This article was downloaded by: [Carnegie Mellon University]

On: 14 October 2014, At: 23:21

Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,
37-41 Mortimer Street, London W1T 3JH, UK

Chemical Engineering Communications

Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/gcec20

A MATHEMATICAL MODEL OF A BUBBLE COLUMN

REACTOR FOR THE CUMENE OXIDATION
a
A. CHIANESE
a
L'Aquila University Monteluco-Roio L'Aquila , Italy
Published online: 03 Apr 2007.

To cite this article: A. CHIANESE (1982) A MATHEMATICAL MODEL OF A BUBBLE COLUMN REACTOR FOR THE CUMENE
OXIDATION, Chemical Engineering Communications, 17:1-6, 261-271, DOI: 10.1080/00986448208911630

To link to this article: http://dx.doi.org/10.1080/00986448208911630

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained
in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no
representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the
Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and
are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and
should be independently verified with primary sources of information. Taylor and Francis shall not be liable for
any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever
or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of
the Content.
This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://
www.tandfonline.com/page/terms-and-conditions
 Chem. Eng. Commun. Vol. 17. pp. 261-271 0 Gordon and Breach. Scicncc Publishers. Inc.. 1982
0098-6445/82/1701-0261%06.50~0 Printed in the U S A .

A MATHEMATICAL MODEL O F A
BUBBLE COLUMN REACTOR FOR T H E
CUMENE OXIDATION
A. CHIANESE
L'Aquila University
Monreluco-Roio L'Aquila, Italy
Downloaded by [Carnegie Mellon University] at 23:21 14 October 2014

(Received August 3. 1981: fn final form May 10. 1982)

Cumene oxidation to cumene hydroperoxide (CHP) and its following cleavage to

phenol and acetone is the most widely used route for phenol production.
Oxidation takes place in an essentially homogeneous organic phase; as catalyst less
than I % aqueous sodium carbonate solution, dispersed in the organic phase, is used.
Since liquid phase reaction rate is relatively slow, reactors commonly used include
mechanically agitated vessel and bubble type column. With commercial processes
bubble column reactors are mostly adopted.
Here a model to predict behaviour of bubble column reactors for cumene oxida-
tion is presented. Design procedures to simulate bubble column behaviour require
the assumptions of a dispersion type model both in gas and liquid phase. A continu-
ous dispersed plug-flow model (DPFM) has been firstly developed by means of a sys-
tem of nonlinear differential equations and relevant boundary conditions. The nu-
merical method proposed by Liu' for solving this nonlinear two point boundary
value problem was retained. Convergence difficulties and high computer times were
to be faced up to. Therefore the back-flow cell model (BFCM) has been used for both
the phases.2t3This model assumes a lumped description of the gas-liquid contactor
apparatus. The application of BFCM to the reactor considerably reduces the compu-
tational requirements. Furthermore results thus achieved fit the behaviour of indus-
trial unit as well as does a continuous dispersion model,

BUBBLE COLUMN REACTOR DESIGN

In Figure 1 the examined reactor section is presented. In the bubble reactor a sparge
ring o r a distributor plate is used to distribute gas upwards throughout a liquid col-
umn. No mechanical mixing, packing or tray devices are used to improve gas-liquid
contact. The turbulence created by the gas flow provides enough mixing and im-
proves mass and heat transfer efficiency between the two phases. Both organic phase
and aqueous sodium carbonate solution are fed to the bottom, so that flows are co-
current. The liquid phase, from the top of the reactor, is separated from the gas phase
and, after cooling, partially recycled to the reactor. A recycle ratio greater than 3 : 1 is
required to assure the reactor temperature stability.
Downloaded by [Carnegie Mellon University] at 23:21 14 October 2014
 BUBBLE REACTOR FOR CUMENE OXIDATION 263

REACTION SYSTEM

It is well known that liquid phase oxidation of cumene proceeds with a complex radi-
cal mechanism. Hattori kinetic model4 has been assumed here. Oxygen absorption in
liquid phase follows a "slow reaction regime," i.e. a negligible portion of the ab-
sorbed oxygen reacts within the diffusion film at gas-liquid interface.
Main reactions occurring in liquid-phase are:
Downloaded by [Carnegie Mellon University] at 23:21 14 October 2014

