Intensi fi cation of Deep Hydrodesulfurization Through a Two-stage Combination of Monolith and Trickle Bed Reactors☆

Min Xu,HuiLiu*,Shengfu Ji,Chengyue Li

State Key Laboratory of Chemical Resource Engineering,Beijing University of Chemical Technology,Beijing 100029,China

Intensi fi cation of Deep Hydrodesulfurization Through a Two-stage Combination of Monolith and Trickle Bed Reactors☆

Min Xu,HuiLiu*,Shengfu Ji,Chengyue Li

State Key Laboratory of Chemical Resource Engineering,Beijing University of Chemical Technology,Beijing 100029,China

A R T I C L E I N F o

Article history:

Received 5 January 2014

Received in revised form 17 March 2014

Accepted 6 April2014

Available online 18 June 2014

Hydrodesulfurization(HDS)

Kinetics

Mass transfer

Monolithic reactor

Trickle bed reactor

Reactor modeling

Deep hydrodesulfurization(HDS)is an important process to produce high quality liquid fuels with ultra-low sulfur.Process intensification for deep HDS could be implemented by developing new active catalysts and/or new types of reactors.In this work,the kinetics of dibenzothiophene(DBT)hydrodesulfurization over Ni-P/SBA-15/ cordierite catalyst was investigated at 340-380°C and 3.0-5.0 MPa.The first-order reaction model with respect to both DBT and H2was used to fit the kinetics data in a batch recycle operation system.It is found that both the activation energy and rate constant over the Ni-P monolithic catalyst under our operating conditions are close to those over conventionally used HDS catalysts.Comparative performance studies of two types of reactors, i.e.,trickle bed reactor and monolithic reactor,were performed based on reactor modeling and simulation.The results indicate that the productivity of the monolithic reactor is 3 times higher than that of the trickle bed reactor on a catalyst weight basis since effective utilization of the catalyst is higher in the monolithic reactor,but the volumetric productivity of the monolithic reactor is lower for HDS of DBT.Based on simulation results,a tworeactor-in-series configuration for hydrodesulfurization is proposed,in which a monolithic reactor is followed by a tickled bed reactor so as to attain intensified performance of the system converting fuel oil of different sulfur-containing compounds.It is illustrated that the two reactor scheme outperforms the trickle bed reactor both on reactor volume and catalyst mass bases while the content of sulfur is reduced from 200μg·g-1to about 10μg·g-1.

?2014 The Chemical Industry and Engineering Society of China,and Chemical Industry Press.Allrights reserved.

1.Introduction

In recent years,the allowed sulfur content of diesel has been subjected to more stringent environmental regulations in order to control sulfur emission.The sulfur-content of diesel is limited to below 5×10-5g·g-1in most of the developed countries,and it will be reduced to 1×10-5g·g-1or less in the near future[1].To produce high quality fuels with ultra-low sulfur content,process intensi fi cation(PI)of deep hydrodesulfurization(HDS)is required.In the past decades,PIof deep HDS was mainly focused on new catalysts and new types ofreactors,extensive reviews of which could be found in Furimsky[2],Babich and Moulijn[3],and Song and Ma[4].

In the literature,many new catalyst supports(e.g.MCM-41[5,6],SBA-15[7],USY[8],ZMS-5[9]and TiO2[10-12])and new active compounds (e.g.,metal carbides[13,14],metalnitrides[13],and metal phosphide [15-21])were studied to substitute the commercial Co(Ni)-Mo/Al2O3catalysts.Oyama[17]prepared a series of transitional metal phosphides and compared their HDS activities,showing that nickel phosphide catalysts(Ni2P/SiO2)exhibitbetter activity in hydroprocessing in comparison to a commercialNi-Mo-S/Al2O3catalyst.However,there is no any kinetic study of HDS on nickel phosphide catalyst being conducted yet,so it is necessary to study the kinetics over the new HDS catalysts for deep HDS process design.

As to the HDS reactor design,many improvements of the conventionally used trickle bed reactors(TBR)were proposed,such as counter-current reactor[22]and two-phase fi xed bed reactor with pre-saturation[1].However,inherent disadvantages such as high pressure drop,partial wetting and poor mass transfer exist in TBRs packed with catalyst particles.On the other hand,monolithic reactors (MRs)have attracted an ever increasing attention as new“process intensified”three-phase reactors,due to their advantages in comparison to the conventionally used TBRs and slurry bubble columns[23-26]. Comparison studies between performances of MRs and other threephase reactors were carried out by experiment or simulation for various reactions,including hydrogenation of many compounds,such as butyne-1,4-diol[27],2-butyne-1,4-diol[28],styrene[29],α-methylstyrene[30,31],nitrobenzoic acid[24],3-hydroxypropanal[32],2-ethylanthraquinone[33]and glucose[34].These results showed a clearprepotency for MRs,because of low mass transfer resistance,higher wetting efficiency and minimum axial dispersion.Irandoust et al. [35,36]studied the HDS kinetics of thiophene and dibenzothiophene (DBT)on a monolithic catalyst,only kinetic expressions were obtained but the applicability of MRs to HDS was not clarified.In view of the fact that the weight of catalyst per unit volume packed in MRs is less than that in TBRs,it is obvious that MRs for HDS need bigger reactor volumes than TBRs in order to achieve the same conversion of sulfur-containing compounds in diesel.Nevertheless, there are different sulfur compounds in diesel,including thiophene, DBT and 4,6-DMDBT,the chemical reactivities of which are quite different,while the HDS of thiophene is a severely mass transfer controlled fast reaction on the contrary.Here we propose then,a combination of enhanced mass transfer and effective catalyst utilization in MRs and bigger catalysts loading in TBRs,anticipating that process intensification might be achieved for deep HDS.

In the presentwork,first we carried out kinetic experiments of HDS for DBT over a nickel phosphide/SBA-15/cordierite catalyst and obtained the reaction rate expression.Hereafter,based on the kinetics,two reactor models for MRs and TBRs,respectively,were set up and the simulation results were used to compare HDS performances of the two reactors.Finally,a two-stage reactor configuration for intensification of deep HDS was proposed,which consists ofa monolithic catalyst reactor and following TBRs packed with catalyst particles.

