1971 - Mass Transfer With Chemical Reaction From Spherical One or Two Component Bubbles or Drops

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 22

Chemical Engineering Science, 197 1,Vol. 26, pp. 647-668. Pergamon Press. Printed in Great Britain.

Mass transfer with chemical reaction from spherical one or two


component bubbles or drops

ELI RUCKENSTEIN, VI-DUONG DANG and WILLIAM N. GILL


Chemical Engineering Department, Clarkson College of Technology, Potsdam, N.Y. 13676, U.S.A.

(Received 3 August 1970)

Abstract-Unsteady state mass transfer between a one or two component bubble (or drop) and the
continuous phase with a chemical reaction occurring either in the continuous or in the dispersed phase
is examined. The main assumption for binary bubbles is that the rate determining step is diffusion in the
continuous phase. Two limiting velocity fields, Hadamard flow (Re + 1) and potential flow (Re s I)
are used in the calculations.
For the single component bubble the solution obtained previously by Ruckenstein for mass transfer
without chemical reaction is generalized to include chemical reaction by using Danckwerts’ trans-
formation and the Duhamel theorem and relatively simple analytical expressions are deduced for
transient and steady state average Sherwood numbers. For binary bubbles the concentration of the
continuous phase depends on the dispersed phase concentration which is a function of time. Con-
sequently the same procedure used for the single component case, together with a material balance on
the bubble, generates an integral equation which is solved by an iterative technique for the concentra-
tion inside the bubble.
To determine the accuracy of simplified models of the type suggested by Higbie an extended pene-
tration model is proposed and used for the situation in which a chemical reaction occurs in the con-
tinuous phase. The results of the simplified model are compared with the exact solutions for one and
two component systems, and the range of parameters for which it is accurate is determined.
The accuracy of the quasi-steady state assumption, based on the steady state equations for the
mass transfer coefficient and on an unsteady mass balance for the bubble, is also examined and it is
shown that this assumption may introduce significant errors under certain conditions. Criteria giving
the conditions under which the quasi-steady state assumption may be used are established.

INTRODUCTION In the first, the rate determining step is mass


MASS TRANSFER to or from bubbles or drops transfer in the continuous phase and in the
dispersed in a continuous phase is of practical second, the dispersed phase controls the rate.
importance in a variety of industrial processes. Various approximations of the velocity field have
Therefore, as one would expect, numerous been used to study these mass transfer problems.
studies related to such systems have been re- When the rate determining step is mass trans-
ported. Because of its complexity, various fer in the continuous phase, the following situa-
simplifying assumptions have been used to make tions have been analyzed: Boussinesq[ 11
the problem tractable for mathematical analysis. examined the case of Re S 1 by assuming
This is necessary even when a single bubble or potential flow. Levich [ 15, p. 801 and Friedlander
drop is considered. Published analyses primarily [4,5] studied the case of Pe % 1 and Re + 1
relate to spherical, constant pressure, single with the Stokes flow approximation. By using
component bubbles without chemical reaction. Hadamard’s velocity profile (Re < 1) Levich
In contrast, the present work deals with both [ 15, p. 4041 examined the case Pe 8=-1. Griffith
one and two component unsteady state systems [7] considered Pe s 1 and used both the Stokes
with chemical reaction. and Hadamard velocity profiles. He also has
Previously, two limiting cases of steady state treated the case of large Reynolds numbers.
mass transfer between a single component Lochiel and Calderbank [ 171 obtained theoretical
spherical bubble (or drop) and the continuous equations for flow around any solid or rapidly
phase through which it moves have been studied. circulating axisymmetric bodies of revolution.
647
E. RUCKENSTEIN, V.-D. DANG and W. N. GlLL

In the presence of capillary active substances numerically the steady state problem for both
the interface between the continuous phase and Stokes and Hadamard velocity fields. Later
the bubble may behave as a solid because the Johnson, Hamielec and Houghton[ 131 extended
tangential velocity component becomes zero the numerical computations of Johnson and
there. In pure systems (without capillary active Akehata to larger Reynolds number and con-
substances) the tangential velocity component sidered second order chemical reaction also.
differs from zero at the interface. Considering Estrin and Schmidt [3] considered chemical
this, Ruckenstein[l9] established equations for reactions of the first order outside the bubble by
the mass transfer coefficient in the continuous using Higbie’s penetration model and they took
phase in which the velocity at the interface into account the time dependent concentration
appears as a parameter, both for small and large of the dispersed phase.
Reynolds numbers. The aims of the present paper are:
When the rate of mass transfer is governed by (1) To solve the unsteady state problem with
the dispersed phase, different flow patterns also first order irreversible chemical reaction for a
have been assumed to apply inside the bubbles single component spherical bubble.
or drops. An equation obtained by Newman[ 181 (2) To study the unsteady mass transfer from
neglects convection inside the bubble completely. two component spherical bubbles or drops with
Kronig and Brink [ 141 have taken into account first order irreversible chemical reaction either in
the circulation inside the drop for small Reynolds the continuous or in the dispersed phase by
number, but large Peclet number, by using assuming that the continuous phase controls the
Hadamard’s velocity field. Handlos and Baron rate.
[9] proposed a turbulent flow pattern inside the (3) To present an extended penetration model
drop by superimposing an assumed random radial and to compare it with the exact solutions.
motion, caused by internal vibration, on the (4) To examine the conditions under which the
internal circulation. Johns and Beckman[l l] quasi-steady state assumption may be used.
have obtained numerical solutions using the Problems (1) and (2) will be solved for two
Hadamard stream function. limiting flow fields, small Reynolds number
Recently the unsteady situation was examined (Hadamard flow) and potential flow (Re 9 1) by
by Ruckenstein[21] for the continuous and dis- combining the transient solution without
persed phases for Hadamard flow and potential chemical reaction for a single component bubble
flow both for single and two component drops. obtained in reference[21] with Duhamel’s
However, the analysis for two component theorem. In problem (3) an Eulerian modification
bubbles was developed for very short times. of Higbie’s penetration theory is combined with
Dang, Ruckenstein and Gi11[2] have examined Duhamel’s theorem and the results are compared
the unsteady mass transfer for binary bubbles by with the more exact solutions under (I) and (2).
assuming that the rate determining step is in the Finally the quasi-steady state assumption, which
continuous phase. is very useful for design purposes, is analyzed.
Chemical reaction in the continuous phase
has been included in the following references: MATHEMATICAL ANALYSIS
Lightfoot [ 161 extended Higbie’s model [ lo] to Consider a two component spherical bubble
include chemical reaction in the continuous moving in a liquid. It is assumed that the system
phase. Gill [6], neglecting the effect of curvature is isothermal; densities, viscosities and diffusivi-
and assuming the flow field to be independent of ties are constant: the shape of the bubble or drop
position and time, obtained an analytical solution always remains spherical; the fluids are New-
for the unsteady state problem with first order tonian and the flow is axisymmetric; first order
irreversible chemical reaction. irreversible chemical reaction occurs in the con-
Johnson and Akehata[l2] have studied tinuous phase and the heat of reaction is negli-
648
Mass transfer with chemical reaction

