Professional Documents
Culture Documents
Article
Comparison of the Economy and Controllability of
Pressure Swing Distillation with Two Energy-Saving
Modes for Separating a Binary Azeotrope Containing
Lower Alcohols
Liping Lv 1,2 , Hang Li 2 , Zheng Zhang 3, * and Huisheng Huang 1
1 School of Chemistry and Chemical Engineering, Yangtze Normal University, Fuling 408100, China;
lvliping@yznu.edu.cn (L.L.); h.s.huang@hotmail.com (H.H.)
2 Key Laboratory of Gas Process Engineering, School of Chemistry and Chemical Engineering, Southwest
Petroleum University, Chengdu 610500, China; lihang0213@163.com
3 China Petroleum Engineering & Construction Corporation Southwest Company, Chengdu 610041, China
* Correspondence: zhangzheng_sw@cnpc.com.cn
Received: 30 July 2019; Accepted: 8 October 2019; Published: 12 October 2019
Abstract: The pressure swing distillation (PSD) with two different energy-saving modes are put
forward to separate a binary azeotrope containing lower alcohols: benzene/methanol. A comparison
of the economy and controllability for the partial and fully heat integrated pressure swing distillation
(HIPSD) is made by detailed simulation analysis. The optimal operating parameters of partial and
fully HIPSD processes are obtained by minimizing total annual cost (TAC). These results show that the
fully HIPSD mode saves 5.88% TAC compared with the partial HIPSD mode. Meanwhile, this paper
proposes that the composition slope profile can help to select the temperature control stage (TCS),
when the temperature profile in the column is rising rapidly near the bottom and the maximum of
temperature slope value occurs in the bottom of the column. Several control structures are developed
to check the rationality of the selection of the TCS and evaluate the industrial application. These results
illustrate the composition/temperature cascade control structure for the PSD with two energy-saving
modes can both get good control performances, and the purities of benzene and methanol can be
brought close back to the initial value. However, the fully HIPSD mode can only handle much smaller
composition disturbances (<10%) compared with the partial HIPSD mode. Therefore, the selection of
energy-saving modes for the separation process should weigh economy against controllability.
1. Introduction
Several binary azeotropic systems containing low carbon alcohols, such as benzene and methanol,
are widely used in the chemical and petrol chemical industries [1–3]. To separate and recycle the
azeotropic mixture, several enhanced distillations including pressure swing distillation (PSD) [4–6],
azeotropic distillation (AD) [7,8], and extractive distillation (ED) [9–11] have been applied to separate
the azeotrope successfully. For example, Shen et al. put emphasis on the separation of azeotropic
systems for the ED method [12]. Shirsat and Wang et al. optimized the ED and PSD method of
separating the isobutyl alcohol/isobutyl acetate and tetrahydrofuran/ethanol azeotropic system, and
indicated that the cost and product contamination should be concerned when selecting the separation
method [13,14]. Compared with AD and ED methods, PSD has drawn more and more attention for the
benefit of no third component added into the azeotropic system. Fulgueras and Maloney [15] pointed
out that PSD method was economically feasible when the mass fraction of azeotropic composition
changed more than 5% (10% is better) in the range of pressure 10 bar. The PSD process has two columns
which includes low pressure column (LPC) and high pressure column (HPC). For the minimum
boiling azeotropes, the high-purity component can be obtained at the bottom of LPC and HPC. For the
maximum boiling azeotropes, the opposite is true.
To reduce heating and cooling loads of thermal utilities and improve heat recovery capacity of
the system, some methods such as heat integration are proposed. In general, two modes of heat
integration are usually applied in the internal system [6,16,17]. One mode is the heat integration
between condenser and reboiler, which includes partial and fully heat integration. The heat integration
between the rectifying and stripping section is another mode. In the past years, the economy of
PSD processes was compared with different heat integrated modes [1,18,19]. Luyben [1] explored
the separation of tetrahydrofuran/water azeotropic system by PSD method with two different heat
integrated modes. Wang et al. [19] made a comparison of partial and fully heat integrated pressure
swing distillation (HIPSD) modes for the separation of methanol/tetrahydrofuran azeotropic system.
These results show that the economy of the PSD process depends on the different azeotropic systems
and heat-integrated modes.
Not only the economy but also the controllability of the separation process should be considered
in the determination of a suitable separation method. Many researchers pointed out that the economy
of the fully HIPSD mode was better but the controllability was weaker than that of the partial HIPSD
mode [1,17,19]. The main reason is that the distillate stream of the HPC is used to provide the heat
source of the reboiler in the LPC and no auxiliary reboiler is needed, leading to a degree of freedom
missed in the full HIPSD process. Luyben [1] studied the energy consumption and controllability of
partial and fully HIPSD modes for the separating water/tetrahydrofuran azeotropic system. It was
revealed that the energy consumption for the fully HIPSD mode was lowest among the three heat
integration modes, but its dynamic controllability was worse compared with the other two heat
integration modes. As a result, it is necessary to weigh the energy consumption and controllability for
different heat integrated modes. Zhang et al. explored the separation of a methyl acetate/methanol
azeotropic system by different heat-integrated modes. It can be seen that the fully HIPSD mode has a
little economic benefit compared with partial HIPSD mode, but its controllable performance is also
poor [6]. Zhu et al. contrasted the economics and controllability for separating of ethanol and toluene
azeotropic system by PSD process with three modes of heat integration [3]. The results indicated that
the total annual cost (TAC) for the fully heat integration is lowest, but this kind of heat integration
mode can handle only small disturbances very well. Similar studies were done for separating an
n-butyl alcohol and toluene system [20], a methanol and tetrahydrofuran system [2], and an ethanol
and chloroform system [21].
The selection of the temperature control stage (TCS) is very important to achieve the excellent
control performance because the signal of TCS is detected by the temperature controller and transmitted
to the adjusted variable. In general, five criteria including product fluctuation minimum criterion,
constant temperature criterion, singular value decomposition, sensitivity criterion, and slope criterion
are usually used to select the suitable TCS. Zhang et al. used singular value decomposition
and sensitivity criterion to select the TCS [6]. Hosgor et al. explored the separation of the
chloroform/methanol azeotropic system by fully HIPSD process and used slope criterion to select the
TCS [22]. Zhang et al. presented a derivative slope criterion to choose the TCS when the temperature
distribution is quite close to linear [23].
So far, there has been no article published that compares PSD with different heat integrated
modes for separating benzene/methanol azeotrope through economy and controllability angles. In
this work, the feasibility of the PSD process with two energy-saving modes for the separation of the
benzene/methanol azeotrope were proved, and a comparison of the economy and controllability for
the partial and fully HIPSD process were made in detail. Firstly, the partial and fully HIPSD processes
were optimized based on the minimization of the TAC. Secondly, a composition slope criterion was
presented to assist the selection of the suitable TCS for the HPC, when the temperature distribution in
Processes 2019, 7, 730 3 of 25
Processes 2019, 7, 730 3 of 25
in the
the bottom
column of the
is rising column.
rapidly nearThe different
the bottom anddegrees of the of
the maximum disturbance
temperature inslope
the feed
valueflowrate
occurs inand
the
composition
bottom of thewere introduced
column. to verify
The different theofrationality
degrees of theinselection
the disturbance the feed of TCS and
flowrate and evaluate the
composition
industrial
were application
introduced of two
to verify processes. of the selection of TCS and evaluate the industrial application
the rationality
of two processes.
2. Steady State Design and Optimization
2. Steady State Design and Optimization
2.1. Property Method
2.1. Property Method
The accuracy of the steady-state simulation depends on the selection of the thermodynamic
The
model. Sinceaccuracy of the steady-state
the benzene/methanol simulation
azeotrope dependsnon-ideal
is a strongly on the selection of the thermodynamic
system, several thermodynamic
model. Since the benzene/methanol azeotrope is a strongly non-ideal
models including Wilson, NRTL and UNIQUAC equations are used for calculating the system, several thermodynamic
vapor-liquid
models including Wilson, NRTL and UNIQUAC equations are used for calculating
equilibrium (VLE) data at 1 bar. The average relative error between calculated and experimental the vapor-liquid
data
equilibrium (VLE) data at 1 bar. The average relative error between calculated and
is listed in the Table 1. As seen from this table, the average relative error of the Wilson equation experimental data
is
is listed in the Table 1. As seen from this table, the average relative
minimum. Figure 1 shows the T-x-y diagram of benzene/methanol azeotropic system with error of the Wilson equation is
minimum.
experimental Figure
data1and
shows the T-x-ydata
calculated diagram of benzene/methanol
at 1 bar. From Figure 1, the azeotropic
calculatedsystem
data with experimental
and experimental
data
data and calculated
are in data at 1 bar.
good agreement. From Figure
Therefore, 1, the calculated
the Wilson equation data
was and experimental
selected data areproperty
as the suitable in good
agreement.
method. Therefore, the Wilson equation was selected as the suitable property method.
80 Experimental data
Calculated data
75
70
T /℃
65
60
55
0.0 0.2 0.4 0.6 0.8 1.0
Molefrac of methanol
Figure 1. T-x-y diagram of benzene/methanol system with experimental data and calculated data at
Figure 1. T-x-y diagram of benzene/methanol system with experimental data and calculated data at 1 bar.
1 bar.