2.Experimental

2.1.Reactor system

A schematic representation of the experimental setup is depicted in Fig.1.It mainly consists of a kinetic experimental system and products analysis system.The experimental system was operated in a semibatch mode with continuous gas fl ows.The system consists of a stainless steel reactor(D=9 mm,L=50 cm)packed with a Ni-P monolithic catalyst and a well-mixed tank with 180 mlliquid reactants.Liquid phase in the well-mixed tank was introduced to the reactor by a position pump at superficial velocity of 0.1 cm·s-1,and then it returned to the wellmixed tank to be recycled,while hydrogen passed through the reactor and a backpressure regulator.HDS experiments were carried out at 340-380°C and 3-5 MPa.

The liquid samples were withdrawn at an interval of 1 h,and then analyzed by a gas chromatography(SP2100,Beijing Beifenruili Analytic Instrument Co.)with a packed capillary column(HJ,PONA, 50 m×0.20 mm×0.5μm).

2.2.Catalyst

A nickel phosphide/SBA-15/cordierite monolithic catalyst with a diameter of8 mm and length of50 mm was packed in the stainless steel reactor.The Ni2P/SBA-15/cordierite catalystwas prepared in the following way.First we impregnated about 10 ml of an aqueous solution of(NH4)2HPO4(0.5 mol·L-1)and Ni(NO3)2·6H2O(1 mol·L-1)with P/Nimolar ratio of1/2 onto SBA-15,which is a silica mesoporous material commonly used as catalyst support.The catalyst preparation and characterization methods were described in detailin[37].The cordierite was then dipped into the prepared nickel phosphide/SBA-15 slurry.The cordierite consists of 25 vertical,parallel square channels with a cell density of 4305(4305 cells per square meter),5 cm long,with the dividing walls being 0.1 mm thick.The loading of Ni was 12.4%(by mass)on a basis of the whole monolithic catalyst,or total0.204 g active catalyst was coated onto washcoat with a thickness of90μm in each channel.The nickel phosphide was well dispersed on washcoat layersof the monolith with a BET surface area of154 m2·g-1.The catalystwas activated before HDS reaction by temperature programmed reduction in a fi xed-bed continuous flow reactor at atmospheric pressure at a H2flow rate of100 ml·min-1.

Fig.1.Schematic diagram of the HDS reaction system.

Table 1 Physical properties of the liquid phase

2.3.Physical properties of the liquid phase

The liquid reactantis a mixture of1%(by mass),DBT(99.5%,by mass, Sigma-Aldrich)dissolved in decalin(99.5%,by mass,Sinopharm Chemical Reagent Beijing Co.)solvent.High purity hydrogen(99.999% purity,Beijing Haipu Gas Co.)is used in all experimentalz runs.The physical properties of the reaction mixture were estimated on the basis of the solvent,decalin,alone;density,viscosity and vapor pressure of the solvent were found in[38].The estimates of the liquid properties are given in Table 1.Hydrogen and DBT diffusivities were estimated by Wilke-Chang formula.Hydrogen solubility was estimated by Henry coefficient correlations provided by Korsten and Hoffmann[39].

3.Kinetics of HDS of DBT

A total of 13 reaction runs were performed.Nine runs were performed according to a complete 32-factorial design at hydrogen pressures 3,4,and 5 MPa and reactor temperatures 613,633,and 653 K.And two runs were performed using initial solutions 0.5%and 1.5%(by mass)DBT in decalin and at 653 K and 5 MPa.Two more runs in which H2flow rates were 300 and 500 ml·min-1were done to evaluate the effects of externalmass transfer on HDS.Other runs in which liquid flow rates were 60 and 120 ml·h-1were also done to evaluate the effects of liquid side mass transfer on HDS.Also,the gas-liquid flow regimes were identified as film flow at allused conditions on the basis of the experimental results of Liu et al.[40].

3.1.Effect of external and internal mass transfer

The observed initialrate of HDS of DBT in the first hour of the reaction was used to testthe absence of mass transport limitations.The initial rate of HDS of DBT observed in the well-mixed tank for three separate batch experiments with different hydrogen flow rates over the same catalyst is shown in Fig.2.When the H2flow rate reaches 300 ml·min-1,the initialrate isinsensitive to the increase in H2flow rate.Then,the initialrate is also unchanged when the liquid flow rate increases from 60 ml·h-1to 120 ml·h-1.These observations suggest that the effect of the external mass transfer is also negligible under the reported conditions.Therefore, allkinetic experiments in this paper were performed at the H2flow rate of300 ml·min-1and liquid flow rate of60 ml·h-1.

Fig.2.Effect of hydrogen flow rate on DBT HDS(P=3 MPa,T=653 K,uL=0.1 cm·s-1, CDBT=47.3 mol·m-3).

Also,by using the Weisz-Prater Criterion[41],the effect of the internal diffusion resistance wasinvestigated.The Weisz modulusΦisde fined as

where robsis the observed initialrate in the well-mixed tank;L is the diffusion length or washcoat thickness;Deis effective diffusion coefficient of the reactants;and Csis the concentration of H2or DBT on the catalyst surface.Rough calculations show that the parameterΦis less than 10-3,indicating that the internal diffusion resistance is negligible.

3.2.Effect of initial DBT concentration and hydrogen pressure

Fig.3 shows the effect of DBT initial concentration on the initial reaction rate.It can be observed that the initial rate increases in proportion to DBT initial concentration at concentrations below 47.3 mol·m-3. This observation indicates that,the reaction shows a first order behavior in DBT over the Ni-P/SBA-15 monolithic catalyst.Many kinetic studies at low DBT concentration also suggested that the HDS of DBT is first order in DBT[6,42,43]and at higher concentrations the initial reaction rate levels off.Singhal et al.[44]obtained similar results and the Langmuir-Hinshelwood kinetic model was used to fi t their experimental data.Fig.4 shows the effect of hydrogen pressure on the initial rate;at different temperatures,the first-order behavior in H2can be observed.

