Integracion de Procesos EtilbencenoStireno

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

Separation and Purification Technology 228 (2019) 115760

Contents lists available at ScienceDirect

Separation and Purification Technology


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

Process synthesis and simulation-based optimization of ethylbenzene/ T


styrene separation using double-effect heat integration and self-heat
recuperation technology: A techno-economic analysis

Xingang Li, Chengtian Cui, Hong Li, Xin Gao
School of Chemical Engineering and Technology, National Engineering Research Center of Distillation Technology, Collaborative Innovation Center of Chemical Science and
Engineering (Tianjin), Tianjin University, 300072, PR China

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

Keywords: Ethylbenzene/styrene separation is a typical example of the most costly processes in the chemical industry due to
Ethylbenzene/styrene separation their close-boiling points of these two species. To reduce the cost of their separation, the concepts of double-
Process synthesis effect distillation (DED) and self-heat recuperation technology (SHRT) have been implemented to produce three
Simulation-based optimization energy-efficient distillation configurations in either an individual or synergistic manner. To obtain a fair com-
Double-effect distillation
parison, all the candidate processes were optimized to a minimum in total annual cost (TAC), using a simulation-
Self-heat recuperation technology
Techno-economic analysis
based optimization framework on a simulator Aspen Plus and an optimizer programmed in MATLAB with a
metaheuristic algorithm. For a small treatment capacity (100 kmol/h) and a short payback period (PBP, 3 years),
the DED and SHRT configurations were found to reduce the TAC by ~8% compared to the conventional process
design. The TAC reduction can be improved to as much as ~28% with a larger capacity (1000 kmol/h). The
synergistic DED-SHRT configuration has the lowest energy consumption, but its high capital investment makes it
only economically viable for longer PBPs. The best scheme under either short or long PBP appeared to be SHRT.
Since the conventional design and SHRT share close technical parameters, the former will benefit more when
retrofitted into the latter when minimum process modifications are applied.

1. Introduction as raw materials for downstream treatment [4–8]. The separation of the
desired SM product from lights and styrene tar is relatively easy, using
Styrene monomer (SM) is commonly recognized as one of the es- conventional distillation columns (CDiCs). However, separating SM
sential monomers for synthesizing various polymers and copolymers, from EB is considerably difficult due to their close-boiling points (the
such as polystyrene, acrylonitrile-butadienestyrene copolymers (ABS), normal boiling points of EB and SM are 136.2 and 145.7 °C, respec-
styrene-acrylonitrile copolymers (SAN), styrene-butadiene rubber tively) so that their relative volatilities are very close to unity (1.27 at
(SBR), styrene block copolymers (SBC), styrene-butadiene latex (SBL), 1 atm and 1.33 at 0.3 atm). Depending on the different distillation se-
and adhesives [1]. The global demand for SM has been estimated to be quences and separation targets, the EB/SM column usually operates at a
more than 25 million tons annually and it is growing rapidly [2]. reflux ratio of 5–15 [2,5,8,9]. In a conventional styrene distillation unit,
SM is predominantly produced via direct dehydrogenation of the EB/SM column alone consumes more than 60% of the overall en-
ethylbenzene (EB) on an iron-based catalyst [3]. Typically, the con- ergy cost [2].
version rate of EB to SM is around 50–70% per pass across the dehy- The energy intensity of EB/SM separation has prompted researchers
drogenation reactor [4]. In normal operations, the dehydrogenated to explore alternative separation methods such as extractive distillation
product is a mixture containing substantial portions of EB and SM as [9,10], membrane separation [11,12], and adsorption [13,14]. Despite
well as minor amounts of undesirable byproducts, such as the lights, the attractiveness of these alternative technologies, conventional en-
aromatics benzene and toluene, and other heavies known as styrene tar trainer-free distillation remains the primary method used for industrial-
[5]. The reactor effluents are then treated in a styrene distillation unit, scale EB/SM separation [2]. For close-boiling components, energy-ef-
where the unreacted EB is recovered from the dehydrogenated product ficient distillation technologies such as multi-effect distillation (MED)
and then recycled to the reactor, while other components are separated [15,16], heat pump assisted distillation (HPAD) [17,18], and internally


Corresponding author.
E-mail address: gaoxin@tju.edu.cn (X. Gao).

