+ All Categories
Home > Documents > CFD study of diesel oil hydrotreating process in thenon-isothermal trickle bed reactor

CFD study of diesel oil hydrotreating process in thenon-isothermal trickle bed reactor

Date post: 12-Nov-2015
Category:
Upload: amirchemeng
View: 253 times
Download: 2 times
Share this document with a friend
Description:
In the present study, the Eulerian–Eulerian multiphase approach was implemented to simulate the hydrotreatingprocesses (hydrodesulfurization (HDS) and hydrodearomatization (HDA)) in the trickle bed reactor (TBR) by means ofcomputational fluid dynamics (CFD) technique. A new gas–liquid interphase heat transfer coefficient was used in theCFD model to predict the reactor performance at non-isothermal conditions. The effects of feed inlet temperature,gas and liquid velocities, operational pressure and hydrogen sulfide concentration of the gas phase were investigatedto calculate the reactions conversions and the bed temperature distribution. Also, the influence of bed porosity andadiabatic operational conditions on the reactor temperature and HDS reaction conversion was discussed. The resultsshowed at adiabatic reactor, HDS reaction conversion increased about 9% compared to constant wall temperaturecondition. Furthermore, it was found that neglecting the gas–liquid interphase heat transfer effect in the CFD modelincreases the average relative error between numerical and the experimental data at prediction of HDS conversionabout 5% and also affects the bed temperature pattern.
Popular Tags:
16
chemical engineering research and design 9 4 ( 2 0 1 5 ) 549–564 Contents lists available at ScienceDirect Chemical Engineering Research and Design j ourna l h omepage: www.elsevier.com/locate/cherd CFD study of diesel oil hydrotreating process in the non-isothermal trickle bed reactor Amir Heidari, Seyed Hassan Hashemabadi Computational Fluid Dynamics (CFD) Research Laboratory, School of Chemical Engineering, Iran University of Science and Technology (IUST), Narmak, Tehran 16846-13114, Iran a b s t r a c t In the present study, the Eulerian–Eulerian multiphase approach was implemented to simulate the hydrotreating processes (hydrodesulfurization (HDS) and hydrodearomatization (HDA)) in the trickle bed reactor (TBR) by means of computational fluid dynamics (CFD) technique. A new gas–liquid interphase heat transfer coefficient was used in the CFD model to predict the reactor performance at non-isothermal conditions. The effects of feed inlet temperature, gas and liquid velocities, operational pressure and hydrogen sulfide concentration of the gas phase were investigated to calculate the reactions conversions and the bed temperature distribution. Also, the influence of bed porosity and adiabatic operational conditions on the reactor temperature and HDS reaction conversion was discussed. The results showed at adiabatic reactor, HDS reaction conversion increased about 9% compared to constant wall temperature condition. Furthermore, it was found that neglecting the gas–liquid interphase heat transfer effect in the CFD model increases the average relative error between numerical and the experimental data at prediction of HDS conversion about 5% and also affects the bed temperature pattern. © 2014 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved. Keywords: Computational fluid dynamics (CFD); Hydrotreating; Non-isothermal condition; Trickle bed reactor; Gas–liquid heat transfer 1. Introduction Hydrodesulfurization (HDS) and hydrodearomatization (HDA) are the well-known hydrotreatment (HDT) processes of fos- sil fuels impurities reduction in the oil industries. The HDT processes are generally performed in the packed bed reac- tors known as trickle bed reactor (TBR) (Ranade et al., 2011). In majority of the TBRs, gas and liquid phases move continuously in the co-current downflow mode through the high pressure bed (Mederos et al., 2009b; Ranade et al., 2011; Wang et al., 2013). During the HDT process, hydrogen transfers from the gas to liquid phase and reacts with the target components to reduce oil impurities. TBRs are generally characterized to have low pressure drop and low catalyst loss. There are no moving parts and TBRs have comparatively lower maintenance costs. On the other hand, issues such as catalyst deactivation and Corresponding author. Tel.: +98 21 77240376; fax: +98 21 77240495. E-mail address: [email protected] (S.H. Hashemabadi). Received 8 April 2014; Received in revised form 16 August 2014; Accepted 22 September 2014 Available online 30 September 2014 cementations are also TBRs’ main characteristics (Mederos and Ancheyta, 2007; Satterfield, 1975). In the past, many attempts were made to find out about dif- ferent aspects of TBR performance using numerical methods. Reviewing published studies shows that most of the research works were focused on development of appropriate models to predict bed hydrodynamic properties by means of CFD tech- nique. Investigating the effect of pressure drop (Atta et al., 2010; Attou and Ferschneider, 2000; Aydin and Larachi, 2005, 2008; Bazmi et al., 2011, 2012; Boyer et al., 2007; Brkljac et al., 2007; Iliuta et al., 2002; Iliuta and Larachi, 2002, 2005; Iliuta et al., 2000a,b; Lopes and Quinta-Ferreira, 2008, 2009b; Propp et al., 2000; Salimi et al., 2013; van der Merwe et al., 2007), mal-distribution and liquid holdup (Bazmi et al., 2012; Gunjal et al., 2003; Lappalainen et al., 2009; Lopes and Quinta-Ferreira, 2009a; Martínez et al., 2012; Mederos et al., 2009a; Strasser, http://dx.doi.org/10.1016/j.cherd.2014.09.016 0263-8762/© 2014 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.
Transcript
  • chemical engineering research and design 9 4 ( 2 0 1 5 ) 549564

    Contents lists available at ScienceDirect

    Chemical Engineering Research and Design

    j ourna l h omepage: www.elsev ier .com/ locate /cherd

    CFD s g pnon-i

    Amir HeComputation nginScience and

    a

    In as i

    pr DA)

    co erph

    CFD model to predict the reactor performance at non-isothermal conditions. The effects of feed inlet temperature,

    gas and liquid velocities, operational pressure and hydrogen sulde concentration of the gas phase were investigated

    to calculate the reactions conversions and the bed temperature distribution. Also, the inuence of bed porosity and

    adiabatic operational conditions on the reactor temperature and HDS reaction conversion was discussed. The results

    showed at adiabatic reactor, HDS reaction conversion increased about 9% compared to constant wall temperature

    co

    in

    ab

    Ke

    G

    1. Int

    Hydrodesuare the wesil fuels improcesses ators knownmajority ofin the co-cbed (Meder2013). Durigas to liquireduce oil imlow pressuparts and TOn the oth

    CorresponE-mail aReceivedAvailable

    http://dx.do0263-8762/ndition. Furthermore, it was found that neglecting the gasliquid interphase heat transfer effect in the CFD model

    creases the average relative error between numerical and the experimental data at prediction of HDS conversion

    out 5% and also affects the bed temperature pattern.

    2014 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.

    ywords: Computational uid dynamics (CFD); Hydrotreating; Non-isothermal condition; Trickle bed reactor;

    asliquid heat transfer

    roduction

    lfurization (HDS) and hydrodearomatization (HDA)ll-known hydrotreatment (HDT) processes of fos-purities reduction in the oil industries. The HDTre generally performed in the packed bed reac-

    as trickle bed reactor (TBR) (Ranade et al., 2011). In the TBRs, gas and liquid phases move continuouslyurrent downow mode through the high pressureos et al., 2009b; Ranade et al., 2011; Wang et al.,ng the HDT process, hydrogen transfers from thed phase and reacts with the target components to

    purities. TBRs are generally characterized to havere drop and low catalyst loss. There are no movingBRs have comparatively lower maintenance costs.er hand, issues such as catalyst deactivation and

    ding author. Tel.: +98 21 77240376; fax: +98 21 77240495.ddress: [email protected] (S.H. Hashemabadi).

    8 April 2014; Received in revised form 16 August 2014; Accepted 22 September 2014 online 30 September 2014

    cementations are also TBRs main characteristics (Mederosand Ancheyta, 2007; Sattereld, 1975).

    In the past, many attempts were made to nd out about dif-ferent aspects of TBR performance using numerical methods.Reviewing published studies shows that most of the researchworks were focused on development of appropriate models topredict bed hydrodynamic properties by means of CFD tech-nique. Investigating the effect of pressure drop (Atta et al.,2010; Attou and Ferschneider, 2000; Aydin and Larachi, 2005,2008; Bazmi et al., 2011, 2012; Boyer et al., 2007; Brkljac et al.,2007; Iliuta et al., 2002; Iliuta and Larachi, 2002, 2005; Iliutaet al., 2000a,b; Lopes and Quinta-Ferreira, 2008, 2009b; Proppet al., 2000; Salimi et al., 2013; van der Merwe et al., 2007),mal-distribution and liquid holdup (Bazmi et al., 2012; Gunjalet al., 2003; Lappalainen et al., 2009; Lopes and Quinta-Ferreira,2009a; Martnez et al., 2012; Mederos et al., 2009a; Strasser,

    i.org/10.1016/j.cherd.2014.09.016 2014 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.tudy of diesel oil hydrotreatinsothermal trickle bed reactor

    idari, Seyed Hassan Hashemabadi

    al Fluid Dynamics (CFD) Research Laboratory, School of Chemical ETechnology (IUST), Narmak, Tehran 16846-13114, Iran

    b s t r a c t

    the present study, the EulerianEulerian multiphase approach w

    ocesses (hydrodesulfurization (HDS) and hydrodearomatization (H

    mputational uid dynamics (CFD) technique. A new gasliquid introcess in the

    eering, Iran University of

    mplemented to simulate the hydrotreating

    ) in the trickle bed reactor (TBR) by means of

    ase heat transfer coefcient was used in the

  • 550 chemical engineering research and design 9 4 ( 2 0 1 5 ) 549564

    Nomenclature

    a, b and C constants (dimensionless)aGL specic gasliquid interface (m1)as specic surface of a particle (m1)CL,i concentration of ith species at liquid phase

