is this true, or am I doing something wrong? -- John Willett

                                  • Re: Simplified Simulation of a Pivot (or Ball Joint) Between Parts?
                                    Ryan Dark

                                    Hi John,

                                    I think the crux of your question lies here...

                                    I wanted to confirm there is no reason why your ball-joint simulation has to connect two beams.  One beam can connect two other objects, either solids or shells, with one end of the beam set as a hinge (the pivot between the beam and one object) and the other set as bonded to the other object.  Right?

                                    You are right.  You can bond the joint on the end of the beam to a solid or a shell and the end of the beam will remain a hinge there.  There is no reason this has to be between beams.

                                     

                                    As for your contact set, the beam end does not bond itself onto shell or solid bodies.  The beam joint must be bonded to the shell or solid body.  You would need to use this filter (below) to select the beam joint in the blue box then the shell/solid in the pink box.