chemical engineering research and design 1 0 2 ( 2 0 1 5 ) 138149

Contents lists available at ScienceDirect

Chemical Engineering Research and Design

journal homepage: www.elsevier.com/locate/cherd

Applications of dividing wall column technology

to industrial-scale cumene production

Jian Zhai, Yuliang Liu, Lumin Li, Yi Zhu, Wang Zhong, Lanyi Sun
State Key Laboratory of Heavy Oil Processing, China University of Petroleum, Qingdao 266580, China

a r t i c l e i n f o a b s t r a c t

Article history: Recent advances in dividing wall column (DWC) have led to renewed interest in the design
Received 3 March 2015 or redesign of many industrial processes. For the distillation system of cumene production,
Received in revised form 25 May the existing design alternatives have the potentials of energy savings and consequently
2015 the reduction of total annual cost (TAC) and energy consumption. In this paper, cumene
Accepted 11 June 2015 production based on the real industrial process is simulated in a conventional process
Available online 19 June 2015 using commercial process simulator Aspen Plus, and two DWC distillation processes are
introduced. Then optimization study for the proposed DWC distillation processes is per-
Keywords: formed based on the TAC calculation with sensitivity analysis to investigate the effects
Cumene production of operating parameters. The results indicate that the optimum DWC distillation process
Dividing wall column works well for the production of cumene, and the corresponding TAC is reduced signicantly
Energy savings accompanied by substantial energy savings.
Total annual cost 2015 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.
Aspen Plus

1. Introduction and the DIPB was discarded as a byproduct stream without

As an important organic material, cumene is not only used
as the thinner for paints and varnishes, but also can be used
as the component of aviation fuel. Moreover, cumene is an
industrial intermediate mainly used for the production of phe-
nol and acetone, which is produced in a high-temperature
and high-pressure gas phase reactor by the reaction of ben-
zene and propylene (Sharma et al., 2013). The FriedelCrafts
alkylation of benzene with propylene is carried out to produce
cumene in the chemical industry. The cumene manufactur-
ing process has received a lot of attention in recent years
and effort was invested in developing alternative catalyst sys-
tems in order to optimize the cumene production (Perego and
Ingallina, 2002; Degnan, 2003). Turton (2009) described the
cumene process in detail without considering the optimiza-
tion. Luyben developed a steady-state economic optimum
design of the cumene process (Luyben, 2009). Nevertheless,
Turton et al. and Luyben didnt use a transalkylation reactor
to convert the di-isopropyl benzene (DIPB) back to cumene,
being processed further. However, the use of transalkylation
reactor is a standard practice in a conventional cumene man-
ufacturing process (Pohl and Ram, 2005). Pathak et al. (2011)
developed a reactive distillation process with transalkyla-
tion reactor for cumene manufacture, which was about 47%
cheaper for the total annual cost (TAC) compared with the
corresponding conventional distillation process design. Flegiel
et al. (2015) modied the cumene process design for reducing
the raw materials, product losses, and energy integration. Both
two designs they proposed are optimized using multiobjective
optimization method which required less energy requirement
than other literature designs. Much work was invested in
developing the cumene production, nevertheless few ow-
sheet based on the heat integration has been announced.
Distillation, as a workhorse of chemical process industries,
is an energy-intensive process and, therefore, it is the rst to
be addressed to improve the energy efciency over the short-
and long-term. Dividing wall column (DWC) distillation, which
is thermodynamically equivalent to the Petlyuk column, has

Corresponding author. Tel.: +86 13854208340.
E-mail address: sunlanyi@upc.edu.cn (L. Sun).
http://dx.doi.org/10.1016/j.cherd.2015.06.020
0263-8762/ 2015 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.
 chemical engineering research and design 1 0 2 ( 2 0 1 5 ) 138149 139

