Professional Documents
Culture Documents
Solvent Improvement For Separating C4 With ACN
Solvent Improvement For Separating C4 With ACN
cn
Received 20 November 2000; received in revised form 9 October 2001; accepted 4 March 2002
Abstract
This paper reports the work to find the best additives added to acetonitrile (ACN) for improving the separation of C4 mixtures
with the help of computer-aided molecular design (CAMD). CAMD is programmed to provide necessary information for our
search of the additives. The two types of additives we seek, liquid solvents and salts, are designed by CAMD. The selected
additives are then tested by experiments. It is found that the addition of salt is more efficient than the addition of liquid solvents
(including water) for improving the separation ability of ACN. After taking implicit properties into consideration, salts NaSCN
and KSCN, are the best candidates to improve ACN. As a strong tool, CAMD greatly reduces experimental working in
separating C4 by extractive distillation with ACN. © 2002 Elsevier Science Ireland Ltd. All rights reserved.
0098-1354/02/$ - see front matter © 2002 Elsevier Science Ireland Ltd. All rights reserved.
PII: S 0 0 9 8 - 1 3 5 4 ( 0 2 ) 0 0 0 3 5 - 2 转载
中国科技论文在线 http://www.paper.edu.cn
1214 Z. Lei et al. / Computers and Chemical Engineering 26 (2002) 1213–1221
Many studies in this field have been made, yet most for the synthesis of liquid solvents. CAMD for liquid
are limited to unit operations, such as gas absorption, solvents is conducted in the following four steps.
liquid–liquid extraction, etc. The application of
CAMD in extractive distillation has seldom been sys-
tematically reported in references (Pretel, Lopez, Bot-
tini & Brignole, 1994). Moreover, in general, CAMD
2.1.1. Group sorting and pre-selection
focuses on the design of organic molecules instead of
Molecular design makes full use of the group con-
salts. In recent years, extractive distillation with salt
cept raised by Franklin (1949), which is built on the
addition has been applied more and more widely in
UNIFAC groups (Jorgensen, Kolbe, Gmehling &
the industry (Duan, Lei & Zhou, 1980; Lei, Duan &
Xu, 1982; Zhang, Qian & Jiang, 1984) and it is very Rasmussen, 1979; Gmehling, Rasmussen & Fre-
important to extend the range of CAMD to the addi- denslund, 1982; Macedo, Weidlich, Gmehling & Ras-
tives including salts. In this work, liquid solvents and mussen, 1983). The groups must be systematically
salts are both designed by CAMD and then tested by ordered to facilitate their use.
experiments. On this basis, the best additives for C4 There are different approaches to sorting the
separation by extractive distillation with ACN are se- groups (Gani & Brignole, 1983; Gani, Nielsen & Fre-
lected. denslund, 1991; Gani & Fredenslund, 1993; Pretel et
al., 1994). The approach we adopt is that a certain
group is determined by the number of attachments
2. Selection of liquid solvents present in a given group (or the valence number of
the group) and the degree of difficulty that the group
2.1. Procedure of CAMD for the liquid sol6ents combines other groups (or the type of attachment).
The UNIFAC groups are sorted and listed in Table 1.
The UNIFAC groups are used as building blocks If a group belongs to several different classes, it
means that this group takes on different types of
Table 1 characterization in different molecules.
Attachment type characterization of UNIFAC groups
Since the groups in Table 1 are based on the UNI-
Type of Groups FAC definition of functional group, they can be used
attachment to obtain the values of some properties related to
UNIFAC-based methods. For the computation of
(N,0) CH3OH CH3NO2 H2O CH3NH2
other properties based on certain group contributions,
CF3 CH3SH CH3CN Furfural
HCOOH DMSO (CH2OH)2 CH2Cl2 but not related to UNIFAC-based methods, new
CHCl3 ACRY DMF NMP parameter tables for the groups in Table 1 have been
CH3NH2 CS2 CCl3F CHCl2F developed using information on the group contribu-
CHClF2 CClF3 CCl2F2 C4H4S
tions and their corresponding parameter tables
Morph AmHCH3 Am (CH3)2
(Joback & Reid, 1987; Reid, Prausnitz & Poling,
(M,1) (CH3)
1987).
