Bubble column reactor

Jamshid Behin
Tahereh Shojaeimehr
Department of Chemical
Engineering, Faculty of
Engineering, Razi University,
Kermanshah, Iran.

819

Research Article

Modeling of Multistage Bubble Column

Reactor for Oxidation Reaction
In this work, a mathematical model based on axial dispersion has been suggested
to simulate the behavior of a multistage bubble column reactor. A six-stage pilotscale reactor with an inner diameter of 0.35 m and a height of 12 m was used for
hydrogen peroxide production through the direct oxidation of isopropyl alcohol
at isothermal condition. Steady-state and dynamic simulations were performed
to predict the concentration of all the reactants in gas and liquid phases. It was
observed that for steady state-conditions the simulation results were consistent
with the experimental results. Dynamic models involving liquid back-mixing can
be used for the simulation of start-up, shut-down or transition operations in this
kind of a rector.
Keywords: Axial dispersion model, Bubble column reactor, Hydrogen peroxide, Modeling,
Simulation
Received: May 11, 2012; revised: February 01, 2013; accepted: February 01, 2013
DOI: 10.1002/ceat.201200275

Introduction

Bubble columns reactors (BCR) are mass-transfer and reaction

equipment in which one or several gases disperse through a
deep pool of liquid and react with the liquid phase itself or
with a component dissolved or suspended in it [13]. These
reactors are widely used in chemical, biochemical, and petrochemical industries for oxidation, hydrogenation, chlorination,
alkylation, polymerization, methanol production and FischerTropsch synthesis [48]. The most important advantages of
BCRs are their simple construction, isothermal operation,
good interface contact, excellent heat- and mass transfer properties, low pressure drop, and low energy consumption
[2, 9, 10]. Despite their simple construction, the reactor design
and scale-up are still very difficult because of their complex
hydrodynamics and mass transfer phenomena [4, 11]. The two
ideal flow regime models, i.e., completely stirred tank and plug
flow, are not convenient for the prediction of phenomena
occurring in BCRs. In such reactors, the axial dispersion model
(ADM) or the multi-cell model (MCM) are traditionally used
to describe and quantify the extent of liquid back-mixing
[5, 1012].
A number of both experimental and simulation works on
BCRs have been published. In order to simulate the catalytic
chlorination of toluene in a batch BCR, Lohse et al. employed

Correspondence: Prof. J. Behin (Jamshid_Behin@yahoo.com), Department of Chemical Engineering, Faculty of Engineering, Razi University,
Kermanshah, Iran.

Chem. Eng. Technol. 2013, 36, No. 5, 819828

a completely back-mixed model for the liquid phase and a plug

flow model for the gas phase. Simulation results showed good
agreement with experimental data in the determination of
chlorine conversion [13]. Muroyama et al. used the ADM for
liquid phase, and plug flow model for gas phase, in order to
simulate the BCR used for treating drinking water by ozone
[7]. Simulation of the BCR used for wet air oxidation of industrial wastewater has been done by Debellefontaine et al. They
employed the ADM in both gas and liquid phases [6]. Mecklenburg and Hartland [14] and Schluter [15] used a numerical
simulation, and they found that the predictions of the ADM
and MCM were nearly equivalent. Abu-Reesh and Abu-Sharkh
showed that the ADM is suitable for the analysis of enzymatic
reactions during reactive extraction, using first-order enzyme
kinetics [16]. Xi and Gang employed ADM in both gas and liquid phases for p-xylene oxidation in industrial BCRs. They
tested the model with experimental results and reported that
the ADM is suitable for the modeling of BCRs [17]. Deckwer
et al. also used the ADM for the measurement of the gas-liquid
mass transfer and depicted an effective usage to analyze mass
transfer coefficients in BCRs [18].
The multistage bubble column reactor (MBCR) is a type of
BCR divided into discrete sections by perforated plates. It can
significantly improve the mass transfer characteristics and substantially reduce the degree of back-mixing in the contacted
phase at the same time. Therefore, high conversion levels can
be achieved in this reactor [10, 19, 20]. It has been demonstrated that it is an efficient way to break the described gas-liquid
phase macrocirculation pattern by creating independent mixed
stages between the sections [21, 22]. Alvar et al. studied the
effect of tray design and operating conditions on the overall

2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.cet-journal.com

820

J. Behin, T. Shojaeimehr

stant pressure of about 22 bar. Evaporated components are

totally condensed at the top and are recycled into the 6th stage
of the reactor. A sampling device is placed on the outlet of the
liquid in each stage of the reactor and the experimental data is
available only for these points. The amount of HP in the sample is determined by iodometry. Other components are analyzed using gaseous chromatography.

liquid phase mixing in a bench-scale trayed BCR. A reduction

in the liquid back-mixing was achieved in the trayed column,
as compared to the column without the tray. The ADM and
the MCM were compared to interpret the experimental results
[10].
This study aimed to develop a multi-component, onedimensional, dynamic mathematical model for pilot-scale
MBCR by employing ADM in both liquid and gas phases. This
reactor is used to produce hydrogen peroxide (HP) by the
autoxidation of isopropyl alcohol (IPA). Molecular oxygen
(O2) is the oxidation agent and acetone (ACTN) is produced
as a by-product of the reaction.