2.2. Optimization
2.2. Optimization Method
Method
TAC including
TAC including both
both equipment
equipment costcost and
and annual
annual operating
operating cost is aa common
cost is common measure
measure forfor the
the
economic evaluation in chemical and pharmaceutical process [18]. Equations
economic evaluation in chemical and pharmaceutical process [18]. Equations (1)–(3) show the detailed (1)–(3) show the
detailed calculation.
calculation. Note thatNote that the payback
the payback period
period time time
is set as is set as and
3 years 3 years
the and the operation
operation time
time is set at is set
8000
at 8000 The
h/year. h/year. The equipment
equipment costpartial
cost for the for the
andpartial and fully
fully HIPSD HIPSD mainly
processes processes mainly
includes twoincludes two
distillation
distillation columns and heat exchangers. Compared with heat exchangers and distillation
columns and heat exchangers. Compared with heat exchangers and distillation columns, the costs for columns,
the pumps,
the costs forreflux
the pumps, refluxand
drum, pipes drum, pipes
valves areand
veryvalves are very
low, which canlow,
evenwhich can even
be ignored be conceptual
in the ignored in
the conceptual design phase [18,24]. The diameters of HPC and LPC are computed
design phase [18,24]. The diameters of HPC and LPC are computed via the “tray sizing” function via the “trayin
sizing” function in the Aspen plus. The annual operating cost mainly includes the
the Aspen plus. The annual operating cost mainly includes the annual steam cost and cooling water annual steam cost
and cooling
cost. water
The related cost. The formulae
calculating related calculating formulae
and parameters and parameters
are listed in the Tableare listed in the
2 according Table 2
to previous
according to previous work of
work of Luyben and Douglas [19,24]. Luyben and Douglas [19,24].
total equipment cost
TAC = equipment cos
total +tannual operating cost , (1)
payback period
TAC = + annual operating cos t , (1)
payback period
Total equipment cost = distillation columns + heat exchangers cost, (2)
Total equipment cost = distillation columns + heat exchangers cost, (2)
Annual operating cost = annual steam cost + annual cooling water cost. (3)
Processes 2019, 7, 730 4 of 25
Annual operating cost = annual steam cost + annual cooling water cost. (3)
Heat-transfer Q
Heatcoefficient (KC) Ac
= 0.852
= (KkW/K·m
2
transfer area:
c ×∆t)
Total equipment cost Heat transfer area: Ac =× (Ac, m2 )0.65
Reboilers Equipment cost = 7296 ( ×∆ )
Reboilers Heat-transfer
Equipmentcoefficient ) = 0.568
cost = 7296(K×R(Ac, m2)0.65kW/K·m2
Q
Heatcoefficient
Heat-transfer transfer area:
(KR)Ac = (K kW/K·m
= 0.568 2
R ×∆t)
0.9
composition 160
0.8 140
Liquid molefrac benzene
temperature
120
0.7
100
T/ ℃
0.6 80
60
0.5
40
0.4 20
0
0 1 2 3 4 5 6 7 8 9 10
P/bar
Figure 2. Effect
Figure ofof
2. Effect pressure
pressureon
onazeotropic compositionand
azeotropic composition and temperature.
temperature.
Furthermore,
Furthermore, thethe
keykeyof of
PSDPSD isisthe
theselection
selection of the
the pressures
pressuresofoftwotwo columns.
columns. TheThe difference
difference of of
azeotropic
azeotropic composition
composition increases
increases withthe
with theincrease
increase ofof pressure
pressuredifference,
difference, which
whichleads to decrease
leads of of
to decrease
the material
the material circulation
circulation andand energy
energy consumption.
consumption. However,ititincreases
However, increasesthethe cost
cost of
of the
the equipment.
equipment. For
For the
the LPC, theLPC, the complexity
complexity andof
and cost cost
the ofrefrigeration
the refrigeration systemincrease
system increasewhen
when thetheoperating
operating pressure
pressure is
is too low. In general, the pressure of the LPC is set as 1 bar, and the top vapor stream (about 57 °C)
too low. In general, the pressure of the LPC is set as 1 bar, and the top vapor stream (about 57 ◦ C) of
of the LPC can be condensed by the cooling water (32 °C). For the HPC, the selection principle of the
the LPC can be condensed by the cooling water (32 ◦ C). For the HPC, the selection principle of the
pressure is that the top stream can be used to heat the bottom liquid stream of LPC [6,25]. Table 3
pressure
givesisthe
thatTAC
theoftop stream
several canwith
cases be used to heat
different the bottom
pressures liquid
in HPC. Notestream
that theof LPC [6,25].difference
temperature Table 3 gives
the TAC
between the bottom liquid stream (about 67 °C at 1 bar) and the top vapor stream (about 116 °C between
of several cases with different pressures in HPC. Note that the temperature difference at 6
the bottom liquid stream ◦ C at 1 bar) and the top vapor stream (about 116 ◦ C at 6 bar) is
bar) is more than 25 °C.(about 67 heat
Thus, the integration can be implemented in the internal system. From
moreTable
than 3, ◦ C.TAC
25the Thus, the heat integration
is minimum for the HPCcan at 6be
barimplemented in theofinternal
while the pressure HPC at system.
range of From Table 3,
7–10 bar.
The is
the TAC main reason isfor
minimum that theHPC
the equipment
at 6 bar cost and the
while operating costof
pressure have
HPC a significant
at range of increase when
7–10 bar. themain
The
operation pressure of HPC is greater than or equal to 7 bar. Therefore, the optimum operation
pressure of HPC is selected as 6 bar.
Processes 2019, 7, 730 5 of 25
reason is that the equipment cost and operating cost have a significant increase when the operation
pressure of HPC is greater than or equal to 7 bar. Therefore, the optimum operation pressure of HPC is
selected as 6 bar.
Table 3. Case studies of pressure swing distillation (PSD) with partial heat integration.
2.4. Optimization
Table 4. Economic comparison for partial and fully heat integrated pressure swing distillation (HIPSD)
processes with different distillation sequences.
condenser for HPC (QC2 ). When the material composition and the operating pressures are fixed,
there are still several variables including the number of stages in LPC and HPC (NT1 and NT2 ), and
the feed positions (NF1 , NF2 , NR ) and the reflux ratios of the LPC and HPC (RR1 and RR2 ) need to
be optimized via
Processes 2019, set the minimum TAC as the objective function. The purities of the benzene
7, 730 6 of 25 and
methanol as the constraint variables are set as 99.9 wt% via the “design specification/vary” function
whose constraint
manipulatingvariables are set is
variables asthe
99.9distillate
wt% viatothe “design
feed ratios specification/vary”
of two columns. function
Figure 3 whose
shows the
manipulating variables is the distillate to feed ratios of two columns. Figure 3 shows the sequential
sequential iterative optimization procedure for partial HIPSD process. From this figure, RR1 and RR2
iterative optimization procedure for partial HIPSD process. From this figure, RR1 and RR2 are the
are the innermost loops, NF1 , NF2 , and NR are the secondary loops, and NT1 and NT2 are the outermost
innermost loops, NF1, NF2, and NR are the secondary loops, and NT1 and NT2 are the outermost loops.
loops.TheThe optimized
optimized operating
operating parameters
parameters are acquired
are acquired through through multipleof
multiple iterations iterations of cycle
the internal the internal
to
cycleexternal
to external cycle parameter.
cycle parameter.
Give NF1
Give NT2
Give NF2
Give NT1
NO
Give RR1
Is TAC minimal with
NT1, NT2, NR fixed ?
Adjust D1 :F1 and D2 :F2
to meet the Design
Yes specifications of HPC and
LPC
NO
Is TAC minimal with
NT1 and NT2fixed ?
Is TAC minimal NO
withNT1,N T2,N F1 ,N R ,N F2 ,
Yes RR2fixed ?
NO Yes
Is TAC minimal with NT2
fixed ?
Is TAC minimal NO
with NT1,N T2,N F1,N R ,N F2
Yes fixed ?
NO Yes
Is TAC minimal ?
Figure 3. Sequential
Figure iterative
3. Sequential iterativeoptimization procedurefor
optimization procedure forthe
the partial
partial HIPSD
HIPSD process.
process.
Figure 4 shows
Figure thethe
4 shows effect
effectofofthe
the seven variablesonon
seven variables TACTAC in the
in the partial
partial HIPSDHIPSD process.
process. As can
As can be
be seen
seenininthese figures,the
these figures, theeffects
effects of each
of each variable
variable onwas
on TAC TAC was
first first decrease
decrease thenwithin
then increase increase within
limits.
The
limits. The detailed
detailedoperating
operatingparameters
parameters andand
equipment
equipmentparameters
parametersfor the
forpartial HIPSD
the partial process
HIPSD is
process is
displayed
displayed in Figure
in Figure 5. From
5. From thisthis figure,
figure, about331.283
about 331.283kW
kWenergy
energy toto the
the base
baseof
ofthe
theLPC
LPCisissupplied
supplied by
by vapor
the top the topstream
vapor stream of theand
of the HPC HPCthe and the extra
extra 39.87139.871 kW energy
kW energy is supplied
is supplied by by
thethe low-pressuresteam
low-pressure
steam to remain the energy requirement. The TAC of the partial HIPSD process is $198,891.
to remain the energy requirement. The TAC of the partial HIPSD process is $198,891. Compared with
Compared with the no heat integrated PSD process, the equipment cost and annual energy cost of
the no heat integrated PSD process, the equipment cost and annual energy cost of this process have
this process have saved $48,503 and $77,607, respectively.
saved $48,503 and $77,607,
Compared with the respectively.
partial HIPSD mode, there is no auxiliary reboiler needed in the fully HIPSD
mode. The primary reason is that the top vapor stream of the HPC just can provide the required
Processes 2019, 7, 730 7 of 25
Compared with the partial HIPSD mode, there is no auxiliary reboiler needed in the fully HIPSD
mode. The primary reason is that the top vapor stream of the HPC just can provide the required energy
Processes 2019, 7, 730 7 of 25
of the reboiler of LPC. Only one heat exchanger serves as the condenser and reboiler of two columns.
