14 Replies Latest reply on Apr 2, 2013 6:38 AM by Mark Thompson

    Concentric Shaft Arrangement

    Mark Thompson

      Good afternoon,

       

      I'm looking for some advice on the best way to perform a static analysis and in the event of large stresses and/or deflections (unlikely) a non-linear analysis of the following  concentric shaft and bearing problem. Effectively I have a series of large concentric aluminum pipe sections which are designed to be supported within one another by a pair of four point contact bearings (i.e. locating LHS and non locating RHS) housing within the ends of each parent (i.e. outer) shaft. Please see attached screen grabs for an illustration of the arrangement, geometry and applied loads.

       

      My question is whether the bearing connections are constraining the model as intended or whether I should opt for a differnt connection (e.g. rigid, spring, pin etc) or approximate with a cylindrical fixture (e.g. locating LHS = fixed radial/axial translation and non locating RHS = fixed radial translation). Obviously I've run simulations on both the bearing connections and cylindrical fixtures and the figures are close in each case, however, in both instances the 'bearing' restrains appear to maintain parallel axial alignment between every selected face restraint and the original assembly (datum) axis, obviously a crude approximation of such a deflection.

       

      I've already preformed an analysis of each independent shaft and its applied loads including the equivalent reaction loads of the internal shafts/loads of its supported  bearing centers, however, this does not take account of the resultant deflection, although small, hence, a full assembly would be preferred.

       

      Does anyone have any suggestions as to a more appropriate bearing restrain whose effective axis follows the curvature of each shaft/housing?

       

      Thank you in advance.

       

       

      Mark.

        • Re: Concentric Shaft Arrangement
          Jared Conway

          are you saying you want the bearing to remain stiff but bend with the shaft?

           

          whether you use a bearing restraint or cylindrical restraint, that fixes your assembly with respect to the environment. the environment doesn't move so the bearings stay where they started.

           

          if the bearings need to move, what causes them to move? sounds like you'd need another part and to make the bearing a spring between where you have the bearing now and that part.

           

          i think a diagram of what you're expecting and the next parts in the assembly might be helpful.

            • Re: Concentric Shaft Arrangement
              Mark Thompson

              Morning Jared,


              Understood, I though as much. Please see attached section drawing for an illustration of the arrangement:Pipe Sections.png

               

              1) Static Hub (aluminum 6082 T6)

              2) Primary (aluminum 6082 T6)

              3) Secondary (aluminum 6082 T6)

              4) Tertiary (aluminum 6082 T6)

              5) Static Neutral (aluminum 6082 T6)

              6,7,8,9) Thin Section Four Point Contact Bearings

               

              The primary, secondary and tertiary shafts are designed to rotate at different speeds relative to one another, however, if we consider the static case both the shafts self weight and the application of external loads will induce curvature along their lengths thus modify the orientation of the counter bore bearing housings at each end (obviously more so on the RHS).

               

              I tried replacing the bearing connections with rigid connections over the weekend, these results appear to be more representative, however, I can't be sure of my assumptions. I also attempted the spring connections with 1000 (N/m)/m^2 but this must represent to many DOF as the solve time maxed out at 40% ~4hr....

               

              Any thoughts?

            • Re: Concentric Shaft Arrangement
              Bill McEachern

              I ran some simple tests on bearing connectors and rigid connector. Now to be clear, this is the bearing connector under the connections folder in the tree - not the one under the fixtures "folder". The one under the fixtures "folder" is not for part to part connections but part to ground which did not seem appropraite here. I got exactly the same answers for both  ( rigid & bearing) when the alllow rotation was not enabled in the bearing connector when the rigid option was used as oppsed to flexible. This was reassuring to me at least. Thus,  I would use the rigid at the locating end and the felxible at the non locating end with the axial stiffness set to zero and the radial stiffness set to some really big number - unless of course you know the actual bearing stiffnesses, then use those.

              On another note: I found that the interface on the enable rotation button in the bearing connector dialog, at least to me seems backwards. I thought the enable rotation was on by default as it looked in the depressed state. However my tests revealed that it was the opposite. They should consider changing the interface to be less ambiguous - such as a radio button.

                • Re: Concentric Shaft Arrangement
                  Mark Thompson

                  Hi Bill,

                   

                  That's correct, only connections rather than fixtures are considered in either case. The 'allow rotation' button certainly is ambiguous, I made the same assumption but didn't think to check the figures, not very thorough. I presume my results vary between the bearing and rigid connection configurations due to this assumption or maybe the size/complexity of my model; perhaps some marginal variability in the discretization of each body? Please see attached stress/deflection results for the previous and revised bearing configurations as instructed:

                  • Re: Concentric Shaft Arrangement
                    Mark Thompson

                    These figures, although within operational limits, are a little more than was anticipated. I'm wondering if I applied the 'armature' loads correctly? To clarify we have three separate armatures coupled to the RHS of each shaft section with their CoM aligned to their and the assembly axis. I considered this a remote load for simplicity and used direct transfer. Is this correct?