3.1 Time-Dependent Nonlinear Kinetics

Due to the important role of reaction kinetics in the reactor
operation, robust kinetic equations involving all influent variables were needed. The main reactions of HP formation by
IPA oxidation are as follows:

Oxidation Reactor

Fig. 1 shows the schematic diagram of the MBCR that is used

for the production of HP in the pilot plant sited at the Petrochemical Research and Technology Company (Arak/Iran). This
cylindrical reactor, with an inner diameter of 0.35 m, has six
reaction stages (sections). In each stage, compressed air disperses in a pool of liquid (about 1.8 m in height) from the bottom by means of a bubble cap tray. Continuous flow of gas
and liquid are in counter-current and oxidation reaction
occurs. Incoming gas from the lower stage disperses into the
liquid, firstly through 27 holes of 4 mm diameter under the
distributor (cap), and secondly through 80 holes of 2 mm
diameter on top of the cap. The down-flow liquid falls into
another stage (tray) by means of a downcomer under the cap.
The inlet liquid is an azeotropic mixture containing IPA-H2O
(83.2 wt % IPA) and low amounts of HP (nearly 0.1 wt %) as a
reaction initiator. In order to control temperature changes at a
constant level of 130 C, each stage of the reactor is equipped
with heating and cooling jackets. The reactor operates at a con-

O2 gas O2 diss:

(1)

O2 diss: CH3 2 CHOHl H2 O2 l CH3 2 COl

(2)

The reaction is not catalytic, but a small amount of HP is

needed for initiating the reaction. The non-elementary kinetic
rate equations derived by Kunugi et al. are as follows [23]:
rHP;l

dCHP;l
1=2
kCHP;l CIPA;l
dt

rO2 ;l

dCO2 ;l

dt

kCHP;l

(3)

dCIPA;l dCAc;l

dt
dt

1=2

kCHP;l CIPA;l kCHP;l

(4)

Liquid

Outlet gas

Inlet
liquid

Modeling

Gas

Gas

Condensed
liquid
6

TC

TC

TC

TC

TC

180 cm
Top view of cap

Inlet
gas

Gas

Gas
Under view of cap

Outlet liquid

www.cet-journal.com

Liquid

2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Figure 1. Schematic diagram of the IPA oxidation

reactor.

Chem. Eng. Technol. 2013, 36, No. 5, 819828

Bubble column reactor

The formation of a small

amount of acetic acid (AA) leads
to the decomposition of the HP
produced during the reaction and
is considered as an undesirable byproduct. Although the concentration of AA as an independent component is ignored, the effect of this
by-product on the decomposition
of HP cannot be neglected. In this
paper, the impact of AA was studied and its considerable effect on
reducing the production efficiency
of HP was identified. Therefore,
the relationship between the variations of rate constants with the
concentration of acid was presented as a function of time and
temperature:

821

Outlet gas

Inlet liquid

Liquid phase

Gas phase

dx

k 287:844 exp



12950
8440
0:9028t
exp
RT
RT
(5)

k 1:8064 exp



8440
8440
1:8064t
exp
RT
RT
(6)

3.2

Assumptions

Outlet liquid

Figure 2. Gas-liquid contact in the MBCR (left) and the model of the differential element of stage
j (right).

The mathematical model is based on the following assumptions:

The reactor operates at isothermal condition; therefore, no
energy balance equation is needed.
Each component entering into the successive stage, in gaseous or liquid form, disperses and its concentration does
not change while leaving the stage.
There is no resistance against gas diffusion; therefore, the
concentration of component i1) in the gaseous bulk phase is
equal to its gaseous concentration at the gas-liquid interface.
Nitrogen is an inert gas and does not participate in the reactions.
H2O does not participate in the reactions, and its production is just due to HP decomposition.

3.3

Inlet gas

3.3.1 Liquid Phase Equations

The component mass balance equation on element dx is written as:




n_ conv:
n_ diss:
n_ conv:
n_ diss:
il;j
il;j 
il;j
il;j 
xdx
x
(7)
i
Ni;j ri;j el;j S dx
t
+ for the production of component i
for the consumption of component i
The convective and the dispersive terms of mass transfer
shown in the differential element of stage j for liquid phase
are:




n_ conv:
il;j  Sul:j Cil;j x

Model Development

To develop the dynamic model of the MBCR based on the

ADM in both gas and liquid phases, mass balance equations
are written on a differential element for each stage of the reactor (Fig. 2).