(J,2) (CH2)
(L,1) (CH2Cl) The attachment types are indicated as ordered pairs
(L,2) (CHCl) (i, j ), where i is the type of attachment and j, the
(K,1) (CH2 =CH) (OH) (CH3CO) (CHO) number of i attachment. Five types of attachments
(CH3COO) (HCOO) (CH3O) (CH2NH2) are put forward for non-aromatic groups:
(CH3NH) (C5H4N) (CH2CN) (COOH)
(CHCl2) (CCl3) (CH2NO2) (CH2SH)
N= Single molecules as a group having no at-
(I) (BR) (CH =C) Cl(C= C)
(SiH3) tachment with other groups, such as H2O
K= Severely restricted attachment, such as OH
(K,2) (CH= CH) (CH2 = C) (CH3N) (C5H3N)
(CCl2) (CHNO2) (C= C) (DMF-2)
L= partially restricted attachment, such as
(COO) (SiH2) (SiH2O) CH2Cl
(K,1) (L,1) (CH2CO) (CH2COO) (CH2O) (CHNH2) M= unrestricted carbon attachment in linear dual
(CH2NH) (FCH2O) valence or single valence groups, such as
(I,1) (ACH) (ACF) CH3
(H,1) (ACCH3) (ACOH) (ACNH2) (ACCl)
J= unrestricted carbon attachment in radial dual
(ACNO2) valence groups, such as CH2
(H,1) (M,1) (ACCH2)
(H,1) (J,2) (ACCH) For aromatic molecules, two new attachments are
introduced:
中国科技论文在线 http://www.paper.edu.cn
flZ. Lei et al. / Computers and Chemical Engineering 26 (2002) 1213–1221 1215
I= aromatic carbon ring attachment, such as 1991; Gmehling, Li & Schilleer, 1993) and most of
ACH the UNIFAC binary parameters and group parame-
H= substituted aromatic carbon ring attachment, ters are known to us. Many properties can be
such as ACCl calculated from these parameters.
4. In general, a molecule is composed of not more than
Types M and J attachments are extended to aromatic eight groups and the polar groups cannot be over
groups. three. The groups in a molecule must accord with
the following attachment criterion: K= M+ J/2+2
M= unrestricted attachment in a carbon linked to for aliphatic compounds; I+ H= 6, H5 2 or H53
an aromatic carbon, such as ACCH2 for aromatic compounds; for aliphatic–aromatic
J= unrestricted attachment in a radial carbon compounds, these restriction must be satisfied simul-
linked to an aromatic carbon, such as ACCH taneously. Otherwise, this molecule is unsteady un-
der normal condition.
The number of the molecules combined by groups at
The chemically feasible molecules will be generated random is often too large to be processed by a com-
from the above characterized UNIFAC groups in terms puter in a short time, which is the so-called combina-
of the rules that must be met in the formation of a tion explosion. In the molecular design, we should be
molecule and are to be mentioned afterwards. Not all cautious to avoid its occurrence.
of the groups in Table 1 need to be used in CAMD for Assuming that there are n kinds of different groups
a specific application. The groups may be screened in and the molecule combined by them has only m groups,
advance. The pre-selection of groups is conducted on then the number of ways to combine the molecules
two criteria: without other restrictions is:
1. Availability of interaction parameters. The predic-
tion of solvent properties requires the availability of (n+ m− 1)!
UNIFAC binary parameters. Therefore, the first (n− 1)!m!
criterion of the selection is the availability of binary
parameters for the synthesis groups. When the number of groups is in the range of 2–m,
2. Elimination of unsteady compounds and corrosive, the number of such molecules is:
toxic materials. Sometimes for a specialized separa-
tion process, we need to bear in mind that a lot of m
(n+m− 1)!
groups could be excluded in advance. For instance, %
m = 2 (n− 1)!m!
some groups that may cause corrosion to the equip-
ment are avoided for distillation processes. If n= 12 and m= 6, the number of combination
By sticking to these criteria, the number of partici- ways is very high with 18,551. Therefore, the assembled
pating groups is reduced and the time the computer rules must be considered to effectively decrease the
takes to fulfil the combination of groups is saved. combination ways.
Table 3
Results of CAMD for liquid solvent
properties. Even though the prediction methods have 3. Selection of the salt
been extensively used, they should be considered only
as an approximation for the estimation of solvent prop-
erties. This is where the limitations of CAMD lie. 3.1. CAMD for the salts
Table 5
Experimental values of relative volatility of C4 mixtures at different additives concentration
are listed in Table 7, in which the values of relative KI. On this basis, some experiments, just as we carried
volatility and selectivity are closely related with salting out for liquid solvents, were performed to investigate
coefficients. Salting coefficients are calculated by the the effect of salts. Table 9 shows the experimental
scaled particle theory (Masterton & Lee, 1970; Pierotti, values at 30 °C with salt concentration 10 wt%. The
1976). The scaled particle theory, deduced from ther- subscripts 1–5 represent n-butane, 1-butene, trans-2-
modynamics and statistical physics, has a defined phys- butene, cis-2-butene and 1,3-butadiene, respectively.
ical meaning and the required molecular parameters are
readily available. In particular, in recent years the study
of scaled particle theory has been further developed.
