computers and chemical engineering -...

8
Computers and Chemical Engineering 73 (2015) 183–190 Contents lists available at ScienceDirect Computers and Chemical Engineering j ourna l ho me pa g e: www.elsevier.com/locate/compchemeng A semicontinuous approach for heterogeneous azeotropic distillation processes Amir Tabari a , Arshad Ahmad a,b,a Institute of Hydrogen Economy, Universiti Teknologi Malaysia, 81310 Johor Bahru, Malaysia b Faculty of Chemical Engineering, Universiti Teknologi Malaysia, 81310 Johor Bahru, Malaysia a r t i c l e i n f o Article history: Received 17 March 2014 Received in revised form 16 October 2014 Accepted 10 December 2014 Available online 19 December 2014 Keywords: HAc dehydration Heterogeneous azeotropic distillation Dividing-wall distillation columns Process optimization Semicontinuous approach a b s t r a c t The separation of azeotropes has substantial energy and investment costs, and the available methods require high capital costs for reconstruction of process plants. As an alternative, a semicontinuous con- figuration that utilizes an existing plant with minor modifications has been explored. In this paper, a semicontinuous, heterogeneous azeotropic distillation process is proposed and acetic acid dehydration process is used as a case study. To carry out the simulation work, Aspen HYSYS ® simulation software is used along with MATLAB ® and an interface program to handle the mode-transition of the semicontinu- ous process. Sensitivity analyses on operating parameters are performed to identify the process limits. Comparisons are made to conventional heterogeneous azeotropic distillation, and dividing-wall distilla- tion column on the annual cost. The results proved that the semicontinuous system is the best setup in terms of total annual costs and energy requirements. © 2014 Elsevier Ltd. All rights reserved. 1. Introduction Separation processes are important for chemical plants to recover products, by-products, and unreacted raw materials. Among the many alternative processes, distillation is still preferred despite its high energy requirement that accounts for almost 3% of the world’s energy consumption (Hewitt et al., 1999). The pro- cess is based on the difference in the boiling points of components within a mixture; when they are close together, the separation is more difficult. In cases where an azeotrope is formed, complete separation cannot be realized using ordinary fractionations, thus requiring specialized distillation systems to produce high-purity products. An example of these specialized processes is azeotropic distilla- tion, which utilizes two or more consecutive columns with a variety of designs depending on the feed and desirable product specifica- tions. Through this distillation system, a low-boiling azeotrope is formed with one of the main constituents by adding a third compo- nent (entrainer) to the binary mixture. While the desired product specifications can be obtained, this process not only complicates the design and control but also imposes extra costs due to the need for Corresponding author at: Institute of Hydrogen Economy, Universiti Teknologi Malaysia, 81310 Johor Bahru, Malaysia. Tel.: +60 75535610; fax: +60 75586166. E-mail address: [email protected] (A. Ahmad). additional columns as well as large amounts of energy. An extensive review of this issue is provided by Widagdo and Seider (1996). The heterogeneous azeotropic distillation (HAD) column is a special case of an azeotropic distillation system for the separa- tion of heteroazeotropes in which a vapor coexists with two liquid phases. In a traditional HAD process, a two-column system is used; the first column is for product separation, and the second is for entrainer recovery (Gaubert et al., 2001; Urdaneta et al., 2002). In addition to the two-column system, some studies provided config- urations with three or four consecutive columns, such as in the case of the separation of an ethanol–water mixture using benzene as the entrainer (Ryan and Doherty, 1989; Pham and Doherty, 1990). Another technique used to separate azeotropic mixtures is the dividing wall distillation column (DWC), which is favorable in terms of energy requirements and capital investment (Wright, 1949). The applications of the DWC has been commercially analyzed with results showing that replacing the conventional column with a DWC leads to 28% and 30% savings in capital and energy demands, respectively (Becker et al., 2001; Schultz et al., 2002). More recently, a semicontinuous approach to azeotropic dis- tillation columns was introduced (Phimister and Seider, 2000a,b, 2001). This cyclic system offered significant potential for the iso- lation of high-purity products using existing plants, thus lowering the required capital investment costs with a more flexible process option. These systems were invented (Haynes et al., 1992, 1995) and gradually developed (Sorensen and Skogestad, 1994) toward http://dx.doi.org/10.1016/j.compchemeng.2014.12.005 0098-1354/© 2014 Elsevier Ltd. All rights reserved.

Upload: vanliem

Post on 12-Apr-2018

216 views

Category:

Documents


2 download

TRANSCRIPT

Page 1: Computers and Chemical Engineering - FCEEfcee.utm.my/arshad/wp-content/blogs.dir/86/files/2015/04/...aration of acetic acid from water using conventional systems is not economical

