Conce atimpro ys

SumanDepartment wai,

a

R vity b

re ne c

a acta

m ts (e

a . The

h sible

T rope

e nden

K distil

1. Introduction

Integrationdistinct advreaction focombinatiounit. Variouin case ofintegrationcial applicaSharma anof the studtions, selecapplicationalcohol (Poization of iin which umediate prexperimen

Reactiveselectivity ouct is desirereactions. Rmixture or

tants can lead to reduction in the rates of side reactions. In ourearlierwork on single reactant system (Agarwal et al., 2008a,b),of reaction with separation in a single unit offersantages over conventional sequential approach of

llowed by separation. Reactive distillation (RD) is an of chemical reaction with distillation in a singles advantages of RD include increased conversionequilibrium reactions, improved selectivity, heat, avoidance of azeotropes, etc. Various commer-tions of RD have been reviewed elsewhere (e.g. seed Mahajani, 2002; Hiwale et al., 2004). While, mosties in the past target equilibrium controlled reac-tivity engineering is also emerging as a promisingof RD. Aldol condensation of acetone to diacetone

drebarac et al., 1998; Thotla et al., 2007) and dimer-sobutene (Talwalkar et al., 2007) are the processesse of RD to improve selectivity towards the inter-oduct has been demonstrated successfully throughtal studies and theoretical analysis.distillation can be effectively used to improve thef a reaction especiallywhen an intermediate prod-d in the series or combination of series andparallelemoving one of the products from the reactionmaintaining low concentration of one of the reac-

we showed that RD can be used as a promising tool to improvethe selectivity of the desired product in a single reactant system.For a non-azeotropic reacting system with series (ABC)and series-parallel (2AB; A+BC) reactions, Damkhlernumber can be manipulated such that maximum possibleselectivity close to 100% can be achieved at any conversion.As a general rule, for a non-azeotropic system, it has beenrecommended that if the reactant (A) is more volatile thanthe desired product then the reactive section of RD should beplaced in the rectifying zone of the reactive distillation col-umn and if it is less volatile than the desired product thereactive section should be placed in the stripping section. TheDamkhler number per stage should be as small as possibleand the column should be operated at sufciently high re-boiler duty. Essentially the choice of parameters is governedby distillation requirements, and any decision that favors dis-tillative separation of the intermediate product (B) from thereactive zone is recommended for the improvement in theperformance. The separation of the intermediate product (B)from the reactive zone helps maintain low concentration ofthis component on the catalyst thereby suppressing its fur-ther side reaction(s). In all these congurations the column isptual design of reactive distillvement in multiple reactant s

Thotla, Sanjay M. Mahajaniof Chemical Engineering, Indian Institute of Technology, Bombay, Po

b s t r a c t

eactive distillation (RD) can be effectively used to improve selecti

actants and products inside the reactive zone of a RD column. O

zeotropes, if any, donot limit the separation. As against the single re

ore design options are available in the case of multiple reactan

nd we propose design guidelines for such systems in this work

ypothetical reaction systems, involving both reversible and irrever

he cases associated with non-ideality through formation of azeot

xtended to real processes involving reactions such as cross aldol co

eywords: Chemical reactors; Conceptual design; Design; Reactiveion for selectivitytems

Mumbai 400 076, India

y manipulating the composition proles of

an obtain close to quantitative yields if the

nt systems (i.e. condensation, dimerization),

.g. esterication, cross aldol condensation)

parametric studies for the representative

reactions, form the basis of these guidelines.

s are also discussed. The analysis is further

sation, esterication and transesterication.

lation; Selectivity; Azeotrope

62

Nomenc

aCFHHcatklL, VPrkv

x, y

Greek let

SubscriptcatijlNC

Superscrij

AbbreviaDaRDHSRSHSRRSBRSHRRP

SHRSP

Graphica

operated eiing on thedistillationproduct strwhen the re

idue curve map), a multi-product hybrid congurationcapmideory2008lature

activityconcentrationfeed ow rate (mol/s)

the resthat isin theThe thet al. (hold upcatalyst hold up (kg)rate constantL/V ratioliquid, vapor ow rate, respectively (mol/s)pressurerate for kth reaction, mol/(s g) catalyst loadingV/L ratioliquid, vapor composition, respectively

terwarped time

scatalystith componentjth stageliquid sidetotal number of components

ptjth stage

tionsDamkhler Numberreactive distillationsemi-batch reactive strippingsemi-batch reactive recticationsemi-batch reactorsemi-batch reactive rectication with productremovalsemi-batch reactive stripping with productremoval

l symbols

semi-batch reactive rectication

semi-batch reactive stripping

semi-batch reactive rectication with prod-uct removal

semi-batch reactive stripping with productremoval

SBR

ther in total reux or in total re-boil mode depend-volatility of the reactant. The proposed reactivecolumn thus has a single feed stream and singleeam (Fig. 1a and b). In particular cases however,actant is intermediate boiling (i.e. a saddle node in

studies byThe rea

such as hytion, crossof systemsto side reaindustrialdesigned Rcantly. Thewith side iwith 100% yin the reactversion levthe desiredest. Thougselectivity iation requilow feed rator leadingwith contintion mixturto 100% yiemance. In tof the secodesign as ibetter undetual designthis subjec

In the ptive exampreactions aRD conguThe analystillation (Scontinuousoped for hto selectedsuch as croethyl ketonglyoxal andbutanol.

The rstant systemis not necesproduct remnicantly. Iand demonyields.

2. Exproduct r

A reacting sreactions is

desired rea

undesiredable of maintaining the saddle node compositiondle reactive zone is recommended (Fig. 1c and d).developed for a single reactant system by Agarwala,b) has been validated through the experimentalThotla et al. (2007) and Talwalkar et al. (2007).ctions with multiple reactants (i.e. two reactants)dration, alkylation, trans-esterication, esterica-aldol condensation, etherication etc. form a classwhich often suffer from selectivity issues due

ctions. Table 1 lists various potentially importantreactions which fall in this category. A properlyD column can improve the performance signi-conventional reactors, e.g. batch reactors or PFR

njection, fall short in producing desired productsield because the product formed is always presention mixture, which is prone to either limit the con-el due to reversibility or decrease the selectivity ofproduct under the operating conditions of inter-

h theoretically it is also possible to achieve 100%n these conventional reactors, the design and oper-rements at times, are unfavorable, e.g. extremelyte of one of the reactants in the semi-batch reac-to very long reaction times. On the other hand,uous removal of the desired product from the reac-e through reactive distillation, one can reach closeld of the desired product and enhance the perfor-he reactions involving two reactants, the presencend reactant has a considerable inuence on thets presence also opens up new design options. Arstanding of such cases and a systematic concep-method is not evident in the existing literature ont.resent work, we consider three different illustra-les of multiple reactant systems involving sidend propose design guidelines so as to obtain aration capable of offering close to 100% yield.is is performed for a semi-batch reactive dis-BRD) conguration and later the correspondingversions are suggested. The methodology devel-

ypothetical reactions has been further extendedcommercially important processes from Table 1,

ss aldol condensation of acetaldehyde with methyle, esterication of glutaric acid, acetalization oftransesterication of cyclohexyl acetate with n-

t illustrative example considers the multiple reac-s when simultaneous product removal from SBRDsarywhereas in the second example simultaneousoval of theproduct enhances theperformance sig-

n the last example we consider azeotropic systemsstrate the effect of azeotrope on the attainable

ample 1: SBRD without simultaneousemoval

ystem involving following irreversible liquid phaseconsidered as the rst illustrative example:

ction : A + B C

reaction : B + C D

63

Fig. 1 (a)congurati

The reactiois given by

r1 = k1xAxB

r2 = k2xCxB

The ratio oThe conv

per molestant B usedB reacted. Y

Fig. 2 (a)(HSRS).Reactive rectication (A>B, C), (b) reactive stripping (A

64

Table 1 Industrially important reactions for improving selectivity of desired product(s)

Reactions Desired product(s) Reactionconditions/volatility

order

References

Esterications(1)

Adipic acid (A) +methanol (B)monomethyl adipate(C) +water (D)

Dimethyl adipateIon exchange resin,60120 C/B>A>C>D

Hung et al.(2007)

Monomethyl adipate (C) +methanol (B)dimethyladipate+water (D)

(2)Oxalic acid (A) +n-butanol (B)monobutyl oxalate(C) +water (D)

Dibutyl oxalatePTSA on granular activecarbon/D>B>C>E>A

Sheng-wenand Ji-zhong(2005)Monobutyl oxalate (C) +n-butanol (B)dibutyl oxalate

(E) +water (D)(3)

Glyoxalic acid (A) +n-butanol (B)butyl glyoxalate(C) +water (D)

Dibutoxy butylglyoxalate

Ion exchange resin,100120 C/D>B>C>E>A

Mahajani(2000)

Butyl glyoxalate (C)(C) +n-butanol (B)dibutoxy butylglyoxalate (E) +water (D)

(4)Glyoxalic acid (A) + isoamyl alcohol (B) isoamylglyoxalate (C) +water (D)

Diisoamoxy isoamylglyoxalate

Ion exchange resin,100120 C/D>B>C>E>A

Mahajani(2000)

Isoamyl glyoxalate (C) + isoamyl alcohol(B)diisoamoxy isoamyl glyoxalate (E) +water (D)

(5)Glutaric acid (A) +methanol (B)monomethylglutarate (C) +water (D)

Dimethyl glutarateIon exchange resin,90180 C/B>A>C>D

Hung et al.(2007)

Monomethyl glutarate (C) +methanol (B)dimethylglutarate (E) +water (D)

(6)Maleic acid (A) + ethanol (B)monoethyl maleate(C) +water (D)

Diethyl maleateIon exchange resin,5080 C/B>A>C>D

Yadav andThathagar(2002)Monoethyl maleate (C) + ethanol (B)diethyl maleate

(E) +water (D)(7)

Ethylene glycol (A) + acetic acid (B)ethylene glycolmonoacetate (C) +water (D)

Ethylene glycoldiacetate

Ion exchange resin,2540 C/B>A>C>D

Schmid et al.(2008)

Ethylene glycol monoacetate (C) +Acetic acid (B)Ethyleneglycol diacetate (E) +Water (D)

(8)Glycerol (A) + acetic acid (B)monoacetine (C) +water(D) Triacetine

Ion exchange resin,60100 C/B>D>A>C>E> F

Gelosa et al.(2003)

