www.elsevier.com/locate/ces
Received 23 April 2007; received in revised form 7 July 2007; accepted 13 August 2007
Available online 24 August 2007
Abstract
A bubble column slurry reactor (BCSR) model has been developed for the reductive alkylation of p-phenylenediamine (PPDA) with methyl
ethyl ketone (MEK) to N, N -di-secondary-alkyl-p-phenylenediamine (Di-amine). This particular reaction system is commercially relevant and
involves a combination of parallel and consecutive reactions comprising equilibrium non-catalytic (homogeneous) and catalytic (heterogeneous)
steps. The proposed model is based on the mixing cell approach. In this work the mixing cell approach has been extended by including a
liquid backow stream from all but the bottommost mixing cell. The model incorporates the contributions of gasliquid and liquidsolid mass
transfer, heat effects, and complex multistep reaction kinetics. CFD model is used to estimate the extent of backow among mixing cells and
its dependence on operating parameters. The effect of gas and liquid velocities, catalyst loading, inlet PPDA concentration, and temperature on
the conversion, selectivity, global rate of hydrogenation, and temperature rise is discussed. The comparison of the current approach with the
traditional mixing cell model is discussed. The BCSR model presented here will be useful to provide guidelines for designing and improving
overall performance of bubble column reactors.
2007 Elsevier Ltd. All rights reserved.
liquid phases is also important in deciding the overall efciency to N, N -di-sec-alkyl-p-phenylenediamine, using Pt/Al2 O3
of such reactors. While the analysis of multiphase catalytic re- catalyst (Patil et al., 2007), wherein two amine functionalities
actors has been well developed from the theoretical perspective are available for the alkylation reaction. A schematic of the
for simple reaction schemes, information on the analysis of re- reaction scheme considered in this work is shown in Fig. 1. In
actor performance for industrially useful processes in BCSRs this paper, we have reported a detailed analysis of a BCSR for
is very limited. In previous work on the modeling of BCSRs, reductive alkylation of PPDA.
detailed consideration has been given to hydrodynamic mod- The BCSR model was derived using the detailed kinetics
eling, particularly CFD modeling (Krishna and Maretto, 1998; reported in our earlier work (Patil et al., 2007) and mixing
Krishna et al., 2001; Rampure et al., 2007), evaluation of mix- cell approach. It incorporates the gasliquid, liquidsolid mass
ing (Govindarao and Chidambaram, 1983), and mass transfer transfer, heat effects, and complex multistep reaction kinetics.
parameters (Kawakami et al., 1981; Dassori, 1998). Although The changes in gas and liquid velocities inuence hydrodynam-
reactor performance models have been proposed for several ics of BCSR and therefore alter degree of back mixing in liq-
reactions, such as the FischerTropsch synthesis (Mills et al., uid phase. Inuence of operating conditions on hydrodynamics
1996; Krishna and Maretto, 1998; Krishna et al., 2001), hy- and liquid phase mixing was quantied using a computational
drogenation of glucose (Brahme et al., 1984), and the hydro- uid dynamics (CFD) model in the literature (Ekambara and
genation of butynediol (Jaganathan et al., 1987), in most of Joshi, 2003; Buwa and Ranade, 2003; Rampure et al., 2007).
these cases, only single reactions with simplied kinetics have The model of Rampure et al. (2007) was used for this purpose.
been considered. Thakar et al. (2003) have modeled the hy- Liquid phase backmixing was accounted by including a back-
drogenation of maleic acid to tetrahydrofuran (THF) in BCSR, ow (obtained using CFD) among mixing cells. The effect of
taking into account the distribution pattern of the product THF gas and liquid velocities, catalyst loading, inlet substrate con-
in gas and liquid phases. Industrial BCSRs typically involve centration, and temperature on the conversion, selectivity, and
complex multistep catalytic reactions with complexities such as temperature rise is discussed. The model presented here is use-
non-linear kinetics, catalyst deactivation, exothermic reactions, ful for developing insight into the effects of various operating
and non-uniform distribution of catalyst particles. One such ap- parameters on the overall reactor performance.
plication is the reductive alkylation of amine compounds for
the manufacture of higher alkylated (secondary and tertiary) 2. Reactor model
amine derivatives, which nd applications as intermediates in
ne chemicals and speciality products (Lehtonen et al., 1998). A detailed analysis of a BCSR model has been presented
The reductive alkylation reaction goes through a condensa- using a mixing cell approach. A mixing cell approach pro-
tion reaction between an amine compound or its precursor and posed earlier by Ramachandran and Smith (1979), Brahme
a carbonyl compound or alcohol to form an iminea Shiff et al. (1984), and Jaganathan et al. (1987) has been extended
Base(Ege, 1994), which is hydrogenated in the presence of in this work by including a liquid backow stream from all but
a metallic catalyst to N-alkylated products. Reductive alkyla- the bottommost mixing cell. This model incorporated the com-
tion of nitro or amine compounds has been investigated using a plexities of reaction kinetics coupled with mass transfer and
wide range of alkylating agents and catalysts (Greeneld, 1994; the thermal effects.
