Control comparison of conventional and thermally coupled ternary extractive distillation processes

Control comparison of conventional and thermally coupled ternary extractive distillation processes

Accepted Manuscript Title: Control Comparison of Conventional and Thermally Coupled Ternary Extractive Distillation Processes Author: William L. Luybe...

510KB Sizes 0 Downloads 47 Views

Accepted Manuscript Title: Control Comparison of Conventional and Thermally Coupled Ternary Extractive Distillation Processes Author: William L. Luyben PII: DOI: Reference:

S0263-8762(15)00490-6 http://dx.doi.org/doi:10.1016/j.cherd.2015.11.021 CHERD 2104

To appear in: Received date: Revised date: Accepted date:

13-10-2015 24-11-2015 27-11-2015

Please cite this article as: Luyben, W.L.,Control Comparison of Conventional and Thermally Coupled Ternary Extractive Distillation Processes, Chemical Engineering Research and Design (2015), http://dx.doi.org/10.1016/j.cherd.2015.11.021 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

cr

ip t

Paper submitted to Chemical Engineering Research and Design

M

an

us

Control Comparison of Conventional and Thermally Coupled Ternary Extractive Distillation Processes

d

William L. Luyben

Ac ce p

te

Department of Chemical Engineering Lehigh University Bethlehem, PA 18015 USA

October 10, 2015 Revised November 17, 2015

[email protected]; 610-758-4256; FAX 610-758-5057

Page 1 of 31

Abstract Extractive distillation is widely used to separate binary mixtures with azeotropes that prevent separation in a single distillation column. The conventional binary extractive

ip t

distillation configuration uses two columns with one key component going overhead in the extractive column and the other key component going either overhead or out the

cr

bottom of the solvent recovery column, depending on the effect of the solvent on the volatilities of the key components.

us

Extractive distillation can also be applied to separate ternary mixtures with azeotropes. The conventional ternary extractive process requires three columns.

an

Unconventional processes can also be used, such as thermally coupled

sidestream/rectifier columns. Economic advantages of the thermally coupled process have been reported for some separations.

M

The controllability of extractive distillation system for binary mixtures has been explored in many papers. The novel contribution of this paper is the exploration of the

d

dynamic controllability of extractive distillation systems for a ternary mixture. The

Ac ce p

column/rectifier process.

te

dynamics of a conventional three-column process are compared with a thermally coupled

Key Words

Extractive distillation, ternary systems, controllability

1. Introduction

There is a rich literature covering the use of extractive distillation to separate

binary mixtures with azeotropes. Many papers and books1,2,3,4 have appeared that discuss the various types of configurations that can be used. The conventional configuration uses two distillation columns with a solvent fed near the top of the first column, and one of the key components taken overhead. The essentially binary mixture leaving in the bottoms is separated in the second column with the solvent recycled from either the bottom (highboiling solvent) or the top (intermediate-boiling solvent) of the column.

2

Page 2 of 31

Nonconventional configurations for binary mixtures have also been explored5,6 The E-DWC features a single divided-wall column with the wall extending from the top of the stripping section all the way to the top of the column. The bottoms stream is the high-boiling solvent, which is recycled to the near the top of the feed side of the wall.

ip t

Each side of the wall has its independent condenser and reflux. The two distillate streams are the high-purity key components of the binary feed mixture. An important application

cr

is the dehydration of ethanol using ethylene glycol solvent.

However, the single-column E-DWC cannot be applied when the initial feed is a

us

ternary mixture since three product streams must be produced. In a very comprehensive and insightful paper, Timoshenko et al7 discuss the many alternative extractive

an

distillation configurations for ternary mixtures, both conventional three-column flowsheets and partially thermally coupled flowsheets, for a variety of phase equilibrium relationships. A case study is presented for the ternary mixture of benzene (boiling point

M

78.11 oC), cyclohexane (80.72 oC) and toluene (110.6 oC) using N-methylpyrrolidone as a heavy solvent (204.3 oC). The conventional three-column flowsheet was shown to have

d

13% higher total reboiler duty than a non-conventional two-column flowsheet with a side rectifier on the extractive column. No discussion of the dynamic controllability of the

Ac ce p

te

alternative flowsheets was presented. That is the purpose of this paper.

2. Process Studied

A conventional three-column configuration (Figure 1) is compared with a two-

column configuration with a side rectifier (Figure 2). We study the same ternary mixture of key components considered by Timoshenko et al7 (benzene, cyclohexane and toluene). The minimum-boiling azeotrope in the benzene/cyclohexane system is shown in Figure 3. Timoshenko used the high-boiling solvent N-methylpyrrolidone (NMP). In our Aspen simulations using NRTL physical properties, we were unable to duplicate the Timoshenko designs since the separation in the extractive column between cyclohexane and benzene using NMP was inadequate to achieve the desired product purities even with very large solvent-to-feed ratios. Instead we switched to dimethyl formamide (DMF).

