5 Replies Latest reply on Mar 6, 2013 1:46 AM by Jared Conway

    Varying simulation results - Desktop vs Desktop and Desktop vs Laptop

    Yuvaraj Dhanasekar

      Simulation results vary from machine to machine.  try yourself  ..run a simple "plate with hole" in desktop and with same setup in  laptop.  Results vary drastically.  The same happens between the system configuration of different RAM size.  An iterative solver should be the same ..no matter whatever processor it is running upon.
      In one of my case i got FOS value above 1 in desktop ...but below 0.7 in laptop....:(.
      I dont know whether my design is safe or not....For files ...you may try with any file..that are in standard simulation tutorial.

        • Re: Varying simulation results - Desktop vs Desktop and Desktop vs Laptop
          Dave Laban

          I get consistent results between 3 different desktops and a latop.  Sounds more like an issue with your particular hardware / installs.

           

          What version of SW?  Is ther service pack the same on both?  What hardware are you running?

            • Re: Varying simulation results - Desktop vs Desktop and Desktop vs Laptop
              Yuvaraj Dhanasekar

              Hi Dave.

               

              SW 2013 sp 1 on desktop as well as in laptop. My desktop runs with 4GB of RAM while laptop with 8GB - both are windows 7 with core i7 processor  .  My concern is how does the result change with different systems.  It should be the same no matter whatever system (configurations minimum to run sw) that i run it in. Is solvers accuracy connected with the processor that i run in??...

                • Re: Varying simulation results - Desktop vs Desktop and Desktop vs Laptop
                  Josh Brady

                  Is this simulation or simulation express?

                   

                  My guess is that the mesh size is being automatically optimized by Simulation for your PC's performance capability.  The smaller the mesh, the more accurate the results (generally). 

                   

                  If you are relying on simulation to decide if your design is safe or not, don't.  I really hope you don't mean "safe" as in "won't kill people".

                   

                  If you are using Express, it's not intended for anything more than a rough-pass, sanity check.  If your FOS is less than 5 on express, you should do the hand calcs.  If you can't do the hand calcs, hire a pro.

                   

                  If you are not using Express, you should not accept anything less than FOS of 5 unless you know enough about FEA theory to understand on your own why you are getting different results on multiple machines.  Again, do the hand calcs.  If you can't, hire a pro.

                    • Re: Varying simulation results - Desktop vs Desktop and Desktop vs Laptop
                      Yuvaraj Dhanasekar

                      Hey Josh,

                       

                      I am running on simulaiton premium.

                       

                      How come system configuration affect the mesh?? .  A solver or a mesher  should give the same answer no matter whatever config system i run in (given: i have minimum config) . My concern is how come system config affect the Solvers or Meshers settings. 

                       

                      No i dont rely on simulation alone.I really meant "safe" as "design limit" or "Safety limit"..:)

                       

                      I know some of FEA theories that are used to predict material failure. And yes as you said .. I am just researching on my own what could cause the mesher to give different values..and thats why i am here to get some help..:).. I can do some hand calcs...  I am trying to become a Pro..:P

                        • Re: Varying simulation results - Desktop vs Desktop and Desktop vs Laptop
                          Jared Conway

                          Hi Yuvaraj, can you please upload the example file from each computer and a report from each computers study? That should help us pinpoint the differences.

                           

                          I'm making the assumption you're moving the file from computer to computer and rerunning vs setting up the problem on each. If that is the case, everything should be the same. Mesh is a good place to check but it should be the same from computer to computer if you are running the same study. My guess is you have soft spring or something similar enabled in the study.

                           

                           

                          Also, if you haven't already done so, my recommendation is to run a verification problem so that you have the closed form solution. Or calculate yourself so that you know what the answer should be and what computer should be right.