ASIA-PACIFIC JOURNAL OF CHEMICAL ENGINEERING

Asia-Pac. J. Chem. Eng. 2016; 11: 958972

Published online 5 September 2016 in Wiley Online Library
(wileyonlinelibrary.com) DOI: 10.1002/apj.2030

Research article
Side draw optimisation of a high-purity, multi-component
distillation column
Michael A. Kraller,1 Isuru A. Udugama,2 Robert Kirkpatrick,2 Wei Yu2* and Brent R. Young2
1
Chemical Engineering Department, Technical University of Munich, Munich, Germany
2
Chemical and Materials Engineering Department, The University of Auckland, Auckland, New Zealand

Received 23 February 2016; Revised 12 July 2016; Accepted 2 August 2016

ABSTRACT: Distillation columns with side draws are widely used in the process industry to rene crude methanol into
high-purity methanol. Historically, industrial methanol manufacturers have concentrated on meeting strict product
specications, while methanol yield and reboiler duty optimisation has often been neglected. In this work, a steady-state
model of an industrial high-purity multi-component methanol distillation column was developed using a commercial process
simulator. To achieve higher recovery at optimal reboiler duty, the side draw location and side draw mass ow rates were
identied as two important factors. It was determined that a combination of lowering the side draw location while reducing
the side draw mass ow rate will lead to the most optimal outcome. A design of experiment was also carried out to evaluate
the stability of the column at recovery rates of 97.7% (current recovery) and 99.5% (proposed high recovery). Disturbances
in the feed stream were identied as having the greatest effect, while the product purity was observed to become more
sensitive towards all disturbances when operating at an enhanced recovery. 2016 Curtin University of Technology and
John Wiley & Sons, Ltd.

KEYWORDS: high-purity distillation column; side draw optimisation; methanol distillation

INTRODUCTION and isobutanol is distilled into federal AA grade

The global demand for methanol has grown markedly
over the past decade, nearly doubling from 40 million
tons in 2004 to 70 million tons in 2013. This growth
is predicted to continue into the future with current
estimates for global demand in 2016 exceeding 90
million tons.[1] To satisfy both this growing global
demand and demand from new emerging markets,
methanol plants with capacities of up to 10 000 metric
tons per day (MTPD) are considered and planned.[2]
In order for these proposed plants to meet the future
demand for methanol while still increasing the prot
margins for methanol producers, they must operate at
higher methanol recovery rates and in a more energy
efcient manner than current plants. In this paper, we
will investigate applying recovery enhancement
optimisation to a high-purity methanol distillation
column with a side draw. The operation effects that
inherently come with this kind of optimisation, such as
increased disturbance sensitivity, will also be explored.
In industrial methanol production, crude methanol
consisting of methanol, water and trace levels of ethanol
*Correspondence to: Wei Yu, Chemical and Materials Engineering
Department, The University of Auckland, Auckland, New Zealand.
E-mail: w.yu@auckland.ac.nz
2016 Curtin University of Technology and John Wiley & Sons, Ltd.
Curtin University is a trademark of Curtin University of Technology
methanol, which should consist of <10 ppm of ethanol.
Moreover, the bottoms discharge from the column is
required to have <10 ppm of methanol to avoid costly
bio treatment. To separate this multi-component
mixture and to make high-purity products as well as
bottoms discharge, these columns employ a side stream.
In addition, the column is operated above a recovery
rate of 97.7% for more prot. In the oil and gas industry,
distillation operations can use up to 40% of total energy
consumption in a plant.[3] Given this, optimising the
columns energy consumption and product methanol
yield is of great economic importance for operators.
Optimising this distillation process is challenging
because the mixture is highly non-ideal and high
purities need to be achieved at both top and bottom of
the column. In the past, methanol product recovery
and the energy input into the distillation column have
often been overlooked in favour of meeting the strict
<10 ppm of product ethanol specication.
The subject of rening crude methanol into federal
AA grade specication (<10 ppm ethanol), ultra-high-
purity methanol is very scarcely discussed in published
literature. Zhang et al.[4] look at a crude methanol
rening and the feasibility of employing multiple
methanol rening columns to reduce the overall energy
demand of the process. However, they do not consider