Besides above-listed reactions, chain branching reactions bringing about some by-
products such as a-methylstyrene and dicumylperoxide may take place.
Following equation, as proposed by Hattori, could be properly used to evaluate
cumene hydroperoxide overall production rate:

Here k depends on temperature and on oxygen partial pressure. The term k c c C&P ~
represents the C H P production rate, while the second term k' CCHP gives hydroperox-
ide decomposition rate. As shown by reactions 2 and 3, C H P decomposes into
dimethylphenylcarbinol and to acetophenone. Hattori too estimated the ratio of the
rates of the two reactions. However since in his experimental runs the consumed
amount of cumene and C H P was too small, an accurate value of this ratio was not
obtained. It was estimated as about 0.4 regardless of temperature.
Here the three main reactions only have been considered and, based on industrial
reactor yield data, the C H P decomposition ratio equal to 0.66 was assumed.
Thanks to Hattori kinetic model and above assumptions, following rate equations
were derived for reactions 1, 2, 3, respectively:

MODEL EQUATIONS

As pointed out BFCM was used. Model is based on the flow situation, as schemati-
cally shown in Figure 2. BFCM consists of a number of equal sized mixed cells in se-
ries; the first/last cells are fictitious ones, thus imposing boundary conditions of
 Downloaded by [Carnegie Mellon University] at 23:21 14 October 2014

FIGURE 2
Flow diagram of the column reactor.
 BUBBLE REACTOR FOR CUMENE OXIDATION 265
DPFM in finite terms.' The model is based on following assumptions: (a) steady-state
and isothermal conditions, (b) liquid and gas partially backmixing in axial direction,
(c) no radial concentration gradients neither in gas nor in liquid, (d) gas velocity, gas
hold-up and interfacial area changing with axial length, (e) owing to the hydrostatic
head, pressure linear dependence height is assumed. (f) carbonate aqueous solution
and organic liquid physical properties d o not affect each other, (g) nitrogen is not ab-
sorbed into the liquid. With assumption (b) back flows are introduced for simulating
the liquid and gas dispersion. The dimensionless back flow ratio (back flow to con-
vective flow) is obtained from the well-known relations:'
Downloaded by [Carnegie Mellon University] at 23:21 14 October 2014

Material balances are required for each organic component in each cell, but disper-
sion in both phases is taken into account for cumene and cumene hydroperoxide,
only. The approximation on concentration profiles of DMFC, A C F and MET is al-
lowable. since these compounds d o not affect the reaction rates and have only a mod-
erate influence on gas and liquid flowrates and physical properties. In each cell oxy-
gen material balance is separately written for gas and liquid phase.
Cumene and cumene hydroperoxide balance over the j-th cell yields:

DMFC, A C F and MET balance over the j-th cell yields:

Dissolved gas component balance over liquid phase of the j-th cell is:

while over the gas phase is:

 266 A. CHIANESE
Nitrogen overall flow throughout the cell is constant and equal to inlet one. For the
cell j the following equation is written:

Congruence condition with gas phase leaving the j-th cell yields:

y?= 1 -(yp?+ c i k j x j )

Overall molar balance over the j-th cell yields:

Downloaded by [Carnegie Mellon University] at 23:21 14 October 2014

Dankwerts boundary conditions in finite termsS are applied to cumene, curnene hy-
droperoxide and oxygen in liquid phase and for oxygen and nitrogen in gas phase.
For each liquid component we have:

At reactor inlet section

Xo=(xin f YO XI)/(^ f yo)
at reactor outlet section
XN = XOUT

For oxygen and nitrogen in gas phase we have:

At reactor inlet section

yo = (yi" + po y,)/(l + po)
at reactor outlet section
Y N = YOUT

Boundary condition equations allow unknowns evaluation at the reactor inlet and
outlet sections, but are not used when working out solutions for equation set.