3

Page 3 of 31

As shown in Figure 4, DMF has a stronger effect on the volatility between benzene and cyclohexane. The ordinate and abscissa are the normalized mole fractions in the vapor and liquid phases excluding the solvent. The plots are generated by using a Flash2 Aspen model. The flash pressure and the feed composition are fixed. For

ip t

example, with a feed of 50 mol% DMF, 25 mol% benzene and 25 mol% cyclohexane, the normalized composition is 50 mol% benzene. The vapor fraction in the Flash2 is

cr

specified to be very small (10-5), so the liquid composition is the same as the feed. The

resulting calculated vapor composition in equilibrium with the known liquid composition

us

is then used to calculate the normalized vapor benzene composition.

Since our objective is to make a comparative study of dynamic controllability

an

between the two processes, we feel the using DMF instead of NMP should have little effect on the comparison.

Notice in Figure 4 that the normalized xy curve is below the 45-degree line. This

M

means that cyclohexane goes out the top of the extractive column even though it is higher boiling than benzene. Figure 5 gives a ternary diagram for the benzene/cyclohexane/DMF

d

system with the distillate (D1), fresh feed (F), solvent (S), mix point (M) and bottoms (B1) points are indicated (neglecting the third key component toluene). In the region

te

below the isovolatility line as we go towards more solvent, the relative volatility of

Ac ce p

benzene to cyclohexane is less than unity. So the cyclohexane goes overhead in the extractive column.

2.1 Conventional Process

Figure 1 gives the flowsheet of the conventional process studied. The fresh feed is

100 kmol/ h with composition 30 mol% benzene (B), 30 mol% cyclohexane (CH) and 40 mol% toluene (T). Fresh fed is fed on Stage 40 of an 80-stage column. The DMF solvent is fed on Stage 20 at 89.74 kmol/h with composition 0.41 mol% toluene and 99.59 mol% DMF. The operating pressure of the extractive column (C1) is set at 0.7 bar because there is an azeotrope between toluene and DMF at pressures higher than 0.7 bar that adversely affects the separation. At 1 atm this azeotrope is 1.54 mol% DMF at 110 oC. At 0.7 bar, the azeotrope disappears.

4

Page 4 of 31

It was found that many stages are needed in the extractive column to achieve the specified purity of 99 mol% cyclohexane in the distillate and to keep the cyclohexane impurity in the bottoms low enough to be able to achieve the 99 mol% benzene specification in the distillate of the next column.

ip t

The extractive column C1 has three design degrees of freedom once the pressure, total number of stages, location of the feed and location of the solvent are fixed. Three

cr

Aspen Design spec/Vary functions are used.

1. The impurity of cyclohexane in the bottoms is set at 0.09 mol% by varying the

us

distillate flowrate.

2. The impurity of DMF in the distillate is set at 0.84 mol% by varying the reflux

an

ratio.

3. The impurity of benzene in the distillate is set at 0.16 mol% by varying the solvent flowrate.

M

The required reboiler duty to achieve this performance is 1.731 MW. The second column C2 operates at 1.1 bar and has 41 stages. The two design

d

specifications are 0.5 mol% toluene in the distillate and 0.05 mol% benzene in the bottoms. These are achieved by varying the distillate flowrate and the reflux ratio. The

te

required reboiler duty to achieve this performance is 1.820 MW. The third column C3 operates at 0.12 bar because of the toluene/DMF azeotrope.

Ac ce p

At this pressure the reflux-drum temperature is 50 oF, so cooling water can be used in the condenser. The column has 21 stages. The two design specifications are 0.61 mol% DMF in the distillate and 0.41 mol% toluene in the bottoms. These are achieved by varying the distillate and the reflux ratio. The required reboiler duty to achieve this performance is 0.6785 MW.

The total reboiler duty for the three reboilers in the conventional process is 4.230

MW. As we report in the next section, the total reboiler duty for the two reboilers in the non-conventional partially thermal coupled columns/rectifier process is 3.632 MW. This represents a 14 % reduction in energy, which is identical to that reported by Timoshenko et al7.

2.2 Non-conventional Rectifier Process

5

Page 5 of 31

Figure 2 gives the flowsheet for the partially thermally coupled process with a side rectifier. Notice that a vapor stream is removed from Stage 60 of the extractive column C1 and fed to the base of the rectifier C2. The pressure in C1 is set at 0.8 bar and the pressure in the rectifier is set lower at 0.4 bar so that the flowrate of the vapor can be

ip t

controlled by a valve.

The distillate from C1 is the cyclohexane product. The distillate from the rectifier

cr

is the benzene product. The distillate from the solvent recovery column C3 is the toluene product. The bottoms from the solvent recovery column is the DMF recycled back to the

us

extractive column.