gible; the continuous phase is nonvolatile; boundary layer on the potential flow, has shown
thermodynamic equilibrium exists at the fluid that for sufficiently large Reynolds numbers the
-fluid interface. Consequently we can write the contribution of the boundary layer to the velocity
convective diffusion equation for the continuous field is negligibly small. Therefore for sufficiently
phase in the form: large Reynolds numbers, one may use the
potential flow distribution

=D a2c+z!E+ &$(sinO$$j]-kc. (1) ,Re % 1. (3)


[ a9 r ar
Since the Schmidt number is large for a liquid
phase the diffusion boundary layer is restricted to Since the depth of penetration of the diffusing
the vicinity of the surface of the bubble and the component is very small, only the velocity field
concentration, C, varies appreciably only inside in the vicinity of the bubble surface is of import-
this thin layer. The region of interest is then with- ance for the diffusion process. Consequently,
in y 4 a where y is the distance measured from one can approximate the velocity components by
the surface of the bubble and a is the radius of the those valid near the interface and replace the
bubble. term (v,/r)(aC/aO) by (v,/a)(aC/at?). Molecular
In contrast to fluid-solid interfaces, at fluid- diffusion in the 6 direction also may be neglected.
fluid interfaces the tangential velocity of the By expanding the velocity components in terms
continuous phase is not zero because internal of y/a, and truncating the higher order terms, one
circulation occurs in the absence of contaminants. obtains
For small Reynolds numbers, in systems free of v~=2voYcos8
capillary active impurities, Hadamard’s expres- ,Ree 1 (4)
vO= -v. sin 19
sions for the velocity components[8] may be
used
for slow flows, and

v,=3UYcos8
,Re+ 1 (5)
ve=-9Usin0

for high Reynolds number flows.


With these approximations, Eq. (1) is
v. = [(y-T)(;)-”

+(7-T)(:)-‘- U] sin0
For either single or binary component bubbles,
where the initial and boundary conditions are
IJL
vo=_zP+&. C(0, r, cl) = co (7)
C(r, m, 0) = Coe+ (8)
When the bubble is moving in the continuous C(t, 0,O) = c*(t) (9)
phase at very large Reynolds number, the effect
of the viscous force on the flow field is not very where it is assumed that the initial concentration
important. Levich [ 15, p. 4431, superposing a of the continuous phase is a constant, CO.
649
CES Vol. 26 No. 5 -F
E. RUCKENSTEIN. V.-D. DANG and W. N. GILL

Because the thickness of the diffusion boundary ~(7, Y, 0) = ~(7, Y, B)epR7 (15)
layer is small, one can assume that the system is
semi-infinite. The bulk concentration is a de- where u is defined by
creasing function of time due to the effect of
chemical reaction as indicated in Eq. (8). Equa-
sin$$--2YcosO* (16)
tion (9) is a consequence of the assumption of aY
equilibrium at the interface; for single com-
ponent bubbles C” is constant but for binary u(0, Y, 0) = 00 (17)
bubbles it varies with time in an unknown U(T,~,@ =wg (18)
fashion. To determine C*(t) a material balance
u(7,0,0) = eRl. (19)
over the bubble is used. This is

4 ,dCd With this transformation, we have eliminated the


r,ac
3 dY I 21=0
chemical reaction term in Eq. (11) and intro-
-Tadt= I0
duced a time dependent interfacial boundary
%=ra2sin 6de -$a3klCd. (10) condition in Eq. (19). To solve these transformed
equations for u we introduce a new function, F,
I. First order chemical reaction in continuous which is related to u by the Duhamel theorem.
phase with single component bubbles or drops Therefore the interfacial boundary condition for
F can be considered to be constant by replacing
By introducing the following dimensionless
variables T with the constant parameter A. Consequently,
the system of equations which defines F is

w= c C* constant
C*’ $=g+Pei sint9%-2YcoseE aY (20)
( >

y=:=Y F(0, Y, 0) = o. (21)


F(T, ~0~0) = w,, (22)
7=- tD
a2 F(T, 0,19) = eRh. (23)
R _ ka2 R _ ka2 Equation (20) was solved previously for constant
D” D
boundary conditions similar to those of Eqs. (2 l),
Pe =1 -- Al.’ 2aU ,Pez=2D 3aU (22) and (23) by Ruckenstein[21] by means of
’ ~/A+-/.L’ D the similarity transformation,

we can rewrite Eqs. (6) (7), (8) and (9) in F(T, Y, 6) = F(q) (24)
dimensionless form as
where

sinfI$--2Ycos0s -Rw, Y
> ’ = 8(0, 7) (25)
i= 1,2 (11)
w(0, Y, 0) = 00 (12) and the solution is
w (7, co, 13)= ti0eeR7 (13)
F(r)) = w,+ (eRA-W,) erfc& (26)
W(T, 0, e) = 1. (14)
To solve this set of equations, first let where 6 is determined by
650
Mass transfer with chemical reaction

( .
>I1
1 - tan2 (O/2) e2Peir -- 1 1 - tan2 (o/2) ezPeiT3
Szsin4B=j$ costi-_Scos3H - (27)
zK > [ + tan2 ( f3/2)e2Pei7 3 1+ tan2 (e/2)e2Pe1’
1

With the F function known, we can write the solution for u immediately, by using the Duhamel
theorem, as
a 7
F(X, T-X, Y, B)dh. (28)
u=G Jo

Thus, from Eqs. (15), (26), (27) and (28), the solution of the original set of Eqs. (1 l), (12), (13) and
(14) can be obtained as

~(7, Y,e) = eeRr-$ [ [wo+(eRA-co,)


0

YSi sin2 8

HI
X erfc 3 ,,2 dh. (29)
2 [ c0se--fcos'8- I+ tan2
[ 1 -tan2 (O/2) e2pei(T-A)
(e/2)e2Pei(T-A)_i 3( 11 --an2
+ tan2 (e/2)e2Pei(T-A)
(O/2) e2Pei(T-A)

