You are on page 1of 11

Chemical Engineering & Processing: Process Intensification 144 (2019) 107651

Contents lists available at ScienceDirect

Chemical Engineering & Processing: Process Intensification


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

Separation of propylene oxide-methanol-water mixture via enhanced T


extractive distillation: Design and control

Zhishan Zhang , Chao Wang, Chao Guang, Chen Wang
College of Chemical and Environmental Engineering, Shandong University of Science and Technology, Qingdao 266590, China

A R T I C LE I N FO A B S T R A C T

Keywords: The ternary azeotropic mixture of propylene oxide-methanol-water generally occurs in the direct oxidation of
Extractive distillation hydrogen peroxide and needs to be separated effectively. This paper explores two energy-efficient configurations
Process intensification with water as solvent, namely extractive dividing-wall column (EDWC) and side-stream extractive distillation
Process control (SSED), in terms of the potential economic advantages compared to conventional extractive distillation (CED)
Azeotropic mixtures
and the dynamic control. Several steady-state designs with the minimum total annual cost (TAC) by using a
Propylene oxide-methanol-water
sequential iterative optimization method are achieved and compared. Regarding the controllability, four control
structures for the SSED are developed and examined via introducing the feed flow and composition disturbances
as well as one control structure for the EDWC. Similar to the EDWC, the SSED can be well controllable via an
optimal control structure relying on only temperature measurements.

1. Introduction component azeotropic mixtures. Anokhina and Timoshenko [21] com-


pared the energy efficiency of traditional and partially thermally cou-
Propylene oxide (PO) plays an important role in organic chemical pling extractive distillation flowsheets and proposed a few evaluation
materials and fine chemicals. The epoxidation of propylene, using hy- criterions for the effectiveness of complex extractive distillation column
drogen peroxide as oxidant and methanol as solvent is a common PO via investigating the separation of seven different azeotropic mixtures.
production technology named HPPO. During this process, it is difficult Wang et al. [22] studied the steady simulation and dynamic control of
for separating the high purity PO product from the mixture of PO-me- the EDWC for separating a binary azeotrope of acetonitrile-N-propanol.
thanol-water due to the presence of one minimum-boiling azeotrope of Zhao et al. [23] investigated two thermally coupled ternary extractive
methanol-PO. Hence, some special distillation methods have to be used, distillation processes for separating tetrahydrofuran-ethanol-water
such as pressure-swing distillation [1–5], extractive distillation [6–11] ternary azeotropic mixtures. Tututi-Avila et al. [24] firstly investigated
and azeotropic distillation [12–16]. an energy-efficient extractive distillation configuration intensified with
Among these alternatives, extractive distillation is an attractive side-stream (SSED), which is illustrated with the separation of three
option since no extra substance is introduced when selecting water as binary azeotropes. Wang et al. [25] proposed and explored ten ther-
solvent. Currently, some extractive distillation enhanced configurations mally coupled extractive distillation separation sequences based on
via dividing-wall column and side-stream column have been widely SSED and EDWC for separating methanol-acetonitrile-benzene ternary
studied for achieving the efficient separation of different systems. Kiss mixture with three binary azeotropes. In addition, Wang et al. [26]
and Suszwalak [17] explored the application of the dividing-wall applied the side-stream technology to pressure-swing distillation and
column in extractive distillation (EDWC) for enhancing ethanol dehy- proposed a few energy-enhanced schemes for the binary separation of
dration. Kiss and Ignat [18] developed a thermally integrated extractive ethyl acetate-ethanol, methanol-chloroform, and ethylenediamine-
distillation configuration in the downstream bioethanol fuel produc- water.
tion. Modla [19] investigated a comparison of conventional, thermally The application of any high efficient separation configuration to a
integrated and dividing-wall extractive distillation processes for se- new system would likely be dismissed as far as implementation due to
parating a two-component azeotrope of methanol/toluene. Timoshenko operational and control issues. The optimal design and operational
et al. [20] explored the practicability of diverse partially thermal-cou- control for EDWC have been investigated extensively for a long period.
pled extractive distillation configurations for splitting several triple- Zhang et al. [27] studied the economic and dynamic performances of


Corresponding author.
E-mail address: tjzza@163.com (Z. Zhang).

https://doi.org/10.1016/j.cep.2019.107651
Received 20 June 2019; Received in revised form 10 August 2019; Accepted 30 August 2019
Available online 31 August 2019
0255-2701/ © 2019 Elsevier B.V. All rights reserved.
Z. Zhang, et al. Chemical Engineering & Processing: Process Intensification 144 (2019) 107651

