5 replies to this topic

[eng4ever]

hello
could anyone help me please, i want to make a research about UNIFAC interaction parameter, how they calculate it?? ,or how they reached to that parameter???
I'm already knows how to use UNIFAC model and how to calculate the activity coefficient, and i read several books in thermodynamic talking a little bit about UNIFAC but i couldn't find any thing about interaction parameter,
i searched for days in internet for "Vapor-liquid equilibria using UNIFAC : a group contribution method" by Aage Fredenslund cause he was the 1st one who developed that theory but i couldn't find it for free .
my request is, could someone had that book upload it or anyone could help me with a suggestion to solve my problem
i'm hopeless any information could help........
thanks

[Chellani]

Hi,

You seem a bit disappointed so first cheer-up as you’ve already done half of the job by knowing how to use UNIFAC.
As you already know how to use UNIFAC, I hope you are able to build a basic function in excel / Matlab / scilab or any other program which would estimate T & y for P and x as input or estimate P & y for T and x as input. All you need to do is use this function and treat UNIFAC Binary Interaction Parameters (BIPs) as adjustable variable to meet specific objective function. For sake of simplification, you can start targeting lngamma i.e. to minimize (lngamma(exp) – lngamma(correlated))^2.
I’ve attached a simple excel sheet which uses NRTL and excel’s solver tool has been used to get BIPs. Look at ‘1-propanol-water’ worksheet in which objective function (cell N28) is being minimized using solver by adjusting K2 and K3. K4 is not being adjusted as for NRTL, mostly its value is fixed i.e. 0.3 / 0.45.
You need to build similar function for UNIFAC and your BIPs would be specific to groups rather than components.

You can think of going one step ahead i.e. rather than minimizing error in lngamma, minimize error in experimental and correlated T & y or P & y based on your data. You’ll need to add bubble point / dew point algorithm for that.

Hope this would be helpful

Attached Files

•  Database.xls   97KB   223 downloads

[eng4ever]

Hi,

You seem a bit disappointed so first cheer-up as you've already done half of the job by knowing how to use UNIFAC.
As you already know how to use UNIFAC, I hope you are able to build a basic function in excel / Matlab / scilab or any other program which would estimate T & y for P and x as input or estimate P & y for T and x as input. All you need to do is use this function and treat UNIFAC Binary Interaction Parameters (BIPs) as adjustable variable to meet specific objective function. For sake of simplification, you can start targeting lngamma i.e. to minimize (lngamma(exp) – lngamma(correlated))^2.
I've attached a simple excel sheet which uses NRTL and excel's solver tool has been used to get BIPs. Look at '1-propanol-water' worksheet in which objective function (cell N28) is being minimized using solver by adjusting K2 and K3. K4 is not being adjusted as for NRTL, mostly its value is fixed i.e. 0.3 / 0.45.
You need to build similar function for UNIFAC and your BIPs would be specific to groups rather than components.

You can think of going one step ahead i.e. rather than minimizing error in lngamma, minimize error in experimental and correlated T & y or P & y based on your data. You'll need to add bubble point / dew point algorithm for that.

Hope this would be helpful

first thanks for reply me and thanks a lot for the information and sheet, actually i was disappointed cause searching without getting thing doesn't make anyone happy, but now i think i know from where to start , by using trial and error and made a back calculation for UNIFAC model i think i can calculate the Interaction Parameters theoretically but now how about experimentally how they calculate it????

Edited by eng4ever, 10 May 2011 - 06:33 AM.

[breizh]

Hi ,
let you consider this resource:

http://www.fsalazar..../xls/UNIFAC.xls

Hope this helps
Breizh

[eng4ever]

Hi ,
let you consider this resource:

http://www.fsalazar..../xls/UNIFAC.xls

Hope this helps
Breizh

thanks i saw that sheet when i studied the UNIFAC model
but now the quotation is how can i find the Interaction Parameters experimentally???

[MrShorty]

how can i find the Interaction Parameters experimentally???

Your response to Chellani suggest to me that you didn't fully understand his response. To summarize:

1) Obtain experimental VLE (P, T, x, y) data. I would expect that these data should come from a variety of molecules containing the groups of interest to provide the widest applicability.
2) Program computer algorithms using UNIFAC to calculate the relationship between P, T, x, y. I prefer algorithms that start with T and x as input and calculate P and y.
3) Select an objective function. Chellani discussed this.
4) Select/program an optimization algorithm that will change the UNIFAC group interaction parameter to minimize the objective function.

If you still have further questions, perhaps you can be more specific. Which step in the process are you having difficulty with?