It will be shown that the mass transfer results obtained from this complicated solution can be
simplified markedly for small and large values of 7.
The local mass transfer rate is

sin2 f3dX
X (30)
)I1
l- tan2 (e/2)e2Pq(T-A) 1_-an2 (ep)e2PetfT-A) 3 l/2.
c0sB-+0s38--
I+ tan2 (e/2) ezpe+)-7 1 + tan2 (e/2) e2Pe@h)

and the average rate of mass transfer over the surface of the bubble or drop is
P

J 0
2rra2N0 sin Ode = &?ee-R%.t[ (eR"-_o)

X
s
0
77

c0se--~cus"e-
1_-an2 (e/2)
sin3 Bdt?dh
e2Pei(r_A)_l 1 --an2

1 + tan2 (O/2) e2Pei(T-A) 3 1 + tan2 (O/2) ezPei(T-*)


(e/2) e2Ped-A)

HI
3 l/2' (31)

If one defines the mass transfer coefficient of the liquid phase as

kL = c* _;,e” ’ (32)

then the Sherwood number can be calculated by means of the following equation

sin3 BdOdh

HI
X
I_ tan2 (e/2) e2Pei(7-A) 1 1--an!2 (e/2)e2Pei(T-A) 3 112
. (33)
--
c0se--fc0s3e-
1 + tan2 (e/2) e2PeicT-A) 3 1 + tan2 (e/2) ezPei(T-*)
651
E. RUCKENSTEIN, V.-D. DANG and W. N. GILL

After some algebraic manipulations, one gets

Sh,= ’
1 _ W0e_R7

sin3 Ode
X
cos+oss8-- 11 +
- tan2 (e/2) e2PW-- 3
1 11 +
- tan2 (e/2) e2PW l/2

(f9/2)e2P”i’ (O/2)ezPei7

+ReR’J: e~Rzdz~o~~cos~_~cos3R_[l-tan2 ($r~Z~~z_l(l-tan2 (~,2)e2pe~?“I)“‘](34’


3 1 + tan2 (O/2) ezPe@ 3 1 + tan2 (19/2)ezpeiz

Equation (34) can be integrated numerically. or


However, for the limiting conditions of small or
large times new analytical expressions may be ShL = l_;oe-R~E(2Eetig
obtained.
For T 4 0*_5/Pei, the integrals with respect to 8 +-%e-O~5R!Prj__&,,oe-Rr . (36)
may be replaced by 2/a[21], and Eq. (34) is >
approximated as
As R + 0, Eq. (36) reduces to Levich’s equation
e-Rr
[lS,p. 4081 -
ShL =
1 -woe-R7
Sh, = $==(-$--g)“‘~, Re * 1 (37)

and for large Re,

Sh,=k (38)

For T % 0*5/Pei, the integral with respect to 8 in Equations (35) and (36) are two of the new
the first term of Eq. (34) may be replaced by results obtained in the present study.
4/fi[21] and the double integral in the second
term may be split into two parts: for z < 0*5/Pei, II. First order chemical reaction in continuous
the integral with respect to 19will be replaced by and dispersed phase with binary bubbles or drops
2/e for z > 0*5/Pei, the value 4/ti will be The physical problem treated in this section is
used for the same integral. Equation (34) can similar to that in part I except a binary rather
then be approximated very well as than a single component bubble is considered. If
the rate is controlled by the continuous phase,
e-Rr the conditions for this will be discussed later, one
Sh, =
1-6~~e-~T can assume that the concentration of the diffusive
species is uniform throughout the dispersed
.5R/Peize-R~&
+ReRT - phase and is a function of time only.
(P 0 The governing equations for the concentration
distribution in the continuous phase are the same
+ Ae-R"dz
s O.BRIPq ti )I as Eqs. (1 I)-( 14) except, to reflect the time
dependence of the bubble concentration, Eq. ( 14)
652
Mass transfer with chemical reaction

is replaced by The dispersed phase concentration ~~(7) is


determined by a mass balance over the bubble
W(T, 0, 0) = 7QJd(7) (39, given
by

= 2na2 sin 0dO-$ra3k,o,


where H is the distribution coefficient and all the YO

dimensionless variables are defined as before (44)


except w = C/Cd,(C,, is the initial concentration
of the diffusive component of the dispersed which, after evaluating the concentration gradient
phase). and integrating with respect to time, leads to the
Again, we introduce the W-U transformation following integral equation

sin3 OdBdA
x
1 _ tan2 (e/2) eH+?i(~-M 1_ tan2 (o/2) eZPei(T-A)
cos l9---Icos3,- (45)
3 1+ tan2 (e/2) e2Pei(7-A) 1+ tan2 (o/2) ezPe@*) .

given in Eq. (15) and in this case u is defined by It is difficult to obtain a solution of this integral
Eqs. (16), (17) and (IS) together with equation in analytical form; therefore it will be
solved by numerical methods.
%(‘)
u(7,0,13) = He ,$7
. (40)
III. Extended penetration model, first order
chemical reactions in continuous phase with
As before, the U-F transformation given by single or binary component bubbles (or drops)
Eq. (28) is employed and F is defined by Eqs.
In a large number of papers the penetration
(20), (2 1), and (22) together with
theory has been used as a simplified model for
estimating the rate of mass transfer. It uses a
F(r, 0,19) = TeRA. (41) Lagrangian description of the process and
assumes that in each element of liquid flowing
along the bubble surface unsteady mass transfer
Therefore the solution for F is
takes place as if the element were a solid body.
We shall use in what follows a simplified Eulerian
F(T, Y, 0) = w,+ Wd(;eRh-~O description of the mass transfer process which
>
neglects radial convection and assumes that the
Y tangential velocity can be replaced by the velocity
Xerfc6(8,T) (42)
of the center of mass of the bubble. Our purpose
is to determine by comparison with the exact
and the concentration distribution of the con-
solution the accuracy of such simplified models.
tinuous phase is given by
Consequently the convective diffusion equation
for the continuous phase in which a single
Wd(A)e'*_
~(7, Y, 0) = epRT& H