Table 6
Therefore, it is reasonable to deal with salting effect of Ions of CAMD for salts
extractive distillation with this theory, which is a great
theory and application prospect and should be given Ions Groups
more attention.
Cation Li+ Na+ K+ Rb+
Cs+ Ag+ NH+ 4 Cu+
Mg2+ Ca2+ Sr2+ Ba2+
3.2. Results of CAMD Be2+ Fe2+ Zn 2+
Cu2+
Al3+ Fe3+
The additive constraints for the system of ACN/C4 Anion OH− F− Cl− Br−
are listed in the following order. I− SCN− NO−3 NO−2
1. Pre-selected group type number: 8 (Li+, Na+, K+, CH3COO− ClO−
4 BF−
4
Joback, K. G., & Reid, R. C. (1987). Estimation of pure-component Millero, F. J. (1971). The molar volumes of electrolytes. Chem. Re6.,
properties from group-contributions. Chem. Eng. Comm., 57, 71, 147 – 176.
233 – 243. Ourique, J. E., & Telles, A. S. (1998). Computer-aided molecular
Joback, K. G., & Stephanopoulos, G. (1995). Searching spaces of design with simulated annealing and molecular graphs. Comput.
discrete solutions: the design of molecules possessing desired Chem. Eng., 22, S615 – S618.
physical properties. Ad6. Chem. Eng., 21, 257 –311. Pierotti, R. A. (1976). A scaled particle theory of aqueous and
Jorgensen, S. S., Kolbe, B., Gmehling, J., & Rasmussen, P. (1979). nonaqueous solutions. Chem. Re6., 76 (6), 717 – 726.
Vapor-liquid equilibria by UNIFAC group contribution. Ind. Pistikopoulos, E. N., & Stefanis, S. K. (1998). Optimal solvent design
Eng. Chem. Proc. Des. De6., 18, 714 –722. for environmental impact minimization. Comput. Chem. Eng.,
Lei, L., Duan, Z., & Xu, Y. (1982). Studies of extractive distillation 22 (6), 717 – 733.
with salt (I). Petrochem. Technol. (China), 11, 404 –409. Pretel, E. J., Lopez, P. A., Bottini, S. B., & Brignole, E. A. (1994).
Computer-aided molecular design of solvents for separation pro-
Leroi, J. C., Masson, J. C., & Renon, H. (1977). Accurate measure-
cesses. Am. Inst. Chem. Eng. J., 40 (8), 1349 – 1360.
ment of activity coefficients at infinite dilution by inert gas
Raman, V. S., & Maranas, C. D. (1998). Optimization in product
stripping and gas chromatography. Ind. Eng. Chem. Proc. Des.
design with properties correlated with topological indices. Com-
De6., 16 (1), 139 – 144.
put. Chem. Eng., 22 (6), 747 – 763.
Macedo, E. A., Weidlich, U., Gmehling, J., & Rasmussen, P. (1983).
Reid, R. C., Prausnitz, J. M., & Poling, B. E. (1987). The Properties
Vapor-liquid equilibria by UNIFAC group contribution. Ind. of Gases and Liquids. New York: McGraw-Hill.
Eng. Chem. Proc. Des. De6., 22, 676 –678. Sinha, M., Achenie, L. E. K., & Ostrovsky, G. M. (1999). Environ-
Marcoulaki, E. C., & Kokossis, A. C. (1998). Molecular design mentally benign solvent design by global optimization. Comput.
synthesis using stochastic optimisation as a tool for scoping and Chem. Eng., 23 (10), 1381 – 1394.
screening. Comp. Chem. Eng., 22, S11 –S18. Tiegs, D., Gmehling, J., Rasmussen, P., & Fredenslund, A. (1987).
Masterton, W. L., & Lee, T. P. (1970). Salting coefficients from Vapor-liquid equilibria by UNIFAC group contribution revision
scaled particle theory. J. Phys. Chem., 74 (8), 1776 –1782. and extension 4. Ind. Eng. Chem. Res., 26, 159 – 161.
Mavrovouniotis, M. L. (1998). Design of chemical compounds. Venkatasubramanian, V., Chan, K., & Caruthers, J. M. (1995).
Comp. Chem. Eng., 22, 713 –715. Genetic algorithmic approach for computer-aided molecular de-
Meniai, A. H., Newsham, D. M. T., & Khalfaoui, B. (1998). Solvent sign. ACS Symp. Ser., 589, 396 – 414.
design for liquid extraction using calculated molecular interaction Zhang, Q., Qian, W., & Jiang, W. (1984). Studies of extractive
parameters. Trans. IchemE, 76 (A11), 942 –950. distillation with salt (I). Petrochem. Technol. (China), 13, 1–9.