14 Replies Latest reply on Nov 20, 2017 3:43 AM by Andrew Cuttle

    FEA Simulation of a tube through test path

    Nick Canfield

      I need some help setting up an FEA simulation. I have a tube with an internal edge that we want to simulate the stress on as it moves through a bending fixture with different edge radii. I have tried to do this using the motion works, but the tube just pierces the fixture. The the most success I have had is using a static FEA with an applied for to the force to the tip and a fixed proximal end. This gave us some useful information but we really need the simulation to better match the actual testing. Any help in setting up the proper study would be appreciated.

        • Re: FEA Simulation of a tube through test path
          Bill McEachern

          non-linear static with contact should do the trick. What are you measuring in the experiment that you need to match?

          • Re: FEA Simulation of a tube through test path
            Chris Clouser

            nick, I PM'd you but now I see the reply button!  Our internet is worse than dial up.

             

            Bill is right, if you are permanently deforming the material, you must use a non-linear FEA. 

            • Re: FEA Simulation of a tube through test path
              Andrew Cuttle

              Hi Nick,
              If I understand what you're doing, you're interested in tube forming by inserting a tube into a bending form?

               

              I think you'd be better off doing this by hand surely?

              Accumulated plastic strain is probably your dominant worry and that will be easily hand calc'ed?

               

              I can think of a few issues the FEA probably isn't going to cover:
              If you're driving the tube in as the arrow suggests you'll have an a potential buckling issue when it hits the outer edge.

              As the tube passes through around the radius you're likely to get issues with ovalisation, the FEA might capture that but I wouldn't be sure.

              If you drive the tube around the bend capstan effect on semi-rigid bodies is very challenging to model. Generally specialist software is required and even then there are a lot of unknowns due to fiction etc.

               

              Sorry that's not an easy answer but I spent a long time in a previous job modelling tubes (on a different scale) passing around bends and beyond a certain point it's not a simple issue. Static post forming stress/strain is pretty easy to calc (barring the ovality) but during the process, I wouldn't touch it with FEA. Hand calcs are the only way to go and even then brute forcing it with iterative calcs seems is probably the best option (unless your maths are really on point).

                • Re: FEA Simulation of a tube through test path
                  Bill McEachern

                  Hi Andrew,

                  While I do not disagree that a simple hand calc based on curvature could probably do the trick.....a NL calc may handle the buckling, worst case is it can't progress past the buckling point which in itself is probably useful. SWX Sim might not be the best code for this but Abaqus/standard or MARC would be able to solve this without much issue I would think. It could get more complicated if very large deformations occur and the element distortions get too large and then you would need in solution remeshing. All this assumes that the spring back needs to be modeled correctly. If it doesn't it would be a relatively easy solve in an explicit code.

                    • Re: FEA Simulation of a tube through test path
                      Andrew Cuttle

                      Aye, I've seen people do stuff with Abaqus but not used it myself. I think the variables are so great w.r.t buckling directions etc. that detailed analysis is probably so complex to be meaningless beyond that particular analysis. Friction/entry-direction/etc will have so much impact and be uncontrollable.

                        • Re: FEA Simulation of a tube through test path
                          Nick Canfield

                          For this setup, friction doesn't really matter and I am only interested in motion along the 2D X-Y plane. Real units do go to the top or bottom of the fixture depending on how they are inserted but that is a factor I don't need to account for right now. We really need to show the difference in the stress levels for the different geometries to verify that the design changes we are making will greatly reduce the stress loads compared to what we have currently been making.

                          • Re: FEA Simulation of a tube through test path
                            Bill McEachern

                            HI Andrew,

                            Not trying to beat this to death, and with all due respect to you, I have done many post buckling analysis, even with SWX Sim and you can get very close on buckling numbers (~1%) and other parameters can be equally good. It depends on the uncertainties about what you know and your approximating set up to get that close. Sometimes you just don't have good enough definition to get that close but it is certainly possible with the method. I have done many food containers - think a yogurt tub and gotten within 1% on predicting the buckling load. They don't want them to collapse in shipping but they sell yogurt not plastic so the less plastic the better. I have done similar problems but the definition is not quite as certain and you miss by considerably more. Things like the stress stain curve is for a generic resin and not the exact resin, tapered  walls are approximated as straight, thing like that. Arc length methods can get by buckling points frequently with out issue but sometime it can be a bear depending the on the specifics of the problem, particularly when contact is present. Contact makes these kinds of problem far more complicated but without contact the SWX NL product is pretty reasonable. With contact not so predictable. Even ABAQUS and MARC can have trouble when the contact is tricky and in my view its always tricky. Don't get me wrong I agree you can to the stress side of this with a hand calc and it will be a very good estimate of what's going on - buckling not so much.

                              • Re: FEA Simulation of a tube through test path
                                Andrew Cuttle

                                Hi Bill,
                                Sorry, I perhaps wasn't very clear. I quite agree with you and don't have an issue with buckling analysis in FEA.

                                 

                                More that this problems is effectively a motion study defined by material behaviour (coming from FEA or whatever) and the extent of variables makes it probably an excessively challenging requirement, the FEA can predict buckling behaviour but the direction of buckling could be within a 360° axial range and influenced by a wide range of factors. The direction of deformation, whether curvature or buckling then significantly impacts the subsequent deformation and motion study components.

                                 

                                Considering the analysis as individual elements most are manageable in isolation but together... that's something very different.

                          • Re: FEA Simulation of a tube through test path
                            Nick Canfield

                            This is actually the setup of a test fixture to make sure our tubes can survive a 10mm bend radius as they are inserted and removed during use. We have had issues with some of the parts (4/100) breaking at the target location as soon as the unit starts to bend in the fixture. We would like to set up an FEA to analyze the stress along the target line as it more realistically bends in the fixture constraints. We are trying to change both the internal feature's distance from the tip and add a fillet radius to optimize the stress performance.