SOLUTION METHOD

Number (10 N) algebraic equations are needed to determine (10 N) unknowns, i.e. xj,
y?', y?, G, and L,. As Eqs. (5) to (14) are all nonlinear, an iterative solution proce-
dure is required. Computational time is reduced by tearing equations in two sets.
Marquardt's method6 is applied for solving the first set of equations including Eqs.
( 5 ) , (6), (10) and (1 I), as a result values of x?, xFHP,xpzand y? for each cell are ob-
tained. The derivatives of the first term of equations, as required by Marquardt's
 BUBBLE REACTOR FOR CUMENE OXIDATION 267
method, are analytically calculated. The solution of the second set, including Eqs. (7).
(8). (9). (12). (13) and (14). is carried out inside the solution method of the first set of
equations, by means of a sequential procedure.
Comprehensive numerical calculations were performed for various reactor pa-
rameter set, aimed to establish a proper number of cells. For reactor with larger di-
ameter, that is more than 2 m, the backmixing is very effective and the behaviour is
similar to the continuous stirred reactor one, so that 3-5 cells are deemed enough.
The simulation of reactors with small diameters needs a greater number of cells. In
Figure 3 the dependence of the obtained cumene conversions from the assumed cell
number for an 1 m reactor diameter is shown. It is clear that in this case 8-10cells are
Downloaded by [Carnegie Mellon University] at 23:21 14 October 2014

sufficient to accurately describe the reactor behaviour.

This result is mostly a consequence of fluodynamic regime in liquid phase. Gas
phase fluodynamic model has on the contrary a negligible~ffecton the reactor per-
formances, since the kinetic model is slightly influenced by oxygen partial p r e ~ s u r e . ~
With the examined reactor, simulation runs are carried out on the basis either of dis-
persion either of plug flow model in gas phase: the relative deviation between the cu-
mene conversion with both models is about 1%.
Overall procedure leads to a stable and quick convergence, when sum of quadratic
deviations between the two sides of equations being less than lo-'. With ten cells
computing time is about 3 sec with an UNIVAC 1100/80 computer.
Convergence is still easily reached with the two extreme flow conditions: namely
with gas and liquid phase complete mixing and with the plug flow in both the phases.
On other hand, simulation of these two ideal conditions turns out to be a rather diffi-
cult task, when DPFM is assumed.

N , cell number
FIGURE 3 Influence of the assumed cell number on cumene conversion for a reactor without cold recy-
cle (DR= 1 m. length = 10 m).
 A. CHIANESE

REACTOR PARAMETERS

With a specific bubble column reactor transport parameters are affected by vessel
size and gas and liquid flow rates. These parameters are: heat and mass transfer coef-
ficients, axial dispersion coefficient and gas hold-up. In this field most of work has
been performed with a bench scale laboratory equipments, even if some papers report
results obtained in a plant scale equipments. For oxygen diffusion provision in liquid
phase, the procedure suggested by air' has been assumed: (kr a ) value for reacting
system is derived from the value of this group as calculated for the air/water system
in reactor operating condition, that is
Downloaded by [Carnegie Mellon University] at 23:21 14 October 2014

The mentioned paper also reports the assumed relationship to predict the (kra)airwster
value.
Using a proper axial mixing correlation is outstanding feature for the correct de-
sign of a bubble reactor. More extensive measurements were made by ow ell' in large
diameter columns (maximum I m diameter). Empirical correlations presented in that
work are used in this model. Gas hold up has been extensively investigated, in this
model the relationship of ~ k i t is
a ~used.

SIMULATION O F INDUSTRIAL UNITS

Cumene oxidation reactor section consists of two to four reactors in series. Liquid
cumene and C H P are fed from a reactor to the next one. Bubble column reactors
have sufficient volume to allow a stepwise conversion of cumene up to 20-30% C H P
a t the outlet of the last reactor. Normal temperatures and pressures are in 90-120°C
and 5-8 atm abs ranges."
A train of three industrial reactors in series. with temperatures and pressures in the
above ranges, have been simulated. Reactor diameters between 2.5 and 3.3 m with 4
as diameterAength ratio were used. Comparison between experimental and calculated
results is presented in Table I. Selectivity of all the reaction product is fairly well pre-
TABLE I

Comparison between experimental and calculated reactors yields

'Cumene Selectivity. %
Reactor Stage Conversion. % CHP DMFC ACF
No. Exp Calc Exp Calc Exp Calc Exp Calc

'Calculated on the feeded and recycled cumene.