Asia-Pacic Journal of Chemical Engineering
industrial and economic factors and do not investigate
distillation columns with side draws. Columns with a
side draw are more complex in design than simple
distillation columns, as they have more design variables
ascribed to additional degrees of freedom. These types
of columns were initially designed by hand; this meant
that the number of trays required and the side stream
location and ow rate may not be the optimal condition.
In 1985, Glinos and Malone[5] proposed equations and a
design algorithm for side stream columns, which
allowed accurate estimations for the fundamental
parameters of these type columns. These equations only
apply on ideal mixtures, as they are based on the classic
equations introduced by Underwood.[6] Bausa and
Watzdorf presented shortcut calculation methods for
non-ideal mixtures with up to three or four
components.[7,8] These methods were also capable of
being applied to side stream columns. Modern process
simulators allow a column design and optimisation with
much higher accuracy. Many methanol distillation
columns were built several decades ago, when the
distillation design methods were not as advanced as
today. Because of this fact, it is potentially benecial
to more closely examine existing columns with modern
engineering tools, allowing for both the product
methanol yield, column energy input and, hence, the
overall efciency of the column to be optimised.
Modelling industrial chemical processes allows for
the study and optimisation of existing, large-scale,
industrial processes without the need for eld trials. A
recent study successfully modelled the Fischer
Tropsch process to examine how a novel type of a
reactor can improve key operational parameters.[9] In
another study, a kinetic model of a methanol to olen
plant was developed and employed to optimise the
operating conditions.[10] In both studies, an appreciable
level of agreement between the computational models
and experimental data was achieved. Using process
models to optimise a columns operational parameters
is cost efcient and time saving. These models also
contribute to the general understanding of the process
and helps highlight areas of potential research interest.
Conversely, plant trials and small scale pilot plants, in
addition to being costly and time consuming, do not
allow process engineers to explore the entire range of
operating scenarios because of safety concerns and
other practical limitations.
A literature search on methanol distillation does,
however, turn up a handful of papers that investigate
the high-purity separation of a water/methanol binary
mixture. In the study of Luyben[11,12] and Sadgehi
and Ahangar,[13] the authors carry out process control
and optimisation work on this type binary methanol
distillation process using dynamic process simulations
as well as mathematical models. The optimisation of
distillation units in general has also been long studied.
In the study of Shahandeh et al.,[14] the authors carry
2016 Curtin University of Technology and John Wiley & Sons, Ltd.
SIDE DRAW OPTIMISATION OF A MULTI-COMPONENT
DISTILLATION COLUMN 959
out optimisation work on a distillation column design
on a benzenetoluene separation. In another study of
Shahandeh et al.,[15] the same authors carried out an
optimisation case study on a methanol/water binary
separation, where ve heat pump-assisted distillation
column schemes were rst optimised and then
compared with each other. In the study of Shahandeh
et al.,[14,15] the authors represented the column using
a mathematical model. In the study of You et al.,[16]
the authors modelled an extractive distillation process
of acetonemethanol with water using an industrial
process simulation and minimised the total energy
consumption. In the study of Palmeira et al.,[17] the
authors also use an industrial process simulation to
simulate and optimise a methanol-water-glycol system.
However, the authors in this instance carried out a
sensitivity analysis rather than a true optimisation study
in order to optimise the operations of the column. In the
study of Tian et al.,[18] the authors use an
experimentally backed mathematical model and plot
data in a viable operating range to nd the optimal feed
stage and reboiler duty usage. Despite being in the
general area of optimisation of distillation columns,
these works lack one/more of the following aspects:
The columns do not include a side draw.
The columns do not require a higher purity at the top
and bottom of the column.
The columns do not require high recovery rates.
The operation of each of the columns is not limited
by industrial and economic considerations.
In contrast to the columns studied in these previous
works, the industrial methanol columns studied in this
paper have multi-component feeds (consisting of
methanol, water and ethanol), include a side draw and
need to produce product methanol with <10 ppm of
ethanol. On top of all this, the results proposed in this
work are easily implementable in an industrial setting.
In this paper, we have modelled a real life methanol
distillation column in Aspen HYSYS to optimise the
methanol recovery and energy efciency of the
process. Employing this model, we also carried out a
design of experiment, to analyse the stability of the
model and its sensitivity to disturbances. Through this,
conclusions on the feasibility of the increase in
recovery were made.
This paper is organised as follows. In the section
immediately following this, the basic operating
principles of columns with side draws are discussed,
and key concepts are dened. In the next section,
industrial distillation processes are described, allowing
for the input parameters for the process simulator to be
specied. Following this, a qualitative analysis of
efcient yield enhancement and the optimisation of an
industrial distillation column are performed. Next, a
description and evaluation of the experimental
Asia-Pac. J. Chem. Eng. 2016; 11: 958972
DOI: 10.1002/apj
960 M. A. KRALLER ET AL.
procedure is given. Subsequent to this, the stability and
feasibility of operating the column at increased
recovery rates is discussed. Finally, a summary of the
most important results found concludes this paper.
DISTILLATION COLUMNS WITH SIDE DRAWS
Distillation columns with side draws can be used to
separate any nonazeotropic multicomponent mixture
into three (or more) product streams. The high energy
input for the primary (more difcult) separation is also
used to carry out the secondary separation in addition.
The design of side stream distillation columns was
analysed in detail by Glinos and Malone.[5]
A columns side stream can be placed either above or
below the feed point location; on a similar note, the
side stream can be drawn as either a liquid or a vapour.
Within this work, the side stream is located below the
feed tray and is drawn as a liquid. The side stream
location was chosen on the basis that a high fractional
recovery and high purity of the light key component
need to be achieved in the column. The quality of the
side stream was chosen to lower the vapour rate and,
therefore, lower the operational costs of the column.
A nonazeotropic quaternary mixture with the
components (X1/X2/X3/X4) is considered. The boiling
point increases from X1 to X4.
Figure 1 shows that a column with a side stream is
effectively divided into three sections. In Introduction
and Distillation Columns with Side Draws sections,
Figure 1. Different sections of a distillation column with
side draw.
2016 Curtin University of Technology and John Wiley & Sons, Ltd.
Asia-Pacic Journal of Chemical Engineering
the primary separation between X1 and X2/X3/X4 takes
place, and in Model Description section, the secondary
separation between X2 and X3/X4. Generally, the
column can be considered as two separate columns
with two denable boil-up ratios for each hypothetical
column. The boil-up ratio at the top of the column for
the primary separation is given by Equation [1]:
M_ W
; (1)
M_ B M_ S
and the boil-up ratio at the bottom of the column for the
secondary separation is give by Equation [2]:
M_ V
: (2)
M_ B
In reality, the separations within the column are
never as sharp as what is assumed herethere are
always impurities in each of the different streams.
Degrees of freedom
To understand the difculties of designing and
optimising side stream distillation columns, it is
important to analyse the degrees of freedom (DOF).
For a simple distillation column with one feed, no side
draws and the other column parameters (number of
stages, the feed plate location, etc.) remaining xed,
there are usually ve DOF that need to be controlled.
Because the pressure and holdups in the condenser
and reux drum are assumed to be constant at steady-
state operation, only two DOF need to be controlled
for a simple column.[19] The presence of a side draw
means an additional DOF needs to adjusted, making
the design and optimisation of the column considerably
more complicated. In total, there are three DOF that
need to be controlled with a single side draw column.
These are the following:
reboiler duty;
distillate ow rate; and
side draw ow rate.
The remaining ow rate (bottoms) is determined by
an overall mass balance around the entire column.
The three DOF are used to adjust the following column
outputs:
ethanol mass fraction at the top;
methanol mass fraction at the bottom; and
recovery.
Optimise energy input
As for simple distillation columns, the energy input
of a side stream column is a function of the number
Asia-Pac. J. Chem. Eng. 2016; 11: 958972
DOI: 10.1002/apj