Monoacetine (C) + acetic acid (B)diacetine (E) +water(D)Diacetine (E) + acetic acid (B) triacetine (F) +water (D)

(9)Citric acid (A) + ethanol (B)monoethyl citrate(C) +water (D) Trietyl citrate

Ion exchange resin,78120 C/B>D>A>C>E> F

Kolah et al.(2007)

Monoethyl citrate (C) + ethanol (B)diethyl citrate(E) +water (D)Diethyl citrate (E) + ethanol (B) triethyl citrate(F) +water (D)

Acetalizations(1)

Glyoxal (A) + alcohola (B)monoacetal (C) +water (D)Diacetal

Resin,6085 C/B>D>C>E>A(for methanol)D>B>C>E>A (for otheralcohols)

Mahajaniand Sharma(1997)

Monoacetal (C) + alcohola (B)diacetal (E) +water (D)

Trans-vinylation(1)

Vinyl acetate (A) + stearic acid (B)vinyl stearate(C) + acetic acid (D)

Vinyl stearateMercuric acetate andsulphric acid,7095 C/D>A>C>E>B

Geelen andWijffels(1965)Vinyl acetate (A) + acetic acid (D)ethylidene acetate

(E)Etherications

(1)Ethylene glycol (EG) (A) + isobutylene(B)mono-tert-butyl ether of EG (C) Diethers of ethylene

glycolIon exchange resin,5090 C/B>E> F>C>D>A

Klepacova etal. (2007)Mono-tert-butyl ether of EG (C) + isobutylene

(B)Di-tert-butyl ether of EG (D)

65

Table 1 (Continued )

Reactions Desired product(s) Reactionconditions/volatility

order

References

Isobutylene (B) + isobutylene (B)diisobutylene (E)Diisobutylene (E) + isobutylene (B) triisobutylene (F)

(2)Propylene glycol (PG) (A) + isobutylene(B)mono-tert-butyl ether of PG (C) Di ethers of

propylene glycolIon exchange resin,5070 C/B>E> F>C>D>A

Jayadeokarand Sharma(1993)

Mono-tert-butyl ether of PG (C) + isobutylene(B)di-tert-butyl ether of PG (D)Isobutylene (B) + isobutylene (B)diisobutylene (E)Diisobutylene (E) + isobutylene (B) triisobutylene (F)

(3)Glycerol (A) + isobutylene (B)mono-tert-butyl ether ofglycerol (MTBG) (C)

Tri-tert-butyl etherof glycerol

Ion exchange resin,5090 C/B> F>G>C>D>E>A

Klepacova etal. (2007)

MTBG (C) + isobutylene (B)di-tert-butyl ether ofglycerol (DTBG) (D)DTBG (D) + isobutylene (B) tri-tert-butyl ether ofglycerol (TTBG) (E)Isobutylene (B) + isobutylene (B)diisobutylene (F)Diisobutylene (E) + isobutylene (B) triisobutylene (G)

Trans-esterications(1)

Dimethyl carbonate (A) + ethanol (B)monomethylethyl carbonate (C) +methanol (D)

Diethyl carbonateIon exchange resin,75 C/D>B>A>C>E

Luo and Xiao(2001)

Monomethyl ethyl carbonate (C) + ethanol (B)diethylcarbonate (E) +methanol (D)

(2)Dimethyl oxalate (A) + phenol (B)monomethylphenyl oxalate (C) +methanol (D)

Mmethyl phenyloxalate

Homogeneousorganotin compounds,120220 C/D>B>C>A>E

Nishihira etal. (2000)

Monomethyl phenyl oxalate (C) + phenol (B)diphenyloxalate (E) +methanol (D)

(3)Cyclohexyl acetate (A) +n-butanol (B) cyclohexanol(C) +n-butyl acetate (D)

CyclohexanolIon exchange resin,120160 C/F>B>D>C>A>E

Chakrabartiand Sharma(1992a)Cyclohexanol (C) +n-butanol (B)butyl cyclohexyl

ether (E) +water (F)(4)

Triacetine (A) +methanol (B)diacetine (C) +methylacetate (D) Glycerol

Ion exchange resin,60 C/B>D> F>E>C>A

Lopez et al.(2005)

Diacetine (C) +methanol (B)monoacetine (E) +methylacetate (D)Monoacetine (E) +methanol (B)glycerol (F) +methylacetate (D)

(5)Ethylene carbonate (A) +methanol (B)2-hydroxymethyl carbonate (C)

Dimethyl carbonateIon exchangeresin/B>A>C>D>E

Scott et al.(2003)

2-Hydroxy methyl carbonate (C) +methanol(B)dimethyl carbonate (D) + ethylene glycol (E)

Hydrations(1)

Propene (A) +water (B) isopropyl alcohol (C)Isopropyl alcohol

Ion exchange resin,130 C/A>C>B>D

Petrus et al.(1984)Isopropyl alcohol (C) + propene (A)diisopropyl ether

(D)(2)

Ethylene oxide (A) +water (B)ethylene glycol (C)Ethylene glycol

Ion exchange resin,200300 C/A>B>C>D

Reed et al.(1956)Ethylene glycol (C) + ethylene oxide (A)diethylene

glycol (D)(3)

Isobutene (A) +water (B) tert-butyl alcohol (C)tert-Butyl alcohol

Ion exchange resin,5080 C/A>B>C>D

Zhang et al.(2003)Isobutene (A) + isobutene (A)diisobutylene (D)

Aminations(1)

Methanol (A) + urea (B)monomethyl amine(C) +water (D)

Dimethyl amineDibutyltin dimethoxide,Tetraphenytin,145191 C/A>D>C>E>B

Ryu andGelbein(2002)Methanol (A) +monomethyl amine (C)dimethyl

amine (E) +water (D)

66

Table 1 (C

Reactions duct

(2)Ethylenamine

ineEthylen(C)dEthylenamineEthylenEthylenglycol

Alkylations(1)

p-Cres

-cres2-tert-B2,4-di-IsobutyDiisobu

(2)Phenol(C)

henoPhenol(B)OCycloh(C) oortho/p(B)d

(3)Phenol(B) o2 ortho2 Alph

a Methano

then a semhybrid semThe reboilethe columnhybrid colustages placsufcientlyzone to sustages servdesired proconguratireactive strtwo new cosystems, vihybrid semiof one of th(i.e. B). Likecapable ofration is poand HSRR (reactive zonyield thanwise similaa detailed asuggest themodels fora hypothetthe effectsology at coontinued )

Desired pro

e oxide (A) + ammonia (B)monoethanol(C) +water (D)

Diethanol ame oxide (A) +monoethanol amineiethanol amine (E) +water (D)e oxide (A) + diethanol amine (E) triethanol(F) +water (D)e oxide (A) +water (D)ethylene glycol (G)e oxide (A) + ethylene glycol (G diethylene

(H)

ol (A) + isobutene (B)2-tert-butyl-p-cresol (C)

2-tert-Butyl-putyl-p-cresol (C) + isobutene (B)tert-butyl-p-cresol (D)lene (B) + isobutylene (B)diisobutylene (E)tylene (E) + isobutylene (B) triisobutylene (F)

(A) + cyclohexane (B) cyclohexylphenyl ethero- andp-cyclohexylp(A) + cyclohexane

rtho/Para-cyclohexylphenol (D)exylphenyl etherrtho/Para-cyclohexylphenol (D)ara-Cyclohexylphenol (D) + cyclohexanei-cyclohexylphenol (E)(A) + alpha methyl styrenertho/para-cumylphenol (C)

o- andp-cumylphenol

/para-Cumylphenol (C)Dicumylphenol (D)a methyl styrene (B)dimethyl styrene (E)

l/n-butanol/isopropyl alcohol/2-ethyl hexanol.

i-batch RD conguration shown in Fig. 2b, called asi-batch reactive rectication (HSRR), may be used.r is charged with A, and B is fed continuously toat an appropriate location. Further, by using a

mn conguration with reactive and non reactiveed at the appropriate locations one can maintainlow concentration of product (C) in the reactive

ppress the side reaction. Hence, the non-reactivee the purpose of improving the selectivity of aduct. Similarly if A is the least volatile component,on shown in Fig. 2c, called as hybrid semi-batchipping (HSRS) may be used. Hence, we introducengurations for multiple reactants-single productz. hybrid semi-batch reactive rectication (HSRR) and-batch reactive stripping (HSRS) with continuous feede reactants that participates in both the reactionssemi-batch reactors, these congurations are alsooffering selectivity close to 100%, if perfect sepa-ssible. The difference between semi-batch reactoror HSRS) is the removal of desired product from thee. Hence, HSRR (orHSRS) is expected to offer better

the conventional semi-batch reactor under other-r conditions. In the following section, we presentnalysis of the performances of these RD units anddesign methodology. First we formulate generalHSRR and HSRS. The models are then solved for

ical case and the performance is analyzed throughof different parameters to devise a design method-nceptual level.

2.1. Hyand hybrid

Hybrid semireactive rectnon-reactivreactants ttinuously fre-boiler isin the reacnon-reactivreactive strreactive zoreactive zo(Agarwal etand reactivebelow:

(a) In theis consnon-reathe reare-boilelibrium

(b) The mois assuto convbased oHSRR, a(s) Reactionconditions/volatility

order

References

Absence of catalystDiGuilio andMcKinney(2000)

olIon exchange resin,3045/B>A>E> F>C>D

Santacesariaet al. (1988)

lIon exchange resin,4580 C/B>A>C>D>E

Chakrabartiand Sharma(1992b)Ion exchange resin,60100 C/B>A>C>D>E

Chaudari andSharma(1991)

brid semi-batch reactive rectication (HSRR)semi-batch reactive stripping (HSRS)

-batch reactive rectication (Fig. 2b) is similar to theication unit (Fig. 1a and c) except that HSRR hase stages below the reactive zone and one of thehat is participating in both the reactions is con-ed to the column at an appropriate location. Thenon-reactive and the reaction takes place only

tive zone placed between the condenser and thee zone. On the other hand, in hybrid semi-batchipping (HSRS), condenser is non-reactive and thene is placed between the reboiler and the non-ne. The assumptions made in our earlier workal., 2008a,b),while developing the reactive condenserre-boiler models apply here as well and are given

reactive rectication (HSRR) model, condenseridered as a total condenser and re-boiler as active equilibrium stage. On the other hand, inctive stripping (HSRS) model, re-boiler is a totalr and the condenser acts as a non-reactive equi-stage.lar holdup of a stage on which reaction takes placemed to be negligible. This assumption enables useniently calculate the conversion and selectivityn the holdup in the reboiler alone, in the case ofnd condenser alone, in the case of HSRS.