Nomenclature NFS Solvent location


NLS Side liquid location
AC Heat transfer area of condenser [m2] NV Side vapor location
AR Heat transfer area of reboiler [m2] PO Propylene oxide
CED Conventional extractive distillation QC Condenser duty [kW]
CSCinst Installed cost of column shell [$] QR Reboiler duty [kW]
EDC Extractive distillation column RR Reflux ratio
EDC-SS Extractive distillation column with a side stream SC Side column
EDWC Extractive dividing-wall column SRC Solvent recovery column
FS Solvent flow [kmol/h] SSED Side-stream extractive distillation
hcol Column height [m] S:F Solvent to feed ratio
htray Column height between top and bottom tray [m] TAC Total annual cost [$/a]
ID Diameter [m] TACC Total annual capital costs[$/a]
FLS Side liquid flow[kmol/h] TAEC Total annual energy costs[$/a]
FV Side vapor flow [kmol/h] TCinst Installed cost of trays [$]
MC Main column ΔT Temperature difference [K]
NT Number of stages UC Heat transfer coefficient of condenser [kW/(K·m2)]
NF Feed location UR Heat transfer coefficient of rebolier [kW/(K·m2)]

the optimal three-column EDWC for separating ethyl acetate and iso- distillation configurations (regular extractive distillation, vacuum ex-
propyl alcohol. Tututi-Avila et al. [28] investigated the same issues for tractive distillation and EDWC) for separating a 2-methoxyethanol/to-
separating ethanol-water binary azeotrope system with ethylene glycol luene mixture. Wang et al. [32] researched the controllability of EDWC
as solvent. Luyben [29] provided a dynamic performance investigation for separating methanol-acetonitrile-benzene ternary mixtures with
of regular and thermally coupled three components extractive distilla- chlorobenzene as solvent. Feng et al. [33] proposed three different
tion process for separating benzene-cyclohexane-toluene mixture with temperature control schemes for the proportional-integral control of
N-methylpyrrolidone as solvent. Dai et al. [30] studied the similar is- EDWC for separating toluene-2-methoxyethanol mixture with dimethyl
sues of separating benzene and cyclohexane mixture by extractive sulfoxide as solvent, and evaluated their dynamic performances. Sun
distillation with o-xylene and sulfolane mixture as entrainer. Li et al. et al. [34] studied the design and control of the separation of benzene
[31] compared the economics and controllability of three extractive and cyclohexane via the EDWC process. To the best of our knowledge,

Fig. 1. The consistency between the predicted and experimental phase equilibrium data (a) methanol-PO, (b) water-PO and (c) water-methanol.

2
Z. Zhang, et al. Chemical Engineering & Processing: Process Intensification 144 (2019) 107651

Table 1 economic and energy performances. In order to improve the opera-


The detailed formulation of TAC. tional performance of SSED and avoid on-line composition measure-
Parameters Formula or data Units ments, four control structures are proposed and analyzed. For com-
parison, one common-used control structure of EDWC is also developed.
Condenser: All control structures are evaluated via introducing feed disturbances in
Heat transfer 0.852 kW/(K·m2)
the flowrate and composition.
coefficient (UC)
Heat transfer area(AC) QC/(UC×△Tcon) m2
Heat transfer 13.9 K 2. Steady-state design
temperature
difference (△TCon)
Capital cost(C1) (M&S/280) ×1609.13×AC0.65 $ 2.1. Design basis
Reboiler:
Heat transfer 0.568 kW/(K·m2) 2.1.1. Thermodynamic model
coefficient (UR) The NRTL model is chosen to describe the phase behaviors of this
Heat transfer area(AR) QR/(UR×△T) m2
Heat transfer 34.8 K
azeotropic system and all binary interactive parameters from Aspen
temperature Plus V8.4 are listed in Table S1 in Supplementary information. Fig. 1
difference (△TReb) shows the extreme consistency of the model prediction data with the
Capital cost(C2) (M&S/280)×1775.26×AR0.65 $ experimental data from the references [36–38].
Column:
hcol 1.2 × 0.6096×(NT-2) m
htray 0.6096×(NT-2) m 2.1.2. Economic evaluation
TCinst (M&S/280)×97.243×ID1.55×htray $ TAC is minimized to obtain one optimal separation configuration. It
CSCinst (M&S/280)×3919.32×ID1.066×hcol0.802 $
Capital cost(C3) TCinst+CSCinst $
is made up of total annual energy costs (TAEC) and total annual capital
Payback period 3 a costs (TACC). The major capital costs include the costs of condensers,
Energy price 4.7 × 10−6 $/kJ reboilers and columns. The column costs cover the installed shell cost
TACC (C1+C2+C3)/Payback period $/a (CSCinst) and the installed tray cost (TCinst). The minor items are neg-
TAEC QR×3600 × 365 × 24×Energy price $/a
ligible due to a lower expenditure, for instance, reflux drum, pump,
TAC TACC + TAEC $/a
valve, and pipe. The TAEC only accounts for the steam consumption
and the costs of solvent makeup, electricity and cooling water are un-
there are no more efforts to the control performance for the high effi- covered. Table 1 summarizes the formulation of the TAC calculation
cient SSED except for Ma et al. [35]. In that reference, the dynamic [39–41].
control of SSED for the separation of methanol-acetone binary mixtures All capital cost data are based on Marshall and Swift equipment cost
was conducted effectively through evaluating the dynamic perfor- index for 2016 (M&S = 1582.3). The CSCinst is calculated as a function
mances of various control structures. of the column height (hcol) and the column diameter (Dcol). The TCinst is
This article aims to explore the application of two energy-saving calculated as a function of the height of the column occupied by trays
extractive distillation configurations (EDWC and SSED) with water as (htray) and Dcol. The cost of the heat exchanger is related to the heat
solvent for separating and purifying the PO product from the PO-me- exchanger area (A) that is estimated from the heat duty (Q), and the
thanol-water mixture. The intensified designs and the conventional temperature difference (ΔT) and the appropriate overall heat transfer
extractive distillation process (CED) are optimized through the search coefficients (U). Simultaneously, the individual cost correction factors
of the minimum total annual costs (TAC) and assessed in terms of associated with the pressure effect, the construction material and
equipment type are fully considered. The basis of equipment material is

