6 Replies Latest reply on Nov 29, 2018 11:41 AM by Gary Lucas

    How can I model a twisted belt that runs on offset pulleys?

    Mike Hissong

      Hi all.  I am working on a motion transfer assembly.  I am trying to accurately model a timing belt that runs on offset pulleys (yes, I know this is an atypical application).  I need accurate bends and twists in order to model appropriate intermediate rollers where the belt changes direction.

      Belt path was developed by making an intersection curve from intersecting surfaces - see attached picture.  I've attempted to model with sweeps and lofts but don't know how to approach the necessary twisting that occurs.  Anyone have a good direction to suggest?

        • Re: How can I model a twisted belt that runs on offset pulleys?
          Matt Lombard

          Are there capstans or something to direct the path of the belt as shown? If the belt is free between pulleys, it won't look like that.

           

          Anyway, I just swept along your 3D path with a rectangle as shown below. I added a sketch point and placed it at the midpoint of the side of the rectangle, and used the Pierce constraint. It will compress the inside and extend the outside radiuses of the belt. Should be a reasonable assumption for a rubber belt. It depends on how accurate you need it to be. Belt teeth are not shown, but would add material to the inside of the belt, obviously.

           

          Oh, and learn how to use Pack and Go, so you get all the assembly and parts in a single zip file.

           

           

            • Re: How can I model a twisted belt that runs on offset pulleys?
              Mike Hissong

              Thanks for the quick reply Matt - perhaps too quick...

              My question deals with the capstans you are asking about.  "I need accurate bends and twists in order to model appropriate intermediate rollers where the belt changes direction."

                • Re: How can I model a twisted belt that runs on offset pulleys?
                  Matt Lombard

                  Ok.

                   

                  Well, the capstans complicate things. In that case, I placed a plane at the capstan and reoriented the sketch (actually I used a block to simplify things) for that location, using the same midpoint pierce constraint, and rotating the block appropriately. Then I used a series of sweeps (for the two pulley ends) and centerline lofts for the twist. So the blue faces are swept and the gray are lofted. Actually you'd have to add two profiles for each capstan to be really accurate, one at the beginning of the arc and one at the end, and then adjust the features as necessary.

                   

                  You must have a couple of rollers really close to one another. The twist at Plane 5 and 7 are not quite right. Anyway, this gives the idea. Use lofts for twists, use sweep for everything else.

                   

                  The loft looks like what is shown at the bottom. You have to use the SelectionManager to get the sweep and the loft Centerline option to accept partial loops. There are ways to avoid the Centerline option using Guide Curve instead, but you'd have to put a sketch endpoint on the path. Sketch midpoint or sketch point doesn't count.

                   

              • Re: How can I model a twisted belt that runs on offset pulleys?
                Steven McCallion

                Model the belt between the two pulleys, and then use a model of a capstan with the "Indent" tool?

                • Re: How can I model a twisted belt that runs on offset pulleys?
                  Gary Lucas

                  I used to do this in Rhino for a V-belt drive that had a bend away from the plane of the two end pulleys in the middle of one leg to go around a drive pulley.  I used V-belt profiles at the tangent points of a center line curve oriented to the pulleys and then used an oriented sweep or loft of those profiles. So I think a similar approach would work here.  That was more than 15 years ago, and they are still running.