Safety enhancement by transposition of the nitration of toluene from semi-batch reactor to continuous intensified heat exchanger reactor

O

pen

A

rchive

T

OULOUSE

A

rchive

O

uverte (

OATAO

)

OATAO is an open access repository that collects the work of Toulouse researchers and

makes it freely available over the web where possible.

This is an author-deposited version published in :

http://oatao.univ-toulouse.fr/

Eprints ID : 15849

To link to this article : DOI:10.1016/j.cherd.2014.07.029

URL :

http://dx.doi.org/10.1016/j.cherd.2014.07.029

To cite this version :

Di Miceli Raimondi, Nathalie and Olivier

Maget, Nelly and Gabas, Nadine and Cabassud, Michel and

Gourdon, Christophe Safety enhancement by transposition of the

nitration of toluene from semi-batch reactor to continuous

intensified heat exchanger reactor. (2015) Chemical Engineering

Research and Design, vol. 94. pp. 182-193. ISSN 0263-8762

Any correspondence concerning this service should be sent to the repository

administrator:

staff-oatao@listes-diff.inp-toulouse.fr

Safety

enhancement

by

transposition

of

the

nitration

of

toluene

from

semi-batch

reactor

to

continuous

intensified

heat

exchanger

reactor

N.

Di

Miceli

Raimondi

a,b,∗

,

N.

Olivier-Maget

a,b

,

N.

Gabas

a,b

,

M.

Cabassud

a,b

,

C.

Gourdon

a,b

a_{UniversitédeToulouse,INPT,UPS,LaboratoiredeGénieChimique,4,AlléeEmileMonso,F-31030Toulouse,France} b_{CNRS,LaboratoiredeGénieChimique,F-31030Toulouse,France}

a

b

s

t

r

a

c

t

Thebehaviorofacontinuousintensifiedheatexchanger(HEX)reactorincaseofprocessfailureisanalyzedand comparedtothebehaviorofasemi-continuousreactor.Thenitrationoftolueneisconsideredastestreactionto identifythemainfailurescenariosthatcanleadtothermalrunawayinbothprocessesusingtheHAZOPmethod. Noflowrateofprocessfluidandutilityfluidinthecontinuousprocess.Nostirringduringfeedingofthereactor followedbynormalstirringforthesemi-continuousreactor.Thesescenariosaresimulatedforbothprocessesand thetemperatureprofilesareobserved.Thisstudyshowsthatthetemperatureisbettercontrolledinthecontinuous processbecauseoftheintrinsiccharacteristicsoftheHEXreactor.Infact,thisdevicehasalowreactivevolume relativetothemassofthereactor,allowingagooddissipationoftheheatproducedbythereaction,evenincaseof failure.Thischaracteristicoftheintensifiedreactorisconfirmedbyanexperimentalwork.

Keywords: Processintensification;Heat-exchangerreactor;Nitration;Safety;Simulation;Failurescenarios

1.

Introduction

Processintensificationaimsatofferingdrasticimprovements in chemical manufacturing particularly in terms of cost, energyconsumption,safety,quantityandqualityofwastes. Thepurposeistodevelopnewmedia(ionicliquids, super-criticalfluids,etc.),newmethodsofactivation(microwaves, ultrasounds,etc.)andnewtechnologies(microreactors,hybrid separators, etc.) to allow these improvements. One of the basicconceptsofprocessintensificationistheuseof multi-functionalapparatuseswheremorethanoneunitoperation are performed in a unique equipment such as intensified heatexchangers(HEX)reactors(Anxionnazetal.,2008).These apparatusesareverypromisingalternativestobatchor semi-batchreactors mainlyusedinfine chemicalsmanufacture. Theirprospectsareadrasticreductionofunitsizeand sol-vent consumption while safety is increased due to their

∗

Correspondingauthorat:UniversitédeToulouse,INPT,UPS,LaboratoiredeGénieChimique,4,AlléeEmileMonso,F-31030Toulouse, France.Tel.:+33562258920;fax:+33562258891.

E-mailaddress:nathalie.raimondi@iut-tlse3.fr(N.DiMiceliRaimondi).

remarkableheattransfercapabilities.However,Ebrahimietal. (2012) pointed out the fact that process intensification by miniaturizationimprovessafetyinmanycasesbutthatthis trendcannotbegeneralized.Therefore, asafetyanalysisis requiredpriortotheimplementationofanintensifiedprocess basedonanewtechnology.

The present work aimstodemonstrate the intrinsically saferbehaviorofaHEXreactordevelopedbytheLaboratoire deGénieChimique(LGC–Toulouse,France)andtheBoostec companyspecializedinthemanufactureofequipmentsmade ofsiliconcarbide(SiC).Thismaterialpresentsexcellent chem-icalresistance,highmechanicalstrengthandstiffness,high thermalresistanceandgood conductivity(180Wm−1_K−1 _at 20◦_{C). The performancesof this device havealready been} demonstratedtohandleexothermicreactionssuchasadirect fluorination(Elgueetal., 2012)andapharmaceutical appli-cation(Despènes etal.,2012).However,itsbehaviorincase

Nomenclature

a interfacialarea(m2_m−3₎ A heatexchangearea(m2₎ C concentration(molL−1₎ CP heatcapacity(Jkg−1K−1)

d diameter(m)

D diffusioncoefficient(m2_s−1₎

E enhancementfactor

Ea apparentactivationenergyofthe

decomposi-tionreaction(Jmol−1₎ F molarflowrate(molh−₁₎

h height(m)

h_int heat transfer coefficient on process side (Wm−2_K−1₎

Ha Hattanumber

ID molarisomerdistribution

K intrinsickineticconstant(Lmol−1_s−1₎

L length(m)

m mass(kg)

M molarweight(kgmol−1₎

MTSR MaximumTemperatureattainablebythe syn-thesisreaction(K)

n numberofmoles(mol)

Ni parameterusedinthemodelingofthekinetic

rate

NC numberofcompounds

Qe heatfluxexchanged(W)

Qp heatfluxproduced(W)

r reactionrate(molL−1_s−1₎ R gasconstant(Jmol−1_K−1₎ S molarselectivity

t time(s)

tr residencetime(s)

tR reactivemediumthickness(m)

tW wallthickness(m)

T temperature(K)

Tb normalboilingtemperature(◦C)

Tm normalmeltingtemperature(◦C)

TMR timetomaximalrate(h)

U heattransfercoefficient(Wm−2_K−1₎

V volume(L)

w massfraction

wR reactivemediumwidth(m)

W acidstrength

Greekletters

1Hr reactionenthalpy(Jmol−1)

1T temperaturerise(◦_C)  thermalconductivity(Wm−1_K−1₎  density(kgL−1₎ ε phasehold-up  conversionrate Subscripts ad inadiabaticconditions aq aqueousphase bulk bulk film film i compoundi Ntol nitrotoluene

R reactivemediumorfluid

tol toluene

W reactorwall

Superscripts

0 atthereactorinletoratinitialtime

a inaqueousphase

b inthebulk f inthefilm i attheinterface

o inorganicphase

offailurehasnotbeeninvestigatedyet.Inthiscontext,the nitration of tolueneis considered as test reaction. This is a liquid–liquid reaction that presents high risk of thermal runaway due toits exothermicity and the low decomposi-tion temperatureofnitrocompounds(ChenandWu,1996). Aromatic nitrationisan intermediatereaction forthe pro-ductionofmanycompoundssuchaspharmaceuticals,dyes, pesticides,explosives.Attheindustrialscale,thisreactionis mostlyoperatedinbatchorsemi-batchreactors(Booth,2000; Dugal,2005).Ithasbeenusedinseveralworkstodemonstrate thecapabilitiesoftechnologiessuchasmicroand millistruc-turedHEX reactors (Burnsand Ramshaw, 2002;Henke and Winterbauer,2005;Halderetal.,2007;Ruslietal.,2013).