0 component bubble is moving will be approxi-


mated by
Y
erfc 6(8, 7_ h) dh. (43) $$-+.%=,$-kc (46)
I
653
E. RUCKENSTEIN, V.-D. DANG and W. N. GILL

where x is the coordinate measured along the hand, at distances less than tU steady state con-
surface of the bubble downward from the front vection prevails. When t 2 2a/U Eq. (50) pre-
stagnation point. The initial and boundary con- dicts that the entire system is at steady state.
ditions are the same as Eqs. (7), (8) and (9) Actually, as will be discussed later, real system
except the 8 coordinate is replaced by the x behavior departs from this simplified picture
coordinate and the boundary condition at the because of the radial convective effect and be-
front stagnation point of the bubble, which is the cause the tangential velocity depends on distance
same as the condition at infinity in the continuous from the forward stagnation point. However it
phase, is seems interesting to determine the accuracy of
this simpler viewpoint.
C(t, y, 0) = C,ehk’. (47) The mass transfer rate can be obtained by
evaluating the gradient at the surface of the
For x 3 Ut, if one uses the transformation bubble. This yields,
x,=x--t
NILS.= C*D
Eq. (46) becomes

For x d Ur, another transformation is used and moreover

tl = t-x/u

Eq. (46) is then


s Ut. (52)
dCax= D$-kc. t 3 x/U. (49)
The mass transfer coefficient is defined as

Solution of Eq. (48) and boundary conditions


1
(7), (8), (9) and solution of Eq. (49) and boundary kL =
2a(C* - COePkt)[I D“tN~.Jr+
1;; Ldr].
conditions (8), (9), (47) are well known and
these can be combined to give the complete exact
solution of Eq. (46) as
t G$ (53a)

,-t&E erfc ~-d&]+e~~erfc[*fi]+~e~kterf--J==},

c
-=- 1 ,-UGD erfc Y ES +e arvm erfc Y +h (50)
c* 2 1 [ 2-u- $1 U [ 2m 0 E

C Y ,t&
-t-oe-kt et-f
C* 2-u I U’

Equation (50) predicts that the region at a and moreover


distance greater than tU from the front stagna- 2a
tion point behaves as though the only transport 1
kL = N,.,.dx, t 2 $. (53b)
mechanism is molecular diffusion. On the other 2a (C* - COeakt) I 0

654
Mass transfer with chemical reaction

Consequently, the Sherwood number is and

shL=F= 1_ le_RT w(7, X, Y) = ewRTPe-$


WO
od (7 - X/Pe + A)eR(7-X’Pe)eRh
IviiEerfdiG+2(~erfGG+e-RI
vz >
1
Y
-2e -R’(l+Per) , t G$ X erfc dh for-r 2 E (56a)
(54) 2m
>
and
cod(T) has to be determined by taking a mass
Sh, = F = 1 _ ioe_, balance over the bubble as in Eq. (44) and this
leads to

1 dud _ -;
--- C-R'
3 dr
UPe wd (T _ l/pe + 1) eR(~-l/Pe+h)

where J( H
-00
>

aU (57)
Pe=E.

forPer > 1,and


When R + 0, Eqs. (54) and (55) reduce to

(54a)
---
1dWd_
3 dr
pe e-R~
-wo

and --&--$$l-Pe7)

ShL = UW (58)

These new results will be compared with the


for Pe T < 1. After some algebraic manipulation
exact analysis of part I; in this way the magnitude
and integration with respect to time, we finally
of the errors embodied in the extended penetra-
obtain the integral equation for the dispersed
tion theory model may be estimated.
phase concentration as:
For a binary component bubble, the initial and
boundary conditions are the same as Eqs. (7), (8),
(39) and (47). Consequently, we generalize the
solution given by Eq. (50) to satisfy Eqs. (39)
wd (7) = e-MT-l/f+) wd

(A >
and (47) by employing the Duhamel theorem.
i-R e-R(T-t)ud(t) dt
This yields the solutions for the two regions as
T

~(7, X, Y) = e-RT&
J[ (%@)eR”_o >
w +
0
H
0

1
erfc ~
; od(7r _~2)e-R(~--r'+D*)
dAforrcg (56) -woe-RT dpdr’ (59)
2m H >
655
E. RUCKENSTEIN, V.-D. DANG and W. N. GILL

forPer > 1 and

~~(7) = eVRT+R [ e-R’7-th(t) dt-$=[2Pe ffi

for Pe T 5 1. These two integral equations are no significant errors are introduced in the
then solved numerically to determine ~~(7). average Sherwood number by using the velocity
distribution which is accurate only for the region
DISCUSSION
near the interface, and by neglecting the term
Single component results (2/t-)(X/h), which was included in the numeri-
It seems that no solutions for unsteady state cal computation of Johnson and Akehata[ 121.
mass transfer with chemical reaction between a Let us evaluate the time required for the steady
single or binary component bubble or drop and state to be achieved in the case of a single com-
the continuous phase exist. However, numerical ponent bubble (only in this case is the steady
solutions for steady state mass transfer with state achieved in fact). An order of magnitude
single component bubbles or drops are given by estimate of this time can be obtained from the
Johnson and Akehata[l2] for Stokes and Hada- intersection of the expressions for the Sherwood
mard flows and Johnson, Hamielec and Hough- numbers for large and small time, Eqs. (35) and
ton [ 131 for intermediate Reynolds numbers. (36), and for simplicity consider the case w,, = 0.
Therefore it is useful to compare, for Hadamard This leads to the general conclusion that the
flow, the asymptotic behavior of Eq. (34) for dimensionless time 7cs required for the steady
large times with the steady state numerical state to be achieved, times the dimensionless
solution of reference [12]. The comparison is rate constant R is a function of RIPei only. That
made in Table 1 which shows that the differences is
between the corresponding Sherwood numbers
(61)
are small. Consequently, one can conclude that
Table 1. Comparison of steady state Sherwood numbers from the two models for Re 4 I
and Re S 1 and the numerical results obtained by Johnson and Akehata[l2] with o0 = 0

ShL
Sh, (Hadamard flow) Sh, (Potential flow) (Extended ShL
Penetration (Johnson
2aU Model and
RD 0%. (55)) Akehata)
Eq. (34) Eq. (36) l% (34) Eq. (36)

I I03 20.65 2066 35.66 35.72 35.73 21.73


104 65.06 65.17 112.67 I l2S5 112.85 64.63
I05 205.69 206.02 356.75 356.83 356.83
IO” 650.42 65 I .47 1126.6 1128.4 1128.4 -
102 IO 20.03 20.00 - 22.02
102 20.22 19.97 - - - 22.2
103 27.02 26.33 39.53 39.08 40.26 28.27
104 67.02 67.03 113.92 113.93 114.34 66.02
105 206.38 206.61 356.65 357.17 357.30
10” 650.64 65166 1126.7 1128.5 1128.5 -
104 10s 200.33 200@0 200.44 199.99 202.8 202.22
104 202.34 199.72 215.04 208.79 225.2 204.45
105 270. I4 263.25 395.33 390.81 40260
106 672.27 670.27 1139.2 1139.3 1143.4

