12 Replies Latest reply on Apr 6, 2016 6:02 AM by James Riddell

    How do you set up FEA on Extruded Al T-Slot Frames - Non Welded Construction?

    Allan Bowers

      I am designing a frame with T-Slot extruded Al which has several types of connection fixtures. These can be T-Nuts, T-Bolts, Gussets etc and are bolted connections.

       

      See eg here: http://www.3dcontentcentral.com/parts/supplier/Bosch-Rexroth/6.aspx or here: http://www.3dcontentcentral.com/parts/supplier/80%2020-Inc.aspx

       

      The frame is required to mount a vertical plate on one side and connect to L-Track on the other side where the L-Track is fixed.

       

      I assume I should use Beams for the FEA, and as the connections are bolted together I will need to add Bolt fixtures?

       

      I have to have a deflection of less than 0.5 inches at the load point in the design and the reaction load and moment limits for the L-Track must be within tolerances, (I am unsure how to conduct the FEA for. the L-Track).

       

      A minimum factor of safety for yield and a factor of safety for ultimate stress is required for the analysis too.

       

      I am quite new to FEA. I can set up a static study OK for a part but am unsure how to approach this problem of a bolted together Al T-Slot frame.

       

      The atch pic gives and idea of a butt connection fixture. Other connections are more simplified T-Nuts and Bolts etc.

       

      EAS T-Slot Hinge and Butt Joint.JPG

       

      Here is the L-Track with its connector type:

       

      EAS Mil Track & T-Slot.JPG

       

      Any help on how to set up such an FEA would be much appreciated.

        • Re: How do you set up FEA on Extruded Al T-Slot Frames - Non Welded Construction?
          James Riddell

          I'm not certain it is even possible to use bolted connections with beam elements since I believe you have to always define the bolt joint with geometry.

           

          However, if you have a small enough model and plenty of computing power, you could attempt this structure as solid geometry (with a fine mesh size of course.)  Read and heed well the warning on bolted joint analysis! 

           

          A viable alternative sometimes is to set up a contact surface where the bolted joint interface would be and bond those together and leave the remainder of the surfaces in non-penetration to one another (contact surfaces also have a high computing cost.)

           

          As a first pass I'd recommend ignoring the bolted connections, leave everything globally bonded and see if there is any great stress in the structure itself.  Then you could proceed to analyzing the bolted joints on a smaller scale.

            • Re: How do you set up FEA on Extruded Al T-Slot Frames - Non Welded Construction?
              Allan Bowers

              Thanks James. I have done the initial model as an assembly and modelled each piece of the frame so I can move them around as needed. Maybe I should make the frame as a Weldment to do the initial FEA first pass so that all the contacts get created automatically. I think they can then be represented as beams?? so that the sim runs expeditiously.

               

              I could include the T-Nuts and or corner gussets as part of the extruded geo as these faces effectively contact each piece of the extruded Al frame where they join, thus they should be able to be meshed with the frame elements.

               

              The lengths are relatively short for this design. Max around 40 inches across by 8 inches in depth and 20 inches height and the force applied is no more than 130 lbf at the end of the structure.

                • Re: How do you set up FEA on Extruded Al T-Slot Frames - Non Welded Construction?
                  James Riddell

                  That sounds about right.  We frequently do a quick 'load case 0' with just the main structure and gravity (don't forget gravity even on such a small frame if deflection is any concern.)  If the parts are making good contact and there are no slivers/gaps then a solid model is usually easiest to set up.  I run simplified weldments all the time.

                   

                  Yes, it is certainly possible to model the T-Nuts, etc. and use their contact to transfer forces.  It won't be 100% accurate but will get you close to the real answer.  The geometry will make the sim run much faster than bolted connections or contact sets.

                   

                  As Amit says, you could run these as a welded beam analysis but since the size is so small it doesn't make much sense to do things that way.  It is also easier to constrain the geometry logically - just don't over-constrain any surface/part.

              • Re: How do you set up FEA on Extruded Al T-Slot Frames - Non Welded Construction?
                Amit Katz

                For your initial analysis you can probably assume the beams will bend and yield for before your bolted connections will, since those connectors are typically steel. From personal experience those extruded aluminum beams tend to sag quite easily when they get longer than a few feet. I would start with a truss analysis. Once you isolate your major stresses you can look at individual features like holes in the beams for thru bolts etc.

                • Re: How do you set up FEA on Extruded Al T-Slot Frames - Non Welded Construction?
                  Allan Bowers

                  A quick update. After messing around trying to get a mixed mesh with beams and solids to work I discovered that end to end beam joint contacts work well but if you want a solid to contact a beam perpendicularly to its length the FEA runs into trouble.

                   

                  So I started with one long section of T-Slot and added my other pieces using the Bolt connectors. As I added I ran a simple sim and things worked well.

                   

                  Now I have the assy nearly done and ALL of the joints will be No Penetration and they all have various BC's tying the design together.

                   

                  I wondered if the fol approach would work correctly, (it seemed to when I ran the FEA).

                   

                  As all the contacts are No Pen I set the Global Contact to No Pen for the whole assy. The FEA ran fast.

                   

                  Then I set the global to bonded and created Contact Sets for all the No Pen contacts. It bogged down, (first point) then kept giving me Equilibrium not met errors.

                   

                  So it begs the question about setting the global to No Pen. It doesn't have to wade thru 30+ contact sets and if you make a simple mistake electing Set 1 vs Set 2 the FEA can get quirky real fast and sometimes not work.

                   

                  Any thoughts?

                    • Re: How do you set up FEA on Extruded Al T-Slot Frames - Non Welded Construction?
                      James Riddell

                      That's an interesting concept.  If it has proven to work reasonably well for small set ups then there is no reason to expect that the whole model would act any different.  When I have run fairly large models (12M DoF & 80-100 bolted connections) and it was all non-pen contact sets I got very good results.  I've run others w/ some global and they do take MUCH longer for the DoF numbers.  However, I've had to make sure that the solver was NOT FFE since that would crash.

                        • Re: How do you set up FEA on Extruded Al T-Slot Frames - Non Welded Construction?
                          Allan Bowers

                          Hi James, it all seems to be working well using a global Non Pen set. The SW does a search for all possible contacts. The Help menu does state that it usually takes longer to solve when set up this way, so I tried making NP contact sets.

                           

                          It gets real finicky when I do this, even using the Auto mode.  The sim runs, seemingly a little quicker BUT it keeps giving Equilibrium Not Reached Save up until Now errors. If you say Yes to save it keeps chugging on but this means sitting through the whole solve as it happens two to three times.

                           

                          So I will stick to the other method for now since ALL the connections are bolted.

                           

                          One dilemma I have though is how to replicate a grub screw where the bond created between two surfaces is by the pressure applied by the conic end of the grub screw onto a face of one component and transferred to the other by the thread contact.

                           

                          The T Bar is a method of extending the length of two sections of T-Slot

                           

                          PastedGraphic-1.png

                          PastedGraphic-2.png

                           

                          PastedGraphic-3.png