This preview shows page 1. Sign up to view the full content.
Unformatted text preview: Distillation Troubleshooting
Stanislaw K. Wasylkiewicz
Aspen Technology, Inc. A tool called azeotrope pressure
sensitivity analysis (APSA) can shed light
on the dependence of an azeotrope’s composition
and temperature on changes in pressure. R igorous, robust and reliable thermodynamic models are crucial for the synthesis and design of
separation systems for azeotropic mixtures.
Incorrect prediction of azeotropes often leads to infeasible separation sequences.
Although much progress in the modeling of physical
and chemical equilibrium has been made, there is no universal model that can be used for any mixture at any conditions without adjusting various empirical parameters.
Furthermore, because we often neglect some terms that we
think are insignificant in rigorous thermodynamic models,
the adjustable empirical parameters must account for not
only what they were supposed to describe (e.g., interactions between molecules in liquid phase), but also all
effects that have been neglected (1). This makes the models less predictable and their extrapolation to multicomponent mixtures questionable. Even if model predictions for
all binary pairs are consistent with experimental vaporliquid equilibrium (VLE) data, we have to verify behavior
of the whole mixture before using a thermodynamic package for rigorous simulations.
Several excellent tools have recently been developed
for selecting, verifying and troubleshooting thermodynamic models. Phase diagrams, residue curve maps
(RCMs) and distillation region diagrams (DRDs) are
invaluable for this purpose because of their visualization
capabilities. However, they are restricted to ternary and
There are no such restrictions for the azeotrope pres22 www.cepmagazine.org June 2006 CEP sure sensitivity analysis (APSA), where temperature and
composition bifurcation diagrams can be created for mixtures of any number of components. Since pressure
changes from stage to stage in distillation columns, APSA
is an excellent new tool for quick model verification and
troubleshooting of real, non-isobaric distillation columns.
It qualitatively reveals changes in the separation space that
are difficult to find using an RCM or a DRD.
This article describes, through several case studies, some
frequently encountered modeling problems during simulation of non-isobaric azeotropic distillation columns, and
applies the new troubleshooting tools to these simulations. An overview of the tools
The shape of the separation space, including distillation
boundaries, distillation regions and azeotropes (singular
points of the maps), can be easily examined on RCMs (2).
Visual analysis of an RCM’s structure provides valuable
insight into the underlying thermodynamics, making
RCMs useful for screening binary VLE parameters based
on alternative databanks and regressions strategies (3).
Because RCMs depict VLE behavior in a ternary composition space, they can be more relevant than binary phase
diagrams in assessing the validity of various parameter
sets for distillation modeling.
Reference 3 shows examples of troubleshooting phase
equilibria that include using data outside their intended
range, employing VLE instead of vapor-liquid-liquid equilibrium (VLLE) calculations, using parameter sets opti- mized for different regression objectives, and choosing the
appropriate model for the vapor phase. Reference 4 presents examples of utilizing RCMs and VLLE diagrams for
selection of the appropriate thermodynamic model and
verification of interaction parameters, and Refs. 5 and 6
explain how these methods can be applied in troubleshooting distillation columns. Conveniently, RCMs can now be
readily generated using commercial process simulation or
synthesis software. Moreover, the individual points of the
maps can be calculated for any number of components and
used to test phase equilibrium models.
Azeotropes determine distillation regions and distillation boundaries. Therefore, knowledge of temperatures
and compositions of all the azeotropes in a mixture at a
specified pressure is critical for both the design and operation of distillation systems. Calculating them is complicated by the strong non-linearities encountered in VLLE and
the presence of multiple solutions, both real and spurious.
One can attempt to find all of the solutions by using a
nonlinear solver with several starting points. However, even
with extreme calculation effort, this approach does not guarantee that all of the azeotropes will be found. Several techniques have been proposed to solve this problem, e.g., the
Levenberg-Marquardt algorithm (7), the interval Newton
method (8), a global optimization method (9), etc.
The most reliable and robust method for calculating all
the azeotropes in a homogeneous mixture has been proposed by Fidkowski, et al. (10) and later generalized to
include heterogeneous liquids (11). This method combined
with an arc length continuation and rigorous stability
analysis (12) is an efficient way to find all of the homogeneous and heterogeneous azeotropes predicted by a thermodynamic model at a specified pressure.
In the real world, plant operation is not static.