In the present work, a safety analysis is conducted to compare therisksassociatedtothe implementationofthe nitrationoftolueneinasemi-batchreactorandinan inten-sified HEXreactor.For thatpurpose,furthertothereaction descriptionandthepresentationofthecontinuousprocess, risk assessment byHAZOP method iscarried out forboth processesinordertoidentifythefailuresthatcancausethe mostseriousdamages.Then,criticalscenariosaresimulated toobservethetemperatureincreaseinbothreactorsincase ofmajorfailures.TheinherentlysaferdesignoftheSiC inten-sifiedHEXreactoristhenconfirmedbyexperimentsinfaulty mode.

2.

Nitration

of

toluene

2.1. Reactiondescription

Thenitrationoftolueneisperformedwithnitricacidin pres-ence of sulfuric acid and water. This acid mixture is the mostcommonnitratingsystemcurrentlyadoptedinchemical industry(Milleretal.,1964;Harris,1976).Themainreactionis themononitrationcharacterizedbyaheatofreaction1Hrof

−125kJmol−1_{(ChenandWu,1996):}

C7H8+_HNO3→ _C7H7NO2+_{H2O (1)}

Nitrationkineticsandselectivitystronglydependonthe sulfuricacid strength W definedasfollows (Zaldivar et al., 1995): W= wa H2SO4 wa H2SO4+w a H2O (2) wherewa H2SO4andw a

H2Oarethemassfractionsofsulfuricacid

andwaterintheaqueousphaserespectively.Ahighsulfuric acidstrengthfavorsconversionratebutdegradesselectivity. Indeed,the dinitrationoftoluenesignificantly occurs from

Table1–Physicalpropertiesofthecompounds.

Component CASnumber M(gmol−1_{) T}

m(◦C) Tb(◦C) CP(Jkg−1K−1) Water 7732-18-5 18.02 0 100 4183a Nitricacid 7697-37-2 63.01 −42 83 1744a Sulfuricacid 7664-93-9 98.08 10 337 1416a Toluene 108-88-3 92.14 −95 111 1705a 2-Nitrotoluene 88-72-2 137.14 −2 223 1476b 3-Nitrotoluene 99-08-1 137.14 16 231 1473b 4-Nitrotoluene 99-99-0 137.14 52 238 1256d 2,4-Dinitrotoluene 121-14-2 182.13 70 221 1400d,c 2,6-Dinitrotoluene 606-20-2 182.13 65 192 1208d,b SiC 409-21-2 40.10 2557 – 668d

Source:NIST(2013)andHaynes(2013).

a _Dataat25◦_C. b _Dataat30◦_C. c _Dataat52◦_C.

d _{Dataforstableandmetastablecrystalform.}

W=0.8.Thetrinitrationoccursinanhydrousmedium(W=1). Astheacidstrength,temperaturealsopromotesthe forma-tionofbyproducts(FranckandStadelhofer,1987;Booth,2000). Hencethemononitrationoftolueneisgenerallycarriedoutat 30–45◦_{Cwithacidstrengthslightlylowerthan0.8.Theimpact} ofacidstrengthonnitrationkineticmakespossiblethe inhi-bition ofthe reactions by dilution.Therefore nitrationcan bequenchedbyaddingwatertothereactive medium:this methodisadoptedinorder tocollectandanalyze samples atthe reactoroutput.Gas chromatographytechnique with aninternalstandardizationmethodisusedtomeasurethe compositionoftheorganicphaseusinga25mHP-1non-polar capillarycolumnandanionizationflamedetector(detection oftoluene,mononitrotoluenesanddinitrotoluenes).

The physical properties of the compounds used in the presentstudyaregiveninTable1(molarweightM,normal meltingtemperatureTmandboilingtemperatureTbandheat

capacityCP).

2.2. Riskassessment

Themononitrationoftolueneishighly exothermic. There-fore,itsprocessingrequiresacontrolofthetemperatureto managetheriskofthermalrunawayduetothelow decom-positiontemperatureofnitrocompounds.Indeed,Chenand Wu (1996) studied the thermal stability of all the com-poundsbydifferential scanningcalorimetry.They observed thatthedecompositionofnitricacidwasbeginningto120◦_C, andabout240◦_{Cfortolueneandmononitratedcompounds.} Thereforereactionmixturehastobekeptbelow120◦_C.

ChenandWu(1996)alsomeasuredtheself-heatrateofthe reactionmixtureasafunctionoftemperatureusingan accel-eratingratecalorimetertoaccesstothetimetomaximalrate, TMR(TownsendandTou,1980).Thisdatacorrespondstothe amountoftimebeforeareactiongetsoutofcontrolin adia-baticconditions.Foraninitialtemperatureof140◦_C,

TMRis about2h;20minat160◦_C;2minat180◦_{C.Atindustrialscale,} areactionissupposedtobeperformedundersafeconditions ifTMRishigherthan8h(Stoessel,2008).

3.

Intensified

continuous

process

3.1. SiCheatexchangerreactor

Thedeviceusedinthisworkisdesignedasamillistructured plateheatexchangerwitharegularseriesofplatesmadeof siliconcarbide(SiC)whereprocessandutilityfluidscirculate (Fig.1).ThethicknessofaplateofSiCis6mm.Thestructureis heldbetweentwoendplatesofstainlesssteel.Itsbehaviorcan beassimilatedtoacross-flowheatexchanger.Thenitration oftolueneiscarriedoutinthisdevicecomposedof3plates wheretheprocessfluidcirculates(processplates)and4 util-ityplates.Thewholereactorweighs16kgwithapproximately 3kgofSiC.

Theprocess fluid flowsina squaresection meandering channelof2mmindepthandabout3mlongbyplate.The pro-cessvolumebyplateis11.3mL(34mLforthewholereactor). Theratiobetweenthesurfaceofthechannelanditsvolume is2000m2_m−3_{. Theutility fluidflowsin15 parallelsquare}

Fig.2–Patternofaprocessplate.

sectionmeanderingchannelsbyplateof2mmindepth.All thechannelsareengravedintheSiCplates.

Circularsubchannelsof2mmdiameteraredrilledinthe thicknessoftheprocessplates.Theyleadtothemainchannel toenableinjectionoffluidsorinsertionofthermocouples.The thermocouplesaresetupinordertobelocatedatthe chan-nelsurfacefornon-intrusivetemperaturemeasurements.The configurationillustratedinFig.2correspondstothefirst pro-cessplate:theaqueousphaseisintroducedatthemaininput andtolueneisinjectedthroughthesecondaryinput.

