7 Replies Latest reply on Nov 21, 2014 12:17 PM by Jared Conway



      My questions are from 2 streams.  In the first stage, I am running a study with all the modes of heat transfer turned on (radiative surface BC + natural convection, including gravity, etc.) and let's say there is one heat source (net Watts, uniform distribution) applied to an 'internal' component's surface, where there is only conduction.  I then apply a an equation goal, EG, to achieve the heat rate as surface goal SG heat flux * SG area.  Furthermore, the EG is exported onto the goal parameter (i.e., plot) list.   I expect this EG to match the input heat source at conclusion.  Upon running the solver, the initial value of the EG is exactly matching the input value, but at conclusion there is an error of 5-10%:


      Q1. Is this normal performance for integral parameter estimates that coincide, exactly, with the input BC?


      Now in the post-processing or Results interpretation mode, what is meant by the following available surface parameters (I include my interpretation for a point of agreement/disagreement)?


      Q2.  surface parameter Heat transfer rate.  If I select a complete component or body, then should this = 0 net for steady state analysis?  I have a discrepancy error (referred to the input BC value) of 10-20%


      Q3. Net versus leaving radiation rate, there appears to be an error in net.  In the model setup, this is supposed to be, effectively, an n=2 gray body system, where the extent of the computational domain is the large area body and the solid(s) of interest are the small area body.  I expect, therefore, net radiation rate = leaving rate?   I am seeing a large error then


      Q4.  I see that there is a heat transfer coefficient but no actual convection heat rate or flux.  The problem with heat transfer or thin-film convection (apparently?) coefficient is I cannot, properly, account for, its distribution compared to the temperature (temperature potential with respect to the fluid at infinity) distribution.  In other words, I can only can get a rough estimate if I use the product of the average temperature potential and the average heat transfer coefficient!  Is there an actual convection heat rate or flux surface parameter?  I realize, with enough patience, if I select key faces only and not the whole component, which is what I rather do and have used as the implied work flow, then I may use the surface heat flux - net radiation flux to get a the convective heat flux.


      Thanks for your consideration of the above questions.

          Jared Conway

          start with a much simpler example, draw your control volume diagram and do some tests. The software should give you conservation of heat. The only time there is a discrepancy is if the mesh is too coarse. ive done this a dozen times for customers on support and it has never failed. Once your have the technique, apply it to your model.


          some key principals

          -use heat rate for convection

          -choose all surfaces involved in convection and radiation

          -radiation requires some algebra to get the component of interest sometime

          -dont get more complicated than a cube or cylinder at first. Cylinder or sphere is usually easy to see the effect of mesh.

              MAXIM FRAYER



              Thanks for the reply and interest in my questions.  This is my lingering confusion with surface parameters in the Results:


              Q4. Are you stating that the "heat transfer rate" is just convection, and, therefore, If I select the whole component (all faces are selected by default), that the area will be divided into "wetted only" for convection.  I don't see that happening (I include an actual analysis ex. below):


              Global Coordinate System
              Iteration [ ]49
              Local parametersIntegral parameters
              ParameterMinimumMaximumAverageBulk AverageSurface Area [m^2]ParameterValueX-componentY-componentZ-componentSurface Area [m^2]
              Heat Transfer Rate [W]-0.0918623 0.00120803
              Net Radiation Rate [W]0.09094497 0.000657897
              Leaving Radiation Rate [W]0.32522234 0.000657897

              It appears that the heat transfer rate is all forms of heat transfer as a flux (i.e, (q * en) dA.).  In that case what is the exact parameter to isolate convection only?



              For Q1 and Q2. You mention a coarse mesh driving an error in the intergral estimate (temperature potential error appears to be very small, however, when compared with measurement).  What is the amplitude of the error, in your experience?  It appears like 10-20% is, being achieved, in this example: Is that reasonable for a coarse mesh?


              Q3. As per the included example, where radiation appears to automatically select "wetted only" surface patches (or more exactly, those participating in radiation exhange?), what would cause net radiation rate to not equal leaving radiation rate for a n=2 type gray body system?


              I appreciate your patience and attention to these questions, as I would like to get to a definitive answer (from an expert). 

                  Jared Conway

                  q4 > heat transfer rate is convection only, yes. post your simplified model if you aren't seeing that.

                  Q1/Q2 > cannot be predicted without doing a mesh convergence test. this is well discussed on this forum. could be as high as 100% if the mesh is very coarse

                  Q3 > no example was included. i'm not sure what the question is here. post your simplified example.

                      MAXIM FRAYER

                      Q4: are you sure that heat transfer rate is convection only?  The reason I ask, is that in the exported parameter table in the above post, I do not see the surface area reflect "wetted only" size, as it would need to for convection.  Only for the radiation parameters does it reflect "wetted only" (what I have here is essentially a window as the dominant surface area, and the outside plane is wetted, while the inside plane is not, hence approx. double the area as per the table).


                      Q1-2: Thanks for the data point: It is too bad the error is so large for a coarse mesh, as this mesh suffices to get a quick and accurate estimate of the temperature potential.  In other words, it is already painful enough with my hardware setup(s) to do a coarse mesh and then the post-processing steps, etc.


                      Q3. My only point that I am stressing is why does net not approx. equal leaving radiation rate, per the ex. in the table?  For the type of n=2 body radiation system, this is what I expect.  Perhaps the coarse mesh error again?