5 Replies Latest reply on Oct 28, 2013 8:47 PM by Anthony Botting

    Thrmal simulation

    Marco Stegeman

      Hello,

       

      I've started to begin with solidworks thermal simulation.

      I'm goiing to test out a heatsink that is going to cool a LEDlight source.

       

      1. I started out with a thermal study

      2. Apply material (alu alloy 7079)

      3. Adding heat power at the upper surface (6w)

      4. Going to study properties and set to transient (300 sec)

      5. Set inital temperature (293K)

      6. Set convection at all surfaces (8 (W/m2)/K) ambient (293K)

      7. Run simulation.

       

      Result?!?

       

      Result 1.jpg

       

      I don't know what i'm doing wrong.

       

      If I replace the heatpower by a Temperature (403K)

      Then this is the result:

       

      Result 2.jpg

       

      Are there people with some experience with thermal siulation?

       

      Thanks for your help,

        • Re: Thrmal simulation
          Jared Conway

          can you post screenshots of each of your inputs

           

          and a picture of the mesh

           

          and a picture of the study properties

          • Re: Thrmal simulation
            Anthony Botting

            Hi Marco: Try steady-state, first, to make sure it will run through and provide a reasonable temperature distribution at time t=infinity.

            (then later you can do the "on" and "off" transient analyses).

            Perhaps just apply convection on one or two of the fins to see if that works - the "all surfaces" may not solve realistically (i.e., where the LED is mounted you can remove convection from there).

             

            As one check, I recall there is a list result or probe tool so you can check that power out = power in. I believe you will need to manually select all surfaces on which convection is defined, and right-click on the results icon and look for a list function -  and it will provide a total power going out of those surfaces.

              • Re: Thrmal simulation
                Marco Stegeman

                Okay,

                 

                First of all thank you for your replay's.

                I have made a screenshot of all the inputs and I tried first the stready state simulation.

                Here we go:

                1.jpg2.jpg3.jpg4.jpg5.jpg6.jpg

                 

                Okay that's nice!

                Lets try to put convection at all the fins!

                7.jpg

                Okay looks good. The Temperature also looks very realistic.

                Lets try a tranient simulation!

                8.jpg9.jpg11.jpg12.jpg

                 

                What I'm a doing wrong?

                I like the results of the steady state simulation, but i need the results for the transient simulation.

                The temperature is minus degrees?

                 

                I hope this is enough information

                  • Re: Thrmal simulation
                    Jared Conway

                    apply initial temperature to the body instead of the faces

                    • Re: Thrmal simulation
                      Anthony Botting

                      yes do what Jared indicated. There should be a 'entire body' option, or 'whole body'. it will apply temperatures to nodes inside the volume. You may have to select a feature in the tree that represents the whole body.

                       

                      For trouble-shooting, set the element type to be the Tetra4 (the opposite of the 'high quality' mesh). The Tetra4's run very, very fast so you can obtain quick turnarounds and trouble-shoot. Also, I don't believe the first time step will be anywhere near accurate due to the large delta-t, so you could try a 300 second integration at 10 second intervals and see if that gives you something a little better, plots starting at time step number > 1.

                      Also you can probe the 3D plot to generate a 2D graph of the temperature at a node on the fin as a function of time and see if that looks reasonable.

                      Hope that helps.