Fig. 2. The CED optimal design with detailed information.

3
Z. Zhang, et al. Chemical Engineering & Processing: Process Intensification 144 (2019) 107651

Fig. 3. The optimal EDWC configuration (a) constructure schematic, (b) thermodynamic equivalent flowsheet with detailed information.

carbon steel. The payback period of 3 years is commonly used in most the EDC and SRC, respectively. The bottom stream from the SRC that is
references for the TACC calculation, which can provide a safer cost final the high purity water is divided into two parts of the solvent and the
estimate. The assumed operating time is 365 × 24 h. product.
The global sequential iterative optimization algorithm is used to
obtain the optimum design parameters and the minimum TAC, as
2.1.3. Design statement
shown in Fig. S1 of Supporting Information. In the CED process, four
A fresh feed is assumed as 1000 kmol/h with the composition
key design variables including two reflux ratios (RR1 and RR2) and two
50 mol% methanol, 40 mol% water and 10 mol% PO. The specifications
distillate flowrates (D1 and D2) are determined via the “Design Spec/
of both PO and methanol products are 99.99 wt%, and that of the water
Vary” function to meet the separation requirements below.
product is 99.995 wt%. In addition, the operating pressure of the dis-
tillation column extracting PO as the top product is set at 2 atm for the
(1) For the EDC, the mass composition and the recovery of PO in the
purpose of using the cooling water as cooling service. The pressure of
distillate are both set at 99.99% via adjusting RR1 and D1, respec-
the other column is set at 1 atm, and the tray pressure drop is given as
tively.
0.0068 atm.
(2) For the SRC, the methanol purity in the distillate is maintained
99.99 wt% via adjusting RR2, and the water purity in the bottom
2.2. Optimal design for CED outflow reaches 99.995 wt% via adjusting D2.

The CED process consists of the extractive distillation column (EDC) Six optimization variables including the number of stages (NT1 and
and the solvent recovery column (SRC), as shown in Fig. 2. The fresh NT2), three fresh feed stages (NF1, NF2 and NFS) and the solvent flow
feed and the solvent enter the EDC from the lower tray and upper tray, rate (FS) are determined by minimizing the TAC. The recycling solvent
respectively. The PO and methanol products are obtained at the top of

4
Z. Zhang, et al. Chemical Engineering & Processing: Process Intensification 144 (2019) 107651

Fig. 4. The optimal SSED flowsheet with detailed information.

Fig. 5. The temperature profiles of the EDWC configuration (a) MC, (b) SC.

temperature is considered as fixed variable (298.15 K) instead of opti- methanol are obtained as distillate products of the MC and SC, re-
mization variable in this work, which is generally done for simplicity by spectively. The high purity water is produced as bottom product of the
most scholars. Some important optimum parameters of the CED design MC, which is partially recycled back to the MC as solvent.
are given in Fig. 2. The minimum TAC is 3.485 × 106$/a and includes A conventional sequential iterative optimization algorithm is used
TAEC (2.308 × 106$/a) and TACC (1.177 × 106$/a). to search optimal design parameters and the minimum TAC, as pro-
vided in Fig. S.2. of Supplementary Information. For the EDWC process,
five variables are selected as key design parameters, including reflux
2.3. Optimal design for EDWC ratios (RR1 and RR2), the distillate flowrates (D1 and D2) and the side
vapor flowrate (FV). They are determined via the “Design Spec/Vary”
The used-widely EDWC configuration is capable of accomplishing function to reach several separation specifications below:
the above separation for the ternary mixture only containing one
minimum azeotrope. Fig. 3 gives the constructive schematic and the (1) For MC, the mass composition and recovery of PO in the distillate
thermodynamically equivalent flowsheet consisting of a main column reach to 99.99% and 99.99% by adjusting RR1 and FV, respectively,
(MC) and a side column (SC). In the calculation of TACC, the divided- while the mass composition of water in the bottom reach to
wall column equivalent diameter is estimated based on the equivalent 99.995% via adjusting D1.
square method [30]. The fresh feed and solvent enter the MC from the (2) The mass purity of methanol in the stream D2 reaches to 99.99%
different trays. A side vapor stream containing full methanol and part of via manipulating RR2.
water is withdrawn and fed into the SC. The high purity PO and

