A boiling liquid mixture is subjected to distillation to separate its
components according to their different volatilities.1 In situations
where the boiling points of components of a mixture are close to each other,
extractive distillation can be used rather than simple distillation. In
extractive distillation, a non-volatile solvent is used—this alters the relative volatility of the
components without forming an azeotropic mixture with them.2 The
extractive distillation process utilizes the selective accentuation of the
non-ideal characteristics of the liquid phase components to be separated that
are induced by the presence of solvent. The solvent achieves this by altering
the activity coefficient of the individual components.3
These newly formed solvent-component
mixtures can then be separated using normal distillation. The individual
mixtures once separated out as products from the process can then be purified
in another distillation column to separate the solvent from the components.
This solvent can then be recycled back into the first column along with fresh
solvent. To quantify the risks associated with different parameter settings in
a model and optimize the system, a sensitivity analysis can be performed. This
is a valuable tool for understanding the model's strengths and weaknesses and
can be utilized to recommend operational approaches for the system.
Theory. Extractive distillation
must adhere to certain guidelines to be effective:3
For this article, the separation of an ethanol and benzene mixture is
considered. Ethanol and benzene—with their boiling points of 78.24°C and 80.08°C, respectively4—are difficult to
separate using normal distillation. To separate them, p-xylene with its relatively
high boiling point of 138.3°C
is used as a solvent in a dual-feed distillation column.4 It reacts
with ethanol and benzene in a way so that the benzene and p-xylene mixture
separates out from the bottom while relatively pure ethanol is taken out from
the top. This benzene and p-xylene mixture is then sent to a solvent recovery
distillation column where both components are separated out and the p-xylene is
recycled back into the first distillation column as solvent. A simulation study
was conducted using the process simulator DWSIM, an open source software to
simulate steady-state operations.5 Input conditions for the above
mentioned operation were entered into the software to analyze the process.
FIG. 1 shows the T-xy plot for the minimum boiling
ethanol-benzene azeotropic mixture with an azeotropic temperature of around 314
K, which makes it difficult to separate using simple distillation methods.6
As shown in the diagram, a relatively flat region is found near the minimum on
the bubble point curve, which slopes gently in both directions. With increasing
distance from the intersection point, these slopes become steeper. As for the
dewpoint curve, a relatively constant slope characterizes it between the
azeotropic and pure component points.7
Mathematical model. An equilibrium stage
model (EQ) is considered for the extractive distillation process. One EQ stage
corresponds to one section of tray or packaging, depending on the type of
column. To develop this model, certain assumptions must be made, as outlined here:8
FIG. 2 represents the EQ stage model schematically. The
mathematical model for the extractive distillation process at a particular
stage in the column is given by the equations below.9
The mass balance across an EQ stage is given by Eq. 1:
When hold-up of vapor is not considered in the EQ stage, the material balance for the components is determined by Eq. 2:
The sidestream-to-interstage flow ratio for the vapor and liquid phases is (Eq. 3):
To relate the
equilibrium between the vapor and liquid phases, Eq. 4 is used:
where (Eq. 5):
The equation of enthalpy balance across an EQ stage is shown
in Eq. 6:
All time derivatives in
the above equations are zero since the process is said to occur under steady-state
conditions. Additionally, since the various components do not react, the
reaction rates also remain zero.
CASE STUDY MODELING ON DWSIM
An equimolar azeotropic mixture of benzene and ethanol at the rate of
100 kmol/hr (1 atm and 298.15 K) is fed into a 71-tray extractive distillation
column on Tray 50.10 The makeup solvent, which consists of pure
p-xylene, is fed at a flowrate of 0.301 kmol/hr (1 atm and 298.15 K) to a
static mixer and is mixed with the recycle stream from the solvent recovery
column. The outlet of this mixture is fed to the extractive distillation column
on Tray 24. With a reflux ratio of 2.13, the lighter product distills out from
the top and contains ethanol as the key component, while the bottoms is a
mixture of benzene and p-xylene with the benzene mol fraction at 0.189. The
bottoms product is then sent to a 21-tray solvent recovery column where it
enters the column on Tray 8. In this simple distillation where the reflux ratio
is 6.14, benzene is separated as the top product and p-xylene distills out from
the bottom. The bottoms stream from the solvent recovery column is passed
through a cooler, where it is cooled to a temperature of 100°C before being recycled
and fed into the static mixer.
FIG. 3 shows the flowsheet of the entire process. The
products shown in parentheses are the main key components of that stream.
The simulation was conducted using the UNIversal QUAsi‐Chemical (UNIQUAC) activity coefficients in a DWSIM
process simulator.11 For simplicity, the whole system was considered
to be at a constant pressure of 1 atm without any pressure drop. For both of the
extractive distillation and solvent recovery columns, the K-value model used was
DECHEMA, with the ideal gas as the equation of state and Antoine equation used
for the vapor pressure relations.12 In the absence of solvent
recycle flowrate information in the problem, an initial estimate of 200 kmol/hr
was made before beginning the simulation.