Patil et al., 2007).
Most of the previous work on reductive alkylation was fo- 2.1. Intrinsic kinetics
cused on catalysis aspects with the aim of improving catalytic
activity and selectivity of N-alkylated compounds. However, The intrinsic kinetics for the hydrogenation of PPDA to
the reaction system has not been studied much from reaction N, N -di-sec-butyl-p-phenylenediamine with 3% Pt/Al2 O3 cat-
engineering viewpoint. Though, there is an obvious advan- alyst have been recently studied by Patil et al. (2007) using a
tage in conducting the condensation and hydrogenation re- batch slurry reactor. Based on this work, the reaction network
actions in a single step and effectively drive the equilibrium can be described as shown in Fig. 1. The following types of
reaction step to complete conversion, understanding of the rate equations for non-catalytic (R1 , R2 , and R4 ) steps have
reaction engineering aspects is of signicant importance to been found to adequately describe the individual homogeneous
optimize the overall rate of reductive alkylation. There are reaction steps involved,
only a few reports in the literature, which deal with combina-
tions of homogeneousheterogeneous, non-catalyticcatalytic R1 = keq1 Bl keq2 Dl CH2 O , (1)
reactions. Lehtonen et al. (1998) followed by Salmi et al.
R2 = keq3 Dl keq4 Fl CH2 O , (2)
(1999), developed a semi-batch slurry reactor model and loop
reactor model for reductive alkylation of aromatic amines R4 = keq6 Gl keq7 Hl CH2 O , (3)
with short chain aldehydes (carbon chain length less than 3)
using Pt/C catalyst. More detailed kinetic studies were re- where keq1 = keq1m Cm , keq3 = keq3m Cm , keq6 = keq6m Cm and
ported from our group for reductive alkylation of aniline with CH2 O = (Dl + 2Fl + Gl + 2Hl + 2Pl ),
acetone to N-isopropylaniline using Pd/Al2 O3 catalyst (Roy
et al., 2005), and the more complex reductive alkylation of where keq1 , keq3 , keq6 and keq2 , keq4 , keq7 were the forward and
p-phenylenediamine (PPDA) with methyl ethyl ketone (MEK) backward equilibrium rate constants, respectively, for the steps
7292 Amit S. Chaudhari et al. / Chemical Engineering Science 62 (2007) 7290 7304
keq1 C 2H 5
O
H 2N N +
H 2N NH2 + H2O
H 3C C 2H 5 R1 , keq2 CH3
C 2H 5 C2H5 C2H5
N N + H2O H 2N N
H
H3 C CH3 CH3
k8 Catalyst R4 , keq6 O
H2
H3C C2H5
keq7
MEK (Cm)
C 2H 5 C 2H 5
N N + H2O
H
H 3C CH3
Intermediate (H) (CH O)
2
R6 k9
H2 Catalyst
C2H5 C2H5
N N
H H
H3C CH3
Di-alkylated amine (P), (Di-amine)
(N,N'-di-sec-butyl-p-phenylenediamine)
Fig. 1. Detailed reaction scheme of reductive alkylation of PPDA with MEK using 3% Pt/Al2 O3 catalyst.
R1 , R2 , and R4 mentioned in Fig. 1 and Cm represents the In Eqs. (4)(6), As represents the concentration of dissolved
concentration of MEK. For all the kinetic experiments, MEK hydrogen on the catalyst surface. Denitions for the remaining
was used in large excess with respect to PPDA and its change kinetic rate equation variables appear in the Notation section.
in concentration during the course of reaction was negligible. The rate parameters are given in Table 1.
Therefore, all the forward equilibrium steps were considered
to be as pseudo rst order reactions.
2.2. Mixing cell model
The following types of LangmuirHinshelwood rate equa-
tions have been found to adequately describe the individual
The schematic of mixing cell model used in this work is
catalytic reaction steps (R3 , R5 , and R6 ) involved,
shown in Fig. 2. The model was based on following assump-
wk 5 Dl AS tions:
R3 = , (4)
(1 + KA AS ) (1) The BCSR was visualized as reactor consisting of N cells
of stirred tanks in series where the liquid is completely
wk 8 Fl AS
R5 = , (5) backmixed and gas is in plug ow in each cell. Thus,
(1 + KA AS ) for N , plug ow will prevail, while for N = 1,
wk 9 Hl AS the reactor performance will be equal to backmixed slurry
R6 = . (6) reactor. Intermediate values will describe the liquid phase
(1 + KA AS )
ow patterns between these two extremes.
