CFD Online Discussion Forums

CFD Online Discussion Forums (
-   CD-adapco (
-   -   Heat Transfer Coefficients (

Ashish Kumar Gupta November 16, 2001 00:40

Heat Transfer Coefficients
Hello This question is related to computation of heat transfer coefficient in StarCD. What formula does StarCD uses to calculate heat transfer coefficients? For a specific problem with adiabatic wall boundary condition (RANS equations are solved with k. epsilon high Reynolds no. turbulence model with wall function approach), how does StarCD calculates heat transfer coefficients? How to interpret the HTC value that is plotted?

Thanks for a quick reply.


Helge November 16, 2001 06:18

Re: Heat Transfer Coefficients
Refer to page 6-5 in the methodology guide of STAR-CD. Another hint you will find in paper at

Ashish Kumar Gupta November 17, 2001 08:08

Re: Heat Transfer Coefficients
Hi We can get the expression for HTC from Methodology manual as stated by you. But when one looks at 'Load wall post data' (this appears in Prostar main window when we click, Get wall data), one has two options 1) Heat Transfer 2) Hcoefficient. Which of these two uses the expression given on page 6-5, Star methodology guide? It seem from this expression, it is not important whether one give fixed wall temperature boundary condition or adiabatic wall boundary (q across wall is zero in this case and hence HTC should be zero)? Any comment on this please?



star-user November 17, 2001 11:09

Re: Heat Transfer Coefficients
The definition given in the manual is for htran which as you mentioned does not require a wall temperature or heat flux for definition.

The definition of hcoef is more traditional (i.e. Newton's law of cooling). It requires a heat flux across the wall boundary. Hcoef would be the heat transfer coefficient defined in the Nusselt number. It generally has more meaning then htran (at least for me).

Star-CD calculates htran in the solver. Depending on the value of y+ their are two definitions. Be careful when using the two-layer or low reynolds number turbulence models. You need to have an appropriate thermal variation of the conductivity. Hcoef is a post-processing number and requires a bulk temperature in celsius. Usually this is the inlet temperature. Thus hcoef may be less than zero while htran is always greater than zero.

There is a great article in adapco=online which details this important difference.

Ashish Kumar Gupta November 22, 2001 00:10

Re: Heat Transfer Coefficients
Hi, Thanks for the mail and clarification.


Ashish Kumar Gupta November 22, 2001 02:17

Re: Heat Transfer Coefficients
Hi Thanks for the reply. It will be great if some one can tell me how HCoeff is calculated witin StarCD as well as what is Tbulk parameter? As hcoeff is defined as qw/(Tw-Tbulk), whether qw is obtained from k*dT/dy in StarCD or some correlation is used? Again it has been defined in StarCD Methodology guide on page 12.6, but is this true for all types of flows (eg. single phase, turbulent flows etc.).



Helge November 22, 2001 03:50

Re: Heat Transfer Coefficients
1) Again, everything is explained in the methodology guide page 6-5 for turbulent flow using a wall profile.

2) But to explain it in words. htran is a postprocessing value which can be calculated wether the flow is isothermal or not. The first time I really noticed that I was a bit surprised to be honest. But this is simply the consequence of the log law for the velocity and temperature profile used in the standard k-e turbulence model. The consequence on that is of course if the log law is not the right one for your type of flow, htran will also be wrong.

3) The value is calculated using u+ assuming that T+ is a function of u+.

4) If you like to have the traditional heat transfer coefficient used in the Nusselt number for an non isothermal flow (htran), Star works as follows. From u+ T+ (here that means the temperature in the cell which attaches the wall) is derived. Then Star calculates the heat flux using the temperature difference between the wall and the cell adjacent to the wall. Then the user has to define Tbulk for Star to calculate hcoeff.

5) Now it comes to the question what Tbulk is. You are more or less free to give a Tbulk value. To be able to compare the calculated hcoeff with measurements you have to use the same definition for Tbulk as the person used who made the measurements. For a local hcoeff value Tbulk normally is the avarage Temperature in the section perpendicular to the wall (for a pipe flow).

Tbulk=(Integral(rho u c_v T dA))/(massflow c_v)

For an average hcoeff Tbulk will be T_outlet-T_inlet and hcoeff will be the average of your "local" hcoeffs.

6) For more information about the mean Temperature you should look into a book dealing with heat transfer (e.g. Fundamentals of heat and mass transfer, Frank P. Incropera, David P. de Witt, Wiley)

7) In general it should be easier to compare heat fluxes and not hcoeffs

All times are GMT -4. The time now is 18:56.