[jiguozhao]

Dear All:
I would like to simulate gas mass transfer across membrane in gas separation module. There is a membrane separating the module into shell domain and tube domain.
As you know the mass transfer is driven by the pressure difference of feed side and perm side. I introduced the sink and source at both sides of membrane by means of CEL. But the problem is if you generate a mass source in one side, you can just use the parameters within this elements (like pressure) to express the source, but the source is also ralated to the parameters (pressure) at the other side element (the nearest element at the other domain). Can any one tell me how to introduce a source expressed by a equation involving parameters from two different elements (as the attached picture shows). Now I can just treat perm side (in tube domain) pressure as 0 when introducing sink at feed side, and treat feed side pressure as a constance when calculating the source at perm side, but this is not quite accurate.
Is there any one who can help me?
Best regards

Ji Guozhao
g.ji@uq.edu.au

[ghorrocks]

Do you have to use a source/sink? Do any of the available mass transfer across a thin film models do what you want?

[jiguozhao]

Originally Posted by ghorrocks

Do you have to use a source/sink? Do any of the available mass transfer across a thin film models do what you want?

Thank you for your reply!
to the best of my knowledge of CFX, only Darcy's Law is built in thin film model(porous media). unfortunately, it is not able to satisfy my requirement. so I have to introduce a sink at the feed side of the membrane and a source come out from the permeate side of the membrane. the quantity of source or sink is a function of the both pressures at two sides. if i introduce the sink at feed side, I can only use the pressure at feed side. but I can not use the pressure at perm side into the sink CEL.
Do you know how to deal with my problem?

[joey2007]

Guess you have two domains and and interface between. One solution would be to develop an interpolation in user fortran. Some month ago I talked to some people who did that for an implementation of an temperature periodicity. They required intense support by ANSYS and I guess also much time. I discussed it already below:

http://www.cfd-online.com/Forums/cfx...tml#post283314