    (mol/m3)Cp heat capacity (J/kg K)D diffusion coefcient (m2/s)dp, dh, dr particle, hydraulic and reactor diameter (m)E1, E2 Ergun constantsEo Etvs number Eo = Lg(dh)2/ in Eq. (14)

    (dimensionless)Eo modied Etvs number (dimensionless)F interphase momentum exchange coefcient

    (kg/m3 s)fe wetting efciency (dimensionless)Fr Froude number (dimensionless)g gravity (m/s2)H Henrys coefcient (MPa cm3/mol)h specic enthalpy (W/m3)hG,L gasliquid interphase heat transfer (W/m2 K)Ga modied Galileo number (dimensionless)K mass transfer coefcient (m/s)KPoly/Di/Mono dynamic equilibrium constant (dimension-

    less)Kw Watson characterization factor (dimensionless)k thermal conductivity (W/m K)ks HDS reaction constant (((m3)2.16)/kg(kmol)1.16 s)k,k forward and backward reaction rate constants

    (m3/kg s)kad HDS reaction adsorption coefcient (m3/kmol)LHSV liquid hourly supercial velocity (h1)m interphase mass transfer rate (kg/m3 s)Nu Nusselt number (dimensionless)Pr Prandtl number (dimensionless)Re Reynolds number, Re = Udh/ in Eq. (14)

    (dimensionless)Re modied Reynolds number (dimensionless)P pressure (Pa)q heat ux (W/m2)R universal gas constant (J/mol K)Ri reaction rate of ith species (kg/m3 s)r radius (m)S source term (kg/m3 s)SG specic gravity (dimensionless)T temperature (K)TCABP cubic average boiling point (K)TMABP molal average boiling point (K)TMeABP mean average boiling point (K)t time (s)u velocity (m/s)v molar volume (m3/mol)x mass fraction (dimensionless)xmi molar fraction (dimensionless)

    Greek symbols, and constants (dimensionless), and constants (dimensionless)HR heat of reaction (J/mol) volume fraction (dimensionless) average bed porosity (dimensionless)

    Subscriptb G i int k L P r S T

    2010; ZeiseDudukovicKundu et aFerreira, 20Ring and Mthe main reactors. ToperationaHashemabFerreira, 202010), scaleand Krishn(Alvarez anet al., 2012aMurali et alems (Kalli2011; Skalamany reseapublished oheat transfreported. Tcan be sumfer (Flvio Pet al., 2003;ticles to uet al., 2006heat transftive axial (FconductivitLamine et a

    In this cwas simulagasliquid iby Heidari phase momimplementreported Hnd the beiors of reacExtensive seffects of gtemperaturdynamic viscosity (N s/m2)solubility coefcient (N l/kg MPa)ratio of catalyst volume to summation of inertand catalyst volume (dimensionless)density (kg/m3)surface tension (N/m)

    sboiling point or bulk porositygasith speciesinteractionkth phaseliquid phasepressurerth phasesolid phasetemperature

    r et al., 2001) and wetting efciency (Al-Dahhan and, 1995; Baussaron et al., 2007a,b; Cheng et al., 2012;l., 2003; Lappalainen et al., 2008; Lopes and Quinta-10c; Mills and Dudukovic, 1981; Pironti et al., 1999;issen, 1991; van Houwelingen et al., 2006) were

    subjects on the numerical study of trickle bedhe investigations of TBR performance at differentl conditions (Gunjal and Ranade, 2007; Heidari andadi, 2014; Jarullah et al., 2011; Lopes and Quinta-07, 2010a,b,c; Lopes et al., 2007; Mapiour et al.,

    up challenges of TBRs (Hickman et al., 2013; Siea, 1998), reactor performance in the industrial scaled Ancheyta, 2008; Alvarez et al., 2007; Jarullah,b; Mederos and Ancheyta, 2007; Munoz et al., 2005;l., 2007) and study of catalysts deactivation prob-nikos et al., 2008; Kam et al., 2005; Pacheco et al.,

    et al., 1991) were also other interesting subjects forrchers. Despite the fact that many researches weren hydrodynamics and yield of TBRs, studies abouter in the trickle ow regime have not been widely

    he available researches about heat transfer at TBRsmarized in four categories; wall to bed heat trans-into Moreira et al., 2006; Habtu et al., 2011; Larachi

    Mariani et al., 2003; Mousazadeh et al., 2012), par-ids heat transfer (Borremans et al., 2004; Guardo, 2007; Larachi et al., 2003), gasliquid interphaseer (Heidari and Hashemabadi, 2013) and the effec-lvio Pinto Moreira et al., 2006) and radial thermaly (Babu et al., 2007; Flvio Pinto Moreira et al., 2006;l., 1996; Larachi et al., 2003).ontribution, the non-isothermal trickle bed reactorted by CFD method and implementation of a newnterphase heat transfer model that was developedand Hashemabadi (2013). The appropriate inter-entum, mass and heat interaction models were

    ed in EulerianEulerian multiphase approach. TheDS and HDA reactions kinetics were optimized tost values that present intrinsic chemical behav-tions mechanisms based on the reactor real model.imulation works were conducted to examine theas and liquid velocities, operational pressure, feede and gas phase hydrogen sulde concentration on

  • chemical engineering research and design 9 4 ( 2 0 1 5 ) 549564 551

    the reactor performance. The CFD results were compared withreported eximportanceaccuracy ofand bed tem

    2. Go

    2.1. Mo

    To develop lowing assu

    1. Catalyst2. Vaporiza

    to their 3. Two-lm

    fer betw4. Mass tra

    uid interesistan

    5. The solporosity

    6. Accordinet al. (20ow regthe simuforce wa

    2.2. Hy

    In the prapproach wuid (discretfollows:

    (kk)t

    +

    (kkuk)t

    +k

    i=r(u

    The lastbetween phciency, fe, is

    Fint,G = fe

    Fint,L = fe(

    In orderAttou and model. In tin the derivtion to accoin the three

    forms:

    GG +

    +E2

    GG +

    +E2

    1

    G +

    le 1 pties,

    Spe

    Eulen be

    xk,i +

    k

    i=r(m

    corrffusile 1. n infthe e mdisp(2011sion on, p

    havrs. Thn (digatios govchey

    the ity d) are

    oveanc

    ith la fullyriatel deses sizCFD

    abition rese

    et aperimental data (Chowdhury et al., 2002). Also, the of gasliquid interphase heat transfer effect on the

    CFD model at prediction of reactions conversionsperature distribution were studied.

    verning equations

    deling assumption

    the momentum, mass and heat equations the fol-mptions were regarded as:

    s deactivation was ignored in the bed.tions of diesel oil components were neglected duehigh molecular weights.

    theory was used to dene interphase mass trans-een gas and liquid phases.nsfer resistance is controlled by the gas and liq-rface and consequently it was assumed that noce exists between other phases.id phase was taken as third phase with known

    distribution and zero velocity in the domain.g to Lopes and Quinta-Ferreira (2007) and Atta07a,b), due to low gasliquid interaction at trickleime, capillary pressure force can be neglected inlation of TBRs. Accordingly, in this work, capillarys ignored in the CFD model.

    drodynamics equations

    esent study, an EulerianEulerian multiphaseas used to present gas (continuous phase) and liq-e phase) continuity and momentum equations as

    (kkuk) =k

    i=r(mrk mkr) + Sq (1)

    + (kkuk2) = kPk + (kuk) + kkg

    rkmrk ukrmkr) + Fint,k(uk ur) (2)

    term in Eq. (2), Fint,k, shows momentum exchangeases and with considering effect of wetting ef-

    evaluated by (Lappalainen et al., 2009):

    FGL (1 fe)FGS (3)

    FLG FLS) (4)

    to develop interphase momentum exchanges, Fkr,Ferschneider (1999) model was applied in the CFDhis model, effective diameter of particles was usedation of interphase interaction terms as a correc-unt for the presence of the liquid lm. Their model

    phase system can be expressed by the following

    FGL =

    FGS =

    FLS =

    Tabproper

    2.3.

    In the tion ca

    kkt

    +

    Theular diin Tabtillatio

    In describradial et al. dispertributifactorsreactopersioinvestispecieand Anfore, in(porosnelingcausesperformbeds wnot beappropmaticaparticlin the

    Theevaluaferent GunjalL

    (E1G(1 G)2

    G2dp2

    [S

    (1 G)]0.667

    G(uG uL)(1 G)Gdp

    [S

    (1 G)]0.333)

    (5)

    L

    (E1G(1 G)2

    G2dp2

    [S

    (1 G)]0.667

    GuG(1 G)Gdp

    [S

    (1 G)]0.333)

    (6)

    L

    (E1G(S)

    2

    L2dp2

    + E2LuL(S)Ldp

    )(7)

    resents supplementary correlations to estimate oil wetting efciency and porosity distribution.

    cies conservation equations

    rianEulerian multiphase approach, species equa- written as follows:

    (kkxk,iUk) = (kkDi,kxk,i)

    rk,i mkr,i) + fekRk,i (8)

    elations required for calculation of species molec-on, Di, and interphase mass transfer are presentedAlso, diesel oil component concentration and dis-ormation are tabulated in Table 2.current work, molecular diffusion was used toass diffusion instead of packed beds axial andersion coefcients, Eq. (8). According to Ranade) main factors contributing to mass and liquidin the packed beds are non-uniform porosity dis-artial wetting, dead-zones and channeling. Thesee the vital effects on dispersion in the trickle bede main idea to develop axial and radial mass dis-

    ffusive term) models in the packed beds is then of hydrodynamics and porous media effects onerning equations (Jarullah et al., 2012b; Mederosta, 2007; Mederos et al., 2006, 2009a, 2012). There-

    models in which hydrodynamic and bed propertiesistribution, partial wetting, dead-zones and chan-regarded, using axial or radial dispersion coefcientr prediction of species diffusion terms and reactore. Here, it should be noticed that in the packedrge particles diameter, mechanical dispersion can-

    explained by porosity distribution functions and models need to be used to improve the mathe-cription of the reactor. In this work, due to smalle, the effect of mechanical dispersion was ignoredmodel.lity of EulerianEulerian multiphase approach onof axial and radial dispersion was studied by dif-archers (Atta et al., 2007a; Bazmi et al., 2012;l., 2003). They found a good agreement between

  • 552

    chem

    ical

    engin

    eering

    research

    and

    desig

    n

    9 4

    (

    2 0

    1 5

    )

    549564

    Table 1 Correlations to estimate diesel oil properties, bed characteristics and interphase mass transfer correlations.