Previousworksstudiedthehydraulicbehavioroffluidsin meanderingductsorchannels(Josephetal.,1975;Ligraniand Niver,1988;Fellouah etal., 2006). Inthecaseofone-phase flow,theyshowedthatdespitelowReynoldsnumberdueto lowvelocityand smallchannelsize,vorticescanappearin thechannelbends.Thesevorticesavoidpurelaminarflowand allowanenhancementofmixing(Jiangetal.,2004;Anxionnaz etal.,2013)andheattransfer(Rushetal.,1999;Chandratilleke andNursubyakto,2003).Thisparticularhydrodynamic behav-iorinthereactorandtheSiCpropertiesallowstoobtainheat transfercoefficientsupto10000Wm−2_K−1_{(Despènesetal.,}

2012).

3.2. Experimentalsetup

TheexperimentalsetuppresentedinFig.3iscomposedof4 lines:

- Thefeedlineoftheaqueousphaseiscomposedofagear pump(flowrateiscontrolledbythemotorfrequency),aflow meter,temperature and pressure sensorsand apressure safetyvalvecalibratedat16bar.Thislinecanbefedbytwo differenttanks.Thefirstonecontainswaterforstartand

stopoperations(V-101).Thesecondonecontainstheacid mixturecomposedofnitricacid,sulfuricacidandwater (V-102).Forsafetyreasons,itispossibletoflushthereactor withwaterfromtheutilitynetwork(F-105)inordertoavoid astagnationofthereactivemediuminthereactorincase ofpumpfailure.

- Thefeedlineoftolueneiscomposedofatank(V-103),agear pump,aflowmeter,temperatureandpressuresensorsand apressuresafetyvalvecalibratedat16bar.

- Theoutputlineiscomposedofavalvetocollectsamples andacoaxialtubeinwhichcirculatesglycol/watermixture (F-106)at5◦_{Cinordertodrasticallydecreasethe} tempera-tureoftheoutletfluidincaseofincompleteconversionrate inthereactor.TheoutlettankV-104isstirredandpartly filledwithwaterbeforeoperationtoquenchthereaction. - Theutilitylineiscomposedofaflowmeterandtemperature

sensors.Theutilityfluidiswater.Thetemperatureofthe fluidiscontrolledusingaheatedcirculatingbath.

13thermocouplesareinsertedintothereactorinorderto measurethe temperatureofthe processfluidall along the channel.

3.3. Experimentsinnormaloperationmode

Thenitrationoftolueneis carriedout under thefollowing operatingconditions:

- Tolueneisinexcesscomparedtonitricacidwithamolar ratio of 1.5/1 in order to avoid the crystallization of 4-nitrotoluene(Tm=52◦C,seeTable1).

- Nitricacidconcentrationinaqueousphaseis4molL−1_. - Acidstrengthrangesfrom0.75to0.80.

- Tolueneflowraterangesfrom0.9to1.1Lh−1_. - Acidmixtureflowraterangesfrom1.4to1.8Lh−_1. - Utilityfluidtemperaturerangesfrom23to35◦_C. - Utilityfluidflowrateis80Lh−1_.

Table2presentstheresultsobtainedfordifferentoperating conditions(acidstrengthW,temperatureoftheutilityfluid T,residence timetrwhich dependsonthe totalflow rate).

Conversionratecorrespondstothenitricacid conversion rate(reactantindefault)obtainedfromEq.(3).SelectivityS is definedasthe ratio betweenthe sumofthe molar out-let flow rates of the three mononitrotoluene isomers and themolarflowrateoftolueneconverted(Eq.(4)).Themolar

Table2–ExperimentalresultsofthenitrationoftolueneobtainedintheSiCHEXreactor.

W T(◦_C)

tr(s) (%) S(%) ID(%)

2-Nitrotoluene 3-Nitrotoluene 4-Nitrotoluene

0.75 23 40 8.5 98.0 57.6 4.0 38.5 0.75 23 50 9.2 98.0 57.7 4.0 38.4 0.75 30 40 9.9 98.2 57.5 4.2 38.3 0.75 30 50 9.3 98.3 57.4 4.2 38.5 0.75 35 40 11.9 98.6 57.6 4.1 38.4 0.75 35 50 13.9 98.6 57.6 4.2 38.2 0.80 28 40 26.7 94.8 55.9 4.1 40.0 0.80 27 50 33.7 95.1 54.8 4.6 40.5

isomerdistributionIDofmononitratedproductsisobtained byEq.(5): = Fin HNO3−F out HNO3 Fin HNO3 (3) S= Fout

2Ntol+Fout3Ntol+Fout4Ntol Fin

tol−Ftolout

(4)

IDi=

Fout i

Fout_2Ntol+Fout_3Ntol+F_4Ntolout (5)

ForW=0.75,theconversionrateincreasesfromabout8.5%to 14%whenthetemperatureisincreasedfrom23◦_Cto35◦_C. Selectivity isabout 98%. As expected, conversion rate and selectivityarehighlysensitivetoacidstrength.ForW=0.80at 28◦_{C,conversionrateisabout30%foraselectivityof95%.The} molarisomerdistributionforthemononitrationisconsistent withliterature.Indeed,Harris(1976)andMolgaetal.(1993)

obtained55–65%of2-nitrotoluene,3.5–4.5%of3-nitrotoluene and34–40%of4-nitrotoluene.Theseexperimentsdemonstrate thecapabilityofthedevicetohandlethenitrationunder typ-icaloperating conditions(W=0.8,T=30◦_{C).Itispossibleto} achieveconversionratesof30%withlessthan1minof res-idencetime.Ithastobementionedthattheresidencetime cansimplybeincreasedbyaddingplatestotheHEXreactor.

4.

Transposition

from

semi-batch

to

continuous

process:

impact

on

safety

4.1. Nitrationoftolueneinsemi-batchprocess

Inordertocontrolthetemperatureriseduetothe exother-micityofthe reaction, thenitrationoftolueneisgenerally carriedout inafedbatch(semi-batch) reactor.Thereactor isinitiallyfilledwithtolueneandtheacidmixtureisfed.The totaloperationtimeisbetween2and4h.Ittakesintoaccount the feedingtimeofthe acid mixture,around 1.5–3h (Rusli etal.,2013;D’Angeloetal.,2003).D’Angeloetal.(2003) car-riedoutthereactionatalaboratoryscaleinajacketedreactor of1.5Lwherethemonofluidisaglycol–watermixture.The experimentalsetupisgiveninFig.4.Themonofluid temper-atureiscontrolledusingaheating–coolingsystemcomposed ofanelectricalresistanceandtwoplateheat-exchangers(the firstoneuseswaterat15◦_{C,thesecondoneaglycol–water} mixtureat−₁₀◦_{C).Theoperatingconditionsaredescribedin}

Table3.

4.2. RiskassessmentbyHAZOPmethod

Thethermalrunawayriskfortheprocessconsideredinthis workcanbeevaluatedaccordingtotwocriteria:

- Theadiabatictemperaturerise
T_adwhichcorrespondsto thetemperature riseattotalconversion rateinadiabatic conditions: 1T_ad= −1HrF0 toluene NC

X

i=1 F0 iCP,iMi (6)

- Thetimetomaximumrate(TMR)evaluatedatthemaximum temperatureattainablebythesynthesisreaction(MTSR)(Eq.

(7)).Thistemperatureisreachedwhenthedesiredchemical reactioniscarriedoutunderadiabaticconditions,starting attheprocesstemperatureT(Eq.(8)).

