- Email: [email protected]
Energy-saving dividing-wall column design and control for benzene extraction distillation via mixed entrainer
Chemical Engineering and Processing 100 (2016) 49–64
Contents lists available at ScienceDirect
Chemical Engineering and Processing: Process Intensification journal homepage: www.elsevier.com/locate/cep
Energy-saving dividing-wall column design and control for benzene extraction distillation via mixed entrainer Xin Dai, Qing Ye* , Jiwei Qin, Hao Yu, Xiaomeng Suo, Rui Li Jiangsu Key Laboratory of Advanced Catalytic Materials and Technology, School of Petrochemical Engineering, Changzhou University, Changzhou, Jiangsu 213164, China
A R T I C L E I N F O
A B S T R A C T
Article history: Received 30 August 2015 Received in revised form 5 November 2015 Accepted 29 November 2015 Available online 2 December 2015
Benzene and cyclohexane form an azeotrope (Benzene, 49.7 w.t%, 1 atm), which can be separated by extractive distillation. In this study, o-xylene is added into the sulfolane entrainer for mixed entrainer to make the operation conditions more convenient and controllable. The TAC (total annual cost, generally utilized in global economic optimization) of the process with o-xylene as mixed entrainer is less than that of the process with SULF as entrainer. In addition, as the most energy saving process, the process of extractive dividing wall column (EDWC) is proposed, for it can save 44.0% of energy and 35.8% of TAC, compared with the conventional extractive distillation arrangement with pure entrainer. Besides, two control structures are established for EDWC, and their dynamic performances are evaluated by feed flow rate disturbances and feed composition disturbances. Finally, it is found that the disturbances are well handled with the second control structure. ã 2015 Elsevier B.V. All rights reserved.
Keywords: Benzene Cyclohexane O-xylene Extractive dividing wall column Energy saving
1. Introduction Benzene (BEN) is an important basic chemical materials, which can be used to synthesize a variety of fine organic chemical intermediates. The market has great demand in high purity BEN, so it has great social effects to produce BEN products to suit market demand. In the production of catalytic reforming or pyrolysis hydrogenation, there is usually existing a mixture of BEN and cyclohexane (CYH). BEN and CYH can form azeotrope, which makes their separation via traditional distillation column unrealized. Therefore, an effective alternative separation technique is highly required in industry. There are two common methods to separate this system in industry. One is liquid–liquid extraction. Liquid–liquid extraction is the unit operation which uses different solubility of components in solvents to separate the mixture. It is widely used in chemical, metallurgy, food industry and petroleum refining industry. In the aromatics extraction process, the ideal solvent is sulfolane (SULF). Because it can obtain higher aromatic hydrocarbon recovery rate and can be dissolved into two phase separation and dissolution agent. Schuur et al. studied different entrainer in liquid–liquid extraction, they considered crown ether based extractants, metal complexes, metalloids and so on [1]. Billard et al. discussed
* Corresponding author. Fax: +86 519 86330355. E-mail address: [email protected] (Q. Ye). http://dx.doi.org/10.1016/j.cep.2015.11.014 0255-2701/ ã 2015 Elsevier B.V. All rights reserved.
actinides and lanthanides by using ionic liquids in liquid–liquid extraction [2]. The other method is aromatics (BEN) extraction distillation which is an extractive distillation process. Extractive distillation is a special distillation method, which continuous adding entrainer into the distillation column to change the relative volatility of the separation components. Many different entrainers can be used in extractive distillation. SUFL can also be used in extractive distillation. Qin et al. discussed the use of extractive distillation for the separation of BEN-CYH via pure entrainer [3]. SUFL was used as entrainer and two control structures were proposed. The SUFL would be thermally decomposed when the base temperature was over 200 C. The pressure of the condenser in the entrainer recovery column (ERC) was 0.08 atm and the top temperature of ERC was 15.7 C. Because the temperature was too low to use the convenient water as condensing intermedium, the operating cost would increase. Sun et al. studied separation of BEN–CYH system by extractive dividing wall column with furfural as entrainer [4]. However, because the boiling point of furfural was too low, there were some problems after treatment. Due to the continuous improvement of limited energy resources, the process has been strengthened, which has caused wide attention in the world [5–7]. For further energy savings, the new configuration with integrated two columns into one shell, which is identified as a dividing wall distillation column (DWC) [8]. Gómez-Castro et al. found that 33% reduction of energy had been saved in DWC when they studied six cases of different feed
50
X. Dai et al. / Chemical Engineering and Processing 100 (2016) 49–64
Nomenclature
EDWC DWC BEN CYH SULF OX F R/F EF EDC ERC EDC-C EDC-B RC F2 RR EF/F NF NEF NTEDC-C NTEDC-B NTRC XD1 XD2 HX KU PU Kc
tI
QR TC TP QR/F
Extractive dividing wall column Dividing wall distillation column Benzene Cyclohexane Sulfolane O-xylene; The fresh feed of the extractive distillation column; The mass ratio of reflux flow rate/flow rate The flow rate of entrainer The extractive distillation column The entrainer recovery column The extractive distillation column of CYH The BEN recovery column The entrainer recovery column The feed of ERC The reflux ratio The molar ratio of mixture entrainer to feed The stage of fresh feed The mixture entrainer feed stage The number of total stages of EDC-C The number of total stages of EDC-B The number of total stages of RC The purity of top product in EDC-C The purity of top product in EDC-B Heat exchangers Ultimate gains Ultimate periods Gain Integral time Reboiler duty Temperature controller The top pressure of column Reboiler duty/mole flow rate of F
components of DWC [9]. Emtir et al. found that there was 30% reduction of the total annual cost of DWC when they studied a ternary mixture system with three different feed compositions [10]. Compared with the conventional structures, DWCs can reduce the cost of capital and energy [11–14]. Kiss and Suszwalak [11] studied an extractive dividing wall column (EDWC) for the purpose of bioethanol dehydration, and the result showed that 9.4% reboiler duty can be saved as compared with conventional extractive distillation process. Zhang et al. studied EDWC simulated as three columns model. They found that EDWC can save more energy than conventional extractive distillation arrangement [14]. Another important aspect of EDWC should be considered is dynamic control. EDWC have interactions and inner structures among control loops which make it much more difficult to control than that of conventional extractive distillation arrangement [13]. Sun et al. used the composition control cascaded with the molar ratio of mixture entrainer to feed and the composition control cascaded with temperature control in the EDWC [4]. Zhang et al. studied the separation of ethyl acetate-isopropyl alcohol system by extractive dividing wall column and the temperature control was cooperated with the composition control [14]. Xia et al. studied separation of methylal–methanol system by extractive dividing wall column and two control schemes were all evaluated [15]. In this study, BEN and CYH azeotrope is separated through extractive distillation. The entrainer is the mixture of SULF and oxylene (OX). Therefore, the temperature of bottom is lower than the decomposition temperature of SULF, and water can be utilized
as the cooling medium at the top of the column. In order to save energy, EDWC simulated as a three-column model is studied in this paper. In terms of simulation studies, three-column model is more close to EDWC concept compared with the models simulated as two columns. Total annual cost (TAC) is calculated to obtain the optimal conventional extractive distillation process and EDWC with mixed entrainer. The separation of BEN-CYH using mixed entrainer, which is SULF mixed with OX via EDWC is studied for the first time. Subsequently, two control structures are proposed. There has been no published literature about the detailed control of EDWC with mixed entrainer simulated as a three-column model. 2. Optimization of extractive distillation process 2.1. Thermodynamic model and mixed entrainer screening and selection In this work, NRTL model is used in the Aspen simulations. BEN and CYH have very similar boiling points (80.13 C and 80.78 C), and the BEN/CYH mixture has an azeotrope with composition of 55.01 mol% BEN at atmospheric pressure. In the earlier work [3], BEN–CYH system was separated by extractive distillation using SULF as entrainer. To avoid thermally decomposition, the entrainer recovery column (ERC) was operated at 0.08 atm. However, the top temperature of ERC was 15.7 C, which was too low to use water as cooling medium. To solve this problem, another lower boiling point component is added into the entrainer so that the temperature in bottom of column can be lower than that of thermally decomposition (200 C) and meanwhile the top temperature in column is high enough to use water as cooling medium. O-xylene (OX) has no selectivity for the separation of BEN–CYH system and its boiling point (144.4 C) is between BEN (80.13 C) and SULF (287.3 C). OX had no selectivity for separation of aromatics but can help increase mutual solubility of SUFL and BEN. So, it is used as a cosolvent of SULF in the process and the compatibility is commendable. The concentration of OX in mixed entrainer is important to the effect of separation. Usually the top pressure of the EDWC is designed to be able to use water as the cooling medium. Comparison of the relative volatilities in the presence of different entrainers is a criterion for the entrainer selection [16]. The relative volatility is more higher, the separation process is more easier. Fig. 1 shows the pseudo-binary VLE of BEN/ CYH mixture with SULF and OX to feed mole ratio 1.1 by NRTL
Fig. 1. Pseudo-binary x-y plot for the BEN–CYH system with mixed entrainer to feed molar ratio 1.1.