    Parameter Correlation

    Liquid density (Ahmed, 2007)L = 0 + P + TP = ([0.167 + 16.181 100.04250 ][P 1.450 107] 0.01 [0.2999 + 263 100.06030 ] [P 1.450 107]2) 16.018T = ([0.0133 + 152.4(0 + P)2.54](T 1.8 520) [8.1 106 0.0622 100.764(0+P )](T 1.8 520)2) 16.018

    Liquid viscosity (Ahmed, 2007)

    L = 3.141 107(T 1.8 460)3.444[log10(API)]aa = 10.313[log10(T 1.8 460)] 36.447API = 141.5

    SG15.6 131.5

    Surface tension (Tsonopoulos et al., 1986) = (0.0017237 T0.05873b

    SG0.64927(L G))4/1000Wetting efciency (Lappalainen et al.,

    2008)fe = 0.335Re0.185L Eo0.188Ga0.025G (1 + FrG)0.014

    Diffusion coefcient (Ahmed, 2007; Perryand Green, 2008; Reid et al., 1987; Riazi,2005)

    DL,i = 8.93 108v0.267Lv0.433i

    T

    L

    vi = 0.285v1.048ci or L , vcL = 0.5567T0.2896MeABPSG

    0.766615.6

    TMeABP = TMABP + TCABP2 , TMABP =n

    i=1

    xmiTbi

    TCABP =(

    11.8

    )[ ni=1

    i(1.8Tbi 459.67)1/3]3

    Interphase mass transfer (Goto andSmith, 1975; Korsten and Hoffmann,1996)

    mGL,i = KL,iaGL(

    PG,iHi

    CL,i)

    , mLG,i = mGL,i i : H2 or H2SKL,iaLDL,i

    = 1.8(

    LuLL

    )2(L

    LDL,i

    )0.5 i : H2 or H2S

    Hi =vi

    iL i : H2 or H2S

    H2 =(

    0.5597 0.4294 103(T 273.15) + 3.0753 109 (T 273.15)20 C

    + 1.944 106(T 273.15)2 + 0.835 103(

    120 C

    )2) 106

    H2S = exp(3.3670 0.00847(T 273.15)) 103

    Liquid thermal conductivity (Riazi andFaghri, 1985)

    k = 1.7307(1.8 TMeABP)SG15.5 C = exp(21.78 8.07986t + 1.12981t2 0.05309t3) = 4.13948 + 1.29924t 0.17813t2 + 0.00833t3 = 0.19876 0.0312t 0.00567t2t = (1.8T 460)/100

    Liquid heat capacity (Kesler and Lee, 1976)

    Cp = ( + T) = 1.4651 + 0.2302 Kw = 0.306469 0.16734 SG15.5 C = 0.001467 0.000551 SG15.5 CKw = ((1.8 TMeABP)1/3)/SG15.5 C

    Bed porosityfunction (Bazmiet al., 2011)(constants wereselected for sockbed)

    = (b + D) + (1 (b + D))

    [(exp

    (C r

    dp

    ))2 + 3i=1

    (ai (r/dp )

    2

    (r/dp )(3+2(i1))+bi

    )]i a b C D1 1.803 0.0479 0.1252 0.0452 1.185 0.3566 3 0.02649 0.001925

  • chemical engineering research and design 9 4 ( 2 0 1 5 ) 549564 553

    Table 2 (Chowdhu

    Distillatio(ASTM D2

    IBP (0.5 vol.50 vol.%: 33FBP (99.5 vo434 CSulfur: 16 2Nitride: 218

    CFD and radial dispusing moledynamics consideringtive TBR pegood agreethe same mused in thconsidered

    2.3.1. ReaThe last teduction of ireaction raHDA reactio

    Ar S + 2H2

    Poly-aroma

    Di-aromati

    Mono-arom

    The reashould be nrate constakMono) canet al., 2002)

    2.4. Ene

    In the curreuated for tpacked bedent researcpractical reshould be le(2001) stateisothermalIn this wolent diamedoes not saied TBR. Futhe producels is quitecurrent stubed is necereactor pereach phasemultiphase

    khk)

    k

    i=r(Q

    , , heat

    term

    Ga correweennge

    at trishem

    2.18

    hG,LkG

    dh is

    4 s(1

    Liquwed tor. In ndits ans:

    (1

    fe

    Waansf

    senariaunt

    part

    = Re

    ULDiesel oil (liquid phase) specicationsry et al., 2002).

    n curve887)

    Components xwt 100 (wt%)

    %): 180 C Monoaromatics 17.967 C Diaromatics 8.77l.%): Triaromatics 1.64

    98 ppm Tetraaromatics 0.95 ppm Naphthenes 19.25

    Parafn 49.78

    experimental data for prediction of axial andersion. Gunjal and Ranade (2007) indicated thatcular diffusion coefcient with appropriate hydro-model (EulerianEulerian multiphase model and

    porosity distribution) can properly describe reac-rformance. The results of their CFD model showedment with experimental data. In the current work,odel proposed by Gunjal and Ranade (2007) was

    e CFD model. Furthermore, wetting effects were on the CFD model to improve model accuracy.

    ction kineticsrm in Eq. (8), Rk,i, stands for consumption or pro-th species due to chemical reactions. The followingte expressions were used to describe the HDS andns (Chowdhury et al., 2002):

    Aromat + H2S (9)

    tics + H2 Di-aromatics (10)

    cs + 2H2 Mono-aromatics (11)

    atics + 3H2 Naphthene (12)

    ctions rate equations are tabulated in Table 3. Itoticed that HDA reactions are reversible and thents for the backward reactions (kPoly, kDi and

    be estimated by vant Hoff equation (Chowdhury.

    rgy conservation equation

    nt work, the non-isothermal conditions were eval-he reactor CFD model. The criteria for study a

    at isothermal condition were reported by differ-hers. Doraiswamy and Tajbl (1974) showed that theactor diameter-to-particles diameter ratio (dR/dp)ss than 4 for ignoring radial temperature. Carberryd that the maximum value for assumption of

    t(k

    +

    wherephase source

    2.4.1. A newfer betwide raerties and Ha

    NuGL =

    NuGL =

    where

    dh = a

    2.4.2. Boelhoof solireactoting cothe gafollow

    NuGS =

    NuLS =

    2.4.3. Heat tricantlydata, Mto accobed toas

    Nuwall temperature in the radial direction is dR/dp < 6.rk, the ratio of dR/dp based on catalysts equiva-ter is about 13.6 (dR = 19 mm and dp = 1.4 mm) thattisfy criteria of isothermal operation for the stud-rthermore, according to Carberry and White (1969)t yield in the packed bed reactors using 2D mod-

    sensitive to temperature effects. Therefore, in thedy, evaluation of temperature distribution in thessary to achieve the best results for prediction offormance. The energy conservation equation for

    (gas, liquid and solid) based on EulerianEulerian approach can be developed in the following form:

    ReLsu =

    where ULc abed and pa

    3. CF

    In this stuexperimendeveloped diction of hhydrodesul + (kkUkhk) = (k Tk) + k uk kPkt

    rk + mrkhrk mkrhkr) + Sk (13)

    Qrk and S are thermal conductivity, viscosity, inter- transfer according to Eqs. (14) and (18) and the

    due to reactions heat, respectively.

    sliquid interphase heat transferlation was used to account interphase heat trans-

    gas and liquid phases. The correlation covers the of operational conditions and uids physical prop-ckle ow regime with the following form (Heidariabadi, 2013):

    5(

    ReGReL

    )0.378(PrGPrL

    )0.499Eo0.627 (14)

    dh (15)

    the packed bed hydraulic diameter and dened as

    ) (16)

    uidsolid and gassolid interphase heat transferr et al. (2001) derived a correlation to evaluate rate

    liquid heat transfer at fully wetted trickle bedthe current study, due to catalysts partial wet-ion, their correlation was modied to account ford liquid phases heat transfer with solid phase as

    fe) 0.111Re0.8G Pr1/3G (17)

    0.111Re0.8L Pr1/3L (18)

    ll to bed heat transferer from the reactor wall to the bed at TBRs is signif-sitive to the liquid ow rate. Based on experimentalni et al. (2003) proposed the following correlation

    wall to bed Nusselt number, Nuwall, for differenticle diameter aspect ratios (4.7, 8.2, 17.2 and 34.3)

    0.76Lsu

    Pr1/3L (19)

    c dp

    L(20)

    nd dp are liquid supercial velocity at core of therticle equivalent diameter, respectively.

    D model and numerical algorithms

    dy, as a benchmark, Chowdhury et al. (2002)tal work was used to consider the effect ofinterphase heat transfer correlations in the pre-ydrotreating reactions conversions. They studiedfurization and hydrodearomatization of diesel oil

  • 554 chemical engineering research and design 9 4 ( 2 0 1 5 ) 549564

    Table 3 Hydrodesulfurization and hydrodearomatization reactions kinetics.