Ad

Aa

b

a

ARRAA

KHHDPS

1

rAdocwmsrp

totfnsd

M

h0

Computers and Chemical Engineering 73 (2015) 183–190

Contents lists available at ScienceDirect

Computers and Chemical Engineering

j ourna l ho me pa g e: www.elsev ier .com/ locate /compchemeng

semicontinuous approach for heterogeneous azeotropicistillation processes

mir Tabaria, Arshad Ahmada,b,∗

Institute of Hydrogen Economy, Universiti Teknologi Malaysia, 81310 Johor Bahru, MalaysiaFaculty of Chemical Engineering, Universiti Teknologi Malaysia, 81310 Johor Bahru, Malaysia

r t i c l e i n f o

rticle history:eceived 17 March 2014eceived in revised form 16 October 2014ccepted 10 December 2014vailable online 19 December 2014

a b s t r a c t

The separation of azeotropes has substantial energy and investment costs, and the available methodsrequire high capital costs for reconstruction of process plants. As an alternative, a semicontinuous con-figuration that utilizes an existing plant with minor modifications has been explored. In this paper, asemicontinuous, heterogeneous azeotropic distillation process is proposed and acetic acid dehydrationprocess is used as a case study. To carry out the simulation work, Aspen HYSYS® simulation software is

®

eywords:Ac dehydrationeterogeneous azeotropic distillationividing-wall distillation columnsrocess optimizationemicontinuous approach

used along with MATLAB and an interface program to handle the mode-transition of the semicontinu-ous process. Sensitivity analyses on operating parameters are performed to identify the process limits.Comparisons are made to conventional heterogeneous azeotropic distillation, and dividing-wall distilla-tion column on the annual cost. The results proved that the semicontinuous system is the best setup interms of total annual costs and energy requirements.

© 2014 Elsevier Ltd. All rights reserved.

. Introduction

Separation processes are important for chemical plants toecover products, by-products, and unreacted raw materials.mong the many alternative processes, distillation is still preferredespite its high energy requirement that accounts for almost 3%f the world’s energy consumption (Hewitt et al., 1999). The pro-ess is based on the difference in the boiling points of componentsithin a mixture; when they are close together, the separation isore difficult. In cases where an azeotrope is formed, complete

eparation cannot be realized using ordinary fractionations, thusequiring specialized distillation systems to produce high-purityroducts.

An example of these specialized processes is azeotropic distilla-ion, which utilizes two or more consecutive columns with a varietyf designs depending on the feed and desirable product specifica-ions. Through this distillation system, a low-boiling azeotrope isormed with one of the main constituents by adding a third compo-

ent (entrainer) to the binary mixture. While the desired productpecifications can be obtained, this process not only complicates theesign and control but also imposes extra costs due to the need for

∗ Corresponding author at: Institute of Hydrogen Economy, Universiti Teknologialaysia, 81310 Johor Bahru, Malaysia. Tel.: +60 75535610; fax: +60 75586166.

E-mail address: [email protected] (A. Ahmad).

ttp://dx.doi.org/10.1016/j.compchemeng.2014.12.005098-1354/© 2014 Elsevier Ltd. All rights reserved.

additional columns as well as large amounts of energy. An extensivereview of this issue is provided by Widagdo and Seider (1996).

The heterogeneous azeotropic distillation (HAD) column is aspecial case of an azeotropic distillation system for the separa-tion of heteroazeotropes in which a vapor coexists with two liquidphases. In a traditional HAD process, a two-column system is used;the first column is for product separation, and the second is forentrainer recovery (Gaubert et al., 2001; Urdaneta et al., 2002). Inaddition to the two-column system, some studies provided config-urations with three or four consecutive columns, such as in the caseof the separation of an ethanol–water mixture using benzene as theentrainer (Ryan and Doherty, 1989; Pham and Doherty, 1990).

Another technique used to separate azeotropic mixtures is thedividing wall distillation column (DWC), which is favorable in termsof energy requirements and capital investment (Wright, 1949). Theapplications of the DWC has been commercially analyzed withresults showing that replacing the conventional column with aDWC leads to 28% and 30% savings in capital and energy demands,respectively (Becker et al., 2001; Schultz et al., 2002).

More recently, a semicontinuous approach to azeotropic dis-tillation columns was introduced (Phimister and Seider, 2000a,b,2001). This cyclic system offered significant potential for the iso-

lation of high-purity products using existing plants, thus loweringthe required capital investment costs with a more flexible processoption. These systems were invented (Haynes et al., 1992, 1995)and gradually developed (Sorensen and Skogestad, 1994) toward
Page 2: Computers and Chemical Engineering - FCEEfcee.utm.my/arshad/wp-content/blogs.dir/86/files/2015/04/...aration of acetic acid from water using conventional systems is not economical

184 A. Tabari, A. Ahmad / Computers and Chemical Engineering 73 (2015) 183–190

hia1(

imldtdtisccrp

ttvtqlf

ponents containing HAc, water and small amounts of MA and PX(Wang et al., 2008), only one column is needed in the SHAD pro-

Fig. 1. VLE and relative volatility of HAc–H2O system.

igher performance by introducing alternative configurations,ncluding systems involving middle-vessel (MV) columns (Cheongnd Barton, 1999a,b,c), multi-vessel columns (Skogestad et al.,997), heat-integrated systems and multi-effect batch columnsHasebe et al., 1999).

Successful application of middle-vessel systems extended thentegration of MVs to traditional distillation systems under the

oniker “semicontinuous processes”. The feasibility, operabi-ity and profitability of semicontinuous processes have beenemonstrated in different separation processes, such as reac-ive distillation (Adams and Seider, 2009a,b), pressure-swingistillation (Phimister and Seider, 2000a) and extractive dis-illation (Phimister and Seider, 2000b). Although considerablemprovements were achieved using semicontinuous pressure-wing distillation and semicontinuous extractive distillation,ertain restrictions, such as a limited range of separation (in thease of the pressure-swing distillation) and difficulties for entrainerecovery column (in the case of extractive distillation), made theserocesses unsuitable for a wide range of mixtures and components.