The extractive column and the rectifier are designed simultaneously. With the

an

pressures, total stages and feed locations fixed, there are five remaining variables that can be varied, which are selected to be (1) the flowrate of distillate from the extractive column, (2) the flowrate of distillate from the rectifier, (3) the flowrate of the vapor

M

sidestream fed to the rectifier, (4) the reflux ratio in the extractive column and (5) the solvent flowrate. Therefore, five compositions can be specified. These are selected to be

d

(1) the impurity of benzene in the extractive column distillate, (2) the impurity of DMF in the extractive column distillate, (3) the impurity of cyclohexane in the extractive column

te

bottoms, (4) the impurity of cyclohexane in the rectifier distillate and (5) the impurity of

Ac ce p

toluene in the rectifier distillate.

Converging the system was challenging. The ideal procedure of using five Design

spec/Vary functions was unsuccessful. After trying numerous approaches, the method used consisted of employing two Design spec/Vary functions and manually varying the other three variables. The flowrate of the rectifier distillate was varied to achieve 1 mol% toluene impurity in the rectifier distillate. The flowrate of the extractive column distillate was varied to achieve 99 mol% cyclohexane purity in the extractive column distillate. The remaining three variables were manually adjusted to achieve the desired separations. The final design is shown in Figure 2. The flowrate of the sidestream vapor is 80 kmol/h. The solvent flowrate is 110 kmol/h. The reflux ratio is the extractive column is 4.23. The flowrate of the vapor sidestream has a strong effect on the cyclohexane impurity in the rectifier distillate. Decreasing the vapor sidestream flowrate raises the

6

Page 6 of 31

cyclohexane profile up higher in the extractive column so only a small amount enters the rectifier. The flowrate of the solvent has a strong effect on the benzene impurity in the extractive column distillate. The reflux ratio in the extractive column affects the DMF impurity in the extractive column distillate. The reboiler duty is 2.789 MW. There is no

ip t

reboiler in the rectifier.

The solvent recovery column is designed in the same way as in the conventional

cr

process. The solvent flowrate is higher, so the bottoms stream from the solvent recovery column is larger. Slightly higher reflux ratio and reboiler duty (0.843 MW) are required.

us

Note that the total reboiler duty in the two reboilers of the non-conventional partially thermal coupled columns/rectifier process is 3.632 MW, compared to the total

an

reboiler duty for the three reboilers in the conventional process of 4.230 MW. So the nonconventional process is more economical that the conventional. The issue of dynamic controllability is addressed in the following section.

M

The column sizing (number of trays and feed locations) was approximately based on the Timoshenko case study, but the use of a different solvent required some significant

d

modifications. The number of stages in all columns was increased to keep reboiler duties and reflux ratios reasonable given the design specifications on the product streams. No

Ac ce p

dynamic controllability.

te

formal economic optimization was conducted since our objective was to compare

3. Control

The dynamics of binary extractive distillation systems have been studied in a

number of papers. Both conventional two-column systems and non-conventional configurations have been explored. One of the pioneering papers by Grassi8 established a basic structure that has been widely applied for many years. Several recent papers have looked at the control of divided-wall columns9,10 and configurations with rectifiers11. There appears to be little published work dealing with the control of ternary extractive distillation processes. The two flowsheets are exported into Aspen Dynamics as a pressure-driven dynamic simulation after the plumbing (valves and pumps) and liquid holdups have been

7

Page 7 of 31

specified. Reflux-drum and column base holdups are specified to give 5 minute holdup when half full. Proportional level controllers are used (KC = 2). Temperature control loops have 1-minute deadtimes and are tuned by running relay-feedback tests to get ultimate gains and periods and using Tyreus-Luyben tuning rules.

ip t

In both flowsheets, the ternary fresh feed is flow controlled. The solvent flow is ratioed to the fresh feed .flowrate. Reflux-drum levels are controlled by manipulating

cr

distillate flowrates in all columns. Base levels are controlled by manipulate bottoms

flowrates except in the solvent recovery column in which base level is controlled by

us

manipulating the flowrate of solvent makeup. The reflux flowrate in each column is ratioed to the feed to that column. Column pressure is controlled by manipulating

an

condenser heat removal.

The main issue in finding an effective control structure is selecting a suitable

M

stage whose temperature should be controlled.

3.1 Conventional Three-Column Process

d

Figure 6 gives the temperature profiles in the three columns. One of the important functions of the extractive column C1 is to keep cyclohexane from dropping out the

te

bottom and subsequently appearing as an impurity in the downstream column C2 distillate. Figure 6 shows that there are stages in each column where temperatures change

Ac ce p

significantly from stage to stage. In the extractive column C1 this break is near the bottom of the column. A temperature controller to maintain Stage 77 temperature is tentatively selected, recognizing that pressure changes in this vacuum column may significantly affect the performance of a simple temperature controller. Reboiler duty is manipulated to control Stage 77 temperature. The other function of C1 is to maintain the purity of the cyclohexane distillate, which achieved by the solvent-to-feed ratio (S/F) and the reflux-to-feed ratio (R1/F).