https://doi.org/10.1016/j.seppur.2019.115760
Received 19 April 2019; Received in revised form 25 June 2019; Accepted 30 June 2019
Available online 02 July 2019
1383-5866/ © 2019 Elsevier B.V. All rights reserved.
X. Li, et al. Separation and Purification Technology 228 (2019) 115760

adiabatically compressed by a heat pump, which elevates its tempera-


ture. Then the heating requirements of the reboiler and feed preheater
are supplied by the discharged high-pressure vapor and bottoms. It is
expected that the heat of the process stream can be circulated without
the need for any additional heat. Unlike HPAD that only uses latent
heat, SHRT utilizes both latent and sensible heat to intensify the column
and thus it is expected to be more efficient. Moreover, instead of ap-
plying SHRT to a single column, it is possible to further strengthen the
DED with SHRT at the cost of more capital investments. Generally
speaking, the synergistic effect of integrating DED with SHRT has not
yet been addressed.
The purpose of this study was to perform a comprehensive economic
evaluation of CDiC, DED, SHRT, and DED-SHRT for the EB/SM se-
paration to obtain a configuration with a minimum separation cost for
Fig. 1. Schematic diagram of the CDiC. SM production. Each of these proposed configurations were in-
dividually optimized using a simulation-based optimization framework
heat-integrated distillation column (HIDiC) [19,20] are expected to and fairly compared under their respective optimum conditions.
save substantially reduce energy consumption as well as total annual
cost (TAC). Cui et al. [21] suggested using double-effect distillation 2. Process synthesis
(DED, the simplest type of MED) and mechanical vapor recompression
(MVR, a kind of HPAD) to intensify close-boiling point distillation 2.1. Process specifications
schemes, since these two methods are most commonly used in industry.
By applying DED and MVR to the conventional EB/SM column, these To explore the influence of feed flow rates on the selection of the
authors observed an energy consumption reduction of 30% and 40%, optimum distillation configuration, two flow rates – 100 kmol/h and
respectively [2]. However, the authors [2] heuristically set the struc- 1000 kmol/h – with 40/60 mol% EB/SM saturated liquid mixture at
tural and operating parameters that were based on those of an in- 1 atm were used. Employing an annual operating time (AOT) of 8000 h,
dustrial plant without explicit optimization. It is unfair to compare the feed flow rates are approximately equivalent to styrene treatment
different configurations using non-optimum conditions. Utilizing stage- capacities of 50,000 ton/a and 500,000 ton/a, respectively. The product
by-stage heat transfer between rectifying and stripping sections, a specifications were 99.9 mol% of purity for each component. To sup-
HIDiC is expected to produce good energy-saving performance. But in press SM polymerization, the EB/SM column was considered to be
some cases, this structure can consume more energy, which would be operated in the high vacuum condition with low-pressure drop tray. As
restrictive [22]. In addition, due to design, construction and operation discussed in previous works [30,31], the optimum selection of column
complexities, HiDiC has minimal applications in industry [18–21]. pressure is crucial for distillation design. Therefore, the column pres-
Until recently, Wakabayashi et al. [23] developed the world’s first sure was considered to be a design variable. The compressor mechan-
commercial application of HIDiC technology by discretely applying the ical/isentropic efficiency was set at 0.8/0.72. The pressure drops in
heat exchangers to appropriately selected stages in the rectifying and heat exchangers and pipelines were neglected. Cooling water was op-
stripping sections. Recognizing that the practical implementation issues erated at 30–40 °C and low-pressure steam operated 6 bar was used as
of HIDiC have not been fully resolved, this approach was not considered cold and hot utilities, respectively. All the process simulations were
in this present study. performed in the Aspen Plus V8.8 environment with the built-in Peng-
The concept of self-heat recuperation technology (SHRT) was pro- Robinson thermodynamic model [2,5].
posed by the Tsutsumi research group more than a decade ago [24]. To
date, the authors have applied this technology to several thermal pro-
2.2. Process configurations
cesses including the reaction section [25], distillation section [26],
drying section [27], gas separation section [28], etc., and found that it
2.2.1. Configuration 1 – CDiC
produced a considerable degree of energy saving. The basic principle of
Fig. 1 presents a CDiC for EB/SM separation. The decision variables
SHRT is to recover the waste heat from the effluent streams to heat the
to be optimized were column pressure (PT), total stage (NT), and feed
feed streams by adiabatically compressing gas and/or vapor [29]. In the
stage (NF). Reflux ratio (RR) and distillate flow rate (D) were varied to
case of applying SHRT to binary distillation, the overhead vapor is
satisfy the required product specifications by using the Design Spec/

Fig. 2. Schematic diagram of the feed split partial heat integration DED configuration.

