[Vikas gaur]

I have created a UDF of specific heat with temperature dependent exponential function as follows:

#include "udf.h"
DEFINE_SPECIFIC_HEAT(my_user_cp, T, Tref, h, yi)
{
real cp = 2180;
real exponent=(-2*(T -316.65));
real denominator=(1+exp(exponent));
real numerator=(2390-2180);
real second_part=numerator/denominator;
cp=cp+second_part;
*h = cp*(T-Tref);
return cp;
}

above code has worked well but being new user to fluent I want to ensure whether this code is correct and I also want to create a similar function for thermal conductivity using property macro.

[obscureed]

Hi Vikas gaur,

The value of h is not correct, and this could lead to (probably small) heat-balance issues. The Customization manual states: "the enthalpy function should be the temperature integral of the specific heat function". The example given in the manual is for the very easy case of integrating constant Cp. WolframAlpha is your friend: see for example http://www.wolframalpha.com/input/?i...(-k+x+%2B+d)))

Obviously we cannot tell you whether there are typing errors in your specific values. You seem to have a transition from constant value 2180 to constant value 2390 (in J/kg/K), with the transition occurring smoothly and sigmoidally between about 312K and 322K. You have got this far with the UDF (well done!), so I can imagine you will be tempted to keep going, but a piecewise-linear description would be an option worth considering. A handful of points (10 or fewer) would be enough for most purposes here.

The DEFINE_PROPERTY UDF for thermal conductivity is easier, but again piecewise-linear is easy for users to verify and maintain.

Good luck!
Ed

[Vikas gaur]

Thanks Obscureed,

I really appreciate you for giving time to my problem.
Earlier I was using piece-wise function and obtained correct results to my problem. But when I change boundary conditions (for same material properties) I am getting somewhat deviated results. that's why I tried UDF but I don't find significant changes in result.

Is it possible that particular problem can only be solved for certain set of boundary conditions?

[obscureed]

Hi there.

It would take some ingenuity to think of cases that change from "easy answer" to "no possible answer", just by changing boundary conditions. But it is relatively common to find cases that change from "easy answer" to "difficult answer" to "no answer found in the time available". At the very least, you may need to experiment with solver settings such as under-relaxation factors.

Best regards,
Ed