TMR= R ·_MTSR2

!

dT dt

·Ea (7) MTSR=T+1T_ad (8)

Eaistheapparentactivationenergyofthedecomposition

reaction,assumingazero-orderkinetic,and(dT/dt)isthe self-heatrateduetothisreaction.Risthegasconstant.Chenand Wu(1996)determinedtheevolutionofTMRasafunctionof temperature.Consideringtheoperatingconditionsdescribed before,theevaluationoftheadiabatictemperaturerisegives 1T_ad=120◦_{C.ItcorrespondstoTMR=30minwithreference} toChenandWu(1996)data,whichclassifiedthisreactionas highlycriticalregardingtheriskofthermalrunaway(Stoessel, 2008).

Inordertocomparethedangerousnessofthesemi-batch andthecontinuousprocesses,riskassessmentbytheHAZOP method was carried out (Di Miceli Raimondi et al., 2009).

Table3–Operatingconditionsforthenitrationof tolueneinalaboratoryscalesemi-batchreactor.

Initialacidstrength(%) 80

Temperature(◦_{C) 30}

Initialmassoftoluene(g) 260

Totalmassofacidmixtureinjected(g) 767

Masscompositionoftheacidmixture(%) HNO3/H2SO4/H2O:

23.2/61.4/15.4

Feedingtime 2h52

Totaloperationtime 4h

Fig.4– Semi-batchreactor–experimentalsetupatlaboratoryscale. Source_{:D’Angeloetal.(2003).}

Thismethodconsidersthepotentialdeviationsofeach oper-atingparameter(temperature,pressure,flowrate,etc.)and study their causes and consequences (Dunjo et al., 2010).

Table4describesthedeviationsidentifiedforbothprocesses that lead to thermal runaway. It appears that the risk of thermal runaway is not negligible in the continuous pro-cess.However,this methoddoesnottakeinto accountthe characteristicsofthereactors(notablytheratiobetweenthe massofthereactorandthemassofthereactive medium). Simulationsarecarriedouttoestimatethevalueofthe tem-peratureriseandthetemperatureincreasingrateincaseof failure.

4.3. Simulationsofseverefailurescenarios

4.3.1. Descriptionofthescenarios

AmongthescenariospresentedinTable4,theonesleadingto themostdramaticconsequenceshavebeensimulated. - Forthesemi-batchprocess,failure3issimulated:LESS

tem-peratureinthereactorwhilenormalfeeding(forinstanceno stirringsolowinterfacialarea)followedbyMORE tempera-tureinthereactor(returntonormalstirringconditionafter thefeedingiscompleted)causedbyNOutilityflowrateand MOREreactantinthereactor.Thisscenariocorrespondsto

Table4–Failuresthatleadtothermalrunway(HAZOPmethod).

Semi-batchprocess(fedstirredtankreactor) Continuousprocess(heat-exchangerreactor)

Failure1:MOREreactantfeedingflowrate Failure1:NOutilityflowrate

-Exampleofcause:feedingpumpfailure -Exampleofcause:piperupture

-Consequence:increaseoftheheatproductionbythereaction leadingtoariseofthefluidtemperature

-Consequence:degradationofheattransfer (utilityside)leadingtoariseofthefluid temperature

Failure2:MOREtemperatureinthereactor Failure2:NOprocessflowrate

-Exampleofcause:failureofthetemperatureregulationsystem -Exampleofcause:bothfeedlinepumpsfailure

-Consequence:riseofthefluidtemperature -Consequence:degradationofheattransfer

(processside)leadingtoariseofthefluid temperature

Failure3:LESStemperatureinthereactor Failure3:MOREtemperatureinthereactor

-Exampleofcause:lessornostirring -Exampleofcause:thermostatedbathpump

breakdown -Consequence:decreaseofthereactionrateleadingtoan

accumulationofthereactantsinthereactorthatcanbedramaticin caseofreturntonormaloperatingconditions

Table5–Semi-batchandcontinuousprocessesconsideredforthesimulationofthefailurescenario.

Process Semi-batch Continuous

Reactorgeometry VR=7.1L VR=0.034L hR=0.21m LR=8.5m dR=0.21m wR=0.002m tW=0.006m tR=0.002m mW=3.3kg tW=0.004m mW=0.43kg

Heattransfercoefficients U=967.7Wm−2_K−1 _U=99.8Wm−2_K−1

A=0.123m2 _A=0.034m2

Masstransfercoefficients kL=1e−5ms−1 kL=5e−4ms−1

a=2400m2_m−3 _a=2400m2_m−3 kLa=0.024s−1 kLa=1.2s−1 Da tot=1e−9m2s−1 Datot=1e−9m2s−1 Da HNO3=3e−9m2s−1 DaHNO3=3e−9m2s−1

abatchreactorinadiabaticcondition(noheatistransferred tothestagnantutilityfluid).

- Forthecontinuousprocess,failures2and3aresimulated simultaneously:MOREtemperatureinthereactorcausedby NOutilityflowrateandNOprocessflowrate.Thereforethe reactorcanalsobeassimilatedtoabatchreactorinadiabatic condition(noheatistransferredtothestagnantutilityfluid).

Inordertocomparebothprocesses,thesemi-batchreactor volumeiscalculatedtoobtainthesameproductivitythanin thecontinuousprocess,i.e.2.4Lh−1_{oftotalflowrate(0.9Lh}−1 intolueneand 1.4Lh−1 _{inacid mixture).Typical operation} time infed batch forthe nitration oftoluene is about 3h (Ruslietal.,2013):hencethereactivemediumvolumeofthe semi-batch reactor is 7.1L. The reactor wall for both pro-cessesissupposedmadeofSiCinordertoconsiderthesame thermalcharacteristics ofthematerial(W=180Wm−1K−1;

CP,W=668Jkg−1K−1; W=3.16kgL−1). Table 5 illustrates the

configurationandparametersofbothscenarios.Ithastobe mentionedthatthehypothesesforthesimulationsare unfa-vorableforthecontinuousprocess(overestimationoftheheat producedandunderestimationoftheheatexchanged) com-paredto the semi-batch process. The massof the wall in thecontinuousprocess isunderestimatedcomparedtothe totalmassoftheHEXreactor.Indeed,onlythewallsbetween theutilityfluidandtheprocessfluidaretakenintoaccount. ThereforetheexchangeareaAisalsounderestimated(halfof thetotalareaisconsidered).Thishypothesistendsto min-imizetheheatfluxtransferredtothe reactorwallinorder tooverestimatetheriskofthermalrunawayforthisprocess configuration.

Heattransfercoefficientestimationisbasedonthe follow-ingassumptions:

- For both technologies, it is assumed that heat transfer betweentheliquidphasesisnotlimiting.

- Semi-batchreactor:theheattransfercoefficienth_inton pro-cesssideis1000Wm−2_K−1_{.Itcorrespondstoatypicalvalue} forsmallvolumestirredtank.Uiscalculatedconsideringthe resistancetotransferinthefluidandthewall:

U=



₁ hint + tW W



−1 (9) - HEXreactor:inthisfailurescenario,processfluidis stag-nant.Thereforeheattransferissupposedtooccurbypure conduction.Thishypothesisleadstothefollowingequation:

U=



_tR 2R + tW W



−1 (10) Thethermalconductivityofthefluidischosenlow,equals to0.1Wm−_1K−_{1 whichcorrespondstotypicalvalueof} con-ductivityinorganicphase(toluene:=0.1Wm−1_K−1_at25◦_C). Aqueousphaseisgenerallycharacterizedbyhigherthermal conductivity(water:=0.6Wm−1_K−1_at25◦_C).

