CFD Online Discussion Forums

CFD Online Discussion Forums (
-   CFX (
-   -   Heat ransfer coef in natural convection (

Atit Koonsrisuk March 27, 2003 06:01

Heat ransfer coef in natural convection
I am using CFX5.5.1 to simulate the laminar flow over the constant temperature vertical flat plate. Physically the boudary layer must occur along the plate. Then I apply the inflation mesh near the plate surface. The temperature and velocity profile seem reasonable. However the value of heat transfer coefficient is significantly larger than the theoretical value. It say, in CFX manual, that the heat transfer coefficient is computed as

h = q / (T_s - T_p)

where T_p is the temperature at the internal near-wall boundary element center node.

I would like to capture the boundary layer, then the first inflation mesh layer is very thin. Consequently the temperature in the first layer, and few succeeding layers also, are closely the same as the temperature at the wall. So no surprise why the value of heat transfer coef is very large.

When I remove the inflation mesh, the value of heat transfer coef is closed to the theoretical value, but the velocity and tempeture profiles are wrong.

What should I do to get the right flow profile; and the right heat transfer in the same time?

Thank you very much.


Eve April 4, 2003 08:15

Re: Heat ransfer coef in natural convection
To me the h-value is defined as q=h(Ts-Tamb). You get then the h-value by dividing the wall heat flux by the temperature differense between the solid and the ambient temperature.

Hope this answers your question! /Eve

BAK_FLOW April 4, 2003 16:59

Re: Heat ransfer coef in natural convection

yes this is an interesting issue.

For all flows with constant properties the heat transfer coef in non-dimensional form is the Nusselt number and is a function only of the Reynolds Number and Prandtl number. This is useful and has resulted in correlations of experimental data. In CFD we directly determine the temperatures and heat fluxes as they are what what we usually want to know. To then convert this back to a heat transfer coeficient is sometimes going the wrong way!

We can do it, however but have to be careful that the temperature scales we use are meaningful. For simple flows the logical reference temperature to use is some bulk temperature or temperature at infinity. This can be forced in CFX 5 by setting the expert parameter:

tbulk for htc = ??.??[K]

for example.

This will over-ride the use of the nearest node temperature.

Hope this helps,


Henriqueg June 17, 2009 16:07

Atit, im working in a similar problem and im also in trouble. But ive noticed the velocity profile changes significantly with diferent time scales (auto time scale, physical time scale or local time scale). May i ask you wich time scale are you using? And, for curiosity, what boundary conditions are you using? Are you working with the openning condition?


augusto_bassan August 19, 2009 22:09

Convection Heat Transfer
hi someone have or know where I can found the solution manual of book Convection Heat Transfer, third edition of author Adrian Bejan, because I need very much in my pos- graduate , thanks.
my e mail is

ghorrocks August 20, 2009 07:19

Can't you visit your uni library? If they don't have it they should be able to arrange an inter-library loan from one which does.

All times are GMT -4. The time now is 19:41.