8 Replies Latest reply on Jun 15, 2009 1:41 PM by Robert Stupplebeen

    Large Structure Analysis

    Russ Johnston
      I have an analysis probalem that I am struggling to solve and I am hoping someone can steer me in the right direction. Here's the details:

      A large structual platform 45' x 25'
      Currently the top structure is made of W18 x 50 beams with a 1" plate welded to the top flange
      Under the top structure is another W18 x 50 structure that will contain a lifting mechanism
      The load (100,000 lbs) is applied over 6 pads on the platform top surface (indicated by the red squares in the attached images)
      The lower structure is essential supported by the end beams only (indicated in blue)
      Both frames have been modeled using weldments

      What is the best approach to solve this analysis? I am mostly concerned with deflections (the goal is less than 1/8" deviation between all 6 load pads).

        • Large Structure Analysis
          David Anderson
          are you using fea?

          if so simply apply the load to the 6 pads with a total of 100k (dont forget to apply gravity), constrain the underside where the lifting mech is attached and run a static analysis. i do not think you will need to run a large deflection analysis as the delfections are small compared to the model size.
            • Large Structure Analysis
              Russ Johnston
              yes...I am usings Simulation....

              I tried this approach. The mesh fails everytime.
                • Large Structure Analysis
                  David Anderson
                  then you have geometry that the mesher cannot handle. look for slivers or geometry that would create elements with large aspect ratios. i assume you are meshing with high order tets?
                  for the beams you have, using a beam element will certainly reduce the element count and will increse run speed. is this on big solid or an assembly?
                  if it is an assemby which would be the best way to model this using contact elements, try meshing each part individually to until you find the problematic geometry.

                  alternatively, you can refine the mesh around holes and sharp transistions, but this will drastically increase your model size. first, try defeaturing, that is removing features that have no impact on the analysis such as bolt holes, fasteners etc.
                    • Large Structure Analysis
                      Derek Bishop
                      You should be able to run this as a beam and shell mesh analysis. You will need to define contacts between each of the beams and the shell surface. Give us more detail on what is causing the failure or the failure messages. How are you meshing this model at the moment? What are you computer specifications?
                • Large Structure Analysis
                  Robert Stupplebeen
                  You also have 1 plane of symmetry which will speed up your solution very well. You could probably simulate this with only shell elements by smearing the properties of the beams into the shells locally by making sure the that "I" is appropriate and fudging the mass. I hope this helps.

                  Rob Stupplebeen
                    • Large Structure Analysis
                      Russ Johnston
                      I tried running this as mixed beam and solid mesh but the result don't seem right. I get very small deflections (less then 0.001"). The stange things is that if I run it without the deck plate I get beam deflections in the range of 1".

                      The 100,000 lbs will actaully be offset side to side and front to back so I can't really use symmetry.

                      Can you give me some direction on how to do this as a shell mesh?

                      I a m running this on Core 2 Duo, 4 GB RAM, XP 32-bit
                        • Large Structure Analysis
                          Derek Bishop
                          If you are not familiar with shell mesh you might find it best to start with a simple model of a plate and beams and play around with it for a while to work out what does and doesn't work. Basically you add a surface to the plate and then exclude the solid plate from the FEA. I normally do this by adding a delete feature and deleting the plate. Beams should be modelled as weldments. The beams and surface are then automatically recognised in the study.

                          The biggest problem with shells I've found is defining contacts. Contacts between the sides of beams and shells are easy enough. Just select the beam icon and the surface in the contact property maager. Shell to shell contacts can get tricky. You will know when you haven't been successful because pieces of the model 'fall off' when you run the study or you get the message saying the model is not properly restrained.

                          Beam and shell mesh studies are relatively quick to run. What kind of CPU have you got?
                      • Large Structure Analysis
                        Robert Stupplebeen
                        Check out the section describing "I" moment of inertia in your strength of materials text book. The difficulties come into what area of the plate has the modified properties and what is the neutral bending axis.