The overall rate of hydrogenation using Eqs. (4)(6) can be (2) The catalyst particles were assumed to be dispersed uni-
given as formly throughout the reactor. To verify this assumption,
RA = R3 + R5 + R6 . the approach of Kato et al. (1972) was followed.
(3) The liquid phase components are non-volatile since the
w(k5 Dl + k8 Fl + k9 Hl ) AS bubble point of mixture was much larger than the operating
RA = . (7)
(1 + KA AS ) temperature.
Amit S. Chaudhari et al. / Chemical Engineering Science 62 (2007) 7290 7304 7293
Table 1
Kinetic rate equation parameters
Temp. (K) keq1 (103 ) keq2 (104 ) keq3 (104 ) keq4 (104 ) k5 (102 ) keq6 (104 ) keq7 (105 ) k8 (102 ) K9 (103 ) KA
373 0.1248 0.03508 1.214 0.0692 0.0639 0.2539 0.072 0.2035 0.4110 1.605
393 0.2091 0.2562 2.910 0.3732 0.1091 0.2018 0.2397 0.3980 0.5147 1.111
413 0.2440 0.8807 2.982 0.3981 0.1481 0.6497 1.267 0.3923 0.6609 1.6395
Tgf Gas Liquid isothermal effects. To verify the rst assumption, the cri-
ulu C iLf terion proposed by Mears (1971) was used.
A gf
A Lf TLf
Cell N The extent of backow among mixing cells was estimated using
the CFD model.
u lb
ul = u lu + ulb . . 2.3. Model equations
Table 2
Dimensionless parameters used in model
The term cH2 O in Eqs. (14)(19) is the dimensionless concen- N th (Top) cell:
tration of water and can be represented as
He (T ) = 1.2e3 exp(2.7e3 T ) for 5% PPDA, (20) The solution of the preceding set of non-linear algebraic
equations allow the prediction of the concentrations of all
He (T ) = 1.8e3 exp(2.2e3 T ) for 10% PPDA, (21)
species and the temperature at the exit of each cell as well
He (T ) = 3e4 exp(6.9e3 T ) for 15% PPDA. (22) as the N th cell (exit of the reactor) for any given set of inlet
parameters. Mass balance of hydrogen can be calculated by
The change in rate and equilibrium constants with temperature
integrating Eq. (9) to give
can be expressed as
agj alj
Ei 1 1 = exp(A z). (28)
ki (To ) = ki (Ti ) exp , (23) agj 1 alj
Rg Ti To
By substituting z = 1/N , the dimensionless concentration of A
EAD 1 1 in the stream leaving the cell j is
Ki (To ) = Ki (Ti ) exp . (24)
Rg Ti To
A
agj = agj 1 exp + alj . (29)
N
The heat evolved during the reaction was assumed to be carried
away by the heat transfer to the reactor wall, which is charac- Substituting Eq. (28) in Eqs. (10), (11), and (12) and solving
terized by the bed-to-wall heat transfer coefcient, Uw . Under the resulting integral equation gives the following expression
such conditions where interphase and intraparticle heat transfer for the dimensionless concentration of A in the liquid stream
resistances were assumed to be negligible, the energy balance leaving cell j
in the reactor can be expressed as First cell:
First cell:
ls
(alj 1 + alj +1 ) + A agj 1 + asj
alj = N
R5 (dlj + k85 flj + k95 hlj ) asj ls . (30)
= 1++ + A
N qB 1 1 + ka asj N
2 Middle cell:
+ [R1 blj R2 dlj cH2 O ]
N
ls
((1 + )alj 1 + alj +1 ) + A agj 1 + asj
+ 3 [R3 dlj R4 flj cH2 O ] + 4 [R6 glj R7 hlj cH2 O ] alj = N . (31)
N N ls
5 1 + 2 + + A
(j w ) + 6 j +1 + 7 j 1 . (25) N
N
N th (Top) cell:
Middle cells:
ls
((1 + )alj 1 ) + A agj 1 + asj
R5 (dlj + k85 flj + k95 hlj ) asj alj = N . (32)
= ls
N qB 8 1 + ka asj 1++ + A
N
9
+ [R1 blj R2 dlj cH2 O ] is the parameter dened in Eqs. (29)(32) and can be repre-
N sented as
+ 10 [R3 dlj R4 flj cH2 O ] A
N = 1 exp .
