6 Replies Latest reply on Mar 14, 2013 2:18 AM by Jared Conway

    Meshing and Analysis of Large Models

    Cory Langenkamp

      I have a tubular drilling mast structure that is around 50 ft tall That I am trying to analyze. I have tried a beam analysis, but the results I got back were questionable, and my supervisor has decided not to persue that method of analysis.

       

      The mast has been joined into a large single part. The mesh was difficult to create, and as such is rather fine do to the thickness of the tube. The model has approximately 30 million DOF, due to the large mesh. Applied loads are only forces, and restraints are simple fixed hinges and roller-sliders. The mesh itself takes about 25 minutes to complete, and the analysis takes about 15 minutes.

       

      My problem arises when I try to view the results. When trying to open any results files, Solidworks enters a state of fluctuating lock-ups, and the results, if they decide to show up, take a bout 30 minutes to do so. And if they do appear, they are usually garbage. A change in any simulation parameters displays the EXACT SAME results, regardless of the machine the analysis is run on. (I have tried on 2 seperate machines with similar capabilities, transfering the model only, and none of the results files)

       

      Has anyone ever seen this before? Does anyone have any ideas as to why this is happening?

       

      Attached are two images of the stress results. I'm sorry, but I cannot post the model for legal reasons.

       

      The machines I ran this on are running Windos 7 65-bit on Intel i7 3.40 GHz processors, with 32GB memory. The working directory is on a SSD.

       

      Thanks for any help you can provide.

        • Re: Meshing and Analysis of Large Models
          Bill McEachern

          It looks like a beam problem to me. I suppose you could use shells to get the dof down to something sensible. What was the problem with the beam analysis?

          At 30 million DOF this is the largest problem I have heard of being done with the tool. Its remarkable that it provided any results. They do look like junk I have to agree.

            • Re: Meshing and Analysis of Large Models
              Cory Langenkamp

              The beam analysis had very poor stress flow through it. What I mean by that is the stress results would flip almost completely from full compression to full tension across ajacent nodes. I don't have those result to show you anymore. It could have been a result of having contact between plate gussets and the beams in the simulation. This was one of my first attempts at beam analysis, and my Supervisor and I both did not feel comfortable relying on those results.

               

              This is my 4th analysis grouping of a model this large, and I have never had this problem before. This large of analysis is actually quite common at my company. I am currently revising the model in an attempt to shrink the mesh size.

               

              Thanks for your response.

            • Re: Meshing and Analysis of Large Models
              Cory Langenkamp

              For anyone interested, I solved the issue. The analysis model was a 2nd generation Pack-and-Go. On a large model, Solidworks seems to have trouble with multiple generations of pack-and-go's, and numerous geometric problems arose. I was able to fix them, and the mesh and analysis both went off without a hitch, and appears to be presenting valid results.

               

              What I think was happening was I had a very fine mesh to compensate for the geometric errors in the model. This mesh was inerently flawed, and resulted in skewed results.

               

              This of course is mostly speculation. All I know for certain is the model ran fine after I fixed the errors in the model.

               

              I hope this helps anyone with a similar problem.