You are on page 1of 6

Chemical

Engineering Science
PERGAMON Chemical Engineering Science 54 (1999) 1499-1504
Mass transfer and kinetics in ethoxylation
spray tower loop reactors
E. Santacesaria, M. Di Serio, P. Iengo
Dipartimento di Chimica, Universita di Napoli "Federico II", via Mezzocannone, 4 (80134) Napoli, Italy
Abstract - A mathematical model has been developed for describing the behaviour of the spray tower loop reactors.
Th reactors are characterised by the fact that mass transfer and reaction occurs independently in two distinct zone of
the reactor. The model has been used for simulating the performance of a pilot plant both by assuming liquid drops
internally stagnant and turbulent. The last assumption gives the better results. 1999 Elsevier Science Ltd. All
rights reserved.
Keywords: Spray reactors, Ethoxylation, Mass transfer, Kinetics
INTRODUCTION
Ethoxylation is performed in industry for producing
surfactants and polyglycols. The reaction is normally
promoted by alkaline catalyst (NaOH, KOH) and is
performed in gas-liquid well-mixed reactors, Venturi
loop reactors and spray tower loop reactors.
. In the fIrst two types of reactors the gas phase is
dispersed into the liquid phase; on the contrary, in the
last case, the liquid phase is dispersed in the gaseous
one.
Kinetics and mass transfer models for simulating the
performances of well stirred gas-liquid reactors are
reported in the literature (Di Serio et a!., 1995). A model
to describe the behaviour of well stirred spray reactor has
also been published (Hall and Agrawal ,1990). On the
contrary very few paper have been published on the use
of Venturi loop reactor and spray tower loop reactor
(Santacesaria et al. 1995).
In the present paper the ethoxylation performed in spray
tower loop reactors is studied and a model able to
simulate the behaviour of these reactors is reported.
A MATHEMATICAL MODEL TO DESCRIBE
SPRAY TOWER LOOP REACTOR
PERFORMANCES
A scheme of a spray tower loop reactor is reported in
Fig. 1. As it can be seen, reacting liquid, containing the
catalyst, is sprayed in an atmosphere of gaseous ethylene
oxide (EO). Small drops of liquid are formed and are
promptly saturated of the gaseous reagent.
For small drops the flight times are normally enough for
achieving saturation of ethylene oxide, but are negligible
for the occurrence of the reaction (Di Serio et a!., 1995;
Santacesaria et a!., 1995). The reaction occurs in the
liquid phase and ethylene oxide is consumed, more or
less, from the top to the bottom of the liquid column,
according to the recirculating feed rate. Therefore, we
can recognize that in this type of reactors, mass transfer
and chemical reaction occur independently in two
different zones of the reactor.
.I
-I ~ : ~ - 1 .
----
EO
~ ,
- - I
ex
_'_ _ _ J_
I r
I
roo ......
I xm
h*
Qj
Cl
c:
I
co
rm h ~
cu
(;j
v---- --
cu
.c:
'--r--
hi
... I'
Fig. 1 Spray tower loop reactor scheme
The reaction is exothermic and the temperature of the
liquid column increases from the top to the bottom. An
heat exchanger cools the liquid stream before the
injection in the spray nozzle, placed at the top of the
mass transfer zone, as it can be seen in Fig. 1
Description of the Mass Transfer zone
The mean concentration of EO in the falling drops at
the end of the flight (C\o), can be calculated by
integrating the following equation:
(1)
0009-2509/99/$-see front matter 1999 Elsevier Science Ltd. All rights reserved.
1500 E. Santacesaria et al.lChemical Engineering Science 54 (1999) 1499-1504
from 0 to the average flight time (ttin).
To solve this equation the following data must be
known: (i) the equilibrium solubility of EO (C*EO); (ii)
the average specific surface of the drops (aim); (iii) the
average flight time (iv) the average mass transfer
coefficient (v) the concentration of EO in the
sprayed liquid (c
B
EO)'
Let us consider now each point:
(i)The value of the equilibrium EO solubility (C*Eo)
depends on the temperature, pressure and liquid
composition[1 ].
The solubility of EO in respectively nonylphenol,
dodeacanol, dodecanoic acid and their ethoxylated
oligomers with different number of EO adducts shows a
high non - ideal behaviour (Di Serio et aI., 1995;
Santacesaria et al., 1995).
By comparing different methods for calculating activity
coefficients such as: UNIF AC (Fredeslund et al., 1977) ,
NRTL (Renon and Prausnitz, 1968) and Wilson (Wilson,
1964), Di Serio et al. (1995) observed that Wilson
method was the best one for the description of EO
solubility in binary and multicomponent mixtures.