Nomenclature
XD cumene composition in the distillate of main
Greek letters column
hT height of the tower
V the latent heat of the steam (kJ/kg)
received considerable attention as a means of process intensi-
Abbreviations
cation in the eld of chemical engineering (Dejanovic et al.,
C-1 rst column of the separation unit of the whole
2010; Staak et al., 2014). Since BASF in Ludwigshafen set up
process
the rst production scale column in 1985, many researches
C-2 second column of the separation unit of the
have been proposed to retrot conventional columns to DWC
whole process
(Premkumar and Rangaiah, 2009; Long and Lee, 2013a, 2013b;
C-3 third column of the separation unit of the whole
Chew et al., 2014; Staak et al., 2014). Apart from the poten-
process
tial of both energy and capital cost savings, DWC also requires
CC capital cost ($)
less space, shorter piping and electrical runs, therefore, the
CP specic heat capacity of the cooling water
DWC shows great advantages over conventional two-column
D diameter of the column
conguration (Long et al., 2010; Cossio-Vargas et al., 2012;
DIPB di-isopropyl benzene
Sun et al., 2014). With the integration of azeotropic, extrac-
DWC dividing wall column
tive and reactive distillation principles, DWC can be utilized
F fresh feed ow rate (kg/h)
as azeotropic dividing wall column (A-DWC), extractive divid-
FL1 ow rate of the liquid stream withdrawn from
ing wall column (E-DWC) and reactive dividing wall distillation
main column (kg/h)
(R-DWC) which shows a remarkable reduction in terms of total
FV2 ow rate of the vapor stream withdrawn from
annual cost and energy requirement (Bravo-Bravo et al., 2010;
main column (kg/h)
Murrieta-Duenas et al., 2011).
ID industrial data
Fig. 1(a) illustrates the integration scheme of DWC for sepa-
LC height of the column (m)
rating ternary mixtures, which consists of two parts separated
M&S Marshall & Swift index
by a dividing wall. Ternary mixtures are introduced into the
N1 the number of C-1
left part of the column. The light component A and inter-
N2 the number of C-2
mediate component B move upward, while component B and
N3 the number of C-3
the heavy component C ow down out of the left part. These
Nactual the number of actual stages
compounds move to the right part of the DWC for further
NF feed tray location of the DWC
separation. It is noticed that the left and right parts share
NF1 feed tray location of C-1
the common rectifying and stripping sections, which conse-
NF2 feed tray location of C-2
quently reduces the capital cost as well as the utility cost
NF3 feed tray location of C-3
compared with the conventional distillation sequence.
NHV the net heating value of a fuel (kJ/kg)
Process simulation technology is an effective tool to ana-
NL1 source stage of the liquid stream withdrawn
lyze the performance of a chemical engineering process.
from main column
Aspen Plus, which contains strong databases, complete sets of
NRTL non-random two liquid
modules and exible simulation tools, is one of the standard
NS side stream tray location of the DWC
process simulators in chemical industries. Fig. 1(b) shows
NT1 number of stages in the main column
the equivalent scheme for DWC implemented in Aspen Plus,
NT2 number of stages in the prefractionator
which consists of a prefractionator and a main column. This
NV2 source stage of the vapor stream withdrawn
equivalent scheme is applied in the study of this paper
from main column
for its feasibility of modeling in Aspen Plus. Schultz et al.
OP operating cost ($/year)
(2009) proposed a process with the use of DWC for producing
PIPBs polyisopropyl benzene
cumene, and the fresh benzene fed to the alkylation reactor
P1 operating pressure of the main column (MPa)
directly doesnt have any impurities, however, the industrial
P2 operating pressure of the prefractionator (MPa)
feed benzene contains impurities such as inert n-propane
PT1 operating pressure of C-1 (bar)
and water, which has a signicant inuence on the activ-
PT2 operating pressure of C-1 (bar)
ity and stability of catalyst (Van Bokhoven et al., 2002). The
PT3 operating pressure of C-1 (bar)
reduction of CO2 emissions from distillation systems is an
QC condenser heat duty (kW)
absolute necessity and an expensive challenge in the chemical
QR reboiler heat duty of the DWC (kW)
industries (Gutirrez-Guerra et al., 2009; Ibarra-Snchez and
QR1 reboiler heat duty of C-1 (kW)
Segovia-Hernndez, 2010). It is desirable to nd other feasible
QR2 reboiler heat duty of C-2 (kW)
distillation systems for industrial cumene production that can
QR3 reboiler heat duty of C-3 (kW)
achieve the optimum economic design.
R reux ratio in the main column
Dividing wall columns (DWC) for the distillation of multi-
R1 reux ratio in C-1
component mixtures have received much attention in the past
R2 reux ratio in C-2
15 years and have experienced a booming development. Since
R3 reux ratio in C-3
DWC demand higher efforts in equipment design and process
SR simulated results
control, few open literature can be found where DWC are used
TAC total annual cost ($/year)
in industrial operation of cumene production.
U heat transfer coefcient (kW/(K m2 )
Several schemes have been proposed to produce cumene
(Lei et al., 2004; Pohl and Ram, 2005; Turton, 2009; Luyben,
140 chemical engineering research and design 1 0 2 ( 2 0 1 5 ) 138149
Fig. 1 (a) Integration scheme and (b) equivalent scheme of DWC for separating ternary mixtures.
2009; Pathak et al., 2011; Sharma et al., 2013). However, the
fresh benzene contains impurities such as water which will
deactivate the activity of catalyst and this was not consid-
ered in the literature. This paper investigates a conguration
suitable for improving the energy efciency and reducing the
total annual cost of the cumene production process mainly via
the utilization of DWC in industrial purpose. The conventional
distillation process is simulated in terms of actual industrial
conditions using commercial process simulator Aspen Plus,
which not only adopts a transalkylation reactor to convert the
DIPB to cumene, but also adopts a distillation column to rene
the fresh feed benzene rather than introducing it to the alkyl-
ation reactor directly. The simulation results are compared to
real operating data from the industrial production. Two DWC
distillation processes for cumene production are proposed to
nd the potential benets derived from DWC. The results show
that utilizing DWC can not only signicantly achieve energy
savings, but also minimize total annual cost.
The rest of this paper is organized as follows. First, a real
industrial process for cumene production is briey described,
and this conventional distillation process is simulated using
the process simulator Aspen Plus. Then, two DWC distilla-
tion processes for cumene production are proposed to achieve
energy savings in contrast to the conventional distillation
sequence. The optimization of the two new DWC distillation
processes and the conventional distillation sequence are per-
formed based on the total annual cost. Finally, the comparison
between the conventional three-column design and the two
DWC congurations are made to nd the optimum design for
industrial-scale cumene production.
2. Conventional distillation process for
cumene production
2.1. Description of the cumene plant
In a typical commercial process, the FriedelCrafts alkylation
of benzene with propylene is the most common route to pro-
duce cumene, which can be further alkylated to DIPB and
heavier polyisopropyl benzenes (PIPBs) as byproducts. Accord-
ing to the real industrial process, the formation of PIPBs is
considered in this study which is different from the process
design in the literature (Turton, 2009; Luyben, 2009). In order
to suppress further alkylation of cumene to side products,
a molar excess of benzene throughout the reaction zone is
maintained. In addition, a transalkylation reactor is employed
to convert the DIPB and PIPBs back to cumene. The alkylator
and transalkylator efuents are traditionally separated using
three distillation columns to obtain a high-purity cumene
product, and the schematic diagram of conventional distilla-
tion process for cumene production is given in Fig. 2.
In this study, cumene is the main product of the whole pro-
cess with the separation unit consisting of three distillation
columns. As shown in Fig. 2, the fresh benzene, which contains
impurities such as water, is fed to the rst column (C-1), and
the inert n-propane, water and small amounts of light compo-
nents are separated as vapor distillate of C-1. The benzene is
withdrawn from the middle section of C-1, and the bottoms are
further distilled in the second column (C-2) to recover cumene
product as overhead, and DIPB and PIPBs are further separated
in the third column (C-3). Since the distillate of C-3 contains
some cumene that has a negative inuence on the transalky-
lation process, the recycle stream is withdrawn from the C-3
and fed to the transalkylation reactor reacting with benzene,
and the high boiling bottoms, which largely contains PIPBs, is
stored for further usage.
2.2. Modeling and simulation of the whole process
The whole conventional distillation process at industrial scale
is simulated using commercial process simulator Aspen Plus.
The feed conditions, product specications, column operat-
ing conditions are available from the plant upon the situation.
To begin the steady-state simulation, the number of theoret-
ical trays of the three columns are 71, 39 and 19, respectively
(including the condensers and reboilers), as shown in Fig. 2. To
simulate the whole process, the overall tray efciency for each
column is assumed to be 75% (Mller and Marquardt, 1997; Sun
et al., 2014). Thus, the three columns contain 95, 52 and 25
actual trays (sieve tray), respectively. The Design Spec/Vary
feature in Aspen Plus is used to achieve the desired product
quality requirements. Design specication is set up to meet
the desired cumene product purity of 99.9 wt% by manipulat-
ing the reux ratio for the column C-2. To predict the activity
coefcients of the components, the Non-Random Two Liquid
(NRTL) thermodynamic model is adopted for the rst column
because of the strong interaction between water and other
components, and Peng Robinson (PENG-ROB) thermodynamic
model is adopted for the second and third columns due to
the properties of the components. Tables 1 and 2 show the
operating conditions for the reactors and distillation columns,
respectively, and these conditions are consistent with the real
 chemical engineering research and design 1 0 2 ( 2 0 1 5 ) 138149 141