                      • Re: Concentric Shaft Arrangement
                        Bill McEachern

                        I don't think I have a complete grasp from you description but my first thought would be to use the rigid connection option as, I would assume, the armature wouldbe quite rigid do to its larger diameter adn ti would keep the attachment from deforming much.The direct transfer option will allow the nodes on attaching face to deform relative to each other, the rigid will not.

                        • Re: Concentric Shaft Arrangement
                          Jared Conway

                          Mark, what do you want those loads to represent?

                           

                          Regarding the bearing behavior, have you checked out this article in the help?

                           

                          http://help.solidworks.com/2013/English/SolidWorks/cworks/r_Connector_-_Bearing.htm

                           

                          It covers the implementation of the bearing connector. I'd suggest starting a bit smaller, say 2 shafts and get things setup and running to the point where things are working as you expect. That way you can better understand the ramifications of the different options in the connector. In the end, the bearing connector just connects 2 parts together with springs and if chosen, will restrict rotation relative to each other. Generally you'd have at least one with rotation locked if all of your parts were connected with these types of connectors or a bearing fixture with it locked.

                      • Re: Concentric Shaft Arrangement
                        Mark Thompson

                        To clarify, I'm attempting to simulate five concentric shafts supported within one another by a series of non-self aligning bearings  (i.e. four point contact bearings) housed within the counter bored ends of each shaft in order to support their own self weight and the loads of three separate balanced armatures located at each of the highlighted split faces. In effect these armatures can be assumed as concentric hubs lightly press fit to the end of each shaft (i.e. CoM in line with shaft axis).

                         

                        In reality these bearings are very thin and although they offer a tremendous amount of radial/axial capacity for their size they are only as rigid as their housing supports permit them to be (i.e. you could easily crush the largest 10" bore bearing in your hand). Depending on fit, which will be close but not actually a press fit in practice (i.e. to avoid pre load on the bearing), each hub should increase the radial rigidity of these particular housings, however, in this case I'd like a conservative figure so ideally want to eliminate the hubs additional support (i.e. assume the shaft housing offers the only support). In this case, is the direct transfer is correct?

                         

                        Jared, I've reviewed the help menu a couple of times but unfortunately its hard to tell whether I've extrapolated the bearing connection assumptions for my problem corretly (i.e. a series of concentric shafts rather than a relatively simple shaft supported by two fixed housings).

                          • Re: Concentric Shaft Arrangement
                            Mark Thompson

                            Please see attached comparison between the three bearing assumptions (A.C.W. from top right):

                             

                            1) All self aligning rigid bearings and direct transfer load.

                             

                            2) All non-self aligning rigid bearings and direct transfer load.

                             

                            3) All non-self aligning bearings (LHS rigid) and (RHS flexible, 0N/m Axial and 1GN/m Radial) and direct transfer load.

                             

                            Stress 2.png

                             

                            Deflection 2.png

                             

                            This assumes the counter intuitive 'self aligning' button point made earlier in the thread is correct...

                            • Re: Concentric Shaft Arrangement
                              Mark Thompson

                              Apologies, forgot to attach the following to the previous reply...

                              Shaft Arrangement.png

                                • Re: Concentric Shaft Arrangement
                                  Jared Conway

                                  Mark, you seem to have a great wealth of knowledge about your application but I think trying to start with your full assembly is leading you to some concerns that may not be valid and/or a result of the assumptions made by the boundary conditions.

                                   

                                  I'd recommend you start a bit simpler and add one connection/load at a time. If the assembly isn't moving as expected dig into the different settings for the load/connection to see which matches what you want and then we can discuss why the options are the best for your setup. Without seeing the settings you've chosen or the model, it's diffcult to tell if the results you're seeing are because of a bad assumption or a bad input.

                                    • Re: Concentric Shaft Arrangement
                                      Mark Thompson

                                      Hi Jared,

                                       

                                      Sounds sensible, I'll get back to you once I've had a play with the setup for a simpler arrangement. Perhaps I need to either assume the parent shaft (i.e. outer) as rigid (i.e. no reactive curvature) or model the child (i.e. inner) in its deformed state (i.e deflected housings) in order to calculate the cumulative superposition of each deflection through each generations.

                                       

                                      Mark.