In order to make
energy the −QC1
of the reboiler equalOnly
of LPC. to Qone
R2 , heat
the RR 2 is varied
exchanger toas
serves implement
the condenserthe and
fully heat integration.
reboiler of two
columns.the
Meanwhile, Indistillate
order to make
to feed -QC1ofequal
theratios HPC to andQR2LPC
, theare
RR2adjusted
is variedtotomaintain
implement thethe fully heat
specific purities.
integration. Meanwhile, the distillate to feed ratios of HPC and LPC are adjusted
The number of the variables in this process is less than the partial HIPSD process. The six variables to maintain the
specific purities. The number of the variables in this process is less than the partial
including NT1 , NT2, NF1 , NF2 , NR , and RR1 should be optimized to obtain the best configuration by HIPSD process.
The six variables including NT1, NT2, NF1, NF2, NR, and RR1 should be optimized to obtain the best
minimizing TAC. The optimization procedure and detailed optimizing flowsheet for the fully HIPSD
configuration by minimizing TAC. The optimization procedure and detailed optimizing flowsheet
mode are displayed in the Figures 6 and 7. The results indicate that the fully HIPSD process with a
for the fully HIPSD mode are displayed in the Figures 6 and 7. The results indicate that the fully
TAC of $187,189
HIPSD processhaswith
more economic
a TAC benefit
of $187,189 has than
morethe partialbenefit
economic HIPSD process.
than the partial HIPSD process.
3.52 3.135
TAC(10 $ /y)
TAC(10 $ /y)
3.48
3.130
5
5
3.44
3.40
3.125
3.36
3.120
0.5 0.6 0.7 0.8 0.9 0.8 0.9 1.0 1.1 1.2
RR1 RR2
(a) (b)
3.11
3.09
TAC(10 $ /y)
TAC(10 $ /y)
3.10
5
5
3.08
3.09
3.07 3.08
9 10 11 12 13 14 11 12 13 14 15 16
NF1 NF2
(c) (d)
3.11
3.10
TAC(10 $ /y)
5
3.09
3.08
3.07
6 7 8 9 10 11
NR
(e)
3.18
3.081
3.15
TAC(10 $ /y)
TAC(10 $ /y)
3.078
5
5
3.12
3.075
3.09 3.072
3.06 3.069
18 19 20 21 22 23 19 20 21 22 23
NT1 NT2
(f) (g)
Figure
Figure 4. The
4. The effect
effect of the
of the seven
seven variableson
variables ontotal
total annual
annualcost
cost(TAC)
(TAC)inin
thethe
partial HIPSD
partial process:
HIPSD (a) (a)
process:
RR1, (b) RR2, (c) NF1, (d) NF2, (e) NR, (f) NT1, (g) NT2.
RR1, (b) RR2, (c) NF1, (d) NF2, (e) NR, (f) NT1, (g) NT2.
Processes 2019, 7, 730 8 of 25
Processes 2019, 7, 730 8 of 25
Processes 2019, 7, 730 8 of 25
-398.6 kW
-398.6
57.94kW
℃
57.94 ℃
D1 1269.83 kg/h
D1 1269.83 kg/h
0.604 Benzene
0.604 Benzene
0.396 Methanol
2 0.396 Methanol 2
2 2
R 970.251 kg/h
R 970.251 kg/h
0.483 Benzene HPC
0.483 Benzene
0.517 Methanol LPC HPC
0.517 Methanol LPC9
9 P1=1 bar 14 P2=6 bar
P1RR
=11bar 14 P2RR
=62bar
=0.85
=0.54
12 RRID1=0.54
1=0.47m
RRID2=0.85
2=0.39 m
F 1000 kg/h 12 ID1=0.47m ID2=0.39 m
F 1000
0.30kg/h
Benzene
0.30 Benzene
0.70 Methanol 11
0.70 Methanol 26 11
26
18 20
1837 2013
37 331.283 kW 13 403.43 kW
39.871 kW 331.283 kW
67.10 ℃ 403.43 kW
146.83 ℃
39.871
67.10kW
℃ 67.10 ℃ 146.83 ℃
Auxiliary 67.10 ℃
B2 299.579 kg/h
Auxiliary
Reboiler B2 299.579 kg/h
0.001 Methanol
Reboiler 0.001 Methanol
B1 700.421 kg/h 0.999 Benzene
B1 700.421 kg/h 0.999 Benzene
0.999 Methanol
0.999 Methanol
0.001 Benzene
0.001 Benzene
Fix P1=0.1MPa
Fix PP
1=0.1MPa
2=0.6MPa
P2=0.6MPa
Give NF1
Give NF1
Give NT2
Give NT2
Give NR
Give NR
Give NT1
Give NT1
Give NF2
Give NF2
NO
NO Give RR1
Give RR1
NO NO
NO Is TAC minimal with NO
Is TAC minimal with NT2
Is TAC minimal
fixed ?with NT2 NIs TAC
,N ,N
T1 T2 F1
minimal
,N R ,N with
F2 fixed ?
fixed ? NT1,N T2,N F1 ,N R ,N F2 fixed ?
Yes Yes
Yes Yes
NO NO
NO Is TAC minimal with
IsNTAC minimal with NO
Is TAC minimal ? T1,N T2,N F1 ,N R fixed ?
Is TAC minimal ? NT1,N T2,N F1 ,N R fixed ?
Yes Yes
Yes Yes
Get the optimal
Get,Nthe,N
N optimal Is TAC minimal with NO
T1 T2 F1,N R,N F2,
NT1,N T2and
,N F1TAC
,N R,N F2, Is TAC
NT2,Nminimal with NO
T1,N F1 fixed ?
and TAC NT2,N T1,N F1 fixed ?
Yes
Yes
Figure6.6.Sequential
Figure Sequentialiterative
iterativeoptimization
optimizationprocedure
procedurefor
forthe
thefully
fullyHIPSD
HIPSDprocess.
process.
Figure 6. Sequential iterative optimization procedure for the fully HIPSD process.
Processes 2019, 7, 730 9 of 25
Processes 2019, 7, 730 9 of 25
-370.316 kW
57.94℃
D1 1251.752 kg/h
0.603 Benzene
Processes 2019, 7, 730 0.397 Methanol 9 of 25
2 2
R 952.112 kg/h
0.479 Benzene HPC
0.521 Methanol LPC -370.316 kW
8 57.94℃
P1=1 bar 15 P2=6 bar
RR1=0.60 D1 1251.752 kg/h RR2=1.11
11 ID1=0.453m 0.603 Benzene ID2=0.40 m
F 1000 kg/h 0.397 Methanol
0.30 Benzene 2 2
0.70 Methanol
R 952.112 kg/h 11
26
0.479 Benzene HPC
0.521 Methanol LPC
18 8 20
37 P1=1 bar 15 13
P2=6 bar
343.929 kW
RR1=0.60 415.096 kW
RR2=1.11
11 67.10ID℃
1=0.453m 146.83
ID2=0.40 m ℃
F 1000 kg/h
0.30 Benzene
0.70 Methanol
B1 700.36 kg/h 11
26 B2 299.64 kg/h
0.999 Methanol 0.001 Methanol
0.001 Benzene 18 20
37 0.999 Benzene
13
343.929 kW 415.096 kW
67.10 ℃ 146.83 ℃
68 10
1.4
slope value/℃
68 8 8
1.4
Temperature/℃
66 1.2
8 8
Temperature/℃
66 6 1.2
1.0
64 6
64 6 1.0
0.8 6
62 4 4
0.6
0.8
62
Temperature
60 4 0.4 4
2 0.6 2
58 0.2
60 0.4
2
0 0.0 0 2
0 2 4 6 8 10 12 14 16 18 20 0 2
0.2 4 6 8 10 12 14 16 18 20
58 Stage Stage
0 0.0 0
0 2 4 6 8 10 12 14 16 18 20 (a) 0 2 4 6 8 10 12 14 16 18 20
1 Stage 1
Stage
Temperature slope value/℃
145 20
8 (a) 8
Temperature/℃
140 1
1 15 6
6
Temperature slope value/℃
135
145
130 8 1020 4 8
4
Temperature/℃
140
125 515
2
6 2 6
135 120
010 0
130 0 4
0 2 4 6 8 10 12 14 16 18 20 224 0 2 4 6 8 10 12 14 16 18 20 22
Stage Stage
125 5
2 2
(b)
120
0 0
0
Figure 8. Temperature distribution and their slope
0 2 4 6 8 10 12 14 16 18 20 22
distribution:
0 2 (a)
4 6high pressure
8 10 column
12 14 16 (HPC),
18 20 22 (b)
low pressure column (LPC). Stage Stage
(b)
Processes 2019, 7, 730 10 of 25
Figure 8. Temperature distribution and their slope distribution: (a) high pressure column (HPC), (b)
Processes 2019, 7, 730 10 of 25
low pressure column (LPC).
The principleofofselecting
The principle selectingTCS TCS based
based on slope
on slope criterion
criterion is thatisthe
that the temperature
temperature difference
difference between
between two adjacent
two adjacent plates and plates
theand the composition
composition variation
variation of keyof key component
component correspond,
correspond, oneone
one by by one
[24].