Fig. 2 Process ow diagram of synthesis and separation of the cumene product.

Table 1 Operating conditions for alkylation reactor and transalkylation reactor.

Temperature (K) Pressure Space velocity Recycle Benzene/propylene Benzene/DIPB
(MPa) (h1 ) ratio mole ratio mole ratio

Alkylation reactor 438.15 3.1 7.5 3.0 3.5 \

Transalkylation reactor 453.15 1.6 3.0 \ \ 6.0

process. The simple conversion reactor models are used, and electricity is not included in this work because they are much
the cumene products are specied as mole fraction of 21.3% lower than the costs of the heat duties. The capital cost (CC)
and 17.6% to dene the alkylation and transalkylation reac- includes installed column shell cost, tray cost and exchanger
tions, respectively. cost. The small equipment of this process such as reux
Table 3 gives the comparison of all streams in both drums, pumps, valves and pipes are not considered because
simulated and industrial processes. As shown in Table 3, of their lower costs compared with the column vessel (includ-
the simulated results are highly consistent with the indus- ing column internals) and heat exchangers (condensers and
trial data, which indicates that the thermodynamic models reboilers) (Luyben, 2013).
adopted in the simulation process can obtain satisfactory pre- A capital payback of 3 years is assumed. The diameter
diction for thermal and caloric properties in the conventional of the column is determined by tray sizing option in RAD-
distillation process, thus the thermodynamic models can be FRAC block of Aspen Plus, and the materials of construction
used in the following study. are carbon steel. Sieve tray is chosen as the tray type, and
the tray spacing is specied as 0.6 m for both the DWC and
2.3. Economic optimization for the conventional conventional distillation column. The heat transfer areas for
distillation process condenser, reboiler, cooler and heater are calculated based on
the overall heat transfer coefcient and differential temper-
Total annual cost (TAC) is employed as the objective function ature. The overall heat transfer coefcients are assumed to
to achieve the economically optimum design by adjusting the be 0.852 kW/(K m2 ) for the condenser, heater and cooler, and
design parameters, including the number of trays in each col- 0.568 kW/(K m2 ) for the reboiler. All the data are taken from
umn, the feed location in the column, etc. The TAC (Douglas, Luybens book (Luyben, 2013), and the major equipment costs
1988) is dened as: are estimated by Douglas (1988). High-pressure steam with
a pressure of 41 barg and cooling water with a temperature
TAC = operating cost +
capital cost
(1) of 25 C are supposed to be available in the plant. The util-
payback year ity prices showed in Table 4 are taken from the Smiths book
(Smith, 2005) and the paper reported by Wu et al. (2013). In
where the operating cost (OC) usually includes the cost of
addition, its reasonable to add an assumed penalty to the col-
steam, cooling water and electricity, however the cost of
umn shell and trays because the construction and installation
of a DWC might be much more difcult than that of a conven-
Table 2 Operating conditions for the three distillation tional distillation column (Xia et al., 2012). Hence, a penalty of
columns. roughly 10% is assumed.
Column Condenser Reboiler Operating Reux The equipment is sized as follows:
number temperature temperature pressure ratio
(K) (K) (MPa) (1) The heat transfer area A (m2 ) in the condenser and
C-1 318.15 474.49 0.2 278.63 reboiler:
C-2 487.35 552.67 0.4 0.87
Q
C-3 531.11 569.23 0.4 44.39 A[m2 ] = (2)
U T
 142
Table 3 Comparison of the streams between the simulated results (SR) and industrial data (ID).
Stream ID 1 2 3 4 5 6 7 8 9

ID SR ID SR ID SR ID SR ID SR ID SR ID SR ID SR ID

Temperature 343.15 343.15 313.15 313.15 438.15 438.15 453.15 453.15 418.15 418.15 384.28 384.28 384.28 384.28 384.28 384.28 318.15
(K)
Pressure 0.4 0.4 1.7 1.7 3.1 3.1 1.6 1.6 1 1 0.24 0.24 0.24 0.24 0.24 0.24 0.2
(MPa)
Total ow 8383.75 8383.75 4527.33 4527.33 34,132.30 34,132.63 13,175.25 13,175.74 47,308.31 47,308.37 39,261 39,261.99 29,605 29,605.3 9655.5 9655.69 33.87
(kg h1 )