Masstransfercoefficientestimationisbasedonthe follow-ingassumptions:

- The interfacial area is calculated for droplets of organic phaseof1mmdiameter.In microand millichannels,for dropletsgeneratedinaT-junction(equivalenttothe config-urationusedintheHEXreactor),theirdiameterisgenerally limitedbythediameteroftheinjectionchannel(2mmin the present work) as observed byGarstecki et al. (2006)

andGupta etal.(2013). Adropletdiametertwicesmaller than theinjection channelisconsidered inthisstudy to increasethe toluenetransferandso theheatproduction inthecontinuousprocess.Forthesemi-batchprocess,this hypothesistendstominimizethemassfluxsincedroplets with diameter smaller than 100mm can be generatedin stirredcontactors(FernandesandSharma,1967).

- Mass transfercoefficientkL isobtainedfrom typical

andmillichannels(Ghainietal.(2010)obtainedkLavalues

ofabout 1s−1 _{in1mmdiametercapillaries). Itshould be} notedthatthecoefficientkLaconsideredforthe

continu-ousprocessislargeinrelationtothescenarioimposinga stagnant processfluid.Indeed,aphenomenonofsettling canoccur,therebyreducingtheinterfacialareabetweenthe twophases.Theprogressofthereactionandthereforethe temperatureriseareoverestimatedinthisscenario. - Diffusioncoefficientsoftolueneandnitricacidinaqueous

phaseareconstant.Thevaluesusedforthesimulationsare extrapolatedfromZaldivaretal.(1995)andYehandWills (1971)works.

4.3.2. Mathematicalmodeling

Themathematicalmodelisbasedontheequationsdescribed afterwards.Theyareexplicitlyintegratedovertimeaccording toafirst-orderaccuratealgorithm.

4.3.2.1. Material balances. It is assumed that the nitration oftolueneisirreversible and occurs inthe aqueousphase (Zaldivaretal.,1996).Reactantsconsumptiondependsonthe intrinsickineticrateandmasstransferrateoftoluenefrom theorganicphasetotheaqueousphase.Masstransfer estima-tionisbasedonthefilmtheory(TrambouzeandEuzen,2002). Theenhancementofmasstransferbythereactioninthefilm istakenintoaccountbyintroducingtheenhancementfactorE linkedtotheHattanumberHa(seeAppendixA).Thesolubility oftolueneinaqueousphaseisverylowsothatmasstransfer limitationsintheorganicphasecanbeneglected(Levenspiel, 1999).Indeed,themodeldevelopedbyZaldivaretal.(1995)

givesCa,i_tol∼0.01Co

tol,whereC a,i

tolistheconcentrationoftoluene atthe interfaceoftheaqueousphaseand Co

tol the concen-trationoftolueneinorganicphase.Materialbalancesineach phaseareexpressedasfollows:

Partialmolarbalancesintoluene: Organicphase: dno tol dt =−kLaE(C a,i tol−C a,b tol)VR (11) Aqueousphase: dna tol dt =kLaE(C a,i tol−C a,b

tol)VR−rfilmVfilm−rbulkVbulk (12) Partialmolarbalancesinnitricacid:

Organicphase: dno HNO3 dt =0 (13) Aqueousphase: dna HNO3

dt =−rfilmVfilm−rbulkVbulk (14)

Partialmolarbalancesinwater: Organicphase: dno_H 2O dt =0 (15) Aqueousphase: dna H2O

dt =rfilmVfilm+rbulkVbulk (16)

Partialmolarbalancesinsulfuricacid: Organicphase: dno H2SO4 dt =0 (17) Aqueousphase: dna H2SO4 dt =0 (18)

Partialmolarbalancesinnitrotoluene: Organicphase:

dno Ntol

dt =rfilmVfilm+rbulkVbulk (19)

Aqueousphase: dna

Ntol

dt =0 (20)

wherenisthenumberofmolesandCtheconcentration.Only themononitrationisconsideredsincethedinitrationisnot significantforacidstrengthupto80%.Themasstransfer driv-ingforcecorrespondstothedifferencebetweenthetoluene concentrationsattheinterfaceandinthebulkintheaqueous phase,Ca,i_tolandCa,b_tolrespectively.risthereactionrate.V_filmand Vbulkstandforthevolumesofthefilmandofthebulkinthe aqueousphaserespectively.Theyarerelatedtothevolumeof aqueousphaseVaqbythefollowingequations:

Vfilm=Vaq aDa_tol

εaqkL (21)

V_bulk=Vaq−V_film (22)

εaqistheaqueousphasehold-up,equalto0.6inthe

sim-ulations.Thereactioninthefilmistakenintoaccount.The reactionratesareestimatedusingthefollowingequations: r_film=KCa,f tolCaHNO3 (23) Ca,f_tol=(C a,i tol+C a,b tol) 2 (24)

r_bulk=KCa,b_tolCa_HNO

3 (25)

TheconcentrationoftolueneinthefilmCa,f_tol corresponds toa meanvaluecalculated from theconcentrations atthe interface and inthe bulk. The intrinsic kinetic constant K and Ca,i_tol areobtainedfromthe modelproposedbyZaldivar et al. (1995), validated byexperiments in semi-batch reac-tor (D’Angelo et al.,2003). Theconcentrationof nitric acid issupposedhomogeneousinthewholeaqueousphase.This hypothesisisjustifiedinAppendixA.1.Theconcentrationof tolueneattheinterfaceiscalculatedfromtheconcentration oftolueneintheorganicphase(supposedhomogeneous)and theequilibriumconstantobtainedfromthemodelofZaldivar etal.(1995).TheestimationoftheenhancementfactorEand theconcentrationoftolueneinthebulkdependonthe reac-tionregime.TheirexpressionsaregiveninAppendixA. 4.3.2.2. Thermal balances. Heat transfer resistance at the liquid–liquidinterfaceisneglected.Henceitisassumedthat the temperature inthe fluidisuniform. Theheat capacity

Table6–Initialcompositionofeachphaseinmolar fraction.

Component Aqueousphase Organicphase

Toluene 0.00 1.00

Nitricacid 0.13 0.00

Water 0.50 0.00

Sulfuricacid 0.37 0.00

Nitrotoluene 0.00 0.00

oftheprocess fluidiscalculatedfrom theheat capacityof the compounds givenin Table1 (influence oftemperature neglected).Enthalpiesofmixingareneglected.

Thermalbalancesintheprocessfluidandinthereactor wallaregivenbythefollowingequations:

NC

X

i=1 (miCP,i) dTR dt −Qp+Qe=0 (26) mWCP,W dTW dt −Qe=0 (27)

Qp=−1Hr(r_filmVfilm+rbulkVbulk) (28)

Qe=UA(TR−TW) (29)

wheremi is the massof component i. Qp anQe stand for

the heat flux produced by the reaction and the heat flux exchangedbetweenthefluidandthewall.TR andTW

corre-spondtothereactivemediumandthewalltemperatures. 4.3.2.3. Initialconditions. Theinitialtemperaturesofthewall andthefluidarethetemperatureatnormaloperating condi-tions,i.e.30◦_{C.Theinitialmolarcompositionforeachphase} isgiveninTable6.Itisconsideredhomogeneousbetweenthe filmandthebulk.