2
X. Li, et al. Separation and Purification Technology 228 (2019) 115760

Fig. 3. Schematic diagrams of the (a) MVR, (b) MVR-PCS-CIP-FP, (c) MVR-PCS, (d) MVR-CIP, (e) MVR-FP, (f) MVR-PCS-CIP, (g) MVR-PCS-FP, and (h) MVR-CIP-FP.

high-pressure column (HPC) to drive the reboiler of the low-pressure


column (LPC). To produce binary separation, Cui et al. [16] summar-
ized five commonly used DED configurations – feed split, light split
forward, light split reverse, heavy split forward, and heavy split reverse.
Based on a case study of the separation of close-boiling components,
feed split proved to be more economical than the other DED config-
urations [16]. Feed split was selected, because the relative volatility
and feed composition of the previous case were close to the metrics set
in this study. DED can be conceptually classified as full and partial heat
integration, where the former is a particular case of the latter. Fig. 2
shows a feed split partial heat integration DED configuration. When the
duty of the LPC trim reboiler approaches to zero, the partial heat in-
tegration is reduced to a full phase. The selection of a partial or a full
type depends on the following economic optimization. The decision
variables for this configuration include column pressures (PT1 and PT2),
Fig. 4. A process superstructure considering MVR with PCS, CIP, and FP. feed stages (NF1 and NF2), total stages (NT1 and NT2), and feed split
flow rate (F1 or F2). Also, the reflux ratio (RR1 and RR2) and distillate
Vary option in Aspen Plus. The CDiC was used as the baseline case for flow rate (D1 and D2) are varied to satisfy the desired product specifi-
comparison with other configurations. cations.

2.2.2. Configuration 2 – DED 2.2.3. Configuration 3 – SHRT


The basic principle of DED is the use of the overhead vapor from the The implementation of a SHRT in EB/SM column is similar to the

3
X. Li, et al. Separation and Purification Technology 228 (2019) 115760

Fig. 5. Schematic diagram of the DED-SHRT.

Fig. 6. Aspen Plus/MATLAB optimization framework with a SADDE algorithm.

traditional MVR configuration (Fig. 3a), in which the overhead vapor is bottoms’ temperature was too low to preheat feed stream and thus was
compressed to a higher temperature to vaporize the bottom material, not considered. The decision variables for the superstructure are
and the high-pressure condensate is then expanded to the top pressure column pressure (PT), feed stage (NF), total stage (NT), PCS flow rate
by a valve before it enters the column. During the compression/ex- (F1 or F2), CIP outlet temperature (T2), steam preheater outlet tem-
pansion process, partial saturated vapor/liquid can be condensed/fla- perature (T3), compressor outlet pressure (PC), and feed vapor fraction
shed, so an optional heater/cooler is required to preheat/condense the (q).
stream. If the latent heat of the pressurized vapor is not fully utilized in
the reboiler and exits the heat exchanger as a two-phase mixture, a flow
splitter should to be added to pre-split part of the vapor before it enters 2.2.4. Configuration 4 – DED-SHRT
the compressor. This measure is called pre-compressor splitting (PCS, Fig. 5 shows the synergistic configuration that integrates DED with
Fig. 3c) [32]. To further intensify the MVR, with the exception of the SHRT. The LPC is driven by the HPC top vapor, while the LPC top vapor
latent heat, the remaining sensible heat can be further used for com- is reversely compressed to drive HPC, and its sensible heat that exits the
pressor inlet preheating (CIP, Fig. 3d) and feed preheating (FP, Fig. 3e) HPC reboiler is used sequentially for CIP and FP. Note that all the heat
[33]. By optionally combining PCS, CIP, and FP, eight different con- of the process stream is circulated without the need for any additional
figurations can be generated, as demonstrated in Fig. 3 (Note that heat. In other words, this configuration is entirely driven by electrical
Fig. 3b is a combination of PCS, CIP, and FP). To optimize all these power. The decision variables considered include column pressures (PT1
configurations once and for all, a process superstructure comprised of and PT2), feed stages (NF1 and NF2), total stages (NT1 and NT2), feed
the MVR with PCS, CIP, and FP was synthesized as shown in Fig. 4. In split flow rate (F1 or F2), PCS flow rate (F3 or F4), CIP outlet tem-
the original SHRT column proposed by Matsuda et al. [26], all the feed perature (T1), compressor outlet pressure (PC), HPC feed vapor fraction
and products were set at standard conditions (25 °C and 0.10 MPag), (q1), and LPC feed vapor fraction (q2).
allowing the feed preheating by the bottoms. However, in this case, the

4
X. Li, et al. Separation and Purification Technology 228 (2019) 115760

Fig. 7. (a) The optimum CDiC configuration, (b) T-H composite curve, and (c) grand composite curve.