One of the functions of the benzene column C2 is to keep benzene from dropping out the bottom. A significant temperature break occurs at Stage 8, so it is selected. Reboiler duty is manipulated to control Stage 8 temperature. The other function of C2 is to maintain the purity of the benzene distillate, which achieved by the reflux-to-feed ratio (R2/B1).

8

Page 8 of 31

One of the functions of the solvent recovery column C3 is to keep toluene from dropping out the bottom. A significant temperature break occurs at Stage 6, so it is selected. Reboiler duty is manipulated to control Stage 6 temperature. The other function of C3 is to maintain the purity of the toluene distillate, which achieved by the reflux-to-

ip t

feed ratio (R3/B2).

Figure 7 shows the dynamic responses of the three columns to 10% changes in

cr

fresh feed flowrate. Solid lines are 10% increases, and dashed lines are 10% decreases. Stable regulatory control is achieved. The purity of the extractive column distillate

us

cyclohexane product xD1(CH) and the purity of the solvent recovery column distillate toluene product xD3(T) are held near their specifications. However, the purity of the

specification at the two final steady states.

an

benzene column distillate products xD2(B) shows significant departurse from its

The use of plain temperature control in a vacuum column using a tray near the

M

bottom would be expected to cause some problems because of the changes in pressure that occur when vapor flowrates change. The use of pressure-compensated temperature

d

control has been extensively used for many years to solve this problem. With a binary system, procedures have been proposed to handle this problem12. The present system has

te

four components, so developing an appropriate pressure correction to apply to the

Ac ce p

measured tray temperature is less obvious. An empirical approach was taken. The simulation was run for a 10% increase in

fresh feed flowrate, holding the temperature on Stage 77 in the extractive column constant at 89.31 oC. The pressure on Stage 77 increased from the initial steady state pressure of 0.89763 bar to 0.91628 bar as the reboiler duty and vapor flowrates increased. The composition of cyclohexane on Stage 77 change from the initial 1.75 mol% to 3.21 mol%, which caused the cyclohexane impurity in the downstream C2 distillate to increase from 0.4916 to 0.9400 mol% CH. This decreased the benzene purity of the C2 distillate. Then the setpoint of the temperature controller was increased until cyclohexane impurity in the D2 distillate dropped to 0.496 mol% CH. The composition of cyclohexane on Stage 77 was 1.76 mol% CH. The new temperature was 90 oC. So a 0.019 bar change in pressure required a 0.7 oC change in Stage 77 temperature to keep the

9

Page 9 of 31

same cyclohexane composition on Stage 77. Aspen Dynamics Flowsheet Equations are used to implement a strategy in which the setpoint of the Stage 77 temperature controller (TC1) is changed as the pressure on Stage 77 changes. Note that this is a different procedure than is conventionally used in which the setpoint (SP) is fixed and the process

ip t

variable signal (PV) is calculated using both the temperature measurement on Stage 77 and the pressure measurement on Stage 77.

cr

Figure 8 illustrates the effectiveness of this pressure-compensated setpoint

approach. The composition of the benzene product xD2(B) is maintained very close to its

us

specification.

an

3.2 Non-Conventional Two-Column/Rectifier Process

The key to successful operation of the rectifier is to keep cyclohexane from dropping down to the tray where the vapor sidestream is withdrawn. Once cyclohexane

M

gets into the rectifier, it will go out with the rectifier distillate benzene product. The two heavier components toluene and DMF entering the bottom of the rectifier can be driven

d

back down the column using rectifier reflux.

te

The problem is to detect the location of the cyclohexane profile in upper part of the extractive column. Figure 9A gives the composition profiles in the extractive column

Ac ce p

C1. The compositions of the four components change significantly throughout the column. However, Figure 9B shows that the temperature profile is very flat in the upper part of the column where the cyclohexane compositions must decrease as we move down the column towards the vapor sidestream withdrawal stage. There are four components in the column, so the ability to use temperature as an indicator of cyclohexane composition is diminished compared to a system with fewer components. Therefore the use of temperature control is problematic in a ternary extractive distillation system. Simulation results confirm this expectation. The temperature on Stage 55 in the extractive column is controlled by manipulating the flowrate of the vapor sidestream. A flow controller is installed on the sidestream vapor and put on “cascade” with its remote setpoint coming from the output signal from the temperature controller, which is a “direct acting” controller. An increase in temperature indicates less cyclohexane coming down the column, so more vapor can be fed to the rectifier. The temperature on Stage 61 is

10

Page 10 of 31

controlled by manipulating reboiler duty. Extraction column reflux is ratioed to fresh feed. Temperature loops have 1-minute deadtimes. Responses using the T55 temperature controller are given in Figures 10 and 11 for 10% changes in fresh feed flowrate. In Figure 10, the rectifier reflux R2 is ratioed to the

ip t

fresh feed. The purity of the rectifier distillate benzene product xD2(B) is poor, with large transient departures from the specified 99 mol%. In addition, there are steady-state offsets

cr

from the desired purity. Dynamic performance is improved somewhat by switching to a

rectifier reflux-to-vapor sidestream ratio R2/V2, as shown in Figure 11, but steady-state

us

offsets remain.