656
Mass transfer with chemical reaction

For small values of R/Pe* this relationship the Peclet number and in this case the range of
reduces to RIPe,is, 0 < RjPe, G 10. As in Fig. 1 we find
that Eqs. (62) and (63) provide useful estimates
t cs zzla2L ifR 4 1.
(62)
of the time required for steady state.
4 D Pei Pei It is desirable to have explicit analytical cor-
relations for the Sherwood number as are given
When RIPei is large the right hand side of Eq. by Eqs. (35) and (36). To determine the time
(61) becomes constant and the time required for range in which these expressions are accurate,
the steady state to be achieved is inversely some comparisons are made with the exact
proportional to k, solutions obtained numerically in Figs. 3 and 4.
Clearly, Eq. (35) agrees very well with the exact
2 constant results at small values of r, and Eq. (36) is an
t C8 if$s 1. (63)
k z excellent approximation of the exact steady state
values. Additional data comparing Eq. (36) with
Figure 1 gives the numerical solution of Eq. exact values are given in Table 1.
(34) and illustrates the case of a small value of Figures 1 and 2 also present a comparison
the Peclet number, Pe, = 50, and 0 G R/Pe, G between the exact results of Eq. (34) and the
200. It is seen that the time required for the extended penetration theory (EPT) Eqs. (54)

IO” I I ! I I 1 I
- Exact Analysis Pe,=50

Extended Penetration Model

Fig. I. Sh, vs. time. Comparison between the exact analysis and the extended
penetration model for single component bubbles, or drops (Pe, = 50). Results for
both models for R = 103, I O4 are essentially indistinguishable.

steady state decreases rapidly as the rate of and (55). EPT is seen to be an excellent approxi-
reaction increases. This is in accord with the mation for all values of r for both Hadamard and
criteria discussed above. If one were to define potential flow providing R is sufficiently large,
the steady state time as the time required for the say RIPei & 5. For smaller values of RIPe, the
Sherwood number to be within 5 per cent of its agreement depends on the type of flow and the
steady state value, then for RIPe, %- 1, the value of r under consideration.
constant in Eq. (63) can be chosen to be approxi- For potential flow Eqs. (38) and (55a) show
mately unity. that the exact and EPT steady state results are
Figure 2 gives the results for a large value of identical in the limit of R/Pe -+ 0. A simple cal-
657
E. RUCKENSTEIN, V.-D. DANG and W. N. GILL

IO' I

- Exact Analysis

I -.- Extended Penetrotion Model 1 WO’O I

Fig. 2. ShL vs. time. Comparison between the exact analysis and the
extended penetration model for single component bubbles or drops
(Pe, = 105). Results for both models for R = lo6 are essentially
indistinguishable.

culation further shows that Eqs. (36) and (55) large and the EPT is seen to provide a reasonable
also are identical as R/Pe + ~0 and both reduce approximation in general, as is illustrated
to quantitatively in Table I.
With Hadamard flow the situation is quite
ShL = 2fi. (64) different. That is, in the limit of RIPe + 0, the
ratio of the exact to EPT steady state Sherwood
Thus, it is only in the intermediate range of numbers is (l/fl)(p’/(c1.+~‘)) whereas they
RIPez, say 0.1 Q RIPe, < 5, that differences both reduce to Eq. (64) as R/Pe+ w. Thus, the
exist between the exact potential flow and EPT EPT is a poor approximation of Hadamard flow,
results. However, the differences are not very unless RIPe, is large enough in which case the

- Exact Anolysis, Equation (34)

Asymptotic Expressions,

ons (35) and (36)

Fig. 3. Sh, vs. time. Comparison between the exact analysis


(Eq. (34)) and the asymptotic equations (Eqs. (35) and (36))
(PQ = 2 X 105). Results for Eqs. (34) and (3.5) for R = 5 X loj, IO6
are essentially indistinguishable.

658
Mass transfer with chemical reaction

- Exact Analysis, Equation (34) Pe,=50


- l - Asymptotic Expressions, w,=o
Equations (35) and (36)

Fig. 4. Sh, vs. time. Comparison between the exact analysis (Eq. (34)) and the
asymptotic equations (Eqs. (35) and (36)) (PeI = 50). Results for Eqs. (34) and
(35) for R = 500,I03, 5 X 103, IO4 are essentially indistinguishable.

Sherwood number is independent of hydro- trols if the approximate inequality


dynamical parameters as shown in Eq. (64).
Finally, it should be noted that in the calcula-
tions for all of the single component results
presented above, the effect of a moving boundary,
which causes the radial velocity at the inter-
face to be finite, has been neglected.

Two component numerical results is satisfied. For gas mixtures, D/D is on the
With rapid chemical reaction in the continuous order of 1O4or 1O5 and therefore this equality is
phase, and therefore with very rapid rates of satisfied unless the reaction parameter R is very
mass transfer, the rate of collapse of single large. However, it may be very restrictive for
bubbles may significantly affect the rate of mass liquid mixture for which D’/D is of the order of
transfer. In contrast to single component bubbles, unity.
the volume of binary bubbles is approximately Figure 5 illustrates the effect of chemical
constant if the initial concentration of the trans- reaction on the time dependence of the bubble
ferring component is dilute. Thus it is not as concentration for Re ==@ 1. Figure 6 demonstrates
important to consider this moving boundary the same effect for Re 9 1.
effect with binary bubbles. Figures 7 and 8 show the effect of solubility on
We first consider the case with reaction in the the change in bubble concentration with time for
continuous phase only so that k, = 0. Approxi- Re < 1. Figures 9 and 10 demonstrate the same
mate criteria for the validity of the assumption of effect for Re & 1. These figures will be discussed
continuous phase control can be established by in more detail in the next section.
using the additivity of resistence concept with
the steady state expression for the dispersed and The quasi-steady state assumption QSSA
continuous phase mass transfer coefficients. Clearly exact analysis of mass transfer to
This approach shows that the liquid phase con- binary bubbles is quite complicated and it is
659
E. RUCKENSTEIN. V.-D. DANG and W. N. GILL

Exact Analysis
- l - Quasi Steady State Assumption
1.0

OE
li:
%
3

0.2

0.c

I
I
0 0.2 0.4 0.6 0.8
I- x IO5

Fig. 5. Dimensionless bubble concentration vs. dimensionless time for different values of
continuous phase chemical reaction parameter, R, for Re Q 1 (Pe, = 50).