Changes in product quality and quantity often require
operation at different pressures. Even when operating
near steady state, many columns exhibit significant pressure gradients between the distillate and the bottoms.
Thus, it is critical to understand the pressure sensitivity
of the azeotropes in a mixture.
When upper and lower pressure limits are specified, the
analysis can start from one pressure limit and then the
azeotrope composition can be calculated for gradually
increasing or decreasing pressures. This simple parametercontinuation procedure can follow any known azeotrope.
However, it cannot find new azeotropes that appear as the
An alternative is to apply rigorous azeotrope calculation procedures at each pressure interval. Such an
approach, though, would be extremely computationally
intensive and time consuming. Table 1. Temperatures and compositions of binary
azeotrope WA at selected pressures. Pressure,
of W Mole
of A 80.64
0.047468 To overcome these difficulties, we have developed a new
method called azeotrope pressure sensitivity analysis
(APSA) that reveals how the compositions and temperatures
of azeotropes change with pressure. It applies bifurcation
theory together with an arc length continuation to perform
the calculations (13). This allows us to not only follow individual azeotropes but also to find any new azeotrope that
appears in a specified pressure interval. Therefore, all bifurcation pressures, where azeotropes appear or disappear, can
be found. The computation time required for the entire pressure sensitivity analysis is similar to the time necessary for
the homotopy-continuation method to find all azeotropes at a
The pressure-sensitivity of azeotropes provides new
opportunities in the design of azeotropic distillation
sequences. Increasing or decreasing operating pressures in
individual columns changes the composition of azeotropes
and positions of distillation boundaries, and can even
cause their appearance or disappearance. This can have
enormous effects on the topology of the RCM and the feasibility of distillation sequences.
APSA shows opportunities for pressure-swing distillation and heat integration between columns in the
sequence. Pressure-swing distillation can often be an
attractive alternative for breaking homogeneous
azeotropes and sometimes can considerably simplify complex separation systems.
Pressure-sensitivity information can also be important
in the design and troubleshooting of real distillation
columns, especially where there is a substantial pressure
drop in the column. In some cases, this can lead to a
switch in topology of distillation regions inside the column and cause serious problems in convergence of simulators in steady-state and dynamic modes. Troubleshooting non-isobaric
Let us consider a typical separation problem. A mixture
containing mostly components A and W is to be separated.
There is a binary azeotrope between components A and W.
Since the azeotrope is not sensitive to pressure change
(Table 1), an entrainer, C, was selected to separate the binaCEP June 2006 www.cepmagazine.org 23 Distillation Azeotrope pressure
The sensitivity of
azeotropes to changes in
pressure has been known
and studied for years. The
magnitude of pressure
effects depends on the mixture. Sometimes, composiLower
tions of azeotropes change
very little (e.g., the ethanolA+W
water azeotrope (13)).
However, there are mixtures
where compositions of some
azeotropes change rapidly
with pressure and even
■ Figure. Extractive distillation is used to break the binary azeotrope.
azeotropes that appear or
disappear as pressure varies.
Knapp (14) introduced the concept of a bifurcation
ry mixture in an extractive distillation sequence of two dispressure, where an azeotrope appears or disappears, and
tillation columns (Figure 1). The first column was set up at
showed the necessary conditions for a homogeneous binaa constant pressure of 150 kPa and converged quickly to the
ry azeotrope to bifurcate. An azeotrope exists on one side
desired products: high-purity A and a binary mixture of W
of a bifurcation pressure and a tangent pinch on the other
and C. Then a pressure drop in the column was taken into
side. As the pressure increases or decreases away from the
account. This time, the required high recovery of compobifurcation pressure, the severity of the tangent pinch
nent A in the bottom of the column was never achieved,
decreases. Therefore, a distillation column should not
even with an extreme reflux and a large number of stages.
operate near a bifurcation pressure.