It appears that an on-line composition measurement is required in this ternary

an

extractive distillation system with four components in the extractive column since the temperature profile is not rich enough in content to enable composition estimation. However, composition measurement usually entails larger deadtime (3 minutes) in the

M

control loop than temperature measurement (1 minute), so dynamic performance may be adversely affected.

d

Several composition control structures were tested. Figure 12 gives results when the cyclohexane composition on Stage 55 was measured and controlled by manipulating

te

the flowrate of the sidestream vapor fed to the rectifier. The cyclohexane composition of

Ac ce p

Stage 55 is 4.53 mol%. Rectifier reflux is ratioed to the flowrate of the vapor to the rectifier. An optimistically small deadtime of 1 minute is used. The responses are much improved compared to the temperature control in terms of both smaller transient deviations and smaller steady-state offsets. However, a comparison of Figure 8 (conventional process with pressure-

compensated temperature control) with Figure 12 (rectifier process with 1-minute composition measurement deadtime) demonstrates that the dynamic performance of the conventional three-column process is better than that of the non-conventional partially thermally integrated rectifier process, even when an optimistically small deadtime is assumed. The main difference is in the purity of the benzene product stream. The results shown in Figure 12 are when the cyclohexane composition on Stage 55 is controlled. To explore the question of what stage should be used for composition control, the control tray is moved up in the column to Stage 52 where the cyclohexane

11

Page 11 of 31

composition is larger (15.5 mol%). This is further away from the sidestream drawoff tray, which may make it less effective. On the other hand, the larger initial composition might improve the load response by avoiding saturation (cyclohexane composition approaching zero).

ip t

Figure 13 gives a direct comparison of using Stage 52 or Stage 55 on the

composition of the benzene product xD2(B) leaving in the distillate of the rectifier. The

cr

left graphs are for feed flowrate disturbances. The right graphs are for feed composition disturbances. In the upper right graph, the feed composition is changed from 30/30/40

us

(B/CH/T mol%) to 35/25/40. In the lower right graph, the feed composition is changed from 30/30/40 to 25/35/40. Control of Stage 55 composition is more effective than using

an

Stage 52 in terms of both peak dynamic transients and steady-state offsets (particularly for feed composition disturbances).

A direct comparison between the conventional three-column process and the non-

M

conventional two-column/rectifier process is shown in Figure 14 for both feed flowrate and feed composition disturbances. The inferiority of the dynamic responses of the non-

d

conventional process is clearly demonstrated. Remember that a 1-minute composition deadtime has been used, so the dynamic performance would be even worse than that

te

shown if a more realistic 3-minute deadtime were used.

Ac ce p

This ternary extractive distillation process provides a good example of the conflict (trade-off) between steady-state economics and dynamic controllability. The nonconventional process has lower energy cost than the conventional process, but its dynamic response to disturbances is worse.

4. Conclusion

Ternary extractive distillation systems can involve quite complex process configurations whose dynamic controllability needs to be studied. A very efficient process that cannot effectively handle the inevitable disturbances that occur during operation may suffer from major product quality, safety and environmental issues. A numerical example has been presented that illustrates the degradation in dynamic controllability that sometimes occurs when more complex non-conventional process

12

Page 12 of 31

configurations are employed. In the separation of a ternary mixture using a fourth component as solvent, the existence of four components in the extractive column makes

Ac ce p

te

d

M

an

us

cr

ip t

the use of temperature to infer composition problematic.

13

Page 13 of 31

References 1. Wankat, P. C. Equilibrium Stage Separations (1988) Elsevier

ip t

2. Stichlmair, J. G., Fair, J. R. Distillation Principles and Practices (1998) WileyVCH.

cr

3. Doherty, M. D., Malone, M. D. Conceptual Design of Distillation Systems (2001) McGraw-Hill.

us

4. Luyben, W. L., Chien, I-Lung Design and Control of Distillation Systems for Separating Azeotropes 2010Wiley.

an

5. Kiss, A., Suszwalak,D. “Enhanced bioethanol dehydration by extractive and azeotropic distillation in a dividing-wall column” Sep. Pur. Tech. 86 (2012) 7078.

M

6. Wu, Y. C., Hsu, P. H., Chien, I. “Critical assessment of the energy-saving potential of an extractive dividing-wall column” Ind. Eng. Chem. Res. 52 (2013)

d

5384-5399.

te

7. Timoshenko, A.V, Anokhina, E. A., Morgunov, A. V., Rudakov, D. G. “Application of the partially thermally coupled distillation flowsheets for the

Ac ce p

extractive distillation of ternary azeotropic mixtures” Chemical Engineering Research an Design 104 (2015) 139-155.

