[bharat.cmeri]

Hi all,

Can someone please suggest me how to use harmonic mean for diffusion coefficient for diffusion term of a UDS equation ? Presently it uses only central differencing scheme for its discretization which uses linear average for calculating face values of diffusion coefficient.

regards,

Bharat

[sbaffini]

Are you sure about that? I'm pretty sure Fluent uses an harmonic mean for the diffusion coefficients of the regular equations and would be strange that it does differently for the UDS. Do you have a reference on that?

[bharat.cmeri]

Originally Posted by sbaffini

Are you sure about that? I'm pretty sure Fluent uses an harmonic mean for the diffusion coefficients of the regular equations and would be strange that it does differently for the UDS. Do you have a reference on that?

fluent theory guide: look in paragraph on spatial discretization where it says that "the diffusion terms in Equation18.2-2
are central-differenced and are always second-
18.2.2(discretized form of General UDS equation) are central-differenced and are always second order accurate ". ( 12.0 version theory guide)

I am not sure if they uses harmonic mean or not but no where it is written that they do so.... Anyhow I have written to customer portal to know how to implement/use harmonic mean for diffusion coefficients. If you know how to do so please let me know, else I will tell you what they said. Hope you are right....


Bharat

[sbaffini]

They clearly do not refer to the diffusion coefficient when talking about the discretization. It just appears as a numerical value (gamma_f) whose interpolation is not defined (there, at least).

I remember reading a paper from Fluent people talking about the multi-grid method for multi-material stuff, and they mention the harmonic mean. However i cannot find it now. Nonetheless, i would be surprised of anybody using the classical average.

Still, i don't think you have any choice in the selection of the diffusion part.

P.S. Good luck with the Fluent support (you better write yourself the code from scratch)

[bharat.cmeri]

Originally Posted by sbaffini

They clearly do not refer to the diffusion coefficient when talking about the discretization. It just appears as a numerical value (gamma_f) whose interpolation is not defined (there, at least).

I remember reading a paper from Fluent people talking about the multi-grid method for multi-material stuff, and they mention the harmonic mean. However i cannot find it now. Nonetheless, i would be surprised of anybody using the classical average.

Still, i don't think you have any choice in the selection of the diffusion part.

P.S. Good luck with the Fluent support (you better write yourself the code from scratch)

I talked to fluent guy....I dont think they will be able to help me out....If you get to remember anything about the paper you mentioned....like its author etc....then pls share with me.
Anyhow I now know I have to work on OpenFoam only.

if you have any other suggestion please give it to me. Would be helpful to me.
Thanks & Regards,

Bharat

[sbaffini]

Dear Bharat,

i have the really bad habit of not renaming any of the paper i download (over 100 Gb right now). Thus, while i'm usually good at this (so that i clearly remember the article), it is not really straightforward to find such paper among the Others, 5+ years later.

What i can be 99.99% sure is that if the paper (as i remember) is from Fluent people, it also surely has at least one of the following authors (in order of probability):

Mathur, Murthy, Choudhury, Makarov, Caraeni, S.-E. Kim

Moreover, i also remember that the topic was about the use/development of the multi-grid for multi-material interfaces. This is all i can remember. There are chances (but i'm really guessing here) that it was an AIAA paper or similar AIAA publication. Second guess would be Numerical Heat Transfer, Part B. Just for the record.

Despite being really astonishing that nobody in the Fluent support was able to tell you "yes (no), we (don't) use the harmonic mean", that's exactly equivalent to my last experience with them (back in 2005).

Good luck with openfoam too

[bharat.cmeri]

Originally Posted by sbaffini

Dear Bharat,

i have the really bad habit of not renaming any of the paper i download (over 100 Gb right now). Thus, while i'm usually good at this (so that i clearly remember the article), it is not really straightforward to find such paper among the Others, 5+ years later.

What i can be 99.99% sure is that if the paper (as i remember) is from Fluent people, it also surely has at least one of the following authors (in order of probability):

Mathur, Murthy, Choudhury, Makarov, Caraeni, S.-E. Kim

Moreover, i also remember that the topic was about the use/development of the multi-grid for multi-material interfaces. This is all i can remember. There are chances (but i'm really guessing here) that it was an AIAA paper or similar AIAA publication. Second guess would be Numerical Heat Transfer, Part B. Just for the record.

Despite being really astonishing that nobody in the Fluent support was able to tell you "yes (no), we (don't) use the harmonic mean", that's exactly equivalent to my last experience with them (back in 2005).

Good luck with openfoam too

Hi,

Thanks for the information.

Regards,

Bharat