X. Dai et al. / Chemical Engineering and Processing 100 (2016) 49–64
51
2.2. The steady-state design of conventional extractive distillation process with mixed entrainer
Fig. 2. The effect of mole fraction of OX on temperature of bottom in EDWC.
model. The less the mole fraction of OX use in mixed entrainer, the larger relative volatility gets. Fig. 2 shows the effect of mole fraction of OX on temperature of bottom in EDWC when the top pressure of column is 0.6 atm. When the mole fraction of OX is more than 10%, the base temperature of the column is lower than that of the thermally decomposition (200 C). The top temperature of EDWC is 64.4 C at 0.6 atm so that water can be used as cooling medium. So, the 10% mole fraction of OX is the best choice.
In this section, conventional extractive distillation process with mixed entrainer is proposed. The process is shown in Fig. 3. The entrainer and the mixture of BEN–CYH feed in the extractive distillation column (EDC) from the upper tray and the lower tray, respectively. The CYH as the top product exits at the top of the EDC, while the mixture entrainer and the BEN exit at the bottom of the EDC. Then the BEN and the mixture entrainer are separated in the entrainer recovery column (ERC). BEN of the lower boiling point is distillated from the top of the ERC, and the mixture entrainer is produced from the base of the ERC. The mixture entrainer product is recycled to the EDC after being cooled at the entrainer cooler. To compensate the loss of the mixture entrainer, a small quantity of fresh mixture entrainer is made up to the entrainer feed. In previous work [3], the steady state design had done some optimization on the operating conditions. To compare with the earlier work [4], the flow sheet in this work is similar with it. The mixture feed flowrate of BEN-CYH is 100 kmol/h and the mole fraction of BEN is 0.5. The optimal mole fraction of OX is 0.1 in the mixed entrainer. The mole fraction of products are both specified at 0.998 (kmol/kmol). The RadFrac model in Aspen plus is used to simulate the steady state of extractive distillation system. Through the economic optimization, the optimal design parameters can be obtained. Furthermore, the minimal total annual cost can be obtained. The global optimization of this process is shown in Fig. 4. The final steady-state flowsheet and the detail information of the flowsheet are illustrated in Fig. 3. The temperature profiles of conventional extractive distillation process are shown in Figs. 5 and 6, respectively.
Fig. 3. The optimal steady state of the two-column flowsheet.
52
X. Dai et al. / Chemical Engineering and Processing 100 (2016) 49–64
Fig. 4. The optimization procedure of the extractive distillation process.
2.3. The steady-state design of extractive dividing wall (edwc) column process with mixed entrianer In this part, DWC will be designed with the same feed conditions and product specifications. For saving the investment cost and operating cost, a dividing-wall is added into the process which is based on the conventional extractive distillation process. In this work, the EDWC concept is applied to the same system with OX as the mixed entrainer. Fig. 7 shows the configuration of an EDWC. EDWC consists of three parts. The left side of the dividingwall is included in part 1. On the right side of the dividing-wall is part 2, while part 3 is under part 1 and part 2. The fresh feed and mixed entrainer are fed into part 1. CYH and BEN are distillate stream in part 1 and part 2, respectively. The mixed entrainer is get
from the bottom and recycled back to part 1. Only one reboiler is used in this model. The vapor from the bottom is split by the wall into part1 and part 2. It is depicted in Fig. 7 that vapor split ratio (av) is defined as V1/V3. Because the top pressure of column is 0.6 atm and the drop of tray pressure is 0.0068 atm in the three columns, the compressor in the model is just an illusionary equipment to improve the pressure of vapor coming from the top of column RC so that the vapor can go back to EDC-B and EDC-C. In this paper, a three-column model is selected to simulate EDWC displayed in Fig. 8. In terms of simulation studies, threecolumn model is more close to EDWC concept compared with the models simulated as two columns. In this model, av can be set directly. Then, the model is closer to EDWC structure. This model has three columns: the extractive distillation column of CYH (EDC-
X. Dai et al. / Chemical Engineering and Processing 100 (2016) 49–64
53
Fig. 5. The temperature profile of EDC.
Fig. 8. Process flow diagram of EDWC.
both are stage 1. The total stages of the entrainer recovery column (RC) is 6, the reboiler is stage 6. 2.4. Global optimization of EDWC
Fig. 6. The temperature profile of ERC.
C), the recovery column of BEN (EDC-B) and the recovery column (RC). It is assumed that EDC-C and EDC-B have a total of 36 stages and 25 stages, respectively. The condenser of EDC-C and EDC-B
On the base of the Luyben’s papers [17,18], the total annual cost (TAC) includes the cost of main devices and the cost of the energy consumption. The main devices include the column vessels, the condensers, the reboilers and the entrainer cooler. The other accessorial small equipments (the valves and the pumps) can be neglected. The related calculation data and formulas of the TAC are based on William Luyben (2008). The total capital cost (TAC) including the capital and operating cost, it has the following statement: TAC ¼
Fig. 7. Configuration of main column.
total capital cost þ total operating cost pay back period
Fig. 9. The calculation of EDWC diameter.
ð1Þ
54
X. Dai et al. / Chemical Engineering and Processing 100 (2016) 49–64
It is assumed to be a three-year payback period. Douglas [19] and Luyben [20] suggested that a lot of parameters of EDWC can be optimized. 0.568 and 0.852 (kW/K m2) [21–23] are heat transfer coefficients of reboiler and condenser, respectively. EDWC configuration has only one shell and the equivalent diameter [23] of EDWC is calculated as illuminated in Fig. 9. The formula of equivalent diameter is as follow: SðEDC C Þ ¼
p DðEDC C Þ2 4
SðEDC BÞ ¼
p DðEDC BÞ2 4
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi DðEDWCÞ ¼ 2 S½ðEDC C Þ þ SðEDC BÞ=p qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ DðEDC C Þ2 þ DðEDC BÞ2
ð2Þ be
ð3Þ
ð4Þ
The total stages of EDWC include those of EDC-C and RC. It can observed that the relatively higher cost related to
Fig. 10. The process of the TAC calculation.
X. Dai et al. / Chemical Engineering and Processing 100 (2016) 49–64
Fig. 11. Effects of RR1 and EF (koml/h) in the EDWC on (a) the CYH composition (XD1) in the EDC-C and (b) XD1 in the EDC-B.
Fig. 12. The effects of entrainer flow rate (EF) and NTEDC-C on TAC.
Fig. 14. Effects of NTEDC-B on the TAC of EDWC.
Fig. 13. The effects of the feed stage (NF) and the entrainer feed stage (NEF) on the TAC of EDWC.
Fig. 15. Effects of NTRC on the TAC of EDWC.
55
56
X. Dai et al. / Chemical Engineering and Processing 100 (2016) 49–64
Fig. 16. The optimized conditions of EDWC.
manufacturing EDWC should be taken into consideration in the economic analysis. It has been assumed that the capital cost of EDWC is 1.2 times more than that of conventional distillation column. Zhang et al. has done some research of the TAC of EDWC simulated as three columns [14]. According to their previous work, sequence for EDWC simulated as three columns is indicated as Fig. 10. Based on the product specifications, EDWC process gets the optimal TAC by adjusting the following parameters: the fresh feed stage (NF), the entrainer feed stage (NEF), the stages of EDC-C (NTEDC-C), the stages of EDC-B (NTEDC-B) the stages of RC (NRC), and the av. To clearly, the sequence of calculating optimal TAC is shown in Fig. 10. The optimization procedure of EDWC is as follow: (1) Give the top pressure of EDC-C and EDC-B (PEDC-C = 0.6 atm, PEDC-B = 0.6 atm); (2) Give the total stages of RC (NTRC); (3) Give the total stages of EDC-C (NTEDC-C); (4) Give the total stages of EDC-B (NTEDC-B); (5) Obtain the flow rate of EF by the steady-state simulation in Aspen plus, and select the operating EF; (6) Give the feed stage (NF) and the entrainer feed stage (NEF) of the EDWC;
Fig. 17. Stage temperature of EDWC as one column shell.