5
Z. Zhang, et al. Chemical Engineering & Processing: Process Intensification 144 (2019) 107651

Fig. 6. The EDWC configuration (a) control structure, (b) feed flowrate dynamic responses, and (c) feed composition dynamic responses.

Fig. 7. The temperature profiles of the SSED configuration (a) EDC-SS, (b) SSRC.

The six variables including the number of trays (NT1 and NT2), the 2.4. Optimal design for SSED
raw materials and solvent feed stages (NF1, and NFS), the solvent
flowrate (FS) and side vapor stream withdrawn location (NV) are opti- The lately-developed SSED configuration is also adapted to the se-
mized. Some key design parameters of the EDWC optimum configura- paration of this system. Fig. 4 shows the SSED separation flowsheet that
tion are given in Fig. 3. TAC is minimized as 3.457 × 106$/a, including includes an extractive distillation column with side stream (EDC-SS)
TAEC (2.361 × 106$/a) and TACC (1.096 × 106$/a). and a column for recovering solvent (SRC). PO and methanol are

6
Z. Zhang, et al. Chemical Engineering & Processing: Process Intensification 144 (2019) 107651

Fig. 8. The SSED configuration (a) control structure-CS1, (b) feed flowrate dynamic responses, and (c) feed composition dynamic responses.

obtained as the overhead products of EDC-SS and SRC respectively. The are given in Fig. 4, with the minimum TAC of 3.445 × 106$/a, and
differences from the EDWC configuration are the removal of some made up to the TAEC (2.277 × 106$/a) and (TACC of 1.168 × 106$/a).
water as the bottom product in the EDC-SS and the extraction of a side In comparison with the EDWC, the SSED has advantages to some extent
liquid stream containing full methanol and the remaining water. in terms of economic costs.
A sequential iterative optimization algorithm is used to obtain the
optimum design parameters and the minimum TAC, as given in Fig. S.3 3. Dynamic control
of Supplementary Information. For the SSED configuration, five key
design variables of reflux ratios (RR1 and RR2), the distillate flowrates The control and operational issues of the implementation following
(D1 and D2) and a side liquid flowrate (FLS) are chosen as key design the steady-state optimum design must be discussed. In this section, the
parameters. They are determined via the “Design Spec/Vary” function control structures of the EDWC for one and the SSED for four are
to reach the several separation specifications below: constructed. Their dynamic properties are analyzed via introducing the
disturbances in feed flowrates and feed compositions. The magnitude of
(1) For EDC-SS, the mass composition and recovery of PO in the dis- the disturbances is the same one in the literature [42]. The steady-state
tillate are set as 99.99% and 99.99% via adjusting RR1 and FLS simulation is exported into Aspen Dynamics as a pressure-driven si-
respectively, while the mass composition of water in the bottom mulation after the needful equipment sizes are calculated. At the same
reaches 99.995% via adjusting D2. time, the adequate pressure drop is provided by selecting appropriate
(2) For SRC, the methanol purity in the distillate is maintained pumps and valves. Additionally, the relay-feedback testing and Tyreus-
99.99 wt% via adjusting RR2, and the water purity in the bottom Luyben tuning rules are used to get the ultimate gain KC and period τI
stream reaches 99.995 wt% via adjusting D2. for all controllers in this paper.

The selected seven variables are optimized including the number of 3.1. Control of EDWC configuration
stages (NT1, NT2), fresh feed stages (NF1, NT2, NFS), a side liquid lo-
cation (NLS) and a solvent flowrate (FS) by searching for the minimum The slope criterion is adopted for the selection of the sensitivity tray
TAC. Some key design parameters of the optimum SSED configuration [43]. The temperature profiles of EDWC are shown in Fig. 5. For the

7
Z. Zhang, et al. Chemical Engineering & Processing: Process Intensification 144 (2019) 107651

Fig. 9. The SSED configuration (a) control structure-CS2, (b) feed flowrate dynamic responses, and (c) feed composition dynamic responses.

