CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   FLUENT (https://www.cfd-online.com/Forums/fluent/)
-   -   Estimation of a diffusion coefficient at a face (https://www.cfd-online.com/Forums/fluent/68557-estimation-diffusion-coefficient-face.html)

pifou74 September 23, 2009 16:17

Estimation of a diffusion coefficient at a face
 
Hello,

I would like to calculate precisely a term such as div(Gf.grad(phi)) in order to create a source term for an equation.
To achieve this, I need to know how fluent evaluate the value of the coefficient Gf at faces, knowing Gc at cell centers around the face.
I have tried to estimate Gf from surrounding cells c0 and c1 like:
Gf=(2Gc0*Gc1/(Gc0+Gc1)) but it doesn't work very well. I have also tried simple linear interpolation with bad results.

If someone has an idea or an information it would help me a lot.

Thanks

arhn September 26, 2009 06:51

hi
according to Patankar book (p. 45) you should use harmonic mean with respect to distance between center points and interface.
for interface that palace in meddle your formula is correct but in general:
Gf = G0*G1/( G1(1-fe)+G0(fe) )
fe = (distance between G0 and interface)/( distance between G0 and G1 )

pifou74 September 26, 2009 07:47

Thank you Arhn for your answer. Do you think that, in fluent, it is made like you explain?
Of course I know the very good Patankar's book. You're suggestion is good, nevertheless hard to apply in unstructured mesh, due to the fact that it is quite hard to estimate precisely fe in this case especially when grid is not orthogonal .
Have you any experience about this?

arhn September 26, 2009 08:19

I write some MATLAB codes that use unstructured meshes like Delaunay or Voronoi and use that formulation. I since don't use FLUENT, sorry.

pifou74 September 26, 2009 08:39

Ok thank you arhn.

Have a nice day

pifou74 September 26, 2009 16:56

Re Hello Arhn,

I have think about your message. In a Delaunay grid, I think that you can have some triangles which are not orthogonal?
It would help my understanding if you can tell me, in this case how you estimate fe?
Indeed, in this case (non orthogonal grid) the face center, and the cells centers are not on a same line. Moreover I don't know for you but in my case I can only access to the coordinates of the centers of the face and cells. Do you make a kind of approximation?

Thank you for your help

arhn September 26, 2009 17:40

Hi
In my code I had coordinate of cell centroids and vertexes of each triangle.
With command "polyxpoly" by using of coordinate of 4 point (2 cell centroids and 2 vertexes) I find the cross point of line between cell centroids and interface line then with respect to the point's coordinates I calculate Fe.

pifou74 September 27, 2009 04:08

Thank you for all, I will try

arhn September 27, 2009 04:51

You are welcome.


All times are GMT -4. The time now is 14:50.