X. Dai et al. / Chemical Engineering and Processing 100 (2016) 49–64
57
Table 1 The optimization of the TAC of two columns with/without mixed entrainer and EDWC. Items
N NF NEF EF/F Column diameter (m) The condencer duty QC (kW) The reboiler duty QR (kW) Capital investment ($1000) Operation cost ($1000) TAC ($1000/yr)
The conventional extractive distillation with pure entrainer
The conventional extractive distillation with mixed entrainer
The process of EDWC with mixed entrainer
EDC
ERC
EDC
ERC
EDC-C
29 20 4 1.1 0.98 922.00 304.360 590.915 583.395 780.367
10 5 10 1.1 1.460 714.00 1091.000 722.346 387.852 628.634
42 28 10 1.1 1.003 944.70 413.183 487.879 338.637 501.269
28 14
42 25
0.808 607.24 945.115
1.467 661.44 1084.328
EDC-B
422.89
Fig. 18. The basic control structure (CS1).
(7) Adjust the reflux ratios (RR) of EDC-C to make the two products satisfy the purities requirement; (8) Go back to the step 6 and give new conditions until TAC of EDWC is the minimal; (9) Go back to the step 5 and give new conditions until TAC of EDWC is the minimal; (10) Go back to the step 4 and select new NTEDC-B until TAC of EDWC is the minimal; (11) Go back to the step 3 and select new NTEDC-C until TAC of EDWC is the minimal; (12) Go back to the step 2 and give new NTRC until TAC of EDWC is the minimal.
The effects of changing the entrainer flow rate (EF) and RR1 on the concentrations of components in distillate stream (D) from the EDC-C and EDC-B are studied. The results have been plotted in Fig. 11, in which the results are obtained when CYH product purity is satisfied with number of total stages of NEDC-C fixed at 34. The Table 2 Parameters of all temperature controllers. Parameters
TC
TC-HX
Ultimate gain Ultimate period (min) Gain (KC) Integral time (min)
4.9341 3.6 1.5419 7.92
1.2944 1.8 0.4045 3.96
58
X. Dai et al. / Chemical Engineering and Processing 100 (2016) 49–64
Fig. 19. Dynamic responses (CS1) to 20% feed flow rate disturbances.
plots in Fig. 11 show that, for a given EF, the product CYH purity reaches a maximum value with an optimum RR1. Fig. 12 shows the effect of entrainer flow rate (EF) and NTEDC-C on TAC. When NTEDC-C is 34 and EF is 110 kmol/h, TAC is minimal. The effects of the feed stage (NF) and the entrainer feed stage (NEF) on the TAC of EDWC are shown in Fig. 13, which reveals that minimum TAC exits when NF and NEF are recommended as 25 and 10, respectively. Figs. 14 and 15 show the effect of stages of NTEDC-B and NTRC on the TAC of EDWC, it is suggested that the minimal TAC existing when NTERC is 25 and NTRC is 6. Fig. 16 displays the optimized
conditions of EDWC. The temperature profiles of EDWC are shown in Fig. 17. Table 1 summarizes the results that the optimization of the TAC with/without mixed entrainer in conventional extractive distillation process and EDWC with mixed entrainer, the data of the process with pure entrainer comes from earlier work [3]. The reboiler of EDC is heated by low pressure (LP) steam of 6 bar and the reboiler of ERC is heated by high pressure (HP) steam of 42 bar and the condenser of ERC is cooled by refrigerant at 20 C in the earlier flowsheet of two columns. The conventional extractive distillation process with mixed entrainer can save 33.5% operating
X. Dai et al. / Chemical Engineering and Processing 100 (2016) 49–64
cost than that of the conventional extractive distillation without mixed entrainer. The operating cost of EDWC process can also be reduced by 44.0% that of the conventional extractive distillation with pure entrainer. The TAC of EDWC can save 35.8% than that of the conventional extractive distillation with pure entrainer. 3. Control strategy for the EDWC The EDWC control was studied in many papers, and most of them studied the special separation system and proposed the control structures [24–29]. The control of EDWC is more difficult than that of conventional extractive distillation arrangement. Moreover, the control of EDWC with mixed entrainer is much more difficult than that of EDWC. There is no published literature about the detailed control of EDWC with mixed entrainer simulated as
59
three columns model. Two control structures are proposed in this part. The sizes of EDWC are necessary to convert the steady-state simulation into a dynamic simulation. The Tray Sizing in Aspen plus is used to calculate the diameter of the columns. Reflux drums and the column bases are sized to provide 10 min of holdup when full. Dynamic process is completely pressure-driven when valves and pumps are inserted to the dynamic simulation. Then, it is ready to export the steady-state simulation to Aspen Dynamic. 3.1. Selecting temperature control trays Firstly, the tray temperature control is considered for the EDWC. Temperature control stages selection is important to temperature control structure. There are a lot of methods to select temperature control trays, such as the singular value decomposition (SVD)
Fig. 20. Dynamic responses of CS1 to composition disturbance.
60
X. Dai et al. / Chemical Engineering and Processing 100 (2016) 49–64
Fig. 21. The control structure of CS2.
method, the slope rule and sensitivity criterion. The slope rule is the easiest method in these three ways. The slope rule is chosen to select temperature control stages for EDWC in this work. As shown in Fig. 17, the 41th of EDWC is chosen as temperature control stage. Due to the addition of mixed entrainer, the temperature changing are not obvious in EDC-C and EDC-B, the sensitive stage is not selected. So, the reflux ratio control is used in EDC-C and EDC-B. 3.2. Control structures design 3.2.1. Basic control structure 1 (CS1) The basic control structure established for the three column system and the controller face plate are present in Fig. 18. All control loops of the basic control structure are listed as follows. (1) F is flow controlled (reserve acting). (2) The levels of the two reflux drums are controlled by manipulating the distillates, respectively (direct acting). (3) The column base levels of the EDC-C and EDC-B are controlled by manipulating the bottoms (direct acting) while the base level of RC is controlled by manipulating the flowrate of the makeup entrainer (reverse acting). (4) Top pressures of EDC-C and EDC-B are controlled by manipulating the heat removal duty of the two condensers (direct acting), respectively. (5) The mixed entrainer feed is flow controlled (reverse acting), and it is ratioed to F by cascade control. (6) The reflux flow rate of EDC-C and EDC-B are ratioed to F, respectively. (7) The 5th stage temperature of RC is controlled by manipulating reboiler duty (reverse acting). (8) The recycle entrainer temperature is maintained by manipulating the HX heat removal duty (reserve acting).
(9) Dead time of 1 min is inserted to all temperature control loops to match time drag in real plant environment. For all the flows, pressures and temperatures, conventional proportional and integral (PI) controllers are used. All level loops are P-only controllers with gain (Kc) = 2. For the pressure PI controllers, tuning parameters of Kc = 20 and integral time (t I) = 12 min are used, while Kc = 0.5 and t I = 0.3 min are used for the flow PI controllers. Considering the practical operation, a 1-min deadtime is inserted into each temperature control loop. Relayfeedback test and Tyreus-Luben turning are utilized to get the ultimate gains (KU) and ultimate periods (PU) of controllers for all of the control loops with dead time [30]. Parameters of all temperature controllers are presented in Table 2. To test the effectiveness of the CS1, large disturbances in the fresh feed flow rate and feed composition are made. Fig. 19 illustrates the dynamic responses of CS1 to 20% step changes in fresh feed flow rate at 0.5 h. Flow rate of CYH (D1), BEN (D2) are changed 20% correspondingly as well. However, the purities of CYH (XD1) and BEN (XD2) have large transient deviations. The transient deviation of XD1 and XD2 reaches about 0.935. The mole fraction of OX has large transient deviations when the feed flowrate undergoing 20% disturbances after 0.5 h. The transient deviation reaches about 0.035. Fig. 20 shows the results of basic control structure when undergoing the feed composition disturbances. The composition disturbances are also introduced at 0.5 h. As is shown in Fig. 20, large deviations occur in the purities of the two productions and the purities are all about 0.90 when the process reaches to a new steady state. Moreover, the mole fraction of OX has large deviation and need long time to reach to a new steady state.
X. Dai et al. / Chemical Engineering and Processing 100 (2016) 49–64
61
Fig. 22. Dynamic response (CS2) to flow rate disturbances.