MC, the location of the tray with the larger slopes are the 40th and reflux ratio (RR2).
57th. For the SC, the location of the tray with the larger slope is the
46th. Fig. 6(a) shows the dual-temperature control structure. The dynamic behavior under this control structure is tested when
As for the controller tuning, level controllers are proportional-only facing feed flowrate disturbances ( ± 10%) and feed composition dis-
controllers with large integral time (9999 min) and KC = 2. Flow is turbances from 0.1(PO)/0.5(methanol)/0.4(water) to 0.09/0.52 /0.39 (+Δz)
controlled by using the proportional-integral controller with KC = 0.5 and 0.11/0.48/0.41 (-Δz) at time = 2 h. Fig. 6(b) and (c) respectively
and τ1 = 0.3 min. Pressure controllers are tightly tuned with KC = 20 shows the dynamic responses.
and τ1 = 12 min. The deadtime in the temperature control loop is set as From these figures, two temperature controllers respond rather fast
1 min [44,45]. The control loops are listed as follows: by getting the temperatures back to their setpoints, and all product
purities approaching the initial set values can be maintained. Overall,
(1) Fresh feed flow is controlled. the implementation of the EDWC goes with the good control.
(2) Reflux-drum levels of MC and SC are maintained by manipulating
their distillate flowrates. 3.2. Control of SSED configuration
(3) Base level of SC is maintained by manipulating the flowrate of the
bottom product. 3.2.1. Control structure (CS1)
(4) Notably, base level of MC is controlled by adjusting the water The control of the SSED configuration is less studied than that of the
product flowrate due to the absence of a solvent makeup. EDWC despite more advantages in energy-savings. The temperature
(5) The overhead pressures of MC and SC are controlled by manip- profiles of the EDC-SS and SRC are shown in Fig. 7. It can be seen that
ulating the condenser heat duty. the locations with the larger slope are stages 42 and 56 for the EDC-SS
(6) The ratio of solvent flowrate to fresh feed flowrate is fixed. and the location with the largest slope, is stage 40 for the SRC. Note that
(7) The temperature on stage 40 is kept constant by adjusting the ratio the side-stream flowrate is an important control variable in the control
of side vapor flowrates and reboiler heat duty (VR/QR), while the structure of the SSED that can achieve different control goals. Fig. 8(a)
temperature on stage 57 is controlled by manipulating QR. shows the first control structure (CS1). The following is the control
(8) The temperature on stage 46 in the SC is controlled by operating the loops that need to be highlighted.

8
Z. Zhang, et al. Chemical Engineering & Processing: Process Intensification 144 (2019) 107651

Fig. 10. The SSED configuration (a) control structure-CS3, (b) feed flowrate dynamic responses, and (c) feed composition dynamic responses.

(1) For EDC-SS, temperatures on stage 42 and 56 are controlled by “cascade”. In other words, the methanol mole fraction is connected as
manipulating reflux ratio (RR1) and reboiler duty (QR1), respec- an input signal with the composition controller, where the resulting
tively. output signal changes the ration of FLS/F.
(2) The stage 40 temperature in the SRC is controlled by manipulating Fig. 9(b) and (c) respectively show the dynamic responses to feed
reboiler duty (QR2). flow and composition disturbances after 2 h. After stabilization, all
(3) The side-stream flowrate (FLS) is used to control the purity of me- sensitive tray temperatures can quickly return to their setpoints, and all
thanol product. The deadtime in the composition control loop is set product compositions have a small deviation from the initial set values.
as 3 min. Although the CS2 can effectively address the feed flow disturbances
( ± 10%), the major drawbacks are to require too long time and os-
The dynamic performances are analyzed by disturbing feed flowrate cillate violently before restoring to another stable state.
and composition at the time = 2 h. However, the Aspen Dynamics si-
mulation fails when a large feed flowrate disturbance (-10%) occurs, so
only a +10% feed flowrate disturbance is investigated besides the ± 3.2.3. Control structure (CS3)
5% feed flowrate disturbances. Fig. 8(b) and (c) gives their dynamic The third control structure (CS3) is developed to improve the con-
responses for these feed disturbances. It can be seen that both product trol quality further, as shown in Fig. 10(a). The difference from the CS1
purities and tray temperatures are returned to their specifications but a and CS2 is that the PO composition in the side stream is measured and
long restore time is needed. controlled by manipulating the side-stream flowrate controller on
“cascade”.
Fig. 10(b) and (c) show the dynamic performances of the CS3 when
3.2.2. Control structure (CS2) the feed disturbances occur. It can be noticed that the CS3 handles these
In order to improve the poor performance of CS1, the second control disturbances quickly and effectively. The restoring time to the initial
structure (CS2) is proposed by combining the ratio of side-stream values is significantly shortened and the oscillate phenomenon is
flowrate to feed flowrate (FLS/F) with a composition controller, as greatly weakened as well. However, one of the deficiencies is that the
shown in Fig. 9(a). The FLS is indirectly adjusted to maintain the me- PO composition in the LS stream is so small as to make the detection
thanol product purity and the side-stream flowrate controller is on and control difficult. The other is that the PO composition is inconstant

