10 Replies Latest reply on Jul 24, 2014 3:12 PM by Jared Conway

    Help troubleshooting 2D thermal simulations

    Paul Mark

      Hello again all! More problems yet. I have a 3D thermal simulation, which involves a rod split into 11 concentric rings and a heatsink surrounding it, cooling it down to the case below it. I am loading the inner 10 rings at 10W each for a total of 100W, and the base of the heatsink is held at 20C I believe. The 3D simulation runs fine and gives me the "correct" answers. I want to convert it into a 2D simulation so that I can quickly test a number of configuration changes without having to simulate the entire environment. However, when I do that, everything is broken. The thermal flux is too small by a factor of 2, and when I calculate the heat flux through the bottom of the heatsink I get values like 3W, so that the temperatures, etc. are completely wrong. I am baffled as the 3D version of the same thing works just fine. May I ask for someone's time and effort to help me troubleshoot this issue? I have attached the assembly with simulation files for 2D attached. The issue seems to be the "Indium_4MM" piece or "Crystal_11", at which point the heat seems to disappear, although the heatflux at the "crystal" is still too small by a factor of 2: it should be on average 265000 while the simulation gives me an average of 133000. Thanks so much!

        • Re: Help troubleshooting 2D thermal simulations
          Nicholas Luyster

          Hi Paul,

           

          Before we dig in, did you converge the 3D simulation?

           

          All the best,

           

          Nick Luyster

          SolidWorks Simulation Training

          • Re: Help troubleshooting 2D thermal simulations
            Jared Conway

            is there a reason there are no bcs on any of the other faces? they will be assumed to be insulated

            when you post a model, its a good idea to mention what model and what study to take a look at. i'm looking at the crystal model and the thermal 2d sim.

            is there a reason why you wouldn't choose the 10 rings and do 10w each like you did in the 3d?

            could you also elaborate on how you are doing your checks?

            i ran your model with 100W in and at the 20deg C out, it says 100W so everything looks ok to me from a balance perspective.

            is it possible you have materials incorrectly defined? or maybe some contact not working properly changing the heat path?

            • Re: Help troubleshooting 2D thermal simulations
              Paul Mark

              Nick,

              The 3D simulation was set to steady state and a converagnce tolerance of 0.1%, with the under-relaxation factor set to automatic. The calculation completes sucessfully, and so I assume it has converged to the aforementioned tolerances.

               

              Jared,

              Yes, I should have defined what model I was simulating, it is indeed Crystal2, and the 2D thermal simulation within it. Those other boundaries go to air, and I do not have FlowSimulation to calculate convection, so I left them as insulated. I defined the outer ring as 100W just to simplify things to make sure that wasn't the issue, and it didnt help. The problem isnt the 100 watts, its that the heat flux at the bottom should have an average of about 80,000 W/m^2, given the surface area of the base and the heat power of 100W, but the simulation reports it as instead having an average of 2400 W/m^2. The ratio between these values is almost exactly the Section Depth, but that may just be a coincidence. This is also true at the outer boundary of the "Indium" part, where the flux should be 260,000 W/m^2 but is 8600 W/m^2 on average, again about the same ratio as the Section Depth of the part.

               

              Interestingly, changing the section depth from 30 mm to 15 mm doesnt change these heat fluxes at all. The base is still ~ 2400 W/m^2 and the outer boundary of the "Indium" is still 8600 W/m^2, but they should be double that since the heat power is still the same but I just cut the area in half.

               

              It is possible I have the material incorrectly defined, are there common errors people make? Changing the material to titanium-8Mn which has a similar thermal conductivity didnt help with the heat disappearing.

              • Re: Help troubleshooting 2D thermal simulations
                Paul Mark

                Some more testing seems to indicate that its an error in how SolidWorks is accounting for the 3rd dimension in its calculations. Putting a heat flux of 250,000 W/m^2 on the outside of the "Indium" part gives the correct heat fluxes within the copper heatsink. Putting the same flux on the outside of "Crystal-11" results in the incorrect values mentioned above. When applying the 10W per part to the inner 10 parts and measuring the heat flux between "Crystal-10" and "Crystal-11" gives the correct value of 427530 W/m^2 that would be expected if it surrounded a 100 Watt source.

                 

                Interestingly enough, the heat per meter, arrived at by taking the heat flux and multiplying by the circumference or length of the boundary is always around 100 W/m, which has the same numerical value as the heat load, 100 W, but very different units. The correct value for this quantity, which should be (and isn't) conserved in 2D calculations, is 3333 W/m.

                 

                This may be due to the "Indium" and copper heatsink parts being in an assembly, which was inserted into the assembly in which the simulation occurs, as it seems to be an issue only with these parts.

                Diagram.png

                 

                EDIT: Ah HAH! Dissolving the subassembly which contains the "Indium" and the copper heat sink parts solves the no-flow problem. It doesnt, however, solve the original error, which is that SolidWorks is treating them differently for some reason.