3.2.2. An improved control structure (CS2) Because the OX is added into the process, the temperature of EDC-C is changed smoothly. The result of the temperature control in the EDWC with mixed entrainer is not very well. So, CS1 is not controlled very well. In order to reduce the large transient deviations, the basic control structure must be improved. The composition controller of BEN(CC1) is cascaded with D1/F ratio. The OX of bottom stream (B) composition controller (CC2) cascaded with TC. The corresponding controller faceplates of CS2 are shown in Fig. 21. The effects of entrainer flow rate on XD1 and XD2 have been introduced in sensitivity analysis. The dynamic responses of CS2 to 20% step changes in fresh feed flow rate are shown in Fig. 22 and the dynamic responses of CS2 to 10% step
changes in composition are shown in Fig. 23. Fig. 24. shows the dynamic results of D1, D2, XD1 and XD2 when the flow rate changes. It is noticed that, the dynamic performance of CS2 is much better than CS1. The two product purities are maintained at high purity with CS2 giving smaller transient deviations as well as shorter settling time. The transient deviation of XD1 and XD2 reaches above 0.996. But, CS1 have the large transient when feed flow rate disturbances occur. CS1 does not work well in treating the feed composition disturbance because XD1 and XD2 a bit lower than the set value in a new steady state. However, CS2 can solve the problem. The purities of two products are brought back to the set points, and the deviation is quite small as well as CS2. The mole fraction of OX also has small deviation. The transient deviation
62
X. Dai et al. / Chemical Engineering and Processing 100 (2016) 49–64
Fig. 23. Dynamic responses (CS2) to composition disturbances.
reaches about 0.10. The deviation of mole fraction of OX can go back to the set point and need less time to reach a new steady state than that of CS1. 4. Conclusions The BEN–CYH system can be separated by extractive distillation with SULF as entrainer. However, the operating temperature of the ERC is too low to allow water to be the cooling medium. As a result, a process improvement is made in this study so that cooling water can be utilized as the cooling medium and thermal decomposition of the entrainer can be avoided. For this purpose, OX is added into the entrainer for the mixed entrainer which can moderate the base temperature of the ERC. Then, in order to save energy, EDWC is
applied to the BEN–CYH system. Moreover, the process of EDWC with mixed entrainer simulated to separate BEN-CYH is investigated in this paper. EDWC simulated as a three-column model is used in this paper so that much more energy can be saved. The minimal TAC of EDWC is established. The result shows that EDWC can save 44.0% of energy and 35.8% of TAC, compared with the conventional extractive distillation arrangement with pure entrainer. EDWC is more difficult to control than conventional columns when feed disturbances are introduced. And it is even harder to control EDWC when the OX is added into the process. Two control structures are illustrated in EDWC. Both of two control structures can well solve the feed flow rate separately, except that the basic structure would have transient deviations of XD1 and XD2 under
X. Dai et al. / Chemical Engineering and Processing 100 (2016) 49–64
63
Fig. 24. Dynamic responses (CS1, CS2) of purities to flow rate disturbances.
the feed flow rate disturbance of +20%. However, the improved control structure can avoid the large transient deviations of XD1 and XD2. Conflict of interest The authors declare no conflict of interest. Acknowledgment We are thankful for assistance from the staff at the School of Petrochemical Technology (Changzhou University). References [1] B. Schuur, B.J.V. Verkuijl, A.J. Minnaard, et al., Chiral separation by enantioselective liquid–liquid extraction, Org. Biomol. Chem. 9 (2011) 36–51. [2] I. Billard, A. Ouadi, C. Gaillard, Liquid–liquid extraction of actinides, lanthanides, and fission products by use of ionic liquids: from discovery to understanding, Anal. Bioanal. Chem. 400 (2011) 1555–1566. [3] J. Qin, Q. Ye, X.J. Xiong, Control of benzene–cyclohexane separation system via extractive distillation using sulfolane as entrainer, Ind. Eng. Chem. Res. 52 (2013) 10754–10766. [4] L.Y. Sun, Q.Y. Wang, L.M. Li, et al., Design and control of extractive dividing wall column for separating benzene/cyclohexane mixtures, Ind. Eng. Chem. Res. 53 (2014) 8120–8131. [5] W.Z. An, X.G. Yuan, A simulated annealing-based approach to the optimal synthesis of heat-integrated distillation sequences, Comput. Chem. Eng. 33 (2009) 199–212. [6] M. Errico, B.G. Rong, G. Tola, M. Spano, Optimal synthesis of distillation systems for bioethanol separation. Part 2. Extractive distillation with complex columns, Ind. Eng. Chem. Res. 52 (2013) 1620–1626. [7] Bärbel Kolbe, Sascha Wenzel, Novel distillation concepts using one-shell columns, Chem. Eng. Process. 43 (2004) 339–346.
[8] A.A. Kiss, Advanced Distillation Technologies: Design, Control and Applications, John Wiley & Sons, Ltd., Chichester, 2013. [9] F.I. Gómez-Castro, J.G. Segovia-Hernández, S. Hernandez, et al., Dividing wall distillation columns: optimization and control properties, Chem. Eng. Technol. 31 (2008) 1246–1260. [10] M. Emtir, Z. Fonyo, Rigorous simulation of energy integrated and thermally coupled distillation schemes for ternary mixture, Appl. Therm. Eng. 21 (2001) 1299–1317. [11] A.A. Kiss, D.J. Suszwalak, Enhanced bioethanol dehydration by extractive and azeotropic distillation in dividing-wall columns, Sep. Purif. Technol. 86 (2012) 70–78. [12] A.A. Kiss, R.M. Ignrat, Innovative single step bioethanol dehydration in an extractive dividing-wall column, Sep. Purif. Technol. 98 (2012) 290–297. [13] J.J. Ibarra-Sánchez, J.G. Segovia-Hernández, Reducing energy consumption and CO2 emissions in extractive distillation, part II: dynamic behavior, Chem. Eng. Res. Des. 88 (2010) 135–145. [14] H. Zhang, Q. Ye, J.W. Qin, et al., Design and control of extractive dividing-wall column for separating ethyl acetate–isopropyl alcohol mixture, Ind. Eng. Chem. Res. 53 (2013) 1189–1205. [15] M. Xia, Y.P. Xin, J.W. Luo, et al., Temperature control for extractive dividing-wall column with an adjustable vapor split: methylal/methanol azeotrope separation, Ind. Eng. Chem. Res. 52 (2013) 17996–18013. [16] B. Van Dyk, I. Nieuwoudt, Design of solvents for extractive distillation, Ind. Eng. Chem. Res. 39 (2000) 1423–1429. [17] W.L. Luyben, Comparison of extractive distillation and pressure-swing distillation for acetone–mehanol separation, Ind. Eng. Chem. Res. 47 (2008) 2696–2707. [18] W.L. Luyben, Design and control of the butyl acetate process, Ind. Eng. Chem. Res. 50 (2011) 1247–1263. [19] J.M. Douglas, Conceptual Design of Chemical Processes, McGraw Hill, New York, 1998, pp. 568–577. [20] W.L. Luyben, Design and control of the ethyl bnzene process, AIChE J. 57 (2011) 655–670. [21] W.L. Luyben, Distillation Design and Control Using Aspen Simulation, John Wiley & Sons, 2013. [22] Y.C. Wu, P.H.C. Hsu, I.L. Chien, Critical assessment of the energy-saving potential of an extractive dividing-wall column, Ind. Eng. Chem. Res. 52 (2013) 5384–5399.
64
X. Dai et al. / Chemical Engineering and Processing 100 (2016) 49–64
[23] M.I. Mutalib, R. Smith, Operation and control of dividing wall distillation columns: part 1: degrees of freedom and dynamic simulation, Chem. Eng. Res. Des. 76 (1998) 308–318. [24] W.L. Luyben, Control of the maximum-boiling acetone/chloroform azeotropic distillatio system, Ind. Eng. Chem. Res. 47 (2008) 6140–6149. [25] Y.C.I.L. Wu Chien, Design and control of heterogeneous azeotropic column system for the separation of pyridine and water, Ind. Eng. Chem. Res. 48 (2009) 10564–10576. [26] W.L. Luyben, Economic optimum design of the heterogeneous azeotropic dehydration of ethanol, Ind. Eng. Chem. Res. 51 (2012) 16427–16432.
[27] P. Jordi, F.L. Estela, L. Sonia, et al., Thermodynamic analysis and process simulation of ethanol dehydration via heterogeneous azeotropic distillation, Ind. Eng. Chem. Res. 53 (2014) 6084–6093. [28] Z.G. Lei, C.Y. Li, B.H. Chen, Extractive distillation: a review, Sep. Purif. Technol. Rev. 32 (2003) 121–213. [29] M. Va’zquez-Ojeda, J.G. Segovia-Herna’ndez, S. Herna’ndez, et al., Design and optimization of an ethanol dehydration process using stochastic methods, Sep. Purif. Technol. 105 (2013) 90–97. [30] W.L. Luyben, I.L. Chien, Design and Control of Distillation Systems for Separating Azeotropes, John Wiley & Sons, Inc., New York, 2010.
