CFD Online Discussion Forums

CFD Online Discussion Forums (
-   ANSYS Meshing & Geometry (
-   -   [ANSYS Meshing] Modelling Heat transfer (

khuram87 August 23, 2012 09:00

Modelling Heat transfer
Hi all...

I am trying to simulate a flow in a ribbed channel

periodic boundary conditions are to be used

i have tried various meshing techniques including the size functions etc but still I am getting very high heat transfer values compared to experiments.

i am using k-eps standard EWT, with energy equation (k-w model apparently doesnt give desired velocity profiles)

if I make the cells too close to the wall, I get an error in fluent that "Aspect ratio is very high for some cells and the default code might not work".

The link to the problem description is (

Please reply ASAP

yonchong August 23, 2012 09:25

Plot Graphic and Animations -> Contours -> Turbulence... -> Wall Yplus on surfaces you are trying to get heat transfer. If they are generally close to 1 or less you should be using Enhanced Wall Treatment with k-epsilion otherwise your heat transfer is going to be high.

Alternatively if you have later version of Fluent try Scalable Wall Functions as the EWT will give high heat transfer at high y+ (e.g. over 15) and Scalable Wall Function seems to give better results through wider y+ range.

Also use Relizable k-epsilion rather the standard model.

khuram87 August 23, 2012 11:13

my y-plus values are around 0.5 and I have tried every model..!! :(

but the values are not right...may be there is a problem of meshing..!!

the mesh picture in the above link (Green one) seems to work but i dont know how to make such mesh..

yonchong August 23, 2012 11:24

The link is broken. You might want to upload the pictures to this forum directly.

By the way, y+ 0.5 should be ok to use with Enhanced Wall Treatment.

khuram87 August 23, 2012 12:02

this link works

yonchong August 23, 2012 14:11

Is your model 2-D?

The green mesh look like Quad Dominant (or Hexa-Dominant) mesh depending on whether this is 2 or 3-D.

So where is the solid mesh if you are doing conjugate heat transfer?

Why are you saying your result is wrong?

Are you trying to calculate metal temperature or Heat transfer Coefficient?

khuram87 August 23, 2012 16:46

my simulation is 2D...and i am giving a constant heat flux (1000 W/m^2) from both the top and bottom walls..

the main aim is to check how does the rib affect heat transfer due to production of vortices etc..this is quantified in terms of Nusselt number at the boundary of the wall where convection is taking place due to the flow...

the normalized nusselt number distribution between two successive ribs (Turbulence generators) is very high than the experimental values

Thanks alot for your cooperation

khuram87 August 23, 2012 16:50

these are the values for the normalized nusselt numbers which I should get..

but my maxima is 20..!!:confused:

yonchong August 23, 2012 17:10

So how do you know the green mesh work?

Can you put your case file up?

evcelica August 23, 2012 23:17

How are you calculating nusselt? I've never used Fluent but in CFX the adjacent wall temperature is used to calculate HTC, not bulk temperature,so it gives MUCH higher values than published data which uses bulk temperature.

A7A October 18, 2012 01:53


Have you checked the reference values you are using like length which is must be the hydraulic diameter of the channel, and the reference temperature.

All times are GMT -4. The time now is 20:14.