3. Simulation-based optimization (CTRAY ), heat exchanger (CHEX ), and the compressor (CCOM ), while the
costs of the pump, pipeline, and valve are neglected. Hence:
3.1. Objective function
CAPEX = ∑ (CCOL + CTRAY + CHEX + CCOM ) (3)
The objective function to be minimized is the TAC (US$/a), that The CAPEX of the distillation column shell (CCOL , US$) and tray
takes into consideration the operational and capital expenditures (CTRAY , US$) can be estimated from [34]:
(OPEX and CAPEX, US$/a). An AOT of 8000 h with a payback period
(PBP) of 3 years for capital investment are the initially considered CCOL = 17, 640·Dc1.066 ·Hc0.802 (4)
parameters. Therefore:
CTRAY = 229·DC1.55 ·(NT − 2) (5)
CAPEX
TAC = OPEX + The column diameterDc (m) is calculated using the Aspen Tray
PBP (1)
Sizing option. The column heightHc (m) is estimated as follows:
OPEX includes hot utility (Chu ) for the reboiler, cold utility (Ccu ) for
Hc = 1.2·0.61·(NT − 2) (6)
the condenser, and electrical power (Celec ) for the compressor:
Without considering equipment details, the CAPEX of the heat ex-
OPEX = AOT ·[∑ (Chu·Qhu ) + ∑ (Ccu·Qcu) + ∑ (Celec·W )] (2) changer (CHEX , US$) is estimated by its heat transfer area (A, m2) [34]:

where Qhu , Qcu , and W are the energy consumption of heating, cooling, CHEX = 7, 296·A0.65 (7)
and compression work, respectively, the available utilities include low-
The heat transfer area is calculated based on:
pressure steam (6 bar, 160 °C), cooling water (30–40 °C), and electrical
power. The prices are 7.78, 0.354, and 16.8 US$/GJ, respectively [34]. Q
A=
The CAPEX mainly considers the distillation column (CCOL ), tray U ·LMTD (8)

5
X. Li, et al. Separation and Purification Technology 228 (2019) 115760

Fig. 8. (a) The optimum DED configuration, (b) T-H composite curve, and (c) grand composite curve.

where Q is the heat duty, U the overall heat-transfer coefficient, and discrete (total stage and feed stage) and continuous (column pressure,
LMTD the logarithmic mean temperature difference between the hot stream flow rate, etc.) variables that were formulated as a mixed integer
and cold streams. The adopted overall heat-transfer coefficient for the nonlinear programming (MINLP). Solving the MINLP by deterministic
condenser is 0.852 kW/(°C·m2), while for reboiler and process-to-pro- optimization algorithms is difficult, because the rigorous MESH equa-
cess heat exchanger 0.568 kW/(°C·m2). tions and thermodynamic properties are expressed as highly nonlinear
The compressor cost (CCOM , US$) is calculated as a function of the equations. Population-based metaheuristic optimization algorithms are
work done (Wc , kW) [34]: more suitable for this type of black-box optimization problem, because
as they do not require any derivative information like deterministic
CCOM = 9, 560·Wc 0.82 (9)
algorithms [35–37]. In this work, the applied metaheuristic optimiza-
tion algorithm is a self-adapting dynamic differential evolution
3.2. Constraints (SADDE) [38].
A simulation-based optimization framework integrating Aspen Plus
The four distillation configurations were developed using Aspen and MATLAB is demonstrated in Fig. 6. Aspen Plus tackles the process
Plus, so the constraints of rigorous distillation MESH equations and simulation, and MATLAB with SADDE program is responsible for the
thermodynamic models were seamlessly implemented. The product black-box optimization in which the independent decision variables for
specifications were satisfied by using the Design Spec/Vary option. In each configuration are embedded. The Aspen Plus and MATLAB are
addition, since the SM will self-polymerize at high temperature, the linked through an Active X/COM interface [39].
bottom temperature is set at no more than 120 °C. The SADDE algorithm has three control parameters – population
size NP, mutation factor F (including lower/upper bound Fl/Fu), and
3.3. Optimization framework crossover probability CR (including lower/upper bound CRl/CRu). The
applied SADDE algorithm tuning parameters are as follows: population
The optimization of the proposed configurations contains both size NP = 100, mutation factor [Fl, Fu] = [0.1, 0.9], and crossover rate

6
X. Li, et al. Separation and Purification Technology 228 (2019) 115760

Fig. 9. (a) The optimum SHRT configuration, (b) T-H composite curve, and (c) grand composite curve.