11 N
+ [R6 glj R7 hlj cH2 O ]
N Once alj is known, the concentration of A on the catalyst sur-
face, as , can be obtained from Eq. (13). Eqs. (9)(19) were
12 (j w ) + 13 j 1 + 14 j +1 . (26)
N combined with Eqs. (25)(27) and solved simultaneously using
7296 Amit S. Chaudhari et al. / Chemical Engineering Science 62 (2007) 7290 7304
Table 3
Correlations used for evaluation of kl aB , ks , Uw , and Dm
For any given set of inlet conditions, the concentrations of re- 2.5. Estimation of backow with CFD model ()
actant/product species and temperature were calculated for the
exit of each cell and also for the exit of the reactor. The dimen- The extent of backmixing in bubble column reactors depends
sionless parameters used in the above equations are presented mainly on column dimensions and supercial gas velocity (pro-
in Table 2. The hydrodynamic and mass transfer parameters vided that liquid supercial velocity is rather small). When gas
required for the calculations of the dimensionless parameters is sparged at the bottom of bubble column reactor, bubbles
were calculated from literature correlations presented in move upwards from the central region of the column creating
Table 3. At any given length of the reactor, the fractional a radial gradient of gas hold-up. Such a radial gradient sets up
conversion of PPDA is given as a liquid circulation within the column causing severe backmix-
ing. CFD models can provide quantitative relationship between
XB = 1 blj . (34) internal liquid circulation and design and operating parameters
of bubble columns (Ranade, 2002). In this work therefore we
The global rate of hydrogenation (including hydrogen con- have used a CFD model to simulate hydrodynamics of bubble
sumption in all steps) was calculated as column reactor considered for simulating reductive alkylation
reactions.
RH2 = (glj + 2hlj + 3plj )Bli . (35) The catalyst particle size and loading is fairly small for
the considered case (30 m). Therefore, the hydrodynamics of
In Eq. (35), is the mean residence time of the liquid in the BCSR was modeled as a two-phase ow by considering gas as
reactor ( = ulu /L), ulu is the liquid velocity, L is the length a dispersed phase and slurry as a pseudo-homogeneous contin-
of the reactor, glj , hlj , and plj are the concentrations of M- uous phase. The gasliquid ow in bubble column was sim-
amine, Intermediate, and Di-amine, respectively, at the exit of ulated using the EulerianEulerian approach. In this approach
the reactor. both mass and momentum balances were solved for each phase
The selectivity of M-imine, Di-imine, M-amine, Intermedi- (Ranade, 2002). The standard k mixture turbulence model
ate, and Di-amine was evaluated using the following relation- was used in the present work. Inuence of gas hold-up on ef-
ships: fective drag coefcient was accounted. Appropriate properties
of gas and liquid (slurry) phase (corresponding to 40 bar op-
dlj erating pressure and average operating temperature) were used
SM-imine = 100, (36) in the simulations (gas: density of 2.48 kg/m3 and viscosity of
1 blj
1 105 kg/m s; slurry: density of 845.68 kg/m3 and viscosity
flj of 0.0618 kg/m s). A sieve-plate sparger was assumed to dis-
SDi-imine = 100, (37) tribute the gas uniformly throughout the sparger region and was
1 blj
modeled as inlet. The bubble velocity was set as gas inlet and
glj volume fraction at inlet was set in such a way that it ensures
SM-amine = 100, (38) desired net gas ow rate. The top surface of the column was
1 blj modeled as velocity inlet, applying same boundary condition as
Amit S. Chaudhari et al. / Chemical Engineering Science 62 (2007) 7290 7304 7297
that for sparger. The overall gas hold-up in the column was still Table 4
a free variable. Simulations were carried out till the unsteady Range of operating conditions for BCSR investigated
ow was established with acceptable mass balance for all uid Design parameters/operating conditions Values
phases. The time averaging was started after discarding the ini-
Catalyst 3% Pt/Al2 O3
tial transients. Mass transfer and reactions were not considered
Catalyst loading, w 0.284.6 kg/m3 (0.02510% w/w)
in the CFD model since the focus was on quantifying the ex- Temperature 373413 K
tent of backmixing. The details of model equations and solu- Initial concentration of MAC, Bli 0.51.5 kmol/m3 (515% w/w)
tion strategies were discussed by Rampure et al. (2007). The H2 pressure, P 26.2 MPa
geometry of the considered reactor was modeled using GAM- Solvent MEK
BIT v2.0. The geometry was discretized using 80,000 com- Liquid velocity, ulu 210e 4 m/s
Gas velocity, ug 210e 2 m/s
putational cells. The commercial CFD code FLUENT 6.3 (of Reactor diameter, dT 0.25 m
Ansys-Fluent Inc., 10 Cavendish Court, Lebanon, NH 03766, Total reactor length, L 4m
USA) was used to carry out simulations. Particle diameter, dp 30 m
The CFD simulations were carried out for supercial gas Density of catalyst, p 1260 kg/m3
velocities ranging from 0.02 to 0.10 m/s. For the range of gas
velocities studied, the homogeneous ow regime prevails up
to 0.05 m/s. Transition regime occurs at the supercial gas ve- 1.0
locities near to 0.050.06 m/s and thus extends to churn turbu-
lent regime at gas velocity of 0.1 m/s (Thorat and Joshi, 2004). N =1
Since the maximum liquid velocity considered for the reac- 0.9 N =3
tor simulations was only 0.001 m/s, actual value of liquid su- N = 10
percial velocity will not inuence internal liquid circulation
0.8
within the column. Therefore, all the CFD simulations were
Conversion of PPDA
8
0.6 0.20
T = 373 K
T = 393 K
M-amine
T = 413 K 0.5
0.16
6
0.4
Temperature rise, K
Di-amine 0.12
Selectivity
Selectivity
4 0.3
0.08
0.2 Intermediate
0.0 0.00
0 0 14 28 42 56 70 84
0 2 4 6 8 10 12 Catalyst loading, w, kg/m3
Liquid velocity, ulu x 104, m/s
0.8 0.20
Fig. 4. Temperature rise as a function of liquid velocity: effect of inlet temper-
ature. Reaction conditions: w=84.6 kg/m3 , Bli =1 kmol/m3 , ug =5102 m/s, Intermediate
P = 4 MPa, N = 10.