Asia-Pacic Journal of Chemical Engineering
of trays of the distillation column. The optimum
operating point is always a compromise between
investment costs and operational costs. To minimise
the required energy input for a column, the ideal
feed plate location needs to be found, as an improper
feed location can reduce the performance of the
column. This ideal location is found where the
composition of the feed is identical to the
composition of the internal ows at that point within
the column. Thus, the composition gradient between
feed stream and liquid in the feed tray is
minimised.[20] While there are accepted and
commonly used engineering practices for estimating
the optimum amount of trays and ideal feed tray
for simple columns as specied in the study of
Kister,[21] the inuence of the side draw on the
energy input is not explored as well.
The right placement of the side draw is of major
importance, as it denes the tray ratio between the
primary and secondary separation. Also of great
interest is the mass ow rate of the side stream, as this
affects the boil-up ratio of the primary separation
(Equation [2]) and, therefore, the relationship between
both boil-up ratios. If the primary and secondary
separation sections of the column are not balanced
accordingly by varying the side stream location and
ow rate, the process will operate at a sub-optimal
efciency.
Increasing the recovery
The protability of increasing a columns recovery is
dependent on the additional capital and operating costs
required to increase recovery and the market value of
the produced product. The nancial benet from
selling the extra product has to outweigh the additional
nancial costs.
For a simple distillation column, there are two
possible ways to enhance the degree of separation
and, hence, the recovery: by increasing the energy
input or by mounting additional trays into the column.
This becomes more complicated for side draw
columns, as the mass ow and the location of the side
stream inuence the interaction of the primary and
secondary separation and, therefore, must be
determined for every operation point separately. For a
higher yield, the primary separation needs to be
enhanced accordingly.
For our situation, however, the recovery of the
column cannot be increased at will. This is because
any marginal increase in recovery leads to severe
sensitivity to disturbances. This reects the fact
that, at high recoveries, the column is being
operated at a point close to the theoretical
limitations of separation, where almost 100% of
the methanol is extracted as products at ultra-high-
purity specications.
2016 Curtin University of Technology and John Wiley & Sons, Ltd.
SIDE DRAW OPTIMISATION OF A MULTI-COMPONENT
DISTILLATION COLUMN 961
MODEL DESCRIPTION
For the analysis and optimisation of the side draw of
the high-purity distillation column, the process
simulator Aspen HYSYS was used. The basis for the
steady-state model is the real life distillation unit
employed by an industrial methanol producer. The feed
enters at the middle of the column and consists of a 4:1
methanol to water mass ratio. The feed contains
150 ppm of ethanol; this must be reduced down to
<10 ppm in the nal methanol product. A complete
description of the composition of each component in
the feed is given Table 1. Failure to meet this
specication will result in substandard products that
cannot be sold to customers. In addition to the strict
product ethanol specication, the bottoms methanol
concentration also needs to be monitored and managed
below 10 ppm. Failure to meet this limit will require
costly bio-treatment of the bottoms stream prior to
discharge. Unlike the product and bottoms streams,
the fusel side draw consists of a methanol, ethanol
and water mixture and allows for dual high-purity
specications. To maximise the protability, it is
important to recover as much of the methanol entering
the column as products. In general, the ratio of
methanol in the feed to methanol in the product must
be at a high level (97.5%). Failing to meet this
recovery will also result in and off specication
bottoms methanol specication. Figure 2 shows the
rening column as implemented in the process
simulator Aspen HYSYS. The trays are numbered from
top to bottom, where the condenser is the rst stage and
the reboiler the last. All together, the column has 87
stages.
The key components in the Feed F are methanol,
ethanol, water and isobutanol. All other components
in the real feed are not considered, because the
concentrations are negligibly low. The feed stream
enters the column at stage 53 with the composition
shown in Table 1. The primary separation, which takes
place in the upper section of the column, is between
methanol and ethanol, as dened in Distillation
Columns with Side Draws section. In the lower part
of the column, the secondary separation between
ethanol and the remaining components is carried out.
The high-purity methanol, which is drawn at the top
(Distillate D), must not contain more than 10 ppm
Table 1. Mass fractions in feed.
Symbol Component Mass fraction (kg/kg)
X1 Methanol 0.83
X2 Ethanol 150 ppm
X3 Water 0.17
X4 Isobutanol 95 ppm
Asia-Pac. J. Chem. Eng. 2016; 11: 958972
DOI: 10.1002/apj
962 M. A. KRALLER ET AL.
Figure 2. Distillation column as implemented in Aspen Plus.
ethanol to reach the desired product specications. For
control purposes and to ensure this specication is
met, the industrial distillation units are operated at a
distillate ethanol concentration of 7 ppm. For the
removal of ethanol, a Side Draw S is installed on stage
73. The Bottoms Product B is mostly water. It is
important that the methanol fraction of the Bottoms
Product B does not exceed 10 ppm, because this would
cause the need for expensive waste water treatment,
which requires access to facilities that many methanol
producers do not even have within their plant. The mass
ow rates of the Feed F and the product streams are
listed in Table 2.
Modelling
An initial process model was created on Aspen HYSYS
by the authors in previous work[22] and was used as a
starting point. The model was rst adjusted by choosing
the right property package and Murphree tray
efciencies to achieve the best match between the
simulation and data provided by industry. The Wilson
Table 2. Mass ow rates for column operating at
97.7% recovery.
Stream Mass ow in kg/h
Feed F 142 000
Distillate D 114 600
Side draw S 3113
Bottoms B 24 287
2016 Curtin University of Technology and John Wiley & Sons, Ltd.
Asia-Pacic Journal of Chemical Engineering
VLE property package with the Virial equation of state
and the Poynting correction was used to model the
distillation unit. The Wilsons equation uses two
adjustable parameters that can be set depending on the
chemical characteristics of the mixture.[23] The Wilsons
equation is well suited to model methanol distillation, as
the mixture has a single liquid and gas phase. The Viral
equation of state and the Poynting correction factor was
employed to modify the gas phase calculations to better
reect non-idealities present in the separation.[23]
The binary interaction parameters for the Wilsons
equation can be taken from vapourliquid equilibrium
(VLE) data bases, which, in turn, are based on physical
experimental data. However, when modelling multi-
component mixtures, these experimental data, which
are determined by carrying out experiments on two
components at a time, can be incorrect. This is because
the interaction parameters do not consider the
interactions among all of the components when they
are all present in a single multi-component mixture.[24]
In industrial process simulators, the users have the
option to choose between the pure binary interaction
parameters (as determined by experimental data) or to
use a semi-empirical (UNIFAC) interaction parameters
that attempts to correct for interaction when in a
mixture.[24] As expected, the simulation with the semi-
empirical (UNIFAC) interaction parameters yielded
satisfactory results, while the pure binary interaction
parameters provided a poor data match with the
industrial data.
The semi-empirical interaction parameters, on the
other hand, were able to match the ethanol, water and
methanol proles with industrial data provided. The
isobutanol prole of the simulated column, however,
did not match well with the industrial data provided.
In order to achieve a better match with the isobutanol
prole, we have changed the interaction parameter
between isobutanol and water. It is also important to
note that this change did not have any noticeable effect
on the other concentration proles. The Murphree
efciencies are assumed to be equal for all the trays in
the column and were iteratively determined to be
84.0%. The t between the provided industrial data
and the computational model is shown in Table 3.
INCREASING THE METHANOL YIELD
Because the methanol synthesis is highly energy
and cost intensive, it is of major interest to minimise
the amount of methanol lost during the purication
process. As shown in Table 4, there is still a large
amount of methanol being drawn from the side stream
(2639 kg/h) for the current operation point of 97.7%
recovery.
In this section, we will investigate the possibility of
increasing the methanol yield (methanol recovery) of
Asia-Pac. J. Chem. Eng. 2016; 11: 958972
DOI: 10.1002/apj
 SIDE DRAW OPTIMISATION OF A MULTI-COMPONENT
Asia-Pacic Journal of Chemical Engineering DISTILLATION COLUMN 963
Table 3. Deviation of the Aspen Plus model from the data provided by industry.