4.3.3. Simulationresults

ThescenariosdescribedinSection4.3.1aresimulated.The failuresstartatt=0.

- Semi-batchprocess:

Att<0:thereactorisinitiallyfilledwithtoluene.Theacid mixtureisintroducedwhilethereisNOstirring.Thereaction isveryslowandtheconversionrateisclosetozerobecause

theinterfacialareaislow.Thefailurestartsaftertheendof thedosing.

Att=0:stirringisrestartedbutthereisnoflowrateofthe coolingfluid.Thereactorissimulatedasapurebatchreactor filledwithbothreactants.

- Continuousprocess:

Att<0:normaloperatingconditionsareconsidered. Att=0:thefluidscirculationisstopped(bothutilityand processfluids).Withoutanyflow,thereactorisconsideredas apurebatchreactor.Itisassumedthattheconcentrationof reactantsandproductsinthereactoratt=0correspondstothe inletconcentrations.Thishypothesisoverestimatestheheat producedduringthefailuresincethereactantsinthereactor wouldhavebeenpartlyconsumedduringnormaloperation beforethefailureoccurs.

Fig.5presentstheevolutionofthefluidtemperatureinthe semi-batchandtheHEXreactor.Itappearsthatthe tempera-tureinthesemi-batchprocessovercomesthedecomposition temperatureofnitricacid(i.e.120◦_{C)inlessthan1minwhile} themaximaltemperaturereachedinthecontinuousprocess isabout86◦_{C.Theheatabsorptioncapacityofthereactorwall} (mWCP,W)represents11%ofthetotalheatabsorptioncapacity (mWCP,W+mRCP,R)forthefirstprocessversus77%forthe sec-ondone:itleadstotemperaturesatequilibriumofabout135◦_C inthesemi-batchreactorand55◦_{CintheHEXreactor.Note} thattheobjectiveistocomparetheevolutionofthesynthesis reactioninbothreactorsduringthefailurescenarios.Thatis whythedecompositionreactionoccurringfrom120◦_Cisnot takenintoaccount.

Fig. 5canberelatedtoFig. 6which representsthe heat fluxproducedbythereactionQpandtheheatfluxexchanged

betweentheprocessfluidandthereactorwallQe.Theratio

betweenbothfluxes isaboutsixtimeshigherinthe semi-batchreactor(Qp/Qe∼_1500att=0.1s;Qp/Qe∼_10.4att=20s) than inthe intensified HEX reactor(Qp/Qe∼_{250 at} t=0.1s; Qp/Qe∼1.2att=20s)despiteunfavorableconditionsforthe continuousprocess(lowexchangeareaandtotalmassof reac-torandhighkLacoefficient)andfavorableconditionsforthe

semi-batchprocess(highheatexchangecoefficientandlow kLacoefficient).Theheatproducedisrelatedtothe reactor

volume.Thesemi-batchreactorisaround200timeshigher thantheHEXreactorvolume(seeTable5).Itisconsistentwith theobserveddifferencebetweentheinitialheatproductions. Thetransitionsinthereactionregimeforbothprocessesare

Fig.6–Heatfluxproducedbythereactionandtransferredtothereactorwallversustimein(a)thesemi-batchreactorand (b)theHEXreactor.

mentionedinFig.6.ItappearsthatunliketheHEXreactor,the semi-batchreactorisfirstlyinafastreactionregime,where thereaction ismonitoredbythemasstransferkinetic(see

Table5:kLa=0.024s−1inthesemi-batchprocess–kLa=1.2s−1

in the continuous process). The discontinuity observed in

Fig.6(a)ataboutt=60sisduetothetransitioninthe reac-tion regimefromfast withsecond-order reactionmodel to intermediatewithpseudo-firstordermodel(AppendixA).For t>65sinthesemi-batchreactor,Qpiszerobecausetotal

con-versionrateisreached.

Hence, these simulations highlight the potential safety improvements in the implementation of the nitration of toluene when operated in a continuous process (reaction carriedoutinamillistructuredHEXreactor)insteadofa semi-batchprocess.Thetemperatureriseincaseoffailureisbetter controlledintheintensifieddevicethankstoagood dissipa-tionoftheheatproducedbythereactioninthereactorwall. Thiscanbeexplainedbytheratiobetweenthemassofthe reactorwallandthevolumeofthe reactivemediumwhich is0.5kgL−1_{forthesemi-batchreactorand12.6kgL}−1_forthe HEXreactorregardingTable5.

4.3.4. Experimentalresults

Tocompletethetheoreticalstudybasedonsimulations, exper-iments are carried out to investigate the behavior of the intensifiedSiCHEXreactorinfaultymode.Thenitrationof tolueneiscarriedoutundernormaloperatingconditions(see Section3.3)at28◦_{Cwithaninputacidstrengthof80%until} voluntaryfailuresarecausedontheutilityfluidlinewhilethe processfluidcirculatesnormally.Twofailuresarestudied:(i) LESSutilityflowratebydecreasingitfrom80Lh−1_to20Lh−1 followedby(ii)NOutilityflowrate.Fig.7presentstheevolution ofthetemperatureoftheprocessfluidallalongthereactor duringthefailures.Itappearsthatthetemperaturesstabilize 300saftertheutilityflowratereduction.Themean tempera-tureinthereactorhasincreasedofabout1◦_{C.Duringnormal} conditions,thereactortemperatureiscontrolledbytheutility temperature,around28◦_{C.Thistemperatureisalmost} con-stantbecauseofhighflowratecondition.Thetemperatureat theinletofthereactorTI1isslightlyhigherbecausethekinetic rateismaximalduetohighconcentrationsofthereactants (hotspot).

WithLESSutilityflowrate(t=4450sinFig.7),TI1doesnot change becausethe reactantsconcentrations and the inlet temperatureofthe utilityfluidarestillthesame.However, sincetheutilityflowrateislower,itstemperatureisnolonger

constantandincreasesallalongthereactor.Itexplainswhy TI6<TI13.

Inthecaseofthesecondfailure(t=5050sinFig.7),the temperatureinthereactorisnolongercontrolledbythe tem-peratureoftheutilityfluid.TI1isthelowestonebecausethe processfluidentersthereactoratroomtemperature.Then, thetemperatureincreasesbecauseofthereaction.Itappears thatthetemperaturedecreasesatthereactorend(TI13<TI6) because the reactant concentration is lower and the heat released by the reaction is dissipated in the reactor mass (Benaissaet al.,2008).Thetemperatureoftheprocessfluid progressivelyincreasesbecausetheheatproducedisnolonger evacuatedbytheutilityfluid.Therefore,thereactorwallsand thestagnantutilityfluidcontinuouslywarmup.However,the temperature riseisslow,ofabout0.4◦_Cmin−1_.Considering thisvalue,thecriticaltemperatureof120◦_{Cwouldbereached} inmorethan3h30assumingaconstantrateoftemperature rise.Moreover,thisexperimentshowsthatthetemperature decreaseisveryfastincaseofreturntonormalconditions, withaninitialrateof−₁₀◦_Cmin−1_.