 BUBBLE REACTOR FOR CUMENE OXIDATION 269
Downloaded by [Carnegie Mellon University] at 23:21 14 October 2014

%reactar l enght
FIGURE 4 CHP in the liquid phase and oxygen in the gas phase computed profiles

dicted by the model. Deviations between calculated and experimental cumene con-
version values for two reactors are not negligible. However the difference is less than
lo%, that is within the same order of magnitude of an usual equipment overdesign.
Large recycle and high dispersion values in the liquid phase of each reactor provide
quite flat profiles of organic compounds liquid concentrations, whereas a remarkable
decrease of oxygen in gas and liquid phase along reactor axis occurs. Calculated pro-
files of CHP mole fraction (in liquid phase) and of oxygen (in gas phase) presented in
Figure 4 refer to stage no. 1 of the reactor section.

CONCLUSIONS

BFCM is a valid tool in simulating a bubble column reactor used for cumene
oxidation.
Contrarily with the continuous dispersion model, here no convergence difficulties
and instability problems were observed. Results confirm that BFCM provides an effi-
cient and convenient method to design gas-liquid flow contactors operating in var-
ious conditions."
The close agreement between operating plant and simulated reactor performance
testifies the validity of the kinetic model for cumene oxidation in liquid phase as pro-
posed by Hattori.
 270 BUBBLE REACTOR FOR CUMENE OXIDATION

NOMENCLATURE

interfacial area, m2/m3

concentration in liquid phase, kmoles/m3
reactor diameter, m
molecular diffusivity of gas i into liquid, m2/h
axial dispersion coefficient, m2/h
gas flow rate, kmoles/h
Henry's law constant. atm-'
Downloaded by [Carnegie Mellon University] at 23:21 14 October 2014

liquid side mass transfer coefficient, m/h

liquid flow rate. kmoles/h
entire cell number
pressure, atm
Peclet number
reaction velocity, kmoles/m3 h
superficial velocity, m/h
liquid molar volume, m3/kmoles
reactor volume, m3
mole fraction in liquid phase
mole fraction in gas phase

Greek lerrers

a stechiometric coefficient
Y backflow ratio in liquid phase
C gas hold up
P viscosity, poise
P backflow ratio in gas phase

Indices

ACF acetophenone
CHP cumene hydroperoxide
CY cumene
DM F C dimethylphenylcarbinol
MET methanol
G gas phase
 I component index
in reactor inlet section
J cell number
L liquid phase
m reaction index
OUT reactor outlet section
B backflow
Downloaded by [Carnegie Mellon University] at 23:21 14 October 2014

REFERENCES

I. Shean-Lin. Liu. Chem. Eng. Science. 22 (1967) 871.

2. Deckwer. W:D.. Inr. Chem. Eng.. 19 (1979) 21.
3. Chianese. A.. Annesini. M.C.. De Santis, R.. and Marrelli. L., The Chem. Eng. JI.. 22 (1981) 151.
4. Hattori. K.. Tanaka. Y.. Suzuki. H.. Ikawa. T.. and Kubota. H.. JI. o/Chem. Eng. o/Japan. 3. (1970)
. -.
77
Hartland. S.. a n d Mecklenbourgh. J.C., Chem. Eng. Science. 21 (1966) 1209.
5.
Marquardt. D.W.. JI. o/ Sac. /or Ind. App. Morh. No. 2 (1963) 431.
6.
Fair. J.R..Chem. Eng.. 74 (1967) 67.
7.
8.
Towell. G . D . . a n d Ackerman. G.H.. Proceeding of the 5th European Symp. o n Chem. Rcac. Eng..
Amsterdam (1972).
9. Akita. K.. a n d Yoshida, F.. Ind. Eng. Chem. Proc. Des. Devel.. 12 (1973) 76.
10. Chem Systems Inc.. "Phenol," Reporr no. 74-5 (1975).
11. Deckwer, W.-D., Hallensieben. J., and Popovic, M.. Conlribution t o 6th C H l S A Congress. Praha
(1978).