In this paper, a semicontinuous heterogeneous azeotropic dis-illation (SHAD) process is presented as a promising alternative tohe conventional configurations. In this approach, some middle-essels are attached to a principal column to cyclically carry outhe separation tasks through consecutive modes, producing highuality products with savings in investment costs and energy uti-

ization. The dehydration of acetic acid (HAc) is used as a case studyor the proposed method, and the performance is compared with

Fig. 2. Residue curve map of HAc–IBA–H2O system.

heterogeneous azeotropic distillation and the dividing wall distil-lation process.

2. Case study – HAc dehydration

One of the important operations in the aromatic acid produc-tion (e.g., terephthalic acid) is the dehydration of acetic acid (HAc)(Chien and Kuo, 2006; Gau, 2005; Parten and Ure, 1999). In thisprocess, despite the reasonably large differences in boiling pointsbetween acetic acid and water at atmospheric pressure, the sep-aration of acetic acid from water using conventional systems isnot economical due to its close volatility in dilute aqueous solu-tion (Fig. 1). Consequently, high reflux ratios, large numbers oftrays and large cross sections are necessary when high-purity prod-ucts are desired. In an HAc dehydration system, the entrainer usedforms a heteroazeotrope with the water as shown in Fig. 2. Thisheteroazeotrope is separated in a decanter after the condensa-tion process. The organic phase is then refluxed to the principalcolumn as the source of entrainer and the aqueous phase is sentout for supplementary treatments. For this process, studies on theentrainer selection pointed to the suitability of IBA (Luyben andChien, 2010), and this result is in agreement with industrial appli-cations (Costantini et al., 1981; Parten and Ure, 1999). In addition toHAc and water, the feed contains other impurities, such as methylacetate (MA) and para-xylene (PX), resulting in more difficulties indesign and operation.

3. Semicontinuous heterogeneous azeotropic distillation(SHAD) process

The proposed SHAD process satisfies the current requirementsfor inherent safety and process intensification, which call forcompact and efficient designs so that better energy efficiency isachieved while maintaining safety, product quality and environ-mental protection requirements (Loperena and Ramirez, 2004;Maleta et al., 2011). While in some real industrial applications,three columns configurations are used to separate the feed com-

cess as shown in Fig. 3. To add to its functionality, middle vessels areadded to the principal column at specified locations based on the

Page 3: Computers and Chemical Engineering - FCEEfcee.utm.my/arshad/wp-content/blogs.dir/86/files/2015/04/...aration of acetic acid from water using conventional systems is not economical

A. Tabari, A. Ahmad / Computers and Chem

Fl

of

3

ppltip

4faar

S6) located at stage 6. The bottom product of mode 2 (stream S9)and the distillate product of mode 3 (stream S7) are 99.98% H2O and

ig. 3. Flowsheet of semicontinuous approach for heterogeneous azeotropic distil-ation.

perating mode. In addition, conventional utilities and a decanteror separating the two insoluble liquid layers are utilized.

.1. Process design

As shown in Fig. 3, the proposed configuration consists of arincipal column, Tank T1, that serves as a middle vessel and fourroduct tanks (T2, T3, T4, T5). The process is run batchwise, circu-

ating the fresh feed that is blended with the recycled stream untilhe desired product purity is achieved. The number of modes andts functions are determined based on the desire products and therocess conditions.

For the HAc case study being investigated, the SHAD column has1 stages along with the middle vessels (T1–T5) and a decanter. Theeed is comprised of an equimolar mixture of HAc and water with

flow rate of 2000 kmol/h at 70 ◦C and 1 atm. The flow rates of MAnd PX in the feed are assumed to be 30 kmol/h and 0.3 kmol/h,espectively.

The operating strategy adopted a three-mode scheme:

i. Mode 1 – HAc productionA stream (S1) that contains four feed components, i.e., HAc,

water, MA and PX, is charged into tank T1, which is then fed tothe distillation column initially charged with butyl acetate. Thecolumn top vapor is condensed and fed to the decanter (streamS3) when the operation approaches the low-boiling point ofthe butyl acetate and water azeotrope. In the decanter, the twoliquid layers are separated, and the organic liquid phase is fedback from the reflux at the top of the column. The aqueous liq-uid phase is sent to tank T1. The bottom product consists of99.5 mol% HAc and is collected in tank T2.

The collection of PX is realized through a side-stream (streamS6) located at a selected column stage, where the PX compo-sition is highest. In their study, Chien et al. (2005) discussedvarious issues associated with the PX impurity accumulationproblem.

ii. Mode 2 – water recoveryIn this mode, the contents of tank T1 are fed to the column to

recover the water as the bottom product and collected in tankT3. The distillate of the column consisting of MA, IBA and a smallamount of water are sent to tank T1 as stream S4 to be processedin the next mode.

ical Engineering 73 (2015) 183–190 185

iii. Mode 3 – MA recoveryIn this mode, the column is charged with the contents of tank

T1 in stream S2. The distillate containing the MA and wateris collected in tank T5, while the bottom product, which con-tains water and IBA, are recycled as stream S5 to the decanter.Meanwhile, the entrainer is added to the decanter so thatthe entrainer loss through the outlet streams is compensated.Finally, tank T1 is recharged with the feed, and the processreturns to Mode 1.