8. Grassi, V. G. “Process design and control of extractive distillation” In Practical Distillation Control, Van Nostrand Reinhold Press, New York (1992), 370-404.

9. Tututi-Avila, S., Jimenez-Gutierrez, A., Hahn, J. “Control analysis of an extractive dividing-wall column used for ethanol dehydration” Chem. Eng. Processing 82 (2014) 88-100.

10. Kiss, A., Suszwalak,D. “Enhanced bioethanol dehydration by extractive and azeotropic distillation in a dividing-wall column” Sep. Pur. Tech. 86 (2012) 7078. 11. Segovia-Hernandez, J. G., Vazquez-Ojeda, M., Gomez-Castro, F. I., RamiezMarquez, C., Errico, M., Tronci, S. Rong, B.G. “Process control analysis for intensified bioethanol separation systems” Chem. Eng. Proc. 75 (2014) 119-125.

14

Page 14 of 31

12. Luyben, W. L. Distillation Design and Control using Aspen Simulation, 2nd

ip t

Ed (2013) Wiley.

Figure Captions

cr

Figure 1 – Conventional ternary extractive distillation Figure 2 – Non-conventional process with rectifier

us

Figure 3 – Txy diagram at 1 atm; benzene/cyclohexane

Figure 4 – Normalized xy diagram; effect of solvent DMF and NMP

Figure 6 – Temperature profile; Conventional Figure 7 – 10% feed flowrate; conventional

an

Figure 5 – Ternary diagram for extractive column C1

M

Figure 8 - 10% feed flowrate; conventional with pressure compensation Figure 9A – Extractive column composition profiles; rectifier process

d

Figure 9B – Temperature profiles; rectifier process

Figure 10 – 10% feed flowrate; rectifier; control T55 with R2/F

te

Figure 11 – 10% feed flowrate; rectifier; control T55 with R2/V2

Ac ce p

Figure 12 – 10% feed flowrate; rectifier; control x55(CH) with D=1 and R2/V2 Figure 13 – Rectifier; Control x55 or x52 Figure 14 – Comparison; conventional and x55 rectifier; benzene purity

15

Page 15 of 31

Ac ce p

te

d

M

an

us

cr

ip t

Highlights  Controllability of non-conventional ternary extractive distillation can be poor.  Four components in the extractive column make temperature control problematic.  On-line composition measurement may be required.