Product Data Model Percentage error

Methanol kg/kg 0.9999 0.9999 0
Ethanol ppm 7.00 7.00 0
Isobutanol ppm 0 0
Water ppm 85 50 41.2
Mass ow kg/h 114 600 114 598 0.01
Side draw
Methanol kg/kg 0.8554 0.8478 0.89
Ethanol kg/kg 0.0066 0.00658 0.25
Isobutanol ppm 47 55.8 18.7
Water kg/kg 0.138 0.141 2.40
Mass ow kg/h 3113 3113 0.00
Bottoms
Methanol ppm 7.00 6.96 0.60
Ethanol ppm 7.00 4 48.26
Isobutanol ppm 500 500 0
Water kg/kg 0.9999 0.9999 0.01
Mass ow kg/h 24287 24264 0.09
Energy
Reboiler MW 87.2 86.3 1.04
Condenser MW 87.7 87.0 0.80

Table 4. Methanol mass ow rates for column
operating at 97.7% recovery.
Stream Methanol ow in kg/h
Feed F 117 236
Distillate D 114 597
Side draw S 2639
Bottoms B <1
the column. From an economic point of view, higher
methanol recovery will result in higher revenue.
However, from a plant stability point of view,
operating the column at high recoveries will result in
an increased sensitivity to disturbances (as discussed
in Design of Experiment section). Based on these two
facts and input from plant management, we have
decided to investigate the possibility of operating the
column at a recovery ratio of 99.5%. This recovery rate
represents a relatively high, yet realistic/achievable
rate. To increase the methanol yield of the column,
the separation capacity has to be increased. For a
simple distillation column without a side draw, this
means that either the reboiler duty must be increased
or additional trays must be introduced into the column.
For a distillation column with a side draw, it is also
necessary to balance the primary and secondary
separation to achieve optimal yield at a new operation
point. There are two possible ways:
i. By reducing mass ow of the side draw
By lowering the mass ow rate of the side draw to
733 kg/h, the amount of methanol that reaches the
2016 Curtin University of Technology and John Wiley & Sons, Ltd.
top of the column is increased. This is a
consequence of the fact that the degree of primary
separation is inversely proportional to the side draw
mass ow rate, as predicted by Equation [1]. With
additional methanol to the top of the column, the
distillate ow rate is increased by 2073 kg/h (a
1.8% increase). This increased ow rate was able
to be maintained while still achieving a distillate
product specication of 7 ppm of ethanol. A list of
the key process parameters for the column operating
at a standard recovery of 97.7% and the column
operating at the highest achievable recovery rate of
99.5% is shown in Table 5. It should also be noted
that reducing the mass ow rate of the side draw
decreased the amount of methanol lost via the side
draw by 77.8%
ii. Lowering side draw location
By lowering the side draw location, the number of
trays provided for the primary separation increases,
while the number of trays provided for the
secondary separation decreases. Although
increasing the number of trays for the primary
separation favours the purication of methanol
within the column, a greater amount of impurities
are still observed at the bottom of the column. This
is illustrated in Fig. 3, where both the methanol and
ethanol proles are shifted down in the column as a
result of lowering the side draw position. In order to
compensate for the reduced secondary separation,
the reboiler duty of the column must be increased.
It was found that by lowering the side draw to stage
81 and keeping the mass ow rate of the side draw
Asia-Pac. J. Chem. Eng. 2016; 11: 958972
DOI: 10.1002/apj
964 M. A. KRALLER ET AL. Asia-Pacic Journal of Chemical Engineering
Table 5. Process parameters for 99.5% recovery rate. varied. This will allow for the column to operate at
the most efcient point possible for a given
By reducing By lowering
mass ow of side draw recovery rate.
Product the side draw location
Methanol kg/kg 0.9999 0.9999
Ethanol ppm 7 7 OPTIMISATION OF THE SIDE DRAW COLUMN
Isobutanol ppm 0 0
Water ppm 75 70 A previous study conducted by Udugama et al.[22] has
Mass ow kg/h 116 671 116 671 successfully shown that the side draw location can be
Side draw optimised to reduce the required energy input for a
Methanol kg/kg 0.79 0.275 certain yield. This study did not vary, however, the
Ethanol kg/kg 0.0277 0.0066
Isobutanol ppm 80 55.8 mass ow rate of the side drawn or the number of
Water kg/kg 0.181 0.725 stages in the column.
Mass ow kg/h 733 3113 To nd the most efcient operation point for a given
Bottoms recovery, both the side draw mass ow rate and the
Methanol ppm 7.2 6.9
Ethanol ppm 0.2 0 side draw location must be simultaneously optimised,
Isobutanol ppm 558 500 resulting in the lowest possible required energy input
Water kg/kg 0.9999 0.9999 being achieved. As previously explained in Distillation
Mass ow kg/h 24 586 22 216 Columns with Side Draws section, this optimisation
Reboiler MW 88.5 88.2 process is inherent complicated as a result of the fact
that changing either one of these variables results in a
corresponding change in numerous other variables.
In the following section, the mass ow rate and
location of the side stream will be optimised for both
the actual operation point at 97.7% recovery and for
an enhanced recovery of 99.5%.

Optimum side stream location

Before optimising the side draw parameters, the

Figure 3. Methanol and ethanol concentration proles in
the column with different side draw positions.
unchanged, a reboiler duty of 88.1 MW was
sufcient to achieve a recovery rate of 99.5% as
shown in Table 5. Compared with changing the
mass ow rate of the side draw, the relative impact
of changing the side draw location is greater than
that of changing the side draw mass ow rate on
the balance between the primary and secondary
separations.In order to change the recovery of the
column while still achieving on-specication
products, the correct balance between the primary
and secondary separations must be maintained.
Achieving this perfect balance between the two
separations requires both the mass ow rate and
position of the side draw to be simultaneously
2016 Curtin University of Technology and John Wiley & Sons, Ltd.
dependence of the reboiler duty on the side draw
location needed to be determined. Although the effect
of the side draw location on the required reboiler duty
was the key relationship being observed here, the mass
ow rate of the side draw, which needs to be at a
certain level to maintain the desired recovery, was also
monitored to ensure that it did not exceed certain limits.
Figure 4 shows that, by lowering the side draw (the
feed plate location remained unchanged), the reboiler
duty can be reduced to a minima point. Unfortunately,
by reducing the side draw location, the side draw mass
ow rate is also increased. This is because the
secondary separation is handicapped by reducing the
height of the bottom section of the column. Based on
plant experience, a side draw mass ow rate between
3000 and 5000 kg/h can be tolerated; a ow rate greater
Figure 4. Reboiler duties for different side draw locations
for an operation with 99.5% recovery.
Asia-Pac. J. Chem. Eng. 2016; 11: 958972
DOI: 10.1002/apj