chemical engineering research and design 1 0 2 ( 2 0 1 5 ) 138149

Mass fraction (wt%)
Propylene 0 0 99.95 99.95 0 0 0 0 0 0 0 0 0 0 0 0 0
Propane 0 0 0.05 0.05 0.07 0.07 0 0 0.05 0.05 0 0 0.002 0.002 0.002 0.002 61.21
2- 0.50 0.50 0 0 0 0 0 0 0 0 0 0 0 7.8E 09 0 7.8E 09 26.25
Methylpentane
Benzene 99.44 99.44 0 0 64.43 64.44 64.75 64.73 64.51 64.52 98.89 98.9 98.91 98.90 98.91 98.90 11.07
Methylbenzene 0 0 0 0 0.85 0.85 0.73 0.74 0.82 0.82 0.99 0.99 0.99 0.99 0.99 0.99 0
Cumene 0 0 0 0 28.65 28.36 23.35 23.29 26.95 26.95 0.12 0.11 0.23 0.11 0.23 0.11 0
DIPB 0 0 0 0 5.60 5.81 10.52 10.58 7.14 7.14 0 0 0 3.8E 09 0 3.8E 09 0
PIPBs 0 0 0 0 0.45 0.47 0.65 0.66 0.53 0.52 0 0 0 5.1E 14 0 5.1E 14 0
Water 0.06 0.06 0 0 0 0 0 0 0 0 0 0 0 6.1E 13 0 6.1E 13 1.47

Stream ID 9 10 11 12 13 14 15 16 17

SR ID SR ID SR ID SR ID SR ID SR ID SR ID SR ID SR

Temperature 318.15 318.15 318.15 318.15 318.15 474.63 474.63 313.15 313.15 495.15 495.15 318.15 318.15 318.15 318.15 313.15 313.15
(K)
Pressure 0.2 0.2 0.2 0.2 0.2 0.4 0.4 0.4 0.4 0.4 0.4 0.4 0.4 0.4 0.4 0.4 0.4
(MPa)
Total ow 31.25 5.73 4.01 68.75 65.19 16,330.54 16,330.68 12670 12699.9 3660.78 3660.78 67.58 67.58 73.15 73.15 3520.05 3520.05
(kg h1 )

Mass fraction (wt%)

Propylene 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Propane 61.86 21.00 0 6.78 5.63 0 7.1E 29 0 0 0 0 0 0 0 0 0 0
2- 25.76 0.18 0 51.08 51.96 0 4.05E 25 0 0 0 0 0 0 0 0 0 0
Methylpentane
Benzene 11.02 0.53 0.1 41.91 41.49 0 2.15E 09 0 0 0 0 0 0 0 0 0 0
Methylbenzene 0 0 0 0 0 0 6.14E 06 0 0 0 0 0 0 0 0 0 0
Cumene 0 0 0 0 0 77.88 77.81 99.90 99.9997 0.97 0.99 42.02 18.96 0 0 0 0.67
DIPB 0 0 0 0 0 20.64 20.68 0 0 92.25 92.25 57.98 80.18 55.72 33.34 93.87 93.71
PIPBs 0 0 0 0 0 1.48 1.51 0 0 6.78 6.76 0 0.86 44.28 66.66 6.13 5.62
Water 1.36 78.29 99.9 0.23 0.92 0 4.2E 27 0 0 0 0 0 0 0 0 0 0
 chemical engineering research and design 1 0 2 ( 2 0 1 5 ) 138149 143

Table 4 Utility prices. Table 5 Optimized design of conventional distillation

process.
Utility Price
Design variables Value
High-pressure steam (41 barg) 14.89 $/103 kg
Cooling water 0.03 $/103 gal Number of stages
N1 69
N2 34
Here, T (K) is the logarithmic mean temperature differ-
N3 16
ence; Q (kW) is the heat duty; U (kW/(K m2 )) is the heat
Molar reux ratio
transfer coefcient. R1 278.63
(2) The height of the column: R2 0.33
R3 44.39
Lc [m] = (Nactual 1) 0.6 + 6 (3) Total energy consumption (kW) 7594.38
Capital cost (103 $) 2272.58
Operating cost (103 $/year) 1899.05
Here 0.6 m is the tray spacing and 6 m is the disengagement.
TAC (103 $/year) 2656.58
The CC is calculated as follows:

A number of design variables for conventional distillation

(1) Installed column cost
process to be determined include the total number stages of
Installed column shells cost[$] three columns (N1 , N2 and N3 ), feed tray locations of three
columns (NF1 , NF2 and NF3 ), and reux ratio of three columns
M&S
= 937.636 D1.066 L0.802
c (2.18 + Fc ) (4) (R1 , R2 and R3 ). Design specication is set up to meet the
280
desired cumene product purity of 99.9 wt% by manipulating
the reux ratio (R2 ) for the column C-2. So many design vari-
Here Fc = Fm Fp ; Fm and Fp are the construction mate- ables should be optimized to nd the most economical design
rial and pressure range, respectively. The ratio of Marshall under the specied product purity and yields. It is important
& Swift index (M&S) is specied at 1518.1 (in year 2011) to simplify the optimization procedure. Therefore, the sequen-
(Marshall and Swift, 2012). tial iterative optimization procedure is used to nd the optimal
design with the Ni (i.e., N1 , N2 or N3 ) as the outer iterative loop
M&S and NFi (i.e., NF1 , NF2 or NF3 ) as the inner iterative loop. The
Installed column trays cost[$] = 97.243
280 optimization procedure to minimize the TAC which is shown
D1.55 hT Fc (5) in Fig. 3 is summarized below. Table 5 gives the gives the opti-
mized design parameters of conventional distillation process.
A. Optimization for rst column (C-1):
Here Fc = Fs + Ft + Fm , and hT [m] = (Nactual 1) 0.6.
(2) Installed heat exchanger cost (1) Fix three column pressure (PT1 , PT2 and PT3 ) at 2 bar, 4 bar
and 4 bar, respectively.
(2) Guess the total stages of C-2 (N2 ) and C-3 (N3 ).
M&S (3) Guess the total stages of C-1 (N1 ).
Insalled heat exchanger cost[$] = 474.668 A0.65
280 (4) Guess the fresh benzene feed tray location (NF1 ).
(2.29 + Fc ) (6) (5) Use the Design Spec/Vary feature to adjust the reux
ratio (R1 ) until the specication can be met.
(6) Go back to step 4 and change NF1 until reboiler duties of
Here Fc = (Fd + Fp ) Fm . three columns (QR1 , QR2 and QR3 ) is minimized.
Steam cost and cooling water cost are calculated as follows: (7) Go back to step 3 and change N1 until TAC is minimized.