1)




Cil;j 

n_ diss:

Se
D

l;j
l;j
il;j
x
x x

(8)

(9)

The mass transfer equation for the diffusion of component i

from gas into liquid phase is:

List of symbols at the end of the paper.

Chem. Eng. Technol. 2013, 36, No. 5, 819828

2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.cet-journal.com

822

J. Behin, T. Shojaeimehr



Ni;j x Kli aj Cil;j

Cil;j Sdx

(10)



n_ diss:
ig;j 
x

 can
Cil;j



Cig;j 
x x

(21)

be calculated by means of phase equilibrium law:

ui;j Cl;j RT
C
ci;j Pisat: ig;j


Cil;j

Seg;j Dg;j

(11)

Substitution of Eqs. (8)(10) in Eq. (7) and simplification

gives:

Substitution of Eqs. (20), (21) and (10) in Eq. (19) and simplification gives:
eg;j Dg;j

2 Cig;j
x2

ug;j Cig;j
x


Kli aj Cil;j

Cil;j eg;j

Cig;j
t
(22)

Cil;j ul;j Cil;j


Kli aj Cil;j

x
x2
Cil;j
el;j
t

el;j Dl;j

with boundary conditions:


Cil;j 
0
x x0


Cil;j xL Cil;j1 x0


el;j Dl;j Cil;j 
ul;j
x xl

and the initial condition:


Cil;j t0 0

Cil;j ri;j el;j

(12)


Cig;j 

i1


Cig;j 
0
x xL

8t > 0

(14)

and initial condition:


Cig;j t0

(15)

Cil;j
(16)

Cl;j
ri;j el;j el;j
t

8t > 0

(17)



ul;j t0 ul;0 t0

8t > 0

(18)

x0

3.3.2 Gas Phase Equations

The component mass balance equation in the gas phase on element dx is written as:


i


_ diss:
_ diss:
(19)
n_ conv:
n_ conv:
Ni;j
ig;j n
ig;j 
ig;j n
ig;j 
t
x
xdx
The convective and dispersive terms of mass transfer shown
in the differential element of stage j for gas phase are:




n_ conv:
(20)
ig;j  Sug;j Cig;j x
x

www.cet-journal.com

8t > 0

(23)

8t > 0

(24)

8x > 0

(25)

The total mass balance equation in the gas phase is:

ug;j Cg;j
x

n
X


Kli aj Cil;j

Cil;j eg;j

i1

Cg;j
t

(26)

with boundary and initial conditions as follows:



ug;j x0 ug;j 1 xL

8t > 0

(27)


ug;j t0 0

8x > 0

(28)

3.4

with boundary and initial conditions:



ul;j 
ul;j1 
xL

Cig;j


eg;j Dg;j Cig;j 

xL
ug;j
x x0

(13)

Because of the gas components solubility in the liquid phase

and the evaporation of liquid components, the mass flow rate
of the gas and the liquid along the reactor is not constant and
another equation is needed to explain the change of liquid
velocity. The total mass balance equation in the liquid phase is
used for this purpose:

n 
X

x0

8t > 0

8t > 0

n
2 Cl;j ul;j Cl;j X

el;j Dl;j

Kli aj Cil;j
x
x2
i1

with boundary conditions:

Parameters Estimation

The UNIFAC method was chosen for the prediction of ci,j. Solubility of O2 and N2 in IPA were predicted by Henrys law
[24]. Different hydrodynamic parameters were predicted by
empirical or semi-empirical equations. Tab. 1 shows the relations which were chosen for the estimation of these parameters.

3.5

Solving the Model Equations

The set of model equations included 10 second-order (component mass balance) and 2 first-order (total mass balance) nonlinear partial differential equations that had to be solved
simultaneously. There were 12 variables to be determined, including the concentration of O2, N2, HP, IPA, ACTN and H2O,
in two phases. To apply numerical methods, each stage of the
reactor was divided into 10 small equidistant elements; there-

2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Chem. Eng. Technol. 2013, 36, No. 5, 819828

Bubble column reactor

Table 1. Hydrodynamic parameters used in the model.

Parameter

Equation

Gas phase dispersion coefficient

Field & Davidson [25]

Dg 9:36 10

Liquid phase dispersion coefficient

Joshi & Sharma [26]

D1 0:31dR1:5 u1;C ;

Bubble rise velocity

Lin et al., 1996 [27]
Sauter-mean bubble diameter
Akita & Yoshida [28]
Gas holdup
Akita & Yoshida [28]

ug
eg

poses to O2 and H2O. HP decomposition was more than the level

obtained in experiments. Considering the IPA, the difference between
the model and the experimental
data is as slight as that of HP.