Fig.3.Effect of initial concentration of DBT on DBT HDS(P=3 MPa,T=653 K,uG= 0.5 cm·s-1,uL=0.1 cm·s-1).

Fig.4.Effect of H2pressure on DBT HDS(T=653 K,CDBT=47.3 mol·m-3, QG=300 ml·min-1,uL=0.1 cm·s-1).

3.3.Kinetic modeling

The analysis herein before confirmed that the HDS of DBT was first order in both hydrogen and DBT if the concentration of DBT is relatively low.Likewise,the effect of H2S was neglected since in the present semibatch system a low concentration level of H2S was maintained. Therefore,the following kinetic rate expression was used:

where k=k0exp(-Ea/RT)is the rate constant.The data from the experiments were fitted to the kinetic modelembedded in the following mass balance model of the recycle monolithic reactor system.A mass balance on the well-mixed tank gives

where C1is the concentration of DBT leaving the monolithic reactor,C2is the concentration of DBT in the mixed tank and entering the reactor,QLis the liquid flow rate,and VTis the volume of the mixed tank.

Using a plug flow pseudohomogeneous reactor model,the mass balance equation of the monolithic reactor is as follows:

The DBT concentration at inlet of the monolithic reactor is equal to that in the well-mixed tank,i.e.,when z=0,C(z=0,t)=C2(t).Besides, the lag time in the transportline is negligible in view of the large volume of the well-mixed tank.

Eqs.(2)-(4)were solved and the results were fitted to the experimental data by using the least-square method.The preexponential factor(k0) and apparent activation energy(Ea)can be calculated for Ni2P/SBA-15 monolithic catalyst:k0=0.457±0.0162 m6·mol-1·(kg cat)-1·s-1, Ea=72.9±7.69 kJ·mol-1.

Fig.6.Result of DBT HDS over Ni-P/SBA-15 monolithic catalyst in batch recycle reactor (P=3 MPa,T=653 K,CDBT=47.3 mol·m-3,uG=0.5 cm·s-1,uL=0.1 cm·s-1).

3.4.Comparison of the Ni2P/SBA-15 monolithic catalyst and Co(Ni)-Mo catalyst

Firstly,the pathway of the HDS of DBT is discussed.Generally,the HDS of DBT takes place through two different pathways:a direct desulfurization(DDS)route and a hydrogenation(HYD)route.Biphenyl(BP)and cyclohexyl benzene(CHB)are the main products through the DDS and HYD routes.Fig.5 shows the overall network of HDS of DBT proposed by Egorova and Prins[45].Their data indicated that different main products were obtained using CoMo,NiMo and Mo as catalysts,which suggests that different reaction pathways occurred over different catalysts.

Fig.6 shows a typicalresult of HDS ofDBT on Ni-P/SBA-15 monolithic catalyst.Under our operating conditions,the selectivity to BP is more than 90%.It indicates that the HDS of DBT mainly proceeds through the DDS pathway.

Besides BP,a small amount of CHB and trace amount of tetrahydrodibenzothiophene(H4-DBT)were observed over the Ni-P/SBA-15 monolithic catalyst.However,hexahydro-dibenzothiophene(H6-DBT), perhydrodibenzothiophene and bicyclohexylwere not detected,which suggests that over the Ni-P/SBA-15 monolithic catalyst the HDS of DBT most probably occurs via reactions 1,2,3,4 and 7 given in Fig.5.In the HYD pathway,the hydrogenation of the first phenylring of DBT to H4-DBT(Path 2)is the rate-determining step,while the desulfurization of H6-DBT(Path 4)is rather faster.Therefore,the yield of hydrogenated intermediates is very low.In addition,hydrogenation of the other phenyl ring of DBT is difficult since perhydrodibenzothiophene and bicyclohexyl were not detected over the catalyst.Similar results were reported by Egorova and Prins[45]over Co-Mo catalyst and Ni-Mo catalyst,buttheselectivity to BP overNi-P/SBA-15monolithic catalyst in our work is much higher(＞90%).

Fig.5.Pathways of DBT HDS(suggested by Egorova and Prins[45]).

The apparent a ctivation energy and the first-order rate constant for the HDS of DBT are shown in Table 2 and compared with other catalysts in the literature.

Likewise,Table3shows a comparison of the first-order rate constant for the HDS of DBT over Co-Mo catalyst presented in the literature and Ni-Pmonolithic catalyst in this work.It can be seen that both activation energy and rate constant over Ni-P monolithic catalyst under our operating conditions are close to those over conventionally used HDS catalysts(Co(Ni)-Mo).The results demonstrate that Ni-P catalysts are more attractive since they can be prepared readily frominexpensive phosphate by reduction in hydrogen in comparison to conventionally used double metal catalysts.It is worth noting that the kinetic models in the literature also indicate the first-order behavior in DBT,soadirect comparison of various kinetic parameters is possible.

Table2 Comparison of activation energies in this work and the data in literature

4.Reactor modeling and performance comparison

The kinetics obtained above and mass transfer models reported in the literature were used to set up a full-scale monolithic reactor model and atrickle-bed reactor model for HDS.The models for both reactors subject to the following general assumptions:(1)the reactors are isothermal;(2)the reactor is operated under steady state conditions;and(3)the liquidenteringthereactorissaturated withhydrogen.

ThemonolithicreactormodelforHDSwasdevelopedintheTaylor flowregime,andthefollowingassumptionsweremadeinderiving thegoverningequations:uniformdistributioninallchannels,plug flowoffluidphases,anduniformlydistributedactivesitesinthe washcoatlayer.ThemasstransfermodeldevelopedbyKreutzeretal. [49]wasused.Basedontheseassumptions,themassbalanceequations forDBTandhydrogeninliquidphase,thecatalystsurfaceandthemass transfercorrelationsusedinthemodelsaregiveninAppendixA.