3.2. Process simulation

A flow chart of the case study was developed in an Aspen-Hysys simulation environment. To extend its capability, so thatadditional mathematical tools can be used in the analyses, MATLABsoftware was interfaced with the simulator to provide interactivedata exchange between the two software packages.

3.2.1. Thermodynamic modelAn important step in process flowsheeting is to select suitable

thermodynamic model to represent the components involved. Forthe system being considered, the nonrandom, two-liquid (NRTL)activity coefficient model was used for the vapor–liquid–liquidequilibrium (VLL) of this system. The Aspen HYSYS built-inassociation parameters were employed to calculate the fugacitycoefficients. The Hayden-O’Connell second virial coefficient model(Hayden and O’Connell, 1975) with the association parameterswas used to account for the dimerization of acetic acid in thevapor phase. The NRTL parameters for the system are listed inTable 1. The set of NRTL parameters were obtained from Chienet al. (2004), Gau (2005) and Wang (2004). Wang et al. (2008)presented a comparison showing satisfactory agreement betweenthe experimental data (Gmehling et al., 2004) and the data pre-dicted by the NRTL model in terms of azeotropic compositions andtemperatures.

3.2.2. Software linking procedureFig. 4 illustrates the three-step software linking strategy. First,

the required initial variables of the first mode were properly setin the interface program (MATLAB). This program was then linkedto AspenHysysTM to transfer the data to the spreadsheet, whichwas previously correlated with the relevant column variables. Sec-ond, the variables imported to the spreadsheet were applied tothe column. After the simulation was run, the output values ofthe first mode (e.g., the specifications of the products and oper-ating parameters) were kept in the spreadsheet to be transferredto the interface. The simulation results were then examined basedon different criteria, such as the feasibility of the process, con-straints, etc. Finally, new design parameters were determinedbased on the data analysis and transferred to the interface pro-gram. This procedure was interactively performed for all processmodes.

3.2.3. Simulation resultsThe simulation results of the operating conditions of the HAc

dehydration process are shown in Table 2. In the first mode, the HAcproduct specifications are satisfied with a slight amount of water inthe bottom. PX is mostly withdrawn through the side draw (stream

87.92% MA, respectively. In mode 3, the entrainer loss through theside stream and the bottom products (streams S6, S8 and S9) arecompensated by adding IBA make-up to the decanter at the flowrate of 0.446 kmol/h.

Page 4: Computers and Chemical Engineering - FCEEfcee.utm.my/arshad/wp-content/blogs.dir/86/files/2015/04/...aration of acetic acid from water using conventional systems is not economical

186 A. Tabari, A. Ahmad / Computers and Chemical Engineering 73 (2015) 183–190

Table 1NRTL parameter values for the system.

Component i Component j aij aji bij (K) bji (K) ˛

HAc H2O 0 0 −211.31 652.995 0.3HAc IBA 0 0 90.286 194.416 0.3HAc MA 0 0 −239.2462 415.270 0.3HAc PX 0 0 466.217 215.826 0.3H2O IBA 0 0 1809.079 489.609 0.250H2O MA 0 0 860.256 442.401 0.383H2O PX 5.91818 −6.03013 784.86 2909.308 0.162IBA MA 0 0 −137.569 187.680 0.3IBA PX 0 0 1025.975 −724.011 0.101MA PX 0 0 1353.844 −439.072 0.3

Table 2Simulation results for the HAc dehydration process.

Stream tag Temperature (◦C) Mole flow rate (kmol/h) Mole percent

HAc H2O IBA MA PX HAc H2O IBA MA PX

S1 70 1000 1000 0.000 30 0.300 49.253 49.253 0.000 1.477 0.014S2 (mode 1) 70 1000 1000 0.000 30 0.300 49.253 49.253 0.000 1.477 0.014S2 (mode 2) 96.2 0.192 997.249 0.426 29.993 0.014 0.019 97.021 0.041 2.918 0.000S2 (mode 3) 67.5 0.000 6.519 0.423 29.990 0.014 0.000 17.644 1.144 81.172 0.037S3 97.1 0.500 1054.653 157.412 571.311 104.518 0.026 55.849 8.335 30.253 5.534S4 68.8 0.000 6.519 0.423 29.99 0.014 0.000 17.644 1.144 81.172 0.037S5 84.5 0.000 2.410 0.423 0.000 0.001 0.000 85.040 14.927 0.000 0.033S6 97.2 0.333 0.140 0.443 0.007 0.277 27.779 11.642 36.907 0.585 23.086S7 56.1 0.000 4.108 0.000 29.989 0.013 0.000 12.044 0.000 87.918 0.000

0 0.000 0.000 99.500 0.500 0.000 0.000 0.0003 0.003 0.000 0.020 99.980 0.000 0.000 0.0006 0.000 0.000 0.000 0.000 100 0.000 0.000

3

fcs

3

dHnAaidpt

3

p

0.9

0.91

0.92

0.93

0.94

0.95

0.96

0.97

0.98

0.99

1

16 19 22 25 28 31 34 37 41 44 47 50

Max

imum

HAc

com

posi

�on

Number of stages

S8 124.2 999.474 5.022 0.00S9 101.6 0.192 990.730 0.00Entrainer make-up 90 0.000 0.000 0.44

.3. Sensitivity analysis

To understand the effect of the process variables on the per-ormance of the semicontinuous process, sensitivity analyses werearried out, which focused on the number of column stages, feedtages and the boilup ratio.