[CRl, CRu] = [0.1, 0.9]. The number of generations is 300 for each utilization of half amount of waste heat by creation of a temperature
optimization. cascade, as shown in the pocket of the grand composite curve. In this
manner, the OPEX was halved at the expense of more CAPEX associated
4. Techno-economic analysis with the additional column. The significant reduction in the OPEX re-
sulted in an overall 7.76% TAC reduction compared to the CDiC. The
4.1. Feed flow rate = 100 kmol/h resulting optimum configuration demonstrated that partial heat in-
tegration is slightly better than the full integration since only light duty
A techno-economic analysis was performed to evaluate the separa- is required from the added trim reboiler. However, from a control
tion cost for producing SM using different configurations with the feed perspective, partial integration can provide much more robust control
flow rate of 100 kmol/h. The optimum CDiC, DED, SHRT, and DED- compared to a full integration process, because the former has more
SHRT configurations, as well as their T-H composite curves and grand degrees of freedom in control than the latter [40].
composite curves are shown in Figs. 7–10, respectively. A thorough When SHRT is incorporated into the design, the optimum config-
comparison of the optimum configuration is given in Table 1. uration shows that the FP is eliminated, leaving the PCS and CIP se-
For the optimum CDiC, the amount of low-grade waste heat dis- quentially implemented. As demonstrated in the grand composite
carded in the condenser was about the same amount as the supplied curve, the heat cascade pocket is close, meaning that the entire waste
high-quality energy, as demonstrated in the T-H diagrams. Its OPEX heat is upgraded and circulated without the need for any additional
strongly influences the TAC, whereas, for the rest of the configurations steam utility. Since the column is entirely driven by electricity in
employing waste heat recovery measures, the annual CAPEX was steady-state mode, it is possible to use renewable energy sources in-
dominant and consisted of a reduction of 50–77% in OPEX. stead of fossil fuel combustion to improve plant sustainability. The
The DED design improved the performance of the CDiC by self- SHRT appears to be the configuration with the lowest TAC of all the

7
X. Li, et al. Separation and Purification Technology 228 (2019) 115760

Fig. 10. (a) The optimum DED-SHRT configuration, (b) T-H composite curve, and (c) grand composite curve.

Table 1 throughput of the DED-SHRT compressor (183.683 kmol/h) was less


Comparison of four distillation configurations with PBP of 3 years (feed flow than in the SHRT (381.571 kmol/h), but the former had a larger
rate = 100 kmol/h). compression ratio (4.3) than the latter (2.1). Compared to the T-H
CDiC DED SHRT DED-SHRT diagrams of the SHRT configuration, the DED-SHRT was sufficient to
divide the large pocket of the SHRT into two small pockets. However,
Cooling water duty (kW) 4015 2159 537 560 even though the DED-SHRT provided the lowest OPEX, the soaring
Steam duty (kW) 3819 1902 0 0
CAPEX made it even worse than a CDiC. Also, the DED-SHRT is difficult
Electrical power (kW) 0 0 444 416
OPEX (106 US$/a) 0.8966 0.4482 0.2204 0.2069 to control due to the strong interaction between the two columns.
Total CAPEX (106 US$) 2.0524 3.0300 3.6979 4.7067 Therefore, from both the economic and dynamic aspects, DED-SHRT
Annual CAPEX (106 US$/a) 0.6841 1.0100 1.2326 1.5689 should be eliminated in the conceptual design stage.
TAC (106 US$/a) 1.5808 1.4582 1.4530 1.7758 With an optimization objective function of the PBP set at 3 years,
OPEX/TAC ratio 0.567 0.307 0.152 0.117
TAC reduction – 7.76% 8.08% −12.34%
determining the economic performance of each configuration in the
long-term production is an interesting exercise. Fig. 11 shows the effect
of PBP on TAC for each configuration in the range of 3–15 years. When
advanced configurations. On the other hand, it is also important to the PBP is longer than four years, the CDiC is found to be the least
consider the start-up procedure, for which two methods can be em- competitive configuration. With the increase of PBP, the TAC of DED-
ployed. One is the use of an electric heater, in which case the com- SHRT is significantly reduced and begins to outperform the DED at
pressor with the adiabatic efficiency of zero can be used as the electric seven years, because it has the lowest OPEX. However, the SHRT con-
heater. The other is the installation of an auxiliary boiler or a fired figuration always beats the other competitors in the given PBP range,
heater as a back-up heat source. In the case of vamping the current and is found to be the best configuration from the perspective of TAC. In
distillation column, the boiler or the fired heater can be left as it is and addition, it should be noted that the technical parameters of the column
be reused as the back-up heater. (pressure, diameter, stage number, etc.) in the SHRT configuration are
The DED-SHRT configuration that was also completely driven by close to that in the CDiC, which means that the existing CDiC can be
electricity produced the lowest OPEX, which demonstrated the sy- reused and retrofitted to the SHRT with minor modifications. In detail,
nergistic effect of combining two energy-efficient measures. The the existing column, condenser, and reboiler can be maintained, so that

8
X. Li, et al. Separation and Purification Technology 228 (2019) 115760

only a new CIP heat exchanger and a compressor are needed, with a
total capital investment of 1,466,000 US$. The OPEX saving is 676,200
US$ each year, which means it will take only 2.17 years to recover the
additional capital investment. Chemical/Petrochemical plants normally
have an annual maintenance period of 10–14 days for major repair and
replacement [41]. The proposed retrofit from CDiC to SHRT is the
preferred approach within this period to avoid production loss.

