9 Replies Latest reply on Nov 12, 2008 7:44 PM by sandeep pawar

    Boundary layer?

      I am not clearn on whether or not something special needs to be done with FW in order to ensure that the simulation takes into account boundary layer conditions/attachment/detachment. The test subject is a high performance heatsink where boundary layer effects could alter efficiency. Any help would be appreciated.

      Thanks,

      -Martin
        • Boundary layer?
          I think I just answered my own question. I found the section on Turbulence in FW help. It seems to indicate that the boundary layer effects should be part of the standard calculation.

          So, the next question is: How good are the results?

          Thanks,

          -Martin
            • Boundary layer?
              Bill McEachern
              In my experience floworks provides exceptional results for this class of problem and this can't be done if the B'layer modeling isn't pretty solid. Check the supplied technical reference for how the drag over a cylinder stacks up against R'number. It is down deep in the install directory under floworks/doc/lang/eng/....
                • Boundary layer?
                  OK, thanks, I'll investigate this.

                  I am trying to study claims made by heatsink manufacturers that boundary layer control can result in a heatsink efficiency boost of up to 20%. While I understand the issues surrounding heat transfer and boundary layer "insulation" a performance gain of 20% seems overstated to me.

                  I threw together a couple of simulations with features designed to turbulate or break-up the boundary layer. So far I haven't seen results that match the claims.

                  -Martin
              • Boundary layer?
                Basil Gello
                Martin and Bill,

                AFAIK the Boundary Layer model used in CFW and in many other even high-level codes without the turbulence accounting and the real viscous friction will give the error level at 20-30 per cent from the test data. Especially it will take its effect on heatsinks where there are lots of channels and parallel grids making the interferential flow. And once I got the situation that I had to manually determine ALL narrow channels with viscous effects in them - and only then I got the prety realistic visualization.

                Regards, Basil
                  • Boundary layer?
                    Bill McEachern
                    Hi Basil,

                    I am not sure I understand your above post. However, in my experience, getting solid, experimentally verifiable answers to heat transfer problems is pretty straight forward in Floworks - in the worst case (ie most onerous) you need to ensure that you have at least 4 cells across small channels. That said I have had pretty decent results with less, though I wouldn't rely on less and I would check the results of 4 with a run against 5 or more cells in a narrow channel.

                    To the best of my knowledge viscous effects are on all the time in Floworks. I don't even think there is a way to turn them off. With respect to turbulent effects in B'layers you do have the option to force either laminar or turbulent. Personally I just leave the default and Floworks figures it out for you (transition included).

                    Further what is AFAIK? To the best of my knowledge Floworks uses the very common K-e (2 equation) model for turbulence modeling and for wall effects it use the so called "modified wall functions approach". AS far as I am aware the modified wall functions approach is a pretty state-of-the-art way to go. The fundementals.pdf give the following information:

                    "The modified wall function uses a Van Driest's profile instead of a logarithmic profile. If the size of the mesh cell near the wall is more than the boundary layer thickness the integral boundary layer technology is used. The model provides accurate velocity and temperature boundary conditions for the above mentioned conservation equations."

                    Which is why, at least in my view, Floworks can give good results for relatively low cell counts across narrow channels at least with respect to temperature.
                  • Boundary layer?
                    Basil Gello
                    Hello Bill,

                    AFAIK means As Far As I Know...

                    And, as far as I understand from my experience and some theoretical background, the reason of a decrease of a heatsink's overall performance (especially with the increase of the airflow velocity) is that the real flow rate differs from the classic solution, where it is assumed that the flow in narrow channel is straight and solid (and pretty much incompressible), and this difference is obvious due to some hard-to-be-known turbulent effects causing at front and rear ends of a channel. Even if the assumption that INSIDE the channel the flow is strictly laminar and incompressible (and if the thermal load on every wall is different it is probably not the same on each surface of a channel), the turbulence on the ends may increase the drag and lead to lock of some channels, same as they would be filled by insects or sand. So, the drag force increases, some channels are lost and we lose the performance.

                    About the B'Layer control - I think that the main reason of its necessarity is what I have said above. In some cases turbulence can "charge up" the flow rate, but in many others it only reduces the effective velocity. And, as far as the B'Layer is not constant throught the channel, we have to specify the ranges of its possible change...

                    About the very accurate results according to temperature in very narrow channels - I can agree with it if the integral B'Layer is used (that means that there is no space for turbulence). But in case with heatsinks with pretty big gaps it is wrong, IMHO.

                    On the whole, the turbulence question in grid profiles (like heatsinks, grills or wing systems on old biplanes) is quite complex, and maybe the Vortex - Lattice method (VLM) or BLM(Boltzmann-Lattice method) will be able to solve it. And I'm very doubt that it can be done in the classic objective.

                    Regards,Basil
                    • Boundary layer?
                      Basil Gello
                      Chris,

                      thanks for correction!
                      Of course, the original phrase is "you have sown", but for a long time listening I have always been singing "you have soul" - quite a nice variant, isn't it? So it was a yet another kick for me to visit my favourite www.pink-floyd-lyrics.com.

                      Regards, Basil