.3.1. Number of stagesThe number of theoretical stages of the principal column was

efined as the number of stages required to produce the desiredAc purity in the first mode because fewer theoretical stages wereeeded to recover water and MA in Mode 2 and 3, respectively.s shown in Fig. 5, with the feed and side stream flow rates fixedt 2030.3 and 1.2 kgmol/h, respectively, the achievable HAc purityncreases with increasing number of stages, but the sensitivityecreases after stage 41. At this point, the composition of HAclateaus at 99.5 mol%. Note that the feed stage location was set byhe maximum HAc purity that was obtained, as illustrated in Fig. 5.

.3.2. Feed stageFig. 6 shows the effect of the feed location on the HAc com-

osition for different numbers of stages in Mode 1. Similar to the

Fig. 4. Schematic steps of t

Fig. 5. Effect of number of stages on the HAc composition.

number of required stages, the feed stage location was also selectedbased on the maximum HAc composition produced. As shown inFig. 6, the sensitivity of the HAc composition to the changes in the

feed stage location is significant when the location is set above themiddle of the column. However, less sensitivity is noticed when thefeed location is in the bottom half of the column.

he linking procedure.

Page 5: Computers and Chemical Engineering - FCEEfcee.utm.my/arshad/wp-content/blogs.dir/86/files/2015/04/...aration of acetic acid from water using conventional systems is not economical

A. Tabari, A. Ahmad / Computers and Chemical Engineering 73 (2015) 183–190 187

0.850.860.870.880.89

0.90.910.920.930.940.950.960.970.980.99

1

6 9 12 14 15 16 18 19 21 23 24 26 27 28 30

HAc

com

posi

�on

Feed stage

41

37

34

31

28

25

22

19

Fig. 6. Effects of feed stage on the HAc composition for columns with differentnumber of stages.

0.7

0.75

0.8

0.85

0.9

0.95

1

0.7

0.75

0.8

0.85

0.9

0.95

1

6 9 12 15 18 21 24 27 30

Com

posi

�on

(%m

ol)

Feed stage

H2O compo si�on (Mod e 2)

MA composi�on (Mode 3)

F

poctcobctt

3

webTrr

ctdtitna

1.5 1.75 2 2. 25 2. 5 2.75 3 3. 25 3. 5 3.75 4 4.25 4.5 515

18

21

24

27

30

33

36

0.5

0.6

0.7

0.8

0.9

1

Rebo

iler h

eat d

uty

(MM

kcal

/h)

HAc

com

posi

�on

(%m

ol)

boilup ra�o ( Mod e 1)

HAc composi�on

Reboiler heat duty

Fig. 8. Effect of boilup ratio on HAc composition and reboiler heat duty in Mode 1.

0.7

0.75

0.8

0.85

0.9

0.95

1

0.5

0.55

0.6

0.65

0.7

0.75

0.8

0.85

0.9

0.95

1

1.5 1.75 2 2. 25 2. 5 2.75 3 3. 25 3.5

Com

posi

�on

(%m

ol)

Boilup ra�o

H2O compo si�on (Mod e 2)

MA composi�on (Mode 3)

ig. 7. Effect of feed stage on the H2O and MA compositions in Mode 2 and Mode 3.

While the main objective of the designed process was to com-letely separate HAc in the first mode, the effects of the feed stagen the water and MA production in the succeeding modes were alsoonsidered. Fig. 7 shows the influence of the feed stage location onhe production of water and MA in Modes 2 and 3, respectively. For aolumn with 41 stages, it can be observed that the desired puritiesf H2O and MA can be obtained when the feed stage location iselow the column middle stage. From Figs. 6 and 7, it can be con-luded that the variation in the feed stage location mainly affectshe HAc composition in the first mode, justifying the selection ofhe feed location.

.3.3. Boilup ratioThe boilup ratio had a direct effect on the product purities

hen all of the other variables in the first mode were fixed. Theffect is illustrated in Fig. 8, which shows the influence of theoilup ratio on the HAc composition and the reboiler heat duty.he higher the desired HAc composition, the higher the boilupatio should be, and consequently, a higher reboiler heat duty isequired.

Fig. 9 shows the effects of the boilup ratio on the water and MAompositions in the subsequent modes. The number of stages andhe feed stage location were fixed at 41 and 30, respectively. Asepicted in Fig. 9, the water purity increases in the second mode as

he boilup ratio increases. A similar trend is observed in the increas-ng boilup ratio with the MA composition in Mode 3. However, inhe case of the MA composition in Mode 3, a lower boilup ratio waseeded to reach the lower target value of the MA composition ofpproximately 88 mol%.

Fig. 9. Effect of boilup ratio on H2O and MA compositions through Mode 2 and Mode3.

4. Process optimization and comparison of performances

To evaluate the proposed SHAD configuration, comparisonswith the performance of the conventional HAD column and theDWC were carried out. The design of both the DWC and HAD pro-cesses were based on the settings used by Wang et al. (2008).The HAD process was designed based on three consecutive col-umn using isobutyl acetate (IBA) as the entrainer (see Fig. 10), andthe DWC configuration was designed by integrating a dividing-wallcolumn (modeled as one prefractionator and a main column) intoa HAD Column (see Fig. 11).