[24]. Therefore, this paper proposes that the composition and its slope profile
Therefore, this paper proposes that the composition and its slope profile can be used to assist the can be used to assist
the selection
selection for for
the the
TCS.TCS. Figure
Figure 9 displays
9 displays the the composition
composition distribution
distribution andand its slope
its slope distribution
distribution of
of the
the
HPC.HPC.FromFromthisthis figure,
figure, only
only one one fairly
fairly sharpcomposition
sharp compositionbreak breakappears
appearson onstage
stage 20.
20. Meanwhile,
Meanwhile,
as
as can
can bebe seen
seen in
in Figure
Figure 8b,
8b, the
the temperature
temperature slope
slope value
value ofof stage
stage 2020 is
is the
the maxima,
maxima, whenwhen removing
removing
the
the value
value ofof the
the temperature
temperature slopeslope onon stage
stage 21.
21. In
In summary,
summary, stagestage 20
20 in
in the
the HPC
HPC isis selected
selected asas the
the
temperature sensitive plate. In the next section, the rationality of selecting temperature
temperature sensitive plate. In the next section, the rationality of selecting temperature sensitive plate sensitive plate
will
will be
be confirmed
confirmed by by introducing
introducing the the different
different degree
degree disturbances
disturbances with with flowrate
flowrate andand composition
composition in in
the
the control
control structures.
structures.
1.0
Methanol 0.2 Methanol
0.8
0.1
0.6
0.0
0.4
-0.1
0.2
-0.2
0.0
0 2 4 6 8 10 12 14 16 18 20 22 0 2 4 6 8 10 12 14 16 18 20 22
Stage Stage
Figure
Figure 9.
9. Composition
Composition distribution
distribution and
and its
its slope
slope distribution
distribution of
of HPC.
HPC.
3.2. Control
3.2. Control Strategies
Strategies for
for Partial
Partial HIPSD
HIPSD Process
Process
Table
Table 5.
5. The
The size
size of
of devices
devices for
for the
the partial
partial HIPSD
HIPSD process.
process.
Equipment
Equipment
Low Pressure Column HighHigh
Low Pressure Column
Pressure Column
Pressure Column
Volume flow (m3/s)3 0.000722942 0.00073135
Volume flow (m /s) 0.000722942 0.00073135
Reflux tank Volume (m3) 3 0.4340.434 0.439
0.439
Reflux tank Volume (m )
Diameter/length (m)(m)
Diameter/length 0.65/1.30
0.65/1.30 0.65/1.31
0.65/1.31
Volume flow (m3/s)3 0.000720035 0.00138284
Volume flow (m /s) 0.000720035 0.00138284
Sump
Sump Volume (m 3) 3
Volume (m ) 0.4320.432 0.830
0.830
Diameter/length (m)(m)
Diameter/length 0.65/1.30
0.65/1.30 0.81/1.62
0.81/1.62
The basic control structure of the partial HIPSD process is built after checking the pressure of the
steady-state flowsheet. The corresponding flowsheet, control panels, and the flowsheet equation are
shown in the Figure 10. It can be seen that four kinds of controller including flow controller, pressure
Processes 2019, 7, 730 11 of 25
The basic control structure of the partial HIPSD process is built after checking the pressure of
Processes 2019, 7, 730 11 of 25
the steady-state flowsheet. The corresponding flowsheet, control panels, and the flowsheet equation
are shown in the Figure 10. It can be seen that four kinds of controller including flow controller,
pressure
controller,controller, level controller,
level controller, and temperature
and temperature controllercontroller
are used in aretheused in the flowsheet.
flowsheet. For the
For the LPC,
LPC, the feed is
feed flowrate flowrate
controlled is controlled by flow controller.
by flow controller. The levels Theof levels of the
the reflux reflux
rank andrank and the
the sump aresump
keptare
by
kept by adjusting
adjusting the stream theflowrate
stream flowrate
of the topofand
the sump.
top andThe sump. The pressure
pressure is controlled
is controlled by adjustingby adjusting
the heat
the heat removal
removal rate of the rate of the condenser.
condenser. The temperature
The temperature of the
of the 13th 13thisstage
stage is adjusted
adjusted by the by theduty
heat heatofduty
the
of the auxiliary
auxiliary reboiler reboiler
in the LPC.in theHowever,
LPC. However, for thethe
for the HPC, HPC, the pressure
pressure controller controller
has only has only afunction
a display display
function
because thebecause
pressurethe is
pressure
floatingisasfloating as theoperating
the different different operating
condition,condition, and the temperature
and the temperature controller
controller
TC2 is usedTC2 is used the
to control to control the temperature
temperature of the 20th of the by
stage 20th stage bythe
adjusting adjusting
heat duty theofheat
the duty of the
reboiler. In
reboiler. In addition, two lag time modules with 1 min are embedded in
addition, two lag time modules with 1 min are embedded in the temperature control loops. According the temperature control
loops. According to the
to the Tyreus–Luyben Tyreus–Luyben
tuning tuning
by Luyben [24], theby Luyben
tuning [24], the of
parameters tuning
TC1 and parameters of TC1 and
TC2 are determined
TC2 arerelay-feedback
by the determined byrules. the relay-feedback rules. results
The relevant tuning The relevant
are giventuning results
in Table 6. are given in Table 6.
Table 6.
Table Tuning parameters
6. Tuning parameters of
of temperature
temperature controllers
controllers TC1
TC1 and
and TC2.
TC2.
Parameter Parameter TC1TC1 TC2 TC2
Control variable
Control variable T1,13T1,13 T2,20 T2,20
Operational variable
Operational variable QAR1 QAR1 QR2 QR2
Control variable range/◦ C 0–126.62 0–248.89
Control variable range/°C 0–126.62 0–248.89
Operating variable range/GJ/h 0–0.285 0–2.91
Operating
Final gain/%/% variable range/GJ/h 0–0.285
15.36 0–2.91 10.02
Final gain/%/%
Final cycle/min 3.615.36 10.02 3.6
Gain/%/% Final cycle/min 4.80 3.6 3.6 3.13
Integration time/min
Gain/%/% 7.924.80 3.13 7.92
Integration time/min 7.92 7.92
Processes 2019, 7, 730 12 of 25
1.000
0.9995
XB2(Benzene)
0.998
XB1(Methanol)
0.9990 0.996
+5% -5%
0.994 +10% -10%
+5% -5% +20% -20%
+10% -10%
0.9985 0.992
+20% -20%
0 5 10 15 20 0 5 10 15 20
Time(h) Time(h)
126
+5% -5% +5% -5%
64 +10% -10% +10% -10%
+20% -20% +20% -20%
(℃)
T1,13(℃)
125
2,20
T
63
124
0 5 10 15 20 0 5 10 15 20
Time(h) Time(h)
1.5 1.6
+5% -5%
+10% -10% +5% -5%
+20% -20% +10% -10%
1.4 1.5 +20% -20%
QR2(GJ/h)
QR1(GJ/h)
1.3 1.4
1.2 1.3
0 5 10 15 20 0 5 10 15 20
Time(h) Time(h)
1.0000
1.000
0.9995
XB2(Benzene)
XB1(Methanol)
0.995
0.9990
0.990
0.9985
0.985
+5% -5%
0.9980 0.980 +10% -10%
+5% -5% +20% -20%
0.9975 +10% -10% 0.975
+20% -20%
0.9970 0.970
0 5 10 15 20 0 5 10 15 20
Time(h) Time(h)
65 127
+5% -5%
+5% -5%
+10% -10% +10% -10%
64 +20% -20%
126 +20% -20%
(℃)
T1,13(℃)
2,20
63 125
T
62 124
0 5 10 15 20 0 5 10 15 20
Time(h) Time(h)
1.6
1.6
1.5
1.5
1.4
QR1(GJ/h)
QR2(GJ/h)
1.3 1.4
1.2 1.3
+5% -5%
1.1 +5% -5% 1.2
+10% -10%
+10% -10%
1.0 1.1 +20% -20%
+20% -20%
0.9 1.0
0 5 10 15 20 0 5 10 15 20
Time(h) Time(h)
The results
The resultsillustrate
illustrate
thatthat
the the selection
selection of theoftemperature
the temperature
sensitivesensitive plate
plate in the HPCin isthe HPC is
reasonable.
reasonable.the
However, However,
benzene the benzene
purity cannotpurity
turn cannot
back toturn back to the
the specified specified
purity, which purity, which
is mainly is mainly
because the
because of
pressure thethe
pressure of the
HPC is not fixed HPC is operating
as the not fixed condition.
as the operating condition.
An improved Anstructure
control improved control
should be
structure should
developed be developed
to overcome to overcome this problem.
this problem.
3.2.2.
3.2.2. Control
Control Structure
Structure CS2
CS2
As shown in
As shown in Figure
Figure11, 11,a alarge
largetransient
transient deviation
deviation appeared
appeared in the
in the benzene
benzene product
product purity
purity and
and a long
a long timetime is needed
is needed to restore
to restore a newa new steady
steady statestate
when when this system
this system facesfaces a great
a great disturbance.
disturbance. The
The pressure-compensated
pressure-compensated temperature
temperature control
control structure,
structure, which
which hashasbeen
beendescribed
describedin in detail
detail and
and
confirmed
confirmed in in some
some literature
literature [1,17,25],
[1,17,25], are
are applied
applied to to improve
improve thisthis problem
problem according
according to to the
the previous
previous
control
control scheme.
scheme.