4.2. Feed flow rate = 1000 kmol/h

For a large treatment capacity, a comparison of different distillation


configurations is given in Table 2. The detailed design parameters are
provided in the Supporting Information. As these data show, as the feed
flow rate is increased, the alternative configurations become much
more attractive over the base case since the TAC reductions are larger.
This means that the process scale greatly affects the final economics.
Compared to the data in Tables 1 and 2, the OPEX is nearly propor-
tional to the treatment capacity, so the TAC reductions are derived
primarily from the decrease in CAPEX. For different treatment capa-
Fig. 11. The effect of PBP on TAC for four distillation configurations (feed flow cities, the approximate CAPEX can be estimated by:
rate = 100 kmol/h).
CAPEX 1 F1 α
=⎛ ⎞
CAPEX 2 ⎝ F2 ⎠ (10)
Table 2
Comparison of four distillation configurations with PBP of 3 years (feed flow where F denotes the treatment capacity and α the scale factor. Nor-
rate = 1000 kmol/h). mally, the scale factor varies from 0.3 to 1, but the average is very close
to 0.6, thus it is called the rule of six-tenths [42]. If α = 1, the CAPEX
CDiC DED SHRT DED-SHRT
will be proportional to the capacity like the OPEX. Apparently, the
Cooling water duty (kW) 38,680 20,859 5256 5341 lower the scale factor, the greater the economic benefits obtained by the
Steam duty (kW) 36,584 18,209 0 0 increased process scale. Table 3 provides the calculated scale factor
Electrical power (kW) 0 0 4294 3814 based on the two treatment capacities. Note that the CDiC and DED
OPEX (106 US$/a) 8.5914 4.2925 2.1311 1.8996
have scale factors close to 0.6, but the configurations with heat pump
Total CAPEX (106 US$) 8.8442 12.0865 18.7075 22.5216
Annual CAPEX (106 US$/a) 2.9481 4.0288 6.2358 7.5072 have larger factors. The factor difference is mainly the result of the
TAC (106 US$/a) 11.5395 8.3213 8.3669 9.4068 conservative compressor cost correlation given in Eq. (9). Recently,
OPEX/TAC ratio 0.745 0.516 0.255 0.202 Luyben [43] has showed that the estimation of compressor CAPEX has a
TAC reduction – 27.89% 27.49% 18.48% great influence in conceptual design stage. After comparing the dif-
ferent correlations listed in this literature report [43], we found that Eq.
(9) predicts a very high value of compressor CAPEX. However, even
Table 3
with this conservative estimation of the heat pump cost, the SHRT still
Calculated scale factor by using feed flow rates of 100 kmol/h and 1000 kmol/
outperformed the other configurations in long-term production, as
h.
shown in Fig. 12. Therefore, the SHRT configuration is recommended as
CDiC DED SHRT DED-SHRT the most desirable energy-efficient design for industrial EB/SM se-
Size factor α 0.63 0.60 0.70 0.68
paration.

5. Conclusions

This study described a process synthesis and optimization metho-


dology for improving the cost-intensive EB/SM separation. Three dif-
ferent distillation configurations employing DED and SHRT concepts
were proposed, optimized, and compared in terms of their capability to
conserve product separation costs. Using a short PBP of 3 years with a
small feed flow rate (100 kmol/h), the DED and SHRT configurations
were found to reduce the TAC by ~8% compared to the base case. The
TAC reduction can be improved up to ~28% with a larger capacity
(100 kmol/h) due to the effects of scaling. However, with the increase
in PBP, SHRT was found to be the best choice from the perspective of
economics. The synergistic DED-SHRT consumes the lowest OPEX, but
the overwhelmingly high CAPEX makes it only economically viable
under a long-term production. Also, it was observed that the technical
parameters of the CDiC are close to that in the SHRT, which meant that
the former can be retrofitted to the latter with minor modifications.
This proposed retrofit could provide chemical engineers with an alter-
native way to enhance the conventional EB/SM column. Since the de-
sign and optimization methods proposed in this reported work can si-
multaneously reduce the costs and energy consumptions, it is
Fig. 12. The effect of PBP on TAC for four distillation configurations (feed flow
recommended that they can also be implemented to other energy-in-
rate = 1000 kmol/h).
tensive close-boiling distillation processes.