0.6 M-amine 0.15
Di-amine
Selectivity
Selectivity
temperature rise (at the exit of the reactor), when the inlet PPDA
concentration, catalyst loading, and H2 pressure were varied. 0.4 0.10
In predicting the reactor performance, we had used the corre-
lation proposed by Yoshida and Akita (1965) for calculating the
gasliquid mass transfer coefcient. It was found that there was M-imine
0.05
0.2
no signicant change in the reactor performance when the mass
transfer coefcient was varied in 10-fold. It may be noted Di-imine
here the gasliquid mass transfer coefcients predicted by liter-
ature correlations (Behkish et al., 2002; Jordan and Schumpe, 0.0 0.00
0 14 28 42 56 70 84
2001; Hikita et al., 1981; Koide et al., 1984) varied only 60%
Catalyst loading, w, kg/m3
to that predicted by Yoshida and Akita (1965).
The effect of liquid velocity on temperature rise for differ- Fig. 5. Selectivity as a function of catalyst loading: effect of liquid velocity.
ent inlet temperatures is shown in Fig. 4. The temperature rise Reaction conditions: Bli =1 kmol/m3 , ug =5102 m/s, T =393 K, P =4 MPa,
was calculated for different reaction conditions and found to N = 10. (a) ulu = 2 104 m/s; (b) ulu = 10 104 m/s.
increase with the increase in inlet temperature. The maximum
temperature rise was observed for the highest substrate con-
centration. The temperature rise was found to be insignicant, 50%) can be achieved. The selectivity of M-amine decreased
might be due to the equilibrium steps involved in the overall with increase in inlet PPDA concentration whereas selectiv-
reaction scheme and better heat transfer efciency of BCSR. ity of Intermediate increased and Di-amine remained constant.
The effects of catalyst loading and inlet PPDA concentration With increase in PPDA concentration, M-amine concentration
on the selectivity of products are shown in Figs. 5 and 6, re- increases which then transforms to Intermediate by the equilib-
spectively, for different inlet liquid velocities. It was observed rium reaction (Fig. 1). As the kinetic modeling data presented
that the selectivity of M-amine and Di-amine increases with in Table 1 clearly indicate the slow hydrogenation rate for the
increase in the catalyst loading (Fig. 5). It is clear from Figs. Intermediate to Di-amine, in a continuous ow reactor the re-
5a and b that selectivity of M-amine increases steeply within sults presented in Fig. 6 were expected.
a catalyst range of 0.22 kg/m3 , which indicates that the reac- The product concentration proles at the reactor exit as a
tion is in kinetic regime. The reaction slowly moves to mass function of N, for minimum and maximum liquid velocity are
transfer regime beyond this catalyst loading. It should be noted presented in Figs. 7a and b, respectively. The concentrations
here that the selectivity toward the products were independent of all species are constant after N = 10, indicating plug ow
at higher catalyst loading (more than 84 kg/m3 ). At lower liq- conditions. As the liquid velocity increased to its maximum
uid velocities and catalyst loading, a high selectivity (up to value, the concentration of M-amine and Di-amine decreases
Amit S. Chaudhari et al. / Chemical Engineering Science 62 (2007) 7290 7304 7299
Concentration x 10 2, kmol/m3
0.032 0.8 0.20
0.6
Concentration, kmol/m3
0.024
Selectivity
0.6 0.15
Selectivity
0.4
0.2
0.008 0.2 0.05
Intermediate
0.0 0.000
0.0 0.00
0.5 1.0 1.5
04 8 12 16
Inlet PPDA concentration, Bli, kmol/m3
Cell No., N
Concentration, kmol/m3
0.12 0.6
Selectivity
Selectivity
0.6 0.06
Di-amine
0.08 0.4
0.4 0.04
Di-imine
0.04 0.2
0.2 0.02
M-imine
0.00 0.0
0.5 1.0 1.5 0.0 0.00
3 0 4 8 12 16
Inlet PPDA concentration, Bli, kmol/m
Cell No., N
Fig. 6. Selectivity as a function of inlet PPDA concentration: effect of liquid
Fig. 7. Effect of N on the species concentration prole for different
velocity. Reaction conditions: w = 84.6 kg/m3 , ug = 5 102 m/s, T = 393 K,
liquid velocities. Reaction conditions: w = 84.6 kg/m3 , Bli = 1 kmol/m3 ,
P = 4 MPa, N = 10. (a) ulu = 2 104 m/s; (b) ulu = 10 104 m/s. ug = 5 102 m/s, T = 393 K, P = 4 MPa. (a) ulu = 2 104 m/s; (b)
ulu = 10 104 m/s.