(1) Steam cost B. Optimization for second column (C-2):

QR
Steam cost [$/year] = Cs 8000 3600 (7) (1) Fix total stage of C-1 (N1 ) what has been optimized and
V
guess the total stages of C-3 (N3 ).
Here Cs ($/kg) is the saturated steam price; V (kJ/kg) is (2) Guess the total stages of C-2 (N2 ).
the latent heat of the steam which depends on the bottom (3) Guess the feed tray location (NF2 ).
temperature; QR (kW) is the reboiler heat duty. (4) Use the Design Spec/Vary feature to adjust the reux
(2) Cooling water cost ratio (R2 ) until the specication can be met.
(5) Go back to step 3 and change NF2 until reboiler duties of
three columns (QR1 , QR2 and QR3 ) is minimized.
0.03 QC (6) Go back to step 2 and change N2 until the TAC is mini-
Cooling water cost [$/year] =
3.785 T Cp 1000 mized.
8000 3600 (8)
C. Optimization for third column (C-3):

Here QC (kW) is the condenser heat duty; Cp (kJ/(kg K)) is the (1) Fix total stages of C-1 (N1 ) and C-2 (N2 ) that have been
specic heat capacity of the cooling water; T (K) is the tem- optimized.
perature difference between the outlet and inlet of the cooling (2) Guess the total stages of C-3 (N3 ).
water. (3) Guess the feed tray location (NF3 ).
144 chemical engineering research and design 1 0 2 ( 2 0 1 5 ) 138149

Fixed PT1 = 0.2 MPa,

PT2 =PT3= 0.4 MPa

Obtain the Obtain the

Given N2, N3 optimum NF1, N1 optimum NF2, N2

Given N1 Change N2 Change N3

Given NF1 Given NF2 Given NF3

Vary R1 to meet Vary R2 to meet Vary R3 to meet

specification of C-1 specification of C-2 specification of C-3

Is QR Is QR Is QR
No minimized No minimized No minimized
with N1, N2 with N1, N2 with N1, N2
N3 fixed? fixed? N3 fixed?

Yes Yes Yes

Calculate TAC Calculate TAC Calculate TAC

Obtain the optimum N1,

Is TAC Is TAC N2, N3 and minimum TAC
No minimized Yes minimized Yes
with N1, N2 with N1, N2
N3 fixed? N3 fixed? Over

Fig. 3 Flowchart for the optimization of conventional distillation process.

(4) Use the Design Spec/Vary feature to adjust the reux
ratio (R3 ) until the specication can be met.
(5) Go back to step 3 and change NF3 until reboiler duties of
three columns (QR1 , QR2 and QR3 ) is minimized.
(6) Go back to step 2 and change N3 until the TAC is mini-
mized.
Table 5 gives the optimal design variables and the mini-
mum TAC of this conventional distillation process.
3. Cumene production by dividing wall
column process
3.1. Replacing C-2 and C-3 with a single dividing wall
column
3.1.1. Steady-state design for the dividing wall column
distillation process
From the view of energy efciency point, integrated and cou-
pled structures are usually better than non-integrated ones
(Emtir and Fonyo, 2001). In this section, the new DWC dis-
tillation process is proposed for synthesis and separation of
high-purity cumene, which has the same feed conditions and
operating conditions with the conventional distillation pro-
cess, and Fig. 4 gives the process ow diagram of the DWC
sequence by replacing C-2 and C-3 with a single dividing wall
column. The DWC structure is initially designed with a short-
cut design procedure based on the similarity between the
conventional column conguration and the DWC sequences.
As shown in Fig. 4, the cumene product with 99.9 wt% purity is
distilled from the top of the DWC, and the DIPB is withdrawn
from the middle section of the DWC, which is recycled back to
the transalkylation reactor. The bottom stream of the DWC is
cooled and stored as other usage.
3.1.2. Economical optimization of the dividing wall column
Since the design obtained by the shortcut method might not
be the optimal design, an optimization is essential. Its a con-
vention to use the total annual cost including annual capital
costs and operating costs as objective function to screen pro-
cess candidates by adjusting the design parameters of DWC,
thus the economical optimization of the DWC is carried out
on the basis of the TAC.
Compared with the conventional distillation process, the
last two columns are replaced by a column with a dividing wall
in the DWC distillation process, which is the main difference
between the two processes, thus in this work the calculation
of TAC is focus on the separation unit.
3.1.2.1. Sensitivity analysis. The design parameters of the
DWC include the number of stages in the main column (NT1 )
and in the prefractionator (NT2 ), reux ratio in the main col-
umn, feed tray location of the DWC (NF ), etc., which should
be varied to nd the most economical design. In addition,
withdrawal locations (NL1 , NV2 ) and ow rates (FL1 , FV2 ) of the
streams withdrawn from the main column have signicant
effects on the composition proles and reboiler heat duty of
the DWC (QR ), sensitivity analysis should be carried out to seek
the optimum NL1 , NV2 and set the initial values of FL1 , FV2 .
3.1.2.1.1. Liquid interconnection ow rate FL1 . Fig. 5 illus-
trates the effect of NL1 and FL1 on the cumene composition
in the distillate of main column (XD) and the QR . As shown
in Fig. 5(a), for a given FL1 , there exists an optimum NL1 value
that gives the maximum cumene purity, and for a xed NL1 , the
higher the value of FL1 , the higher the cumene purity that can
be obtained. To achieve the desired cumene purity of 99.9 wt%,
FL1 must be above 25 kmol/h with NL1 among 11 to 13. Fig. 5(b)
shows the relationship between QR , NL1 and FL1 . As shown in
Fig. 5(b), QR reaches a peak when FL1 is 25 kmol/h with NL1
 chemical engineering research and design 1 0 2 ( 2 0 1 5 ) 138149 145