67

(c) The extent of the reaction on the reactive stage is depen-dent oncatalystThe reaence of

Da = kr

(d) The hewith reallowsliquid model.

(e) The enstagesHSRR awith retion.

As againb), two addistages andmodels. Folshown in F

(a) Overall(stage 1

1H1

d(Hd

d(xi,1)d

where d(b) Overall

(for j=m

F + Lj+1

FxF + L

+k=Rk=1

Assum

F Lj +

FxF + L

+k=Rk=1

Assumget

Hjcat =

idin

lj +

f + lj

= 0

ere ferallge (j

+ lj+

1xi,j+

= 0

ere lthele to

2 to perallge (t

Im +

= I

cond

= x

ng wlid fo

corbrid

Cacas

he rereaction.or baeterC;B

he se

2 is provetratenc

eedn. ThtiontierimaDamkhler number (Da) which is based on theweight and not on the volume or molar holdup.

ction takes place in liquid phase and in the pres-catalyst only.

efHTcat

V(3)

at duty is adjusted such that the vapor ow ratespect to time is constant for the HSRR model. Thisus to club the vapor ow rate in Da. Similarly theow rate is assumed to be constant for the HSRS

ergy balances on both reactive and non-reactiveare ignored for simplicity. The liquid ow rate innd vapor ow rate in HSRS, however, may changespect to time in the case of a non-equimolar reac-

st reactive rectication and stripping (Fig. 1a andtional parameters, viz., the number of non-reactivethe feed ow rate are introduced in the presentlowing are the model equations applicable to HSRRig. 2b.

and component material balances on the re-boiler)

1) = 1 + L2 + FV1

(4)

= (xi,1 yi,1) +L2 + F

V1(xi,2 xi,1)

for i = 1,2, . . . ,NC 1 (5)

= (V1/H1) dt.and component material balances on feed stage)

Lj + Vj1 Vj +k=Rk=1

i,krk(xj)Hjcat = 0 (6)

j+1xi,j+1 Ljxi,j + Vj1yi,j1 Vjyi,j

i,krk(xj)Hjcat = 0 for i = 1,2, . . . ,NC 1 (7)

ing Vj1 =Vj =V, we get

Lj+1 +k=Rk=1

i,krk(xj)Hjcat = 0 (8)

j+1xi,j+1 Ljxi,j + V(yi,j1 yi,j)

i,krk(xj)Hjcat = 0 for i = 1,2, . . . ,NC 1 (9)

ing catalyst loading on each stage to be same, we

HTcatN 1

Div

fj

fjx

wh(c) Ov

sta

lj

lj+

whAllcabj=

(d) Ovsta

1 =

yi,m

As

yi,m

Alovathehy

2.1.1.In thisthan tbatchconditmodeparamA+BFrom tof k1/kTo impconcenzone. Huous fcolumcentratall recbe ming by V, we get

lj+1 +Da

N 1

k=Rk=1

T,krk(xj) = 0 (10)

+1xi,j+1 ljxi,j + yi,j1 yi,j +Da

N 1

k=Rk=1

i,krk(xj)

for i = 1,2, . . . ,NC 1 (11)

j = F/V, lj = Lj/V, lj+1 = (F+ Lj+1)/V and Da = krefHTcat/V.and component material balances below the feed) (for j = p to m1) in non-dimensionalized form

1 +Da

N 1

k=Rk=1

T,krk(xj) = 0 (12)

1 ljxi,j + yi,j1 yi,j +Da

N 1

k=Rk=1

i,krk(xj)

for i = 1,2, . . . ,NC 1 (13)

j = Lj/V, lj+1 = Lj+1/V and Da = krefHTcat/V.above equations without reaction term are appli-the stages in the non-reactive section for stages.and component material balance on the m+1thotal condenser)

D

V(14)

mxi,m +Dxi,DVm

(15)

enser is the total condenser, we can write

i,m = xi,D (16)

ith these equations, the VLE equations are alsor each stage except the total condenser. Similarlyresponding model equations can be written forsemi-batch reactive stripping (HSRS).

se 1: volatility order C, D

68

Fig. 3 Comsimilar conloading=5

The relativsimulation

2.1.2. EffFrom the meters whichi.e. feed locreactive staduty). Fig.case that cocontinuoustillate ratenon-reactivHSRR and ctime=0 toIt can be setive zone Hrange. Anoobtained inSBR. For exin SBR in thin an increOn the othethereby slo

2.1.2.1. Feeics the chvolatility ofis less volatlocation mwise similabe fed at thvolatile theexpected, i100% convereboiler to

2.1.2.2. Effow rate oreactive stoperating ccentrationchange in t

Effect of feed location on selectivity of C for numbertive stages=1; FB =5mol/h; total number of stages=3;stage=0.8.

Effect of FB on selectivity parameter for number ofe stages=1; Da per stage=0.4.

roduct (D). Conditions thus favor the second reactiony resulting in a decrease in selectivity of C as shown6. Hence, to obtain a better selectivity under similaring conditions, lower feed rate of B is recommended.results are similar to that recommended in the case oftional semi-batch reactor (SBR). However, as explainedparison of performance of SRR with SBR underditions of FB =5mol/h; total catalyst0 gms; total batch time=10h.

e volatility of A, B and C with respect to D used fors for this case are 4.4, 2.5 and 1.6, respectively.

ect of various parameters on yield vs conversionodel equations, we have six independent param-inuence the selectivity of the desired product,

ation and ow rate, location and number of non-ges, Da per stage and vapor ow rate (i.e. reboiler3 shows the conversion vs yield plot for a baserresponds to semi-batch reactive recticationwithfeed of B under total reux condition (i.e. dis-

= 0), and with one reactive stage and without ae section. The plot is obtained by solving bothonventional semi-batch reactor (SBR) models fromthe time till 100% conversion is obtained in HSRR.en that due to product (C) removal from the reac-SRR outperforms SBR over the entire conversionther important observation is that the conversionHSRR in a given time is also higher than that in

ample HSRR conversion in 15h is 100% against 92%e same period. This is because removal of C resultsase in the concentration of A in the reactive zone.

Fig. 4of reacDa per

Fig. 5reactiv

sired ptherebin Fig.operatTheseconvenr hand, accumulation of C in SBR results in dilutionwing down the reaction as time proceeds.

d location. For the givenelementary reactionkinet-oice of feed location primarily depends on theBwith respect to A. Fig. 4 shows the results when Bile than A. The selectivity of C increases if the feedoves from reboiler to the condenser under other-r conditions. SinceB is less volatile thanA, it shoulde top of the reactive zone. Similarly if B is moren it should be fed below the reactive zone. Also, ast was observed that the quantity of B required forrsion of A reduces as B feed location shifts from

condenser.

ect of feed ow rate. Fig. 5 shows the effect of feedf B on selectivity parameter, calculated on the

age, for different conversion levels. Under givenonditions, by increasing the feed ow rate of B, con-of B on the reactive stage increases with negligiblehe concentration of C leading to formation of unde-

later due tothe reactiv

Fig. 6 Effestages=1;simultaneous removal of the desired product frome zone, the feed ow rate of B in absolute terms,

ct of FB on selectivity of C for number of reactiveDa per stage=0.4.

69

Fig. 7 Effect of location of non-reactive stage on reactivestage on selectivity parameter for FB =5mol/h, Da perstage=0.4.

required to attain same yield is much less than that in SBR. Inotherwordin a given t

2.1.2.3. Effreactive zoseparate thtive zone. Ivolatile thanon-reactivversion levezone, concenegligible cstage leadihence a detive stagesthe reactiveplacing nonserves theintermediaplacing nonimprove th

2.1.2.4. EffFig. 8, as the

Fig. 8 Effestages on sFB =5mol/h

Fig. 9 Effeparameterper stage=

Effect of number of reactive stages on selectivity of

B =5mol/h; Da=0.4; conversion of A=50%.

he desired product increases up to a particular point foren feed ow rate andDaper stage. Thereafterwith a fur-crease in the number of non-reactive stages, there is non the selectivity of the desired product. This is because

vity parameter, which depends on volatility of compo-and kinetic parameters, becomes almost constant overs, for the same feedow rate of B, the yield obtainedime period is higher in SBRD than that in SBR.

ect of the location of non-reactive zone. The non-ne may be introduced in the column to effectivelye desired product C and keep it away from the reac-t should be noted that C in the present case is lessn A and B. Fig. 7 shows the effect of location ofe stage(s) on selectivity parameter at different con-ls. By placing non-reactive zone above the reactiventration of C for a given conversion increases withhange in the concentration of B on the reactiveng to formation of the undesired product D, andcrease in selectivity of C. By placing non reac-below the reactive stage, amount of C present onstage decreases to give higher selectivity. Hence,-reactive stages below the reactive zone for HSRRpurpose of improving the selectivity of a desiredte product (C). Similarly in the case of HSRS by-reactive stages above the reactive zone one can

e selectivity of a desired intermediate product (C).

ect of number of non-reactive stages. As shown innumber of non-reactive stages increases, selectiv-

Fig. 10C for F

ity of tthe givther ineffect oselectinentsct of non-reactive stages below the reactiveelectivity of C for number of reactive stages=1;; Da per stage=0.4.

the entire cpositions ofurther cha

2.1.2.5. Effreaction isentire reactcentrationis mainly dmole fractitop of the rthat the totthe numbeside producunder theow rate othe area unincreases. Ttive stagesct of non-reactive stages on selectivityfor number of reactive stages=1; FB =5 mol/h; Da0.4.onversion range as shown in Fig. 9, i.e. the com-f A and C on the reactive stages are insensitive tonge in number of non-reactive stages.

ect of number of reactive stages. The extent of sideproportional to

wo

xAxC dW evaluated over theive zone. If one can maintain relatively large con-ofA in the reactive zone (xA 1) then the selectivityetermined by

wo

xC dW. Fig. 10 shows the plot ofon of C vs. catalyst loading as one travels from theeactive zone to the bottom of it. It should be notedal catalyst loading in each case is constant and onlyr of reactive stages has been varied. The amount oft formed in each case is proportional to the areacorresponding prole in Fig. 10. For a given feedf B, by increasing the number of reactive stages,der the curve decreases and the selectivity of Chis is because, with increase in number of reac-and for the same catalyst loading, the distillation

70

Fig. 11 Ef mbloading=5 reas

effect is enhthe reactive

2.1.2.6. Effcan be redreboiler duincreasingcatalyst loarate of B, unow rate pzone makinin Fig. 11ative stage (multiple reselectivity.