For the three key components (W, A and C), distillaIn real distillation columns, pressure changes from
tion region diagrams were created for three selected
stage to stage. This can lead to a switch in topology of dispressures to examine the three-component space for
tillation regions and cause serious problems in both real
azeotropes and distillation boundaries, as shown in
operation and in convergence of calculations for simulaFigure 2. By examining the DRDs, one can easily conclude that it is impossible to achieve complete recovery
tors in steady-state or dynamic modes. In such cases,
of component A in the bottom product of the first colinformation about the bifurcation pressures could be
umn at 50 kPa because of the presence of a distillation
extremely useful in troubleshooting both for simulators
boundary at this pressure. But in our simulation case,
and real operations.
pressure at the top of the column does not drop as low
The APSA method (15) is based on bifurcation theory
— it always stays above 100 kPa. To troubleshoot this
together with an arc length continuation. The method finds
case further, we will use APSA.
all bifurcation pressures within specified pressure limits
and allows one not only to follow individual azeotropes, but also to efficientA
ly find any azeotrope that appears within the pressure range. (More details are
available in Ref. 13.) This article
applies the APSA tool to create various
bifurcation diagrams, which are
extremely useful in troubleshooting of
azeotropic distillation columns.
The pressure sensitivity analysis of
azeotropes starts with a branch for each
pure component and each azeotrope
■ Figure 2. Distillation region diagrams for ternary mixture W-A-C at 50, 100 and150 kPa,
found at 50 kPa. Each branch is folbased on the NRTL-Ideal model.
Top Vapor 24 www.cepmagazine.org June 2006 Distillate CEP Top Vapor 1.0 A Mole Fraction NRTL
50–150 kPa 0.8
0.7 Branch WA
Branch WC 0.6 Branch WAC 0.5 90 95 100 105 110 115
Pressure, kPa 120 125 130 ■ Figure 5. Composition branches WC and WAC for pressure
analysis of azeotropes for ternary mixture W-A-C between 85 and
130 kPa, based on the NRTL-Ideal model. 0.1 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 W ■ Figure 3. Homotopy continuation branches WA, WC and WAC
for pressure analysis of azeotropes for ternary mixture W-A-C
between 50 and 150 kPa, based on the NRTL-Ideal model. 106
Temperature, °C 0.4 0.0
85 0.2 102
Branch WAC 98
85 Branch WAC,
Component C Branch WAC,
Component A 0.3 C 0.6 0.2 0.4 0.0
0.0 Branch WAC,
Component W 0.8 0.9 Branch WC 1.0 90 95 100 105 110 115 120 125 130
Pressure, kPa ■ Figure 4. Temperature branches WA, WC and WAC for pressure
analysis of azeotropes for ternary mixture W-A-C between 85 and
130 kPa, based on the NRTL-Ideal model. lowed and continually checked for new bifurcations
between 50 and 150 kPa or until the branch disappears. The
corresponding homotopy continuation branches are shown
in Figure 3, temperature branches in Figure 4, and composition branches in Figure 5.
The binary azeotrope WA is not sensitive to pressure
change (see also Table 1). On the other hand, the WC and
WAC azeotropes are very pressure sensitive and exist only in
a narrow pressure range (WAC between 89 and 127 kPa, WC
only in the proximity of 127 kPa). There are two bifurcation
points where the WAC branches collapses: one on branch
WA (at 112 kPa), the other on branch WC (at 127 kPa). There is also a turning point on branch WAC at 89 kPa.
Notice how much more information the APSA provides compared to DRD diagrams created for a few
selected pressures (Figure 2). The DRDs confirm the
existence of only one azeotrope, WA. The binary
azeotrope WC exists only in a very narrow range of pressures (see Figure 4 and Figure 5) and is practically visible
as a homotopy continuation branch WC only in Figure 3.
The ternary azeotrope WAC was not found by the
azeotrope searching algorithm (the 100-kPa DRD in
Figure 2) because the homotopy method used in the algorithm cannot find isolas — i.e., continuation paths not
connected to any homotopy branch that starts from a pure
component (10). The model actually predicts two WAC
azeotropes at 100 kPa (Figure 5) — one that is a saddle,
and another that is an unstable node. Because two
azeotropes have been missed and their contributions cancel each other, the consistency test is fulfilled and there is
no warning during creation of DRD at 100 kPa. Troubleshooting the distillation of
Another problem in the simulation of distillation
columns has been reported for the mixture of cyclohexanone (CC6one), cyclohexanol (CC6ol) and phenol.
APSA has been applied to the mixture to verify the
accuracy of predictions of two VLE models throughout
a pressure range of industrial interest. The NRTL (nonrandom two-liquid) model predicts the existence of the
ternary maximum-boiling azeotrope in a wide pressure
range (only slightly pressure-sensitive), whereas the
Wilson model predicts the existence of the ternary saddle azeotrope only in a very narrow range of pressures
(extremely pressure sensitive).