Fig. 4 Process ow diagram in which C-2 and C-3 are replaced with a DWC.

near stage 11. Thus stage 13 is a good choice for NL1 to obtain
a lower reboiler heat input.
3.1.2.1.2. Vapor interconnection ow rate FV2 . Fig. 6 illus-
trates the effect of NV2 and FV2 on XD and QR . It is revealed
that NV2 must be kept between 18 and 24 to give the desired
cumene purity of 99.9 wt% from Fig. 6(a). Fig. 6(b) shows the
effect of NV2 and FV2 on QR . It can be seen that QR reaches a
low point when FL1 is 168 kmol/h with NV2 near stage 22, thus
NV2 is xed at 22 and FL1 at 168 kmol/h.
3.1.2.2. Optimization. The design of the dividing wall column
is more complex that conventional columns because of the
greater number of degrees of freedom. The number of stages
in main column and prefractionator, the reux ratio, the dis-
tillate ow rate, the feed tray location, feed stream condition,
ow rates and withdrawal locations of vapor and liquid inter-
connecting streams must all be initialized, or the rigorous
simulation cannot be performed. These are important degrees
of freedom. Based on the studies of Luyben and Chien (2010),
Xia et al. (2012), and Wang et al. (2014), a global sequential iter-
ative optimization procedure is used to optimize the dividing
wall column. A global optimization sequence for a DWC and
a conventional distillation column are carried out to illustrate
the optimization. As shown in Fig. 7, for the main column of
DWC, NL1 and NV2 are used as the inner iterative loop; and
NT1 is used as the outer iterative loop. For the prefraction-
ator, NF needs to be determined as the iterative variable of
the inner iterative loop, and NT2 is the iterative variable of the
outer iterative loop. Here, NL1 , NV2 , NT1 , NF and NT2 represent
source stage of the liquid stream withdrawn from main col-
umn, source stage of the vapor stream withdrawn from main
column, number of stages in the main column, feed tray loca-
tion of the DWC and number of stages in the prefractionator,
respectively. The conventional distillation columns are also
optimized to compare other processes fairly.
3.1.2.2.1. Selection of NT1 . It is well known that NT1
greatly affects the TAC for the design of the sequence, thus
Fig. 8 demonstrates how the CC, OC and TAC of DWC cong-
uration are affected by NT1 . As shown in Fig. 8(a), it indicates
that the CC rises and OC drops with the increasing NT1 , which
seems to be reasonable. Its because that the shell length of
the main column increases with the rise in NT1 , which will
lead to an increase of CC, and the drop of reboiler duty will
make the decrease of OC. It can also be seen that there exists
an optimum NT1 that gives the minimum TAC as shown in
Fig. 8(b).
3.1.2.2.2. Selection of NT2 . As is revealed in Fig. 9, the TAC
of the DWC decreases signicantly when NT2 is less than 11,

Fig. 5 Effects of NL1 and FL1 on (a) cumene composition in the top of the main column (XD), (b) reboiler heat duty of the
DWC (QR ).
146 chemical engineering research and design 1 0 2 ( 2 0 1 5 ) 138149

Fig. 6 Effects of NV2 and FV2 on (a) cumene composition in the top of the main column (XD), (b) reboiler heat duty of the
DWC (QR ).

Fig. 9 Effects of NT2 on the TAC of the DWC.

varied from 3.2 to 4.4 bar for DWC process. The relation-
ship between the TAC and the operating pressure of DWC is
presented in Table 6 as shown blow. The optimal operating
pressure of DWC was determined from the minimum TAC.
According to Table 6, the optimal pressure of DWC is 3.2 bar.

3.2. Replacing C-1 and C-2 with a single dividing wall

column
Fig. 7 Flowchart for the optimization of DWC.

Another possible option for heat integration using dividing

and there exists a plateau when NT2 is equal to or larger than
11, which determines the optimum NT2 of 11 to give the TAC
of 2198.48 103 $/year.
3.1.2.2.3. Optimum design of the dividing wall column.
To obtain the best conditions, the operating pressure was
wall column is the replacement of the rst column and the
second column with a single dividing wall column. The whole
process ow diagram of the DWC where the two columns, C-1
and C-2 are integrated is shown in Fig. 10. Thus the separa-
tion unit consists of a DWC and a conventional distillation

Fig. 8 Effects of NT1 on (a) the OC and CC (b) the TAC of the DWC.
 chemical engineering research and design 1 0 2 ( 2 0 1 5 ) 138149 147

Table 6 Simulated results for DWC at different pressure of replacing C-2 and C-3 with a single DWC process.
Operating pressure (bar)

Parameters 3.2 3.4 3.6 3.8 4.0 4.2 4.4

Number of stages in C-1 69 70 71 69 69 71 69

Number of stages in the main column 21 21 20 21 27 22 24
Number of stages in the prefractionator 14 16 14 14 11 11 15
Reux ratio for DWC 0.6870 0.6977 0.7026 0.7111 0.7377 0.7432 0.7442
Source stage of the liquid stream 15 15 14 15 15 15 15
withdrawn from main column
Flow rate of the liquid stream withdrawn 38 40 58 54 40 60 60
from main column (kg/h)
Source stage of the vapor stream 17 17 17 17 19 17 19
withdrawn from main column
Flow rate of the vapor stream withdrawn 158 162 165 169 163 177 180
from main column (kg/h)
Total energy requirement (kW) 6638.45 6672.33 6699.23 6728.83 6750.30 6799.55 6819.15
Capital cost (103 $) 2173.77 2178.30 2171.98 2184.67 2205.84 2202.19 2222.30
Operating cost (103 $/year) 1433.85 1441.14 1446.92 1453.29 1463.20 1468.51 1472.72
TAC (103 $/year) 2158.44 2167.24 2170.91 2181.51 2198.48 2202.57 2213.49