A variety of algorithms have been used by previous researchersto optimize the design of azeotropic distillation columns (Menezesand Cataluna, 2008; Pommier et al., 2008; Szitkai et al., 2002). Inthis study, a particle swarm optimization (PSO) algorithm was used.The PSO is an evolutionary computation technique (Kennedy &Eberhart, 1995) by which random solutions (populations) are ini-tially selected, and then, the generations are progressively updatedin the search for the optimal solution. This technique provides fasterconvergence as illustrated in the work of other researchers (Coelloet al., 2004; Wang et al., 2011). Note that the tuning parametersof the PSO (e.g., population size, cognitive and social parameters,maximum velocity, etc.) were obtained from Patel and Rao (2010).

4.1. Formulation of the optimization problem

In this work, a detailed cost function covering all significantparameters was firstly formed. The total annual cost (TAC) isthe main cost criterion, which consists of the capital costs forthe equipment and all of the operating expenditures. Douglas(1989) presented a formula for calculating the annual capital cost

Page 6: Computers and Chemical Engineering - FCEEfcee.utm.my/arshad/wp-content/blogs.dir/86/files/2015/04/...aration of acetic acid from water using conventional systems is not economical

188 A. Tabari, A. Ahmad / Computers and Chemical Engineering 73 (2015) 183–190

Fig. 10. Flowsheet of the conventional plant-wide design.R

(f

A

hsct

R

eproduced from Wang et al. (2008).

ACC) using the Marshall–Swift index for cost actualization asollows:

CCcolumn =(

M&S

280

)∗ 120 ∗ (d ∗ 3.28) ∗ (H ∗ 3.28)0.8

∗(

2.18 + Fc

LCT

)(1)

Here, M&S is the Marshall–Swift index (M&S = 1483), H is the

eight of the column (H = N*0.8 [m]) where N is the number oftages, LCT is the Life Cycle Time (LCT = 3 years), Fc is the designonsideration (Fc = 1.6 and for the DWC Fc = 2.08 because the capi-al cost of the DWC is 30% greater than the conventional column, as

Fig. 11. Flowsheeteproduced from Wang et al. (2008).

estimated in previous reports (Dejanovic et al., 2010, 2011)). Thediameter of the column d [m] was computed using Eq. (2) below:

d =√(

4 ∗ V ∗ (RR + 1)�ϕvg

)(2)

where RR is reflux ratio, V is volumetric flow rate of the distillatein vapor phase [m3/s], ϕ is the ratio of the free cross section area ofthe column for the vapor flow (ϕ = 0.7), and vg is linear velocity ofthe vapor [m/s] (vg = 0.833 m/s).

of the DWC.

Page 7: Computers and Chemical Engineering - FCEEfcee.utm.my/arshad/wp-content/blogs.dir/86/files/2015/04/...aration of acetic acid from water using conventional systems is not economical

A. Tabari, A. Ahmad / Computers and Chemical Engineering 73 (2015) 183–190 189

Table 3Optimal design variables of the three systems.

Design variables C – HAD DW – HAD SHAD

C1 C2 C3 C1 DWC Mode 1 Mode 2 Mode 3

Number of stages 45 8 15 45 15 41Feed stage location 9 4 7 9 8 30Reflux ratio 0.7 1.1 1.4 0.71 1.6 1.8 1.3 1.1Side draw flow rate (kmol/h) 1.2 – – 1.1 1.2 1.2 – –Liquid split ratio – – – – 1.1 – – –Vapor split ratio – – – – 3.5 – – –Qc (MJ) 31,192.5 5019.6 2077.7 31,192.5 5369.1 36,128.6Qr (MJ) 41,451.2 36,455.8 35,924.4ACCcolumn (k$/year) 271 95 135 271 180 370

w

A

A

A

wr[rc

at

T

at(

P

wmpap

esomvicut

4

Ai

ACCutilities (k$/year) 138 85 56

AOC (k$/year) 1970 1100 830

TAC (k$/year) 4680 4097

The annual capital cost of the utilities (condenser and reboiler)as determined using the following relationships:

CCutilities =(

M&S

280

)∗ 101.3 ∗ [(Ac ∗ 3.28)0.65

+ (Ar ∗ 3.28)0.65] ∗(

2.29 + Fc

LCT

)(3)

c =(

Qc

Uc ∗ LMTD

)(4)

r =(

Qr

Ur ∗ LMTD

)(5)

here Qc and Qr are the condenser and reboiler heat duty [kJ/h],espectively, Ac and Ar are the area of the condenser and reboilerm2], respectively, Ur is the overall heat transfer coefficient of theeboilers (Ur = 3400 kJ/m2 h ◦C), and Uc is the overall heat transferoefficient of the condensers (Uc = 2800 kJ/m2 h ◦C).