Asia-Pacic Journal of Chemical Engineering
than this range can never be tolerated, however. This
acceptable range is coloured red in Fig. 4.
For a recovery of 99.5%, the minimum reboiler duty
is achieved with a side draw at stage 79 (Fig. 4). Given
that the mass ow rate of the side draw at this stage is
1100 kg/h and, thus, does not fall above the acceptable
side draw mass ow rate range, this ow rate is not of
concern. The optimum feed tray and the required
reboiler duty were found to be stage 54 and 87.8 MW,
respectively. This represents an energy saving of
0.5% due to side draw optimisation. Compared with a
column with an optimised side stream operating at a
recovery of 97.7%, a column with an optimised side
stream operating at a recovery of 99.5% demands
2.3% more reboiler duty.
Flooding factor
As the reboiler rate increases to increase recovery, the
vapour trafc in the column increases. Eventually, the
column reaches a point where the liquid can no longer
ow down the column properly because of the higher
vapour ow and tray pressure drop. This leads to liquid
accumulating within the column. This situation is
known as ooding. The ooding factor (f-factor)
numerically quanties the likelihood of a single tray
to ood. In order to calculate the f-factor of each tray,
the speed of the upward travelling vapour in the
column must be determined.[25] This was able to be
determined in Aspen HYSYS by providing the process
simulator with the sizing of the trays used in an actual
methanol plant. To achieve a higher recovery, an
increase in reboiler duty is necessary. By having the
reboiler produce more steam and by increasing the
column trafc while the dimensions of the column
remain unchanged, the column operates closer to its
ooding point, as illustrated in Fig. 5. Although the
optimisation of the side stream eases this ooding
issue, the resulting maximum f-factor is still
 0:5
kg
0:999ms 3 , which is 2% higher than that at
m
 0:5
kg
standard operation 0:98ms 3 . However, a slight
m
change in the operating pressure of the column can
counteract this increase in f-factor.[22] In this case,
Figure 5. F-factor prole of the column at different
operation points.
2016 Curtin University of Technology and John Wiley & Sons, Ltd.
SIDE DRAW OPTIMISATION OF A MULTI-COMPONENT
DISTILLATION COLUMN 965
increasing the top pressure by 100 mbar remedies this
increase in the f-factor. Consequently, the effect of
optimising the side draw parameters on the columns
f-factor can be neglected.
Increase number of trays
If ultra-high levels of product recovery are desired at
the design phase of a production facility, the plant
management has the option of installing additional
trays in the main rening column. Figure 6 illustrates
how the required reboiler duty varies with the number
of trays in the column. The feed and side draw location
for each point has been optimised. The red standard
line indicates the reboiler duty of the current operation
point with a recovery of 97.7%. Looking at Fig. 6, it is
apparent that by operating the column at 99.5%
recovery and leaving the reboiler duty unchanged, 27
extra trays are required, resulting in a column with
115 stages. To determine the most economically
optimal number of trays, a nancial optimisation
process that considers both the increased operational
costs associated with increasing reboiler duty and the
increased capital costs associated with installing
additional trays must be performed.
DESIGN OF EXPERIMENT
It is an established concept that, for distillation, columns
are more sensitive to process disturbances at increased
levels of product recovery. However, operating a
column at ultra-high rates of recovery rates (99.5%)
might illustrate interesting observations in comparison
with operating the column at high rates of recovery
(97.5%). The objective of this section is to use the
steady-state model to quantify the increased sensitivity
of the column as a result of increasing recovery. The
implications of this increased sensitivity on
implementing ultra-high-purity distillation operations
in an industrial setting will also be explored. To
examine the steady-state models performance and
stability at an increased recovery (compared with the
current operation), a Design of Experiment (DOE)
was executed. The way in which the steady-state model
Figure 6. Reboiler duties with operation at 99.5% recovery
for different column heights.
Asia-Pac. J. Chem. Eng. 2016; 11: 958972
DOI: 10.1002/apj
966 M. A. KRALLER ET AL.
reacted to certain input disturbances and, by extension,
the process variables that most greatly inuenced this
transient response, was investigated.
Experimental set-up and specications
Within the DOE method, a higher recovery is achieved
without changing the distillation set-up by adjusting the
mass ow rate of the distillate and the side draw, as
well as by varying the reboiler duty. For the analysis,
changes in feed ow M_ F , reboiler duty Q_ Reb, methanol
XF,Meth and ethanol fraction XF,Eth in the feed were
considered. Since the experiment is run at two levels
of disturbance (high and low), this results in a 24
factorial design.
The values for the disturbances are based on the
experience of the real life plant and are as follows:
2.5% in feed ow M_ F
5% in reboiler duty Q_ Reb
1% in methanol fraction XF,Meth of the feed
50% in ethanol fraction XF,Eth of the feed.
For the execution of the DOE, the distillate and side
draw mass ow rates are kept constant. As a result, a
change in feed ow rate only affects the bottoms ow
rate. To balance the changes in XF,Meth and XF,Eth for
the different runs, the water content of the feed XF,
H2O was adjusted accordingly.
Table 6 gives the high and low levels of the input
variables for the DOEs. For both DOEs, the feed ow
and composition are the same, but the reboiler duty is
higher in the case of the column running at a higher
recovery. This is because a greater amount of reux is
required to achieve the desired separation.
The response variables for this setup were ethanol
XD,Eth and methanol XD,Meth concentrations in the
distillate, as well as the methanol concentration in the
bottoms XB,Meth and the recovery Rec. The values of
these variables were monitored during the disturbance
tests. The response variables for the two different
modes of operation without any disturbance are shown
in Table 7.
Table 6. Values of the input variables for both points
of operation.
Main effect Standard + 
A 141 976 145 525 138 426
M_ F
B [97.7] % 86.20 90.50 81.89
Q_ Reb
B [99.5] % 88.53 92.95 84.10
Q_ Reb
C XF,Meth 0.8258 0.8340 0.8175
D XF,Eth 1.50E04 2.25E04 7.50E05
2016 Curtin University of Technology and John Wiley & Sons, Ltd.
Asia-Pacic Journal of Chemical Engineering
Table 7. Response variables for standard and
increased recovery without disturbance.
a XD,Eth 7.00E06 7.00E06
b XB,Meth 7.26E06 6.99E06
c XD,Meth 0.999993 0.999993
d Rec 97.7 99.5
The results of the 24 factorial design for operation at
97.7% recovery and for operation at 99.5% recovery
are shown in Tables A1 and A2, respectively, in the
Appendix. Once the changes in the response variables
to different process disturbances were determined, the
effects e on the response variables could be calculated.
The effects are shown in Table A3 (for a recovery of
97.7%) and in Table A4 (for a recovery of 99.5%) in
the Appendix.
Analysis and acceptance of the results
To judge the signicance of the different disturbances,
Q-Q plots were generated in MATLAB for the two
operation points. The plots can be found in Figs. 7
and 8. Insignicant effects almost behave like a random
sample drawn from a normal distribution with zero
mean, which implies the plotted effects will lie
approximately on a straight line. The effects that
deviate from the line are those of concern.
It can be seen in Fig. 7 that the ethanol concentration
XD,Eth is barely affected by the disturbances for a
recovery of 97.7%, as the plotted effects almost form
a straight line. The other three response variables XB,
Meth, XD,Meth and Rec seem to be affected most by the
feed ow (A), followed by the methanol concentration
in the feed (C). The interaction (AC) of both effects is
also distinctive.
Figure 8 shows that, for a recovery of 99.5%, the
ethanol concentration XD,Eth is affected by disturbances
in the feed ow (A) and the ethanol concentration in the
feed (D), where at the effect of the feed ow dominates.
The effect of the interaction (AD) is visible, too. XB,Meth,
XD,Meth and Rec are, again, disturbed severely by the
feed ow (A) and the methanol concentration in the
feed (C), with the interaction (AC) of both.
The Q-Q plots demonstrate clearly that, for an
operation at increased recovery, the variable XD,Eth
develops a distinct sensitivity towards changes in feed
ow and the ethanol concentration in the feed. The
disturbances in feed ow become perceptible, because
at 99.5% recovery, the column is operating very close
to the absolute methanol recovery. As the methanol feed
ow rate drops (while keeping the distillate ow rate
and reboiler duty constant), a larger fraction of the
methanol entering the column top of the tower in the
distillate. As the methanol feed ow rate drops (while
keeping the distillate ow rate and reboiler duty
Asia-Pac. J. Chem. Eng. 2016; 11: 958972
DOI: 10.1002/apj
 SIDE DRAW OPTIMISATION OF A MULTI-COMPONENT
Asia-Pacic Journal of Chemical Engineering DISTILLATION COLUMN 967