16

Page 16 of 31

Makeup

cr

Fig. 1 – Conventional Ternary Extractive Distillation

i

Figure

0.50 kmol/h DMF

D1 30.15 kmol/h 0.0016 B 0.9900 CH trace T 0.0084 DMF

69 oC 0.7bar 1.251 MW

D3

D2

M an

89.74 kmol/h 50 oC 0.0041 T 0.9959 DMF

us

Solvent

oC

83 1.1bar

30.18 kmol/h 0.9900 B 0.0050 CH 0.0050 T trace DMF

1.608 MW

C1

ce pt

40

20

Ac

79

RR = 3.88 ID = 1.19 m

40

1.137 MW

C3

ed

C2

20

50 oC 0.12 bar

40.16 kmol/h 0.0017B trace CH 0.9922 T 0.0061 DMF

15 RR = 5.26 ID = 1.08 m

20

RR = 1.8 ID = 1.764 m

1.731 MW

Feed 100 kmol/h 50 oC 0.3 B 0.3 CH 0.4 T

108

1.820 MW

0.6785 MW

oC

145 oC

B1 159.6 kmol/h 0.1877 B 0.0009 CH 0.2530 T 0.5584 DMF

94 oC

B2

B3

129.4 kmol/h 0.0005 B trace CH 0.3108 T 0.6887 DMF

89.24 kmol/h 0.0041 T 0.9959 DMF

Page 17 of 31

i

Makeup

cr

Fig. 2 – Non-Conventional Process with Rectifier

0.50 kmol/h DMF

D1 30.14 kmol/h 0.001 B 0.990 CH trace T 0.009 DMF

73 oC 0.8bar 1.124MW

D3

D2

M an

110.0 kmol/h 50 oC 0.0003 T 0.9997 DMF

us

Solvent

oC

53 0.4 bar

30.29 kmol/h 0.990 B 104 ppm CH 0.010 T trace DMF

0.837 MW

C1

ce pt

40

79

RR = 4.23 ID = 1.57 m

2.789 MW

Feed 100 kmol/h 50 oC 0.3 B 0.3 CH 0.4 T

20

V2 - 80 kmol/h

Ac

60

128

oC

1.216 MW

C3

ed

C2

20

50 oC 0.12 bar

39.50 kmol/h 9 ppm B trace CH 0.994 T 0.006 DMF

11

73 oC RR = 2.07 ID = 1.01 m

B2

20

49.71 kmol/h 0.286 B 16 ppm CH 0.596 T 0.117 DMF

RR = 2 ID = 1.72 m

0.843 MW 94 oC

B1

B3

149.4 kmol/h Trace B trace CH 0.266 T 0.734 DMF

109.5 kmol/h 0.003 T 0.997 DMF

Page 18 of 31

cr

i

Fig. 3 – Txy Diagram at 1 atm; Benzene/Cyclohexane T-xy diagram for B/CH

80.75

M an

80.50 80.25 80.00

ed

79.75 79.50

ce pt

79.25 79.00 78.75 78.50

Ac

Temperature, C

x 1.0133 bar y 1.0133 bar

us

81.00

78.25 78.00 77.75 77.50 0.00

0.05

0.10

0.15

0.20

0.25

0.30

0.35

0.40

0.45 0.50 0.55 0.60 Liquid/vapor mole fraction, B

0.65

0.70

0.75

0.80

0.85

0.90

0.95

1.00

Page 19 of 31

us

cr

i

Fig. 4 – Normalized xy Diagram; Effect of Solvents DMF and NMP B/CH; 1 atm; S/F = 1; DMF/NMP

M an

1 0.9 0.8

ed

Binary Azeotrope

0.6

ce pt

0.5 0.4

No Solvent

0.3

NMP DMF

Ac

ystar (B)

0.7

0.2 0.1

0

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

xstar (B) Page 20 of 31

us

cr

i

Fig. 5 – Ternary Diagram for Extractive Column C1

CH(80.78 C)

D1 0.9

M an

Ternary Map (Mole Basis)

Isovolatility

0.8

ed

0.7

ce pt

0.6 0.5

F Azeotrope

77.54 C

0.4 0.3

Ac

M

S

0.2 0.1

DMF (151.77 C)

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

B (80.13 C)

B1 Page 21 of 31

i cr

us

Fig. 6 – Temperature Profiles; Conventional Conventional

M an

160

140

C2

ed ce pt

100

C3 80

60

40

C1

Ac

Temp (C)

120

0

10

20

30

40

50

60

70

80

Stage Page 22 of 31

0.995 0.99

73 0

1

2

3

4

72

5

1

0

1

2

3

cr 4

5

100 95

+10

1

0.995 0.99 0.985

3

4

5

0

1

2

3

4

5

76

Ac

xD3(T)

1

2

75 74

73 0

1

2

3

Time (h)

4

5

72

0

1

2

3

Time (h)

1.6

4

5

0

1

2

3

4

5

3

4

5

3

4

5

2.2

+10

2 1.8

1.6 1.4

QR3 (MW)

0

90

T3 (C)

0.985

+10

1.8

1.4

QR2 (MW)

0.99

T2 (C)

10

0.995

2

10

105

ce pt

xD2(B)

74

ed

0.985

75

M an

T1 (C)

xD1(CH)

76

QR1 (MW)

us

Conventional; +/-10 % Feed 1

i

Fig. 7 – 10% Feed Flowrate; Conventional

10 0

1

2

0.8

+10 0.75 0.7

10

0.65 0

1

2

Time (h) Page 23 of 31

1

2

3

4

5

1

2

3

4

5

0

1

3

4

0.99

0

1

2

3

4

1.6

10 75 74

0

1

2

0

1

2

3

Time (h)

4

5

73

0

1

2

3

Time (h)

4

5

4

5

3

4

5

3

4

5

2.2

+10 2

1.8

10 0

1

2

0.8

+10 0.75 0.7

10

0.65

+10 0.985

3

2.4

1.6

5

76

T3 (C)

0.995

95 90

5

Ac

2

100

QR3 (MW)

0.99

+10

1.8

1.4

QR2 (MW)

T2 (C)

0.995

2

10

105

1

xD3(T)

0

ce pt

xD2(B)

1

0.985

73

ed

0

74

QR1 (MW)

0.99

75

M an

0.995

0.985

76

T1 (C)

xD1(CH)

Conventional; Pressure Comp; +/-10 % Feed 1

us

cr

i

Fig. 8 – 10% Feed Flowrate; Conventional with Pressure Compensation

0

1

2

Time (h) Page 24 of 31

us

cr

i

Fig. 9A – Extractive Column Composition Profiles; Rectifier Process C1 in Rectifier Process 1

M an

Cyclohexane 0.9

ed

0.7

0.6

0.4

DMF

0.3

0.2

0.1

0

Benzene

ce pt

0.5

Toluene

Ac

Composition

0.8

10

20

30

40

50

60

70

80

Stage Page 25 of 31

i

us

cr

Fig. 9B – Temperature Profiles; Rectifier Process Rectifier

M an

140

130

120

ed

100

C3

ce pt

90

80

70

60

C2

50

40

C1

Ac

Temp (C)