Contents lists available at ScienceDirect
Chemical Engineering and Processing: Process Intensification journal homepage: www.elsevier.com/locate/cep
Energy-saving dividing-wall column design and control for benzene extraction distillation via mixed entrainer Xin Dai, Qing Ye* , Jiwei Qin, Hao Yu, Xiaomeng Suo, Rui Li Jiangsu Key Laboratory of Advanced Catalytic Materials and Technology, School of Petrochemical Engineering, Changzhou University, Changzhou, Jiangsu 213164, China
A R T I C L E I N F O
A B S T R A C T
Article history: Received 30 August 2015 Received in revised form 5 November 2015 Accepted 29 November 2015 Available online 2 December 2015
Benzene and cyclohexane form an azeotrope (Benzene, 49.7 w.t%, 1 atm), which can be separated by extractive distillation. In this study, o-xylene is added into the sulfolane entrainer for mixed entrainer to make the operation conditions more convenient and controllable. The TAC (total annual cost, generally utilized in global economic optimization) of the process with o-xylene as mixed entrainer is less than that of the process with SULF as entrainer. In addition, as the most energy saving process, the process of extractive dividing wall column (EDWC) is proposed, for it can save 44.0% of energy and 35.8% of TAC, compared with the conventional extractive distillation arrangement with pure entrainer. Besides, two control structures are established for EDWC, and their dynamic performances are evaluated by feed flow rate disturbances and feed composition disturbances. Finally, it is found that the disturbances are well handled with the second control structure. ã 2015 Elsevier B.V. All rights reserved.
Keywords: Benzene Cyclohexane O-xylene Extractive dividing wall column Energy saving
1. Introduction Benzene (BEN) is an important basic chemical materials, which can be used to synthesize a variety of fine organic chemical intermediates. The market has great demand in high purity BEN, so it has great social effects to produce BEN products to suit market demand. In the production of catalytic reforming or pyrolysis hydrogenation, there is usually existing a mixture of BEN and cyclohexane (CYH). BEN and CYH can form azeotrope, which makes their separation via traditional distillation column unrealized. Therefore, an effective alternative separation technique is highly required in industry. There are two common methods to separate this system in industry. One is liquid–liquid extraction. Liquid–liquid extraction is the unit operation which uses different solubility of components in solvents to separate the mixture. It is widely used in chemical, metallurgy, food industry and petroleum refining industry. In the aromatics extraction process, the ideal solvent is sulfolane (SULF). Because it can obtain higher aromatic hydrocarbon recovery rate and can be dissolved into two phase separation and dissolution agent. Schuur et al. studied different entrainer in liquid–liquid extraction, they considered crown ether based extractants, metal complexes, metalloids and so on [1]. Billard et al. discussed
* Corresponding author. Fax: +86 519 86330355. E-mail address: [email protected] (Q. Ye). http://dx.doi.org/10.1016/j.cep.2015.11.014 0255-2701/ ã 2015 Elsevier B.V. All rights reserved.
actinides and lanthanides by using ionic liquids in liquid–liquid extraction [2]. The other method is aromatics (BEN) extraction distillation which is an extractive distillation process. Extractive distillation is a special distillation method, which continuous adding entrainer into the distillation column to change the relative volatility of the separation components. Many different entrainers can be used in extractive distillation. SUFL can also be used in extractive distillation. Qin et al. discussed the use of extractive distillation for the separation of BEN-CYH via pure entrainer [3]. SUFL was used as entrainer and two control structures were proposed. The SUFL would be thermally decomposed when the base temperature was over 200 C. The pressure of the condenser in the entrainer recovery column (ERC) was 0.08 atm and the top temperature of ERC was 15.7 C. Because the temperature was too low to use the convenient water as condensing intermedium, the operating cost would increase. Sun et al. studied separation of BEN–CYH system by extractive dividing wall column with furfural as entrainer [4]. However, because the boiling point of furfural was too low, there were some problems after treatment. Due to the continuous improvement of limited energy resources, the process has been strengthened, which has caused wide attention in the world [5–7]. For further energy savings, the new configuration with integrated two columns into one shell, which is identified as a dividing wall distillation column (DWC) [8]. Gómez-Castro et al. found that 33% reduction of energy had been saved in DWC when they studied six cases of different feed
50
X. Dai et al. / Chemical Engineering and Processing 100 (2016) 49–64
Nomenclature
EDWC DWC BEN CYH SULF OX F R/F EF EDC ERC EDC-C EDC-B RC F2 RR EF/F NF NEF NTEDC-C NTEDC-B NTRC XD1 XD2 HX KU PU Kc
tI
QR TC TP QR/F
Extractive dividing wall column Dividing wall distillation column Benzene Cyclohexane Sulfolane O-xylene; The fresh feed of the extractive distillation column; The mass ratio of reflux flow rate/flow rate The flow rate of entrainer The extractive distillation column The entrainer recovery column The extractive distillation column of CYH The BEN recovery column The entrainer recovery column The feed of ERC The reflux ratio The molar ratio of mixture entrainer to feed The stage of fresh feed The mixture entrainer feed stage The number of total stages of EDC-C The number of total stages of EDC-B The number of total stages of RC The purity of top product in EDC-C The purity of top product in EDC-B Heat exchangers Ultimate gains Ultimate periods Gain Integral time Reboiler duty Temperature controller The top pressure of column Reboiler duty/mole flow rate of F
components of DWC [9]. Emtir et al. found that there was 30% reduction of the total annual cost of DWC when they studied a ternary mixture system with three different feed compositions [10]. Compared with the conventional structures, DWCs can reduce the cost of capital and energy [11–14]. Kiss and Suszwalak [11] studied an extractive dividing wall column (EDWC) for the purpose of bioethanol dehydration, and the result showed that 9.4% reboiler duty can be saved as compared with conventional extractive distillation process. Zhang et al. studied EDWC simulated as three columns model. They found that EDWC can save more energy than conventional extractive distillation arrangement [14]. Another important aspect of EDWC should be considered is dynamic control. EDWC have interactions and inner structures among control loops which make it much more difficult to control than that of conventional extractive distillation arrangement [13]. Sun et al. used the composition control cascaded with the molar ratio of mixture entrainer to feed and the composition control cascaded with temperature control in the EDWC [4]. Zhang et al. studied the separation of ethyl acetate-isopropyl alcohol system by extractive dividing wall column and the temperature control was cooperated with the composition control [14]. Xia et al. studied separation of methylal–methanol system by extractive dividing wall column and two control schemes were all evaluated [15]. In this study, BEN and CYH azeotrope is separated through extractive distillation. The entrainer is the mixture of SULF and oxylene (OX). Therefore, the temperature of bottom is lower than the decomposition temperature of SULF, and water can be utilized
as the cooling medium at the top of the column. In order to save energy, EDWC simulated as a three-column model is studied in this paper. In terms of simulation studies, three-column model is more close to EDWC concept compared with the models simulated as two columns. Total annual cost (TAC) is calculated to obtain the optimal conventional extractive distillation process and EDWC with mixed entrainer. The separation of BEN-CYH using mixed entrainer, which is SULF mixed with OX via EDWC is studied for the first time. Subsequently, two control structures are proposed. There has been no published literature about the detailed control of EDWC with mixed entrainer simulated as a three-column model. 2. Optimization of extractive distillation process 2.1. Thermodynamic model and mixed entrainer screening and selection In this work, NRTL model is used in the Aspen simulations. BEN and CYH have very similar boiling points (80.13 C and 80.78 C), and the BEN/CYH mixture has an azeotrope with composition of 55.01 mol% BEN at atmospheric pressure. In the earlier work [3], BEN–CYH system was separated by extractive distillation using SULF as entrainer. To avoid thermally decomposition, the entrainer recovery column (ERC) was operated at 0.08 atm. However, the top temperature of ERC was 15.7 C, which was too low to use water as cooling medium. To solve this problem, another lower boiling point component is added into the entrainer so that the temperature in bottom of column can be lower than that of thermally decomposition (200 C) and meanwhile the top temperature in column is high enough to use water as cooling medium. O-xylene (OX) has no selectivity for the separation of BEN–CYH system and its boiling point (144.4 C) is between BEN (80.13 C) and SULF (287.3 C). OX had no selectivity for separation of aromatics but can help increase mutual solubility of SUFL and BEN. So, it is used as a cosolvent of SULF in the process and the compatibility is commendable. The concentration of OX in mixed entrainer is important to the effect of separation. Usually the top pressure of the EDWC is designed to be able to use water as the cooling medium. Comparison of the relative volatilities in the presence of different entrainers is a criterion for the entrainer selection [16]. The relative volatility is more higher, the separation process is more easier. Fig. 1 shows the pseudo-binary VLE of BEN/ CYH mixture with SULF and OX to feed mole ratio 1.1 by NRTL
Fig. 1. Pseudo-binary x-y plot for the BEN–CYH system with mixed entrainer to feed molar ratio 1.1.