2.1.2.7. Disstudy is peFig. 12 showand 5th rearelatively his observedtively bettethe exact dan optimiz

From thcan be concsystem ofof 100% seadopting oincreasinglyst loadingincreasingthe feed o

ocespossed wsitiohenfect of vapor ow rate on composition of (a) B and (b) C for nu0gms; reactive stage: stage 2 ( indicates change due to inc

anced resulting in an efcient separation ofC fromzone.

ect of vapor ow rate (reboiler duty). Da per stage

ous prnot beis linkcompoeterswuced either by increasing the vapor ow rate (i.e.

ty) for the same catalyst loading per stage or bythe number of reactive stages for the given totalding and the vapor ow rate. For a given feed owder total reux condition an increase in the vaporulls the desired product down from the reactiveg it rich in reactant A and decient in C as shown

and b for a base case conguration with one reac-stage number 2). Similar behavior is observed withactive stages. This results in an increase in the

tributed feed in reactive zone. So far parametricrformed by considering single feed to the column.s the effect of distributed feed of B between 2ndctive stage. For a single feed of B to the column,igh concentration of B and C on the reactive zoneas compared to the distributed feed. Hence, a rela-r performance is seen in the latter case. Obtainingistribution of the feed to give cost effective design isation problem and is out of the scope of this paper.e foregoing discussion on the parametric studies, itluded that the attainable region for non-azeotropicinterest can be stretched to its maximum limitlectivity towards the intermediate component byne or more of the following design strategies: (1)number of reactive stages for the same total cata-, (2) increasing number of non-reactive stages, (3)

the vapor ow rate (i.e. reboiler duty), (4) decreasingw rate of B. It should be noted that, in a continu-

shows a caparametersentire convotherwise iseen that thconventionis much lesin HSRR is mtants due t

Fig. 12 EfselectivityFB =7mol/her of reactive stages=1; FB =5mol/h; total catalyste in vapor ow rate).

s, the last option, i.e. reduced ow rate of B mayible to implement in all the cases as this decisionith the desired production rate. Fig. 13 shows then proles of all the components for a set of param-conversion and selectivity are close to 100%. Fig. 14

se wherein, for the suitable design and operating, the desirable 100% selectivity is achieved over theersion range as compared with SBR operated atdentical batch times and ow rates of B. It can bee yield of B is much higher than that offered by theal SBR. The conversion in SBR over the same periods than that of HSRR. This is because the reactionuch faster as the catalyst is exposed to pure reac-

o in situ separation of product. For a case wherein,

fect of distributed feed of B on reactive stages onof C for number of reactive stages=5;; Da per stage=0.8.

71

Fig. 13 Coreactive staFB =5mol/h

volatility orthe desired

2.1.3. DesBased on tsuggestedumn for thi.e. A+Bsteps to be

1. The choof the d

Fig. 14 Coconditions

(i.e. B) which is involved in both the reactions. If the productis less vHSRS. Ifreactantis involv

2. The reacis fed inthe coluthan thevalue as

3. Adjust tthat desthat if threquiredcatalystconsider

4. Add nonand abo

nsiteasel catrovee attyiel

ds, ie-boy wio itsdesiosedinse5. Incr

totaimp

6. If thableworof rdutit ttheimpmposition prole of column for number ofges=2; number of non reactive stages=8;; Da per stage=0.04.

der is C, D>A, B, HSRS shown in Fig. 2c can offerattainable region.

ign methodologyhe parametric studies, a methodology has beento conceptually design a reactive distillation col-e reactions falling in the above class of systems,C, B+CD wherein, C is the desired product. Thefollowed are as follows:

ice betweenHSRRorHSRSdepends on the volatilityesired product (i.e. C) with respect to the reactant

mparison of SBR and SBRD under similarof FB =5mol/h; total catalyst loading=50g.

By folloachieved atgiven reactconventionditions.

The samtion system

1. MultipleA+BC

2. MultipleA+BCselectiviRD is thing onepossible

3. MultipleA+BCThe remthe perfthe equiconversconvent

In the foogydevelopreactants a

2.1.4. EstGlutaric acmonomethcation ofthe reactioolatile than the reactants then use HSRR else, usethe desired product (i.e. C) is less volatile than thethen charge the reboiler with reactant (i.e. A) thated only in the desired reaction.tant which is common in both the reactions (i.e. B)a continuous manner at an appropriate location inmn, i.e. above the reactive stages if it is less volatileother reactant (i.e. A). Set the feed ow rate to alow as possible.

he total catalyst loading (or Da) in the column suchired conversion can be obtained. It should be notede feed ow rate is high, a large catalyst loading isand a suitable combination of feed ow rate andloading may be decided based on the economications.-reactive stages below the reactive zone for HSRRve the reactive zone for HSRS until selectivity isive to the change in number of non-reactive stages.the number of reactive stages by keeping the

alyst loading (or Da) constant. The selectivity will.ainable yield is not same as the maximum achiev-d dened by the reaction stoichiometry or in otherf the selectivity is not close to 100%, then the effectiler duty may be examined. Increase in re-boilerll further expand the feasible region and stretchmaximum limit dened by the stoichiometry ofred reaction if there are no separation limitationsby azeotropes, if any.

wing this methodology, higher selectivity ishigher conversion and the feasible region for the

ion system is much larger than that offered by aal semi-batch reactor operated under similar con-

e methodology is applicable to the following reac-s:

reactants-multiple products irreversible reactions+D; A+CE+D, with C as the desired product.reactants-single product reversible reactions

; C+AD, when the objective is to increase thety of C. In this case an added advantage of usingat it shifts the equilibrium of the reaction allow-to obtain 100% conversion which is otherwise notwith SBR or any conventional reactor.reactants-multiple products reversible reactions+D, A+CE+D, with C as the desired product.oval of one of the products (D) strongly inuencesormance. The advantage of using RD is that it shiftslibrium of the reaction allowing one to obtain 100%ion which otherwise is not possible with SBR or anyional reactor.

llowing section we extend the design methodol-ed above to real reaction system involvingmultiplend single or multiple products.

erication of gluteric acidid (GA) on esterication with methanol givesyl glutarate (MMG) and water. Further, the esteri-MMG produces dimethyl glutarate (DMG). As bothns are reversible, the conversion in conventional

72

reactors is limited so it is impossible to get pure DMG fromGA. The reaction scheme falls in the category of A+BC+D,A+CE+D and is given by Eqs. (17) and (18). MMG (i.e.monobasic ester) has a wide range of applications in pharma-ceutical industry. The objective is to improve the selectivityfor MMG.

glutaric acid + methanolk1k2

monomethyl glutarate

+water (17)

monomethyl glutarate + methanolk3k4

dimethyl glutarate

+water

Thermodybinary intefrom ASPEConcentratfor the simVLE and k(Tables A1a

r1 = k1CGAC

r2 = k3CMM

S = k1C(k2CMM

This is a nois less volaThe expresthat MMGto achievedesign guidtive recticperformanusing RD inand hencethe reactiveof the reaccomparisonformance iconditions.

3. Exproduct r

In all theremoved co

Table 2 operating

Total glutaFeed ow rTemperatuTotal catalyReactive stNon-reactiConversionYield of MMTime of op

In some cases, this may be desired and is possible to imple-ment if volexamples wof multiplefrom eitherof the casecial case ofalso necessstrate B isgiven below

A + B C +

A + C E +

ld bn disry thnol (movaver, iThed theons:

xAxB

xAxC

(k1xxExD

k3 anis dtantg deeterto itioncondesiditiotion

e strifacovine) or. In te (D)he rein F

is chuous

Semal (H

is sb) usxpla

ociat, in me a

nt ishe se(18)

namics is modeled using NRTL equation withraction parameters estimated by UNIFAC modelN PROPERTY PLUS (Aspen Technology Inc., 2001).ion based kinetic model (Eqs. (19) and (20)) usedulations is taken from Hung et al. (2007). The

inetic parameters are given in the Appendix Aand A1b):

MeOH k2CMMGCwater (19)

GCMeOH k4CDMGCwater (20)

GA + (k4CDMGCwater/CMeOH)GCwater/CMeOH) + k3CMMGCMeOH

(21)

n-azeotropic system and desired product, i.e. MMGtile than glutaric acid which is fed to the reboiler.sion for the selectivity parameter (Eq. (21)) suggestsneeds to be removed from the reactive zone so asselectivity close to 100%. Based on the proposedelines it is expected that hybrid semi-batch reac-ation (HSRR), shown in Fig. 2b, would offer the bestce in terms of selectivity to MMG. The objective ofthis case is to improve the conversion of reactants

the yield by simultaneous removal of product fromzone even with close to stoichiometric mole ratios

tants. Table 2 shows the operating conditions andof SBR with SBRD. It can be seen that SBRD per-

s superior to that of SBR under otherwise similar

ample 2: SBRD with simultaneousemoval

previous cases, the product(s) formed are notntinuously from the column during the reaction.