Usingthetrickle-bedmasstransfermodeldiscussedbyRajashekharam etal.[50],wedevelopedthetrickle-bedreactormodelforHDSofDBT. Asdiscussedintheliterature,thecatalystparticlephasewasdivided intothreezones:(1)aflowingdynamicliquidzone,(2)astagnant liquidzoneand(3)adryzone.Thevaryingliquid-solidmasstransfer coefficientswereusedatdifferentzones.Betweenthezonesmass transferexchangesalsooccur.InthemodeltheDBTconcentrationon acatalystsurfaceisassumedtobeuniform.Itindicatesthatthereaction alsooccursontheparticlecatalystinthedryzone.Sincethefractionof thedryzoneisverysmall(＜1%inthepresentcase),theinducederroris consideredignorable.Basedonthemasstransfermodel,thetricklebed reactormodelsforHDSaregiveninAppendixB.

Theconditionsformodelcalculationofbothreactorsareshownin Table4.

Table3 Comparisonof thefirst-orderrateconstantsinthisworkandthedataintheliterature

Table4 Reactorparametersusedformodelingof themonolithicandtricklebedreactors

4.1.Monolithicreactor

Thesimulationresultsforthemonolithicreactorareshownin Fig.7.Asshown,inthemonolithicreactor,theDBTconcentrations onthecatalystsurfacearenearlyequaltothebulkconcentrations. Thehydrogenconcentrationsinbulkphaseandthoseonthecatalyst surfacearealsonearlyequaltothesaturatedconcentration basedonouroperatingconditions.Theobservationsindicate thattheliquid-solidandgas-liquidmasstransfersarenotlimiting stepsinthisMR.

4.2.Trickle-bedreactor

ThesimulationresultsforthetricklebedreactorareshowninFig.8. Itindicatesthatinthefirst1mof thetricklebedreactor,thehydrogen concentrationsinliquidbulkandoncatalystsurfacearesomehow lowerthanthesaturatedconcentrationduetothelowergas-liquid masstransfercoefficient(0.08s-1)thanthatinthemonolithicreactor (5.0s-1).Thedifferencesarenotapparentatthelater2/3lengthof thereactor.Therefore,themasstransferresistanceispresentonlyat thebeginningof thereactorinwhichtheDBTconcentrationishighand accordingly reaction rate is high.Similar results were obtained by Macias and Ancheya[51].

Fig.8.Modeling results forthe trickle bed reactorforthe HDSofDBT(P=5 MPa,T=653 K).

Fig.9.The catalyst effectiveness factor for the monolithic reactor and the trickle bed reactor for HDS of DBT(P=5 MPa,T=653 K).

4.3.Comparison and new reactor con fi guration

As reported in Sections 4.1 and 4.2,the monolithic reactor shows the better mass transfer characteristics,nevertheless it needs about 12 times longer reactor lengths to achieve 95%conversion of all DBT while the liquid feed in the monolithic reactor is fi ve times higher than that in the TBR(deduced from liquid super fi cialvelocity shown in Table 4).Table 5 shows a comparison of the monolithic reactor and the trickle bed reactor productivity upon a basis of catalyst mass and reactor volume,when 95%ofDBT in decalin is converted,i.e.,sulfur content ofabout 10μg·g-1reached.It can be seen that the productivity of the monolithic reactor is 3 times higher than that of the trickle bed reactor on the catalyst weight basis.It indicates that the utilization of the catalyst is much effective.This can be explained in view of the thin washcoat thickness in the monolithic reactor.The effectiveness factor ofcatalysts for DBT in both the dynamic and stagnantzones of the trickle bed reactor is less than 0.70,while that in the monolithic reactor is more than 0.95(see Fig.9).The calculation formulae of the effectiveness factor of the monolithic and particle catalystare shown in Appendixes A and B.The presence of5%dry zone and 25%stagnantliquid zone on all catalyst particles is also attributable to the low effective utilization although it is not very signi fi cant(about 10%productivity increases when stagnantliquid is absent).Besides,ef fi cientexternalmass transfer in the monolithic reactor is also an advantage as discussed previously. However,the monolithic reactorneeds more reactor volume to convert the same liquid feed than the trickle bed reactor since the trickle bed reactor contains about 4 times higher active catalyst than the monolithic reactor.

Table 5 Comparison ofperformance of TBR,MR and two-stage reactor for DBT HDS

Based on the results and discussions above,a combination of the excellentmass transfer in the monolithic reactor and the big catalystloading in the trickle bed emerges as a method ofprocess intensi fi cation for deep HDS.Here we propose a two-reactor-in-series con fi guration for hydrodesulfurization,in which a monolithic reactor is followed by a tickled bed reactorso as to attain intensi fi ed performance of the systems converting fueloilcontaining different sulfur-containing compounds. The calculated productivities of the two-stage reactor for HDS of DBT are listed in Table 5.As shown,the two-stage reactor is able to convert most of the liquid in the same size reactor,and the productivity of the two-stage reactor is close to that of the monolithic reactor on the catalystweightbasis.Itcan be explained by the higher catalysteffectiveness and the negligible mass transfer resistance atthe monolithic reactor in comparison of the beginning part of the trickle bed reactor.For example,to convert 100 m3·h-1liquid feed we need a trickle bed reactor with a diameter of1.88 m and length of2.95 m,or a monolithic reactor with a diameter of0.84 m and a length of19.62 m,which is equivalent to combining a monolithic reactor(D=0.84 m,L=4.86 m)subsequently with a trickle bed reactor(D=1.88 m,L=1.96)to achieve the same result.