!3:56
dR1:33


u1;C 1:4 dR g ug

eg uB



r
2:14
0:505dB g
q1 dB
!
 2
 0:5  3  0:12
ug
dS
d q g
g dR
p

26 R 1
r
dR
t21
g dR
!
1
 2
  31
eg
ug
dR q1 g 8 g dR 12
p

a

4
r
t21
g dR
1 eg

1
3

4.2

uB

3
 1 
3 
k1 a dR2
t1 2 dR ug 5 dR2 q1 g 5
0:452
r
Dil
Dil
t1

fore, a set of 720 algebraic equations was solved in each time

interval. A computer program, using an explicit method, was
developed in Matlab 6-2 to solve all the equations simultaneously.

u2g

Steady-State Simulation

Fig. 4 shows steady-state concentration profiles of all the components in the liquid phase along the
reactor. The concentrations of HP
and ACTN are increased from the
top to the bottom of the reactor,
and this trend is reversed for IPA.
The amount of HP in all stages of
the reactor is always less than that
of ACTN. This small difference in
concentration was predictable and

0:12

a = 2 for non-electrolyte solutions

Volumetric mass transfer coefficient
Kawase et al. [29]

823

!7
60

g dR

Table 2. Operating conditions of the pilot plant oxidation reactor.

Stream

Mass flow rate

[103 kg s1]

Composition
[wt %]

Intlet gas

Results and Discussion

4.1 Validation of Model Results

In order to check the accuracy of the proposed model, the
computer program was run in steady-state operating conditions. Tab. 2 lists the operating conditions of the MBCR. The
inlet mass flow rate of the gas and liquid, and their compositions, temperature and pressure were considered as given and
constant.
Experimental data related to steady-state conditions of the
MBCR and the simulation results are presented in Tab. 3.
Comparing the compositions of the components in the outlet
gas and liquid shows a good agreement of the model with the
experimental data. The major components in outlet gas stream
are N2, O2 and IPA. The maximum error (9.47 %) is related to
O2. Its concentration distribution along the reactor was affected by mass transfer and not by the reaction rate. The minor
components in outlet gas streams are HP, ACTN and H2O.
The observed modeling error (between 710 %) is related to a
tiny amount of these components. The significant components
are HP, IPA, ACTN and also H2O. The predictions for IPA and
ACTN components are very good, with a maximum error of
0.69 %, but the most significant difference was observed in
H2O and the produced HP.
Steady-state concentrations of HP and IPA in the outlet liquid stream collected from each stage of the reactor were compared with the measured data from the reactor (Fig. 3).
Because of the presence of low amounts of AA, HP decom-

Chem. Eng. Technol. 2013, 36, No. 5, 819828

O2

1.475

22.879

N2

4.936

76.563

HP

0.000

0.00

ACTN

0.000

0.00

IPA

0.000

0.00

H2O

0.036

0.558

6.447

100.000

O2

0.000

0.00

N2

0.000

0.00

HP

0.008

0.077

ACTN

0.075

0.719

IPA

8.681

83.247

H2O

1.664

15.957

10.428

100.000

Total
Inlet liquid

Total

*Pressures of inlet and outlet gas: 22 bar(g) and 21 bar(g), respectively. Temperature in each stage: 130 C. Mass flow rate of
inlet liquid: 37.5 kg h1, mass flow rate of inlet gas: 23.2 kg h1.

it is due to the decomposition of HP, whose effect was included in the rate equations. The dependency of rate equations
on HP concentration caused an increase of concentration gradients (for all components) in the lower stages of the reactor.

2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.cet-journal.com

824

J. Behin, T. Shojaeimehr

Table 3. Comparison of the experimental data and the simulation results at steady state conditions.
Stream

Mass flow rate

experimental
[103 kg s1]

after consumption in the reaction,

caused uniformity and its concentration reached nearly zero compared with the other components
in the liquid phase. The concentration of N2 remained nearly constant. There is a linear relationship
between the formation rate of HP
with the concentration of IPA. In
the early stages of the IPA oxidation, HP is not decomposed. In the
later stages, when the accumulation of ACTN in the reaction environment becomes considerable,
ACTN is oxidized to AA and some
amounts of formic acid. This acid
leads to the rapid decomposition
of HP. Acidic degradation products
affect the oxidation rate and cause
some problems in controlling the
reaction rate in high-level conversions. A continuous multistage
process can ameliorate the problem
and improve the yield.
Steady-state concentrations of
gas-phase components versus reactor height are plotted in Fig. 5.
Due to absorption and consumption in the liquid phase, the concentration of gaseous O2 was decreased from the bottom to the top

Composition
model
[103 kg s1]

error
[%]

experimental
[wt %]

model
[wt %]

error
[%]