110

0

10

20

30

40

50

60

70

80

Stage Page 26 of 31

98

0

1

2

3

4

96

5

2

3

4

0

1

2

90 88

86

3

1

4

84

5

0

1

2

3

4

70

10

65

+10

60

0

1

2

3

Time (h)

4

5

55

0

10

2.6

1

2

3

4

5

4

5

4

5

100

+10

90

80

10

70

0

1

2

3

0.95

QR3 (MW)

T3 (C)

Ac

0.99

+10

2.8

60

5

75

0.995

3

0

V2 (kmo/h)

T55 (C)

0.96

3.2

5

92

ce pt

xD2(B)

0.98

10

xD3(T)

1

94

+10

0.985

0

ed

1

0.94

100

QR1 (MW)

0.99

102

M an

0.995

0.985

104

T1 (C)

xD1(CH)

Rectifier; R2/F; TC55; D=1; 10 % Feed 1

us

cr

i

Fig. 10 – 10% Feed Flowrate; Rectifier; Control T55; R2/F

1

2

3

Time (h)

4

5

+10

0.9 0.85 0.8

10

0.75 0.7

0

1

2

3

Time (h) Page 27 of 31

100

0

1

2

3

4

98

5

0.98

4

0

1

2

3

1

4

92

90

86

5

0

1

70

2

3

4

1

2

3

Time (h)

4

5

55

10

+10 0

10

2.6

1

2

3

1

2

Time (h)

4

5

5

4

5

4

5

+10

90 85 80

10 0

1

2

3

0.95

+10

0.9 0.85 0.8

10

0.75 3

4

95

70

5

65 60

0

2.8

0

QR3 (MW)

T3 (C)

Ac

0.99

+10

75

75

0.995

3

5

88

10

xD3(T)

3

3.2

V2 (kmo/h)

0.99

0.985

2

ed T55 (C)

+10

0.97

1

94

ce pt

xD2(B)

1

0

QR1 (MW)

0.99

102

M an

0.995

0.985

104

T1 (C)

xD1(CH)

Rectifier; R2/V2; TC55; D=1; 10 % Feed 1

us

cr

i

Fig. 11 – 10% Feed Flowrate; Rectifier; Control T55 with R2/v2

0.7

0

1

2

3

Time (h) Page 28 of 31

1

3

1

2

3

1

2

3

4

4

6 4

2 0

5

0

0.085 0.08

1

2

3

4

0

1

2

3

Time (h)

4

5

10

65

55

+10 0

2

3

4

5

4

5

4

5

+10

90 85 80

10

75 0

1

2

3

3.2

60

0

1

95

70

5

QR3 (MW)

Ac

0.99

70

10

0.075

5

8

+10

0.09

75

0.995

0.985

1

ed

x55 (%CH)

10

0

0

10

+10

0.98

98

5

0.99

0.97

xD3(T)

4

T3 (C)

xD2(B)

1

2

0.095

V2 (kmo/h)

0

100

QR1 (MW)

0.99

102

M an

T1 (C)

0.995

0.985

104

ce pt

xD1(CH)

Rectifier; R2/V2; CC x55; D=1; 10 % Feed 1

us

cr

i

Fig. 12 – 10% Feed Flowrate; Rectifier; Control x55(CH) with D=1 and R2/V2

1

2

3

Time (h)

4

5

+10 3 2.8

10

2.6

0

1

2

3

Time (h) Page 29 of 31

us

cr

i

Fig. 13 – Rectifier; Control x55 or x52 10% Feed Increase 0.998

0.998

M an

0.996

0.996

xD2(B)

xD2(B)

0.994 0.992 0.99

0.994 0.992

x55

0.99

0.986 1

2

3

4

x52

0.988 0.986

5

ce pt

0

ed

0.988

0.984

0

1

10% Feed Decrease

0.99

0.98

1

2

5

4

5

x55 0.985

0.98

x52 0

4

0.99

x55

0.985

3

zB 30 to 25

Ac

xD2(B)

0.995

2

0.995

xD2(B)

1

0.975

zB 30 to 35

1

x52 3

Time (h)

4

5

0.975

0

1

2

3

Time (h) Page 30 of 31

us

cr

i

Fig. 14 – Comparison; Conventional and Rectifier; Benzene Product Purity +10 % Feed

zB 30 to 35

0.998

1

0.998

Rect

M an

0.994

xD2(B)

xD2(B)

0.996

0.992 0.99

0.994 0.992

1

2

3

4

0.988

5

0

-10 % Feed

0.992

0.988 0.986 0.984

1

2

4

5

3

4

5

0.99

Conv 0.988 0.986 0.984

Rect 0

3

0.992

Conv

0.99

2

zB 30 to 25

Ac

xD2(B)

0.994

1

0.994

xD2(B)

0.996

Conv

0.99

ce pt

0

ed

Conv 0.988 0.986

Rect

0.996

3

Time (h)

4

5

0.982

Rect 0

1

2

Time (h) Page 31 of 31