In general,different sulfur compounds exist in different fuel oils based on their boiling points.For example,in gasoline thiophene is the main sulfur compounds,while alkylbenzonthiophene,DBT and 4,6-dimethyldibenzothiophene(4,6-DMDBT)are also present in diesel. The sulfur compounds show differentreactivities in the HDS process.It has been reported that,under the same reaction conditions,thiophene and benzonthiophene are 1 or 2 orders of magnitude more reactive than DBT,nevertheless DBT is about 20 times more reactive than 4,6-DMDBT[47,48,52-54].Calculations were also performed for HDS of 4, 6-DMDBT,benzonthiophene and thiophene in both the trickle bed and monolithic reactor.In these simulations,the reaction rate varied by 0.05,10 and 100 times,while the correlations of the physicalproperties and the mass transfer as wellas the geometric parameters(reactor length etc.)of the reactor were unaltered.Table 6 shows the results of the monolithic and trickle bed reactors for HDS of other three sulfurcontaining compounds.It can be seen that the monolithic reactor is more superior to the trickle bed reactor with increasing rates of the HDS reaction.Moreover,the productivities of the monolithic reactor are higher than that of the trickle bed reactor both on the reactor volume basis and on the catalyst mass basis for HDS ofbenzonthiophene and thiophene,while for HDS of 4,6-DMDBT negative results are found.Itcan be attributed to the lower internaland gas-liquid external mass transferresistance in the monolithic reactorthan thatin the trickle bed reactor for HDS ofthiophene and benzonthiophene.Figs.10 and 11 show the simulation results of the trickle bed and monolithic reactor forHDS ofthiophene.As shown,the hydrogen concentration in the liquid phase in the monolithic reactor is higher than thatin the trickle bed reactor.It indicates that the gas-liquid mass transfer is signi fi cantly important in the trickle bed reactor at this reaction rate.Moreover,the pressure drops ofboth the monolithic and trickle bed reactor are also shown in Table 6,which illustrates another advantage of the monolithic reactor.So the monolithic reactor is recommended as a superior reactor for HDS of gasoline in which thiophene is the main sulfur-containing compounds.

Table 6 Comparison ofperformance of TBR and MR for HDS of different sulfur-containing compounds

Fig.10.Modeling results for the monolithic reactor for the HDS ofthiophene(P=5 MPa, T=653 K).

Fig.11.Modeling results for the trickle bed reactor for the HDS ofthiophene(P=5 MPa, T=653 K).

Aiming to remove the sulfur-containing compounds ofdifferent reactivity in diesel oil,Mochida et al.[55]and Ma et al.[56]proposed the multi-stage reactorwherein differentcatalysts were packed and different temperatures were set along the varying sections.Likewise,on the basis of the comparison of the trickle bed and monolithic reactor, we propose the two-stage reactor scheme to convert reactive and unreactive sulfur-containing compounds in dieseloils.The monolithic reactor is set to the fi rst stage mainly to remove benzonthiophene and a part of DBT since it has excellent mass transfer characteristics,while the trickle bed reactor is the second stage for HDS of the feeds with the extremely less reactive sulfur compounds(for example 4,6-DMDBT).

5.Conclusions

In the batch recycle reactor the kinetic experiments of the HDS of DBToverthe Ni-P/SBA-15 monolithic reactorwere carried out.Comparative studies of the performances of the two types of reactors(trickle bed and monolithic reactors)were made to determine the best reactor con fi guration for deep HDS offueloilby using the kinetic expression. The main results can be summarized as follows:

(1)The reaction rate is fi rst-order with respect to both DBT and H2, and the reaction activity was compared over Ni-P and Co(Ni)-Mo(W)catalyst.Both activation energy and rate constant over Ni-P monolithic catalystin the investigated operating conditions are relatively close to those over conventionally used HDS catalysts.The Ni-P catalysts are more attractive,since they can save noble metalin comparison to conventionally used double metal catalysts.

(2)A comparison between the performances ofmonolithic and tickle bed reactor for HDS of DBT shows that the productivity of the monolithic reactor is 3 times higher than that of the trickle bed reactor on the catalyst mass basis,but the volumetric productivity of the monolithic reactor is about2.2 times lower than that of the trickle bed reactor.

(3)Simulations for both reactors for HDS of the sulfur-containing compounds of different types and reactivities were also performed.The monolithic reactor is more superior to the trickle bed reactor with increasing rates of the HDS reaction.A twostage reactor is suggested to convertthe fueloilcontaining different sulfur-containing compounds.The monolithic reactor is used as the fi rst stage mainly to remove benzonthiophene and a part of DBT since it has excellent mass transfer characteristics, followed by a trickle bed reactor as the second stage for deep HDS of the feeds with extremely low sulfur contents and less reactive sulfur compounds.The two-stage reactor proposed outperforms the trickle bed reactor both on reactor volume and catalyst mass bases.

Nomenclature

a mass transfer surface area,m2·m-3

abgeometricalsurface area ofcatalystbed in trickle bed,m2·m-3

C concentration,mol·m-3

Ca Capillary number(=μL(uL+uG)/σL)

D diffusion coef fi cient,m2·s-1

Deeffective diffusion coef fi cient,m2·s-1

dhdiameter of monolith,mm

dpcatalyst particle diameter,mm

f friction factor,dimensionless

fdcatalyst fraction in contact with dynamic liquid

fgcatalystfraction in contactwith gas

fscatalyst fraction in contact with stationary liquid

fwwetting effectiveness

GztubeGraetz number based on tube length(=LtubeD/dh2uTP)

g gravity,m2·s-1

H Herry coef fi cient,m3·Pa·mol-1

k reaction rate constant,(m3)2·mol-1·kg-1·s-1

k0preexponentialfactor,(m3)2·mol-1·kg-1·s-1

kexexchange coef fi cient between dynamic and stationary liquid, s-1

kGLgas-liquid volumetric mass transfer coef fi cient,m·s-1

kGSgas-solid volumetric mass transfer coef fi cient,m·s-1

kLSliquid-solid volumetric mass transfer coef fi cient,m·s-1

L diffusion length,m

Ltubelength ofa capillary length,m

Lslugliquid slug length,m

LUCunit celllength,m

P pressure,Pa

P0saturated vapor pressure,Pa

Δp/L unit pressure drop,Pa·m-1

QLliquid flow rate,m3·s-1

R idealgas constant,Pa m3·(mol·K)-1

r reaction rate,mol·(kg-1·s-1)

rvobsobserved volumetric reaction rate,mol·m-3·s-1

Sh Sherwood number(=kL/D)