9
X. Li, et al. Separation and Purification Technology 228 (2019) 115760

Acknowledgement [19] M. Nakaiwa, K. Huang, A. Endo, T. Ohmori, T. Akiya, T. Takamatsu, Internally heat-
integrated distillation columns: a review, Chem. Eng. Res. Des. 81 (2003) 162–177.
[20] B. Suphanit, Design of internally heat-integrated distillation column (HIDiC): uni-
The authors acknowledge financial support from National Key form heat transfer area versus uniform heat distribution, Energy 35 (2010)
Research and Development Program of China (No. 2018YFB0604903), 1505–1514.
and National Nature Science Foundation of China (Nos. 21336007, [21] C. Cui, X. Li, H. Sui, J. Sun, Quick decision-making for close-boiling distillation
schemes, Ind. Eng. Chem. Res. 56 (2017) 5078–5091.
21776202). [22] J. Cabrera-Ruiz, A. Jimenez-Gutierrez, J.G. Segovia-Hernandez, Assessment of the
implementation of heat-integrated distillation columns for the separation of ternary
Appendix A. Supplementary material mixtures, Ind. Eng. Chem. Res. 50 (2011) 2176–2181.
[23] T. Wakabayashi, K. Yoshitani, H. Takahashi, S. Hasebe, Verification of energy
conservation for discretely heat integrated distillation column through commercial
Supplementary data to this article can be found online at https:// operation, Chem. Eng. Res. Des. 142 (2019) 1–12.
doi.org/10.1016/j.seppur.2019.115760. [24] Y. Kansha, N. Tsuru, K. Sato, C. Fushimi, A. Tsutsumi, Self-heat recuperation
technology for energy saving in chemical processes, Ind. Eng. Chem. Res. 48 (2009)
7682–7686.
References [25] Y. Kansha, M. Ishizuka, C. Song, A. Tsutsumi, An innovative methanol synthesis
process based on self-heat recuperation, Appl. Them. Eng. 70 (2014) 1189–1194.
[1] C. Nederlof, Catalytic Dehydrogenations of Ethylbenzene to Styrene, PhD Thesis [26] K. Matsuda, K. Kawazuishi, Y. Kansha, C. Fushimi, M. Nagao, H. Kunikiyo,
Delft University of Technology, 2012. F. Masuda, A. Tsutsumi, Advanced energy saving in distillation process with self-
[2] C. Cui, X. Li, D. Guo, J. Sun, Towards energy efficient styrene distillation scheme: heat recuperation technology, Energy 36 (2011) 4640–4645.
from grassroots design to retrofit, Energy 134 (2017) 193–205. [27] M. Aziz, Y. Kansha, A. Kishimoto, Y. Kotani, Y. Liu, A. Tsutsumi, Advanced energy
[3] H. Ba, S. Podila, Y. Liu, X. Mu, J.M. Nhut, V. Papaefthimiou, S. Zafeiratos, saving in low rank coal drying based on self-heat recuperation technology, Fuel
P. Granger, C. Pham-Huu, Nanodiamond decorated few-layer graphene composite Process. Technol. 104 (2012) 16–22.
as an efficient metal-free dehydrogenation catalyst for styrene production, Catal. [28] Y. Kansha, A. Kishimoto, T. Nakagawa, A. Tsutsumi, A novel cryogenic air se-
Today. 249 (2015) 167–175. paration process based on self-heat recuperation, Sep. Purif. Technol. 77 (2011)
[4] V.A. Welch, Cascade Reboiling of Ethylbenzene/Styrene Columns, US Patent, US 389–396.
6171449 B1, 2001. [29] K. Matsuda, Y. Kansha, C. Fushimi, A. Tsutsumi, A. Kishimoto, Advanced Energy
[5] W.L. Luyben, Design and control of the styrene process, Ind. Eng. Chem. Res. 50 Saving and its Applications in Industry, Springer, 2013.
(2011) 1231–1246. [30] W.L. Luyben, Distillation column pressure selection, Sep. Purif. Technol. 168 (2016)
[6] N.B. King, Ethylbenzene-styrene separation, US Patent, 3904484, 1975. 62–67.
[7] H.H. Sardina, Dehydrogenation process for production of styrene from ethylben- [31] C. Cui, S. Liu, J. Sun, Optimal selection of operating pressure for distillation col-
zene comprising low temperature heat recovery and modification of the ethylben- umns, Chem. Eng. Res. Des. 137 (2018) 291–307.
zene-steam feed therewith, US Patent, 4628136, 1986. [32] M.B. Leo, A. Dutta, S. Farooq, Process synthesis and optimization of heat pump
[8] A.C. Dimian, C.S. Bildea, Energy efficient styrene process: design and plantwide assisted distillation for ethylene-ethane separation, Ind. Eng. Chem. Res. 57 (2018)