CEP June 2006 www.cepmagazine.org 25 Distillation tures of all azeotropes are so
close that any small change
in pressure can change their
type and the structure of the
DRD. At 5 kPa, the CC6olphenol azeotrope is a stable
node and the CC6one-phenol
azeotrope is a saddle, where113.53 118.22
as at 15 kPa the CC6ol-pheCC6one
nol azeotrope is a saddle and
the CC6one-phenol azeotrope
■ Figure 6. Distillation region diagrams for the mixture of cyclohexanone (CC6one), cyclohexanol
is a stable node.
(CC6ol) and phenol at 5, 10 and 15 kPa, based on the Wilson model.
The homotopy continuation method (11) was used to find all azeotropes at par0.8
ticular pressures. The topological consistency was
checked using a topological constraint (10, 18) and all
diagrams presented are topologically consistent. If the
topological constraint is not fulfilled, one or more
azeotropes are missing or their properties have been
estimated incorrectly. However, fulfillment of the topo0.4
logical constraint does not guarantee that all azeotropes
have been found or that the system properties have been
determined correctly. The topological constraint is a
necessary, but not a sufficient, condition for consistency
in azeotrope calculations (11).
APSA was performed for the mixture between 5 kPa
and 15 kPa. Branches of solutions are shown in a compo5
10 11 12 13 14
sition bifurcation diagram (Figure 7). In this pressure
range, two bifurcation points have been found, as shown
in Table 2:
■ Figure 7. Composition changes of azeotropes with pressure for the
• B1 at 8.26 kPa, where the ternary CC6one-CC6olmixture of cyclohexanone (squares), cyclohexanol (diamonds) and
phenol branch bifurcates from the CC6one-phenol branch,
phenol (triangles) between 5 and 15 kPa, based on the Wilson model.
and the CC6one-phenol azeotrope changes type from sadFor this ternary system, it is known from measuredle to stable node
ments (16) that:
• B2 at 10.94 kPa, where the ternary CC6one-CC6ol• there is a maximum-boiling azeotrope in the binary
phenol branch collapses on the CC6ol-phenol branch, and
system CC6one-phenol between 8.00 and 101.32 kPa
the CC6ol-phenol azeotrope changes type from stable
• there is a maximum-boiling azeotrope in the binary
node to saddle.
system CC6ol-phenol between 9.33 and 101.32 kPa
Both binary azeotropes are only slightly sensitive to
• the three-component system does not form a ternary
pressure changes. On the other hand, the ternary azeotrope
azeotrope at 12.00 kPa.
(saddle) is extremely pressure-sensitive and exists only in
Distillation region diagrams for the mixture are presentthe very narrow pressure range, between the two bifurcaed in Figure 6 for a few selected pressures. VLE calculations were based on
Table 2. Results of bifurcation pressure search for the mixture of
the Wilson model. Binary interaction
cyclohexanol (CC6ol), cyclohexanone (CC6one) and phenol
parameters were taken from Ref 17.
between 5 and 15 kPa, based on the Wilson model.
The Wilson model does not predict
the existence of the ternary azeotrope at
5 kPa and 15 kPa. There is, however, a
very narrow range of pressures where the
ternary azeotrope exists. The temperaCC6ol CC6ol 99.25 Mole Fraction 85.19 26 www.cepmagazine.org June 2006 CEP CC6ol 108.19 in the wider pressure range
of 1–100 kPa, all three
azeotropes (two binary sadNRTL
dles and one ternary maximum-boiling) exist and do
not change their types.
APSA was performed for
the mixture in the pressure
range 1–100 kPa. Branches
of solutions predicted by the
■ Figure 8. Distillation region diagram for the mixture of cyclohexanone (CC6one), cyclohexanol
NRTL model are shown in a
(CC6ol) and phenol at 5, 10 and 15 kPa, based on the NRTL model.
composition triangle in
Figure 9. No bifurcation
pressures were found. In this case, the binary CC6ol-phe1.0
nol azeotrope is the most pressure-sensitive, whereas the
ternary maximum-boiling CC6one-CC6ol-phenol
azeotrope is the least sensitive. This contradicts the predicNRTL
tions of the Wilson model (Figure 7).
