Thank you for your contribution to this discussion.
I would like to re-iterate something that you stated. The
UNIFAC method is a predictive method that should only be used
when no data are available. When data are available they should
be fit to a model like UNIQUAC or NRTL. I don't think we can
overemphasize this point. And I also agree that you must be
careful with supercritical components.
This leads me to another point. I stand by my comment that
VLE models should not be used for LLE (including UNIFAC). I
quote from the late Aage Fredenslund, as I mentioned in my
previous post, one of the pioneers of the UNIFAC method:
It has been found that UNIFAC with parameters based on VLE
does not yield quantitative prediction of LLE. ... It is
questionable if anything can be done about [this] limitation.
from Fluid Phase Equilibria, vol. 52, p. 137, (1989).
Another example is the Wilson equation for activity coefficients.
This method is very popular for VLE, but cannot be used for LLE
regardless of the parameter set.
I would be very interested in a reference to a paper that claims
that their set of UNIFAC equations and parameters yields accurate
results for both VLE and LLE.
John Carroll
AQUAlibrium
http://www.telusplanet.net/public/jcarroll/AQUA.HTM
Last Update: Nov. 1, 1998
B. van Dyk wrote:
As far as I am aware there has only ever been a single set of LLE paramaters published for UNIFAC (Magnussen, 1981). This set included
far less groups than the current revision of UNIFAC (5).
I would recommend that you rather use the modified UNIFAC (Dortmund) method. A second revision of the parameters was recently published by Gmehling et al.
The parameters in this publication were fitted on VLE, LLE, hE and infinite dilution activity coefficients and give good predictions of both VLE and LLE.
There are two other versions of modified UNIFAC - the Lyngby and Free Volume
modifications. I have no experience of these, but some claim that the Dortmund
method is superior. (This of course includes the author of the Dortmund method...)
Note that UNIFAC will give only predictions. If the calculations must be accurate,
you would do better to find experimental data to fit UNIQUAC, NRTL or some other
model to.
Also note that if any of the components are above their critical temperatures, you should
use the PSRK method or something similar. The PSRK method combines modified UNIFAC with the SRK EOS and gives very good results.
The calculations are simple, if slightly tedious. You should have no trouble
programming
this yourself. If you prefer to use commercial or freeware software, check the models that
these programs are using as well as the revision of the interaction parameters.
--
Braam van Dyk
Thermal Separations Research
Chemical Engineering
University of Stellenbosch
South Africa
Sysop: | Keyop |
---|---|
Location: | Huddersfield, West Yorkshire, UK |
Users: | 300 |
Nodes: | 16 (2 / 14) |
Uptime: | 01:14:23 |
Calls: | 6,706 |
Calls today: | 6 |
Files: | 12,235 |
Messages: | 5,349,922 |