Comparison of SBRD and SBR for the givenparameters

SBRD SBR

ric acid charged (mol) 6.36 6.36ate of methanol (mol/h) 0.051 0.051re (C) 148 80/105/145st loading (g) 150 150ages 20 ve stages (below reactive stages) 5

99.9 60/73/85G 99.9 40/52.7/64.5

eration (h) 55 55

It shoucatiocategomethaous reMoreonot C.ics anequati

r1 = k1

r2 = k3

S = xAk4

k1, k2,versionof reacforminparamtive isseparaits theof thetwo adrecticareactivrationsby remvolatilFig. 15volatilthan tshownand Bcontin

3.1.remov

HSRRP(Fig. 2tems eis assHencetem, wreactatrols tatilities are right. In this section, we consider suchherein each reaction is associated with formationproducts one of which can be separated in situtop or bottom depending on its volatility. In most

s the second reaction is the desired one. It is a spe-the series parallel reactions, and therst reaction isary for the second reaction to take place. The sub-a bifunctional molecule. The reaction of interest is:

D

D

e noted that the example of glutaric acid esteri-cussed in the earlier section does not fall in thisrough it follows the same stoichiometry. SinceB) is more volatile than water (D), the simultane-l of water from the system is not recommended.n the present case the product of interest is D andreaction is assumed to be elementary. The kinet-selectivity parameter are given by the following

k2xCxD (22)

k4xExD (23)

B + k3xC)+ k2xCxD

(24)

d k4 are assumed to be equal and unity. The con-ened as moles of reactant A consumed per moleA fed, and selectivity as moles of reactant used insired product to total moles reacted. The selectivityis dened by the Eq. (24). When the main objec-ncrease the selectivity of the desired product E,of D from the reactive stages is crucial as it lim-version of reactant A and governs the selectivityred product E. Hence for such cases, we proposenal basic RD units, viz., hybrid semi-batch reactivewith product removal (HSRRP) and hybrid semi-batchpping with product removal (HSRSP). These congu-ilitate the separation of D from the reactive stagesg desired product either from the top (if it is morefrom the bottom (if it is less volatile) as shown inhe following example, one of the products is moreand the other products C and E are less volatileactants. In such case, it is advisable to use HSRRPig. 15a. When the desired product is E, either of Aarged to the reboiler and the other reactant is fedly.

i-batch reactive rectication with productSRRP)

imilar to the hybrid semi-batch reactive recticationed for the multiple reactant-single product sys-ined in the previous section, except that HSRRPed with continuous overhead product removal.ultiple reactant- multiple product reaction sys-

dd one of the reactants continuously, the othercharged to the reboiler and the product which con-lectivity of the desired component is withdrawn

73

Fig. 15 (a val (Hstripping w

continuousof reactanton the selegiven by Eenhanced bconcentrattially chargHSRRP. All tprevious seence is in tow rate is

3.1.1. EffFrom themeters whichi.e. distillaoverall Da,tive stagesrate (i.e. rebuct (C) is dow rate aif the mainconditionsa relativelyrate undersuch as nustages andand hencebase case cous feed oThe unit htion. As cannot 100%. Istraint, whthe efcienavoid the loversion, this becausestage.

3.1.1.1. Efftive is to remthe distillaassuming cmediate co

Comparison of performance of HSRRP with SBRsimilar conditions of FB =0.1mol/h; total catalystg=60g.

the distillate stream is not pure D, some amount of thent may be lost through distillate and conversion close toay not be realized. Hence the reaction and distillation

eters should be manipulated such the distillate is pure

. Location and number of non-reactive stages. It is obvi-at by providing the non reactive zone above the reactiveserves the purpose of removing relatively pure D from) Hybrid semi-batch reactive rectication with product remoith product removal (HSRSP).

ly. As explained in the previous section, the choiceto be charged in continuousor batchmodedependsctivity parameter (S). The selectivity parameter

q. (24) indicates that the selectivity of E can bey maintaining high concentration of A and/or lowion of B in the reactive zone. Hence, reboiler is ini-ed with A and continuous feed of B is given tohe model equations and assumptions made in thection for HSRR apply here as well. The only differ-he equation for the condenser wherein, distillatenite and not equal to zero.

ect of various parameters on yield vs. conversionodel equations, we have eight independent param-inuence the selectivity of the desired product,

te rate, the location of the feed, feed ow rate,location of non-reactive zone, number of reac-

, number of non reactive stages and vapor owoiler duty). As expected, if the intermediate prod-esired, the column should operate at low feednd at high vapor ow rate. On the other hand,objective is to increase the selectivity of E, the

are reversed and the column should operate athigher feed ow rate of B and lower vapor owotherwise similar conditions. Other parametersmber of reactive stages, location of non-reactivedistillate rate mainly inuence the removal of D

Fig. 16underloadin

that ifreacta100% mparamD.

3.1.1.2ous thsectionthese effects are analyzed here. Fig. 16 shows aonversion vs. yield plot of HSRRP with continu-

f B at specic distillate rate and vapor ow rate.as single reactive stage and no non-reactive sec-be seen, in both the cases the conversion level is

n SBR the limit is imposed by the equilibrium con-ereas for HSRRP, with a limited number of stages,cy of removal of product is not good enough toss of reactant loss in the distillate. At higher con-

e yield towards E is higher than that of SBR. Thisof the simultaneous removal of B from the reactive

ect of distillate rate. As explained before the objec-ove the volatileD as the overhead product. Ideally

te ow rate should be equal to the D formed byonversion of AE only, with formation of the inter-mponent (C) to be negligible. However, it is possible

the reactivatively lessshifts in fotivity to Ebelow theotherwiseconcentratprevious seselectivityis no furthreactive stabecomes inAs the distetry of reanumber ofmum combconversionminimumcSRRP) and (b) hybrid semi-batch reactivee zone and enriching the reactive zone with rel-volatile reactants such as A. Hence, the reaction

rward direction and also an increase in the selec-is realized. By providing the non-reactive zonereactive zone, selectivity of E decreases undersimilar conditions because of the relatively highion of D on the reactive stages. As explained in thection, on increasing the non-reactive stages, theof E increases up to a point and thereafter, thereer increase with increase in the number of non-ges. This is because, the distillate composition of Dsensitive to further change in non-reactive stages.illate rate of D is xed based upon the stoichiom-ction, purity of distillate mainly depends on thenon-reactive stages and vapor ow rate. An opti-ination of these two parameters that offers desired, 100% selectivity towards the desired product andost needs to beworked out for a given problemand

74

Fig. 17 (afraction ofrate =0.1mreactive staFA =0.1mo

is out of thhere.

3.1.1.3. Effaration ofresults in amediate prthe reactivunder eachof D presenselectivityFig. 17b, thnumber ofing.

3.1.1.4. Effcase we arandnot theboth the rewould helpeffect of toin the totastages incrple 1, discuwas to impper stage w

From thparameters

Efer of1km

limitlowiberper sprodof t

be reFig. 18numbV=0.2

imumthe folor numthe upif thesectionuct to) Effect of number of reactive stages on moleD for FB =5mol/h; Da (over all) = 1; distillateol/h for conversion of A=40% and (b) effect ofges on E selectivity for k1/k3 =1, k1 =k2 =k4;l/h; Da (over all) = 1; distillate rate =0.1mol/h.

e scope of the conceptual design work presented

ect of number of reactive stages. An effective sep-common product, i.e. D, from the reactive zonen increase in the selectivity of E instead of inter-oduct C. Fig. 17a shows composition of D alonge zone for different number of stages. The areacurve quantitatively represents the adverse effectt on the reactive stages which in turn controls theof E due to equilibrium limitations. As shown ine selectivity of E increases with an increase in thereactive stages for the given total catalyst load-

ect of catalyst loading (Da per stage). In the presente interested in the product of the second reactionintermediate one. Asmentioned before, in a sense,

actions are desired and expediting the rst reactionthe second reaction indirectly. Fig. 18 shows the

tal catalyst loading on the selectivity of E. Increasel catalyst loading for a given number of reactiveeases the selectivity of E. On the contrary, in exam-ssed in the previous section, when our objectiverove selectivity for the intermediate product, Daas recommended to be as small as possible.e foregoing discussion on the effects of various, the feasible region can be stretched to its max-

of non-reacreboiler dumanner.

3.1.2. Dereactant-sincontinuousBased on thdeveloped ttion systemproduct. Th

1. Use thebottomcommoninvolvedmon procommonthe rebotions (i.e

2. The reaadded inin the ctant, coand if itzone.

3. Adjustsuch thaincrease

4. Based osion, xproductrate of E

5. Add noHSRRPincreasefect of catalyst loading on E selectivity forreactive stages=1; k1/k3 =1, k1 =k2 =k4;ol/h; FA =0.1mol/h; distillate rate =0.1mol/h.

of 100% selectivity to E by adopting one or more ofng design strategies: (1) increasing catalyst loadingof reactive stages; (2) placing non-reactive zone inection of the column, i.e. above the reactive zoneuct to be removed is most volatile or in the lowerhe column, i.e. below the reactive zone if the prod-moved is less volatile, and increasing the numbertive stages; (3) decreasing the vapor ow rate (i.e.ty), and if possible introducing B in a distributed

sign methodology for multiplegle/multiple product reaction systems withremoval of common producte parametric studies, design guidelines have beeno design a reactive distillation column for the reac-A+BC+D, A+CE+D when E is the desired

e steps to be followed are given below:

HSRRP with distillate withdrawal or HSRSP withwithdrawal depending upon the volatility of the

product with respect to the reactant which is

in both the reactions. HSRRP is used if the com-duct is more volatile and HSRSP is used when theproduct is less volatile than the reactants. Charge

iler with reactant that is involved in both the reac-. A).ctant which is involved in the single reaction isa continuous manner at an appropriate location

olumn. If it is more volatile than the other reac-ntinuous feed is introduced below reactive zoneis less volatile, the feed is given above the reactive

the total catalyst loading (or Da) in the columnt desired conversion can be obtained. Conversions with the catalyst loading.n the reaction stoichiometry and desired conver-the distillate rate/bottom rate to remove commonfrom top/bottom. This ow rate is set equal to theformed.

n-reactive stages above the reactive stages forand below reactive stages for HSRSP until thein selectivity is independent of number of non-

75

Fig. 19 Coreactive staFB =0.1moregion indi

reactiveand dist

6. Manipulber of nproductvapor oof non-r

7. Increasetotal cat

By folloachieved atgiven reacttion prolein which coby adjustintioned beforeactor andbe seen thters, the deconversionvolatility orgive the de

In the foogy to realsingle or m

3.1.3. EstEthylene gmonoacetaestericatio

Comparison of attainable region for E betweensemi-batch reactive distillation with productal an

). AsventAEGry of8). Dgs ane th

lene

tic a

mposition prole of column for number ofges=3; number of non-reactive stages=14;l/h; Da=1.4; distillate rate =0.2mol/h (shaded

Fig. 20hybridremov

(DAEGin conpure Mcategoand (1coatinimprov

diethy

+acecates the reactive zone).

stages for a given vapor ow rate (i.e. reboiler duty)illate/bottom rate.ate the vapor ow rate (i.e. reboiler duty) and num-on-reactive stages to obtain the pure commonas a distillate or as a bottom product. Reduction inw rate (i.e. reboiler duty) and increase in numbereactive stages lead to increase in selectivity.the number of reactive stages and/or increase thealyst loading. The feasible region will expand.

wing this methodology, higher selectivity ishigher conversion and the feasible region for theion system expands. Fig. 19 shows the composi-s of components obtained by simulation for a casenversion and selectivity close to 100% are achievedg the parameters using the design guidelines men-re. Fig. 20 shows a comparison between semi-batchSBRD under otherwise similar conditions. It can

at for the suitable design and operating parame-sirable 100% selectivity is achieved over the entirerange using reactive distillation. For the case if theder is C or D>B or A, HSRSP shown in Fig. 12b cansired attainable region.llowing section we extend this design methodol-reaction systems involving multiple reactants andultiple products.