T temperature,K

tslugthe time in which a slug passed the electrode,s

tfilmcontact time ofliquid fi lm with bubble,s

u velocity,m·s-1

uLliquid super fi cialvelocity,m·s-1

VTvolume ofmixed tank,m3

We Weber number(=ρu2/σab)

δwashcoatthickness,mm

εbbed porosity

μdynamic viscosity,Pa·s

ρdensity,kg·m-3

σsurface tension,N·m-1

ΦWeisz modulus(=rvobsL2/DeCs)

ψslugdimensionless liquid slug length(=Lslug/dh)

ΧLockhart-Martinelliparameter(=(ΔpL/ΔpG)0.5)

Subscripts

b bulk phase

c catalyst surface

cat catalyst

d dynamic

F friction

H hydrogen

G gas

GL gas-liquid

L liquid

LS liquid-solid

T total

TP two phase

s stationary

sat saturated

Appendix A.Monolithic reactor model

DBT concentration in liquid bulk

Hydrogen concentration in liquid bulk

DBT concentration at the catalyst surface

Hydrogen concentration at the catalyst surface

The geometric,hydrodynamic and mass transfer parameters used in the modelin Taylor flow mode are listin Table A1.The effectiveness factors by means of Thiele modulus are calculated to evaluate the internal mass transfer resistance for the thickness of the washcoat layer for the monolithic catalyst,as shown:

whereφsis Thiele modulus for the catalyst,Vsand Ssare the volume and surface area of the catalyst,respectively.And k is the fi rst-order rate constant given in the forehead.

Table A1 Correlations used for modeling the monolithic reactor in Taylor flow regime

Appendix B.Trickle bed reactor model

DBT concentration in dynamic liquid zone

Hydrogen concentration in dynamic liquid zone

DBT concentration in stagnant liquid zone

Hydrogen concentration in stagnant liquid zone DBT concentration on catalyst surface

Hydrogen at catalyst in contact with dynamic liquid

Hydrogen at catalyst in contact with stagnant liquid

Hydrogen at catalyst in contact with gas phase

Table B1 Correlations used for modeling the trickle bed reactor

Table B1 gives the geometric parameters,hydrodynamic and mass transfer correlations used in the model.The effectiveness factors are also evaluated by using Eqs.(A5)-(A8).But,unlike monolithic catalyst we calculate separately the catalysteffectiveness factors for the catalyst particle in each zone because of different hydrogen concentration on the catalyst surface at these different locations.

[1]C.Schmitz,L.Datsevitch,A.Jess,Deep desulfurization of dieseloil:kinetic studies and process-improvement by the use of a two-phase reactor with pre-saturator, Chem.Eng.Sci.59(14)(2004)2821-2829.

[2]E.Furimsky,Selection of catalysts and reactors for hydroprocessing,Appl.Catal.A Gen.171(2)(1998)177-206.

[3]I.V.Babich,J.A.Moulijn,Science and technology of novel processes for deep desulfurization of oil re fi nery streams:a review,Fuel 82(6)(2003) 607-631.

[4]C.S.Song,X.L.Ma,New design approaches to ultra-clean diesel fuels by deep desulfurization and deep dearomatization,Appl.Catal.B Environ.41(1-2)(2003) 207-238.

[5]C.Song,K.M.Reddy,Mesoporous molecular sieve MCM-41 supported Co-Mo catalyst for hydrodesulfurization of dibenzothiophene in distillate fuels,Appl. Catal.A Gen.176(1)(1999)1-10.

[6]X.Li,A.J.Wang,Z.C.Sun,C.Li,Y.K.Hu,Study on hydrodesulfurization kinetics of dibenzothiophene over Ni2Wsul fi des supported by siliceous MCM-41,Acta Petrolei Sinica(Petroleum Processing Section)19(4)(2003)1-7.

[7]T.I.Korányi,Z.Vít,D.G.Poduval,R.Ryoo,H.S.Kim,E.J.M.Hensen,SBA-15-supported nickel phosphide hydrotreating catalysts,J.Catal.253(1)(2008)119-131.

[8]M.Sugioka,F.Sado,Y.Matsumoto,N.Maesaki,New hydrodesulfurization catalysts:noble metals supported on USY zeolite,Catal.Today 29(1-4)(1996) 255-259.

[9]M.Sugioka,F.Sadoa,T.Kurosakaa,X.Wang,Hydrodesulfurization over noble metals supported on ZSM-5 zeolites,Catal.Today 45(1-4)(1998)327-334.

[10]C.M.Wang,T.C.Tsai,I.Wang,Deep hydrodesulfurization over Co/Mo catalysts supported on oxides containing vanadium,J.Catal.262(2)(2009)206-214.

[11]D.H.Wang,W.H.Qian,A.Ishihara,T.Kabe,Elucidation of sul fi dation state and hydrodesulfurization mechanism on Mo/TiO2catalyst using35S radioisotope tracer methods,Appl.Catal.A Gen.224(1-2)(2002)191-199.

[12]D.H.Wang,W.Li,M.H.Zhang,K.Tao,Promoting effect of fl uorine on titaniasupported cobalt-molybdenum hydrodesulfurization catalysts,Appl.Catal.A Gen. 317(1)(2007)105-112.

[13]E.Furimsky,Metalcarbides and nitrides as potential catalysts for hydroprocessing, Appl.Catal.A Gen.240(2003)1-28.

[14]P.Y.Wu,S.F.Ji,L.H.Hu,J.Q.Zhu,C.Y.Li,Preparation,characterization,and catalytic properties of the Mo2C/SBA-15 catalysts,J.Porous.Mater.15(2)(2008)181-187.

[15]X.Q.Wang,P.Clark,S.T.Oyama,Synthesis,characterization,and hydrotreating activity of several iron group transition metal phosphides,J.Catal.208(2)(2002) 321-331.

[16]Y.Y.Shu,Y.K.Lee,S.T.Oyama,Structure-sensitivity of hydrodesulfurization of 4,6-dimethyldibenzothiophene over silica-supported nickel phosphide catalysts,J. Catal.236(1)(2005)112-121.