I I

- Exact Analysis Pe,=5x105 R,=O Cd


lAJd=-
- 0 - Quasi Steady State Assumption H -0.01 UJo=o C%
I.C

_ 0.6 ,-
I-J
-k
3

0.2 t-

0.c l-

( I.0

Fig. 6. Dimensionless bubble concentration vs. dimensionless time for different values of
continuous phase chemical reaction parameter, R, for Re B I (PC,= 5 x loj).

660
Mass transfer with chemical reaction

,
Exact Analysis Pe,=500 k,=o
-O- Quosi Steody State Assumption R=lOO wo=o wd
IL

Ofj-

T
%
3

0.2

OS

0 0.02 0.04 0.06 0.00


T

Fig. 7. Dimensionless bubble concentration vs. dimensionless time. Effect of solubi-


lity for Re < I (R = IO*). Multiply time scale by IO-* as indicated to obtain dimen-
sionless time.

Exact Analysis R,=O R=103


Cd
0,-c
-*- Quasi Steady State Assumption Pes= 500 wo=o do
I.

0.00 0.01 0.02 0.03 a04 0.0


T x IO

Fig. 8. Dimensionless bubble concentration vs. dimensionless time. Effect of solubility


for Re < I (R = 103). Multiply time scale by I Oe2, I O-4 as indicated to obtain dimension-
less time.

661
E. RUCKENSTEIN, V.-D. DANG and W. N. GILL

T
0.0 0.1 0.2 0.3 0.4 0.5

-a- Quasi Steady Stote Assumption

I I
OI) 0.2 0.4 0.6 0.7 0.8
r x IO’
Fig. 9. Dimensionless bubble concentration vs. dimensionless time. Comparison be-
tween the exact analysis and the results obtained by means of the quasi-steady state
assumptions for chemical reaction in the continuous phase (R = 104) and Re + 1 (Pe, =
5 X 104). Multiply time scale by IO-’ as indicated to obtain dimensionless time. Results
for both models for H = I, IO are essentially indistinguishable.

desirable to have a simpler method for the inter- The solution of Eq. (66) is
pretation of experimental data and for design.
For these purposes it is useful to introduce the od (7) = e-3ShLT/2ff
+ 3Sh,w,
quasi-steady state approximation, QSSA, and 2(R - 3ShJ2H)
therefore this assumption will be examined. x ( e-3ShLdZH _ e-R~)
(67)
The QSSA implies the use of the steady state
mass transfer coefficient for the continuous where the Sherwood number, ShL, can be
phase, even though transient effects exist, obtained from Eq. (36) in the limit 7 ---, ~0as
coupled with the unsteady state mass balance
over the dispersed phase. By using this assump-
ShL = 2fi et-f e + 4E e--O+RI%. (68)
tion a simple expression for the dispersed phase
concentration as a function of time and the para-
meters Pei, H and R is obtained. The validity of When mass is transferred from the continuous
the expressions established in this manner will be phase to the dispersed phase and chemical
checked with the exact solutions. reaction takes place only inside the bubble
When chemical reaction occurs only in the (i.e. k = 0), the unsteady mass balance for the
continuous phase (i.e. k, = O), the unsteady mass concentration in the bubble leads to
balance for the concentration in the bubble leads
to 4
5na3d$ = 4nazkL(co -2) -$a’k,Cd. (69)
4
y-a ad&
- = -4n-a2kL $$- toe-kt .
(66)
dt ( > Solution of Eq. (69) with zero initial bubble
662
Mass transfer with chemical reaction

‘Exact Analysis \=5xlO4 ‘R,=O Cd


w=-
-*- Quasi Steady State Assumption R=105 w,=o o Cd0

Fig. IO. Dimensionless bubble concentration vs. dimensionless time. Comparison


between the exact analysis and the results obtained by means of the quasi-steady state
assumption for chemical reaction in the continuous phase (R = 105) and Re + 1 (Pe, =
5 X 104). Multiply time scale by IO-* as indicated to obtain dimensionless time. Results
for both models for H = I are essentially indistinguishable.

concentration is then the system will behave approximately as


though it were at steady state.
Over the period of time td, which is required
od (7) = R + 3Sh,,2H [ 1 - e-(R+3Sh@Hb] (70) for significant changes to occur in the concentra-
tion of the dispersed phase, if td is large com-
where ShL can be obtained from Eqs. (37) or (38) pared to tcs, the system will appear to pass
as follows through a series of quasi-steady states. Con-
sequently, the criterion for the quasi-steady
state assumption to apply is, td % tc8.
(71) The conditions for the validity of the quasi-
steady state assumption are first discussed for
Before examining numerical results it is the case of chemical reaction occurring only in
desirable to establish qualitative criteria for the the continuous phase. The steady state time t,,
QSSA. If the interfacial concentration is con- has been determined previously and is given by
stant the continuous phase eventually will Eqs. (62) and (63) for small and large values of
achieve a steady state concentration distribution, Ripe,, which is the ratio of the hydrodynamic to
say in time &. On the other hand, if the con- chemical reaction time scale.
centration of the interface varies with time no Equation (67) shows that the time scale for the
true steady state will exist. However, if the rate dispersed phase is given by
of change of the interfacial, or bubble, concentra-
tion is small enough so that its change in con- aH
td”--.
centration during the time t,, is relatively small, k.
663
E. RUCKENSTEIN, V.-D. DANG and W. N. GILL

