I wanted to let air's properties be functions of temperature for convective heat transfer models over the range 290 to 670 K (some properties change by almost 100% in this range). So I let FLUENT use what I understand is its general purpose model for these properties as follows (in a .jou file):

/define/materials/change-create air air y ideal-gas y kinetic-theory y kinetic-theory y kinetic-theory n n n n n n

I worked one impinging jet model this way for several months trying to get agreement with the literature, with frequent conversations with two FLUENT engineers, and the air properties model never came up as a possible issue. But upon finally checking I found the model's air properties were sometimes way off. Thermal conductivity was only half what it should be! So I created my own polynomial and got proper values, and also found that a simple free convection model suddenly agrees with literature correlations, and the impinging jet model is at least much closer. I asked tech support about this and got the following reply:

"Using Kinectic therory to predict the properties is not the best way to provide inputs for the variable.

"The polynomial that you are using now is the most accurate way of providing the properties as a function of Temperature."

My question: is FLUENT's basic model for air properties so wrong? Why??? Or can my application of it be misguided somehow, even though both FLUENT engineers accepted it? Nowhere in the docs is the suggestion that users should develop their own polynomial models - what do users who can't regress polynomials do?

This is a startling lapse, somewhere or other. How can people model complicated turbines and engine interiors and so forth without this coming up?

If you run the journal this way you will notice that the number of degrees of freedom (dof) is set to zero by default. Choosing the right dof is not easy. While we can get away with 3dof for 1atomic gases things are lot more difficult for a mixture of 2atomic gases. Energy can be stored in translatation (3dof), rotation(2 or 3dof) and oscialltion (2dof). Adding all up will get a maximum of 8dof but not all dof are used depending on the temperature. Calculating specific heat with kninetic theroy is very tricky so i would at least use a polynom for cp to get around the dof problem.

RoM