Results and discussions. The simulation was
performed after initializing the parameters to carry out the steady-state
operation. TABLE 1
shows the specifications for this process and the simulation results. The
stages are numbered in descending order, with the first stage at the top. The
reboiler and condenser are considered as stages in both the columns.
FIG. 4 illustrates the composition profile across the
extractive distillation column. An analysis of the curve indicates that the
composition of ethanol is highest at the top of the column: the lighter key
component with a mole fraction of 0.7355. Trace quantities (mole
fraction 0.0045) of p-xylene are also present in the distillate. Another
fractionating column can be used to further refine this distillate. The
bottom fraction contains very small amounts of ethanol. The majority is a
mixture of p-xylene (mol fraction 0.811) and benzene (mol fraction 0.189) which
is then sent to the solvent recovery column as feed.
The composition profile of the solvent recovery column is shown in FIG. 5. Based on the
fact that the column feed only had trace amounts of ethanol, both the
distillate and bottoms contain minimal amounts of ethanol. Benzene (mol
fraction 0.99) is distilled from the top of the column while p-xylene (mol
fraction 0.933) is taken from the bottom. The bottoms from this column are then
recycled back at the rate of 214.697 kmol/hr (calculated in the
simulation), where it combines with fresh solvent before being fed to the first
column.
Sensitivity analysis study. An analysis of the
system's sensitivity is performed by varying parameters of the process, such as
flowrate and mole fraction, to understand how the changes affect ethanol outlet
parameters over specified intervals.
Case
1: Variation of
ethanol production with the mass flowrate of the makeup stream. FIG. 6 shows the flowrate of ethanol production as a
function of the makeup stream flowrate. This flowrate varies between 15 kg/hr and
50 kg/hr without affecting any other parameters. From the figure, it is evident
that the highest production of ethanol occurs around 30 kg/hr of the makeup
stream. Increasing the makeup stream can increase the quantity of ethanol
produced only up to a certain point. The column will be more mass- and heat-laden
after that point, but no extra separation will occur.
Case
2: Variation of
the ethanol production rate and mole fraction with the change in solvent mole
fraction. A change in the concentration of p-xylene in the
solvent used for the first column can be utilized to examine the changes in two
parameters, namely mole fraction and production rate of ethanol. FIG. 7 shows the
effects of varying the solvent mole fraction of p-xylene from 0.85 to 0.99. As
the p-xylene concentration increases, the ethanol production rate decreases,
while the mole fraction of ethanol in the distillate increases. Ideal operation
of the column can be achieved by considering the intersection as an optimum
point between production and purity. The final operation will be determined by
the process requirements, and such tools will aid in optimizing the process.
Takeaway.
Azeotrope mixtures of ethanol
and benzene can be separated with p-xylene utilizing the extractive
distillation method. By using the UNIQUAC thermodynamic model, a steady-state
simulation was performed on DWSIM software using p-xylene as a solvent to
separate the ethanol-benzene mixture. The simulation demonstrated that ultrapure
benzene can be produced in two consecutive distillation columns and a majority
of the solvent (p-xylene) can be recovered and recycled back into the process. Using
sensitivity analysis, certain parameters can be modified to increase the flowrate
of components or enhance purity. System response can be observed by
changing other parameters such as reflux ratios, feed tray placement, feed
temperatures, feed pressures, reboiler duty, etc. Additionally, it may also be
possible to investigate the use of other solvents for better separation of the
components. HP
ACKNOWLEDGEMENT
The authors are grateful to Muhammad
Saif for being the inspiration for this work.10
NOMENCLATURE
LITERATURE CITED
KAUSTUBH LATURKAR works as an Engineer at the Facility for Rare Isotope
Beams, a U.S. Department of Energy (DOE) project in Michigan. He has more than
9 yr of experience working in the field of process engineering, refinery
operations, utility systems design and operation, with a special focus on design
and commissioning of engineering systems. Laturkar earned an MS degree in chemical
engineering from University of Florida and a BE degree in chemical engineering
from Panjab University, Chandigarh, India. The author can be reached at kos19188@gmail.com.
KASTURI LATURKAR works as a Validation Engineer for Validation Associates LLC and has more than 4 yr of experience working in commissioning, qualification and validation of upstream and downstream bioprocessing equipment and critical utilities. Laturkar graduated with an MS degree in chemical engineering from Syracuse University and a B.Tech degree in chemical engineering from Guru Gobind Singh Indraprastha University, Delhi, India. The author can be reached at kasturi.laturkar@gmail.com.