Thetemperatureriseobtainedinthecontinuousprocess whiletheutilityflowrateisstoppedcanbecomparedtothe temperaturerise estimatedinthefedbatchreactorincase ofsimilarfailure.Forthatpurposeasimulationiscarriedout assumingNOutilityflowrateandNORMALoperating condi-tionsonprocesssideinthesemi-batchprocess.Thereactor isinitiallyfilledwithtolueneandtheacidmixisfedduring 3hwithaflowrateof1.4Lh−1_{.Theconfigurationisdescribed} inTable4butwithaninitialvolumeVRof2.8L(initialheight

Fig.7–Experimentinthecontinuousprocess:evolutionof thetemperatureoftheprocessfluidinfaultymodes(LESS andNOutilityflowrate).

Fig.8–Simulationresultinthefed-batchprocess: temperatureriseandconversionrateprofileinthereactor incaseoffaultymode(NOutilityflowrate).

hR=0.08m).Theexchangeareaandthespecificareaincrease

duringfeedingasthe volumeofthe aqueousphaseinthe reactorincreases.The mathematicalmodeling isdescribed inSection4.3.2.Eqs.(14),(16)and (18)are replacedbyEqs.

(30)–(32)totakeintoaccounttheacidmixfeedinginthe mate-rialbalances.

dna HNO3

dt =−rfilmVfilm−rbulkVbulk+F 0

HNO3 (30)

dna H2O

dt =rfilmVfilm+rbulkVbulk+F 0 H2O (31) dna H2SO4 dt =F 0 H2SO4 (32)

Thesimulationshowsthatthereactortemperaturereaches 120◦_{C inabout1h(Fig.8).Itgives ameanrate of} temper-aturerise of1.5◦_Cmin−1_{. Itisthreetimes higherthan the} rateobservedintheexperimentconsideringthecontinuous process.

5.

Conclusion

This work demonstrates the feasibility of handling highly exothermicreactionsinacontinuousprocessinsteadofa fed-batchprocess.Itisillustratedusingthenitrationoftoluene. Thisreactionpresentsahighriskofthermalrunawaybecause ofnitricaciddecompositionwhichstartsat120◦_C.Thedevice usedinthecontinuousprocessisamillistructuredHEXreactor madeofSiC.Conversionratesupto30%arereachedunder typ-icaloperatingconditionsbutwithlowresidencetimes,about 1min.

Thisstudyiscompletedbyasafetyanalysis.An enhance-mentinthetemperaturecontrolincaseoffailureisobserved whenusingtheintensifiedcontinuousprocessinsteadofthe semi-batchreactor. This analysisis performed by simulat-ingmajorfailuresforbothscenariosidentifiedbyapriorrisk assessmentbasedonHAZOPmethod.Unfavorable parame-tersintermsofmasstransferandheattransferareconsidered for the continuous process. However, the simulation work showsthatdespitehighheatproduction,dissipationinthe reactormaterialenablesareasonabletemperatureriseofthe reactivemediuminthecontinuousprocesscomparedtothe semi-batchreactor.Thistendencyisconfirmedwithan exper-imentinfaultymodecarriedoutintheSiCHEXreactor.

Inorderto simulatethe continuousprocess more accu-rately, it is now necessary to precisely characterize the

intensifiedHEXreactorintermsofmassandheattransfers. Thefurtherworkunderprogressisnowtodevelop correla-tionsfortheestimationofthemasstransfercoefficientand theheattransfercoefficientinfunctionoftheoperating con-ditions. Theirimplementationinsteady stateanddynamic simulatorswillprovidearobusttoolforriskanalysis.

Acknowledgments

ThisworkhasbeensupportedbytheANR(AgenceNationale de la Recherche), France: Project ANR PropreSur, ANR-06-BLAN-0381-02. Theexperimentalfacility wassupportedby: the FNADT, Grand Toulouse, Prefecture Midi-Pyrenees and FEDERfundings.

Appendix

A.

Determination

of

the

reaction

regime

Toevaluatetheimpactofmasstransferonthereaction,Hatta numberiscalculatedtodeterminethereactionregime(slow, fastorintermediate)asdescribed byTrambouzeandEuzen (2002)andLevenspiel(1999): Ha=

p

KCa HNO3D a tol kL (A.1)

Kistheintrinsickineticconstantobtainedfromthemodel proposedbyZaldivaretal.(1995).Ca

HNO3isthenitricacid

con-centrationinaqueousphase.

A.1. Slowandintermediatereactionregime:Ha<5

In slow and intermediate reaction regime (Ha<5), it is assumedthatthisconcentrationishomogeneousinthewhole aqueousphase(inthefilmandinthebulk)becauseoffast transfertothefilmandlargeexcesscomparedtotoluene.To validatethishypothesis,itispossibletocalculatethefactor N2: N2= Da HNO3C a HNO3 Da tolC a,i tol (A.2)

Ca,i_tolistheconcentrationoftolueneinaqueousphaseatthe interface.Inthesimulations,N2isgreaterthan10uptoa con-versionrateof97%.Thereforethisassumptionhaslowimpact onthecalculationoftheheatproducedbythereactionasa functionoftime.Howeveritallowstoconsiderthereactionas apseudo-firstorderreactionwhenHa<5.Withthissimplified approach,analyticalsolutionsexisttodescribethereactants consumption(TrambouzeandEuzen,2002).

Assumingpseudo-firstorderreaction,theconcentrationin thebulkandtheenhancementfactorareobtainedfromthe followingequations: N3= 1 ch(Ha)· 1 1+Ha



εaqkL aDa tol −₁



th(Ha) (A.4)

Ca,b_tol=N3Ca,i_tol (A.5)

E= Ha th(Ha)·

1−_(N3/ch(Ha))

1−N3 (A.6)

A.2. Fastreactionregime:Ha>5

The reaction in the film is so fast that Ca,b_tol=_{0. Moreover} nitric acid transfercanbelimiting. Two tendenciescan be observed:(i)if Ha/N2<0.1, E=Ha;(ii) ifHa/N2>10,E=1+N2. Kishinevskiiproposedanexplicitcorrelationtoestimatethe enhancementfactorinfastreactionregimeforsecondorder reaction(TrambouzeandEuzen,2002):

E=1+Ha N4(1 −exp(−0.65Ha

p

N4)) (A.7) N4=Ha N2 +exp



_0.68 Ha − 0.45Ha N2



(A.8)

References

Anxionnaz,Z.,Cabassud,M.,Gourdon,C.,Tochon,P.,2008.Heat exchanger/reactors(HEXreactors):concepts,technologies: state-of-the-art.Chem.Eng.Process.:ProcessIntensif.47, 2029–2050.

Anxionnaz,Z.,Cabassud,M.,Gourdon,C.,Tochon,P.,2013.

Influenceofthemeanderingchannelgeometryonthe thermo-hydraulicperformancesofanintensifiedheat exchanger/reactor.Chem.Eng.Process.:ProcessIntensif.73, 67–80.

Benaissa,W.,Elgue,S.,Gabas,N.,Cabassud,M.,Carson,D., Demissy,M.,2008.Dynamicbehaviourofacontinuousheat exchanger/reactorafterflowfailure.Int.J.Chem.React.Eng. 6,A23.