X. Dai et al. / Chemical Engineering and Processing 100 (2016) 49–64
51
2.2. The steady-state design of conventional extractive distillation process with mixed entrainer
Fig. 2. The effect of mole fraction of OX on temperature of bottom in EDWC.
model. The less the mole fraction of OX use in mixed entrainer, the larger relative volatility gets. Fig. 2 shows the effect of mole fraction of OX on temperature of bottom in EDWC when the top pressure of column is 0.6 atm. When the mole fraction of OX is more than 10%, the base temperature of the column is lower than that of the thermally decomposition (200 C). The top temperature of EDWC is 64.4 C at 0.6 atm so that water can be used as cooling medium. So, the 10% mole fraction of OX is the best choice.
In this section, conventional extractive distillation process with mixed entrainer is proposed. The process is shown in Fig. 3. The entrainer and the mixture of BEN–CYH feed in the extractive distillation column (EDC) from the upper tray and the lower tray, respectively. The CYH as the top product exits at the top of the EDC, while the mixture entrainer and the BEN exit at the bottom of the EDC. Then the BEN and the mixture entrainer are separated in the entrainer recovery column (ERC). BEN of the lower boiling point is distillated from the top of the ERC, and the mixture entrainer is produced from the base of the ERC. The mixture entrainer product is recycled to the EDC after being cooled at the entrainer cooler. To compensate the loss of the mixture entrainer, a small quantity of fresh mixture entrainer is made up to the entrainer feed. In previous work [3], the steady state design had done some optimization on the operating conditions. To compare with the earlier work [4], the flow sheet in this work is similar with it. The mixture feed flowrate of BEN-CYH is 100 kmol/h and the mole fraction of BEN is 0.5. The optimal mole fraction of OX is 0.1 in the mixed entrainer. The mole fraction of products are both specified at 0.998 (kmol/kmol). The RadFrac model in Aspen plus is used to simulate the steady state of extractive distillation system. Through the economic optimization, the optimal design parameters can be obtained. Furthermore, the minimal total annual cost can be obtained. The global optimization of this process is shown in Fig. 4. The final steady-state flowsheet and the detail information of the flowsheet are illustrated in Fig. 3. The temperature profiles of conventional extractive distillation process are shown in Figs. 5 and 6, respectively.
Fig. 3. The optimal steady state of the two-column flowsheet.
52
X. Dai et al. / Chemical Engineering and Processing 100 (2016) 49–64
Fig. 4. The optimization procedure of the extractive distillation process.
2.3. The steady-state design of extractive dividing wall (edwc) column process with mixed entrianer In this part, DWC will be designed with the same feed conditions and product specifications. For saving the investment cost and operating cost, a dividing-wall is added into the process which is based on the conventional extractive distillation process. In this work, the EDWC concept is applied to the same system with OX as the mixed entrainer. Fig. 7 shows the configuration of an EDWC. EDWC consists of three parts. The left side of the dividingwall is included in part 1. On the right side of the dividing-wall is part 2, while part 3 is under part 1 and part 2. The fresh feed and mixed entrainer are fed into part 1. CYH and BEN are distillate stream in part 1 and part 2, respectively. The mixed entrainer is get
from the bottom and recycled back to part 1. Only one reboiler is used in this model. The vapor from the bottom is split by the wall into part1 and part 2. It is depicted in Fig. 7 that vapor split ratio (av) is defined as V1/V3. Because the top pressure of column is 0.6 atm and the drop of tray pressure is 0.0068 atm in the three columns, the compressor in the model is just an illusionary equipment to improve the pressure of vapor coming from the top of column RC so that the vapor can go back to EDC-B and EDC-C. In this paper, a three-column model is selected to simulate EDWC displayed in Fig. 8. In terms of simulation studies, threecolumn model is more close to EDWC concept compared with the models simulated as two columns. In this model, av can be set directly. Then, the model is closer to EDWC structure. This model has three columns: the extractive distillation column of CYH (EDC-
X. Dai et al. / Chemical Engineering and Processing 100 (2016) 49–64
53
Fig. 5. The temperature profile of EDC.
Fig. 8. Process flow diagram of EDWC.
both are stage 1. The total stages of the entrainer recovery column (RC) is 6, the reboiler is stage 6. 2.4. Global optimization of EDWC
Fig. 6. The temperature profile of ERC.
C), the recovery column of BEN (EDC-B) and the recovery column (RC). It is assumed that EDC-C and EDC-B have a total of 36 stages and 25 stages, respectively. The condenser of EDC-C and EDC-B
On the base of the Luyben’s papers [17,18], the total annual cost (TAC) includes the cost of main devices and the cost of the energy consumption. The main devices include the column vessels, the condensers, the reboilers and the entrainer cooler. The other accessorial small equipments (the valves and the pumps) can be neglected. The related calculation data and formulas of the TAC are based on William Luyben (2008). The total capital cost (TAC) including the capital and operating cost, it has the following statement: TAC ¼
Fig. 7. Configuration of main column.
total capital cost þ total operating cost pay back period
Fig. 9. The calculation of EDWC diameter.
ð1Þ
54
X. Dai et al. / Chemical Engineering and Processing 100 (2016) 49–64
It is assumed to be a three-year payback period. Douglas [19] and Luyben [20] suggested that a lot of parameters of EDWC can be optimized. 0.568 and 0.852 (kW/K m2) [21–23] are heat transfer coefficients of reboiler and condenser, respectively. EDWC configuration has only one shell and the equivalent diameter [23] of EDWC is calculated as illuminated in Fig. 9. The formula of equivalent diameter is as follow: SðEDC C Þ ¼
p DðEDC C Þ2 4
SðEDC BÞ ¼
p DðEDC BÞ2 4
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi DðEDWCÞ ¼ 2 S½ðEDC C Þ þ SðEDC BÞ=p qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ DðEDC C Þ2 þ DðEDC BÞ2
ð2Þ be
ð3Þ
ð4Þ
The total stages of EDWC include those of EDC-C and RC. It can observed that the relatively higher cost related to
Fig. 10. The process of the TAC calculation.
X. Dai et al. / Chemical Engineering and Processing 100 (2016) 49–64
Fig. 11. Effects of RR1 and EF (koml/h) in the EDWC on (a) the CYH composition (XD1) in the EDC-C and (b) XD1 in the EDC-B.
Fig. 12. The effects of entrainer flow rate (EF) and NTEDC-C on TAC.
Fig. 14. Effects of NTEDC-B on the TAC of EDWC.
Fig. 13. The effects of the feed stage (NF) and the entrainer feed stage (NEF) on the TAC of EDWC.
Fig. 15. Effects of NTRC on the TAC of EDWC.
55
56
X. Dai et al. / Chemical Engineering and Processing 100 (2016) 49–64
Fig. 16. The optimized conditions of EDWC.
manufacturing EDWC should be taken into consideration in the economic analysis. It has been assumed that the capital cost of EDWC is 1.2 times more than that of conventional distillation column. Zhang et al. has done some research of the TAC of EDWC simulated as three columns [14]. According to their previous work, sequence for EDWC simulated as three columns is indicated as Fig. 10. Based on the product specifications, EDWC process gets the optimal TAC by adjusting the following parameters: the fresh feed stage (NF), the entrainer feed stage (NEF), the stages of EDC-C (NTEDC-C), the stages of EDC-B (NTEDC-B) the stages of RC (NRC), and the av. To clearly, the sequence of calculating optimal TAC is shown in Fig. 10. The optimization procedure of EDWC is as follow: (1) Give the top pressure of EDC-C and EDC-B (PEDC-C = 0.6 atm, PEDC-B = 0.6 atm); (2) Give the total stages of RC (NTRC); (3) Give the total stages of EDC-C (NTEDC-C); (4) Give the total stages of EDC-B (NTEDC-B); (5) Obtain the flow rate of EF by the steady-state simulation in Aspen plus, and select the operating EF; (6) Give the feed stage (NF) and the entrainer feed stage (NEF) of the EDWC;
Fig. 17. Stage temperature of EDWC as one column shell.