    Reactions (Chowdhury et al., 2002) Heat of reactions (HR) (Stanislaus andCooper, 1994; Chowdhury et al., 2002)

    HDS RAr S = ksC

    1.6Ar S

    C0.56H2

    1+50,000CH2S67 000

    HDA (polyaromatics) RPoly = kPolyCPoly + kPolyCDi 66 462HDA (diaromatics) RDi = kDiCDi + kDiCMono 133 024HDA (monoaromatics) RMono = kMonoCMono + kMonoCNaph 199 203

    in a tubular reactor with 500 mm length and 19 mm diame-ter, Fig. 1. The reactor consisted of two non-reactive zones attwo ends lled by inert particles with 0.2 mm average diame-ter. In the reactive zone, the trilobe catalysts (1.6 mm averagediameter and 3.5 mm average length) and inert spherical par-ticles were mixed with ratio of 1:1.25 (vol/vol). In this study, theequivalent diameter of particles was evaluated about 1.4 mmon the basis of Sauter Mean Diameter (SMD) of trilobe cata-lysts. The more detail about effect of particles different size onCFD model results of Chowdhury et al. (2002) reactor was pre-sented by Gunjal and Ranade (2007). The thermal conductivityof the solid phase on the basis of volumetric average of cat-alysts and inert particles thermal conductivity was evaluatedabout 58 W/m K. The reactor wall temperature was controlledby thermal system to prevent high increase in reactor temper-ature. Table 4 shows 18 different operational conditions usedat CFD simulations. The boundary conditions at the inlet zoneare equal with operational conditions, Table 4; furthermore,

    the wall temperature is equal with the bed inlet temperature.Fig. 1 shows other boundary conditions of the CFD model. Theinitial conditions of the simulations were adjusted based onTable 4 data. In order to reduce divergence at the beginning ofthe simulation, the initial hydrogen concentration in the gasphase was dened to zero.

    Due to reactor shape and uniform ow through the bed,a two dimensional (2D) axisymmetric computational domainwas used for CFD simulations, Fig. 1. Governing equations(Eqs. (1), (2), (8) and (13)) were solved by nite volume method(Patankar, 1980). Semi-Implicit Method for Pressure LinkedEquations (SIMPLE) was implemented for evaluating pressureand velocity coupling in the computational domain. In orderto achieve more accurate simulations, all of the convectiveterms in the transport equations were discretized by QUICK(Quadratic Upwind Interpolation for Convective Kinematics)scheme (Patankar, 1980). The simulations were performed atunsteady state conditions with time step of 0.01 s. The energyFig. 1 Reactor dimensions, computational domain and boundary conditions.

  • chemical engineering research and design 9 4 ( 2 0 1 5 ) 549564 555

    Table 4 Operational and boundary conditions in the CFD simulation.

    Parameter Cases Tinlet, Twall (K) P (MPa) LHSV (h1) QG,NTP/QL xvH2S

    Temperature

    sim-T1 573 4 2 200 0.014sim-T2 593 4 2 200 0.014sim-T3 613 4 2 200 0.014sim-T4 653 4 2 200 0.014

    Pressuresim-P1 593 2 2 200 0.014sim-P2 593 4 2 200 0.014sim-P3 593 8 2 200 0.014

    LHSVsim-L1 593 4 sim-L2 593 4 sim-L3 593 4

    Gas ow rate

    sim-Q1 593 4 sim-Q2 593 4 sim-Q3 593 4 sim-Q4 593 4

    Hydrogensuldeconcentration

    sim-H2S1 593 4 sim-H2S2 593 4 sim-H2S3 593 4

    equations prevent nuulations weand averagof solutionsimultaneosteady stattions were 960 (8M Ca

    4. Re

    4.1. CFD

    The hydrodin the authconcisenesIn order totions that rdeveloped,tures (sim-in a model

    Fig.

    rmalith

    relat (Che dev

    with

    Me

    diffeor Ce nnd nondentism m

    denson th

    Mosim-H2S4 593 4

    relaxation factors were dened small enough tomerical instability during the solution. The sim-re terminated when the bed average temperaturee mass fraction of sulfur components at the entire

    domain have no signicant change with the time,usly. The average time that was required to reache conditions was about 10 000 s. All of the simula-done on a PC with 8 GB RAM and Intel CoreTM i7che, 3.20 GHz) processor.

    sults and discussion

    model validation

    ynamic correlations of CFD model were validatedors previous work (Salimi et al., 2013) and for

    s, validation procedure was not presented here. validation of the CFD model based on the condi-eactions kinetics by Chowdhury et al. (2002) were

    the CFD simulations at different inlet tempera-

    isothetrast wmean resultsthat thmance

    4.2.

    Three mesh fing thtime aindepetype saoptimumesh effect

    4.3.

    T1, sim-T2, sim-T3 and sim-T4, Table 4) were done

    with uniform porosity, fully wetted particles and

    2 Validation of HDS reactor CFD model.

    In this invbed non-isoting effect w(ks,Poly,Di,Monkinetics. Aeffects are tion rates wcalculated (2011) poinevaluation tions shou

    Table 5 operation

    Mesh typ

    1 2 3 1 200 0.0142 200 0.0144 200 0.014

    2 100 0.0142 200 0.0142 300 0.0142 500 0.014

    2 200 02 200 0.0142 200 0.032 200 0.08

    bed. Fig. 2 depicts CFD simulations results in con-Chowdhury et al. (2002) experimental data. Theive error of 5.5% between CFD and experimentalowdhury et al., 2002) was obtained which showseloped CFD model can predict HDS reactor perfor-

    appropriate accuracy.

    sh independency

    rent mesh sizes were examined to nd optimumFD simulations. In this sensitivity analysis, achiev-al HDS reaction conversion at the minimum run

    signicant change on its value were used as meshcy criteria. As shown in Table 5, the second mesh

    es the mentioned criteria and was selected as theesh for numerical simulations. Additionally, the

    ity was rened near the bed wall to account walle radial porosity distribution, Fig. 1.

    died reaction kineticsestigation, the inuence of porosity distribution,thermal conditions and also particles partial wet-ere accounted in the reactions kinetics constants

    o Table 3) to nd intrinsic HDS and HDA reactionsccording to Mederos et al. (2009a) if temperatureneglected in the reactor model, it leads to the reac-ith several orders of magnitude greater than thoseat non-isothermal conditions. Also, Ranade et al.ted out that for development of rate equations andof kinetic parameters, all aspect of reactor condi-ld be considered in the mathematical model. To

    Mesh independency results at sim-T1al condition of Table 4.

    e Cellnumbers

    HDS reactionconversion (%)

    Run time(h)

    9000 40 1022 000 42.6 4588 000 42.7 115

  • 556 chemical engineering research and design 9 4 ( 2 0 1 5 ) 549564

    Table 6 New values of reactions kinetic constants fornon-isothermal conditions and partially wettedcatalysts.

    Reaction Isothermal andfully wettedcatalysts

    (Chowdhury etal., 2002)

    Non-isothermaland partially

    wetted catalysts

    HDS (ks), Eq. (9) 2.5 1012 7.166 1012HDA (polyaromatics

    kPoly), Eq. (10)2.66 105 4.23 105

    HDA (diaromatics kDi),Eq. (11)

    8.5 102 1.33 103

    HDA (monoaromatics kMono), Eq. (12)

    6.04 102 9.62 102

    this aim, new values of reactions constants were estimatedby comparing of HDS conversion obtained from numerical(this work) and experimental data (Chowdhury et al., 2002)at different operational temperatures (sim-T1, sim-T2, sim-T3and sim-T4, Table 4). The best values were evaluated by usingbisection mconstants f

    4.4. Fee

    Fig. 3 showperatures (can be inferments withrelative errmodel withabout 1.8%reactions koperationaties of the H

    4.4.1. TemThe axial teHDA reactipresented zone, the bheat by theperature in

    Fig. 3 HDtemperatur

    Fig. 4 Axthe total reTable 4.

    HDS reaction conversion at different operationalres (sim-P1, sim-P2 and sim-P3, Table 4).

    imately at z = 0.31 m and then begins to decreaseh the bed. The reason of reduction in the bed temper-after z = 0.31 m can be found at reactor constant wallrature which was xed at 613 K. When the reactantsthe reactive zone, great amounts of thermal energy

    due to high reaction rates at the beginning of reactivet the same time, increase in the bed temperature accel-

    reactions rates and the bed temperature rises much On the other hand, the bed constant wall temperatureethod. Table 6 presents the new values of reactionsor the HDS and HDA reactions.

    d inlet temperature effect

    s HDS reaction conversion at different inlet tem-sim-T1, sim-T2, sim-T3 and sim-T4, Table 4). As itred, the CFD simulation results show proper agree-

    experimental data (Chowdhury et al., 2002). Theor between the experimental and optimized CFD

    new reactions constants (Section 4.3) was found. Therefore, it can be concluded that the modiedinetics can be used to study the effects of differentl parameters on the performance and local proper-DS reactor, more accurately.

    perature distributionmperature distribution and total heat of HDS andons at sim-T3 operational condition (Table 4) arein Fig. 4. As shown in this gure, in the reactiveed temperature increases rapidly due to generated

    hydrotreating reactions (Eqs. (9)(12)). The tem-creases to the highest value (a hot spot with 623 K)

    Fig. 5 pressu

    approxthrougature tempeenter releasezone. Aeratesfaster.S reaction conversion at different inletes (sim-T1, sim-T2, sim-T3 and sim-T4, Table 4).

    which was shown in Fafter the inin the reactthe heat thperature derate of redz > 0.4 m, wenergy equ

    4.5. Th

    Fig. 5 depicsim-P2 andtion. The rial temperature distribution (at r = 0) and heat ofactions through the bed at sim-T3 conditions,set at 613 K eliminates generated heat in the bed. Asig. 4, the total heat of reactions begins to decreaselet part of reactive zone which is due to reductionants concentrations. On the other hand, removingrough the reactors wall nally causes the bed tem-creases through the bed at z > 0.31 m. Lastly, the

    uction in the temperature occurs more rapidly athere the reactive zone reaches to its end and theation (Eq. (13)) is solved with no heat source terms.

    e bed operational pressure effect

    ts the effect of the bed operational pressure (sim-P1, sim-P3, Table 4) on the conversion of HDS reac-esults show a good agreement between reported