9
Z. Zhang, et al. Chemical Engineering & Processing: Process Intensification 144 (2019) 107651

Fig. 11. The SSED configuration (a) control structure-CS4, (b) feed flowrate dynamic response, and (c) feed composition dynamic responses.

when facing the greater feed composition disturbances, which has a magnitude of the change.
significant impact on the methanol product purity. Moreover, most
industrial applications try to avoid the use of inferential composition
controller due to the delay and high-cost of on-line composition mea- 4. Conclusions
surements.
In this article, two energy-efficient extractive distillation alter-
natives, namely EDWC and SSED are adopted for the first time to se-
3.2.4. Control structure (CS4) parate the PO-methanol-water mixture with water as solvent. The
By analysis, the fourth control structure (CS4) relying on only steady optimization results reveal that both EDWC and SSED have
temperature measurements is established, as shown in Fig. 11(a). In this economic advantages to great extent compared to the CED, which
structure, the side stream flowrate control (FC) is cascaded with the causes a reduction of 0.28 × 105$/a and 0.4 × 105$/a TAC, respec-
temperature control on stage 56. In addition, a proportional control tively.
between the reboiler duty of EDC-SS (QR1) and the fresh feed flowrate In the control analysis, the EDWC configuration is readily controlled
is implemented to meet the limitation of degrees of freedom. via a common-used control structure and is performed with good dy-
Fig. 11(b) and (c) give the dynamic responses for the identical feed namic behaviors. For the SSED configuration, four control structures
disturbances. The results show that both product purities and the sen- are successively constructed to improve its controllability according to
sitivity trays temperature are capable of returning to their set points the use of the side-stream flowrate as control variable. CS1 is simply
quickly. Therefore, CS4 is considered the best option for the control of unable to handle a large negative feed flowrate disturbance (-10%).
the SSED configuration. Both CS2 and CS3 can solve this puzzle, but a lengthy running period is
To be clear, the proposed control structures are untested for the required. CS2 simultaneously occurs a violently oscillation phenom-
greater feed flowrate and composition disturbances ( ± 20%), which enon. Different from the other three structures, CS4 only using tem-
maybe affect to some extent the system’s operational flexibility. In fact, perature controllers can smoothly maintain the product purities with
the magnitudes of the feed disturbances in the distillation system are litter deviations and a short recovery time. Hence, the SSED config-
associated with possible deviations at the upstream chemical stage. As uration is an attractive energy-saving alternative for separating the PO/
for the propylene oxide production, it is justified to allow the small methanol/water mixture in addition to the EDWC, which can be well

