10 Replies Latest reply on Sep 19, 2013 11:27 AM by Flavio Gaggi

    Convergence

    Flavio Gaggi

      Hi,

      I've attached the required files

        • Re: Convergence
          Jared Conway

          https://forum.solidworks.com/thread/72015?tstart=0

           

          cross post

           

          going back to some of my original questions, what specifically are you concerned about regarding the data?

           

          what was the reason your simulation completed, was it because of the goals?

            • Re: Convergence
              Flavio Gaggi

              The aim of the work (thesis) is to simulate the behaviour of the three resistences, in particular to find the hottest area, calculate the temperature distribution and heat exchange coefficient. That's why I set the goals as max temperature on every resistance. the other goal is to control the stedy state on the simulation.

              To validate the model I'm going to measure the themperature of the real reistance in 20 points by termocouple. The problem is that I will do the test as soon as possible but actually I don't have any data. I know that is quite difficult to say if a model works or not without any real data,but I would like to build as reasonable good as possible model in SWFS.

                • Re: Convergence
                  Jared Conway

                  ok. but what is the concern with your results?

                    • Re: Convergence
                      Flavio Gaggi

                      if you check the goal data you can see that the difference between max and min temperature (the delta) over the last 94 iteration of the solid temperature is more than che criterion. But if you try to calculate the delta on the last 94 iteration of the outlet bulk average temperature, you can see that it's right. So I don't know what is the right value of the solid temperature...this is what I'm concerned on.

                       

                      I have another question, how can I find the right mesh "convergence"? do you think that the simple impovement of the mesh level(with default settings and narrow channel enabled) can carry to the convergence of successive result or is better way to work on initial mesh settings to achieve this target? I.e. use narrow channel refinement only where is needed i.e. oround the resistance instead of let it works around every object

                      thank you

                        • Re: Convergence
                          Jared Conway

                          I see what you mean. Create a rest file and run it to confirm other problems report properly.

                           

                          If they do, great!

                           

                          Now rerun your model and see if the results come back the same or different.

                           

                          From there I would need to see the file to troubleshoot further.

                           

                          Regarding your other question, there is no right answer. The best mesh is the one that gives you accurate results based on testing or hand calcs. Chicken and the egg baiscally.

                           

                          I would recommend looking at the tutorial on meshing and the solving eng problems doc on recommended meshing practices. There are several articles in the kb as well.