  • chemical engineering research and design 9 4 ( 2 0 1 5 ) 549564 557

    Fig. 6 (A) mass fractiat constantsim-P1 con

    experimenlation resuat constantimproves win the convwhen the opresses anthe same tliquid phasbed, whichresidence thydrogen cfore the hygives rise toa result, grohigh operat

    4.5.1. PerFig. 6 compwall tempeture and sucondition. observed bperature anabout 16 K

    HDS reaction conversion at different LHSV values1, sim-L2 and sim-L3, Table 4).

    d condition compared to constant wall temperature,ratestly, com

    reacFig. 7 (sim-L

    isolateaccelesequensulfur than aAxial temperature distribution (at r = 0) and (B)on of desulfurized components through the bed

    wall temperature and isolated wall in theditions, Table 4.

    tal data (Chowdhury et al., 2002) and CFD simu-lts with 4.9% relative error. As it can be observed,

    gas and liquid ow rates, the reaction conversionith an increase in the bed pressure. Enhancementersion rate can be caused by two reasons; rst,perational pressure increases, the gas phase com-d thus its velocity decreases through the bed. Atime, the momentum interaction between gas andes causes the liquid velocity to reduce through the, in turn, brings about increase in the liquid phaseime. Second, in the high reactor pressures, theoncentration increases in the gas phase and there-drogen interphase mass transfer enhances which

    more conversion in the liquid phase reactants. Aswth in the HDS conversion rate can be observed ational pressures.

    formance of adiabatic reactorares the effect of isolated reactor wall and constantrature on axial distribution of liquid tempera-lfur components concentration at sim-P1 (Table 4)According to Fig. 6A, an ascending difference isetween temperature values at constant wall tem-d isolated conditions with maximum difference

    . The more increase in the bed temperature at

    version for78% while iture condit

    As Fig. decreases constants ain the liquiliquidsolidphase heatspecic timlibrium. Asobserved in

    4.6. Eff

    Fig. 7 showSupercial ric ow rate

    Fig. 8 Axand sim-L3 the reactions rates and reactants conversion. Con-as it can be seen in Fig. 6B, the mass fraction ofponents in the adiabatic reactor decreases moretor with constant wall temperature. The nal con-

    HDS reaction in the adiabatic condition was aboutt is obtained about 69% at constant wall tempera-ion.6A illustrates, the liquid phase temperature

    about 0.5 K after the reactive zone and remainsfter z=0.41 m. The reason of this small reductiond temperature can be found in the liquidgas and

    interphase heat transfer rates. In effect, the inter- transfer is not a phenomenon with innite rate ande is required that phases reach the thermal equi-

    a result, after the reactive zone a small decrease is the liquid phase temperature.

    ect of LHSV on HDS reaction conversion

    s HDS reaction conversion versus Liquid HourlyVelocity (LHSV LHSV is ratio of liquid volumet-

    per reactor volume) at sim-L1, sim-L2 and sim-L3ial temperature distribution at r = 0 in the sim-L1 conditions, Table 4.

  • 558 chemical engineering research and design 9 4 ( 2 0 1 5 ) 549564

    Fig. 9 Totvalues (sim

    conditions,and experiwhich showimental resincreases, chigher valuand accordThe differeLHSV valuemass transinterphaseincrease inmass transtration in tseems thatreal conditdiction in Hto experim

    In Fig. is depictedTable 4). Ashows highof such behand subseqL3. Accordthe sim-L1case, sim-L

    Fig. 10 H (sim

    tive ot pon theoughf cond th-L3.

    the nce t

    HDllusteter se inreasetwe2002lative

    of d dueard rs ovbed non-isothermal condition and vant Hoff equation,

    Fig. 11 (Aal HDA reaction conversion at different LHSV-L1, sim-L2 and sim-L3, Table 4).

    Table 4. The relative error between CFD resultsmental data (Chowdhury et al., 2002) is about 3.8%s good agreement between numerical and exper-ults. As it can be inferred from Fig. 7 when LHSVonversion of HDS reaction decreases. Actually, ates of LHSV the liquid residence time decreaseingly, reaction conversion reduces in the reactor.nce between CFD and experimental data at highs can be explained by effect of hydrogen interphasefer rate on HDS reaction conversion. According to

    mass transfer coefcient term, KL,iaL (Table 1), an the liquid velocity enhances the rate of interphasefer. Therefore, at high LHSVs the hydrogen concen-he liquid phase increases. Based on CFD results, it

    hydrogen concentration is predicted more thanions at higher LHSVs and consequently, more pre-DS conversion occurs at numerical results respectental data.8, axial distribution of temperature in the TBR

    at two different conditions (sim-L1 and sim-L3,s shown, the temperature of the bed at sim-L1er values in comparison with sim-L3. The reasonavior can be found in more conversion at sim-L1uently, more heat generation compared to sim-

    values

    of reachot sptime. Iity thrmost ozone athe simmovesreside

    4.6.1. Fig. 9 iparamincreato dection beet al., age rereasoncan bebackwreduceing to ing to Fig. 8, the bed maximum temperature in condition occurs at z = 0.35 m while in the other3, the highest temperature happens at the outlet

    at high temas forwardconversion

    ) Radial temperature distribution at z = 0.4 m and (B) contour of teDS reaction conversion at different gas ow rate-Q1, sim-Q2, sim-Q3 and sim-Q4, Table 4).

    zone, z = 0.4 m. The cause of this difference in thesitions can be found in the liquid phase residence

    sim-L1, the liquid phase moves with lower veloc- the bed than that of sim-L3. As a result, in sim-L1nversion happens at the entrance of the reactivee hot spot occurs closer to the inlet compared with

    In contrast, the higher velocity at sim-L3 conditionhot spot to the end of the reactive zone due to lowime of reactants.

    A reactions conversionrates the effect of liquid velocity in terms of LHSVon the total HDA conversion. As the gure shows,

    the LHSV values reduces reaction conversion due in reactants residence time. Moreover, the devia-en numerical and experimental data (Chowdhury) is high and CFD results show about 45% aver-

    error with respect to the experimental data. Theeviation between CFD and experimental results

    to inuence of temperature on the rate of HDAeactions which damps forward reactions rates anderall conversion of aromatic components. Accord-peratures HDA backward reactions become as fast reactions. Therefore, reduction in the total HDA

    can be observed in the numerical results.

    mperature in the sim-Q1 condition (Table 4).

  • chemical engineering research and design 9 4 ( 2 0 1 5 ) 549564 559

    4.7. Gas ow rate effect

    In Fig. 10, gas ow raAccording tHDS reactiment can brelative erret al., 2002)

    Increaseuid phase subsequenTherefore, higher gas Such inconvelocities oIn comparities the bedcondition. Cenhances bhydrogen cin Fig. 10, mgas ow rat

    The radzone (z = 0.4nal cross s(Table 4). Aaffected bythe porositin the vicinporosity, duphase masslocations aTable 1. Contration in treactions. in the catations leadstemperaturof the localysts volumand the beity positionwhere the perature shheat transfity to the r(Twall = 593

    In Fig. 11Q1 conditioapproximaposition ofthe bed pothe center and r < 4 mboring locato reduce treactor engcess.

    4.8. Eff

    Fig. 12 illussulde con

    HDS reaction conversion at different hydrogen concentrations (sim-H2S1, sim-H2S2, sim-H2S3 and2S4, Table 4).

    S3 and sim-H2S4, Table 4). The inuence of increase gas phase hydrogen sulde can be seen in the formuction of hydrogen concentration in the gas phaserrespondingly, in the liquid phase. Actually, at lower

    (A) Reactive zone average temperature and (B) totalenerated by HDS and HDA reactions (sim-H2S1,2S2, sim-H2S3 and sim-H2S4, Table 4).the HDS reaction conversion is plotted againstte (sim-Q1, sim-Q2, sim-Q3 and sim-Q4, Table 4).o the gure, increase in the gas ow rate improveson conversion, although no remarkable enhance-e observed. The numerical results show about 6.4%or versus available experimental data (Chowdhury.

    in the gas velocity affects directly on the liq-velocity due to interphase momentum coupling;tly, reactants residence time reduces in the bed.it is expected that reaction conversion decreases atow rates while the reverse behavior is observed.sistency can be explained by inuence of higher gasn increase of hydrogen interphase mass transfer.son with low velocity condition, at higher veloci-

    hydrogen concentration remains close to the inletonsequently, the rate of interphase mass transferetween gas and liquid phases which causes moreoncentration in the liquid phase. Hence, as shownore conversion in the HDS reaction occurs at higheres.ial temperature distribution at the end of reactive

    m) and the contour of temperature on longitudi-ection are shown in Fig. 11 for sim-Q1 conditions Fig. 11A suggests, the temperature prole is

    the bed porosity distribution so that wherevery is high, the bed temperature increases exceptity of the wall. In the locations with high locale to more velocity of liquid phase, hydrogen inter-

    transfer increases about 4% respect to neighboringccording to interphase mass transfer correlation ofsequently, enhancement in the hydrogen concen-he liquid phase increases rates and heat of theOn the other hand, it is expected that decreaselysts volume fraction at high local porosity posi-

    to reduction of the reactions rates and the bede. As a result, it can be concluded that formationl hot spots due to small changes in the cata-e fraction is affected by interphase mass transferd temperatures increases in the high local poros-s far from the wall. In the vicinity of the wall,bed porosity is at the highest value, the bed tem-arply decreases. In this region, the maximumer from the bed occurs because of the proxim-eactor wall which is set at constant temperatureK).B, the contour of bed temperature is shown at sim-n. According to this gure, the hottest region is

    tely formed at the end of the reactive zone. The hot spot depends on velocity of the phases androsity distribution. It forms in the location nearof the bed axis located in the range of r > 2 mmm where the local porosity is more than neigh-tions. The estimation of hot spot position helpshermal runaway problems in the bed and enablesineers to have an appropriate control over the pro-

    ect of gas phase H2S concentration

    trates the HDS conversion at different hydrogencentrations of the gas phase (sim-H2S1, sim-H2S2,