Table 7 Simulated results for DWC at different pressure of replacing C-1 and C-2 with a single DWC process.
Operating pressure (bar)

Parameters 3.4 3.6 3.8 4.0 4.2 4.4

Number of stages in C-3 19 19 18 19 19 20

Number of stages in the main column 62 65 65 65 66 66
Number of stages in the prefractionator 10 10 13 10 11 11
Source stage of the vapor stream withdrawn 59 59 61 59 54 55
from main column
Flow rate of the vapor stream withdrawn from 27,856 33,647 31,229 38,391 29,384 22,372
main column (kg/h)
Total energy consumption (kW) 9036.64 9081.70 9127.26 9097.55 9216.66 9259.66
Capital cost (103 $) 2535.60 2535.61 2536.02 2542.70 2549.60 2556.68
Operating cost (103 $/year) 1952.76 1962.49 1972,28 1981.98 1991.49 2000.73
TAC (103 $/year) 3030.77 3040.50 3050.43 3062.36 3074.17 3085.77

column. The optimization of the proposed process is carried
out to seek the minimum TAC using the same method with
the rst conguration. During the optimization, the pressure
at the dividing wall column top was varied from 3.4 to 4.4 bar.
Table 7 reports the simulated results for the case of replacing
C-1 and C-2 with a single DWC process. According to Table 7,
the optimal operating pressure at the minimum TAC is 3.4 bar.
Table 8 gives the comparison between the conventional
three-column design and the two DWC congurations. As
shown in Table 8, the signicant reduction in TAC and energy

