7 Replies Latest reply on Sep 2, 2014 3:22 AM by Murat Topac

    Simulation results higher than expected

    Stuart Gunn

      Let’s start off with the basics, SW Premium 2014, Flow w/ electronic cooling package added.


      The issue I have is that a simulated assembly of an electronic enclosure is not giving me the expected results, comparing with calculated temperature.


      The assembly in question has numerous components from 2-resistor components, multiple PCB’s, volume heat sources and a bunch of different custom materials like thermal gap pads, aluminum for heat sinks, QFN's, FC PBGA and so on. The data for the custom materials and components are from the manufacturers data sheets. Radiation is on, heat transferred through solids, gravity on, ambient is at 40 (and materials are at 40) and modeled as internal with a sudo box that is environment pressue on all but the bottom (I have found this better in the past than loads of small 'filler' parts).


      There are local mesh’s for the small components, contact resistance, boundary for the enclosure and the simulation itself runs.


      It is modeled off a prototype unit that my company is trying to develop and the issues I have is that the temperatures I get from the simulation are mostly way higher than the model.


      For example, I have a SSOP device that have a volume power of 0.57W that is matted to the PCB (setup as a 6 layer). The device should reach a temperature of around 102C in the 40C ambient, but the simulation is showing it to be 180C+.


      The other reason I know it the simulation is wrong is that the unit was developed by another team in my company and they used a different simulation package than SW and got different results, which line up more with the calculated temperatures.


      I have built the assembly twice, re-made the thermal project 3 times and still get the same answer.

      Has anyone else had a problem where their simulation gives higher than expected temperature results? If so, what was the cause and solution?


      I have attached a 3D PDF of the model, made the CAN and Heat Sink transparent, removed one side and as you can tell, it is very basic in design.

        • Re: Simulation results higher than expected
          Jared Conway

          got a new comptuer and for some reason cant' look at that PDF. it sounds like it is just a 3d model which i don't think i need to take a stab at this.


          have i been here before? yep. too high, too low, you name it, i've been there.


          was it software related? nope..always setup related.


          what i'd suggest is starting to list the assumptions you think your model is making.


          the other thing i'd suggest is taking a look at how the other model was setup in the other software. i've been in situations where an error in the setup of another software put them closer to the physical results but when setup equivalently to flow simulation, they got the same results. i guess what i'm saying is make sure you're comparing apples to apples here.


          since i'm not there to compare against your physical setup, the next thing i'd do is start looking at some parameters other than just that one component temperature. what about other components? other locations? other parameters.....if everything is offset a certain amount within what you would expect them to be related, it definitely points to a setup issue.


          then i'd start doing some variance testing. change the wattage by 10%. does the system respond the way you expect? if it does, again, unlikely software.


          we've gone through this process with a few customers. if you're interested in an experienced eye, don't hesitate to drop me a line at jared@hawkridgesys.com.

            • Re: Simulation results higher than expected
              Stuart Gunn

              Thanks Jared for the comments,


              Unfortunatly I do not have access to the other SW, different company sending me the final data.

              I did try changing the wattage and the temperatures did change accordingly, but they are still higher than expected.


              What I have found out is after putting in a sketch line and plotting the temperature profile of the component to the outer walls (thus through the entire model, intersecting the device). There is a hugh temperature drop between the component and it's mated PCB (currently around 50C).


              I originally had it as a sub-assembly as it was used in other models, and recently moved all the components into the main and rebuilt, but the results are still the same. I've also changed the PCB to copper thinking the PCB material was wrong, that lowered all the temperatures but the large drop is still there. I have also put in a contact thermal resistance and that did not change anything (the contact resistance was 2.9e6 C*m^2/W)


              The thing that puzzles me is that I have a similar component setup elsewhere in the model that works right and shows a good temperature profile through it's PCB. I have not altered this part or it's bond and that's been working all along.

            • Re: Simulation results higher than expected
              Murat Topac

              Hi all I am having same problem with similar setup . I have been trying different mesh density But results are higher than expected also I am trying to find sample tutorial by ECooling user but I couldnt achieve to find it. If any one else can solve the issue or have experience please kindly share it.

              • Re: Simulation results higher than expected
                Kim KyuHyung

                In natural convection, Radiation must be considered.


                and check B/C [Wattage, PCB conductivity, ambient temperature etc]

                • Re: Simulation results higher than expected
                  Murat Topac

                  Thanks for support Jared and Kim


                  I improved the simulation results and I got expected result considering your advice. I improved the mesh density by using manual initial mesh setting. Also I applied radiation to my simulation. the result is better.