Performing both equilibrium and azeotrope calcula0.7
tions for this ternary system is challenging. The two
models give qualitatively different predictions for the
ternary CC6one-CC6ol-phenol azeotrope: an extremely
pressure-sensitive saddle azeotrope in a very narrow
range of pressures for the Wilson model, and a moderately pressure-sensitive maximum-boiling azeotrope in
a wide pressure range for the NRTL model. At the
same time, the binary azeotropes are predicted equally
well by both models and are consistent with binary
More experiments for the ternary system are neces0.0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
sary, especially in the pressure range between the
bifurcation pressures (8.26 and 10.94 kPa), to allow us
■ Figure 9. Branches of pressure sensitivity analysis of
to accept or reject the Wilson model predictions. Based
azeotroopes for the mixture of cyclohexanone (CC6one), cycloon the only one available experimental data point for
hexanol (CC6ol) and phenol between 1 and 100 kPa, based on
the ternary mixture at 12.00 kPa (16), we can conclude
the NRTL model.
that the NRTL model should be corrected (better intertion pressures 8.26 and 10.94 kPa. Experiments have
action parameters should be regressed). Generally, if a
shown that the three-component system does not form the
model is expected to reflect reality, interaction parameternary azeotrope at 12.00 kPa (16). This agrees well with
ters should be regressed for the model using experithe Wilson model predictions. New experiments are needmental data covering the particular pressure range of
ed in the pressure range between the two bifurcation presinterest. In this case, special attention is required for
sures to check the model predictions. It must be emphathe saturated pressure correlation for pure components
sized here that, unlike stable and unstable azeotropes, it is
(e.g., the Antoine equation). Since CC6ol and CC6one
much more difficult to confirm experimentally the exisare close-boiling components, small inaccuracies in
tence of a saddle azeotrope.
prediction of saturated pressure can cause significant
Distillation region diagrams for the mixture calculatchanges in the DRD structure.
ed based on the NRTL model are presented in Figure 8
for the same pressures as for the Wilson model. Binary
interaction parameters were taken from Ref. 17. The
It is extremely important to select the proper therNRTL model predicts the existence of the ternary maximodynamic model and carefully verify the behavior of
mum-boiling azeotrope for all of these pressures. Even
the mixture for a particular set of components before
CC6ol 85.19 CC6ol 99.25 CC6ol 108.19 CEP June 2006 www.cepmagazine.org 27 Distillation any attempt is made to simulate a distillation column.
DRDs are extremely useful for this verification
because of their visualization capabilities. They are,
however, restricted to ternary and quaternary mixtures.
There are no such restrictions for APSA, which
involves calculating bifurcation pressures and creating
temperature and composition bifurcation diagrams for
any numbers of components. This provides an excellent
tool for both design and troubleshooting of real distillation columns.
The extreme sensitivity of azeotropes to pressure
change in the examples discussed here has been detected in several distillation systems, especially when
close-boiling components were present. If detected,
this extreme sensitivity should trigger more careful
examination of the thermodynamic model (applicability, parameters) and perhaps some additional VLE
measurements. At a minimum, it should draw attention
to the situation so the particular pressure range can be
avoided both in the design and operation of the distillation column.
Changes in product quality and quantity often require
operation of distillation columns at different pressures.
Even when operating at a near-steady-state condition,
many columns exhibit significant pressure gradients
between the distillate and bottoms. In some cases, this can
lead to a switch in the topology of the distillation regions
and cause serious problems in convergence of calculations
for steady-state or dynamic simulators.
APSA provides an understanding of the dependence
of an azeotrope’s compositions and temperatures on
changes in pressure. This information can be of critical
importance in the design and troubleshooting of distilla- tion systems. APSA can also be used to verify the suitability and accuracy of the VLE models throughout a
pressure range if the system is known to have particular
azeotropes at particular pressures. Literature Cited
3. 4. 5. 6. 7. 8.
9. 10. 11. 12.