The Wilson parameters resulted dependent in a simple
way on the mean number of EO adduct in the EO-
oligomers mixtures and independent of the temperature.
In same case also UIFAC gives satisfactory results (Di
Serio et aI, 1996) and is commonly used when
experimental data of solubility are not available.
A correct determination of EO solubility is very
important for a correct evaluation of both mass transfer
and kinetics of the ethoxylation reaction, as it has been
stressed in a recently published paper (Di Serio et al.,
1995; Santacesaria et al., 1995).
(ii) The average interfacial surface area drops is
calculated as:
Surface area of drops
a, =
m Volume of drops
"I 3 - n.d.
6 . I I
I
Where d32 is the Sauter diameter defmed as
d
32
= Ln;d; ILn;d;2
I
(2)
The overall surface area of the flying drops can be
calculated as:
(3)
Sauter diameter can nonnally be estimated
experimentally by a Laser scattering technique. Water is
nonnally used in the experiments. The Sauter diameter
obtained for water can be converted in the Sauter
diameter of other sprayed liquids by using the following
relation (Perry and Green, 1984):
d)2 =
(d)2 )wa .. ,
(4)
The Sauter diameter have been assumed as average
drops diameter in the calculus of the average flight time
and the average kim, too.
In Fig. 2 an example of drop size distribution measured
with the laser scattering technique for a spray nozzle
with a cone angle of 90 is reported. The Sauter mean
diameter for the reported distribution is 21 0
1400
100 200 300 400 500
dlame18r !11m)
Fig.2 Typical distribution of the drop size ofa spray
nozzle with a cone angle of 90
The given distribution has been determined just in a
small volume of the spray cone, but this distribution
does not change very much by changing the position of
the volume sample. This means that coalescence has a
small effect on the drop distribution size.
(iii) The average flight time is a function of the initial
drops velocity and of the average path of the drops
(x
m
),
The average path of the sprayed drops, for a cylindrical
reactor depends on: the distance of the liquid level from
the spray (h*); the angle of cone of drops (n); the
radius of the reactor (r). By assuming that all the
possible path have the same probability, the following,
the following general relation can be obtained:
(5)
Where
(6)
(7)
E. Santacesaria et al.lChemical Engineering Science 54 ( 1999) 1499-1504
(8)
All the mentioned geometrical quantities are reported in
the scheme of Fig. 1.
The average flight time can be calculated by integrating
the following equation corresponding to the uniformly
accelerated motion applied to the flying drops:
where Co is the friction coefficient (Foust et aI., 1967)
and Vo is the initial drops velocity corresponding to:
(10)
where, M is the pressure drop across the nozzle(Perry
and Green, 1984).
(iv) In order to predict the average mass-transfer
coefficient we have to know if the drops are internally
stagnant or well mixed.
For internally stagnant drops the problem has been faced
by Crank (1958) that given an analitical solution for the
relation:
oC;o =D [02C;0 + ~ OC;o] (11)
at EO or} rd or
d
Describing the evolution with concentration profile of
the gas adsorbing molecules in a liquid drop of spherical
shape. The profile for each time t can be calculated with
the relation
the mean concentration of EO in the drops at the end of
the flight time tm. can be calculated with the relation:
(13)
The average mass-transfer coefficient for stagnant
drops falling in a continuos phase can be calculated
using the Johnson et al. (1958) correlation:
The prediction of mass transfer coefficient in a
turbulent or well-mixed phase will depend on the
quantitative description of the fluid dynamic of the
phase that is a problem not completely solved.
Srinivasan and Aiken (1988) have shown that to
describe the mass transfer for internally well mixed
drops the Levich theory developed for describing the
mass transfer into a turbulent liquid can be applied.
From this theory they have derived a correlation useful
to predict the mass transfer coefficient for well mixed
drops falling in a continuos phase:
(v) The value of C
B
EO
corresponds to the EO
concentration at the bottom of the liquid column and
must be calculated by an appropriate kinetic model
applied to the reaction zone. This aspect will be
examined in detail in the next section.
Description of the reaction zone
The drops saturated with EO fall on the liquid surface
forming a layer that begins to react and moves toward
the bottom of reactor.
The EO concentration profile along the liquid column
can be calculated together with the temperature profile,
by assuming as a first approximation a plug-flow
behaviour and by integrating the equations of mass and
heat balance from the top of liquid column (z=O) to the
bottom (z=h\). The assumption of a plug flow
behaviour is justified by the observation that the
difference of temperature between the top and the
bottom of the liquid column is relatively small, that is
about lOOC. Therefore, free convection circulation
should be moderate. On the other hand, as the
conversion of ethylene oxide in the liquid column is
relatively low (about 30%) the assumption of more
complicated model does not bring to very different
conclusion.