control, Ind. Eng. Chem. Res. 58 (2019) 4890–4905. 11747–11756.
[9] M.T.G. Jongmans, E. Hermens, M. Raijmakers, J.I.W. Maassen, B. Schuur, A.B. de [33] C.C.E. Christopher, A. Dutta, S. Farooq, I.A. Karimi, Process synthesis and optimi-
Hann, Conceptual process design of extractive distillation process for ethylbenzene/ zation of propylene/propane separation using vapor recompression and self-heat
styrene separation, Chem. Eng. Res. Des. 90 (2012) 2086–2100. recuperation, Ind. Eng. Chem. Res. 56 (2017) 14557–14564.
[10] M.T.G. Jongmans, B. Schuur, A.B. de Haan, Ionic liquid screening for ethylbenzene/ [34] W.L. Luyben, Principles and Case Studies of Simultaneous Design, John Wiley &
styrene separation by extractive distillation, Ind. Eng. Chem. Res. 50 (2011) Sons, Inc., 2011.
10800–10810. [35] C. Cui, J. Sun, Rigorous design and simultaneous optimization of extractive dis-
[11] C.A. Koval, T. Spontarelli, P. Thoen, R.D. Noble, Swelling and thickness effects on tillation systems considering the effect of column pressures, Chem. Eng. Process.
the separation of styrene and ethylbenzene based on facilitated transport through Process Intensif. 139 (2019) 68–77.
ionomer membranes, Ind. Eng. Chem. Res. 31 (1992) 1116–1122. [36] J. Javaloyes-Anton, R. Ruiz-Femenia, J.A. Caballero, Rigorous design of complex
[12] B. Cao, M.A. Henson, Modeling of spiral wound pervaporation modules with ap- distillation columns using process simulators and the particle swarm optimization
plication to the separation of styrene/ethylbenzene mixtures, J. Membr. Sci. 197 algorithm, Ind. Eng. Chem. Res. 52 (2013) 15621–15634.
(2002) 117–146. [37] M. Vazquez-Ojeda, J.G. Segovia-Hernandez, S. Hernandez, A. Hernandez-Aguirre,
[13] M. Maes, F. Vermoortele, L. Alaerts, S. Couck, C.E. Kirschhock, J.F.M. Denayer, A.A. Kiss, Design and optimization of an ethanol dehydration process using sto-
D.E. De Vos, Separation of styrene and ethylbenzene on metal – organic frame- chastic methods, Sep. Purif. Technol. 105 (2013) 90–97.
works: analogous structures with different adsorption mechanisms, J. Am. Chem. [38] C. Cui, X. Zhang, J. Sun, Design and optimization of energy-efficient liquid-only
Soc. 132 (2010) 15277–15285. side-stream distillation configurations using a stochastic algorithm, Chem. Eng. Res.
[14] T. Remy, L. Ma, M. Maes, D.E. De Vos, G.V. Baron, J.F.M. Denayer, Vapor-phase Des. 145 (2019) 48–52.
adsorption and separation of ethylbenzene and styrene on the metal – organic [39] Aspen Technology, Aspen Plus User Guide, Aspen Technology, Inc., 2000.
frameworks MIL-47 and MIL-53 (Al), Ind. Eng. Chem. Res. 51 (2012) 14824–14833. [40] W.L. Luyben, Design and control of a fully heat-integrated pressure-swing azeo-
[15] C. Cui, H. Yin, J. Yang, D. Wei, J. Sun, C. Guo, Selecting suitable energy-saving tropic distillation system, Ind. Eng. Chem. Res. 47 (2008) 2681–2695.
distillation schemes: making quick decisions, Chem. Eng. Process. Process Intens. [41] N.V.D. Long, M. Lee, Optimal retrofit of a side stream column to a dividing wall
107 (2016) 138–150. column for energy efficiency maximization, Chem. Eng. Res. Des. 91 (2013)
[16] C. Cui, Z. Xi, S. Liu, J. Sun, An enumeration-based synthesis framework for multi- 2291–2298.
effect distillation processes, Chem. Eng. Res. Des. 144 (2019) 216–227. [42] R.W. Whitesides, Process Equipment Cost Estimating by Ratio and Proportion,
[17] A.K. Jana, Advances in heat pump assisted distillation column: a review, Energy 2012. < https://www.pdhonline.com/courses/g127/g127content.pdf > .
Convers. Manage. 77 (2014) 287–297. [43] W.L. Luyben, Capital cost of compressors for conceptual design, Chem. Eng.
[18] A.A. Kiss, S.J.F. Landaeta, C.A.I. Ferreira, Towards energy efficient distillation Process. Process Intensif. 126 (2018) 206–209.
technologies – making the right choice, Energy 47 (2012) 531–542.

10

You might also like