Figure 7. Q-Q plots of the effect for 97.7%.

Figure 8. Q-Q plots of the effect for 99.5%.

2016 Curtin University of Technology and John Wiley & Sons, Ltd. Asia-Pac. J. Chem. Eng. 2016; 11: 958972
DOI: 10.1002/apj
968 M. A. KRALLER ET AL.
constant), a larger fraction of the methanol entering the
column will be drawn at the top. This can go above
100% recovery. In such a situation, the methanol
entering the column via the feed is, temporarily, lower
than that leaving the column via the distillate. Thus,
the impurities in the product will increase severely, as
the lack of methanol needs to be balanced with other
components.
The larger the fraction of methanol that is drawn at
the top of the column as distillate and, therefore, the
greater the sensitivity of the column to even small
disturbances (which can be sufcient to cause a
methanol decit). An under-supply of methanol can
force all of the ethanol in the column (as there is only
approximately 10 kg/h in the feed) and even water into
the distillate. At this point, XD,Eth becomes more
sensitive for changes in the ethanol content of the feed,
as the changes are passed on to the distillate
composition to the full extent.
For further analysis, all of the effects of the other
three variables are displayed in the bar chart in Fig. 9,
to allow a convenient comparison between the two
operation points for these variables.
The responses of the variables XB,Meth, XD,Meth and
Rec resemble one another for the two different
operation points, as already observed in the Q-Q plots.
All three show the highest sensitivity towards changes
in the feed ow M_ F , followed by disturbances in the
Figure 9. Effects of the different disturbances on the
response variables.
2016 Curtin University of Technology and John Wiley & Sons, Ltd.
Asia-Pacic Journal of Chemical Engineering
methanol content XF,Meth and the interaction of both
disturbances. All the remaining effects on XB,Meth, XD,
Meth and Rec are insignicantly small. At a higher
recovery, the product purity seems to be affected more
by an increase of the feed ow rate or methanol fraction
of the feed. The column is operating at a higher boil-up
ratio, and the lower side draw ow rate at increased
recovery favours the primary separation more.
Therefore, the column is able to accumulate a larger
fraction of the additional methanol in the distillate. As
a consequence, the recovery is affected less.
To understand the directions of the effects, the
process must be analysed more closely. With an
increase in feed ow, more methanol enters the column
along with the other components. As more methanol is
fed to the column (while keeping the distillate ow rate
constant), the product purity improves. But, with the
reboiler duty being steady and the distillate ow being
limited, most of the excess methanol is drawn at the
bottom. This leads to a higher methanol concentration
at the bottom and a lower methanol recovery.
Figure 9 shows that the effects on the bottoms
methanol composition are fairly high for both
recoveries. Changes in the methanol content of the feed
have a similar impact. When the methanol content in
the feed increases, the methanol purity of the distillate
improves because the separation becomes easier.
However, the constant distillate rate means the product
recovery would be lower while the bottoms purity will
also be worse.
The main observation made for both of the recovery
cases is that both M_ F and XF,Meth have a noticeable
effect on the column performance. M_ F and XF,Meth
affect the columns performance by changing the
amount of methanol entering the column.
Consequently, achieving stable on-specication
operations requires the distillate and bottoms ow rates
to be changed such that a feasible mass balance is
reached. This effect is more apparent in the 99.5%
recovery scenario, where the effect on the product
methanol is higher for these two disturbances. A similar
observation is reported in the study of Fruehauf and
Mahoney,[26] where the author concludes that from a
control point of view, having the correct feed split (mass
balance) is a key aspect in achieving on-specication
operations when operating a column at ultra-high
recovery at dual high purity.
COLUMN SENSITIVITY TOWARDS
DISTURBANCES AT INCREASED RECOVERY
Operational stability
In order to evaluate a distillation columns stability
towards disturbances, short-term changes have to be
Asia-Pac. J. Chem. Eng. 2016; 11: 958972
DOI: 10.1002/apj

