17 Replies Latest reply on Mar 10, 2014 6:07 PM by Jared Conway

    Can't seem to get it right

    Caius Mihai Canea

      Hi,

       

      I'm trying to test if a part that I design is strong enough for what I need but I can't seem to do an FEA on it. I tried making it one single part and I got a message that the model is too unstable, so I activated the soft springs options, but nothing. It told me that the displacement was too big and didn't give me any result.

       

      So I started again, this time make the model out of sepate parts and assembling them. I put in all the fixtures and forces, but now it gives me another error (picture). It suggested using the direct sparse solver, which gave me the "not stable" treatment again.

       

      The main objective was to see if the model is strong enough and if the welding is.

       

      Help would gravely be appreciated

        • Re: Can't seem to get it right
          Shaun Densberger
          1. It looks like you're running this as a non-linear analysis; are you sure you need to be? If not, then I would suggest that you start with a linear static study and only move into non-linear if the results suggest it.
          2. It also looks like you're using shell elements; have to checked to make sure that your mesh is matched at the part interfaces? I know that in some older versions of SW, it seemed as if the user had to go through some additional steps to ensure node sharing at part interfaces.
          3. Just a word of warning; accurately simulating the stress within a weld is rather difficult (and involves fracture mechanics), so I would focus on getting stresses away from the welds and back-calculating the stress in the weld based on non-numerical methods.
            • Re: Can't seem to get it right
              Caius Mihai Canea

              1. I was under the impression that i'm doing a linear static study, at least that's what I selected when I chose new study. why do you say it's a non linear study ?

               

              2. I didn't want to use shell, but SW won't let me put in a weld unless it's a shell. Doesn't work on solid bodies.

               

              3. By taking the stresses away from the welds, you mean redesigning it so that the load doesn't stress the weld itself so much, right ? I'll try doing that. As for back-calculating, don't really know what you mean, so any additional info would really held me.

               

              Thanks

                • Re: Can't seem to get it right
                  Jared Conway

                  you're using static, i think the results options thing threw shaun off, that is new for 2014. (it used to only be in nonlinear)

                   

                  if you're new to sim, i'm guessing you're new to shells, i would start with some really simple shell cases to get a hang of it. and start simple. start either at the load or the fixture with a gravity load and get 2 parts connected to work before you work up to the bolted connections..etc.

                   

                  what shaun's suggestion is to look at the general stresses in the assy with the parts bonded together (solids) and then use to estimate the welds necessary.

              • Re: Can't seem to get it right
                Jared Conway

                start a bit simpler

                switch all the parts to solid elements (right mouse button click them)

                make sure global bonded contact is enabled

                remove your weld connectors

                turn off soft springs

                get it to run

                 

                then you can start looking at other stuff

                • Re: Can't seem to get it right
                  Caius Mihai Canea

                  So,

                   

                  I think i managed to get it right and from what plots it gave me, the design is fullproof. It's actually a bit overdesigned, but oh well.

                   

                  From what I could tell, the problem was that I didn't contrain it enough, so I added a few more fixture (hope that i did that correctly). Also, I forgot to put in the contact sets.

                   

                  I did all of that and the analysis ran without any error. What do you think ?

                  update.JPG

                    • Re: Can't seem to get it right
                      Jared Conway

                      my suggestion is to look at displacement with the deformation scale at decent high number to see if it is bending/twisting the way you expect but the results look reasonable. this will help determine if you overconstrained it or not as well

                      • Re: Can't seem to get it right
                        Jerry Steiger

                        Caihus,

                         

                        I'm not skilled in using SolidWorks Simulation, but are you sure that the constraints you added actually make physical sense in the way your part is used or tested? If you don't get the boundary conditions (constraints and loads) correct, the simulation is useless.

                         

                        Jerry S.

                          • Re: Can't seem to get it right
                            Caius Mihai Canea

                            Jerry,

                             

                            Before putting in those constraints, Solidworks used to tell me that there are not enough and so on (see above discussions). I started watching some tutorials on Simulation and applied some of the things shown in them.

                             

                            There are 3 additional constraints that I added and I don't think they should interfere with the analysis. I applied a cilindrical constraint on the lower tube to stop it from sliding, one on the middle tube to stop it from going sideways and the last one to the upper part to stop it from rotating. When I put them, they actually make sense and I don't see how they could make the analysis null. I could be wrong though...

                             

                            That's why I posted the update, to get opinions...

                        • Re: Can't seem to get it right
                          Caius Mihai Canea

                          Here is the model Jared.

                            • Re: Can't seem to get it right
                              Jared Conway

                              to interpret and comment on your setup, it might be good to create a drawing of how it gets used and how it gets loaded. or at least describe in as much detail as possible.

                                • Re: Can't seem to get it right
                                  Caius Mihai Canea

                                  The rectangular part is inserted onto a rectangular tubing and tightened with the two bolts. In the lower tube, a rod is inserted and on that rod a weight will be placed. So basically, the rod will be pressed down and thus, the lower tube as well.

                                    • Re: Can't seem to get it right
                                      Jared Conway

                                      What is the goal of the analysis?

                                       

                                      Without looking at your files, from your description and the picture, your model is probably over restrained.

                                       

                                      Remember that fixtures fix things with respect to the environment. So your vertical tube will always stay vertical and it looks like your horizontal tube will stay horizontal.

                                       

                                      Did you ever check displacements to make sure they are reasonable?

                                       

                                      Going back to my original al suggestion. Start simple and make sure as you go things are working. This problem should only require a restraint at the bolted connection and something to act like the square tube. The lower tube should have a force on it.

                                        • Re: Can't seem to get it right
                                          Caius Mihai Canea

                                          I simplified the design a bit (don't think it should influence the results too much) and got some different results. The design is still good, but I didn't get such ridiculous numbers anymore.

                                           

                                          What do you think ?

                                          suport teava-Study 2-Results-Displacement1 v2.analysis.jpg

                                            • Re: Can't seem to get it right
                                              Jerry Steiger

                                              Caius,

                                               

                                              That looks better than the previous constraints, but I still would not trust it.

                                               

                                              It looks like you have fixed the top plate, which means you are not going to have any stress in it. It would be more realistic to leave the original shape and fix the top two plates. More realistic again would be to include the square tube that goes in the center and fix its ends.

                                               

                                              It looks like you spread a downard force over the whole lower pipe. You would be better off to put a bearing stress on the lower half of the tube. Better again would be to include the rod and put your weight on the ends of the rod.

                                               

                                              Assuming the load is purely vertical, you could cut your model down the center line and it should run about four times as fast. Assuming the load is perfectly centered, you could cut it into quarters and it should run about 16 times as fast.

                                               

                                              Jerry S.

                                              • Re: Can't seem to get it right
                                                Jared Conway

                                                its getting there. add a deformed scale, taht will help you better visualize it

                                                 

                                                then you can add the next part.