Booth,G.,2000.Nitrocompounds,aromatic.In:Ullmann’s EncyclopediaofIndustrialChemistry.JohnWiley&Sons, Weinheim.

Burns,J.R.,Ramshaw,C.,2002.Amicroreactorforthenitrationof benzeneandtoluene.Chem.Eng.Commun.189,1611–1628.

Chandratilleke,T.T.,2003.Numericalpredictionofsecondary flowandconvectiveheattransferinexternallyheatedcurved rectangularducts.Int.J.Therm.Sci.42,187–198.

Chen,C.Y.,Wu,C.W.,1996.Thermalhazardassessmentand macrokineticsanalysisoftoluenemononitrationinabatch reactor.J.LossPrev.ProcessInd.9,309–316.

D’Angelo,F.A.,Brunet,L.,Cognet,P.,Cabassud,M.,2003.

Modellingandconstraintoptimisationofanaromatic nitrationinliquid–liquidmedium.Chem.Eng.J.91,75–84.

Despènes,L.,Elgue,S.,Gourdon,C.,Cabassud,M.,2012.Impactof thematerialonthethermalbehaviorofheat exchangers-reactors.Chem.Eng.Process.:ProcessIntensif.52,102–111.

DiMiceliRaimondi,N.,Olivier-Maget,N.,Elgue,S.,Gabas,N., Cabassud,M.,2009.Evaluationdesrisquesparlaméthode HAZOPd’unréacteursemi-batchetd’unréacteur-échangeur continu.RécentsProgrèsenGéniedesProcédés98(Article 585).

Dugal,M.,2005.Nitrobenzeneandnitrotoluenes.In:Kirk–Othmer EncyclopediaofChemicalTechnology.JohnWiley&SonsInc., NewYork.

Dunjo,J.,Fthenakis,V.,Vilchez,J.A.,Arnaldos,J.,2010.Hazard andoperability(HAZOP)analysis.Aliteraturereview.J. Hazard.Mater.173,19–32.

Ebrahimi,F.,Virkki-Hatakka,T.,Turunen,I.,2012.Safetyanalysis ofintensifiedprocesses.Chem.Eng.Process.52,28–33.

Elgue,S.,Conte,A.,Gourdon,C.,Bastard,Y.,2012.Direct fluorinationof1,3-dicarbonylcompoundinacontinuousflow reactoratindustrialscale.Chem.Today30,18–21.

Fellouah,H.,Castelain,C.,OuldelMoctar,A.,Peerhossaini,H., 2006.AcriterionfordetectionoftheonsetofDeaninstability inNewtonianfluids.Eur.J.Mech.B:Fluids25,505–531.

Fernandes,J.B.,Sharma,M.M.,1967.Effectiveinterfacialareain agitatedliquid–liquidcontactors.Chem.Eng.Sci.22, 1267–1282.

Franck,H.G.,Stadelhofer,J.W.,1987.Productionandusesof toluenederivatives.In:IndustrialAromaticChemistry. Springer-Verlag,BerlinHeidelberg.

Ghaini,A.,Kashid,M.N.,Agar,D.W.,2010.Effectiveinterfacial areaformasstransferintheliquid–liquidslugflowcapillary microreactors.Chem.Eng.Process.49,358–366.

Garstecki,P.,Fuerstman,M.J.,Stone,H.A.,Whitesides,G.M.,2006.

Formationofdropletsandbubblesinamicrofluidic T-junction–scalingandmechanismofbreak-up.LabChip6, 437–446.

Gupta,R.,Leung,S.S.Y.,Manica,R.,Fletcher,D.F.,Haynes,B.S., 2013.Hydrodynamicsofliquid–liquidTaylorflowin microchannels.Chem.Eng.Sci.92,180–189.

Halder,R.,Lawal,A.,Damavarapu,R.,2007.Nitrationoftoluene inamicroreactor.Catal.Today125,74–80.

Harris,G.F.P.,1976.Isomercontrolinthemononitrationof toluene.In:IndustrialandLaboratoryNitration.ACS SymposiumSeries,vol.22.AmericanChemicalSociety, Washington,pp.300–312.

Haynes,W.M.,(InternetVersion2013)2013.CRCHandbookof ChemistryandPhysics,93rded.CRCPress/Taylor&Francis, BocaRaton.

Henke,L.,Winterbauer,H.,2005.Amodularmicroreactorfor mixedacidnitration.Chem.Eng.Technol.28,749–752.

Jiang,F.,Drese,K.S.,Hardt,S.,Kûpper,M.,Schönfeld,F.,2004.

Helicalflowsandchaoticmixingincurvedmicrochannels. AIChEJ.50,2297–2305.

Joseph,B.,Smith,E.P.,Adler,R.J.,1975.Numericaltreatmentof laminarflowinhelicallycoiledtubesofsquarecrosssection. AIChEJ.21,965–974.

Levenspiel,O.,1999.Fluid–fluidreactions:kinetics.In:Chemical ReactionEngineering,3rded.JohnWiley&Sons,NewYork.

Ligrani,P.M.,Niver,R.D.,1988.Flowvisualizationofdeanvortices inacurvedchannelwith40to1aspectratio.Phys.Fluids31, 3605–3617.

Molga,E.J.,Barcons,C.,Zaldivar,J.M.,1993.Mononitrationof tolueneandquantitativedeterminationoftheisomer distributionbygaschromatography.Afinidad50,15–20.

Miller,R.C.,Noyce,D.S.,Vermeulen,T.,1964.Thekineticsof aromaticnitration.Ind.Eng.Chem.56,43–53.

NIST(NationalInstituteofStandardsandTechnology),2013. WebBookofChemistry.http://webbook.nist.gov/chemistry/

Rush,T.A.,Newell,T.A.,Jacobi,A.M.,1999.Anexperimentalstudy offlowandheattransferinsinusoidalwavypassages.Int.J. HeatMassTransf.42,1541–1553.

Rusli,R.,Shariff,A.M.,Khan,F.I.,2013.Evaluatinghazard conflictsusinginherentlysaferdesignconcept.Saf.Sci.53, 61–72.

Stoessel,F.,2008.ThermalSafetyofChemicalProcesses:Risk AssessmentandProcessDesign.Wiley,Weinheim.

Townsend,D.I.,Tou,J.C.,1980.Thermalhazardevaluation byanacceleratingratecalorimeter.Thermochim.Acta37, 1–30.

Trambouze,P.,Euzen,J.P.,2002.Lesréacteurschimiques,dela conceptionàlamiseenœuvre.Technip,Paris.

Yeh,H.S.,Wills,G.B.,1971.Diffusioncoefficientofaqueousnitric acidat25◦

Casfunctionofconcentrationfrom0.1to1.0M.J. Chem.Eng.Data16,76–77.

Zaldivar,J.M.,Molga,E.,Alos,M.A.,Hernandez,H.,Westerterp, K.R.,1995.Aromaticnitrationsbymixedacid.Slow

liquid–liquidreactionregime.Chem.Eng.Process.34,543–559.

Zaldivar,J.M.,Molga,E.,Alos,M.A.,Hernandez,H.,Westerterp, K.R.,1996.Aromaticnitrationsbymixedacid.Fast