    Fig. 12suldesim-H

    sim-H2in theof redand co

    Fig. 13heat gsim-H

  • 560 chemical engineering research and design 9 4 ( 2 0 1 5 ) 549564

    Fig. 14 Mass fraction of Poly-, Di- and Mono-aromaticsalong the bed at sim-H2S1 conditions, Table 4.

    values of hydrogen mole fraction in the gas phase, rateof interphase mass transfer reduces which causes smallerhydrogen concentration in the liquid phase and thereforereduction in the conversion of reactions. As it is depictedin Fig. 12, the current CFD model which considers for the

    Fig. 15 (Athe CFD prsim-P2 andtemperatur

    effect of interphase heat transfer, provides a better matchto the expresults (GuThe averagtal data (Chstudy and between nof HDS coof adsorptperature inreaction kismall chancient has nmain reasoical resultsHDS reactiocentrations

    Fig. 13 zone and hto Fig. 13Awith incredue to decheat by theshould be dthe releasecomputed,heat by theure shows fraction intion decrearate, the retrend at hiconcentratconsumptiHDA reactitration, thedeclines wsumption bhydrogen c

    utioses. Teat bso aperaghydrcontribincreaated hand albed avferent ) Effect of gasliquid interphase heat transfer inediction of HDS reaction conversion (sim-P1,

    sim-P3, Table 4) and (B) axial (r = 0) bede at sim-P3 conditions.

    4.8.1. DisfractionIn Fig. 14,matic comsim-H2S1 cand di-aromono-arominlet of thereactions kthe case osion of di-increase inoccurs alonAfter that,tion and cofraction ofreactor.erimental data in comparison with other CFDnjal and Ranade, 2007) at isothermal conditions.e of relative error between CFD and experimen-owdhury et al., 2002) is about 13% for the current21% for Gunjal and Ranade (2007). The deviationumerical and experimental results at predictionnversion might be attributed to independencyion coefcient of hydrogen sulde to the tem-

    the HDS reaction kinetics and also developednetics constants. However, it seems that due toges in the bed temperature, adsorption coef-o signicant effect on the errors. Therefore, then of deviation between experimental and numer-

    might be related to the undesired dependency ofn kinetic from hydrogen and hydrogen sulde con-.depicts the average temperature of the reactiveeat of HDS and total HDA reactions. According, the temperature shows no signicant changesase in the hydrogen sulde concentration whilerease in HDS conversion (Fig. 12) and produced

    HDS reaction, it seems that the bed temperatureecreased. To explain constant temperature trend,d heat by HDS and total HDA reactions should be

    simultaneously. In Fig. 13B, values of generated reactions are shown at individual bars. The g-that with increase in the hydrogen sulde volume

    the gas phase, the produced heat by HDS reac-ses. In contrast, due to reduction in HDS reactionleased heat by total HDA reactions takes ascendinggher hydrogen sulde concentrations. At the highion of hydrogen in the bed, the rate of hydrogenon by the HDS reaction is considerably more thanons. On the other hand, at low hydrogen concen-

    rate of hydrogen consumption by HDS reactionhich causes growth in the rate of hydrogen con-y the HDA reactions. With more decrease in theoncentration (sim-H2S1 to sim-H2S4, Table 4), then of HDA reactions to consumption of hydrogenherefore, as seen in Fig. 13B, due to higher gener-y HDA reactions in comparison with HDS reactionproximately constant overall produced heat, the

    e temperature shows no signicant changes at dif-ogen sulde concentration, Fig. 13A.

    tribution of aromatic components axial mass

    distribution of average mass fraction of aro-ponents is shown along the reactor length atonditions. As observed, mass fractions of poly-

    matics components continuously decrease whileatics mass fraction shows a maximum at the

    reactive zone. According to Eqs. (10)(12), HDAsinetics are presented by a chain mechanism. Inf mono-aromatics reaction, due to high conver-aromatics species respect to mono-aromatics, an

    the concentration of mono-aromatics componentsg the bed with a maximum point at z = 0.17 m.

    due to reduction in the di-aromatics concentra-nsumption of mono-aromatics components, mass

    mono-aromatics begins to decrease through the

  • chemical engineering research and design 9 4 ( 2 0 1 5 ) 549564 561

    Fig. 16 Co e heheat transf

    4.9. Gasimulation

    In order toeffect in thtemperatursim-P2 andgasliquid case of noversion preconditions.and liquid pto only excheat loss inalong the bsion showstemperaturinterphaseAs it can bethermal intture distribchange in tas interphatants convdistributioninterphaseTBR perform

    5. Co

    CFD simulby means main objenon-isotheinterphase

    ansfr moonve

    CFD

    reaclopebest ntour of the bed temperature: (A) with gasliquid interphaser at sim-P3 conditions, Table 4.

    sliquid interphase heat transfer effect onresults

    demonstrate gasliquid interphase coefciente prediction of HDS reaction conversion and bede distribution, the CFD simulations at sim-P1,

    sim-P3 (Table 4) were repeated with eliminatinginterfacial heat transfer. As Fig. 15 shows, in the

    heat trtransfetions con the

    Thedevethe gasliquid heat transfer, Fig. 15A, the HDS con-diction is about 5% higher than non-isothermal

    In effect, neglecting the heat transfer between gashases causes the reaction heat in the liquid phasehange with the solid phase. Thus, due to lower

    the liquid phase, the phase temperature increasesed (Fig. 15B) which leads to the reaction conver-

    higher values than expected. In Fig. 16, contour ofe is shown with and without effect of gasliquid

    heat transfer during the CFD simulation at sim-P3. inferred from the gure, neglecting the gasliquideraction affects signicantly on the local tempera-ution, so that the location and values of hot spotshe bed. Subsequently, the entire phenomena suchse mass transfer, phases properties and local reac-ersion are affected by difference at temperature. Therefore, it is safe to say that utilizing gasliquid

    heat transfer signicantly improves the accuracy ofance prediction.

    nclusion

    ations of hydrotreating TBR were carried outof EulerianEulerian multiphase approach. Thective of this work was the CFD simulation ofrmal reactor by implementation of appropriate

    heat transfer coefcient. The modied solid-uids

    reactionsand realcomprom

    It was shby the beat positiomaximuthis zonetemperat

    At the adifferencwas showmaximumtemperatment wacondition

    Finally, iphase heconversioThereforinterpha

    Acknowle

    The author(POGC - Islthis work (Gat transfer and (B) without gasliquid interphase

    er coefcient and a new gasliquid interphase heatdel were developed to accurately predict the reac-rsions and local properties through the bed. Based

    simulations, the following results can be inferred:

    tion kinetics constants were optimized based ond CFD model. The optimization was done to ndvalues that present intrinsic chemical behavior of on the basis of appropriate mathematical model conditions of studied TBR. Results showed goodise between numerical and experimental results.

    own that the radial temperature of the bed affectedd porosity and the highest temperature occurredns near the bed center. Results showed that the

    m heat loss occurs at the vicinity of the wall; in, high velocities of the phases and constant wallure cause the maximum heat loss in the bed.diabatic and constant wall temperature, a largee was observed between the bed temperatures. Itn that there is a difference about 16 K between

    bed temperatures at adiabatic and constant wallure conditions. In addition, about 16% improve-s observed in HDS reaction conversion at adiabatics.

    t was shown that neglecting the gasliquid inter-at transfer affects on the prediction of reactionsn and local parameters such as bed temperature.

    e, using appropriate model to describe gasliquidse heat transfer seems inevitable.

    dgment

    s would like to thank Pars Oil and Gas Companyamic Republic of Iran) for the nancial support ofrant No. 91-203/TP).

  • 562 chemical engineering research and design 9 4 ( 2 0 1 5 ) 549564

    Reference

    Ahmed, T.HApplicatHouston

    Al-Dahhan,efciencSci. 50, 2

    Alvarez, A., differentgasoil hy

    Alvarez, A., quench sreactors

    Atta, A., Roymaldistrconcept

    Atta, A., Royand liqupermeab

    Atta, A., RoyapproachhydrodyChem. E

    Attou, A., Feregime tEng. Sci.

    Attou, A., Femodel fococurren491511.

    Aydin, B., Laregime tand a no

    Aydin, B., La(non-)NeChem. E

    Babu, B.V., Strickle be

    Baussaron, Boyer, C.AIChE J.

    Baussaron, Delmas,measureEng. Che

    Bazmi, M., Hand expetrickle becatalysts

    Bazmi, M., Hand expethrough Mass Tra

    BoelhouwerParticle-Eng. Sci.

    Borremans,and partinuenc1403141

    Boyer, C., Votrickle befor pressexperim

    Brkljac, B., B2007. Moperiodicliquid-ph7011701

    Carberry, J.J.Dover Pu

    Carberry, J.J., White, D., 1969. On the role of transport phenomenaataly

    Z.-M2. Hyting 294.hurytor mel. Aamy

    . Rev.Pintoluatioackenow., Smdup a, A., 6. CFtorsm. E, A., 7. CFackem. E, P.R.,mer. Sci., P.R.,ribut

    sim N., Fde, Ah con

    modi, A., liquacke, 674i, A., hermrodem. Ean, Dcessfg lab

    8715., Grakle-dels. ., Larkle-cat., Larliquct. En., Larparti609.., Larparti. Des, A.T

    matirode, A.T

    n indle crd Eurineer, A.Trotres

    ., 2007. Equations of State and PVT Analysis:ions for Improved Reservoir Modeling. Gulf Pub.,, TX.