Outlet gas
O2

0.169

0.155

8.28

3.275

2.965

9.47

N2

4.838

4.928

1.86

93.760

94.262

0.53

HP

0.000

0.001

0.000

0.018

ACTN

0.025

0.023

8.00

0.485

0.440

9.20

IPA

0.114

0.108

5.26

2.209

2.066

6.50

H2O

0.014

0.013

7.14

0.271

0.249

8.35

5.160

5.228

1.32

O2

0.019

0.018

5.26

N2

0.008

0.008

2.50

HP

1.242

1.325

6.68

ACTN

2.539

2.558

0.75

IPA

6.031

6.032

H2O

1.781

Impurities

Total

100.00

100.00

Outlet liquid

Total

0.155

5.31

0.069

0.069

0.06

10.668

11.374

6.61

21.809

21.959

0.69

0.02

51.804

51.781

0.04

1.708

4.10

15.298

14.662

4.16

0.022

0.000

0.189

0.00

11.642

11.649

100.00

0.06

3,0

100.00

11

10

Axial Dispersion Model

Experimental data

2,5

9
1,5
8
1,0
stage 5 stage 6

0,5
stage 1 stage 2

Water

6
5

stage 1

stage 3 stage 4

stage 2

Acetone
Hydrogen peroxide

Oxygen & Nitrogen

0

Reactor height (m)

The concentration gradient for the upper stages of the reactor

is decreased and that for the 6th stage is nearly zero. Having
more than six stages leads to an increase of the by-product AA,
and thus, impairs the yield of the main product, which is not
economical. Such an effect is evident and it is obviously observed from the first stage of reactor. H2O does not participate
in the reaction and very little changes are attributed to its
evaporation. Small amounts of the remaining dissolved O2,

www.cet-journal.com

stage 6

-1

Figure 3. Liquid concentration of HP and IPA along the reactor

(steady-state conditions).

stage 5

stage 3 stage 4

0,0

Cil (kgmol/m3)

2,0

Isopropyl alcohol

10

CIPA ,l (kgmol/m3)

CHP ,l (kgmol/m3)

0.163

10

12

Reactor height (m)

Figure 4. Concentration of all components in the liquid phase
along the reactor (steady-state conditions).

of the reactor. Its concentration and its gradient in the top

stage of the reactor got near zero due to the relatively high solubility of O2 at high pressure. A high production rate in the
lower stages of the reactor causes a high rate of absorption of
O2 into the liquid. Because of evaporation, IPA and H2O concentrations in the gas phase increased from the bottom to the
top of the reactor. The concentration gradient in the bottom

2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Chem. Eng. Technol. 2013, 36, No. 5, 819828

Bubble column reactor

0,25

3,5
stage 6

0,15

C HP, l (kgmol/m3)

Isopropyl alcohol
Acetone

0,1
Water
0,05
Oxygen

stage 2

stage 3

stage 4 stage 5

stage 6

16 atm

2,5

18 atm

2
1,5
1
22 atm

0,5

24 atm
26 atm

Hydrogen peroxide

0
0

-0,05
0

10

12

Reactor height (m)

Figure 5. Concentration of all components in the gas phase
along the reactor (steady-state conditions).

stages of the reactor is higher than at the top of it. Fresh air entering into the reactor does not contain IPA and H2O; therefore,
there is a high chemical potential for mass transfer from liquid
to gas and the rate of evaporation is at its highest.
The mass transfer process occurs when two gas and liquid
phases are in contact. It causes changes in gas and liquid mass
flow rate. Fig. 6 shows the increasing gas and liquid superficial
velocities from the bottom to the top of the reactor. Absorption of O2 into the liquid and evaporation of IPA and H2O to
gas changes the total gas and liquid volume along the reactor
and then changes the superficial velocities. These changes are
in the tolerance range of column hydrodynamics and have a
negligible effect on the calculated hydrodynamics parameters.
1,5

Gas
1,3
1,1

10

12

Reactor height (m)

Figure 7. Concentration of the liquid phase HP along the reactor
(steady-state conditions).

evaporation decreases, and then the liquid holdup increases in

the reactor. Likewise, HP concentration decreases, but this
trend is reversed for the amount of production. With the
increase in pressure from 22 atm to 26 atm, the mass flow rate
of HP increases from 5.13 kg h1 to 5.30 kg h1. The HP concentration decreases from 2.65 kg mol m3 to 2.21 kg mol m3.
Increasing the pressure leads to a decrease in the evaporation
of liquids, thus the total amount of liquid increases while concentration decreases. Generally, it can be demonstrated that increasing the input gas causes an increase in O2 concentration
in the gaseous phase and an increase in the dissolved O2.
Therefore, the available reacting O2 increases and consequently
increases the amount of production. Such a change causes a
decrease in evaporation of the liquid and the condenser load,
but in turn, it increases the load of the gas compressor.
The effect of temperature on HP concentration is shown in
Fig. 8. Rising the temperature up to 130 C increases the mass
flow rate and the concentration of HP by increasing the reaction rate constants, but increasing from 130 C to 140 C
decreases the mass flow rate of HP. This is due to the increase
in the rate of HP decomposition at high temperatures. Increas-

