8 Replies Latest reply on Jul 20, 2011 7:03 PM by Roman Snell

    Simulation with welded frame question

    Roman Snell

      I have a model of a conferance table with a frame made from aluminum tubing. I'm trying to figure out if the table is pushed if it will buckle. Would I have to run a static test or buckle test? Do I have to call out all the welds? Also the bottom of the legs will not be fixed so do I have to model a floor and add some kind of friction between the floor and the table? I've done simple simulations before but this is more advanced for me.

       

      Thanks!

        • Re: Simulation with welded frame question
          Michael Feeney
          • I'd use weldments (beam elements) and run a buckling analysis.
          • If the table is a frame, you simply draw a wireframe (line drawing of the frame). You then create your cross-secitons for each unique member type (if you only have 1 type of cross-section, then you only need 1 profile). You would then apply the cross-section(s) to the appropriate lines in your wireframe drawing.
          • NOTE: for a buckling analysis you have to specify a force. With that said, you should determine what kind of forces you are expecting (worst case scenarios) before modeling the problem. Another way to approach the problem is to do a stress analysis with these determined loads and see if the stresses are beyond yeild. (This approach is a bit more common because, with tables, the legs are typically short compared to the other dimensions and buckling doesn't really happen before the welds break or the surface collapses. Although, if you have thin legs, you might see buckling first. As far as the welds failing, you can specify those parameters and apply specific yeild values to your joints. You could always model the table without these details and observe your peak stresses at the joints.
          • For the boundary conditions, I'd use a pin for 1 leg and rollers for the other 3 legs. A pin boundary condition eliminates vertical and horizontal motion in a given plane, but allows for rotation. A roller eliminates vertical displacement and allows for both rotation and lateral motion in a given plane. Essentially, all of your legs would be fixed onto a plane, but three legs would be allowed to slide away from the leg with the pin joint. I do believe this configuration is sufficiently constrained. One thing to note, a single column with pins on both ends produces the smallest critical load for buckling. Typically it takes a larger force for a fixed end condition column. In otherwords, you are adding to the factor of saftey by modeling with pin joints.
          • Another configuration I would try would be 1 fixed-end boundary condition on 1 leg and rollers on the rest. You can change the boundary conditions very easily and re-run the simulation. Your report could discuss the effects of the different end condtions and perhaps use the lowest quantity as your safe value.
          • I would not consider friction as it adds an unnecessary complexity to the problem. Just remember, for tables, it's pretty uncommon to design things really close to the point of failure. You would want a pretty decent factor of saftey, right?

          good luck : ]

            • Re: Simulation with welded frame question
              Roman Snell

              I would agree that the welds are the main problem to be looking at here. With that said how do I test this? Do I run this as a staic test? In my connections I enter my "edge welds". This gives me the size of weld I can use. From there I don't know where to go. I'm not an engineer so I'm not sure what exactly I'm looking for. I wish it was a simple this is going to fail or this is going to pass. I know I can use "factor of safety" but I think its more complex than that.

                • Re: Simulation with welded frame question
                  Michael Feeney

                  You would need to know the size of the weld and it's strength properties (which are a function of a lot of parameters, heat treatment, weld type, etc.). In short, I would not model this with the welds. It's really not worth it. If you have a good welder making the table, he should be able to make the weld joint near as strong as the parent metal.

                   

                  Do you know the typical loads the table will see? If you don't then there is no point in modeling the table. If you do know, then do a static analysis with the load type (point load, or distributed load across the table surface, or whatever the loading condition that this table is being designed for). Run the boundary conditions I discussed previously and compare the results (my guess is that you won't see much of a difference).

                    • Re: Simulation with welded frame question
                      Roman Snell

                      We are using aluminum tubing at .125 thick. From my understanding aluminum looses some of its strength after welding. This table will be about 9 feet long with a 200lbs. top. The main concern is if it were to be dragged would the welds hold up.

                        • Re: Simulation with welded frame question
                          Michael Feeney

                          Apply rollers on two adjacent legs and fixed end supports on the other two legs. apply horizonatal load at one end of the table. Iterate the load until you see yeild for aluminum at the joints. Divide that by 2 for a factor of saftey of 2. I say it's done at this point. Remember your fea simulation will capture the peak stresses with beam elements at your joints. Obviously, the peak stresses are going to happen at the joint where the surface meets the legs. The picture attached shows the setup. This is the worst case static scenario (impact will be different). It's as if you are sliding the table and it snags something at the end of the leg. Once again do a static analysis. The table I drew is stupidly simple, but it gives you the idea. I assume you guys probably have some 90's reinforcing the legs?

                          ex.png

                  • Re: Simulation with welded frame question
                    Michael Feeney

                    the only time the weight on the table (the 200 lbs and whatever is sitting on it) matters for a collapse scenario is when there is sufficient bending in the legs which then allows a moment arm to bend the leg further. If you want a detailed analysis, maybe you could do the solid element analysis (shown in the second picture) with a high mesh concentartion around the joint and apply the load shown and also apply a load of 200lbs in-line with the leg. One thing that i do want to stress is that you not be overly concerned about modeling the weld. I think a basic stress analysis at the joint will reveal enough information. If you have reinforcements at the leg joint (like a 90 degree bracket), it'd be difficult to model the problem with beams. You would probably need to do solid elements (in which case you would only model 1 leg and a section of the table, which you would have to deem sufficiently large to keep the effects of the BCs away from the stress concentation).