erication of ethylene glycol with acetic acidlycol (EG) on esterication with acetic acid giveste ethylene glycol (MAEG) and water. Further, then of MAEG produces diacetate of ethylene glycol

monoaceta

+acetic a

The activitis taken frtion paramAppendix A

r1 = k1aDEG

r2 = k3aMAE

S = aAA(kawater(k

Table 3 operating

Total aceticFeed ow rDistillate raWater concTemperatuTotal catalyReactive stNon-reacti

stages)% Conversi% Yield of DTime of opd semi-batch reactor.

both the reactions are reversible, the conversionional reactors is limited so it is impossible to getor DAEG from EG. The reaction scheme falls in theA+BC+D, A+CE+D and is given by Eqs. (17)AEG and MAEG are used in thermoplastic acrylicd specialty chemical industry. The objective is toe selectivity for DAEG.

glycol

cidk1k2

monoacetate ethylene glycol + water (25)

te ethyleneglycol

cidk3k4

diacetate ethyleneglycol + water (26)

y based kinetic model used for the simulationsom Schmid et al. (2008). The UNIQUAC interac-eters and the kinetic parameters are given in the(Tables A2a and A2b):

aAA k2aMAEGawater (27)

GaAA k4aDAEGawater (28)1aDEG + k3CMAEG)2CMAEG + k4CDAEG)

(29)

Comparison of SBRD and SBR for the givenparameters

SBRD (HSRRP) SBR

acid charged (mol) 40 40ate of ethylene glycol (mol/h) 0.4 0.4te (mol/h) 0.806 entration in distillate (% mol) 0.992re (C) 116 116st loading (kg) 5 5ages 10 ve stages (above reactive 60

on of acetic acid 95.4 43AEG 95.4 17.3

eration (h) 5 5

76

This is a nproduct inacid whicheter givenselectivity,Hence, basethat hybridous removgive the bementionedwith excesto improveby simultaneven with cHowever, tlarge numbating condmances.Fig. 21 Continuous versions of (a) SHRR, (b) SHRS,

on-azeotropic system and waterthe commonboth the reactions, is more volatile than aceticis charged to the reboiler. From selectivity param-in Eq. (29), to achieve signicant conversion andwater needs to be removed from the reactive zone.d on the proposed design guidelines, it is expectedsemi-batch reactive rectication with continu-

al of product (HSRRP), as shown in Fig. 15a, willst performance in terms of selectivity to DAEG. Asbefore, SBR is also likely to offer signicant yield

s acetic acid. Objective of using RD in this case isthe conversion of reactants and hence the yieldeous removal of products from the reactive zonelose to stiochiometric mole ratios of the reactants.he separation of water and acetic acid requireser of non-reactive stages. Table 3 shows the oper-itions and comparison of SBR and SBRD perfor-

4. Co

In this sechybrid modbatch/semithat offersous modelthe entireilar guideliversions. Icharged tobelow or abof the reacremoved inapplied tounits.

The genreactive rec(c) SHRRP and (d) SHRSP.

ntinuous models

tion, we describe the continuous versions of theels analyzed before. It is well known that for every-batch operation there is a continuous counterpartsimilar performance. Fig. 21 shows the continu-

s capable of giving close to 100% selectivity overrange of conversion if the system is ideal. Sim-nes can be used for design of these continuousn the continuous version, the reactant which isthe reboiler of HSRR or HSRRP is fed continuouslyove the reactive zone depending on the volatility

tant, and also the bottom product is continuouslythe continuous version. The same logic may be

derive the continuous version of HSRS and HSRSP

eralized model equations for continuous hybridtication are given below:

77

(a) Overall and component material balances on condenser(for j = 1

lj +D

V

ljxi,j1+

where l(b) Overall

jth Stag

lj1 lj

lj1xi,j

for i=1(c) For the

If the sthe staremove

(d) Overallj=N):

lj1 V

lj1xi,j

Similaruous hybrishows the cand for 100ous hybriddiscussed i

5. Ex

In our earliesingle reactreactant anhas been coreactant anobtained fomultiple reby the azeouct(s). In subased on thsystem ofdesign metical azeotro

To demosame illustthere existsdesired prostants andthe Table 4volatile thasemi-batch

Composition prole of column for number ofe stages=5; number of non reactive stages=8;ol/h; FB =5mol/h; Da per stage=0.016.

4a Binary interaction parameters (Aspennology Inc.)

ponent A B C D

0 43.45 38 5537.36 0 10 54.5130 240 0 60

69 60 2 0

eux condition. The choice of reactant to be chargedtinuous mode or batch mode to HSRR depends onlectivity parameter (S). For the liquid phase reactionrest A+BC; B+CD, selectivity parameter is equal/k2xC up to azeotropic composition, At higher conver-e selectivity parameter is nearly equal to k1xAAZ/k2xCAZ

4b Extended Antoines constants (Aspennology Inc.)

ponent A B C D

57 104.65 103 56.19640.44 6995.5 12591.9 9.64E+030 0 0.00E+00 0.00E+000 0 0.00E+00 0.00E+006.1935 12.702 12.7 6.19351.47E06 1.24E05 3.81E06 1.47E062 2 2.00E+00 2.00E+00) of Fig. 11c gives:

1= 0 (30)

(D

V

)xi,j1 yi,1 = 0 for i=1,2, . . . ,NC 1 (31)

j = Lj/V.and component material balances on feed stagee (for j=Nf1 and Nf2)

+ Fj +Da

N 1

Rk=1

T,krk(xj) = 0 (32)

1 ljxi,j+Fjxfi,j + yi,j+1 yi,j+Da

N 1

Rk=1

i,krk(xj)=0

(33)

, 2, . . ., NC1.stages other than feed stages (j /= Nf1 and Nf2):tage is reactive, Fj =0 in Eqs. (32) and (33) and ifge is non-reactive, both Fj and reaction term ared from Eqs. (32) and (33).and Component Material Balance on reboiler (for

B

1 = 0 (34)

1 (

B

V

)xi,j yi,j = 0 for i = 1,2, . . . ,NC 1 (35)

to continuous hybrid reactive rectication, contin-d reactive stripping can also be modeled. Fig. 22olumn proles of components for 100% selectivity% conversion of reactants obtained in a continu-reactive rectication unit for the hypothetical casen Example 2.

ample 3: systems with azeotropes

r work (Agarwal et al., 2008b), we showed that for aant system the presence of azeotrope between thed the desired product limits the selectivity and itncluded that when there is no azeotrope betweend desired product then 100% selectivity can ber the entire range of conversion. Similarly, in theactant systems the selectivity may be inuencedtrope between reactant(s) and the desired prod-ch case a complex conguration may be designede complete knowledge of the VLE behavior of the

interest. In the following section, we extend thehodology developed for ideal system to a hypothet-pic case.nstrate the effect of azeotrope, we consider the

rative example 1 discussed before. In this example,a minimum boiling azeotrope at xC =0.5 between

duct (C) and reactant (A). Extended antoines con-the binary interaction parameters are given in

. In this case, both the products C and D are lessn the reactants A and B. Hence, we choose hybridreactive rectication (HSRR) conguration under

Fig. 22reactivFB =5m

TableTech

Com

ABCD

total rin conthe seof inteto k1xAsion th

TableTech

Com

1234567

78

Fig. 23 Coconditions

because thazeotropicthe ratio ofratio k1/k2.are same aneed to macentrationcharged wibatch reactdeveloped fthe performmance of abetween Bperformanhigher conlower thancase is enriof D, thus lthat the praffects thecompositiowhen azeothe reactanselectivitysition.

In the mazeotropeinvolved inproduct anable selectsections wazeotropes

5.1. Case 1: azeotrope between desired product andreactant

5.1.1. Transesterication of cyclohexyl acetateCyclohexyl acetate (CHA) on transesterication with n-butanol (BuOH) gives cyclohexanol (CH) and butyl acetate(BA). Cyclohexanol is used in the manufacture of plasticizersand adipic acid. Cyclohexanol reacts with n-butanol in a sidereaction to give butyl cyclohexyl ether (BCHE) and water. Inconventional reactors, it is impossible to obtain pure CH fromCHA. The reaction scheme falls in the category ofA+BC+D,A+CE+ F and is given by Eqs. (36) and (37):

exyl

yclo

exan

utyl

odyninteSPE

ntrat38) a,b). Tn th

CCHA

CCHC

CCHACCH

s azele 5.oilinexylnt anxpec), shos of

f sele

Table 5 2001)

Temperat

84.6491.2891.692.0693.498.18116.92159.57mparison of SBR and HSRR under similarof FA =5mol/h; total catalyst loading=50g.

e composition in the reactive zone is equal to thecomposition. Hence, the selectivity ofC depends onxA/xC or xAAZ/xCAZ in the reactive zone for a givenThe denitions of Conversion, selectivity and yield

s that in example 1. To improve selectivity of C, weintain either high concentration of A or less con-of C in the reactive zone. Hence, reboiler is initiallyth A and continuous feed of B is given to the semi-ive rectication column. The design methodologyor HSRR in the previous section is used to compareance with that of SBR. Fig. 23 shows the perfor-HSRR against a SBR when azeotrope is present

and C. It can be seen that there is a decrease in thece of HSRR beyond the azeotropic composition. Atversions, the selectivity obtained in HSRR is muchthat in SBR. This is because reactive zone is in suchchedwith respect to productC leading to formationowering the selectivity. From Fig. 23, it can be seenesence of azeotrope between C and A drasticallyperformance of HSRR when compared to SBR atn higher than the azeotropic composition. Hence,trope is present between the desired product andt that is involved in single reaction, it limits the

in RD which in turn depends on azeotropic compo-

cycloh

k1c

cycloh

k1b

Thermbinaryfrom AConce(Eqs. ((1992agiven i

r1 = k1

r2 = k2

S = k1k2

Variouin Tabdiate bcyclohreactait is e(HSRRin termsion oultiple reactant systems, only the presence ofbetween desired product and the reactant that is

single reaction limits the selectivity of desiredd all other azeotropes do not inuence the attain-ivity of the desired product. In the followinge analyze real examples involving formation of.