The
The first step to
first step to build
build this
thiscontrol
controlscheme
schemeisistotocompute
computethe theVLEVLE data
data of of
thethe liquid
liquid mixture
mixture on
on
stage 20 according to the steady-state results. Equation (5), which is obtained by the bubblebubble
stage 20 according to the steady-state results. Equation (5), which is obtained by the point
point temperature–pressure
temperature–pressure diagram diagram (its slope
(its slope is 5.94765),
is 5.94765), is is usedfor
used forcalculating
calculating thethe corresponding
corresponding
pressure-compensated temperature
pressure-compensated temperature T T , where the temperature T is for the
PCPC, where the temperature T is for the temperaturetemperature ononstage 20 20
stage in
the HPC,
in the HPC,thethe
pressure
pressureP isPfor thethe
is for pressure
pressure in the toptop
in the of of
thethe
HPC.HPC. ToToimplement
implement thisthis
control
controlscheme in
scheme
Aspen
in AspenDynamic,
Dynamic, thethe
third equation
third equationin the “flowsheet
in the “flowsheet equation”
equation” is used, which
is used, is shown
which is shown in Figure 12.
in Figure
The TPCTcalculated
12. The PC calculated is the input
is the inputsignal
signal of of
thethe
deadtime
deadtime block
block“dead2”
“dead2”and andthen
thenisisfed
fedtotoTC2.
TC2. The
The
improved
improved control
control structure
structure isisshown
shownin inFigure
Figure12. 12.
𝑇 = 𝑇 − (𝑃 − 6) × 5.94765. (5)
T PC = T − (P − 6) × 5.94765. (5)
The effectiveness of this improved control structure is estimated in Figure 13 for three different
The of
degrees effectiveness
disturbanceofin this improved
feed flowratecontrol structure is estimated
and composition. in Figurethat
It is noteworthy 13 for
thethree
degreedifferent
of the
degrees
disturbance and time introduced are the same as the previous control scheme. From Figure 13a, the
of disturbance in feed flowrate and composition. It is noteworthy that the degree of the
disturbance andHPC
pressure in the time are
introduced
floated in arethe
therange
same ofas 5–7
the previous
bar whencontrol scheme.
this system From
faces 5%,Figure
10%, and13a,20%
the
pressure in theThe
disturbances. HPC are floated
largest deviationin the range in
of purity of the
5–7 bottom
bar when this system
stream of HPCfaces 5%,to10%,
reduces and wt%
−0.0005 20%
disturbances.
and 0.0005 wt% Theinstead
largest deviation
of −0.007 of purity
wt% andin0.0008
the bottom
wt% stream of HPC
after 20% andreduces to −0.0005
−20% variance in wt% and
the feed
0.0005 wt% instead of −0.007 wt% and 0.0008 wt% after 20% and −20% variance
flowrate. The improvement in the methanol product purity is not obvious when the six kinds of in the feed flowrate.
The improvement
disturbance in the
take place inmethanol product
the time equal to 2purity
h. is not obvious when the six kinds of disturbance
take place in the time equal to 2 h.
Processes 2019, 7, 730 14 of 25
Processes 2019, 7, 730 14 of 25
Processes 2019, 7, 730 14 of 25
0.9992
(Benzene)
(Methanol)
0.9992
0.9990
0.9990
XB2
0.9990
XB1
0.9990
XB2
+5% -5%
XB1
+5% -5%
+10%
+5% -10%
-5% +10%
+5% -10%
-5%
0.9988
+20%
+10% -20%
-10% +20%
+10% -20%
-10%
0.9988 0.9985
+20% -20% +20% -20%
0 5 10 15 20 0.9985 0 5 10 15 20
0 5 10
Time(h) 15 20 0 5 10
Time(h) 15 20
7.0
Time(h) Time(h)
7.0
+5% -5%
64 +10% -10%
+5% -5% 6.5
64 +20% -20%
(℃)
+10% -10%
2,12,1(bar)
6.5
+20% -20%
(℃)
P P (bar )
T1,13
6.0
T1,13
63 6.0
63 +5% -5%
5.5
5.5 +10%
+5% -10%
-5%
+20%
+10% -20%
-10%
62 5.0 +20% -20%
62 0 5 10 15 20
5.0 0 5 10 15 20
0 5 10
Time(h) 15 20 0 5 10
Time(h) 15 20
1.6 Time(h) 1.6 Time(h)
1.6 +5% -5% 1.6
1.5 +10%
+5% -10%
-5%
1.5
(GJ/h)
+10% -10%
1.4 1.5
Q (GJ/h)
+20% -20%
(GJ/h)
1.4
R1
QR2
1.3 1.4
QR1
QR2
1.0000
0.9996
0.9995
XB2(Benzene)
XB1(Methanol)
0.9990 0.9992 +5% -5%
+10% -10%
0.9985 +5% -5% +20% -20%
+10% -10% 0.9988
0.9980 +20% -20%
0.9975 0.9984
0 5 10 15 20 0 5 10 15 20
Time(h) Time(h)
65 7
+5% -5%
+10% -10%
(bar)
64 +20% -20%
T1,13(℃)
2,1
P
63 +5% -5%
5 +10% -10%
+20% -20%
62
0 5 10 15 20 0 5 10 15 20
Time(h) Time(h)
1.6
1.6
1.4
QR1(GJ/h)
QR2(GJ/h)
1.4
1.2
+5% -5%
+5% -5% 1.2 +10% -10%
+10% -10%
1.0 +20% -20%
+20% -20%
0 5 10 15 20 1.0
0 5 10 15 20
Time(h) Time(h)
Figure13b
Figure 13bdemonstrates
demonstratesthe theresponses
responses forfor
5%,5%, 10%,
10%, and
and 20%20% composition
composition disturbances.
disturbances. As Ascancan
be
be seen
seen in this
in this figure,
figure, the pressure
the pressure in theinHPC
the HPC
is alsois not
alsofixed,
not fixed,
but it but it changes
changes with
with the the disturbances.
disturbances. From
From
the the benzene
benzene puritypurity
in the in the bottom
bottom streamstream
B2, theB2, the purity
purity of benzene
of benzene showsshows the value
the value of 99.97
of 99.97 wt% wt%
and
and 99.83
99.83 wt% forwt% 20% forand
20% −20%anddisturbances
−20% disturbances in the composition.
in the composition. Compared Compared
with basicwith basic
control control
structure,
structure,
the transient thedeviation
transientdecreases
deviation decreases
from from
2.8 to 0.17 wt%.2.8 However,
to 0.17 wt%.the However, the methanol
methanol purity in the sixpurity
cases in
is
the six cases
similar is similar
to those found in to the
those found in
previous the previous
control structure. control structure.
These results
These results illustrate
illustrate that
that the
the pressure-compensated
pressure-compensated temperature
temperature control
control structure
structure hashas aa
significantimprovement
significant improvementininthe thetransient
transient deviation
deviation in the
in the xB2x(benzene),
B2 (benzene), butbut
thethe effect
effect in improving
in improving the
the purity
purity of methanol
of methanol is notis obvious
not obvious
when when the same
the same disturbances
disturbances occuroccur at 2 h.
at 2 h.
3.2.3.
3.2.3. Control
Control Structure
Structure CS3
CS3
The
Thepurities
puritiesofofthethemethanol
methanol andand
benzene cannot
benzene be driven
cannot to theto
be driven initial setpoint
the initial when the
setpoint system
when the
encounters
system encountersthe disturbances with different
the disturbances degrees degrees
with different in the feed flowrate
in the and composition
feed flowrate even though
and composition even
the
thoughcontrol
the structure CS2 can CS2
control structure obtaincana obtain
robust acontrol.
robust A composition/temperature
control. A composition/temperature cascade cascade
control
structure, which integrates
control structure, the advantages
which integrates of both composition
the advantages control and temperature
of both composition control and control, was
temperature
put forward to improve the control structure.
control, was put forward to improve the control structure.
To
Toset
setup upthe
thecomposition/temperature
composition/temperaturecontrol controlstructure,
structure,two twocomposition
compositioncontrollers
controllers(CC1
(CC1 and
and
CC2)
CC2) withwith 33 min
min deadtime,
deadtime, which
which detect
detect the
the content
content of of methanol
methanol and and benzene
benzene in in the
the product
product streams
streams
BB11 and
and BB22,, were
were inserted
inserted inin the
the loops
loops of of TC1
TC1 and
and TC2.
TC2. In In the
the composition/temperature
composition/temperature cascade cascade loop,
loop,
the
the composition and temperature controllers are the major and assistant controllers, respectively. ItIt
composition and temperature controllers are the major and assistant controllers, respectively.
is
is interesting
interesting to to note
note that
that the
the operating
operating variables
variables ofof the
the temperature
temperature controllers
controllers areare the
the same
same asas the
the
previous control structure, but the input signal which is the output signal
previous control structure, but the input signal which is the output signal of the composition of the composition controller
is not a constant.
controller is not aThe tuningThe
constant. parameters of the majorofcontrollers
tuning parameters the majorCC1 and CC2
controllers need
CC1 andtoCC2
be determined
need to be
by using the by
determined Tyreus–Luyben tuning rules and
using the Tyreus–Luyben rerunning
tuning rules and the rerunning
relay-feedback tests. The ultimate
the relay-feedback tests.gain
The
and period of for CC1 are 14.96 and 25.08 min, and for CC2 are 15.40 and
ultimate gain and period of for CC1 are 14.96 and 25.08 min, and for CC2 are 15.40 and 18.48 min. 18.48 min. Figure 14 shows
control
Figure 14 structure
shows CS3 andstructure
control control-panels.
CS3 and control-panels.