Besides, by assuming a pseudo steady-state condition
for the mentioned profiles, the mass and the heat
balance can be expressed with the following ordinary
differential equations:
1501
1502 E. Santacesaria et al.lChemical Engineering Science 54 (1999) 1499-1504
dCEO 41Z1"2
--=---r
dz Q/ EO
(16)
dT Ml dCEO
-=----- (17)
As the reaction is exothermic, the temperature increases
and CEO decreases from the top to the bottom of the
liquid column.
Liquid volume and composition are considered constant
during the integration and can be updated at the end of
each integration, solving the related mass balance
equations.
The change of CEO in the external loop has been
neglected because the volume of this loop is small with
respect to the liquid column inside the reactor.
rEO can be determined when a kinetic model and related
parameters are available.
In the case of ethoxylation performed by using as starters
alkylphenols or fatty alcohols in the presence of alkaline
catalysts the following reaction scheme has been
suggested by Di Serio et al. (1995):
RXH+M+OH- RX-M+ + H2o!
RX-M+ + EO RXEO-M+
Catalyst
formation
nitiation
RX(EOrM+ + EO RX(EO);+l-M+ Propagation
i=1,2, ....
Ke
RX-(EO)jM+ + RX(EO) RX(EO)jH + RX(EO) i-M+
Proton transfer
On the basis of this mechanism, the following rates of
EO consumption in each step can be written:
Initiation rate
(18)
Propagation rates
So the overall EO consumption is:
(20)
The concentration of charged species can be calculated
by combining equations of proton transfer equilibria
with those of material and charge balances.
From proton transfer equilibria reaction, we can write:
Ke
[RXH][RX(EO); M+]
[RX- M+][RX(EO)jH]
(21)
By assuming with a reasonable approximation that:
[RXH] + [RXM+] == [RXH] (22)
it results:
Defming:
n
B
O
= j - M+] (25)
j=O
it is possible to write:
then:
[RX- M+] = __ __

j=1
Consequently:
[RX(EO); M+]
[RXH] + Ke t [RX(EO) j H]
j=1
(27)
(28)
Therefore, it is possible to evaluate the concentration of
each ionic couple by introducing the corresponding
equilibrium constant Ke.
By using the equations (27) and (28) the relation (20)
can be rewrite :
ko[RXH] + kpKe t[RX(EO) jH]
j=1 BOC
rEO = ----------":....::....----- EO
[RXH] + Ke t [RX(EO) j H]
j=1
(29)
E. Santacesaria et al.lChemicai Engineering Science 54 (1999) /499-/504
So the rate of EO consumption depends on the
concentration of respectively: catalyst, EO, starter and
ethoxylated oligomers.
SIMULATION OF THE PERFORMANCES OF A
PILOT PLANT REACTOR
By integrating relation (I) the EO concentration at the
top of the liquid column can be determined. The EO
consumption during the absorption into the drops is
negligible because the flight time is of the order of 10-
2
s,
that is, very small if compared with the residence time of
the liquid in the column. The ethylene oxide conversion
from the top to the bottom resulted about 30% in the
simulated runs.
The integration of relations (16) and (18) gives profiles
of ethylene oxide and of temperature along the liquid
column. In particular the EO concentration, at the bottom
of the reactor, is about the same of the liquid sprayed
after recirculation, while the temperature of the liquid, at
the bottom, is adjusted to the desired value by cooling
the liquid stream with the heat exchanger, before
recirculation.
The integration of equations (12-13) has been made by
considering the liquid column divided in N fmite
elements of equal volume. This procedure of
discretization is necessary for increasing the accuracy of
the calculations, considering that for a discretization step
with N=I the gas-phase behaviour can be updated only
after a time equal to the residence time of the liquid,
otherwise, this time becomes lIN the residence time.
By using the described model the behaviour of a pilot
plant reactor has been simulated for the ethoxylation of
nonylphenol. The values of initiation (kj) and
propagation (k
p
) and of the equilibrium proton transfer
(Ke), used in the simulation are reported in a previously
published paper by Di Serio et al. (1995).
The simulation has been made both by assuming the
drops internally stagnant and turbulent. An example of
simulation is reported in Fig. 3 for the amount of EO fed
as a function of time and in Fig. 4 for the temperature
measured at the bottom of the reactor.
lurbul
""
/.