Liquid
0,9
3,5
stage 1 stage 2

0,7

stage 3 stage 4 stage 5

stage 6

3
stage 3 stage 4 stage 5

stage 1 stage 2

stage 6

0,5
0

10

12

Reactor height (m)

Figure 6. Concentration gas and liquid velocity along the reactor
(steady-state conditions).

The effects of operation conditions, such as pressure and

temperature, on reactor performance and production of HP
are also simulated. Fig. 7 shows the steady-state concentration
profiles of HP along the reactor height for various chosen inlet
pressures, while assuming other operation conditions to be
constant. Mass flow rate of HP increases as the pressure
increases. With the increase in pressure, the amount of liquid

Chem. Eng. Technol. 2013, 36, No. 5, 819828

CHP, l (kgmol/m3)

Dimensionless superficial velocity (-)

stage 1

0,2

Cig (kgmol/m3)

stage 5

stage 4

stage 1 stage 2 stage 3

825

2,5

140C

2
130C

1,5
125C

1
0,5

120C
110C

0
0

10

12

Reactor height (m)

Figure 8. Concentration of HP along the reactor at various temperatures (steady-state conditions).

2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.cet-journal.com

826

J. Behin, T. Shojaeimehr

ing the liquid temperature enhances the evaporation of liquids,

thus the total amount of liquid decreases while the concentration increases.

ing the results shows that the mass transfer phenomenon is

faster than the chemical reaction phenomenon; therefore, the
BCR operates at a slow reaction-absorption regime.

4.3

4.4

Start-Up Simulation

Fig. 9 shows the concentration of liquid components in the

outlet stream (bottom) of the reactor versus time from the
start-up to the steady-state moment. The slope of each curve
increases with the time, which in turn increases the rate of
reaction by forming HP. Based on reaction kinetics HP formation affects the reaction of other products. Considering the residence time of liquid in this reactor (18 h) and the presence of
all inlet liquid components in the outlet liquid after about 2 h,
the assumption of the ADM is confirmed.
10
Isopropyl alcohol

The influence of a change in the load variables on the HP concentration profile along the reactor in the steady-state normal
operating condition was investigated. Therefore, a stepwise
change in the inlet concentration of HP from 0.1 wt % (steady
state) to 0.3 wt % was considered. Fig. 11 shows the concentration profile of HP along the reactor 6 h, 12 h and18 h after the
change. The liquid residence time in each stage was 3 h and the
reactor required nearly 18 h to achieve a new steady-state condition. Increasing the HP concentration from 0.1 wt % to
0.3 wt % in the inlet liquid led to an increase in its outlet concentration (less than 0.2 wt %), which had no technical and
economical advantage.

8
7

Water

6
Steady-state
conditions

2,5

3
Acetone

2
1

Hydrogen peroxide

Oxygen & Nitrogen

-1
0

10

12

14

16

18

new steady state conditions

after 18 hr

after 12 hr

1,5
1

after 6 hr

normal operational
steady state conditions
stage 1

Figure 9. Outlet concentration of all components in the liquid

phase vs. time (start-up conditions).

Outlet concentrations of gas phase components versus time

are shown in Fig. 10. The mass transfer phenomenon takes
place in less than 2 h and steady-state conditions establish over
this period because of the low residence time of gas. Compar0,25

stage 2

0
0

10

12

Reactor height (m)

Figure 11. Effect of a step input in the inlet concentration of HP
on its concentration profile in the liquid phase (dynamic conditions).

Fig. 12 shows the influence of stepwise change in IPA concentration of the inlet liquid from 83.2 wt % to 70 wt % on the
HP concentration profile. Decreasing IPA concentration led to
a lower concentration profile of HP up to 0.2 wt %.

Isopropyl alcohol
0,2
0,15

0,5

Time (hr)

Cig,6 (kgmol/m3)

stage 5 stage

stage 3 stage 4

CHP, l (kgmol /m3)

Cil,1 (kgmol/m3)

Dynamic Simulation

Acetone

0,1

Water

0,05
Oxygen

Steady-state conditions
-0,05
0

10

12

14

16

18

Time (hr)
Figure 10. Outlet concentration of all the components in the gas
phase vs. time (start-up conditions).

www.cet-journal.com

Conclusions

The following conclusions were obtained from simulation of

the MBCR for the oxidation of IPA by O2:
The reactor works in a slow reaction-absorption regime and
the rate of reaction is a controlling parameter in the oxidation process.
The production rate of HP at the upper stages is higher than
that at the lower stages of the reactor due to its relatively
higher concentration and its low kinetics rate at the lower
stages.

