CFD Online Discussion Forums

CFD Online Discussion Forums (
-   CFX (
-   -   air varible property (

Studentdrak February 24, 2012 08:50

air varible property

I am trying to simulate forced convection (air) where the film temperature is high enough to cause property variation in air. how can i make the thermal conductivity and viscosity of air, change wrt temperature? if CEL is the only way, may i have the general expression for K and wrt to temperature?

evcelica February 25, 2012 01:15

an example of temperature dependent thermal conductivity would be:


ghorrocks February 25, 2012 06:52

You can define properties as either CEL expressions or tables of values. Properties of air at various conditions are pretty easy to get with google.

evcelica February 25, 2012 10:46

You can also try getting NIST RefProp for fluid properties. Its free to students (only has water, air, and a few others). For only a small price you can buy their full package which has just about every fluid and property at any condition you can imagine.

Studentdrak February 27, 2012 04:19

hi, i get the following error when i try to run my code with CEL
Fatal bounds error detected
Variable: Thermal Conductivity
Locale : air

| ERROR #001100279 has occurred in subroutine ErrAction. |
| Message: |
| Stopped in routine ENFORCE_BOUNDS |
| |
| |
| |
| |
| |

| An error has occurred in cfx5solve: |
| |
| The ANSYS CFX solver exited with return code 1. |
| |
| It gave the following output: |
| |
| Contents of /tmp/pvml.504: |
| |
| [t80040000] 02/27 13:41:03 agni3 ( LINUXX86_64 3- |
| .4.4 |
| [t80040000] 02/27 13:41:03 ready Mon Feb 27 13:41:03 2012 |
| No results file has been created. |

End of solution stage.

my expression for Kth is :
-------------------------------------------------------------------------------------(( - 2*10^(-14))[W/m/K^6)]*T^(5) ) + (( 1*10^(-11))[W*m^-1*K^-5]*T^(4) ) - ( (6*10^(-09))[W*m^-1*K^-4]*T^(3) ) + ( (1*10^(-06))[W*m^-1*K^-3]*T^(2) ) - (( 6*10^(-05))[W*m^-1*K^-2]*T ) + 0.006[W*m^-1*K^-1]


If i need to define it as a reference table, how do i do it?

pl. do reply asap


ghorrocks February 27, 2012 06:05

Variable properties are always harder to converge. But the problem with using fitted equations like you have (which includes temperature up to the fifth power) is that outside of the fitted range the equations frequently go bezerk and lead to rapid divergence.

A table can help here, at least they are better behaved out of the defined region. Use a 1D interpolation function and enter the points in. I usually do it by entering in a few points then editting the CCL of the interpolation function and add all the other points in a text editor. It is much easier to add lots of points in the text editor than the GUI.

Zigainer February 27, 2012 06:33

You can also use the sutherland-law for thermal conductivity and viscosity. This work quite fine for me in the past.

Studentdrak February 27, 2012 07:11

Hello all,

thanks for the reply.

Zigainer, The case is running with sutherlands formula.

ghorrocks, cud u pl. explain me how to give an input as a table? May be i am not able to follow the steps explained.


Zigainer February 27, 2012 07:41


Originally Posted by Studentdrak (Post 346436)
Zigainer, The case is running with sutherlands formula.

Did you use the pre-implemented sutherland formual or did you just created a new expression? Don't know if it makes any differences ....

ghorrocks February 27, 2012 17:50

If you want to use the interpolation function approach have a look it up in the documentation. It is quite easy to use.

evcelica February 27, 2012 23:34

Thermal conductivity is usually fit well by a second order curve fit. What crazy temperature range are you using to need a 5th order??

More importantly, and the reason its failing is:
If you want to use such high order polynomials you have to include much more than one significant figure for your values. They become very important in the higher orders. Increase the sig figs in your curve fit values to as many as you can.

Studentdrak February 28, 2012 08:38

Thank u all

Zigainer, i had used inbuilt sutherlands formula

All times are GMT -4. The time now is 00:43.