Processes
Processes 2019,
2019, 7, 730
7, 730 16 of
16 25
of 25
Processes 2019, 7, 730 16 of 25
(a)
(a) Control structureCS3
Control structure CS3
(b) Control-panel
(b) Control-panel
Figure
Figure
Figure 14.Control
14.
14. Controlstructure
Control structure CS3
structure CS3and
CS3 andcontrol-panel.
and control-panel.
control-panel.
Figure Figure
Figure 1515 15displays
displays
displays the the
the dynamicresponses
dynamic
dynamic responses of
responses of the
the control
controlstructure
control structure
structure encountering
encountering
encountering thethedifferent
the different
different
degrees
degrees of of of disturbance.
disturbance. The solid lines with black, green, and red colors signify the feed flowrate and and
degrees disturbance.The Thesolid
solidlines
lineswith
with black, green, and
black, green, andredredcolors
colorssignify
signifythethefeed
feed flowrate
flowrate and
the methanol composition increases of 5%, 10%, and 20%, and the corresponding dashed lines signify
thethe
methanol
methanol composition
compositionincreases
increasesofof5%, 5%,10%,
10%, and 20%,
20%, and
andthethecorresponding
correspondingdashed dashed lines
lines signify
signify
the feed flowrate and the methanol composition decreases of 5%, 10%, and 20% in Figure 15a,b. As
thethe
feedfeed flowrate
flowrate and and
thethe methanol
methanol composition
composition decreases
decreases ofof
5%, 5%, 10%,
10%, and
and 20%20%inin Figure
Figure 15a,b.AsAscan
15a,b.
can be seen from Figure 15, the temperatures on the two sensitive plates are not maintained at the
be can
seen befrom
seenFigure
from Figure 15, the temperatures ontwo
thesensitive
two sensitive plates aremaintained
initial temperatures of 63.31 and 123.29 °C, but change with the different degrees of disturbance.the
15, the temperatures on the plates are not not maintained at at the
Theinitial
initial temperatures
temperatures of 63.31 and◦123.29 °C, but change with the different degrees of disturbance. The
larger the variation in the flowrate or composition, the larger the deviation in the initial value. Thelarger
of 63.31 and 123.29 C, but change with the different degrees of disturbance. The
larger the
the variation
purities of variation
intwo in
the products the
flowrate have flowrate or
or composition, composition,
small transientthedeviationthe
larger the larger the
deviation
and deviation in
in the initial
can be brought the initial
back tovalue. value.
The purities
the setting The
point of
twopurities
when of
products two
this have products
controlsmall have faces
transient
structure small transient
deviation
the deviation
and
±5%, ±10%, can
andbe and can
brought
±20% beback
brought
disturbances to in back
the to the
setting
the feed setting
point
flowrate point
when
and this
when this
composition.control The structure
results faces
illustrate the
that ±5%,
the ±10%,
control and ±20%
structure CS3disturbances
can provide
control structure faces the ±5%, ±10%, and ±20% disturbances in the feed flowrate and composition. in
a the
good feed
dynamicflowrate
controland
Thecomposition.
performance.
results The results illustrate that the control structure CS3 can provide a good dynamic control
illustrate that the control structure CS3 can provide a good dynamic control performance.
performance. 0.9994
0.9994 0.9992
XB2(Benzene)
XB1(Methanol)
0.9992
0.9992
XB2(Benzene)
XB1(Methanol)
0.9992 0.9990
0.9990
+5% -5% 0.9990 +5% -5%
0.9990 +10% -10% 0.9988 +10% -10%
0.9988
+20% -20% +20% -20%
+5% -5% +5% -5%
0 5 10 15
+10% -10% 20 0.99880 5 10 15 +10% 20
-10%
0.9988
Time(h) +20% -20% Time(h) +20% -20%
130
0 5 10 15 20 0 5 10 15 20
+5% -5%
64 Time(h) +10% -10% Time(h)
130
+20% -20%
T ) (℃)
T1,13(℃)
125
+5% -5%
64
2,20 2,20
+10% -10%
63 +20% -20%
T1,13(℃)
(℃
62 +5% -5%
0 5 10 15 20 1200 5 10 15 +10% 20
-10%
Time(h) Time(h) +20% -20%
1.6 1.6
62
0 5 10 15
+5% -5% 20 0 5 10 15 20
1.5 +10% -10%
Time(h) 1.5 Time(h)
QR1(GJ/h)
+20% -20%
QR2(GJ/h)
1.6 1.6
1.4
+5% -5%
1.5 1.3 +10% -10% 1.4
1.5
QR1(GJ/h)
+20% -20%
QR2(GJ/h)
+5% -5%
1.4 1.2
+10% -10%
1.3 +20% -20%
1.3 1.1 1.4
0 5 10 15 20 0 5 10 15 +5% -5%20
1.2 Time(h) Time(h) +10% -10%
1.3 +20% -20%
1.1
0 5
(a) ±5%,10±10%, and
15
±20%20feed flowrate disturbance.
0 5 10 15 20
Time(h) Time(h)
0.9996
0.9996 +5% -5%
0.9994
XB2(Benzene)
+10% -10%
XB1(Methanol)
0.9994 +20% -20%
0.9992 +5% -5%
0.9992
+10% -10%
+20% -20%
0.9990 0.9990
0.9988 0.9988
0.9986
0.9986
0 5 10 15 20 0 5 10 15 20
Time(h) Time(h)
65
130
+5% -5%
+10% -10%
(℃)
64 +20% -20% 125
T1,13(℃)
2,20
120
T
63 +5% -5%
+10% -10%
115 +20% -20%
62
0 5 10 15 20 0 5 10 15 20
Time(h) Time(h)
1.6
1.6
1.4
QR1(GJ/h)
QR2(GJ/h)
1.4
1.2
+5% -5% +5% -5%
+10% -10% 1.2 +10% -10%
1.0 +20% -20% +20% -20%
0 5 10 15 20 1.0
0 5 10 15 20
Time(h) Time(h)
The disturbances
disturbanceswith withdifferent
differentdegrees
degrees were
were utilized
utilized to test
to test the the control
control performance
performance for theforfully
the
fully
HIPSD HIPSD
process.process.
Figure Figure 17a gives
17a gives the dynamic
the dynamic responses
responses whenwhen 5%, 10%,
5%, 10%, 20% change
20% step step change
in theinfeed
the
feed flowrate
flowrate occuroccur
at theattime
the equal
time equal
to 2 h.toThe2 h.temperatures
The temperatures on two onTCStwoareTCS are maintained
maintained to thetosetthe set
point
pointand
(63.5 and ◦124.8
(63.5124.8 C) in °C) in approximately
approximately 2 h after2 hthe
after the flowrate
flowrate disturbances
disturbances occurred occurred at 2 h.
at 2 h. This This is
is caused
caused by the controllers
by the controllers “TC1”and “TC1”and
“TC2” that “TC2” that canthe
can readjust readjust
RR1 and the
QR2 RR1forand QR2 fora achieving
achieving a new
new steady-state
steady-state
condition. The condition. The product
product purities purities
of five cases (±5%, of±10%,
five andcases−20%(±5%, ±10%,disturbances)
flowrate and −20% flowrate
are held
disturbances)
quite close to thearespecified
held quite closebut
purity, to the
the specified
transient purity,
deviation butinthethetransient
purity ofdeviation
benzene isin0.9thewt%purity
when of
benzene
the flowrateis 0.9 wt% when
changes from the1000flowrate changes
to 1200 kg/h. Thefrom
main1000
culpritto 1200
is thatkg/h. The mainofculprit
the increase the feed is flowrate
that the
increase
leads to aofsharp
the feed flowrate
increase in theleads to a of
pressure sharp
HPCincrease
and resultsin the pressure
in the of HPC of
concentration and results on
methanol in the
concentration
sensitive stage,ofbut methanol on the sensitive
the temperature on stagestage, but the temperature
20 is controlled at a constant. on stage 20 is controlled at a
constant.
Figure 17b illustrates the dynamic responses of the control structure for the feed composition
Figure 17b
disturbance withillustrates the dynamic
three different degrees. responses
The solid ofand
the dashed
control lines
structure for positive
are for the feedandcomposition
negative
disturbance
change, respectively. The black, green, and red lines demonstrate the 3%, 5%, and 10% stepnegative
with three different degrees. The solid and dashed lines are for positive and change
change, respectively.
in the feed composition. The Theblack, green,ofand
results the red lines demonstrate
response for the six cases the 3%, 5%, andto10%
are similar the step change
six cases in
in the feed
Figure 17a. composition.
The larger theThe results
variance inof
thethe response for
composition, thethe six cases
larger are similar
the deviation to the
in the six cases
benzene in
purity.
Figure 17a. The larger the variance in the composition, the larger the
The deviation (0.8 wt%) occurred when the methanol concentration decreases from 70 to 63 wt%. deviation in the benzene purity.