Vmax 1.00
0.90
0.80
0.70
0.60
0.50
0.40
0.30
0.20
0.10
Vmin 0.00
Fig. 8. CFD simulations for the estimation of the backow in bubble column: (a) time averaged axial slurry velocity; (b) instantaneous gas hold-up.
Table 5 1.0
Values of at different gas velocities from CFD simulation
ug 102 (m/s) ulb (m/s) (Min.) (Max.)
0.8 M-amine
2 5.39e 4 0.54 2.69
5 9.56e 4 0.96 4.78
7 5.71e 2 57.07 285.33
10 7.22e 2 72.18 360.90 0.6 phi = 0 (Without backflow)
Selectivity
Di-amine
0.2
0.8
0.0
Conversion of PPDA
0.6 0 4 8 12 16 20
Cell No., N
8 9
Without backflow
7 With backflow
8
6
Temperature rise, K
5
Temperature rise, K
4 6
3
5
phi = 0 (Without backflow)
2
phi = 5
phi = 20 4
1
phi = 72
0
3
0 4 8 12 16 20
0 2 4 6 8 10 12
Cell No., N
Gas velocity, ug x 102, m/s
Fig. 11. Temperature rise as a function of number of cells: effect of
backow stream. Reaction conditions: w = 84.6 kg/m3 , Bli = 1 kmol/m3 , Fig. 13. Temperature rise as a function of number of cells: effect of
ulu = 10 104 m/s, T = 393 K, P = 4 MPa, N = 16. backow stream. Reaction conditions: w = 84.6 kg/m3 , Bli = 1 kmol/m3 ,
ulu = 10 104 m/s, T = 393 K, P = 4 MPa, N = 16.
1.0
Without backflow 0.12 0.8
With backflow With backflow
0.7
0.8 0.10
0.08
0.5
0.6
Selectivity
Selectivity
M-amine
Selectivity
Di-amine
0.06 0.4
Di-amine
0.4 0.3
0.04 M-imine
0.2
0.00 0.0
0 2 4 6 8 10 12
0.0
0 2 4 6 8 10 12 Gas velocity, ug x 102, m/s
Liquid velocity, ulu x 104, m/s
Fig. 14. Selectivity as a function of gas velocity: effect of backow stream.
Fig. 12. Selectivity as a function of inlet liquid velocity: effect of Reaction conditions: w = 84.6 kg/m3 , Bli = 1 kmol/m3 , ulu = 10 104 m/s,
backow stream. Reaction conditions: w = 84.6 kg/m3 , Bli = 1 kmol/m3 , T = 393 K, P = 4 MPa, N = 16.
ug = 5 102 m/s, T = 393 K, P = 4 MPa, N = 16.
1.00
0.79
0.08
0.90 0.50
0.58
0.80 0.37
0.38 0.32
0.70
0.17 0.14 -0.06
0.60
-0.04
-0.04 -0.49
0.50
-0.24
0.40 -0.22 -0.92
-0.45
-0.65 0.30
-0.40 -1.35
-0.86 0.20
-1.27 0.00
Fig. 15. CFD simulations for BCSR with bafe for ug = 0.1 m/s. (a) Time averaged axial liquid velocity; (b) instantaneous gas hold-up; (c) time averaged
axial liquid velocity without and with radial bafe.
Notation
Fig. 16. Effect of bafes on conversion of PPDA and selectivity of products.
