7 Replies Latest reply on Nov 19, 2013 10:42 AM by Mike Pogue

    Thermal plot with air pockets imported into static stress sim

    Justin Strempke

      I ran a simulation with a part having air cavities for which I created air bodies to properly model the heat flux and temp distribution we were concerned about.  The second step is to run a stress analysis on using the temperature distribution, and it requires the same mesh (meaning all bodies included).

       

      My question is since I need to apply mechanical properties to the air, and wanted to get opinion on if using extremely low E and nu values is appropriate (say .1 and .01).  There is a good temp difference (~100-degF) so expansion would be an issue with the solid part and I don't want false stresses imparted due to the 'solid gas'.  I don't have access to SW Flow, so a CFD temp plot isn't an option.  The chamber is very small anyway, and I would believe conduction is taking place but not in a highly influencing manner.

       

      Thanks!

        • Re: Thermal plot with air pockets imported into static stress sim
          Jared Conway

          1. have you tried just excluding the air body? i think that should work.

          2. low E, v and CTE should work as well.

          3. best approach would be to not model the air and replace with the right convection coefficients instead. or CFD > thermal stress.

            • Re: Thermal plot with air pockets imported into static stress sim
              Justin Strempke

              Jared Conway wrote:

               

              1. have you tried just excluding the air body? i think that should work.

              2. low E, v and CTE should work as well.

              3. best approach would be to not model the air and replace with the right convection coefficients instead. or CFD > thermal stress.

              1.  I believe I did, but will try again.  I may have gone about it differently the first time (like supressing the bodies first in the model).

              2.  They appeared to have, but wanted to ask

              3.  I know the convection method would be best, but the bulk ambient temp in the cavity is unknown as it'll change with the surrounding body's... 

               

              UPDATE:

              1.  Just retried it, got a message "STOP: Number of parts is not equal to the thermal model", so no-go.  Guess #2 will have to work!

              • Re: Thermal plot with air pockets imported into static stress sim
                Doug Rawlings

                I'm in a very similar situation.

                 

                 

                Jared Conway wrote:

                 

                1. have you tried just excluding the air body? i think that should work.

                With the air included in the thermal study (to include conduction) but not present when I attempt the static study I get the same message about the number of parts.

                2. low E, v and CTE should work as well.

                In reality air is a compressible fluid; not only can it be sqaushed but it also flows. Even with a low E and zero'd v I believe Simulation treats the body as a solid, even if it has a low elastic modulus and zero Poisson's ratio.

                3. best approach would be to not model the air and replace with the right convection coefficients instead. or CFD > thermal stress.

                But convection coefficients rely on a bulk temperature whereas I'm looking for heat transfer between bodies.

                 

                Potentially I could apply a convection load to each body using a Temperature curve in an attempt to approximate the temperature of the other body?

                 

                It would be nicer if suggestion 1 worked and we could just use the air in the thermal study and then exclude it from the static study.

                 

                I wonder if there's another way to do it?

                  • Re: Thermal plot with air pockets imported into static stress sim
                    Mike Pogue

                    Small pockets of air should be treated as conductive. Not only is there no real thermal gradient to get convection going, the empirical correlations for convection in closed vloumes just don't go down to very small scales. Unless the air pockets you are talking about are (just making up a number, look in your heat transfer book) 4 inches across, Jared's suggeston is appropriate and conservative.

                      • Re: Thermal plot with air pockets imported into static stress sim
                        Doug Rawlings

                        Yes, I'd rather treat them as conductive but I'm not certain how to do so. Which of Jared's suggestions are you advocating?

                         

                        I've tried the first two without success and am hoping to try the third. But maybe I've misunderstood something, I often do!

                         

                        Edit:

                        Yes, I'd done something wrong. Not certain what but may have been a contact issue. Basically I'd set up a small study with a piston loading a body of air, to see what happened. I've just done that again and the results look much more acceptable, should be attached.

                        Air_piston_MBpart_displacement_deformed.JPG

                        The material applied to the body under the piston in this case was a copy of the standard SolidWorks air but with an Elastic Modulus of 0.1 Pa and Poissons Ratio of 0.

                         

                        I'd tried this before comgin across this thread and in my previous model the piston hadn't been displaced and I'd interpreted that as the air being incompressible. I can only think that maybe I'd gotten a contact setting wrong.

                         

                        I feel a bit happier about using this method now, having had some better results and other people suggesting it.

                         

                        Message was edited by: Doug Rawlings