Asia-Pacic Journal of Chemical Engineering
distinguished from long-term changes. Short-term
disturbances occur suddenly and unexpected, while
long-term variations are usually known in advance or
happen over a longer period of time. Examples of said
long-term changes include intended changes to the
methanol production or the bit by bit loss of activity
of the catalyst in the reactor, both affecting the feed
of the distillation column. Because these disturbances
are known and predictable, the parameters of the
distillation can be pre-emptively adjusted to account
for these disturbances. For this reason, these kinds of
changes are of little concern.
It becomes apparent that, for a higher recovery, the
product quality is affected more by disturbances, as
the risk of disturbances causing a lack of methanol
for a certain distillate stream increases. Additionally,
the ethanol prole along the column sharpens for
higher recoveries, as the ethanol content in the side
draw increases. This affects the controllability and
stability of the column heavily, as only a small upward
or downward shift of the prole has a major impact on
the performance of the column.
Nevertheless, the effects of the disturbances
experiences by real columns are not as severe as those
observed in the steady-state model. The uctuations of
concern are mostly of short duration, and it, therefore,
partly buffered by the column itself, which is
essentially a large mass and energy storage unit.
Control
In order to achieve an increase in recovery, the effects
of the disturbances with the biggest inuence must be
reduced. As identied in the DOE, the product
composition has the highest sensitivity to changes in
the feed ow. To improve the performance of the
column, a feed forward control can be applied. By
measuring the actual feed ow and adjusting the
column distillate ow rate, side draw ow rate and
reboiler duty accordingly, more stability and a more
consistent product quality should be achieved as a
result of nullifying the disturbance. The inertia of the
column towards interruptions buys additional response
time. The controller must prevent the column from
experiencing a methanol decit while the distillate ow
rate remains unchanged. Such a situation would result
in the concentration of impurities in the distillate
product greatly exceeding product specication. Feed
forward control is also applicable for disturbances in
feed composition, as this is the disturbance with the
second highest effect.
The DOE shows that for both operation points, a
signicant effect of the disturbances on the bottoms
composition. Naturally, the ethanol concentration at
the top of the column is of the greatest interest, as this
determines the product quality and, hence, the market
value of the nal product. With that being said,
2016 Curtin University of Technology and John Wiley & Sons, Ltd.
SIDE DRAW OPTIMISATION OF A MULTI-COMPONENT
DISTILLATION COLUMN 969
however, it is also of interest to ensure that the bottoms
composition does not contain excessive amounts of
methanol such that costly waste water treatments are
required. The effect of disturbances on the methanol
concentration at the bottom is very distinctive, which
makes a dual composition control highly advisable
for both operation points.
For an operation at 99.5% recovery, the column is
operating close to the theoretical limit of separation,
as determined by a mass balance around the entire
column. In any case, the difference between the
methanol mass ow entering the column and the
distillate mass ow is still higher than 500 kg/h.
Although this difference is not that great relative to
the total mass ow rate of the feed, it does leave
enough room for the feed forward controller to react
to disturbances. Looking at an enhanced recovery of
over 2000 kg/h of methanol, the higher control effort
is denitely worth considering.
CONCLUSION
Feasibility of increasing the product recovery rate of a
high-purity methanol distillation column was studied
by employing a validated steady-state process
simulation. In initial studies, it was established that
location and the mass of the side draw are two critically
important factors in achieving high product recovery.
With the steady-state model, the side draw location
and mass ow have been optimised to achieve a
minimum energy input for columns operating at a
standard recovery rate of 97.7%; these location and
side draw adjustments represent an energy saving of
0.5%. If a methanol production facility is in the design
phase, an addition of 27 extra trays would also result in
an enhanced recovery of 99.5% without the need to
increase reboiler duty. Deciding between these two
options requires comparing the relative energy and
investment costs associated with each option. This
nancial optimisation process will need to be
conducted on a case by case basis for each column. It
is often the case that a combination of both solutions
is the most protable course of action. Finally, the
design of experiment showed that the feed ow,
followed by the feed composition, has the highest
impact on the essential column output parameters. It
was also concluded that the effects of disturbances
become more signicant at higher rates of recovery.
NOMENCLATURE:
B Bottoms Product (-)
D Distillate (-)
E Effect (-)
F Feed (-)
Mass ow rate of the bottoms
M_ B (kg/h)
Asia-Pac. J. Chem. Eng. 2016; 11: 958972
DOI: 10.1002/apj
970 M. A. KRALLER ET AL. Asia-Pacic Journal of Chemical Engineering

Mass ow rate of the distillate Rec Recovery (%)

M_ D (kg/h) S Side Draw (-)
Mass ow rate of the feed (kg/h) V Boil-up column (-)
M_ F W Boil-up primary separation (-)
Methanol Mass ow rate in the XB,Meth Methanol mass fraction at the
M_ F;Meth feed (kg/h) bottom (-)
Mass ow rate of the side draw XD,Eth Ethanol mass fraction at the top
M_ S (kg/h) (-)
Methanol Mass ow rate in the XD,Meth Methanol mass fraction at the
M_ S;Meth feed (kg/h) top (-)
Boil-up rate primary separation XF,Eth Ethanol mass fraction in the
M_ V (kg/h) feed (-)
Boil-up rate seconcary XF,H2O Water mass fraction in the feed
M_ W separation (kg = h) (-)
Q_ Reb Reboiler duty (MW) XF,Meth Methanol mass fraction in the
R Reux Ratio (-) feed (-)

APPENDIX

Table A1. Results of the 24 factorial design for 97.7% recovery.

A B C D a b c d
1     9.06E05 3.61E24 0.987967 100.0
2 +    8.75E06 6.15E02 0.999991 96.3
3  +   9.06E05 1.35E24 0.987967 100.0
4   +  9.39E06 1.26E10 0.999991 99.2
5    + 2.72E04 3.62E24 0.987967 100.0
6 + +   1.33E06 6.01E02 0.999999 96.3
7 +  +  7.08E06 1.46E01 0.999993 94.4
8 +   + 2.63E05 6.17E02 0.999974 96.3
9  + +  1.45E06 1.05E10 0.999999 99.2
10  +  + 2.72E04 1.35E24 0.987967 100.0
11   + + 2.80E05 1.28E10 0.999972 99.2
12 + + +  1.07E06 1.45E01 0.999999 94.4
13 + +  + 4.00E06 6.02E02 0.999996 96.3
14 +  + + 2.11E05 1.46E01 0.999979 94.4
15  + + + 4.31E06 1.06E10 0.999996 99.2
16 + + + + 3.18E06 1.45E01 0.999997 94.4

Table A2. Results of the 24 factorial design for 99.5% recovery.

A B C D a b c d
1     8.90E05 1.90E24 0.970094 100.0
2 +    7.67E06 6.00E02 0.999992 98.1
3  +   8.90E05 8.80E25 0.970094 100.0
4   +  8.90E05 2.09E24 0.989692 100.0
5    + 2.67E04 1.90E24 0.970094 100.0
6 + +   1.12E06 5.97E02 0.999999 98.1
7 +  +  6.18E06 1.45E01 0.999994 96.1
8 +   + 2.30E05 6.01E02 0.999977 98.1
9  + +  8.90E05 9.41E25 0.989692 100.0
10  +  + 2.67E04 8.80E25 0.970094 100.0
11   + + 2.67E04 2.08E24 0.989692 100.0
12 + + +  9.01E07 1.45E01 0.999999 96.1
13 + +  + 3.36E06 5.97E02 0.999997 98.1
14 +  + + 1.84E05 1.45E01 0.999982 96.1
15  + + + 2.67E04 9.40E25 0.989692 100.0
16 + + + + 2.70E06 1.45E01 0.999997 96.1

2016 Curtin University of Technology and John Wiley & Sons, Ltd. Asia-Pac. J. Chem. Eng. 2016; 11: 958972
DOI: 10.1002/apj
 SIDE DRAW OPTIMISATION OF A MULTI-COMPONENT
Asia-Pacic Journal of Chemical Engineering DISTILLATION COLUMN 971
Table A3. Effects of the disturbances on the different response variables at 97.7% recovery.