Consequently, for the dispersed phase respectively. As before the QSSA is valid if
td 9 tcs; hence
(72)
(78)
a2H R
% 1. (73)
=zG? iGi
Figures 1 l-l 3 show comparisons of the
As indicated previously the QSSA will be QSSA with the exact solutions. There are con-
valid when td S t,,, and therefore as criteria for ditions, predicted by the above criteria, for which
the validity of this assumption we obtain the steady state assumption is reasonable, but
there are also conditions for which the error
obtained by its use is more than 100 per cent.
HdF& P 1
When (RJPei) + (2/H)- < O-005, the
quasi steady state assumption gives error about
and
5 per cent or less.
Hfi% 1 if&% 1. (75)
I CONCLUSION
The principal results of this work are:
These criteria suggest that systems involving
sparingly soluble gases are most likely to be (1) the concentration distributions for single
described accurately by the quasi-steady state component bubbles (Eqs. (29)) when diffusion is
assumption. accompanied by a first order chemical reaction.
Figures 5-10 indicate clearly that the criteria (2) the exact values of the Sherwood number
given in Eqs. (74) and (75) properly describe the for the same situation (Eq. (33)).
regions in which the QSSA is valid. It is seen (3) the approximate values of the Sherwood
that errors of approximately 15 per cent or less number for the same situation for small and
are incurred by using the QSSA, for the entire large values of time (Eqs. (35) and (36)).
ranges of the parameters H, Pet and R investi- (4) the concentration distribution for the same
gated, if HflR > 10 for RIPei > 1 or if Hfii situation (Eq. (50)) obtained on the basis of
b 10 for RIPei < 1. On the other hand the the extended penetration model.
QSSA may introduce very serious errors when (5) the values of the Sherwood numbers for the
either Hfi or HVZi is smaller than 10 as same situation obtained on the basis of penetra-
indicated in Figs. 5-10. Thus, QSSA should be tion model (Eqs. (54) and (55)).
used with care, but when it is applicable it cer- (6) the concentration distribution for binary
tainly simplifies the analysis of binary bubble or bubbles given by Eq. (43) together with the time
drop systems very markedly. dependent bubble concentration given by Eq.
When chemical reaction takes place only in- (45). First one must solve integral Eq. (45) and
side the bubble, the characteristic time scales for the results are then used to calculate the local
the dispersed and the continuous phases are distribution by means of Eq. (43). These results
were used to discuss the quasi steady state
assumption.
(76)
(7) the concentration distribution obtained on
the basis of an extended penetration model of
binary bubbles (Eqs. (56) and (56a)) together
and
with the time dependent bubble concentration
t -3a2 (77)
(Eqs. (59) and (60)). The calculation procedure
” - DPei followed that described under (6) above.
664
Mass transfer with chemical reaction

Exact Analysis Pe2=5x10S R=O Cd


-o- Quasi Steady State Assumption H=IO
wd=-&-

I.i ,-

0.E i-

‘2
57

F.-
0.4

0.0
a 3 0 0.08 0.12 0.16
rxlos

Fig. I I. Dimensionless bubble concentration vs. dimensionless time. Comparison


between the exact analysis and the results obtained by means of the quasi-steady
state assumption for different values of reaction parameter inside bubble for Re * I
(R = 0, Pe, = 5 X I05). Multiply time scale by IO-‘, 10m3 as indicated to obtain
dimensionless time and multiply concentration scale by IO-’ as indicated to obtain
dimensionless bubble concentration.

- Exact Analysis Pe,=5x10s R=O Cd


wd=c,
-a- Quasi Steady State Assumption R,=103
0.6.

1.04

0.0 3.00
0.0 0.4 0.8 _ 1.2 1.6
T Itlo=
Fig. 12. Dimensionless bubble concentration vs. dimensionless time. Comparison
between the exact analysis and the results obtained by means of the quasi-steady
state assumption for chemical reaction inside bubble (R, = 103) and Re * I (R = 0,
Pe, = 5 X 105).Multiply time scale by IO-‘, 10m3 as indicated to obtain dimensionless
time and multiply concentration scale by IO-* as indicated to obtain dimensionless
bubble concentration.

665
CESVol. 26No.5-G
E. RUCKENSTEIN, V.-D. DANG and W. N. GILL

T
0.0 0.2 0.4 0.6
- Exacl Analysis
-o- QSSA

Fig. 13. Dimensionless bubble concentration vs. dimensionless time. Comparison be-
tween the exact analysis and the results obtained by means of the quasi-steady state
assumption for chemical reaction inside bubble (R, = 2) and Re 4 1 (R = 0,Pe,= 50).
Multiply time scale by 10e2as indicated to obtain dimensionless time and multiply con-
centration scale by 1O-l,
1O-* as indicated to obtain dimensionless bubble concentration.

Acknowledgments-This work was supported in part by the H distribution coefficient


National Science Foundation Grant GK 699. E. R. held a k reaction rate constant in the continuous
National Science Foundation Visiting Foreign Scientist
Fellowship during the course of this work. phase
k, reaction rate constant in the dispersed
phase
NOTATION kL average mass transfer coefficient
radius of the bubble N average mass transfer rate
concentration of the continuous phase N U.S. mass flux across the rear part of the
equilibrium concentration of the con- bubble interface, Eq. (5 1)
tinuous phase NM. mass flux across the front part of the
initial concentration of the continuous bubble interface, Eq. (52)
phase NB local mass transfer rate
Cd concentration of the dispersed phase 1 j.~’ 2aU
initial concentration of the dispersed Pel ~/L+/.L’ D
cdo
phase 3aU
D molecular diffusivity in the continuous Pe2 --
20
phase aU
Pe
D’ molecular diffusivity in the dispersed 20
phase R ka21D
erfc (x) l-erf(x)=L G zme-@dt Rkla2
1 D
I
F function satisfies Eqs. (20), (21), (22), Sh 2k,a
D
L
(23)
666
Mass transfer with chemical reaction

r radius coordinate of the system mea- h parameter


sured from the center of the bubble 7) Yl6
t time p viscosity of the dispersed phase
t cs time required for the continuous phase P’ viscosity of the continuous phase
to reach steady state r tDla2
fd time required for the dispersed phase T,, dimensionless time required for the
to reach steady state continuous phase to reach steady
u function satisfies Eq. ( 16), (17), (18) and state
(19) 13 polar coordinate (spherical coordinate
U translational velocity of the bubble system)
u-/L w C/C* for one component bubble
v” 2 j_L+p’ o. Co/C* for one component bubble
v, radial velocity component of the con- w C/Cd, for two component bubble with
tinuous phase with respect to the chemical reaction in the continuous
center of the bubble phase only
vg tangential velocity component of the od Cd/Cd0 for two component bubble with
continuous phase with respect to the chemical reaction in the continuous
center of the bubble phase only
x coordinate measured along the surface w C/C, for two component bubble with
of the bubble from the front stagna- chemical reaction in the dispersed
tion point downward phase only
md Cd/Co for IWO COmpOnent bubble with
x& chemical reaction in the dispersed
y r-a phase only
Y yla
Subscripts
Greek symbols 1 for Hadamard flow
6 thickness of diffusion boundary layer 2 for potential flow