    M.H., Dudukovic, M.P., 1995. Catalyst wettingy in trickle-bed reactors at high pressure. Chem. Eng.3772389.Ancheyta, J., 2008. Simulation and analysis of

    quenching alternatives for an industrial vacuumdrotreater. Chem. Eng. Sci. 63, 662673.Ancheyta, J., Munoz, J.A.D., 2007. Comparison ofystems in commercial xed-bed hydroprocessing

    . Energy Fuels 21, 11331144., S., Nigam, K.D.P., 2007a. Investigation of liquidibution in trickle-bed reactors using porous mediain CFD. Chem. Eng. Sci. 62, 70337044., S., Nigam, K.D.P., 2007b. Prediction of pressure dropid holdup in trickle bed reactor using relativeility concept in CFD. Chem. Eng. Sci. 62, 58705879., S., Nigam, K.D.P., 2010. A two-phase Eulerian

    using relative permeability concept for modeling ofnamics in trickle-bed reactors at elevated pressure.ng. Res. Des. 88, 369378.rschneider, G., 1999. A two-uid model for owransition in gasliquid trickle-bed reactors. Chem.

    54, 50315037.rschneider, G., 2000. A two-uid hydrodynamicr the transition between trickle and pulse ow in at gasliquid packed-bed reactor. Chem. Eng. Sci. 55,

    rachi, F., 2005. Trickle bed hydrodynamics and owransition at elevated temperature for a Newtoniann-Newtonian liquid. Chem. Eng. Sci. 60, 66876701.rachi, F., 2008. Trickle bed hydrodynamics forwtonian foaming liquids in non-ambient conditions.ng. J. 143, 236243.hah, K.J., Govardhana Rao, V., 2007. Lateral mixing ind reactors. Chem. Eng. Sci. 62, 70537059.

    L., Julcour-Lebigue, C., Wilhelm, A.-M., Delmas, H.,, 2007a. Wetting topology in trickle bed reactors.53, 18501860.L., Julcour-Lebigue, C., Wilhelm, A.M., Boyer, C.,

    H., 2007b. Partial wetting in trickle bed reactors:ment techniques and global wetting efciency. Ind.m. Res. 46, 83978405.ashemabadi, S.H., Bayat, M., 2011. CFD simulationrimental study for two-phase ow through thed reactors, sock and dense loaded by trilobe. Int. Commun. Heat Mass Transf. 38, 391397.ashemabadi, S.H., Bayat, M., 2012. CFD simulationrimental study of liquid ow mal-distributionthe randomly trickle bed reactors. Int. Commun. Heatnsf. 39, 736743., J.G., Piepers, H.W., Drinkenburg, A.A.H., 2001.liquid heat transfer in trickle-bed reactors. Chem.

    56, 11811187. D., Rode, S., Wild, G., 2004. Liquid ow distributionicleuid heat transfer in trickle-bed reactors: thee of periodic operation. Chem. Eng. Process. 43,0.lpi, C., Ferschneider, G., 2007. Hydrodynamics ofd reactors at high pressure: two-phase ow modelure drop and liquid holdup, formulation andental validation. Chem. Eng. Sci. 62, 70267032.ludowsky, T., Dietrich, W., Grunewald, M., Agar, D.W.,delling of unsteady-state hydrodynamics inally operated trickle-bed reactors: inuence of thease physical properties. Chem. Eng. Sci. 62,9., 2001. Chemical and Catalytic Reaction Engineering.blications, Mineola, NY.

    in cCheng,

    201wet283

    Chowdreacdies

    DoraiswCat

    Flvio Evain pdow

    Goto, SHol

    Guardo200reacChe

    Guardo200in pChe

    GunjalcomEng

    GunjaldistCFD

    Habtu,Ayuwitand

    Heidargasof p104

    HeidarisothydChe

    HickmSucusin152

    Iliuta, Itricmo

    Iliuta, Itricveri

    Iliuta, IgasRea

    Iliuta, Ifor 597

    Iliuta, Ifor Res

    Jarullahestihyd

    Jarullahof awho22nEng

    Jarullahhydtic reactor behavior. Ind. Eng. Chem. 61, 2735.., Kong, X.-M., Zhu, J., Wang, Z.-Y., Jin, J., Huang, Z.-B.,drodynamic modeling on the external liquidsolidefciency in a trickling ow reactor. AIChE J. 59,

    , R., Pedernera, E., Reimert, R., 2002. Trickle-bedodel for desulfurization and dearomatization of

    IChE J. 48, 126135., L.K., Tajbl, D.G., 1974. Laboratory catalytic reactors.

    Sci. Eng. 10, 177219. Moreira, M., do Carmo Ferreira, M., Freire, J.T., 2006.n of pseudohomogeneous models for heat transfer

    d beds with gas ow and gasliquid cocurrent and upow. Chem. Eng. Sci. 61, 20562068.

    ith, J.M., 1975. Trickle-bed reactor performance. Part I.nd mass transfer effects. AIChE J. 21, 706713.Coussirat, M., Recasens, F., Larrayoz, M.A., Escaler, X.,D study on particle-to-uid heat transfer in xed bed: convective heat transfer at low and high pressure.ng. Sci. 61, 43414353.Coussirat, M., Recasens, F., Larrayoz, M.A., Escaler, X.,D studies on particle-to-uid mass and heat transferd beds: free convection effects in supercritical uids.ng. Sci. 62, 55035511.

    Ranade, V.V., 2007. Modeling of laboratory andcial scale hydro-processing reactors using CFD. Chem.

    62, 55125526. Ranade, V.V., Chaudhari, R.V., 2003. Liquidion and RTD in trickle bed reactors: experiments andulations. Can. J. Chem. Eng. 81, 821830.ont, J., Fortuny, A., Bengoa, C., Fabregat, A., Haure, P.,., Stber, F., 2011. Heat transfer in trickle bed columnstant and modulated feed temperature: experimentseling. Chem. Eng. Sci. 66, 33583368.Hashemabadi, S.H., 2013. Numerical evaluation of theid interfacial heat transfer in the trickle ow regimed beds at the micro and meso-scale. Chem. Eng. Sci.689.Hashemabadi, S.H., 2014. CFD simulation ofal diesel oil hydrodesulfurization and

    aromatization in trickle bed reactor. J. Taiwan Inst.ng. 45, 13891402..A., Holbrook, M.T., Mistretta, S., Rozeveld, S.J., 2013.ul scale-up of an industrial trickle bed hydrogenationoratory reactor data. Ind. Eng. Chem. Res. 52,292.ndjean, B.P.A., Larachi, F., 2002. Hydrodynamics ofow reactors: updated slip functions for the slitChem. Eng. Res. Des. 80, 195200.achi, F., 2002. Hydrodynamics of power-law uids inow reactors: mechanistic model, experimentalion and simulations. Chem. Eng. Sci. 57, 19311942.achi, F., 2005. Modelling the hydrodynamics ofid packed beds via slit models: a review. Int. J. Chem.g. 3, 125.achi, F., Al-Dahhan, M.H., 2000a. Double-slit modelally wetted trickle ow hydrodynamics. AIChE J. 46,

    achi, F., Al-Dahhan, M.H., 2000b. Multiple-zone modelally wetted trickle ow hydrodynamics. Chem. Eng.. 78, 982990.., Mujtaba, I.M., Wood, A.S., 2011. Kinetic parameteron and simulation of trickle-bed reactor forsulfurization of crude oil. Chem. Eng. Sci. 66, 859871.., Mujtaba, I.M., Wood, A.S.,2012a. Economic analysisustrial rening unit involving hydrotreatment ofude oil in trickle bed reactor using gproms. In: Theopean Symposium on Computer Aided Processing. Elsevier, London, pp. 652656.., Mujtaba, I.M., Wood, A.S., 2012b. Whole crude oilating from small-scale laboratory pilot plant to

  • chemical engineering research and design 9 4 ( 2 0 1 5 ) 549564 563

    large-scale trickle-bed reactor: analysis of operational issuesthrough

    Kallinikos, Lthe catalusing a m2444244

    Kam, E.K.T.,hydroproperformaFuels 19,

    Kesler, M.G.Fractions153158.

    Korsten, H.,hydrotre1350136

    Kundu, A., Ncharacte

    Lamine, A.Sa packedliquid. C

    LappalainenImprovepressureEng. Che

    Lappalainenof radialmechani207218.

    Larachi, F., Bmass trarecomm42, 2222

    Lopes, R.J.G.in the caSci. 62, 7

    Lopes, R.J.G.numerichigh-pre112120.

    Lopes, R.J.G.multipha342355.

    Lopes, R.J.G.of tricklemodels uRes. 48, 1

    Lopes, R.J.G.EulerEucatalytic160, 293

    Lopes, R.J.G.multiphawater po

    Lopes, R.J.G.catalyst environmRes. 49, 1

    Lopes, R.J.G.modelinoxidation

    Mapiour, M.Effects ogas oil: e2536254

    Mariani, N.JBarreto, heat tranEng. 81, 8

    Martnez, MM.A., Culiquid di4957.

    Mederos, F.S., Ancheyta, J., 2007. Mathematical modeling andulatintercs, F.Sure id, 119s, F.S

    simuvy crs, F.Samicew. Cs, F.Samictors..L., Dtactin8939zadehulatistant682., J.A.Droqurotre

    C., Vudar

    perfol 86, o, M.ctiva. Eng.ar, S.isph.H., Gdboo, F., Midsrmin

    3380R.M.,eric

    333., V.Vctors.C., Pes andon..R., tions.R.,

    or hysure.E., Mid hosure

    M., H3. Nu

    bodyhnol.eld, C., Krile-up. 14, .U.,

    imiad oil:t-plaaus, lysisr, W.-disturbamodeling. Energy Fuels 26, 629641..E., Bellos, G.D., Papayannakos, N.G., 2008. Study ofyst deactivation in an industrial gasoil HDS reactorini-scale laboratory reactor. Fuel 87,

    9. Al-Shamali, M., Juraidan, M., Qabazard, H., 2005. Acessing multicatalyst deactivation and reactornce modelpilot-plant life test applications. Energy

    753764., Lee, B.I., 1976. Improve Prediction of Enthalpy of, Hydrocarbon Processing. Gulf Pub., Houston, pp.

