11 Replies Latest reply on Jul 24, 2014 3:39 AM by Jared Conway

    Torque and Thrust on a Shaft

    Bjorn Sorenson

      I'd like to analyse torque and thrust on a vertical shaft at the same time.  I've set up the problem for torque-only using an "on cylindrical faces" fixture to simulate the bearings (see attached; torque at 1 is equal and opposite to total torque at 2 & 3) and gotten good results.  What I'd like to do is apply (in addition to the torque force) a thrust force located at points 2 & 3 as indicated in the image.

      The problem I'm running in to is that the "on cylindrical faces" fixture won't transmit the bending moment through the fixture, and a bearing fixture won't transmit the torque unless I know the axial and radial stiffness values of the bearings.

      I might be able to get axial stiffness values from the bearing manufacturer, but radial stiffness seems very non-trivial (it is based on the bearing/lubrication itself, RPM, angle of the deflected shaft running through it, etc.).  Simulation will allow me to define a flexible bearing with just axial stiffness, but I don't feel comfortable leaving radial stiffness out of the equation altogether.  Any suggestions?

      I should also note that I'm not set on using the SW Sim bearing fixture if there's a better way to do this.  Any and all input is appreciated!

        • Re: Torque and Thrust on a Shaft
          Jared Conway

          so the shaft is supported by bearings at the 2 split lines that you're restraining

          and then you want to apply torque and a downward force on both other locations?

          what's the end goal for the analysis? where are the torque and thrust forces coming from?

          if you were going to look at this in a test or hand calculations, how would you set it up?

           

          the biggest challenge you're going to have with bearing forces or on cylindrical face restraints is that you'll have instability in your model. sure you can turn on "stabilize' but if you look through posts on this, you'll see why that may not be a good idea

           

          so my first suggestion would be to come up with:

          1. what the problem would look like as a static problem with hand calculations.

          2. what you know about the problem or what additional features you want to include that you can't easily do in a hand calc. in the end if you want to include that bearing stiffness, you don't have much choice but to get the information or do testing to get the information. but if you're looking to compare designs relative to each other. make a good estimate and then run simulations and compare them udner the same assumptions.

            • Re: Torque and Thrust on a Shaft
              Bjorn Sorenson

              Thanks Jared.  I have done the hand calcs for a featureless shaft (including nifty shear and bending moment diagrams!), but I also need to add shoulders, keyways, retaining ring grooves, etc which complicates the problem.  Furthermore, I am really interested in deflections more than stresses, as these are critical to my application.  According to Shigley's Mechanical Engineering Design:

               

              "A deflection analysis is straightforward, but it is lengthy and tedious to carry out manually, particularly for multiple points of interest.  Consequently, practically all shaft deflection analysis will be evaluated with the assistance of software.  Any general-purpose finite-element software can readily handle a shaft problem." (and I would add, "provided you know all your input values")

               

              Perhaps it is time to have a discussion w/ the bearing manufacturer of what a reasonable estimate would be.

               

              BTW, stability isn't an issue with the cylindrical face restraint; all the torques cancel and I added a roller/slider top and bottom (not pictured) to keep it from moving axially.  I'm keeping my fingers crossed that if I can get some good values from the bearing manufacturer, this will continue to be the case.

                • Re: Torque and Thrust on a Shaft
                  Bjorn Sorenson

                  Of course, I spoke too soon.  Stability isn't an issue for the perfectly symmetrical featureless shaft shown in the picture.  However, as soon as you try the same analysis on something with keyways cut into the faces where your torques are input, your loading becomes asymmetrical and the shaft flies off into space (yaay!).

                   

                  I now have a complete set of hand calcs for a featureless shaft (no keyways, shoulders, etc).  Unfortunately, hand calcs are a gross oversimplification of the actual geometry (treating a 3" long bearing as a finite point, for example).

                   

                  The ultimate goal here is to simultaneously take into account the longitudinal shear, bending and torsional shear stresses in a (3-d, realistic) linear static analysis for a shaft with more complex features (keyways, changing diameters, etc).  This would give me my critical displacements as well as loading events for a fatigue analysis.

                   

                  Analysis of machine shafts has to be one of the most common engineering problems; surely someone has developed a methodology in SW Simulation that works consistently?

                   

                  As always, any and all input is greatly appreciated.

              • Re: Torque and Thrust on a Shaft
                Eric Vanarsdall

                I am trying to do almost the same thing, except I am more interested in stress than deflection. My shaft has torque, axial thrust, and a radial load (creating bending - my radial load is what you refer to as "thrust" - my thrust load is along the shaft's axis - creating tension in the shaft). I believe I have the torque and thrust working well separately and in combination, but I cannot get the bending stresses to match theory at all.

                 

                Did you ever figure this out?

                 

                Message was edited by: Eric Vanarsdall

                  • Re: Torque and Thrust on a Shaft
                    Simon Yang

                    Can you please put the model here? I did similar job.

                    • Re: Torque and Thrust on a Shaft
                      Bjorn Sorenson

                      I got distracted by other projects and haven't come back to this particular FEA.

                      Before I stopped working on it, I performed beam theory hand calcs for the shaft that took into account the length of the bearings and other components where loads are applied, treating them as distributed loads rather than treating them as point reactions (a whole lotta integratin' goin' on...).  This yielded bending moment, longitudinal shear force and the resulting deflection.

                      Torsional shear stress was handled as a separate calculation.  Now I have everything I needed for a Mohr's circle (and feel like I'm back in Mechanical Design 101 in Engineering school!).

                      I then plugged the resulting equations into Excel using a lookup table so that I could vary the loads and locations/lengths of application and get theoretical responses to compare my SW Sim model to.

                      So I have one tool to use for comparison, but no, I haven't gotten the SW Sim model to "work" yet (although again, I haven't come back to it since).  It would sure be nice to have another tool with a better UI than graph paper and Excel can offer.

                       

                      I will have to come back to this eventually, so I'm very interested in pursuing this conversation.  I understand the loading you've described for your application; what are you using for fixtures?  Are you using the "bearing fixture", and if so, what kind of boxes do you have checked in the property mgr?  What kind of values are you using for bearing stiffness, etc.? 

                      • Re: Torque and Thrust on a Shaft
                        Jared Conway

                        probably worth starting your own thread and posting what you are comparing with. my guess is that the simulation is right but that your theory may have assumptions that don;t exist in the simulation and that hte simulation is a more "correct" solution.