CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   OpenFOAM (http://www.cfd-online.com/Forums/openfoam/)

 houkensjtu November 2, 2012 01:58

Problem when adding temperature to interFoam

Hi foamers!

I tried to apply temperature calculation to interFoam according to the hot thread:
http://www.cfd-online.com/Forums/ope...interfoam.html

And what I did was exactly same with the author "ovie" wrote in the #14 reply to the above thread.

Everything compiled well, however when I run it on a two-phase flow case, temperature results near interface was higher than heat source temperature which is surely not reliable result.
Also I test the modified solver to solve for single phase heat transfer calculation, although I didnot do precise validation, temperature field seems to be reasonable. Thus I think the main problem here lies in the interface.
On the other hand, from a theoretical view, no matter interface properties(cp, Pr...) is linear interpolated or harmonic interpolated, interface temperature should be no higher than phase temperature surrounding it.

I cant figure out what is the reason for this problem, plz help!

(Also I am trying to debug it by output cp and other properties on the fly, but I am not sure how to do it elegantly, directly to the console or paraView, plz help ...)

 houkensjtu November 5, 2012 00:10

I read the thread post by "ovie" again, and I do find that move the Cp to RHS (i.e., now the TEqn becomes:
fvScalarMatrix TEqn
(
fvm::ddt(rho, T)
+ fvm::div(rhoPhi, T)
- fvm::laplacian((kappa/Cp), T)
);
)help the convergence. But from a theoretical view, I believe it's mathematically equal.
So I got even more confused, I think there must be sth. I misunderstood/didn't understood in the whole process.
plz help!

 nimasam November 5, 2012 01:29

:), i guess it returns to the sensitivity of numerical procedure to calculation of convection terms (flux), those are mathematical the same on the whole domain except at interface :D but at interface they are a little different, and it can be accepted a new model :D

 houkensjtu November 5, 2012 02:06

Quote:
 Originally Posted by nimasam (Post 390233) :), i guess it returns to the sensitivity of numerical procedure to calculation of convection terms (flux), those are mathematical the same on the whole domain except at interface :D but at interface they are a little different, and it can be accepted a new model :D
Could you be a little bit more explicit on the difference of the interface.(or where can I find related information)
Basically I have only read Patankar and Versteeg's book on cfd and both these books didn't focus much on detailed numerical scheme.
(Another newbie question: I want to output Cp in order to see the space distribution of it(just like T and alpha), how can I do it? plz help!)

 nimasam November 5, 2012 03:35

if you want to see Cp
just define a volScalarField like T :) in "creatFields.H"

 All times are GMT -4. The time now is 04:41.