with cyclohbelow theselectivitybutyl cyclovant here. H(reactant) ain HSRR as

Azeotropes present in the transesterication of cyclohexyl aceta

ure (C) CHA BuOH CH

0 0 00.0648 0.1488 00 0.2378 00.1463 0 00 0 00 0 0.08840 0.782 00.2281 0 0.7719acetate + butanol

hexanol + butyl acetate (36)

ol + butanol

cyclohexyl ether + water (37)

amics is modeled using UNIQUAC equation withraction parameters determined by UNIFAC methodN PROPERTY PLUS (Aspen Technology Inc., 2001).ion based kinetic model used for the simulationsnd (39)) is taken from Chakrabarti and Sharmahe binary interaction and kinetic parameters aree Appendix A (Tables A3a and A3b):

CBuOH (38)

BuOH (39)

(40)

otropes present in this reaction system are givenThe desired product (i.e. cyclohexanol) is interme-g compared to the reactants, i.e. n-butanol andacetate. Its volatility is close to the less volatiled hence, based on the proposed design guidelines,ted that hybrid semi-batch reactive recticationwn in Fig. 2b, would offer the best performanceselectivity for cyclohexanol. Based on the expres-ctivity parameter (Eq. (40)), the reboiler is chargedexyl acetate, and n-butanol is fed continuously

reactive stages. As the objective is to improve thefor cyclohexanol, the azeotropes between water orhexyl ether with any other component are not rele-owever, the azeotrope between cyclohexyl acetatend cyclohexanol (desired product) is of relevancethe reactive zone consists of cyclohexanol along

te reaction system (Aspen Technology Inc.,

BA WATER BCHE

0.2674 0.7326 00 0.7865 00 0.7622 00 0.8537 00 0.8938 0.10620 0.9116 00.218 0 00 0 0

79

Fig. 24 Cotransesteri

Table 6 operating

Total CH chFeed ow rTemperatuTotal catalyReactive stNon-reactiReboiler duYield of CHTime of op

with cyclohand SBR forpossible tobetween th

5.2. Ca

5.2.1. Cromethyl ethyThe aldolketone (MEdrates to th(MPO). Howa side reaalso possibcrotanaldehfalls in theis to improintermedia

acetaldehy

k13-me

MPO + acet

2 acetaldeh

Kinetic mLangmuirH

7 Comparison of HSRR and SBR for the giventing parameters

HSRR SBR

MEK charged (mol) 6 6owrate of acetaldehyde (mol/h) 0.1 0.1

erature (C) 7075 75, 90 and 105catalyst loading (g) 416 416ive stages 28 eactive stages (below reactivees)

6

ler duty (kW) 2 version 99.9 85/90/90ld of MPO 97 54/58/60

60.5 60.5

is used for the simulations. The kinetic parameters aren the Appendix A (Tables A4a and A4b):

k1aA

+ KW

k2aA

+ KW

k3a

+ KW

xHeav

f thmparison of SBR and SBRD forcation of cyclohexyl acetate.

Comparison of SBRD and SBR for the givenparameters

SBRD SBR

arged (mol) 3.2 3.2ate of 2-butanol (mol/h) 0.1 0.1re (C) 120140 130st loading (g) 560 560ages 14 ve stages (below reactive stages) 4 ty (kW) 4

96 80eration (h) 33 33

Tableopera

TotalFeed TempTotalReactNon-r

stagReboi% Con% YieTime

(2007),given i

r1 =(1

r2 =(1

r3 =(1

S =k2

VLE oexyl acetate. Fig. 24 shows the comparison of HSRRthe operating conditions given in Table 6. It is notobtain 100% selectivity due to azeotrope formede reactant and the desired product.

se 2: azeotrope between reactant and product

ss aldol condensation of acetaldehyde andl ketonecondensation of acetaldehyde with methyl ethylK) yields an unstable aldol which rapidly dehy-e desired product, i.e. 3-methyl-3-pentene-2-oneever, MPO further reacts with acetaldehyde in

ction to give undesired heavier products. It isle that acetaldehyde on self-condensation givesyde. The reaction scheme given by Eqs. (41)(43),

category of A+BC+D, A+CE+D, 2AE. Aimve selectivity toward MPO (C) as it is an importantte in perfumery industry:

de + MEK

thyl-3-pentene-2-one (MPO) + water (41)

aldehydek2heavier products + water (42)

ydek3crotonaldehyde + water (43)

odel (Eqs. (44)(46)) based on a modiedinselwood mechanism proposed by Mahajan

water formcase, is nreactant aand water >MPO is leslines given(HSRR)woutoward MPO(Eq. (47)), Mduced contis to improvdiate boilinfor an incrHowever, itwater is noaffect the

Table 8 operating

Total methFeed ow r

aqueousTemperatuTotal catalyReactive stNon-reacti

stages)Reboiler du% Conversi% Yield of DTime of opaB

aW )2

(44)

aC

aW )2

(45)

2A

aW )2

(46)

k1xMEKMEK

iesHeavies + k3xAceAce(47)

e reaction system is non-ideal, i.e. MEK andbinary azeotrope at xw =0.35. Water, in this

ot a desired product. The volatility order ofnd products is actaldehyde>azeotrope of MEKMEK>water >MPO>crotanaldehyde>heavies. Ass volatile than the reactants, as per the guide-before, hybrid semi-batch reactive rectication

ld offer the best performance in termsof selectivity. Based on the expression of selectivity parameterEK is fed to the reboiler and acetaldehyde is intro-inuously below the reactive stages. As the objectivee the selectivity for MPO, the presence of interme-g azeotrope between MEK and water is responsibleease in water concentration in the reactive zone.does not inuence the performance of HSRR ast involved in any of the reactions and does not

reaction rates signicantly though it inhibits theComparison of SBRD and SBR for the givenparameters

SBRD SBR

anol charged (mol) 30 30ate of glyoxal (40%, w/w,solution) (g/h)

0.105 0.105

re (C) 6566 65st loading (g) 475 475ages 19 ve stages (below reactive 9

ty (kW) 3 on 99.9 45AG 100 10

eration (h) 12 12

80

reaction to some extent given by the Eqs. (44)(46). It is possi-ble to obtaiconditionshyde. HSRRbefore. HSRoperating cazeotrope dof MPO. It sa choice ofSBR at threcomparison

5.3. Ca

5.3.1. AcGlyoxal (G)of glyoxalproduces dreversible, iventional rof A+BCand DMG (isubstitutesThe objecti

glyoxal + m

monoaceta

Vaporliqution with bmethod fro2001). Concused for th(1997). Thein the Appe

r1 = k1CGC2M

r2 = k3CMAG

S = C2MeOH

Cwater(k

VLE of theglyoxal andity order oof MAG anproduct inwhich is ffrom the reand selectiexpected th(Fig. 2b) wilto DAG. Thindicates thglyoxal shoAs the objeence of inteof glyoxal aHSRRas thetioned befo

excess methanol. The advantage of using RD in this case isrove the conversion of reactants and hence the yieldultaneous removal of products from the reactive zoneith close to stoichiometric mole ratios of the reactants.shows the operating conditions and comparison of the

mances of SBR and SBRD.

Conclusion

ies or series-parallel reactions involving multiple reac-when no azeotrope is present in between reactant andd product, 100% selectivity can be obtained for therange of conversion using reactive distillation. Semi-reactive distillation offers better performance than thetional semi-batch reactor under identical operating

ions. However, when an azeotrope is present betweennt and the desired product, there is a limit on thevity and the performance of RD depends on azeotropicsition. Based on the volatility order of reactants andts, the design guidelines have been developed to obtainproved selectivity for the desired product using variousurations of RD.

ndix A. Binary interaction parametersinetic parameters used as input forus simulations

A1a, A1b, A2a, A2b, A3a, A3b, A4a, A4b, A5a and A5b.

A1a Binary interaction parameters (Aspennology Inc.)

ponent GA MeOH MMG DMG Water

0 1322.397 381.858 901.943 319.112708.714 0 306.754 300.238 838.5936

A1b007)

ete

/kmol/kmol/kmol/kmol

A2amete

ponen higher yields toward MPO in SBR under extremesuch as very small feed ow rate of acetalde-is designed based on the guidelines mentioned

R gives a better performance than SBR for the givenonditions mentioned in Table 7. The presence ofoes not inuence the attainable yields/selectivityhould be noted that unlike HSRR, in SBR, one hasoperating temperature and hence the results one different temperatures are reported for better.

se 3: azeotrope between products

etalization of glyoxalon acetalization with methanol gives monoacetal(MAG) and water. Further, MAG on acetalizationiacetal of glyoxal (DAG). As both the reactions aret is impossible to get pure DAG from glyoxal in con-eactors. The reaction scheme falls in the category+D, A+CE+D and is given by eqs (48)(49). MAG.e. mono and diactal of glyoxal) nd applications asof glyoxal in the various non-aqueous reactions.ve is to improve the selectivity to DAG:

ethanolk1k2

monoacetal of glyoxal + water (48)

l + methanolk3k4

diacetal of glyoxal + water (49)

id equilibrium is modeled using UNIQUAC equa-inary interaction parameters estimated by UNIFACm ASPEN PROPERTY PLUS (Aspen Technology Inc.,entration based kinetic model (Eqs. (50) and (51))e simulations is taken from Mahajani and Sharmabinary interaction andkinetic parameters are givenndix A (Tables A5a and A5b):

eOH k2CMAGCwater (50)

C2MeOH k4CDAGCwater (51)

(k1CG + k3CMAG)2CMAG + k4CDAG)

(52)

reaction system is non-ideal, i.e. monoacetal ofwater form azeotrope at xw =0.975. The volatil-

f reactants and products is methanol > azeotroped water >water >MAG>DAG. Water, the commonboth the reactions, is less volatile than methanoled to the reboiler. Water needs to be removedactive zone so as to achieve signicant conversionvity. Based on the proposed design guidelines it isat hybrid semi-batch reactive rectication (HSRR)l offer the best performance in terms of selectivitye expression for the selectivity parameter (Eq. (52))at reboiler should be charged with methanol anduld be continuously fed above the reactive stages.ctive is to improve the selectivity for DAG, the pres-rmediate boiling azeotrope between mononacetalnd water does not inuence the performance ofazeotrope (xw =0.975) is not relevanthere.Asmen-re, SBR is also likely to offer signicant yield with

to impby simeven wTable 8perfor

6.