10
Z. Zhang, et al. Chemical Engineering & Processing: Process Intensification 144 (2019) 107651

controllable under this newly proposed control strategy. [21] E. Anokhina, A. Timoshenko, Criterion of the energy effectiveness of extractive
distillation in the partially thermally coupled columns, Chem. Eng. Res. Des. 99
(2015) 165–175.
Appendix A. Supplementary data [22] X. Wang, L. Xie, P. Tian, G. Tian, Design and control of extractive dividing wall
column and pressure-swing distillation for separating azeotropic mixture of acet-
Supplementary material related to this article can be found, in the onitrile/N-propanol, Chem. Eng. Process. 110 (2016) 172–187.
[23] Y. Zhao, K. Ma, W. Bai, D. Du, Z. Zhu, Y. Wang, J. Gao, Energy-saving thermally
online version, at doi:https://doi.org/10.1016/j.cep.2019.107651. coupled ternary extractive distillation process by combining with mixed entrainer
for separating ternary mixture containing bioethanol, Energy 148 (2018) 296–308.
References [24] S. Tututi-Avila, N. Medina-Herrera, J. Hahn, A. Jiménez-Gutiérrez, Design of an
energy-efficient side-stream extractive distillation system, Comput. Chem. Eng. 102
(2017) 17–25.
[1] S. Liang, Cao Y, Liu X, X. Li, Y. Zhao, Y. Wang, Y. Wang, Insight into pressure-swing [25] C. Wang, C. Guang, Y. Cui, C. Wang, Z. Zhang, Compared novel thermally coupled
distillation from azeotropic phenomenon to dynamic control, Chem. Eng. Res. Des. extractive distillation sequences for separating multi-azeotropic mixture of acet-
117 (2017) 318–335. onitrile/benzene/methanol, Chem. Eng. Res. Des. 136 (2018) 513–528.
[2] C. Guang, X. Shi, Z. Zhang, C. Wang, C. Wang, J. Gao, Comparison of heterogeneous [26] Y. Wang, K. Ma, M. Yu, Y. Dai, R. Yuan, Z. Zhu, J. Gao, An improvement scheme for
azeotropic and pressure-swing distillations for separating the diisopropylether/ pressure-swing distillation with and without heat integration through an inter-
isopropanol/water mixtures, Chem. Eng. Res. Des. 143 (2019) 249–260. mediate connection to achieve energy savings, Comput. Chem. Eng. 119 (2018)
[3] A. Yang, W. Shen, S. Wei, L. Dong, J. Li, V. Gerbaud, Design and control of pressure- 439–449.
swing distillation for separating ternary systems with three binary minimum [27] H. Zhang, Q. Ye, J. Qin, H. Xu, N. Li, Design and control of extractive dividing-wall
azeotropes, AIChE J. 65 (2019) 1281–1293. column for separating ethyl acetate−isopropyl alcohol mixture, Ind. Eng. Chem.
[4] M. Aurangzeb, A.K. Jana, Pressure-swing dividing-wall column with multiple Res. 53 (2014) 1189–1205.
binary azeotropes: improving energy efficiency and cost savings through vapor [28] S. Tututi-Avila, A. Jiménez-Gutiérrez, J. Hahn, Control analysis of an extractive
recompression, Ind. Eng. Chem. Res. 57 (2018) 4019–4032. dividing-wall column used for ethanol dehydration, Chem. Eng. Process. 82 (2014)
[5] C. Wang, Z. Zhang, X. Zhang, C. Guang, J. Gao, Comparison of pressure-swing 88–100.
distillation with or without crossing curved boundary for separating a multi- [29] W.L. Luyben, Control comparison of conventional and thermally coupled ternary
azeotropic ternary mixture, Sep. Purif. Technol. 220 (2019) 114–125. extractive distillation processes, Chem. Eng. Res. Des. 106 (2016) 253–262.
[6] J. Gu, X. You, C. Tao, J. Li, W. Shen, J. Li, Improved design and optimization for [30] X. Dai, Q. Ye, J. Qin, H. Yu, X. Suo, R. Li, Energy-saving dividing-wall column
separating tetrahydrofuran-water azeotrope through extractive distillation with and design and control for benzene extractive distillation via mixed entrainer, Chem.
without heat integration by varying pressure, Chem. Eng. Res. Des. 133 (2018) Eng. Process. 100 (2016) 49–64.
303–313. [31] L. Li, L. Guo, Y. Tu, N. Yu, L. Sun, Y. Tian, Q. Li, Comparison of different extractive
[7] P.D. Ghuge, N.A. Mali, S.S. Joshi, Comparative analysis of extractive and pressure distillation processes for 2-methoxyethanol/toluene separation: design and control,
swing distillation for separation of THF-water separation, Comput. Chem. Eng. 103 Comput. Chem. Eng. 99 (2017) 117–134.
(2017) 188–200. [32] C. Wang, C. Wang, Y. Cui, C. Guang, Z. Zhang, Economics and controllability of
[8] Y. Wang, G. Bu, X. Geng, Z. Zhu, P. Cui, Z. Liao, Design optimization and operating conventional and intensified extractive distillation configurations for acetonitrile/
pressure effects in the separation of acetonitrile/methanol/water mixture by methanol/benzene mixtures, Ind. Eng. Chem. Res. 57 (2018) 10551–10563.
ternary extractive distillation, J. Clean. Prod. 218 (2019) 212–224. [33] Z. Feng, W. Shen, G.P. Rangaiah, L. Dong, Proportional-integral control and model
