CFD Online Discussion Forums

CFD Online Discussion Forums (
-   CFX (
-   -   mass diffusion (

CN January 22, 2006 10:38

mass diffusion
Hello all,

my question: were in CFX do I declare mass diffusion coefficients ? I searched the manuals, found Kinematic Diffusivity, but I am not able to find the place, in graphical user interface menus, were to specify value for a given media combination ?

Thanks for directions, CN

Jeff January 27, 2006 18:18

Re: mass diffusion
The kinematic diffusivity IS the binary diffusion coefficient for one species (AV or mass fraction) through another (the background fluid).

If you have more than one component, these are very difficult to determine. However, if the flow is turbulent, it won't matter anyway because the turbulent diffusivity will overwhelm the kinematic diffusivity. Just set them all to something like 1.0E-6.


CN February 3, 2006 12:31

Re: mass diffusion
Hello Jeff,

first thanks!

You write:

'..If you have more than one component, these are very difficult to determine..'

is it difficult, but determination can be explained? If can, please feel free to indicate were detail can be found:

in CFX help docu (i am not aware of that) or auxilliary CFX docs (not aware of these as such yet), on a dedicated webspace/user forum run by CFX themselfs?

Did you exercise determination of b.coeffs with success. Would you mind outline how it's being down?

amongst the issues: is CFX taking care of the set of b.coeffs ok internally [each 2x distinct species a b.coeff] [by using fickian laws & add. relevant relations]? As the b.coeffs need be considered in context of interaction of participating media, intending multi component setup, not additional variables or evtl other ways to model tracing of spurious contend, like path marker media.

Greetings, CN

opaque February 3, 2006 16:01

Re: mass diffusion
Dear CN,

I imagine you are using multicomponent material, and you are interested in the molecular diffusion between species. ANSYS CFX uses the dilute approximation, i.e. there is a major presence of one of the species and the diffusion can be modeled as the diffusion of a given species into the "fluid" as whole.

This approximation is good for turbulent flows since the "turbulent mass flux term" dominates and the molecular contribution are either neglected or approximated as above..

For laminar flows, the above is not true, and you may need the calculation of the D_ij coefficients and add the proper terms into the species transport equations.. This is not supported in ANSYS CFX out of the box... Perhaps, someone has already tried via UserFortran.. Have you contacted you help desk?

Good luck, Opaque..

CN February 6, 2006 12:01

Re: mass diffusion

the scope is, to determine mass fraction of each gas in the mixture at positions along flow chanel.

Basically a bulk fluid (background fluid as Jeff put it) receives (by way of > 1 side inlet pipes, 1x per each polutant gas) additional minor gas instreams, which are mixed into the main stream (w/o reaction modeling) with chanel containing a grillage to enhance mixing, further upstream.

As I understand, m.diffusion coeffs are less relevant here if, as you and Jeff noted (and Help Desk as well) turbolent diffusity dominates.

But if CFX models dilute only (not mentioned by Help Desk, perhaps I didn't 'capture' that) - do I get mass(or %volumetric)fractions= f(position) as a output quantity? this is required.

Thanks, CN

All times are GMT -4. The time now is 17:48.