2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Chem. Eng. Technol. 2013, 36, No. 5, 819828

Bubble column reactor

Cil

3
stage 5 stage

stage 4

stage 3

CHP, l (kgmol /m3)

2,5
normal operational
steady state conditions

2
new steady state conditions
after 18 hr

1,5
1
after 12 hr

0,5
stage 2

stage 1

after 6 hr

0
0

10

12

Reactor height (m)

Figure 12. Effect of a step input in the inlet concentration of IPA
on the HP concentration profile in the liquid phase (dynamic
conditions).

Incoming air with no IPA and H2O has high chemical

potential for mass transfer, and can cause high evaporation
of liquids in the lower stages.
Increasing the pressure increases the amount of the main
products that, in turn, leads to a higher load for the gas
compressor as well as the operating cost.
Increasing the temperature increases the production rate,
but it is not a good choice for the process optimization due
to a greater rate of HP decomposition at high temperatures.

Acknowledgment
This study was supported financially by the Petrochemical
Research and Technology Company, for which the authors are
grateful. Mr. G. R. Soltanian is acknowledged for providing
laboratory facilities.
The authors have declared no conflict of interest.

Symbols used
aj

[m2m3]

CAc,l

[kg mol m3]

Cg,j

[kg mol m3]

CHP,l

[kg mol m3]

Cig,e

[kg mol m3]

Cig;j jx0 [kg mol m3]

Cig;j jxL [kg mol m3]

specific interfacial area based

on dispersion volume
ACTN concentration in liquid
phase
total concentration of gas phase
in stage j
HP concentration in liquid
phase
gas phase concentration
of component i exit from top
of the reactor
gas phase concentration of
component i entre to stage j
gas phase concentration of
component i exit from stage j

Chem. Eng. Technol. 2013, 36, No. 5, 819828

827

[kg mol m3]

concentration of component i
in liquid phase
Cil;j jx0 [kg mol m3]
liquid phase concentration
of component i exit from stage j
Cil;j jxL [kg mol m3]
liquid phase concentration
of component i entre to stage j

[kg mol m3]
liquid phase concentration
Cil;j
of component i in equilibrium
with its concentration in gas
phase in stage j.
CIPA,l
[kg mol m3]
IPA concentration in liquid
phase
Cl,j
[kg mol m3]
total concentration of liquid
phase in stage j
CO2 ;l
[kg mol m3]
O2 concentration in liquid phase
dB
[mm]
bubble mean diameter
Dg
[cm2s1]
gas phase dispersion coefficient
Dil
[cm2s1]
diffusion coefficient of
component i in liquid phase
Dl
[cm2s1]
liquid phase dispersion
coefficient
dR
[cm]
reactor diameter
dS
[cm]
Sauter mean bubble diameter
g
[m s2]
acceleration due to gravity
i
[]
subscript for components
j
[]
subscript for stages
k
[m1.5kg mol0.5s1] kinetics constant of reaction rate
k
[s1]
kinetics constant of reaction rate
Kla
[s1]
volumetric overall mass transfer
coefficient based on liquid phase
Ni,j
[kg mol s1]
rate of mass transfer of
component i between two
phases in the element of stage j
n_ l;j jxL [kg mol s1]
mol flow rate of liquid enter
to stage j
n_ g;j jxL [kg mol s1]
mol flow rate of gas exit
from stage j
n_ g;e
[kg mol s1]
mol flow rate of gas exit
from top of the reactor
n_ conv:
[kg mol s1]
gaseous mol flow rate
ig;j jx
of component i enter to element
of stage j due to convection
1
n_ diss:
j
[kg
mol
s
]
gaseous
mol flow rate of
ig;j x
component i enter to element
of stage j due to axial dispersion
1
n_ conv:
j
[kg
mol
s
]
liquid
mol flow rate of
x
il;j
component i exit from element
of stage j due to convection
n_ diss:
[kg mol s1]
liquid mol flow rate of
il;j jx
component i exit from element
of stage j due to axial dispersion
Pisat:
[atm]
vapor pressure of component i
R
[kcal kg mol1K1] gas universal constant
rHP;l [kg mol m3s1]
reaction rate of HP production
in liquid phase
rO2 ;l [kg mol m3s1]
reaction rate of O2 consumption
in liquid phase

2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.cet-journal.com

828

J. Behin, T. Shojaeimehr

ri;j

[kg mol m3s1]