Reaction conditions: w = 84.6 kg/m3 , Bli = 1 kmol/m3 , ulu = 10 104 m/s,
ug = 10 102 m/s, T = 393 K, P = 4 MPa, N = 16. aB gasliquid interfacial area, m2 /m3
al dimensionless concentration of hydrogen in the liquid
phase (Al .HA /Ago )
opening of 0.7D was simulated using the CFD model. The pre- ap external surface area of the pellet [6w/p dp ], m1
dicted results are shown in Figs. 15a and b. Presence of radial A saturation solubility of hydrogen, kmol/m3
bafes signicantly reduced the extent of backow (by almost Al concentration of hydrogen in the liquid phase,
70%, see Fig. 15c). The backow predicted in presence of ra- kmol/m3
dial bafes was then used to simulate the considered reductive As concentration of hydrogen on the catalyst surface,
alkylation reactions. The predicted inuence of radial bafes kmol/m3
on selectivity and conversion is shown in Fig. 16. It can be seen bl dimensionless concentration of PPDA in the liquid
that the presence of bafes showed 20% increase in conversion phase (Bl /Bli1 )
without affecting selectivity. The radial bafes can therefore be Bl concentration of PPDA in the liquid phase, kmol/m3
considered to be useful for controlling and manipulating degree Bli initial concentration of PPDA in the liquid phase,
of backmixing in BCSRs. kmol/m3
Amit S. Chaudhari et al. / Chemical Engineering Science 62 (2007) 7290 7304 7303
Behkish, A., Zhuowu, M., Inga, J.R., Morsi, B.I., 2002. Mass transfer Krishna, R., van Baten, J.M., Urseanu, M.I., Ellenberger, J., 2001. Design and
characteristics in a large-scale slurry bubble column reactor with organic scale up of a bubble column slurry reactor for FischerTropsch synthesis.
mixtures. Chemical Engineering Science 57, 33073324. Chemical Engineering Science 56, 537545.
Bern, L., Hell, M., Schoon, N.H., 1975. Kinetics of the hydrogenation of Lehtonen, J., Salmi, T., Vaori, A., Tirronen, E., 1998. On the principles of
rapeseed oil. Rate equations of chemical reactions. Journal of the American modeling of homogeneousheterogeneous reactions in the production of
Oil Chemists Society 52, 391394. ne chemicals. A case study: reductive alkylation of aromatic amines.
Bischoff, K.B., 1965. Effectiveness factors for general reaction rate forms. Organic Process Research and Development 2, 7885.
A.I.Ch.E. Journal 11, 351355. Mashelkar, R.A., 1970. Bubble columns. British Chemical Engineering 15,
Brahme, P.H., Chaudhari, R.V., Ramachandran, P.A., 1984. Modeling of 1297.
hydrogenation of glucose in a continuous slurry reactor. Industrial and Mathieu, C., Dietrich, E., Delmas, H., Jenck, J., 1992. Hydrogenation of
Engineering Chemistry Process Design and Development 23, 857859. adiponitrile catalysed by raney nickel: use of intrinsic kinetics to measure
Bukur, D.B., 1983. Some comments on models for FischerTropsch reaction gasliquid mass transfer in a gas induced stirred slurry reactor. Chemical
in slurry bubble column reactors. Chemical Engineering Science 38, Engineering Science 47, 22892294.
440446. MATLAB 7, Release 14, 2004. The MathWorks Inc.
Buwa, V.V., Ranade, V.V., 2003. Mixing in bubble columns reactors: role of Mears, D.E., 1971. Tests for transport limitations in experimental catalytic
unsteady ow structures. Canadian Journal of Chemical Engineering 81, reactors. Industrial and Engineering Chemistry Process Design and
402411. Development 10, 541547.
Dassori, C.G., 1998. Three phase reactor modeling with signicant back Mills, P.L., Turner, J.R., Ramchandran, P.A., Duducovic, M.P., 1996. The
mixing in the liquid phase using modied cell model (MCM). Computers FischerTropsch synthesis in slurry bubble column reactors: analysis of
& Chemical Engineering 22 (Suppl. 1), S679S682. reactor performance using the axial dispersion model. Topics in Chemical
Deckwer, W.D., 1980. On the mechanism of heat transfer in bubble column Engineering 8 (339386), 679739.
reactors. Chemical Engineering Science 35, 13411346. Patil, N.G., Roy, D., Chaudhari, A.S., Chaudhari, R.V., 2007. Kinetics of
Deckwer, W.D., 1992. Bubble Column Reactors. Wiley, New York. reductive alkylation of p-phenylenediamine with methyl ethyl ketone
Ege, S., 1994. Organic Chemistry Structure and Reactivity. third ed. D.C. using 3% Pt/Al2 O3 catalyst in a slurry reactor. Industrial and Engineering
Heath and Co., Lexington, KY. p. 535. Chemistry Research 46, 32433254.
Ekambara, K., Joshi, J.B., 2003. CFD simulations of residence time Ramachandran, P.A., Chaudhari, R.V., 1983. Three Phase Catalytic Reactors.
distribution and mixing in bubble column reactors. Canadian Journal of Gorden & Breach, New York.