XD,Eth XB,Meth XD,Meth Rec

A 8.69E05 1.03E01 6.01E03 4.28E02
B 1.07E05 6.86E04 1.07E05 1.16E05
C 8.62E05 4.24E02 6.01E03 1.35E02
D 5.25E05 4.71E05 7.22E06 5.30E06
AB 2.75E06 6.86E04 2.75E06 2.96E06
AC 8.43E05 4.24E02 6.01E03 5.59E03
AD 4.35E05 4.70E05 1.85E06 9.66E07
BC 3.23E06 3.34E05 3.22E06 3.22E06
BD 5.31E06 1.33E05 5.30E06 5.30E06
CD 4.31E05 1.70E05 2.17E06 1.09E06
ABC 4.68E06 3.34E05 4.70E06 5.44E06
ABD 1.38E06 1.33E05 1.37E06 9.66E07
ACD 4.21E05 1.70E05 3.20E06 3.24E06
BCD 1.60E06 1.95E06 1.60E06 1.09E06
ABCD 2.34E06 1.95E06 2.33E06 3.24E06

Table A4. Effects of the disturbances on the different response variables at 99.5% recovery.

XD,Eth XB,Meth XD,Meth Rec

A 1.70E04 1.02E01 2.01E02 2.92E02
B 5.90E06 1.75E04 5.90E06 5.21E06
C 8.71E07 4.25E02 9.80E03 9.71E03
D 9.30E05 2.29E05 3.95E06 3.11E06
AB 5.90E06 1.75E04 5.90E06 5.21E06
AC 8.71E07 4.25E02 9.80E03 9.71E03
AD 8.50E05 2.29E05 3.95E06 3.11E06
BC 6.51E07 1.02E05 6.50E07 1.09E06
BD 2.94E06 1.20E05 2.93E06 3.11E06
CD 4.44E07 5.15E06 4.50E07 1.01E06
ABC 6.51E07 1.02E05 6.50E07 1.09E06
ABD 2.94E06 1.20E05 2.93E06 3.11E06
ACD 4.44E07 5.15E06 4.50E07 1.01E06
BCD 3.33E07 1.40E06 3.25E07 1.01E06
ABCD 3.33E07 1.40E06 3.25E07 1.01E06

REFERENCES [9] Bayat M, Hamidi M, Dehghani Z, Rahimpour M. Sorption-

[1] Johnson, D. Global methanol market review 2014; retrieved
from http://www.ptq.pemex.com/productosyservicios/
eventosdescargas/Documents/Foro%20PEMEX%20Petroqu
[2] Aasberg-Petersen, K, Stub Nielsen, C. Large scale methanol
production from natural gas 2013; (technical report) retrieved
from http://www.topsoe.com/business_areas/methanol/media/
PDF%20les/Methanol/Topsoe_large_scale_methanol_prod_
paper.ashx.
[3] Department of Energy. Distillation column modeling tools
2001; retrieved from. http://www1.eere.energy.gov/
manufacturing/resources/chemicals/pdfs/distillation.pdf.
[4] Zhang J, Liang S, Feng X. A novel multi-effect methanol dis-
tillation process. Chem. Eng. Prog 2010; 49(10): 10311037.
[5] Glinos K, Malone M. Design of side stream distillation
columns. Ind Eng Chem Process Des Dev 1985; 24: 822828.
[6] Underwood A. Fractional distillation of multicomponent
mixtures. Ind. Eng. Chem. 1939; 41: 28442847.
[7] Bausa J, Watzdorf R, Marquardt W. Shortcut methods for
nonideal multicomponent distillation. AIChE J 1998; 44:
21812198.
[8] von Watzdorf R, Bausa J, Marquardt W. Shortcut methods
for nonideal multicomponent distillation: 2. Complex
columns. AIChE J 1999; 45: 16151628.
2016 Curtin University of Technology and John Wiley & Sons, Ltd.
enhanced FischerTropsch synthesis with continuous
adsorbent regeneration in GTL technology: modeling and
optimization. J. Ind. Eng. Chem 2013; 20(3): 858869.
[10] Najafabadi A, Fatemi S, Sohrabi M, Salmasi M. Kinetic
modeling and optimization of the operating condition of
MTO process on SAPO-34 catalyst. J. Ind. Eng. Chem
2012; 18: 2937.
[11] Luyben WL. Distillation Design and Control Using Apsen
Simulation. John Wiley & Sons Inc.: New Jersey, USA,
2006.
[12] Luyben W. Design and control of a methanol reactor/column
process. Ind. Eng. Chem. 2010; 49(13): 61506163.
[13] Sadgehi J, Ahangar M. The study of methanol
separation columns control. Orient. J. Chem 2012; 28(1):
229235.
[14] Shahandeh H, Jafari M, Kasiri N, Ivakpour J.
Economic optimization of heat pump-assisted distillation
columns in methanol-water separation. Energy 2015; 80(1):
496508.
[15] Shahandeh H, Ivakpour J, Kasiri N. Internal and external
HIDiCs (heat-integrated distillation columns) optimization
by genetic algorithm. Energy 2014; 60: 875890.
[16] You X, Rodriguez-Donis I, Gerbaud V. Extractive
distillation process optimisation of the 1.0-1a class system,
Asia-Pac. J. Chem. Eng. 2016; 11: 958972
DOI: 10.1002/apj
972 M. A. KRALLER ET AL.
acetone - methanol with water. Comput. Aided Chem. Eng
2014; 33: 13151320.
[17] Palmeira J, Silva J, Matos H. Optimization studies through
simulation of a methanol/water/glycerol distillation column.
Comput. Aided Chem. Eng 2015; 37: 677682.
[18] Tian H, Zhao S, Zheng H, Huang Z. Optimization of
coproduction of ethyl acetate and n-butyl acetate by reactive
distillation. Proc. Syst. Eng.Proc. Safety 2015; 25: 667674.
[19] Wolff E, Skogestad S. Operation of integrated three-product
(petlyuk) distillation columns. Ind. Eng. Chem. Res 1995;
34: 20942103.
[20] Lee, S, Binkley, M. Optimize design for distillation feed.
Hydrocarb Process 2011.
[21] Kister H. Distillation Design. McGraw-Hill: New York, 1992.
2016 Curtin University of Technology and John Wiley & Sons, Ltd.
Asia-Pacic Journal of Chemical Engineering
[22] Udugama, I, Kirkpatrick, R, Yu, W, Young, B High purity
distillation column: simulation and optimization. Paper
presented at Chemeca 2013, Brisbane, Australia, 2013, Sept.
29 - Oct. 2, 2013, paper 30270.
[23] Prausnitz J. Molecular Thermodynamics of Fluid-phase
Equilibria. Prentice-Hall: New Jersey, 1969 .
[24] Fredenslund A, Jones R, Prausnitz J. Group-contribution
estimation of activity coefcients in nonideal liquid mixtures.
AIChE J 1975; 21: 10861095.
[25] Kister H. Distillation Operation. McGraw-Hill: New York,
1992 .
[26] Fruehauf P, Mahoney D. Distillation column control design
using steady state models: usefulness and limitations. ISA
Trans. 1993; 32(2): 157175.
Asia-Pac. J. Chem. Eng. 2016; 11: 958972
DOI: 10.1002/apj