S
T
t
uB
ug
ul
ul,C

[m2]
[K]
[s]
[cm s1]
[cm s1]
[cm s1]
[cm s1]

rate of chemical reaction

of component i in element
of jth stage
cross section area of reactor
temperature
time
bubble rise velocity
superficial gas velocity
superficial liquid velocity
liquid circulation velocity

Greek letters
a
ci,j

[]
[]

el
eg
tl
r
ql
ui,j

[]
[]
[m2s1]
[mN m1]
[kg m3]
[]

empirical constant
activity coefficient of component
i in liquid mixture of stage j
liquid holdup
gas holdup
kinematic viscosity of liquid
surface tension of liquid
liquid density
fugacity coefficient of
component i in gaseous mixture
of stage j

References
[1] B. Gourich, C. Vial, N. El Azher, M. Belhaj Soulami,
M. Ziyad, Biochem. Eng. J. 2008, 38, 114.
[2] W. D. Deckwer, Models for Bubble Column Reactors, Bubble
Column Reactors, 1st ed., John Wiley & Sons, New York
1992.
[3] N. Kantarci, F. Borak, K. O. Ulgen, Process Biochem. J. 2005,
40, 22632283.
[4] S. Nedeltchev, A. Shaikh, M. Al-Dahhan, Chem. Eng. Technol.
2011, 34, 225233.
[5] N. Rados, M. H. Al-Dahhan, M. P. Dudocovic, Catal. Today
2003, 7980, 211218.
[6] H. Debellefontaine, S. Crispel, P. Reilhac, F. Perie, Chem.
Eng. Sci. 1999, 54, 49534959.
[7] K. Muroyama, T. Norieda, A. Morioka, T. Tsuji, Chem. Eng.
Sci. 1999, 54, 52855292.
[8] M. Perdigoto, F. Larachi, R. Lopes, Chem. Eng. Technol. 2013,
36, 137146.

www.cet-journal.com

[9] H. Marschall, R. Mornhinweg, A. Kossmann, S. Oberhauser,

K. Langbein, O, Hinrichsen, Chem. Eng. Technol. 2011, 34,
13111320.
[10] J. Alvar, M. H. Al-Dahhan, Chem. Eng. Sci. 2006, 61, 1819
1835.
[11] H. Marschall, R. Mornhinweg, A. Kossmann, S. Oberhauser,
K. Langbein, O. Hinrichsen, Chem. Eng. Technol. 2011, 34,
13211327.
[12] L. Han, M. H. Al-Dahhan, Chem. Eng. Sci. 2007, 62, 131
139.
[13] M. Lohse, E. Alper, W. D. Deckwer, Chem. Eng. Sci. 1983, 38,
13991409.
[14] J. C. Mecklenburgh, S. Hartland, Chem. Eng. Sci. 1968, 23,
14211430.
[15] S. Schluter, Chem. Eng. Process. 1995, 34, 127136.
[16] I. M. Abu-Reesh, B. F. Abu-Sharkh, Ind. Eng. Chem. Res.
2003, 42, 54955505.
[17] L. Xi, X. Gang, Chinese J. Chem. Eng. 2004, 12, 214220.
[18] W. D. Deckwer, K. N. Tien, G. Kelkar, T. Shah, AIChE J.
1983, 29 (6), 915922.
[19] C. Maretto, R. Krishna, Catal. Today 2001, 66, 241248.
[20] J. N. Sahu, S. Agarwal, B. C. Meikap, M. N. Biswas, J. Hazard.
Mater. 2009, 161, 317324.
[21] V. K. Patil, J. B. Joshi, M. M. Sharma, Can. J. Chem. Eng.
1984, 62, 228232.
[22] K. Schugerl, J. Todt, J. Lucke, A. Renken, Chem. Eng. Sci.
1977, 32, 369375.
[23] T. Kunugi, T. Matsuura, S. Oguni, Hydrocarbon Process.
1965, 44, 116122.
[24] J. M. Smith, H. C. Van Ness, UNIFAQ method, Introduction
to Chemical Engineering Thermodynamics, 4th ed., Mc-Graw
Hill, New York 1987.
[25] R. W. Field, J. F. Davidson, Trans. Inst. Chem. Eng. 1980, 58,
228236.
[26] J. B. Joshi, M. M. Sharma, Trans. Inst. Chem. Eng. 1979, 57,
244251.
[27] T. J. Lin, J. Reese, T. Hong, L. S. Fan, AIChE J. 1996, 42, 301
318.
[28] K. Akita, F. Yoshida, Ind. Eng. Chem. Process Des. Dev. 1973,
12, 7680.
[29] Y. Kawase, B. Halard, M. Moo-Young, Chem. Eng. Sci. 1987,
42, 16091617.

2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Chem. Eng. Technol. 2013, 36, No. 5, 819828