STANISLAW K. WASYLKIEWICZ, PhD, P.Eng., is a senior advisor at Aspen
Technology, Inc. (900, 125 – 9th Ave. SE, Calgary, Alberta T2G 0P6, Canada;
Phone: (403) 303-1047; E-mail: firstname.lastname@example.org), where
his primary emphasis is conceptual design and simulation of distillation
processes. He is currently working on new tools for petroleum distillation
in the computer program RefSYS, a new product that enables companies
to optimize refinery-wide performance using an integrated model of their
facilities. He is a graduate in chemical engineering from the Technical Univ.
of Wroclaw (Poland), where he received his MSc and PhD and taught for
several years. Prior to joining the Conceptual Design Team of Hyprotech
(Calgary, Canada), he worked as a researcher at the Univ. of
Massachusetts (Amherst), where he was developing Mayflower, software
for the conceptual design of distillation systems. Throughout his
engineering career, he has become involved in several areas of expertise,
including thermodynamics, phase equilibrium, process synthesis,
simulation and optimization, and the design of non-azeotropic, azeotropic,
heterogeneous and reactive distillation columns. He has published over
100 scientific papers and contributed to technical books. He is a member
of AIChE and a registered professional engineer in Alberta, Canada. 28 www.cepmagazine.org June 2006 CEP 13.
14. 15. 16.
17. 18. Agraval, R., et al., “Uncovering the Realities of Simulation
(Parts 1 and 2),” Chem. Eng. Progress, 97 (5), pp. 42–52 and
97 (6), pp. 64–72 (May and June, 2001).
Doherty, M. F., and M. F. Malone, “Conceptual Design of
Distillation Systems,” McGraw-Hill, New York, NY (2001).
Villiers, W. E., et al., “Navigate Phase Equilibria via
Residue Curve Maps,” Chem. Eng. Progress, 98 (11), pp.
66–71 (Nov. 2002).
Wasylkiewicz, S. K., and H. K. Shethna, “VLE Data
Estimation for Synthesis of Separation Systems for
Azeotropic Mixtures,” AIChE Spring National Meeting,
Paper No. 4e, New Orleans, LA (Mar. 2002).
Wasylkiewicz, S. K., “Diagrams Troubleshoot Acetic Acid
Dehydration Simulation,” in “Distillation Troubleshooting,”
by Kister, H. Z., Wiley, Hoboken, NJ (2006).
Wasylkiewicz, S. K., “Modeling Ternary Mixture Using
Binary Interaction Parameters,” in “Distillation
Troubleshooting,”by Kister, H. Z., Wiley, Hoboken, NJ (2006).
Chapman, R. G., and S. P. Goodwin, “A General
Algorithm for the Calculation of Azeotropes in Fluid
Mixtures,” Fluid Phase Equilibria, 85, pp. 55-69 (1993).
Maier, R. W., et al., “Reliable Computation of Homogeneous
Azeotropes,” AIChE J., 44, pp. 1745-1755 (1998).
Harding, S. T., et al., “Locating All Homogeneous
Azeotropes in Multicomponent Mixtures,” Ind. Eng. Chem.
Res., 36, pp. 160-178 (1997).
Fidkowski, Z. T., et al., “Computing Azeotropes in
Multicomponent Mixtures,” Comput. Chem. Eng., 17, pp.
Wasylkiewicz, S. K., et al., “Computing All Homogeneous
and Heterogeneous Azeotropes in Multicomponent
Mixtures,” Ind. Eng. Chem. Res., 38, pp. 4901-4912 (1999).
Wasylkiewicz, S. K., et al., “Global Stability Analysis and
Calculation of Liquid-Liquid Equilibrium in
Multicomponent Mixtures,” Ind. Eng. Chem. Res., 35, pp.
Wasylkiewicz, S. K., et al., “Pressure Sensitivity Analysis of
Azeotropes,” Ind. Eng. Chem. Res., 42, pp. 207-213 (2003).
Knapp, J. P., “Exploiting Pressure Effects in the Distillation
of Homogeneous Azeotropic Mixtures,” PhD Thesis, Univ.
of Massachusetts, Amherst (1991).
Wasylkiewicz, S. K., et al., “Pressure Sensitivity Analysis of
Azeotropes in Synthesis of Distillation Column Sequences,”
Hungarian J. of Industrial Chemistry, 28, pp. 41-45 (2000).
Gmehling, J., et al., “Azeotropic Data,” VCH
Verlagsgesellschaft GmbH. Weinheim (1994).
Gmehling, J., et al., “Vapor-Liquid Equilibrium Data
Collection, DECHEMA Chemistry Data Series,” Vol. I,
DECHEMA, Frankfurt (1981).
Zharov, W. T., and L. A. Serafimov, “Physicochemical
Fundamentals of Distillation and Rectification,” Khimiya,
Leningrad (1975). ...
View Full Document
- Fall '08