[17]S.T.Oyama,Novel catalysts for advanced hydroprocessing:transition metal phosphides,J.Catal.216(1-2)(2003)343-352.

[18]S.T.Oyama,Y.K.Lee,The active site of nickel phosphide catalysts for the hydrodesulfurization of4,6-DMDBT,J.Catal.258(2)(2008)393-400.

[19]S.T.Oyama,T.Gott,H.Y.Zhao,Y.K.Lee,Transition metalphosphide hydroprocessing catalysts:a review,Catal.Today 143(1-2)(2009)94-107.

[20]C.Q.Li,G.D.Sun,C.Y.Li,Y.J.Song,Preparation,characterization,hydrodesulfurization and hydrodenitrogenation activities of alumina-supported tungsten phosphide catalysts,Chin.J.Chem.Eng.14(2)(2006)184-193.

[21]A.W.Burns,A.F.Gaudette,M.E.Bussell,Hydrodesulfurization properties of cobaltnickel phosphide catalysts:Ni-rich materials are highly active,J.Catal.260(2) (2008)262-269.

[22]J.A.Ojeda Nava,R.Krishna,In-situ stripping of H2S in gasoilhydrodesulphurization: reactor design considerations,Chem.Eng.Res.Des.82(2)(2004)208-214.

[23]R.K.Edvinsson,A.Cybulsk,A comparison between the monolithic reactor and the trickle-bed reactor for liquid-phase hydrogenations,Catal.Today 24(1-2)(1995) 173-179.

[24]M.Xu,H.Liu,C.Y.Li,Y.Zhou,S.F.Ji,Connection between liquid distribution and gas-liquid mass transfer in monolithic bed,Chin.J.Chem.Eng.19(5)(2011) 738-746.

[25]T.A.Nijhuis,M.T.Kreutzer,A.C.J.Romijn,F.Kapteijn,J.A.Moulijn,Monolithic catalysts as ef fi cient three-phase reactors,Chem.Eng.Sci.56(3)(2001)823-829.

[26]S.Roy,T.Bauer,M.Al-Dahhan,P.Lehner,T.Turek,Monoliths as multiphase reactors: a review,AICHE J.50(2004)2918-2938.

[27]H.Marwana,J.M.Winterbottom,The selective hydrogenation ofbutyne-1,4-diolby supported palladiums:a comparative study on slurry,fi xed bed,and monolith down flow bubble column reactors,Catal.Today 97(4)(2004)325-330.

[28]R.P.Fishwick,R.Natividad,R.Kulkarni,P.A.McGuire,J.Wood,J.M.Winterbottom,E.H. Stitt,Selective hydrogenation reactions:a comparative study ofmonolith CDC,stirred tank and trickle bed reactors,Catal.Today 128(1-2)(2007)108-114.

[29]T.A.Nijhuis,F.M.Dautzenberg,J.A.Moulijna,Modeling of monolithic and trickle-bed reactors for the hydrogenation of styrene,Chem.Eng.Sci.58(7) (2003)1113-1124.

[30]T.Bauer,R.Guettel,S.Roy,M.Schubert,M.Al-Dahhan,R.Lange,Modelling and simulation of the monolithic reactor for gas-liquid-solid reactions,Chem.Eng.Res. Des.83(7)(2005)811-819.

[31]A.G.Bussard,Y.G.Waghmare,K.M.Dooley,F.C.Knopf,Hydrogenation of α-methylstyrene in a piston-oscillating monolith reactor,Ind.Eng.Chem.Res.47 (14)(2008)4623-4631.

[32]A.Cybulski,A.Stankiewicz,R.K.E.Albers,J.A.Moulijn,Monolithic reactors for fi ne chemicals industries:a comparative analysis of a monolithic reactor and a mechanically agitated slurry reactor,Chem.Eng.Sci.54(13-14)(1999) 2351-2358.

[33]D.S.Liu,J.G.Zhang,D.F.Li,Q.D.Kong,T.Zhang,S.D.Wang,Hydrogenation of 2-ethylanthraquinone under Taylor flow in single square channel monolith reactors, AIChE J.55(3)(2009)726-736.

[34]C.Eisenbeis,R.Guettel,U.Kunz,T.Turek,Monolith loop reactor for hydrogenation of glucose,Catal.Today 147S(2009)S342-S346.

[35]S.Irandoust,O.Gahne,Competitive hydrodesulfurization and hydrogenation in a monolithic reactor,AICHE J.36(5)(1990)746-752.

[36]R.Edvinsson,S.Irandoust,Hydrodesulfurization ofdibenzothiophene in a monolithic catalyst reactor,Ind.Eng.Chem.Res.32(2)(1993)391-395.

[37]N.Wei,S.F.Ji,P.Y.Wu,Y.N.Guo,H.Liu,J.Q.Zhu,C.Y.Li,Preparation of nickel phosphide/SBA-15/cordierite monolithic catalysts and catalytic activity for hydrodesulfurization of dibenzothiophene,Catal.Today 147S(2009)S66-S70.

[38]P.S.Ma,Handbook of Basic Data for Petrochemical Engineering,2nd ed.Chemical Industry Press,Beijing,1993.(in Chinese).

[39]H.Korsten,U.Hoffmann,Three-phase reactor model for hydrotreating in pilot trickle-bed reactors,AIChE J.42(5)(1996)1350-1360.

[40]H.Liu,C.O.Vandu,R.Krishna,Hydrodynamics of Taylor flow in verticalcapillaries: flow regimes,bubble rise velocity,liquid slug length and pressure drop,Ind.Eng. Chem.Res.44(14)(2005)4884-4897.

[41]G.F.Froment,K.B.Bischoff,Chemical Reactor Analysis and Design,2nd ed.John Willey&Sons,New York,1990.

[42]Y.Wang,Z.C.Sun,A.J.Wang,L.F.Ruan,M.H.Lu,J.Ren,X.Li,C.Li,Y.K.Hu,P.J.Yao, Kinetics of hydrodesulfurization of dibenzothiophene catalyzed by sul fi ded Co-Mo/MCM-41,Ind.Eng.Chem.Res.43(10)(2004)2324-2329.