Fig. 10 Process ow diagram in which C-1 and C-2 are replaced with a DWC.
148 chemical engineering research and design 1 0 2 ( 2 0 1 5 ) 138149
Table 8 Comparison between the conventional distillation process design and the DWC distillation process design.
Distillation arrangements Conventional
distillation process
Number of trays
C-1 69
C-2 34
C-3 16
DWC
QR,total (kW) (% saving difference) 7594.38 (0%)
Economic data
Capital cost (103 $) (% saving difference) 2272.58 (0%)
Operating cost(103 $/year) (% saving difference) 1899.05 (0%)
TAC (103 $/year) (% saving difference) 2656.58 (0%)
requirement can be obtained by the replacement of C-2 and
C-3 using a DWC design, which results in reduction in the
capital cost of 4.34% and the operating cost of 24.50% based
on the replacement of column C-2 and column C-3. It is calcu-
lated that about 18.75% savings in TAC and 12.59% reduction
in energy requirement can be achieved by the optimum DWC
design compared with corresponding conventional process.
Thus, from the point of economical design, the DWC process
design seems to be more attractive than the corresponding
conventional distillation process design. However, the inte-
gration of the column C-1 and column C-2 using a single
DWC increases the energy requirement by up to 15.96%, and
the TAC increases about 14.09%. This conguration required
more reboiling energy than that consumed in the conventional
columns. It is noted that the DWC application does not always
take a benet.
4. Conclusions
In this work the design and optimization of two DWC distil-
lation processes for cumene production were studied. First,
the conventional distillation process for cumene production
was simulated based on a real industrial process, and the
simulation results were very close to the industrial data.
Then the rst two columns and the last two columns in the
conventional distillation process were integrated by a DWC,
respectively. The TAC calculation of the two DWC distillation
process were carried out. The results showed that appropri-
ate utilizing DWC can reduce energy requirement and TAC
signicantly. Notably, up to 18.75% of the TAC saving on the
DWC conguration can be achieved compared with the con-
ventional process design, which replaced the last two columns
with a single DWC of the separation unit of the whole process.
However, the replacement of the rst two columns, which
saved capital cost up to 4.34% in comparison with that of the
whole conventional process, required more reboiling energy
than that consumed in the conventional columns.
Acknowledgements
This work was supported by the National Natural Science
Foundation of China (Grant: 21276279; Grant: 21476261); the
Development of Key Technologies Project of Qingdao Eco-
nomic and Technological Development Zone under Grant
2013-1-57; Fundamental Research Funds for the Central Uni-
versities (No. 14CX05030A; No. 14CX06108A). Further the
authors are grateful to the editor and the anonymous review-
ers for their helpful comments and constructive suggestions
with regard to the revision of the paper.
DWC distillation DWC distillation
process(C-2 and C-3 are process(C-1 and C-2 are
replaced by a DWC) replaced by a DWC)
71
19
22 62
6638.45 (12.59%) 9036.64 (+15.96%)
2173.77 (4.34%) 2535.60 (11.44%)
1433.85 (24.50%) 1952.76 (+2.83%)
2158.44 (18.75%) 3030.77 (+14.09%)
References
Bravo-Bravo, C., Segovia-Hernndez, J.G., Gutirrez-Antonio, C.,
Duran, A.L., Bonilla-Petriciolet, A., Briones-Ramrez, A., 2010.
Extractive dividing wall column: design and optimization. Ind.
Eng. Chem. Res. 49, 36723688.
Chew, J.M., Reddy, C.C.S., Rangaiah, G.P., 2014. Improving energy
efciency of dividing-wall columns using heat pumps,
Organic Rankine Cycle and Kalina Cycle. Chem. Eng. Process.:
Process Intensif. 76, 4559.
Cossio-Vargas, E., Barroso-Munoz, F.O., Hernandez, S.,
Segovia-Hernandez, J.G., Cano-Rodriguez, M.I., 2012.
Thermally coupled distillation sequences: steady state
simulation of the esterication of fatty organic acids. Chem.
Eng. Process.: Process Intensif. 62, 176182.
Degnan, T.F., 2003. The implications of the fundamentals of
shape selectivity for the development of catalysts for the
petroleum and petrochemical industries. J. Catal. 216, 3246.
Dejanovic, I., Matijasevic, L., Olujic, Z., 2010. Dividing wall
columna breakthrough towards sustainable distilling.
Chem. Eng. Process.: Process Intensif. 49, 559580.
Douglas, J.M., 1988. Conceptual Design of Chemical Processes.
McGraw Hill, New York, NY.
Emtir, M., Fonyo, Z., 2001. Rigorous simulation of energy
integrated and thermally coupled distillation schemes for
ternary mixture. Appl. Therm. Eng. 21, 12991317.
Flegiel, F., Sharma, S., Rangaiah, G.P., 2015. Development and
multi-objective optimization of improved cumene production
processes. Mater. Manuf. Processes 30, 444457.
Gutirrez-Guerra, R., Segovia-Hernndez, J.G., Hernndez, S.,
2009. Reducing energy consumption and CO2 emissions in
extractive distillation. Chem. Eng. Res. Des. 87, 145152.
Ibarra-Snchez, J.J., Segovia-Hernndez, J.G., 2010. Reducing
energy consumption and CO2 emissions in extractive
distillation: Part II. Dynamic behavior. Chem. Eng. Res. Des.
88, 135145.
Lei, Z., Li, C., Li, J., Chen, B., 2004. Suspension catalytic distillation
of simultaneous alkylation and transalkylation for producing
cumene. Sep. Purif. Technol. 34, 265271.
Long, N.V.D., Lee, S., Lee, M., 2010. Design and optimization of a
dividing wall column for debottlenecking of the acetic acid
purication process. Chem. Eng. Process.: Process Intensif. 49,
825835.
Long, N.V.D., Lee, M., 2013a. Optimal retrot design of extractive
distillation to energy efcient thermally coupled distillation
scheme. AIChE J. 59, 11751182.
Long, N.V.D., Lee, M., 2013b. Optimal retrot of a side stream
column to a dividing wall column for energy efciency
maximization. Chem. Eng. Res. Des. 91, 22912298.
Luyben, W.L., Chien, I., 2010. Design and Control of Distillation
Systems for Separating Azeotropes. John Wiley & Sons, New
York, NY.
Luyben, W.L., 2009. Design and control of the cumene process.
Ind. Eng. Chem. Res. 49, 719734.
Luyben, W.L., 2013. Distillation Design and Control Using Aspen
Simulation. John Wiley & Sons, Chichester.
 chemical engineering research and design 1 0 2 ( 2 0 1 5 ) 138149
Marshall, Swift, 2012. Average M&S index for year 2011. Chem.
Eng. 119, 84.
Mller, D., Marquardt, W., 1997. Experimental verication of
multiple steady states in heterogeneous azeotropic
distillation. Ind. Eng. Chem. Res. 36, 54105418.
Murrieta-Duenas, R., Gutirrez-Guerra, R., Segovia-Hernndez,
J.G., Hernndez, S., 2011. Analysis of control properties of
intensied distillation sequences: reactive and extractive
cases. Chem. Eng. Res. Des. 89, 22152227.
Pathak, A.S., Agarwal, S., Gera, V., Kaistha, N., 2011. Design and
control of a vapor-phase conventional process and reactive
distillation process for cumene production. Ind. Eng. Chem.
Res. 50, 33123326.
Perego, C., Ingallina, P., 2002. Recent advances in the industrial
alkylation of aromatics: new catalysts and new processes.
Catal. Today 73, 322.
Pohl, S., Ram, S., 2005. ABB Lummus Global Cumene Production
via CD Cumene Technology. McGraw Hill, New York, NY.
Premkumar, R., Rangaiah, G.P., 2009. Retrotting conventional
column systems to dividing-wall columns. Chem. Eng. Res.
Des. 87, 4760.
Schultz, M.A., Lankton, S.P., Tagamolila, C.P., 2009. U.S. Patent No.
7,525,005. Washington, DC: U.S. Patent and Trademark Ofce.
Sharma, S., Chao Lim, Z., Rangaiah, G.P., 2013. Process design for
economic, environmental and safety objectives with an
application to the cumene process. In: Rangaiah, G.P.,
Bonilla-Petriciolet, A. (Eds.), Multi-Objective Optimization in
Chemical Engineering: Developments and Applications. John
149
Wiley & Sons, Oxford, UK, http://dx.doi.org/10.1002/
9781118341704.ch16.
Smith, R.M., 2005. Chemical Process Design and Integration. John
Wiley & Sons, New York, NY.
Staak, D., Grtzner, T., Schwegler, B., Roederer, D., 2014. Dividing
wall column for industrial multi purpose use. Chem. Eng.
Process.: Process Intensif. 75, 4857.
Sun, L., Wang, Q., Li, L., Zhai, J., Liu, Y., 2014. Design and control
of extractive dividing wall column for separating
benzene/cyclohexane mixtures. Ind. Eng. Chem. Res. 53,
81208131.
Turton, R., 2009. Analysis, Synthesis and Design of Chemical
Processes, third ed. Prentice Hall, New Jersey, NJ.
Van Bokhoven, J.A., Koningsberger, D.C., Kunkeler, P., Van
Bekkum, H., 2002. Inuence of steam activation on pore
structure and acidity of zeolite Beta: an Al K edge XANES
study of aluminum coordination. J Catal. 211, 540547.
Wang, S.J., Chen, W.Y., Chang, W.T., Hu, C.C., Cheng, S.H., 2014.
Optimal design of mixed acid esterication and isopropanol
dehydration systems via incorporation of dividing-wall
columns. Chem. Eng. Process.: Process Intensif. 85, 108124.
Wu, Y.C., Hsu, P.H.C., Chien, I.L., 2013. Critical assessment of the
energy-saving potential of an extractive dividing-wall column.
Ind. Eng. Chem. Res. 52, 53845399.
Xia, M., Yu, B., Wang, Q., Jiao, H., Xu, C., 2012. Design and control
of extractive dividing-wall column for separating
methylalmethanol mixture. Ind. Eng. Chem. Res. 51,
1601616033.