    Hoffmann, U., 1996. Three-phase reactor model forating in pilot trickle-bed reactors. AIChE J. 42,0.igam, K.D.P., Verma, R.P., 2003. Catalyst wetting

    ristics in trickle-bed reactors. AIChE J. 49, 22532263.., Gerth, L., Le Gall, H., Wild, G., 1996. Heat transfer in

    bed reactor with cocurrent downow of a gas and ahem. Eng. Sci. 51, 38133827., K., Alopaeus, V., Manninen, M., Aittamaa, J., 2008.d hydrodynamic model for wetting efciency,

    drop, and liquid holdup in trickle-bed reactors. Ind.m. Res. 47, 84368444., K., Manninen, M., Alopaeus, V., 2009. CFD modeling

    spreading of ow in trickle-bed reactors due tocal and capillary dispersion. Chem. Eng. Sci. 64,

    elfares, L., Iliuta, I., Grandjean, B.P.A., 2003. Heat andnsfer in cocurrent gasliquid packed beds. Analysis,endations, and new correlations. Ind. Eng. Chem. Res.42., Quinta-Ferreira, R.M., 2007. Trickle-bed CFD studiestalytic wet oxidation of phenolic acids. Chem. Eng.0457052., Quinta-Ferreira, R.M., 2008. Three-dimensionalal simulation of pressure drop and liquid holdup forssure trickle-bed reactor. Chem. Eng. J. 145,

    , Quinta-Ferreira, R.M., 2009a. CFD modelling ofse ow distribution in trickle beds. Chem. Eng. J. 147,

    , Quinta-Ferreira, R.M., 2009b. Numerical simulation-bed reactor hydrodynamics with RANS-basedsing a volume of uid technique. Ind. Eng. Chem.7401748., Quinta-Ferreira, R.M., 2010a. Assessment of CFDler method for trickle-bed reactor modelling in the

    wet oxidation of phenolic wastewaters. Chem. Eng. J.301., Quinta-Ferreira, R.M., 2010b. Evaluation ofse CFD models in gasliquid packed-bed reactors forllution abatement. Chem. Eng. Sci. 65, 291297., Quinta-Ferreira, R.M., 2010c. Numerical studies ofwetting and total organic carbon reaction onentally based trickle-bed reactors. Ind. Eng. Chem.073010743., Silva, A.M.T., Quinta-Ferreira, R.M., 2007. Kineticg and trickle-bed CFD studies in the catalytic wet

    of vanillic acid. Ind. Eng. Chem. Res. 46, 83808387., Sundaramurthy, V., Dalai, A.K., Adjaye, J., 2010.f the operating variables on hydrotreating of heavyxperimental, modeling, and kinetic studies. Fuel 89,3.., Mazza, G.D., Martnez, O.M., Cukierman, A.L.,G.F., 2003. On the inuence of liquid distribution onsfer parameters in trickle bed systems. Can. J. Chem.14820.., Pallares, J., Lpez, J., Lpez, A., Albertos, F., Garca,esta, I., Grau, F.X., 2012. Numerical simulation of thestribution in a trickle-bed reactor. Chem. Eng. Sci. 76,

    simcou

    Mederoens355

    Mederoandhea

    Mederodynrevi

    MederoDynreac

    Mills, Pcon27,

    Mousasimcon675

    MunozMarhyd

    Murali,ChotheFue

    PachecdeaInd

    PatankHem

    Perry, RHan

    PirontiLiqudete379

    Propp, num311

    RanadeRea

    Reid, RGasLon

    Riazi, MFrac

    Riazi, Mvappres

    Ring, Zliqupres

    Salimi,201andTec

    SatterSie, S.T

    ScaEng

    Skala, DRahusepilo

    Stanislcata

    StrasseMaldiston of hydrotreating reactors: cocurrent versusurrent operations. Appl. Catal. A 332, 821.., Ancheyta, J., Chen, J., 2009a. Review on criteria toeal behaviors in trickle-bed reactors. Appl. Catal. A.., Ancheyta, J., Elizalde, I., 2012. Dynamic modelinglation of hydrotreating of gas oil obtained from

    ude oil. Appl. Catal. A 425-426, 1327.., Elizalde, I., Ancheyta, J., 2009b. Steady-state and

    reactor models for hydrotreatment of oil fractions: aat. Rev. Sci. Eng. 51, 485607.., Rodrguez, M.A., Ancheyta, J., Arce, E., 2006.

    modeling and simulation of catalytic hydrotreating Energy Fuels 20, 936945.udukovic, M.P., 1981. Evaluation of liquidsolidg in trickle-bed reactors by tracer methods. AIChE J.04., F., van den Akker, H.E.A., Mudde, R.F., 2012. Eulerian

    on of heat transfer in a trickle bed reactor with wall temperature. Chem. Eng. J. 207208,

    ., Alvarez, A., Ancheyta, J., Rodrguez, M.A.,n, G., 2005. Process heat integration of a heavy crudeatment plant. Catal. Today 109, 214218.oolapalli, R.K., Ravichander, N., Gokak, D.T.,y, N.V., 2007. Trickle bed reactor model to simulatermance of commercial diesel hydrotreating unit.

    11761184.E., Martins Salim, V.M., Pinto, J.C., 2011. Acceleratedtion of hydrotreating catalysts by coke deposition.

    Chem. Res. 50, 59755981.V., 1980. Numerical Heat Transfer and Fluid Flow.ere, London.reen, D.W., 2008. Perrys Chemical Engineersk, 8th ed. McGraw-Hill, New York.izrahi, D., Acosta, A., Gonzalez-Mendizabal, D., 1999.olid wetting factor in trickle-bed reactors: itsation by a physical method. Chem. Eng. Sci. 54,0.

    Colella, P., Crutcheld, W.Y., Day, M.S., 2000. Aal model for trickle bed reactors. J. Comput. Phys. 165,

    ., Chaudhari, R.V., Gunjal, P.R., 2011. Trickle Bed: Reactor Engineering & Applications. Elsevier.rausnitz, J.M., Poling, B.E., 1987. The Properties ofd Liquids, 4th ed. McGraw-Hill, New York;

    2005. Characterization and Properties of Petroleum. ASTM International, W. Conshohocken, PA.

    Faghri, A., 1985. Thermal conductivity of liquid anddrocarbon systems: pentanes and heavier at lows. Ind. Eng. Chem. Process Des. Dev. 24, 398401.issen, R.W., 1991. Trickle-bed reactors: tracer study ofldup and wetting efciency at high temperature and. Can. J. Chem. Eng. 69, 10161020.ashemabadi, S.H., Noroozi, S., Heidari, A., Bazmi, M.,merical and experimental study of catalyst loading

    effects on a gasliquid trickle-ow bed. Chem. Eng. 36, 4352..N., 1975. Trickle-bed reactors. AIChE J. 21, 209228.

    shna, R., 1998. Process development and scale up III. and scale-down of trickle bed processes. Rev. Chem.203252.Saban, M.D., Orlovic, A.M., Meyn, V.W., Severin, D.K.,n, I.G.H., Marjanovic, M.V., 1991. Hydrotreating of

    prediction of industrial trickle-bed operation fromnt data. Ind. Eng. Chem. Res. 30, 20592065.A., Cooper, B.H., 1994. Aromatic hydrogenation: a review. Cat. Rev. Sci. Eng. 36, 75123., 2010. CFD study of an evaporative trickle bed reactor.ribution and thermal runaway induced by feednces. Chem. Eng. J. 161, 257268.

  • 564 chemical engineering research and design 9 4 ( 2 0 1 5 ) 549564

    Tsonopoulos, C., Heidman, J.L., Hwang, S.-C., 1986.Thermodynamic and Transport Properties of Coal Liquids.Wiley, New York.

    van der Merwe, W., Nicol, W., de Beer, F., 2007. Three-dimensionalanalysis of trickle ow hydrodynamics: computedtomography image acquisition and processing. Chem. Eng.Sci. 62, 72337244.

    van Houwelingen, A.J., Sandrock, C., Nicol, W., 2006. Particlewetting distribution in trickle-bed reactors. AIChE J. 52,35323542.

    Wang, Y., Chen, J., Larachi, F., 2013. Modelling and simulation oftrickle-bed reactors using computational uid dynamics:a state-of-the-art review. Can. J. Chem. Eng. 91,136180.

    Zeiser, T., Lammers, P., Klemm, E., Li, Y.W., Bernsdorf, J., Brenner,G., 2001. CFD-calculation of ow, dispersion and reaction in acatalyst lled tube by the lattice Boltzmann method. Chem.Eng. Sci. 56, 16971704.

    CFD study of diesel oil hydrotreating process in the non-isothermal trickle bed reactor1 Introduction2 Governing equations2.1 Modeling assumption2.2 Hydrodynamics equations2.3 Species conservation equations2.3.1 Reaction kinetics

    2.4 Energy conservation equation2.4.1 Gasliquid interphase heat transfer2.4.2 Liquidsolid and gassolid interphase heat transfer2.4.3 Wall to bed heat transfer

    3 CFD model and numerical algorithms4 Results and discussion4.1 CFD model validation4.2 Mesh independency4.3 Modified reaction kinetics4.4 Feed inlet temperature effect4.4.1 Temperature distribution

    4.5 The bed operational pressure effect4.5.1 Performance of adiabatic reactor

    4.6 Effect of LHSV on HDS reaction conversion4.6.1 HDA reactions conversion

    4.7 Gas flow rate effect4.8 Effect of gas phase H2S concentration4.8.1 Distribution of aromatic components axial mass fraction

    4.9 Gasliquid interphase heat transfer effect on simulation results

    5 ConclusionAcknowledgmentReferences


Recommended