X. Dai et al. / Chemical Engineering and Processing 100 (2016) 49–64
57
Table 1 The optimization of the TAC of two columns with/without mixed entrainer and EDWC. Items
N NF NEF EF/F Column diameter (m) The condencer duty QC (kW) The reboiler duty QR (kW) Capital investment ($1000) Operation cost ($1000) TAC ($1000/yr)
The conventional extractive distillation with pure entrainer
The conventional extractive distillation with mixed entrainer
The process of EDWC with mixed entrainer
EDC
ERC
EDC
ERC
EDC-C
29 20 4 1.1 0.98 922.00 304.360 590.915 583.395 780.367
10 5 10 1.1 1.460 714.00 1091.000 722.346 387.852 628.634
42 28 10 1.1 1.003 944.70 413.183 487.879 338.637 501.269
28 14
42 25
0.808 607.24 945.115
1.467 661.44 1084.328
EDC-B
422.89
Fig. 18. The basic control structure (CS1).
(7) Adjust the reflux ratios (RR) of EDC-C to make the two products satisfy the purities requirement; (8) Go back to the step 6 and give new conditions until TAC of EDWC is the minimal; (9) Go back to the step 5 and give new conditions until TAC of EDWC is the minimal; (10) Go back to the step 4 and select new NTEDC-B until TAC of EDWC is the minimal; (11) Go back to the step 3 and select new NTEDC-C until TAC of EDWC is the minimal; (12) Go back to the step 2 and give new NTRC until TAC of EDWC is the minimal.
The effects of changing the entrainer flow rate (EF) and RR1 on the concentrations of components in distillate stream (D) from the EDC-C and EDC-B are studied. The results have been plotted in Fig. 11, in which the results are obtained when CYH product purity is satisfied with number of total stages of NEDC-C fixed at 34. The Table 2 Parameters of all temperature controllers. Parameters
TC
TC-HX
Ultimate gain Ultimate period (min) Gain (KC) Integral time (min)
4.9341 3.6 1.5419 7.92
1.2944 1.8 0.4045 3.96
58
X. Dai et al. / Chemical Engineering and Processing 100 (2016) 49–64
Fig. 19. Dynamic responses (CS1) to 20% feed flow rate disturbances.
plots in Fig. 11 show that, for a given EF, the product CYH purity reaches a maximum value with an optimum RR1. Fig. 12 shows the effect of entrainer flow rate (EF) and NTEDC-C on TAC. When NTEDC-C is 34 and EF is 110 kmol/h, TAC is minimal. The effects of the feed stage (NF) and the entrainer feed stage (NEF) on the TAC of EDWC are shown in Fig. 13, which reveals that minimum TAC exits when NF and NEF are recommended as 25 and 10, respectively. Figs. 14 and 15 show the effect of stages of NTEDC-B and NTRC on the TAC of EDWC, it is suggested that the minimal TAC existing when NTERC is 25 and NTRC is 6. Fig. 16 displays the optimized
conditions of EDWC. The temperature profiles of EDWC are shown in Fig. 17. Table 1 summarizes the results that the optimization of the TAC with/without mixed entrainer in conventional extractive distillation process and EDWC with mixed entrainer, the data of the process with pure entrainer comes from earlier work [3]. The reboiler of EDC is heated by low pressure (LP) steam of 6 bar and the reboiler of ERC is heated by high pressure (HP) steam of 42 bar and the condenser of ERC is cooled by refrigerant at 20 C in the earlier flowsheet of two columns. The conventional extractive distillation process with mixed entrainer can save 33.5% operating
X. Dai et al. / Chemical Engineering and Processing 100 (2016) 49–64
cost than that of the conventional extractive distillation without mixed entrainer. The operating cost of EDWC process can also be reduced by 44.0% that of the conventional extractive distillation with pure entrainer. The TAC of EDWC can save 35.8% than that of the conventional extractive distillation with pure entrainer. 3. Control strategy for the EDWC The EDWC control was studied in many papers, and most of them studied the special separation system and proposed the control structures [24–29]. The control of EDWC is more difficult than that of conventional extractive distillation arrangement. Moreover, the control of EDWC with mixed entrainer is much more difficult than that of EDWC. There is no published literature about the detailed control of EDWC with mixed entrainer simulated as
59
three columns model. Two control structures are proposed in this part. The sizes of EDWC are necessary to convert the steady-state simulation into a dynamic simulation. The Tray Sizing in Aspen plus is used to calculate the diameter of the columns. Reflux drums and the column bases are sized to provide 10 min of holdup when full. Dynamic process is completely pressure-driven when valves and pumps are inserted to the dynamic simulation. Then, it is ready to export the steady-state simulation to Aspen Dynamic. 3.1. Selecting temperature control trays Firstly, the tray temperature control is considered for the EDWC. Temperature control stages selection is important to temperature control structure. There are a lot of methods to select temperature control trays, such as the singular value decomposition (SVD)
Fig. 20. Dynamic responses of CS1 to composition disturbance.
60
X. Dai et al. / Chemical Engineering and Processing 100 (2016) 49–64
Fig. 21. The control structure of CS2.
method, the slope rule and sensitivity criterion. The slope rule is the easiest method in these three ways. The slope rule is chosen to select temperature control stages for EDWC in this work. As shown in Fig. 17, the 41th of EDWC is chosen as temperature control stage. Due to the addition of mixed entrainer, the temperature changing are not obvious in EDC-C and EDC-B, the sensitive stage is not selected. So, the reflux ratio control is used in EDC-C and EDC-B. 3.2. Control structures design 3.2.1. Basic control structure 1 (CS1) The basic control structure established for the three column system and the controller face plate are present in Fig. 18. All control loops of the basic control structure are listed as follows. (1) F is flow controlled (reserve acting). (2) The levels of the two reflux drums are controlled by manipulating the distillates, respectively (direct acting). (3) The column base levels of the EDC-C and EDC-B are controlled by manipulating the bottoms (direct acting) while the base level of RC is controlled by manipulating the flowrate of the makeup entrainer (reverse acting). (4) Top pressures of EDC-C and EDC-B are controlled by manipulating the heat removal duty of the two condensers (direct acting), respectively. (5) The mixed entrainer feed is flow controlled (reverse acting), and it is ratioed to F by cascade control. (6) The reflux flow rate of EDC-C and EDC-B are ratioed to F, respectively. (7) The 5th stage temperature of RC is controlled by manipulating reboiler duty (reverse acting). (8) The recycle entrainer temperature is maintained by manipulating the HX heat removal duty (reserve acting).
(9) Dead time of 1 min is inserted to all temperature control loops to match time drag in real plant environment. For all the flows, pressures and temperatures, conventional proportional and integral (PI) controllers are used. All level loops are P-only controllers with gain (Kc) = 2. For the pressure PI controllers, tuning parameters of Kc = 20 and integral time (t I) = 12 min are used, while Kc = 0.5 and t I = 0.3 min are used for the flow PI controllers. Considering the practical operation, a 1-min deadtime is inserted into each temperature control loop. Relayfeedback test and Tyreus-Luben turning are utilized to get the ultimate gains (KU) and ultimate periods (PU) of controllers for all of the control loops with dead time [30]. Parameters of all temperature controllers are presented in Table 2. To test the effectiveness of the CS1, large disturbances in the fresh feed flow rate and feed composition are made. Fig. 19 illustrates the dynamic responses of CS1 to 20% step changes in fresh feed flow rate at 0.5 h. Flow rate of CYH (D1), BEN (D2) are changed 20% correspondingly as well. However, the purities of CYH (XD1) and BEN (XD2) have large transient deviations. The transient deviation of XD1 and XD2 reaches about 0.935. The mole fraction of OX has large transient deviations when the feed flowrate undergoing 20% disturbances after 0.5 h. The transient deviation reaches about 0.035. Fig. 20 shows the results of basic control structure when undergoing the feed composition disturbances. The composition disturbances are also introduced at 0.5 h. As is shown in Fig. 20, large deviations occur in the purities of the two productions and the purities are all about 0.90 when the process reaches to a new steady state. Moreover, the mole fraction of OX has large deviation and need long time to reach to a new steady state.
X. Dai et al. / Chemical Engineering and Processing 100 (2016) 49–64
61
Fig. 22. Dynamic response (CS2) to flow rate disturbances.