[43]X.Z.Yu,X.Q.Ren,Z.G.Dong,J.Wang,Y.R.Wang,Kinetics of the hydrodesulfurization ofdibenzoth iophene over a commercia NiW/Al2O3catalyst,J.FuelChem.Technol.33 (4)(2005)483-486(in Chinese).

[44]G.H.Singhal,R.L.Espino,J.E.Sobel,G.A.Huff,Hydrodesulfurization of sulfur heterocyclic compounds:kinetics ofdibenzothiophene,J.Catal.67(2)(1981)457-468.

[45]M.Egorova,R.Prins,Hydrodesulfurization of dibenzothiophene and 4,6-dimethyldibenzothiophene over sul fi ded NiMo/γ-Al2O3,CoMo/γ-Al2O3,and Mo/γ-Al2O3catalysts,J.Catal.225(2004)417-427.

[46]P.Steiner,E.A.Blekkan,Catalytic hydrodesulfurization ofa light gas oilover a NiMo catalyst:kinetics ofselected sulfur components,FuelProcess.Technol.79(1)(2002) 1-12.

[47]N.K.Nag,A.V.Sapre,D.H.Broderick,B.C.Gates,Hydrodesulfurization of polycyclic aromatics catalyzed by sul fi de CoO-MoO/Al2O3:the relative reactivities,J.Catal.57 (3)(1979)509-512.

[48]M.Houalla,D.H.Broderick,A.V.Sapre,N.K.Nag,V.H.J.de Beer,B.C.Gates,H.Kwart, Hydrodesulfurization of methyl-substituted dibenzothiophenes catalyzed by sul fi ded Co-Mo/Al2O3,J.Catal.61(2)(1980)523-527.

[49]M.T.Kreutzer,P.Du,J.J.Heiszwolf,F.Kapteijn,J.A.Moulijn,Mass transfer characteristics ofthree-phase monolith reactors,Chem.Eng.Sci.56(21-22)(2001)6015-6023.

[50]M.V.Rajashekharam,R.Jaganathan,R.V.Chaudhari,A trickle-bed reactor model for hydrogenation of 2,4 dinitrotoluene:experimentalveri fi cation,Chem.Eng.Sci.53 (4)(1998)787-805.

[51]M.J.Macías,J.Ancheyta,Simulation of an isothermalhydrodesulfurization smallreactor with different catalyst particle shapes,Catal.Today 98(1-2)(2004)243-252.

[52]M.J.Girgis,B.C.Gates,Reactivities,reaction networks,and kinetics in high-pressure catalytic hydroprocessing,Ind.Eng.Chem.Res.30(9)(1991)2021-2058.

[53]T.Kabe,A.Ishihara,Q.Zhang,Deep desulfurization of light oil.Part 2: hydrodesulfurization of dibenzothiophene,4-methyldibenzothiophene and 4,6-dimethyldibenzothiophene,Appl.Catal.A Gen.97(1)(1993)L1-L9.

[54]R.Sha fi,G.J.Hutchings,Hydrodesulfurization of hindered dibenzothiophenes:an overview,Catal.Today 59(3-4)(2000)423-442.

[55]I.Mochida,K.Sakanishi,X.L.Ma,S.Nagao,T.Isoda,Deep hydrodesulfurization of dieselfuel:design of reaction process and catalysts,Catal.Today 29(1-4)(1996) 185-189.

[56]X.L.Ma,K.Sakanishi,T.Isoda,I.Mochida,Hydrodesulfurization reactivities of narrow-cut fractions in a gas oil,Ind.Eng.Chem.Res.34(3)(1996)748-754.

[57]C.O.Vandu,H.Liu,R.Krishna,Mass transfer from Taylor bubbles rising in single capillaries,Chem.Eng.Sci.60(2005)6430-6437.

[58]J.M.van Baten,R.Krishna,CFD simulations of wall mass transfer for Taylor flow in circular capillaries,Chem.Eng.Sci.60(22)(2005)1117-2126.

[59]M.Xu,H.Huang,X.P.Zhan,H.Liu,S.F.Ji,C.Y.Li,Pressure drop and liquid hold-up in multiphase monolithic reactor with different distributors,Catal.Today 147S(2009) S132-S137.

[60]M.T.Kreutzer,M.G.van der Eijnden,F.Kapteijn,J.A.Moulijn,J.J.Heiszwolf,The pressure drop experiment to determine slug lengths in multiphase monoliths, Catal.Today 105(3-4)(2005)667-672.

[61]S.Goto,J.M.Smith,Trickle-bed reactor performance.Part I.Holdup and mass transfer effects,AIChE J.21(4)(1975)706-713.

[62]C.N.Satter fi eld,M.van Eek,G.S.Bliss,Liquid-solid mass transfer in packed beds with downward concurrent gas-liquid fl ow,AICHE J.24(4)(1978)709-718.

[63]I.Iliuta,F.Larachi,B.P.A.Grandjean,Residence time,mass transfer and back-mixing of the liquid in trickle flow reactors containing porous particles,Chem.Eng.Sci.54 (18)(1999)4099-4109.

[64]P.Z.Lu,J.M.Smith,M.Herskowitz,Gas-particle mass transfer in trickle beds,AIChE J. 30(3)(1984)500-502.

[65]J.M.Hochmann,E.Effron,Two-phase cocurrent down flow in packed beds,Ind.Eng. Chem.Fundam.8(1)(1969)63-71.

[66]P.L.Mills,M.P.Dudukovic,Evaluation of liquid-solid contacting in trickle-bed reactors by tracer methods,AIChE J.27(6)(1981)893-904.

[67]Y.Sato,T.Hirose,F.Takahashi,M.Toda,Flow pattern and pulsation properties of cocurrent gas-liquid down flow in packed beds,J.Chem.Eng.Jpn 6(4)(1973) 147-156.

☆Supported by the State Key Development Program for Basic Research of China (2006CB202503).

*Corresponding author.

E-mailaddress:hliu@mail.buct.edu.cn(H.Liu).