The deviation
When the(0.8 wt%) occurred
methanol when the
concentration methanol
changes moreconcentration decreasesoffrom
than 7%, the flowrate 70 to 63 wt%.
the recycling stream
Whentothe
increases themethanol
point which concentration changes
caused the liquid more
level thanreflux
of the 7%, the
tankflowrate
in the LPCof the recycling
to exceed 100%stream
and
increases
the system towas
the point
forcedwhich
to shut caused
down.the liquid
This level of
is caused bythe reflux
more tank in
benzene inthe
theLPC
feedto exceedmaking
stream 100% and the
the system
product wasbottom
in the forced of toHPC
shut increase
down. This andisthecaused by more
pressure benzene in the feed stream making the
rise rapidly.
product
Thesein the bottom
results of HPC
illustrate increase
that and theofpressure
the selection rise rapidly.
the temperature sensitive plate is suitable, and this
These
control resultscan
structure illustrate
implementthat the
theselection of the temperature
robust control. However, it sensitive plate is suitable,
cannot maintain the purity andof this
the
control structure can implement the robust control. However, it cannot maintain the purity of the
Processes 2019, 7, 730 19 of 25
Processes 2019, 7, 730 19 of 25
1.000
0.9995
XB2(Benzene)
XB1(Methanol)
0.9990 0.996
+5% -5%
0.9985 +10% -10%
+5% -5%
+20% -20%
+10% -10% 0.992
0.9980
+20% -20%
0 5 10 15 20 0 5 10 15 20
Time(h) Time(h)
65 126
+5% -5%
+10% -10%
+20% -20%
(℃)
64
T1,13(℃)
2,20
125
T
63 +5% -5%
+10% -10%
+20% -20%
62 124
0 5 10 15 20 0 5 10 15 20
Time(h) Time(h)
0.8
+5% -5% 1.6
+10% -10%
0.7
+20% -20%
QR2(GJ/h)
RR1
1.5
0.6
0.5 1.4
+5% -5%
+10% -10%
0.4 +20% -20%
0 5 10 15 20 1.3
0 5 10 15 20
Time(h) Time(h)
0.9990 0.996
+3% -3%
0.9987 +6% -6%
0.992 +10% -10%
+3% -3%
+6% -6%
0.9984
+10% -10%
0.988
0 5 10 15 20 0 5 10 15 20
Time(h) Time(h)
65 126
+3% -3%
+6% -6%
+10% -10%
(℃)
T1,13(℃)
64
2,20
125
T
63 +3% -3%
+6% -6%
+10% -10%
62 124
0 5 10 15 20 0 5 10 15 20
Time(h) Time(h)
0.8
+3% -3% 1.6
+6% -6%
0.7
+10% -10%
QR2(GJ/h)
RR1
1.5
0.6
of the HPC,
Processes and
2019, 7, 730the pressure-compensated temperature TPC . In Aspen Dynamic, TPC can be calculated
20 of 25
by the third equation in the “flowsheet equation” function. As shown in Figure 18, the input signal of
input signal
the “dead 2”of the “dead
block is the T2” block is the TPC. Note that the output signal of the temperature controller
PC . Note that the output signal of the temperature controller TC2, which is
TC2, which is the ratio of reboiler
the ratio of reboiler heat duty 1.49297 heatGJ/h
dutyto1.49297 GJ/h to
the flowrate the kg/h,
1000 flowrate 1000 kg/h,
is 0.0015 is 0.0015
and not and not
connected the
connected
reboiler, butthe
thereboiler, butblock.
multiplier the multiplier block. The
The effectiveness effectiveness
of this improvedofstructure
this improved
will be structure
illustratedwill be
in the
illustrated in
next section. the next section.
𝑇 (P −
=𝑇T−−(𝑃
TPC = 6) ×
− 6) × 7.5722.
7.5722. (6)
(6)
The disturbances
The disturbancesin inthe
theflowrate
flowrateand
andcomposition
compositionwith several
with different
several degrees
different were
degrees usedused
were to test
to
the dynamic
test performance.
the dynamic TheThe
performance. obtained results
obtained are demonstrated
results in Figure
are demonstrated 19. Figure
in Figure 19a 19a
19. Figure shows the
shows
response
the of the
response of flowrate withwith
the flowrate 5%, 10%, and 20%
5%, 10%, and step
20% change, and Figure
step change, 19b shows
and Figure the response
19b shows of the
the response
methanol component with 3%, 6%, and 10% step change.
of the methanol component with 3%, 6%, and 10% step change.
1.000
0.9995
XB2(Benzene)
XB1(Methanol)
0.9990 0.999
+5% -5%
0.9985 +10% -10%
0.998 +20% -20%
+5% -5%
+10% -10%
0.9980
+20% -20%
0.997
0 5 10 15 20 0 5 10 15 20
Time(h) Time(h)
65 7
+5% -5%
+10% -10%
64 +20% -20%
T1,13(℃)
(bar)
6
2,1
P
63
+5% -5%
+10% -10%
+20% -20%
62 5
0 5 10 15 20 0 5 10 15 20
Time(h) Time(h)
0.8
+5% -5% 1.7
+10% -10%
0.7 1.6
+20% -20%
QR2(GJ/h)
RR1
0.6 1.5
1.4
0.5 +5% -5%
1.3 +10% -10%
+20% -20%
0.4 1.2
0 5 10 15 20 0 5 10 15 20
Time(h) Time(h)
0.9993
XB2(Benzene)
XB1(Methanol)
0.9990 0.9994
0.9987
+3% -3%
+3% -3%
0.9992 +6% -6%
+6% -6%
0.9984 +10% -10%
+10% -10%
0 5 10 15 20 0 5 10 15 20
Time(h) Time(h)
65 7
+3% -3%
+6% -6%
+10% -10% +3% -3%
64
T1,13(℃)
(bar)
+6% -6%
6 +10% -10%
2,1
P
63
62 5
0 5 10 15 20 0 5 10 15 20
Time(h) Time(h)
0.8 1.7
+3% -3%
+6% -6% 1.6
0.7
+10% -10%
QR2(GJ/h)
1.5
RR1
0.6
1.4
0.5 +3% -3%
1.3 +6% -6%
+10% -10%
0.4 1.2
0 5 10 15 20 0 5 10 15 20
Time(h) Time(h)
In the next part, we will test the dynamic performances by introducing the same disturbances in
flowrate and composition at 2 h to verify the effectiveness of the improved structure.
The responses for the different degree of disturbances are displayed in Figure 21. The different
color lines stand for the same variation in the feed flowrate and composition in the basic control
structure. From Figure 21, the product purity of methanol can be brought back to the specified value
after encountering feed flowrate changes of 50, 100, and 200 kg/h and the composition changes of 2.1,
4.2, and 7 wt% methanol. The larger the disturbances in the flowrate and composition, the larger the
transient deviation in the initial purity. However, the transient deviation is smaller than those in the
previous structure control. Meanwhile, the time to reach a new steady state reduces from 4 to 1 h.
Processes 2019, 7, 730 23 of 25
Processes 2019, 7, 730 23 of 25
1.000
0.9995
XB2(Benzene)
XB1(Methanol)
0.999
0.9990
+5% -5%
+10% -10%
0.998 +20% -20%
0.9985 +5% -5%
+10% -10%
+20% -20%
0.9980 0.997
0 5 10 15 20 0 5 10 15 20
Time(h) Time(h)
65
+5% -5%
+10% -10% 132 +5% -5%
+20% -20% +10% -10%
+20% -20%
(℃)
64
T1,13(℃)
128
2,20
T
63 124
120
62
0 5 10 15 20 0 5 10 15 20
Time(h) Time(h)
0.8
1.7
0.7
1.6
QR2(GJ/h)
0.6
RR1
1.5
0.5
1.4
+5% -5% +5% -5%
0.4 +10% -10% +10% -10%
1.3
+20% -20% +20% -20%
0.3 1.2
0 5 10 15 20 0 5 10 15 20
Time(h) Time(h)
0.9993
XB2(Benzene)
XB1(Methanol)
0.9990 0.9994
0.9987
+3% -3%
+3% -3%
0.9992 +6% -6%
+6% -6%
0.9984 +10% -10%
+10% -10%
0 5 10 15 20 0 5 10 15 20
Time(h) Time(h)
65
+3% -3%
+3% -3% +6% -6%
+6% -6% 130 +10% -10%
+10% -10%
(℃)
64
T1,13(℃)
2,20
125
T
63
120
62
0 5 10 15 20 0 5 10 15 20
Time(h) Time(h)
0.8 1.7
+3% -3%
+6% -6% 1.6
0.7
+10% -10%
QR2(GJ/h)
1.5
RR1
0.6
1.4
0.5 +3% -3%
1.3 +6% -6%
+10% -10%
0.4 1.2
0 5 10 15 20 0 5 10 15 20
Time(h) Time(h)
(b) ±5%, ±6%, and ±10% feed composition disturbance.
Figure 21.
Figure 21. The
The dynamic
dynamic responses
responses of
of control
control structure
structure CS3.
CS3.
4. Conclusions
Compared with previous control structure, the benzene purity is almost unchanged when the
system faces the disturbances in the feed flowrate and composition. The methanol purity deviation
The PSD with two energy-saving modes for the separation of the benzene/methanol azeotropic
in the new steady state is maintained within acceptable range when there is a 3%, 6%, and 10% step
system was studied by Aspen Plus and Aspen Dynamics. We contrasted the economy and
change in the methanol composition. This reveals that the control structure CS3 can provide a good
controllability of the PSD with two heat integration modes. The optimization results show that the
dynamic control performance.
fully HIPSD process with a TAC of $187,189 has some economic benefit compared with the partial
HIPSD
4. process with a TAC of $198,891.
Conclusions
The temperature profile in the HPC rose rapidly near the bottom of this column and the
The PSD
maximum with two energy-saving
of temperature modes forinthe
slope value occurred theseparation
last stage, of the benzene/methanol
which is the bottom of theazeotropic
HPC. The
system was studied by Aspen Plus and Aspen Dynamics. We contrasted the economy andsensitive
composition slope profile was proposed to assist with the selection of the temperature controllability
plate.
of the PSD to
According with
thetwo
two heat integration
criterions, stagesmodes. The optimization
13 and stage 20 in the LPCresults
and HPCshow that
were the fully
chosen HIPSD
as the TCS.