For sertants,desireentirebatchconvenconditreactaselecticompoproducthe imcong

Appeand kvario

Tables

TableTech

Com

GAMeOHMMGDMGWater

Tableal., 2

Param

k10k20k30k40E1 (kJE2 (kJE3 (kJE4 (kJ

Tablepara

Com

DEGAAMAEGDAEGWater1029.935 17.448 0 422.863 129.646459.037 334.455 215.938 0 48.0261010.6 1347.527 129.646 1698.559 0

Kinetic parameters for DMG system (Hung et

r Value

3.346E76.416E33.487E51.985E8

) 36726.26) 702.53) 16978.85) 49672.82

Temperature dependent binary interactionrs (cal/mol) (Schmid et al., 2008)

nt DEG AA MAEG DAEG Water

0 214.98 12.65 141.62 606.48288.43 0 219.36 521.77 300.68236.84 224.50 0 278.52 567.78360.94 1101.82 156.03 0 812.11722.44 180.16 344.69 113.52 0

81

Table A2b Kinetic parameters for MG system (Schmid et al., 2008)

Parameter Value

k10 (mol/(g s)) 539.14k20 (mol/(g s)) 114.41k30 (mol/(g s)) 156.43k40 (mol/(g s)) 95.68E1 (kJ/mol) 39.84E2 (kJ/mol) 39.85E3 (kJ/mol) 39.9E4 (kJ/mol) 39.86

Table A3a Binary interaction parameters (Aspen Technology Inc.)

Component CHA BuOH BA CH BCHE Water

CHA 0 146.63 82.01 111.08 95.82 393.37BuOH 35.69 0 39.79 411.24 265.73 34.22BA 90.28 32.05 0 252.63 37.509 345.06CH 27.96 273.66 169.15 0 287.34 515.15BCHE 107.51 558.40 50.47 536.12 0 1343.45Water 135.24 292.44 232.22 144.35 266.55 0

Table A3b Kinetic parameters for CH system (Chakrabarti and Sharma, 1992a)

Parameter Value

k1 (m3/(kmol s)) 4.814105k2 (m3/(kmol s)) 4.985106

Table A4a Binary interaction parameters (Mahajan, 2007)

Component Acet. MEK MPO Crotonaldehyde Water HiB

Acet. 0 19.85 67.83 114.35 539.28 113.81MEK 5.18 0 32.23 15.34 408.65 1.82MPO 99.06 38.09 0 22.36 298.39 27.71Crotonaldehyde 139.56 0.11 37.35 0 233.86 0.60Water 69.85 4.54 139.44 165.61 0 263.63HiB 182.79 18.24 33.64 21.78 255.97 0

Table A4b Kinetic parameters for MPO system (Mahajan, 2007)

Parameter Value

ln(k10) (gmol/(h g)) 18.57ln(k20) (gmol/(h g)) 10.34ln(k30) (gmol/(h g)) 16.31E1 (kJ/gmol) 57.894E2 (kJ/gmol) 37.894E3 (kJ/gmol) 49.911KW 1.179

Table A5a Binary interaction parameters (Aspen Technology Inc.)

Component Glyoxal MeOH MAG DAG Water

Glyoxal 0 340.19 839.05 464.23 480.81MeOH 306.39 0 7.37 366.75 165.26MAG 313.01 41.4 0 110.35 219.62DAG 78.87 248.46 129.82 0 574.70Water 116.01 254.73 394.57 1770.17 0

82

Table A5b Kinetic parameters for DAG system(Mahajani and Sharma, 1997)

Parameter Value

k1 (kmol2/(kg s)) 0.034k2 (kmol/(kg s)) 0.044k3 (kmol2/(kg s)) 0.0085k4 (kmol/(kg s)) 0.14

References

Agarwal, V., Thotla, S. and Mahajani, S.M., 2008a, Attainableregions of reactive distillation. Part I. Non-azeotropic singlereactant systems. Chem Eng Sci, 63(11): 29462965.

Agarwal, V.,Attainabreactant

Aspen TechChakrabarti

cyclohexresin and107115.

Chakrabartiwith cycacid-trea331340.

Chaudari, Balpha-m1-octenecatalysts

DiGuilio, R.Mdiethano

Geelen, H. aas a chemReaction125134.

Gelosa, D., Rreactorspolymer

Hiwale, R.S.distillati

Hung, S.B., LReactivewith adiGreen Ch

Jayadeokar,catalysewith isob

Klepacova, KEthericAppl Cat

Kolah, A.K.,Reactionwith eth

Levenspeil,(John Wi

Lopez, D.E., Goodwin, J.G., Jr., Bruce, D.A. and Lotero, E., 2005,Transesterifaction of triacetin with methanol on solid acidand base catalysts. Appl Catal A: Gen, 295: 97105.

Luo, H.P. and Xiao, J., 2001, A reactive distillation process for acascade and azeotropic reaction system: carbonylation ofethanol with dimethyl carbonate. Chem Eng Sci, 56(2):403410.

Mahajan, Y.S., 2007. Process development studies in reactivedistillation. Ph.D. Thesis. IIT Bombay.

Mahajani, S.M. and Sharma, M.M., 1997, Reactions of glyoxal withaliphatic alcohols using cation exchange resins as catalysts.Org Proc Res Dev, 1: 97105.

Mahajani, S.M., 2000, Reactions of glyoxylic acid with aliphaticalcohols using cation exchange resins as catalysts. ReactFunct Polym, 43: 253268.

Nishihira, K., Tanaka, S., Harada, K. and Sugise, R., 2000. ProcessproduL., Deticsng a): 433arac,iacetillati.M., WthyleEng. andkyl cesariylatioon-e548., B.,

yleneRes., Joskyl conat

a, M.us anwen,g achen

kar, Sdmaineerns: tm, 4, S., Aductitone7557G.D.withm, 5C.M.slurThotla, S., Kaur, R. and Mahajani, S.M., 2008b,le regions of reactive distillation. Part II. Singleazeotropic systems. Chem Eng Sci, 63(11): 29282945.nology Inc., (2001). ASPEN Process Manual., A. and Sharma, M.M., 1992a, Cyclohexanol fromene via cyclohexyl acetate: catalysis by ion-exchangeacid-treated clay. React Polym, 18:

, A. and Sharma, M.M., 1992b, Alkylation of phenollohexene catalysed by cationic-exchange resins andted clay: O- versus C-alkylation. React Polym, 17:

. and Sharma, M.M., 1991, Alkylation of phenol withethyl styrene, propylene, butenes, isoamylene,and diisobutylene: hetrogeneous vs homogeneous. Ind Eng Chem Res, 30: 227231.. and McKinney, M.W., 2000. Selective production oflamine. US Patent 6,075,168.nd Wijffels, J.B., 1965, The use of a distillation columnical reactor, In Proceedings of the Third Chemical

Engineering Symposium (Pergamon, Oxford), pp.

amaioli, M. and Valente, G., 2003, Chromatographic: esterication of glycerol with acetic acid using acidicic resins. Ind Eng Chem Res, 42: 65366544., et al., 2004, Industrial applications of reactiveon: recent trends. Int J Chem React Eng, 2: 152.ai, I.K., Huang, H.P., Lee, M.J. and Yu, C.C., 2007,distillation for two-stage reaction systems: methanolpic and glutaric acid esterication, In Sec Symp onem Proc TechS.S. and Sharma, M.M., 1993, Ion exchange resind etherication of ethylene and propylene glycolsutylene. React Polym, 20: 5767.., Mravec, D., Kaszonyi, A. and Bajus, M., 2007,

ation of glycerol and ethylene glycol by isobutylene.al A: Gen, 328: 113.Asthana, N.S., Vu, D.T., Lira, C.T. and Miller, D.J., 2007,kinetics of the catalytic esterication of citric acid

anol. Ind Eng Chem Res, 46: 31803187.O., (2002). Chemical Reaction Engineering (3rd edition).ley & Sons), pp. 170198

forPetrus,

Kinstro39(3

Podrebof ddist

Reed, Lof eInd

Ryu, J.Ydial

SantacAlkcati541

SchmidethEng

Scott, Bdialcarb

SharmStat

Sheng-usinYu S

TalwalSunengoleChe

Thotlaproace556

Yadav,acidPoly

Zhang,in acing a diaryl oxalate. US Patent 6,018,072.e Roo, R.W., Stamhuis, E.J. and Joosten, G.E.H., 1984,and equilibria of the hydration of propene over acid ion exchange resign as catalyst. Chem Eng Sci,446.G.G., Ng, F.T.T. and Rempel, G.L., 1998, The productionone alcohol with catalytic distillation. Part I. Catalyticon experiments. Chem Eng Sci, 53(5): 10671075.

etzel, L.A. and Ohara, J.B., 1956, Catalytic hydrationne oxide applying ion exchange resins as catalysts. JChem, 48: 205208.Gelbein, A.P., 2002. Process and catlyst for making

aronates. US Patent 6,392,078.a, E., Silvani, R., Wilkinson, P. and Carra, S., 1988,n of p-cresol with isobutene catalysed byxchange resins: a kinetic study. Ind Eng Chem Res, 27:

Doker, M. and Gmehling, J., 2008, Esterication ofglycol with acetic acid catalyzed by amberlyst 36. Ind

Chem, 47(3): 698703.e, G.S. and Zhaozhong, J., 2003. Co-production ofarbonates and diols with treatment of hydroxyl alkyles. US Patent 6,573,396.M. and Mahajani, S.M., (2002). Reactive Distillation:d Future Directions. (WileyVCH, Weinheim), pp. 329W. and Ji-zhong, L., 2005, Synthesis of dibutyl oxalate

id catalyst supported on granular active carbon. HuaGong, 22(10): 4041.., Mankar, S., Katariya, A., Aghalayam, P., Ivanova, M.,cher, K. and Mahajani, S.M., 2007, Selectivitying with reactive distillation for dimerization of C4heoretical and experimental studies. Ind Eng Res6: 30243034.garwal, V. and Mahajani, S.M., 2007, Simultaneousons of diacetone alcohol and mesityl oxide fromusing reactive distillation. Chem Eng Sci, 62:4.and Thathagar, M.B., 2002, Esterication of maleicethanol over cation-exchange resin catalysts. React

2: 99110., et al., 2003, Isobutene hydration over amberlyst-15ry reactor. Chem Eng Proc, 42(12): 985991.

Conceptual design of reactive distillation for selectivity improvement in multiple reactant systemsIntroductionExample 1: SBRD without simultaneous product removalHybrid semi-batch reactive rectification (HSRR) and hybrid semi-batch reactive stripping (HSRS)Case 1: volatility order C, D