REFERENCES
[II BOUSSINESQ J.. .I. Mrrth. PureAppl. 1905 6 285.
PI DANG V. D., RUCKENSTEIN E. and GILL W. N., ChemicalEngineer, London, No. 241, p. 248,197O.
PI ESTRIN J. and SCHMIDT E. H., Jr.,A.f.Ch.E.J! 1968 14 678.
[41 FRIEDLANDER S. K.,A.I.C)I.E.JI 1957 3 43.
1% FRIEDLANDER S. K.,A./.Ch.E.JI 1961 1347.
161GILL W. N., Chem. Engrs 1962 189.
[71 GRIFFITH R. M., Chem. Engng Sci. 1960 12 198.
PI HADAMARD J., Compt. Rend. 191 I 152 1734.
191 HANDLOS A. E. and BARON T.,A.f.Ch.E.JI 19573 127.
[101 HIGBIE R., Trans. Am. inst. Chem. Engrs 1935 31365.
[I 11 JOHNS L. E. and BECKMAN R. B ..A.1 .Ch.E.J/ 196612 10.
[I21 JOHNSON A. I. and AKEHATA T., Can. J. them. Engng I965 43 10.
[I31 JOHNSON A. I. and HAMIELEC A. E. and HOUGHTON W. T.,A.I.Ch.E.Jl 1967 13 379.
[I41 KRONIG R. and BRINK 1. G..Atml. scient. Res. 1950A-2 142.
[ISI LEVICH V. G., PhysicochemicalHydrodynamics. Prentice Hall, New Jersey 1962.
1161 LIGHTFOOT E. N., A.I.Ch.E.JI 19584499.
LOCHIEL A. C. andCALDERBANK P. H.,Chem.EngngSci. 19647471.
;:i; NEWMAN A. B., Trans. Am. Inst. them. Engrs 193 127 3 10.
[I91 RUCKENSTEIN E., Chem. Engng Sci. I964 19 13 1.
WI RUCKENSTEIN E., Chem. Engng Sci. I968 23 363.
[211 RUCKENSTEIN E., Int. J. Heat Mass Trunsfer 1967 10 1785.

667
E. RUCKENSTEIN. V.-D. DANG and W. N. GILL

R&sun&-On examine le transfert de masse en Ctat non stationnaire entre une bulle (ou goutte) ii un
ou deux composants et la phase continue avec une reaction chimique se produisant soit dans la phase
continue, soit dans la phase dispersee. La principale hypothese pour les bulles binaires suppose que
l’etape determinant de la vitesse est la diffusion dans la phase continue. Deux champs de vitesse limite,
I’ecoulement Hadamard (Re a 1) et l’ecoulement potentiel (Re % I) sont utilises pour les calculs.
En ce qui concerne la bulle a un seul composant, la solution obtenue anterieurement par
Ruckenstein pour le transfert de masse sans reaction chimique est gen6ralisC pour comprendre la
reaction chimique en se servant de la transformation de Danckwert et du theoreme de Duhamel.
Des expressions analytiques relativement simples sont deduites pour des nombres de Sherwood
transitoires et en Ctat stationnaire. En ce qui concerne les bulles binaires, la concentration de la
phase continue depend de la concentration de la phase dispersee qui est une fonction du temps. En
consequence, la m&me pro&de utilist pour la bulle a un composant, avec un bilan de matiere sur
la bulle, engendre une equation integrale qui est resolue par une technique iterative pour la concentra-
tion a I’interieur de la bulle.
En vue de determiner la precision des modeles simplifies du type suggere par Higbie, un modele
de penetration ttendue est propose et utilist pour le cas dans lequel une reaction chimique se produit
en phase continue. Les resultants du modele simplifie sont compares aux solutions exactes pour les
systemes a une et deux composants, et la gamme des parametres pour lequel il est p&is est dtter-
mince.
La precision de I’hypothese de l’etat quasi stationnaire, baste sur les equations en etat stationnaire
pour le coefficient de transfert de masse et sur un bilan de mat&e pour la bulle, est tgalement exam-
ine, et il est demontre que cette hypothese peut introduire des erreurs importantes dans certaines
conditions. Des criteres donnant les conditions dans lesquelles I’hypothbse de I’ttat quasi stationnaire
peut &tre utilise, sont Ctablies.

Zusammenfassung-Stoffiibertragung im nichstatiomiren Zustand zwischen einer Ein- oder Zwei-


Komponponenten-blase (oder einem Tropfen) und der kontinuierlichen Phase, mit Ablauf einer
chemischen Reaktion entweder in der kontinuierlichen oder in der dispergierten Phase, wird unter-
sucht. Die Hauptannahme fiir binare Blasen ist, dass die die Geschwindigkeit bestimmende Stufe
die Diffusion in der kontinuierlichen Phase ist. In den Berechnungen werden zwei begrenzende
Geschwindigkeitsfelder verwendet, nlmlich Hadamard Stromung (Re Q I) und potentielle Stromung
(Re s 1).
Fiir die Ein-Komponenten-blase wird die frtiher von Ruckenstein fur Stoffiibertragung ohne
chemische Reaktion erhaltene Losung so verallgemeinert, dass sie die chemische Reaktion mit
einschliesst, und zwar durch Verwendung der Transformation von Danckwerts und des Duhamel-
schen Theorems, und es werden verhaltnismassig einfache analytische Ausdriicke fiir voriiber-
gehende und stationare durchschnittliche Sherwood Zahlen abgeleitet. Fur binare Blasen h%ngt die
Konzentration der kontinuierlichen Phase von der Konzentration der dispergierten Phase ab, die eine
Funktion der Zeit ist. Die im Fall der Ein-Komponenten-blase angewendeteMethode, zusammen mit
einer Stotlbilanz der Blase, erzeugt folglich eine Integralgleichuna, die mittels der iterativen Methode
fiir die Konzentration innerhalb der Blase gel&t wird- - -
Zur Bestimmung der Genauigkeit vereinfachter Modelle des von Higbie vorgeschlagenen Typs
wird ein erweitertes Penetrationsmodell vorgeschlagen und fur den Fall wo eine chemische Reaktion
in der kontinuierlichen Phase vorkommt verwendet. Die Ergebnisse des vereinfachten Modells
werden mit den ganeuen Liisungen fiir Ein- und Zweikomponentensysteme verglichen, und der
Bereich der Parameter fur den es genau ist wird bestimmt.
Die Genauigkeit der Annahme des quasi-stationlen Zustandes, auf Grund der StationLzustands-
gleichungen fur den Stoffubertragungskoeffizienten und der nichtstationaren Stolfbilanz fur die
Blase, wird ebenfalls untersucht und es wird gezeigt, dass diese Annahme unter gewissen Bedingungen
zur Anwesenheit bedeutender Fehler fiihren kann. Es werden Kriterien aufgestellt, die die Bedin-
gungen geben unter welchen die Annahme des quasistationken Zustands verwendet werden kann.

668

You might also like