[9] Y. Wang, X. Zhang, X. Liu, W. Bai, Z. Zhu, Y. Wang, J. Gao, Control of extractive predictive control of extractive dividing-wall column based on temperature differ-
distillation process for separating heterogenerous ternary azeotropic mixture via ences, Ind. Eng. Chem. Res. 57 (2018) 10572–10590.
adjusting the solvent content, Sep. Purif. Technol. 191 (2018) 8–26. [34] L. Sun, Q. Wang, L. Li, J. Zhai, Y. Liu, Design and control of extractive dividing wall
[10] X. Zhang, Y. Zhao, H. Wang, B. Qin, Z. Zhu, N. Zhang, Y. Wang, Control of a ternary column for separating benzene/cyclohexane mixtures, Ind. Eng. Chem. Res. 53
extractive distillation process with recycle splitting using a mixed entrainer, Ind. (2014) 8120–8131.
Eng. Chem. Res. 57 (2018) 339–351. [35] K. Ma, M. Yu, Y. Dai, Y. Ma, J. Gao, P. Cui, Y. Wang, Control of an energy-saving
[11] Q. Pan, X. Shang, J. Li, S. Ma, L. Li, L. Sun, Energy-efficient separation process and side-stream extractive distillation process with different disturbance conditions,
control scheme for extractive distillation of ethanol-water using deep eutectic sol- Sep. Purif. Technol. 210 (2019) 195–208.
vent, Sep. Purif. Technol. 219 (2019) 113–126. [36] V.N. Vostrikova, T.V. Komarova, V.M. Platonov, L.V. Morozova, T.P. Moiseeva,
[12] W.L. Luyben, Control of a multiunit heterogeneous azeotropic distillation process, M.N. Saenko, M.E. Aerov, Liquid-vapor equilibrium in binary systems containing
AIChE J. 52 (2006) 623–637. propene oxide, Zh. Prikl. Khim 47 (1974) 568.
[13] T. Shi, A. Yang, S. Jin, W. Shen, S. Wei, J. Ren, Comparative optimal design and [37] J.N. Wickert, W.S. Tamplin, R.L. Shank, Phase equilibria in the system propylene
control of two alternative approaches for separating heterogeneous mixtures iso- oxide – water, Chem. Eng. Prog. Symp. Ser 48 (1952) 92–96.
propyl alcohol-isopropyl acetate-water with four azeotropes, Sep. Purif. Technol. [38] C. Yang, S. Ma, X. Yin, Organic salt effect of tetramethylammonium bicarbonate on
225 (2019) 1–17. the vapor liquid equilibrium of the methanol water System, J. Chem. Eng. Data 56
[14] Y. Cui, X. Shi, C. Guang, Z. Zhang, C. Wang, C. Wang, Comparison of pressure-swing (2011) 3747–3751.
distillation and heterogeneous azeotropic distillation for recovering benzene and [39] Ž. Olujić, L. Sun, A.D. Rijke, P.J. Jansens, Conceptual design of an internally heat
isopropanol from wastewater, Process. Saf. Environ. 122 (2019) 1–12. integrated propylene-propane splitter, Energy. 31 (2006) 3083–3096.
[15] A. Yang, S. Jin, W. Shen, P. Cui, I.-L. Chen, J. Ren, Investigation of energy-saving [40] W.L. Luyben, Distillation Design and Control Using AspenTM Simulation, second
azeotropic dividing wall column to achieve cleaner production via heat exchanger edition, (2013) Hoboken, New Jersey.
network and heat pump technique, J. Clean. Prod. 234 (2019) 410–422. [41] C. Wang, C. Wang, C. Guang, Z. Zhang, Comparison of extractive distillation se-
[16] Y. Dai, F. Zheng, B. Xia, P. Cui, Y. Wang, J. Gao, Application of mixed solvent to paration sequences for acetonitrile/methanol/benzene multi-azeotropic mixtures,
achieve an energy-saving hybrid process including liquid−liquid extraction and J. Chem. Technol. Biotechnol. 93 (2018) 3302–3316.
heterogeneous azeotropic distillation, Ind. Eng. Chem. Res. 58 (2019) 2379–2388. [42] W.L. Luyben, Control comparison of conventional extractive distillation with a new
[17] A.A. Kiss, D.J.-P.C. Suszwalak, Enhanced bioethanol dehydration by extractive and split-feed configuration, Chem. Eng. Process. 107 (2016) 29–41.
azeotropic distillation in dividing-wall columns, Sep. Purif. Technol. 86 (2012) [43] W.L. Luyben, Evaluation of criteria for selecting temperature control trays in dis-
70–78. tillation columns, J. Process.Contr. 16 (2006) 115–134.
[18] A.A. Kiss, R.M. Ignat, Innovative single step bioethanol dehydration in an extractive [44] K. Ma, X. Pan, T. Zhao, J. Yang, Z. Zhu, D. Xu, Y. Wang, Dynamic control of hybrid
dividing-wall column, Sep. Purif. Technol. 98 (2012) 290–297. processes with liquid−liquid extraction for propylene glycol methyl ether dehy-
[19] G. Modla, Energy saving methods for the separation of a minimum boiling point dration, Ind. Eng. Chem. Res. 57 (2018) 13811–13820.
azeotrope using an intermediate entrainer, Energy 50 (2013) 103–109. [45] Z. Feng, W. Shen, G.P. Rangaiah, L. Lv, L. Dong, Process development, assessment,
[20] A.V. Timoshenko, E.A. Anokhina, A.V. Morgunov, D.G. Rudakov, Application of the and control of reactive dividing-wall column with vapor recompression for pro-
partially thermally coupled distillation flowsheets for the extractive distillation of ducing n‑Propyl Acetate, Ind. Eng. Chem. Res. 58 (2019) 276–295.
ternary azeotropic mixtures, Chem. Eng. Res. Des. 104 (2015) 139–155.

11

You might also like