;"" \

,,"" stagnanldrop
/'
4"
time
(arbitrary unlls)
Fig. 3 Simulation of EO alimentation for a nonilphenol
ethoxylation in pilot spray loop reactor by using
turbulent drop model and stagnant drop model.
atagnanldrop
lime
(arbitrary un"s)
FigA Simulation of Bottom liquid temperature for a
nonilphenol ethoxylation in pilot spray loop reactor by
using turbulent drop model and stagnant drop model.
As it can be seen the assumption of internally turbulent
drops gives place to a better simulation.
As matter of fact by assuming the drops internally
stagnant an EO consumption lower than the one
experimentally observed can be predicted because the
calculated mass transfer rate is much less active than
the one considering the drops internally turbulent.
We have so demonstrate in agreement with other
authors (Srinivasan and Aiken, 1988) that by using
efficient spray nozzle drops formed are internally well
mixed. The consequence is a very high mass transfer
rate giving drops saturation in a short time and space.
ACKNOWLEDGEMENTS
Thanks are due to Pressindustria SpA and Scientific
Design for the fmancial support.
NOTATION
aim = average specific interfacial surface area of
drops (cm
2
cm-
3
)
at = total interfacial surface area of flying drops
(cm
2
)
BO = catalyst concentration (mol cm
3
)
Co = friction coefficient
C
p
= liquid specific heat U g.1 K
1
)
dj = diameter of i fraction drops (cm)
D
EO
= EO diffusion coefficient (cm
2
S-I)
d32 = Sauter diameter (cm)
(d
32
)water = Sauter diameter of water drops (cm)
CEO = bulk EO concentration (mol cm
3
)
c
r
EO = concentration of EO in the point rd
g
(mol cm
3
)
= interface gas-liquid equilibrium EO
concentration (mol cm-
3
)
= bulk EO concentration at Bottom of liquid
phase (mol cm
3
)
= bulk EO concentration at Top ofliquid
phase (mol cm-
3
)
= gravity acceleration (cm
2
S-2)
1503
1504
E. Santacesaria et al.lChemical Engineering Science 54 (1999) 1499-1504
kbn = average mass transfer gas-liquid coefficient
(cm SI)
leo = constant rate of initiation (cm
3
mor
l
SI)
kp = constant rate of propagation (cm
3
mor
l
SI)
Ke = equilibrium constants of the proton
exchange reaction
nj = number of drops with diameter dj
QI = liquid circulation rate (cm
3
SI)
r = reactor radius (cm)
rd = distance from the center of drop (cm)
rEO = rate of EO reaction (mol cm
3
SI)
T = temperature (K)
tm = average flight time (s)
Xm = average path of sprayed drops (cm)
Vo = initial drops velocity (cm s )
v = drops velocity (em SI )
[RXHO] = initial starter concentration (mol cm
3
)
[RXH] = starter concentration ( mol cm
3
)
[RX(EO)jH] = oligomer concentration with i
- adducts in the chain ( mol cm
3
)
Greek Letter
J.1 = viscosity of substrate (cP)
P = density of substrate (g cm
3
)
PG = density of gas phase (g cm
3
)
(J = interfacial tension of substrate (dyn cm
l
)
cp = spray nozzle efficiency
m = reaction heat (j mor
l
)
REFERENCES
Crank, 1., The Mathematics of diffusion, Clarendon
Press, Oxford, 1958
Di Serio, M., Tesser, R., Felippone, F. and Santacesaria,
E., (1995), Ind. Eng. Chern. Res., 34, 4092
Di Serio, M., Yairo, G., Iengo, P., Felippone, F.,
Santacesaria, E., (1996), Ind. Eng. Chern. Res., 35, 3848
Foust, A. S., Wenzel, L. A., Clump, C. W., Maus, L. and
Anderson, L. B. I principi delle operazioni unitarie,
Ambrosiana, Milano 1967.
Fredeslund, A., Gmenhling, J. and Rasmussen, P.,
Vapour-Liquid Equilibria using UNIFAC: a group-
contribution method method, Elsevier Publishing,
Amsterdam, 1977
Hall, C.A. and Agrawal, P.K., (1990) The Can. J. oj
Chern. Eng., 68, 104
Johnson, A. I., Hamielic, A. E., Ward D. and Golding,
A., (1958), The Can. J. Chern. Eng., 36, 221
Perry, R.H. and Green, D. W., Chemical Engineer's
Handbook, 6
th
edition McGraw-Hill Book Company,
New York, 1984
Renon, H., and Prausnitz, 1., (1968), AIChE J., 14, 135
Santacesaria, E., Di Serio and M., Tesser, R., (1995),
Catalysis Today, 24, 23
Srinivasan, Y. and Aiken, R. C., (1988)Chem. Eng. Sci.,
43,3141
Wilson, G. M., (1964),J. Am. Chern. Soc., 86,127

You might also like