Steam, cooling water and electrical power costs were considereds the annual operation costs of the column (AOC). Consequently,he total annual cost (TAC) of the column takes the form below:

AC = ACCcolumn + ACCutilities + AOC (6)

The objective function contains the total annual cost (TAC) and penalty function (P) to ensure that the desirable compositions ofhe products are achieved. The penalty function is as shown in Eq.7) below:

= K1 ∗ 100000(1/(1−abs(K2−Z))) (7)

here K1 is the proportionality factor (the value of K1 is deter-ined so that the condition P > 2*TAC is assured), K2 is the desired

urity, and Z is the actual purity. This penalty function should bedequately large to move the design parameters into the mostromising operating space (Modla, 2013).

Once the TAC is formed, the minimization of the TAC is consid-red as the objective of the optimization problem. The number oftages, the feed stage location, the reflux ratio and the flow ratef the side stream were used as the general design variables forinimization of the objective function. Moreover, the liquid and

apor ratios were also considered as additional design variablesn the optimization of the DWC. Note that in this study, the purityonstraints were put into the penalty function so that the HAc prod-ct was above 99.5 mol%, and the HAc loss was below 0.025 mol%hrough the aqueous phase product.

.2. Results

All three distillation systems were rigorously simulated inspenHysysTM and then, the comparisons of the optimum operat-

ng conditions of the three separation strategies were made based

138 78 1601970 1460 28503380

on the optimization studies. Table 3 shows the optimal design vari-ables of the three systems considered. Out of these three systems,the proposed SHAD system emerged as the best, followed by theDWC. The SHAD system offers savings of 20% on the ACC and 17%on the AOC over the DWC, resulting in an overall savings of 17%on the TAC. Similarly, when the DWC and the HAD processes werecompared, the former provided a 14% reduction in the ACC and a12% savings in the AOC, resulting in a 12% reduction in the TAC.

The results show that by operating in semicontinuous mode,more savings are obtained. This result was evident from compar-ison of the SHAD process with the conventional HAD process, inwhich a 27% reduction in the TAC was obtained due to 32% and 26%reductions in the ACC and AOC, respectively.

5. Conclusions

A new approach for heterogeneous azeotropic distillationcolumns (i.e., the SHAD process) was proposed in this study toreduce the total annual costs and the energy requirements whilemaintaining a high purity. The study was illustrated by a case studythat was simulated using AspenHysys and MATLAB. The sensitiv-ity analyses carried out have identified the process responses andoperating limits. A comparative study that was based on the opti-mal design configuration of the SHAD process with DWC and HADprocesses revealed the superiority of the SHAD strategy. From aneconomic standpoint, the SHAD column significantly outperformsthe other traditional alternatives with lower investment costs andgreater energy-saving capabilities.

Acknowledgments

This study is partly supported by the Ministry of Education, Gov-ernment of Malaysia through the Prototype Research Grant Schemeand Universiti Teknologi Malaysia through RUGS-07H12.

References

Adams TA, Seider WD. Semicontinuous reactive extraction and reactive distillation.Chem Eng Res Des 2009a;87:245–62.

Adams TA, Seider WD. Design heuristics for semicontinuous separation processeswith chemical reactions. Chem Eng Res Des 2009b;87:263–70.

Becker H, Godorr S, Kreis H, Vaughan J. Partitioned distillation columns – why, whenand how. Chem Eng 2001;108:68–74.

Cheong W, Barton PI. Azeotropic distillation in a middle vessel batch column.I. Model formulation and linear separation boundaries. Ind Eng Chem Res1999a;38:1504–30.

Cheong W, Barton PI. Azeotropic distillation in a middle vessel batch column. II.Nonlinear separation boundaries. Ind Eng Chem Res 1999b;38:1531–48.

Cheong W, Barton PI. Azeotropic distillation in a middle vessel batch column. III.Model validation. Ind Eng Chem Res 1999c;38:1549–64.

Chien IL, Kuo CL. Investigating the need of a pre-concentrator column for aceticacid dehydration system via heterogeneous azeotropic distillation. Chem EngSci 2006;61:569–85.

Page 8: Computers and Chemical Engineering - FCEEfcee.utm.my/arshad/wp-content/blogs.dir/86/files/2015/04/...aration of acetic acid from water using conventional systems is not economical

1 Chem

C

C

C

C

D

D

D

G

G

G

H

H

H

H

HK

L

L

M

90 A. Tabari, A. Ahmad / Computers and

hien I, Zeng K, Chao H, Liu J. Design and control of acetic acid dehydration systemvia heterogeneous azeotropic distillation. Chem Eng Sci 2004;59:4547–67.

hien IL, Huang HP, Gau TK, Wang CH. Influence of feed impurity on the design andoperation of an industrial acetic acid dehydration column. Ind Eng Chem Res2005;44:3510–21.

oello CA, Pulido GT, Lechuga MS. Handling multiple objectives with particle swarmoptimization. IEEE Trans Evol Comput 2004;8:256–79.

ostantini G, Serafini M, Paoli P. Process for the recovery of the solvent and of theby-produced methylacetate in the synthesis of terephthalic acid. U.S. Patent4,250,330; 1981.

ejanovic I, Matijasevic LJ, Halvorsen IJ, Skogestad S, Jansen H, Kaibel B, et al. Design-ing four-product dividing wall columns for separation of a multicomponentaromatics mixture. Chem Eng Res Des 2011;89:1155–67.

ejanovic I, Matijasevic Lj, Olujic Z, Halvorsen IJ, Skogestad S, Jansen H, et al. Concep-tual design and comparison of four-products dividing wall column for separationof a multicomponent aromatics mixture. In: Proceedings of Distillation Absorp-tion, Eindhoven: Eindhoven University of Technology; 2010. p. 85–90.