Chemical Engineering 81, 669676. Ramachandran, P.A., Smith, J.M., 1979. Mixing-cell model for design of
Govindarao, V.M.H., Chidambaram, M., 1983. On the steady state trickle-bed reactors. Chemical Engineering Science 17, 9199.
performance of cocurrent bubble column slurry reactors. Chemical Rampure, M., Kulkarni, A.A., Ranade, V.V., 2007. Hydrodynamics of bubble
Engineering Journal 27, 2936. column reactors at high gas velocity: experiments and CFD simulations.
Greeneld, H., 1994. Side reactions in reductive alkylation of aromatic amines Industrial and Engineering Chemistry Research, doi: 10.1021/ie070079.
with aldehydes and ketones. Chemical Industry 53, 265277. Ranade, V.V., 2002. Computational Flow Modeling for Chemical Reactor
Hikita, H., Asai, S., Tanigawa, K., Segawa, K., Kitao, M., 1981. The Engineering. Academic Press, London.
volumetric liquid phase mass transfer coefcient in bubble columns. Roy, D., Jaganathan, R., Chaudhari, R.V., 2005. Kinetic modeling of aniline
Chemical Engineering Journal 22, 6169. with acetone using 3% Pd/Al2 O3 catalyst in a batch slurry reactor.
Jaganathan, R., Gholap, R.V., Brahme, P.H., Chaudhari, R.V., 1987. Industrial and Engineering Chemistry Research 44, 53885396.
Performance of a continuous bubble column slurry reactor for Salmi, T., Lehtonen, J., Kaplin, J., Vuori, A., Tirronen, E., Haarrio, H.,
hydrogenation of butynediol. Proceedings of ISCRE 2, vol. II. NCL, India, 1999. A homogeneous-heterogeneously catalysed reaction system in a loop
pp. 141155. reactor. Catalysis Today 48, 139145.
Jordan, U., Schumpe, A., 2001. The gas density effect on mass transfer in Sano, Y., Yagamuchi, N., Adachi, T., 1974. Mass transfer coefcients for
bubble columns with organic liquids. Chemical Engineering Science 56, suspended particles in agitated vessels and bubble columns. Journal of
62676272. Chemical Engineering Japan 1, 255.
Kato, Y., Nishiwaki, A., Fakuda, T., Tanaka, S., 1972. The behavior of Saxena, S.C., 1995. Bubble column reactors and FischerTropsch synthesis.
suspended solid particles and liquid in bubble columns. Journal of Catalysis Reviews: Science and Engineering 37, 227309.
Chemical Engineering of Japan 5, 112117. Schumpe, A., Nigam, K.D.P., 1996. Mass transfer in three-phase sparged
Kawakami, K., Aio, T., Kusunoki, K., 1981. Selectivity of butadiene reactors. Topics in Chemical Engineering 8 (169300), 679739.
hydrogenation in a bubble column slurry reactor. Kagaku Kogaku Shah, Y.T., 1979. GasLiquidSolid Reactor Design. McGraw-Hill, New York.
Ronbunshu (Japanese) 7, 105108. Thakar, N.N., Jaganathan, R., Chaudhari, R.V., Mills, P.L., 2003. Modeling of
Khare, A.S., Dharwadkar, S.V., Joshi, J.B., Sharma, M.M., 1990. Liquid- and hydrogenation of maleic acid in a bubble column slurry reactor. A.I.Ch.E.
solid-phase mixing in sectionalized bubble column slurry reactor. Industrial Journal 49, 31993212.
and Engineering Chemistry Research 29, 15031509. Thorat, B.N., Joshi, J.B., 2004. Regime transition in bubble columns:
Koide, K., Takazawa, A., Komura, M., Matsunaga, H., 1984. Gas hold up experimental and predictions. Experimental Thermal and Fluid Science
and volumetric liquid phase mass transfer coefcient in solid suspended 28, 423430.
bubble columns. Journal of Chemical Engineering of Japan 17, 459466. Vinke, H., Hamersma, P.J., Fortuin, J.M.H., 1991. Particle-to-bubble adhesion
Krishna, R., 2000. A scale up strategy for a commercial scale bubble column in gas/liquid/solid slurries. A.I.Ch.E. Journal 37, 18011809.
slurry reactor for FischerTropsch synthesis. Oil and Gas Science and Wilke, C.R., Chang, P., 1955. Correlation of diffusion coefcients in dilute
Technology 55, 359393. solutions. A.I.Ch.E. Journal 1, 264270.
Krishna, R., Maretto, C., 1998. Scale up of a bubble column slurry reactor Yoshida, F., Akita, K., 1965. Performance of gas bubble columns. Volumetric
for FischerTropsch synthesis. Studies in Surface Science and Catalysis liquid phase mass transfer coefcient and gas hold up. A.I.Ch.E. Journal
(Natural Gas Conversion V) 119, 197202. 11, 913.