Several control structures were developed to evaluate the industrial application and check the
rationality of the selection of the temperature sensitive plate. The different degrees of disturbances in
Processes 2019, 7, 730 24 of 25
process with a TAC of $187,189 has some economic benefit compared with the partial HIPSD process
with a TAC of $198,891.
The temperature profile in the HPC rose rapidly near the bottom of this column and the maximum
of temperature slope value occurred in the last stage, which is the bottom of the HPC. The composition
slope profile was proposed to assist with the selection of the temperature sensitive plate. According
to the two criterions, stages 13 and stage 20 in the LPC and HPC were chosen as the TCS. Several
control structures were developed to evaluate the industrial application and check the rationality of the
selection of the temperature sensitive plate. The different degrees of disturbances in the feed flowrate
and composition were introduced in two processes at 2 h. The results illustrate that the selection of
the TCS is reasonable and the composition/temperature control structure can provide rust control
performance for the PSD with two heat integrated modes. Compared with the partial HIPSD mode, the
fully HIPSD mode can handle much smaller composition disturbance despite little economic benefit.
These works can provide a reference for selecting TCS, especially for azeotropic systems whose
maximum value of temperature slope profile occurs in the bottom of the column. Meanwhile, the
findings can also help engineers to weigh the economy against controllability for selection of the
suitable separating process.
Author Contributions: Conception and design: Z.Z., L.L.; Process simulation and optimization: L.L., Z.Z.;
Establishment of control structure: H.L., H.H.; Analysis and interpretation: L.L., H.H.; Data collection: H.L., H.H.;
Writing the article: L.L.; Critical revision of the article: L.L., Z.Z., H.L., H.H.; Final approval of the article: L.L.,
Z.Z., H.L., H.H.
Funding: This research was funded by the project of science and technology research program of Chongqing
Education Commission of China (No. KJQN201901429).
Acknowledgments: The authors gratefully acknowledge the anonymous reviewers for the insightful comments
and suggestions, which improved the quality of this paper to a large extent. Additionally, the authors also
gratefully acknowledge the financial support by the project of science and technology research program of
Chongqing Education Commission of China (No. KJQN201901429).
Conflicts of Interest: The authors declare no conflicts of interest.
References
1. Luyben, W.L. Design and Control of a Fully Heat-Integrated Pressure-Swing Azeotropic Distillation System.
Ind. Eng. Chem. Res. 2008, 47, 2681–2695. [CrossRef]
2. Wang, Y.; Zhang, Z.; Zhang, H.; Zhang, Q. Control of Heat Integrated Pressure-Swing-Distillation Process for
Separating Azeotropic Mixture of Tetrahydrofuran and Methanol. Ind. Eng. Chem. Res. 2015, 54, 1646–1655.
[CrossRef]
3. Zhu, Z.; Wang, L.; Ma, Y.; Wang, W.; Wang, Y. Separating an Azeotropic Mixture of Toluene and Ethanol Via
Heat Integration Pressure Swing Distillation. Comput. Chem. Eng. 2015, 76, 137–149. [CrossRef]
4. Lv, L.Z.L.; Liu, H.; Li, H.; Sun, S. Comparison of Continuous Homogenous Azeotropic and Pressure-Swing
Distillation for a Minimum Azeotropic System Ethyl Acetate N-Hexane Separation. Chin. J. Chem. Eng. 2018,
26, 2023–2033.
5. Li, X.; Zhao, Y.; Qin, B.; Zhang, X.; Wang, Y.; Zhu, Z. Optimization of Pressure-Swing Batch Distillation with
and without Heat Integration for Separating Dichloromethane/Methanol Azeotrope Based on Minimum
Total Annual Cost. Ind. Eng. Chem. Res. 2017, 56, 4104–4112. [CrossRef]
6. Zhang, Z.; Zhang, Q.; Li, G.; Liu, M.; Gao, J. Design and Control of Methyl Acetate-Methanol Separation Via
Heat-Integrated Pressure-Swing Distillation. Chin. J. Chem. Eng. 2016, 24, 1584–1599. [CrossRef]
7. Zhao, L.; Lyu, X.; Wang, W.; Shan, J.; Qiu, T. Comparison of Heterogeneous Azeotropic Distillation and
Extractive Distillation Methods for Ternary Azeotrope Ethanol/Toluene/Water Separation. Comput. Chem.
Eng. 2017, 100, 27–37. [CrossRef]
8. Li, W.; Zhong, L.; He, Y.; Meng, J.; Yao, F.; Guo, Y.; Xu, C. Multiple Steady-States Analysis and Unstable
Operating Point Stabilization in Homogeneous Azeotropic Distillation with Intermediate Entrainer. Ind.
Eng. Chem. Res. 2015, 54, 7668–7686. [CrossRef]
Processes 2019, 7, 730 25 of 25
9. Lladosa, E.; Monton, J.B.; Burguet, M. Separation of di-N-Propyl ether and n-Propyl Alcohol by Extractive
Distillation and Pressure-Swing Distillation: Computer Simulation and Economic Optimization. Chem. Eng.
Process. Process Intensif. 2011, 50, 1266–1274. [CrossRef]
10. Shen, W.; Benyounes, H.; Gerbaud, V. Extension of Thermodynamic Insights on Batch Extractive Distillation
to Continuous Operation. 1. Azeotropic Mixtures with a Heavy Entrainer. Ind. Eng. Chem. Res. 2013, 52,
4606–4622. [CrossRef]
11. Shen, W.; Dong, L.; Wei, S.A.; Li, J.; Benyounes, H.; You, X.; Gerbaud, V. Systematic Design of an Extractive
Distillation for Maximum-Boiling Azeotropes with Heavy Entrainers. Aiche J. 2015, 61, 3898–3910. [CrossRef]
12. Shen, W.; Benyounes, H.; Gerbaud, V. Extractive Distillation: Recent Advances in Operation Strategies. Rev.
Chem. Eng. 2015, 31, 13–26. [CrossRef]
13. Shirsat, S.P. Separation of Isobutyl Alcohol and Isobutyl Acetate by Extractive Distillation and Pressure-Swing
Distillation: Simulation and Optimization. Sep. Purif. Technol. 2015, 154, 366. [CrossRef]
14. Wang, Y.; Cui, P.; Ma, Y.; Zhang, Z. Extractive Distillation and Pressure-Swing Distillation for THF/Ethanol
Separation. J. Chem. Technol. Biotechnol. 2015, 90, 1463–1472. [CrossRef]
15. Fulgueras, A.M.; Poudel, J.; Kim, D.S.; Cho, J. Optimization Study of Pressure-Swing Distillation for the
Separation Process of a Maximum-Boiling Azeotropic System of Water-Ethylenediamine. Korean J. Chem.
Eng. 2015, 33, 46–56. [CrossRef]
16. Suo, X.; Ye, Q.; Li, R.; Dai, X.; Yu, H. The Partial Heat-Integrated Pressure-Swing Reactive Distillation Process
for Transesterification of Methyl Acetate with Isopropanol. Chem Eng. Process. Process Intensif. 2016, 107,
42–57. [CrossRef]
17. Luo, B.; Feng, H.; Sun, D.; Zhong, X. Control of Fully Heat-Integrated Pressure Swing Distillation for
Separating Isobutyl Alcohol and Isobutyl Acetate. Chem. Eng. Process. Process Intensif. 2016, 110, 9–20.
[CrossRef]
18. Luyben, W.L. Methanol/Trimethoxysilane Azeotrope Separation Using Pressure-Swing Distillation. Ind. Eng.
Chem. Res. 2014, 53, 5590–5597. [CrossRef]
19. Wang, Y.; Cui, P.; Zhang, Z. Heat-Integrated Pressure-Swing-Distillation Process for Separation of
Tetrahydrofuran/Methanol with Different Feed Compositions. Ind. Eng. Chem. Res. 2014, 53, 7186–7194.
[CrossRef]
20. Zhang, H. Heat Integration Process for Separating System of Toluene and N-Butyl Alcohol with Pressure Swing
Distillation; University of Science and Technology: Qingdao, China, 2015.
21. Zhang, Q. Dynamics and Control of Pressure Swing Distillation Process for Azeotropic Systems of
Methanol/Ethanol-Tetrahydrofuran Qingdao; University of Science and Technology: Qingdao, China, 2015.
22. Hosgor, E.; Kucuk, T.; Oksal, I.N.; Kaymak, D.B. Design and Control of Distillation Processes for
Methanol–Chloroform Separation. Comput. Chem. Eng. 2014, 67, 166–177. [CrossRef]
23. Zhang, Q.; Peng, J.; Zhang, K. Separation of an Azeotropic Mixture of Dimethyl Carbonate and Methanol Via
Partial Heat Integration Pressure Swing Distillation. Asia Pac. J. Chem. Eng. 2017, 12, 50–64. [CrossRef]
24. Luyben, W.L. Design and Control of Distillation Systems for Separating Azeotropes; John Wiley & Son: Hoboken,
NJ, USA, 2008.
25. Cao, Y.; Li, M.; Wang, Y.; Zhao, T.; Li, X.; Zhu, Z.; Wang, Y. Effect of Feed Temperature on Economics and
Controllability of Pressure-Swing Distillation for Separating Binary Azeotrope. Chem. Eng. Process. Process
Intensif. 2016, 110, 160–171. [CrossRef]
© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).