ouglas JM. Conceptual design of chemical processes. McGraw-Hill: New York;1989.

au TK. [Master’s thesis] Effect of feed impurity on the design of and control ofheterogeneous azeotropic distillation for acetic acid dehydration (II) [Master’sthesis]. Taiwan: Taiwan University; 2005.

aubert MA, Gerbaud V, Joulia X, Peyrigain PS, Pons M. Analysis and multiple steadystates of an industrial heterogeneous azeotropic distillation. Ind Eng Chem Res2001;40(13):2914–24.

mehling J, Menke J, Krafczyk J, Fischer K. Azeotropic data. Weinheim: Wiley-VCH;2004.

asebe S, Noda M, Hashimoto I. Optimal operation policy for total reflux and multi-effect batch distillation systems. Comput Chem Eng 1999;23:523–32.

ayden JG, O’Connell JP. A generalized method for predicting second virialcoefficients. Ind Eng Chem Process Des Dev 1975;14:209.

aynes TN, Georgakis C, Caram HS. The application of reverse flow reactors toendothermic reactions. Chem Eng Sci 1992;47:2927–32.

aynes TN, Georgakis C, Caram HS. The design of reverse flow reactors for catalyticcombustion systems. Chem Eng Sci 1995;50:401–16.

ewitt G, Quarini J, Morrell M. More efficiency distillation. Chem Eng 1999;21.ennedy J, Eberhart RC. Particle swarm optimization. In: Proceedings of IEEE inter-

national conference on neural networks; 1995.operena MR, Ramirez JA. Some aspects of the operation of semicontinuous, middle-

vessel distillation columns. Chem Eng Commun 2004;191:1437–55.uyben WL, Chien IL. Design and control of distillation systems for separating

azeotropes. AlChE, Wiley; 2010.aleta VN, Kiss AA, Taran VM, Maleta BV. Understanding process intensification in

cyclic distillation systems. Chem Eng Process 2011;50:655–64.

ical Engineering 73 (2015) 183–190

Menezes EW, Cataluna R. Optimization of the ETBE (ethyl tert-butyl ether) produc-tion process. Fuel Process Technol 2008;89:1148–52.

Modla G. Energy saving methods for the separation of a minimum boiling pointazeotrope using an intermediate entrainer. Energy 2013;50:103–9.

Parten WD, Ure AM. Dehydration of acetic acid by azeotropic distillation in theproduction of an aromatic acid. U.S. Patent. 5,980,696; 1999.

Patel VK, Rao RV. Design optimization of shell-and-tube heat exchangerusing particle swarm optimization technique. Appl Therm Eng 2010;30:1417–25.

Pham H, Doherty M. Design and synthesis of heterogeneous azeotropic distillations– III. Column sequences. Chem Eng Sci 1990;45:1845–54.

Phimister JR, Seider WD. Semicontinuous, pressure-swing distillation. Ind Eng ChemRes 2000a;39:122–30.

Phimister J, Seider WD. Semicontinuous, middle-vessel, extractive distillation. Com-put Chem Eng 2000b;24:879–85.

Phimister JR, Seider WD. Bridge the gap with semicontinuous distillation. Chem EngProg 2001;97(88):72–8.

Pommier S, Massebeuf S, Kotai B, Lang B, Baudouin O, Floquet P, et al. Heteroge-neous batch distillation processes: real system optimisation. Chem Eng Process2008;47:408–19.

Ryan P, Doherty M. Design/optimization of ternary heterogeneous azeotropic dis-tillation sequences. AIChE J 1989;35:1592–601.

Schultz M, Stewart D, Harris J. Reduce costs with dividing-wall columns. Chem EngProg 2002;98:64–71.

Skogestad S, Wittgens B, Litto R, Sorensen E. Multivessel batch distillation. AIChE J1997;43:971–8.

Sorensen E, Skogestad S. Optimal operating policies of batch distillation with empha-sis on the cyclic operating policy. In: Proceeding of proceedings process systemsengineering (PSE); 1994. p. 449–56.

Szitkai Z, Lelkes Z, Rev E, Fonyo Z. Optimization of hybrid ethanol dehydrationsystems. Chem Eng Process 2002;41:631–46.

Urdaneta RY, Bausa J, Brggemann S, Marquardt W. Analysis and conceptualdesign of ternary heteroazeotropic distillation processes. Ind Eng Chem Res2002;41(16):3849–66.

Wang CH. [Master’s thesis] Influences of feed impurity for heterogeneous azeotropicdistillation [Master’s thesis]. Taiwan: Taiwan University; 2004.

Wang G, Zhao G, Li H, Guan Y. Multi-objective optimization design of the heating-cooling channels of the steam-heating rapid thermal response mold usingparticle swarm optimization. Int J Therm Sci 2011;50:790–802.

Wang S, Lee C, Jang S, Shieh S. Plant-wide design and control of acetic aciddehydration system via heterogeneous azeotropic distillation and divided walldistillation. J Process Control 2008;18:45–60.

Widagdo S, Seider WD. Azeotropic distillation. AIChE J 1996;42:96–130.Wright R. Fractionation apparatus. US Patent, No. 2471134; 1949.