3.2.2. An improved control structure (CS2) Because the OX is added into the process, the temperature of EDC-C is changed smoothly. The result of the temperature control in the EDWC with mixed entrainer is not very well. So, CS1 is not controlled very well. In order to reduce the large transient deviations, the basic control structure must be improved. The composition controller of BEN(CC1) is cascaded with D1/F ratio. The OX of bottom stream (B) composition controller (CC2) cascaded with TC. The corresponding controller faceplates of CS2 are shown in Fig. 21. The effects of entrainer flow rate on XD1 and XD2 have been introduced in sensitivity analysis. The dynamic responses of CS2 to 20% step changes in fresh feed flow rate are shown in Fig. 22 and the dynamic responses of CS2 to 10% step
changes in composition are shown in Fig. 23. Fig. 24. shows the dynamic results of D1, D2, XD1 and XD2 when the flow rate changes. It is noticed that, the dynamic performance of CS2 is much better than CS1. The two product purities are maintained at high purity with CS2 giving smaller transient deviations as well as shorter settling time. The transient deviation of XD1 and XD2 reaches above 0.996. But, CS1 have the large transient when feed flow rate disturbances occur. CS1 does not work well in treating the feed composition disturbance because XD1 and XD2 a bit lower than the set value in a new steady state. However, CS2 can solve the problem. The purities of two products are brought back to the set points, and the deviation is quite small as well as CS2. The mole fraction of OX also has small deviation. The transient deviation
62
X. Dai et al. / Chemical Engineering and Processing 100 (2016) 49–64
Fig. 23. Dynamic responses (CS2) to composition disturbances.
reaches about 0.10. The deviation of mole fraction of OX can go back to the set point and need less time to reach a new steady state than that of CS1. 4. Conclusions The BEN–CYH system can be separated by extractive distillation with SULF as entrainer. However, the operating temperature of the ERC is too low to allow water to be the cooling medium. As a result, a process improvement is made in this study so that cooling water can be utilized as the cooling medium and thermal decomposition of the entrainer can be avoided. For this purpose, OX is added into the entrainer for the mixed entrainer which can moderate the base temperature of the ERC. Then, in order to save energy, EDWC is
applied to the BEN–CYH system. Moreover, the process of EDWC with mixed entrainer simulated to separate BEN-CYH is investigated in this paper. EDWC simulated as a three-column model is used in this paper so that much more energy can be saved. The minimal TAC of EDWC is established. The result shows that EDWC can save 44.0% of energy and 35.8% of TAC, compared with the conventional extractive distillation arrangement with pure entrainer. EDWC is more difficult to control than conventional columns when feed disturbances are introduced. And it is even harder to control EDWC when the OX is added into the process. Two control structures are illustrated in EDWC. Both of two control structures can well solve the feed flow rate separately, except that the basic structure would have transient deviations of XD1 and XD2 under
X. Dai et al. / Chemical Engineering and Processing 100 (2016) 49–64
63
Fig. 24. Dynamic responses (CS1, CS2) of purities to flow rate disturbances.
the feed flow rate disturbance of +20%. However, the improved control structure can avoid the large transient deviations of XD1 and XD2. Conflict of interest The authors declare no conflict of interest. Acknowledgment We are thankful for assistance from the staff at the School of Petrochemical Technology (Changzhou University). References [1] B. Schuur, B.J.V. Verkuijl, A.J. Minnaard, et al., Chiral separation by enantioselective liquid–liquid extraction, Org. Biomol. Chem. 9 (2011) 36–51. [2] I. Billard, A. Ouadi, C. Gaillard, Liquid–liquid extraction of actinides, lanthanides, and fission products by use of ionic liquids: from discovery to understanding, Anal. Bioanal. Chem. 400 (2011) 1555–1566. [3] J. Qin, Q. Ye, X.J. Xiong, Control of benzene–cyclohexane separation system via extractive distillation using sulfolane as entrainer, Ind. Eng. Chem. Res. 52 (2013) 10754–10766. [4] L.Y. Sun, Q.Y. Wang, L.M. Li, et al., Design and control of extractive dividing wall column for separating benzene/cyclohexane mixtures, Ind. Eng. Chem. Res. 53 (2014) 8120–8131. [5] W.Z. An, X.G. Yuan, A simulated annealing-based approach to the optimal synthesis of heat-integrated distillation sequences, Comput. Chem. Eng. 33 (2009) 199–212. [6] M. Errico, B.G. Rong, G. Tola, M. Spano, Optimal synthesis of distillation systems for bioethanol separation. Part 2. Extractive distillation with complex columns, Ind. Eng. Chem. Res. 52 (2013) 1620–1626. [7] Bärbel Kolbe, Sascha Wenzel, Novel distillation concepts using one-shell columns, Chem. Eng. Process. 43 (2004) 339–346.
[8] A.A. Kiss, Advanced Distillation Technologies: Design, Control and Applications, John Wiley & Sons, Ltd., Chichester, 2013. [9] F.I. Gómez-Castro, J.G. Segovia-Hernández, S. Hernandez, et al., Dividing wall distillation columns: optimization and control properties, Chem. Eng. Technol. 31 (2008) 1246–1260. [10] M. Emtir, Z. Fonyo, Rigorous simulation of energy integrated and thermally coupled distillation schemes for ternary mixture, Appl. Therm. Eng. 21 (2001) 1299–1317. [11] A.A. Kiss, D.J. Suszwalak, Enhanced bioethanol dehydration by extractive and azeotropic distillation in dividing-wall columns, Sep. Purif. Technol. 86 (2012) 70–78. [12] A.A. Kiss, R.M. Ignrat, Innovative single step bioethanol dehydration in an extractive dividing-wall column, Sep. Purif. Technol. 98 (2012) 290–297. [13] J.J. Ibarra-Sánchez, J.G. Segovia-Hernández, Reducing energy consumption and CO2 emissions in extractive distillation, part II: dynamic behavior, Chem. Eng. Res. Des. 88 (2010) 135–145. [14] H. Zhang, Q. Ye, J.W. Qin, et al., Design and control of extractive dividing-wall column for separating ethyl acetate–isopropyl alcohol mixture, Ind. Eng. Chem. Res. 53 (2013) 1189–1205. [15] M. Xia, Y.P. Xin, J.W. Luo, et al., Temperature control for extractive dividing-wall column with an adjustable vapor split: methylal/methanol azeotrope separation, Ind. Eng. Chem. Res. 52 (2013) 17996–18013. [16] B. Van Dyk, I. Nieuwoudt, Design of solvents for extractive distillation, Ind. Eng. Chem. Res. 39 (2000) 1423–1429. [17] W.L. Luyben, Comparison of extractive distillation and pressure-swing distillation for acetone–mehanol separation, Ind. Eng. Chem. Res. 47 (2008) 2696–2707. [18] W.L. Luyben, Design and control of the butyl acetate process, Ind. Eng. Chem. Res. 50 (2011) 1247–1263. [19] J.M. Douglas, Conceptual Design of Chemical Processes, McGraw Hill, New York, 1998, pp. 568–577. [20] W.L. Luyben, Design and control of the ethyl bnzene process, AIChE J. 57 (2011) 655–670. [21] W.L. Luyben, Distillation Design and Control Using Aspen Simulation, John Wiley & Sons, 2013. [22] Y.C. Wu, P.H.C. Hsu, I.L. Chien, Critical assessment of the energy-saving potential of an extractive dividing-wall column, Ind. Eng. Chem. Res. 52 (2013) 5384–5399.
64
X. Dai et al. / Chemical Engineering and Processing 100 (2016) 49–64
[23] M.I. Mutalib, R. Smith, Operation and control of dividing wall distillation columns: part 1: degrees of freedom and dynamic simulation, Chem. Eng. Res. Des. 76 (1998) 308–318. [24] W.L. Luyben, Control of the maximum-boiling acetone/chloroform azeotropic distillatio system, Ind. Eng. Chem. Res. 47 (2008) 6140–6149. [25] Y.C.I.L. Wu Chien, Design and control of heterogeneous azeotropic column system for the separation of pyridine and water, Ind. Eng. Chem. Res. 48 (2009) 10564–10576. [26] W.L. Luyben, Economic optimum design of the heterogeneous azeotropic dehydration of ethanol, Ind. Eng. Chem. Res. 51 (2012) 16427–16432.
[27] P. Jordi, F.L. Estela, L. Sonia, et al., Thermodynamic analysis and process simulation of ethanol dehydration via heterogeneous azeotropic distillation, Ind. Eng. Chem. Res. 53 (2014) 6084–6093. [28] Z.G. Lei, C.Y. Li, B.H. Chen, Extractive distillation: a review, Sep. Purif. Technol. Rev. 32 (2003) 121–213. [29] M. Va’zquez-Ojeda, J.G. Segovia-Herna’ndez, S. Herna’ndez, et al., Design and optimization of an ethanol dehydration process using stochastic methods, Sep. Purif. Technol. 105 (2013) 90–97. [30] W.L. Luyben, I.L. Chien, Design and Control of Distillation Systems for Separating Azeotropes, John Wiley & Sons, Inc., New York, 2010.