Professional Documents
Culture Documents
Laminar Mass Transfer From Porous Tubes and Flat Plates With Wall Resistance
Laminar Mass Transfer From Porous Tubes and Flat Plates With Wall Resistance
Laminar Mass Transfer From Porous Tubes and Flat Plates With Wall Resistance
34 Summer/Fall 1978
Abstract
Mass transfer in laminar flow due to combined convection and diffusion through porous,
tubular and fiat membranes which have wall resistance has been examined. The analysis is an
extension of the Graetz method and includes membrane resistance, membrane partition
coefficients and fluid injection and extraction. The results are expressed as a comparison of
mass transfer with wall convection to mass transfer without wall convection. Mass transfer
with wall convection can be stated in terms of an overall mass transfer coefficient alone when
the overall mass transfer coefficient is defined in terms of the wall Sherwood number, modified
Graetz number and membrane partition coefficient.
It was found that mass transfer with wall convection, when expressed as a function of the
dimensionless mass transfer coefficient, is independent of the wall Sherwood number and
membrane partition coefficient for the normal range of mass transfer parameters. A simple
empirical relationship involving only the fluid extraction ratio and dimensionless mass
transfer coefficient fits the composite mass transfer results for both tubes and fiat plates. A
series of experiments utilizing a tubular mass exchanger verified the theoretical predictions.
Nomenclature
A surface area for mass transfer, ~DL
A. orthogonal coefficient
c concentration
c* dimensionless concentration, c/c~
D tube diameter
diffusivity in liquid
E. fluid bulk coefficient
h half channel height
ho overall mass transfer coefficient
k partition coefficient
L tube length
127
128 H.R. DAVIS
subscripts
i inlet
o outlet
w inner wall
b bulk
d outer wall
1. Introduction
The p r o b l e m of h e a t transfer in l a m i n a r flow between parallel plates or
circular tubes with wall mass transfer has been studied by m a n y authors.
E a r l y w o r k b y Terrill a n d W a l k e r [1] for flat plates a n d P e d e r s o n a n d
K i n n e y 1-2] for tubes p r e s e n t e d certain results for h e a t transfer where the
wall c o n d u c t i v i t y was extremely large. R a i t h b y [-3, 4] c o m p l e t e d solutions
for a large range of fluid velocity profiles in b o t h circular tubes a n d
r e c t a n g u l a r ducts b u t w i t h o u t including wall c o n d u c t i v i t y as a p a r a m e t e r .
R a i t h b y suggested that heat transfer d e p e n d s a l m o s t exclusively on the
wall P6cl~t n u m b e r where P e w = R e w P r . In all previous w o r k the results,
either b u l k t e m p e r a t u r e or Nusselt number, have been expressed as a
function of the G r a e t z n u m b e r for various wall P6cl6t numbers.
LAMINAR MASS TRANSFER 129
This paper resulted from the desire to be able to predict the effect of ftuid
transfer from porous tubes on the performance of diffusional mass ex-
changers. In a practical tubular mass exchanger there is considerable
resistance to the movement of a solute species by concentration gradient
controlled diffusion both through the fluid boundary layer and through the
tube wall. A paper by Davis and Parkinson [5] examined the problem of
mass transfer from tubes with wall resistance but without wall convection.
That solution will now be extended to include mass transfer with wall
convection over a range of parameters of interest in the design of mass
exchangers such as artificial kidneys. The results will be expressed in a
general form which is convenient for comparison of mass exchanger
systems, either tubular or flat plate.
Rew DQw
IQwl-< Qi (1)
Re i 4LQi
Thus, in laminar flow, the wall Reynolds number will be very small. Terrill
and Thomas [61 determined the fluid velocity profiles in porous tubes
when the fluid velocity through the wall was constant. They demonstrated
that at a critical wall Reynolds number of approximately 4.6 for the suction
case, the fluid separates from the wall. This paper will consider only fluid
injection or suction through the tube wall for wall Reynolds numbers less
than unity. Hence, the fluid velocity profiles determined by Macey [7] for
negligible inertia terms in the equations of motion will give a reasonable
approximation to the problem. The advantage of utilizing these appro-
ximate velocity profiles results from one dimensionless number, the wall
P6cl6t number, being used to characterize the problem.
130 H.R. DAVIS
The analysis of mass transfer from non-porous tubes given by Davis [5]
for various fluid velocity profiles in the tube showed that there was little
effect of the velocity profile on mass transfer. Thus, it is likely that use of the
approximate velocity profiles of Macey will yield only small errors in the
mass transfer analysis.
~c t3c @ ~3 ( ~ c )
v,~-r +v~ c~z -- r ar r ~ r (2)
dc d2c
Vw~-s = ~ -ds
~- (3)
Here it has been assumed that the wall is thin and that axial diffusion in the
wall can be neglected. The mass concentration in the fluid at the wall can be
related to the mass concentration in the wall at the fluid boundaries by
means of a partition coefficient. This coefficient, k, is defined as the ratio of
the equilibrium concentration in the wall to the concentration in the fluid
at the boundary. The boundary conditions for the porous wall are,
c = k c w, s = 0
(4)
C = kCd, S = Sm
Solving the wall mass transport equation with its boundary conditions
yields,
kc w - c 1 - exp ( V w S / ~ )
(5)
k(c w - ca) 1 - exp(VwSm/~ )
L A M I N A R MASS TRANSFER 131
But the mass flux through the fluid boundary layer must equal the mass
flux through the wall. Equating these fluxes gives,
The boundary condition at the tube wall for the mass transport equation
is given by solving equations (5) and (6) with the assumptions that VwSm/~
< l a n d t h a t c d--0.
~c r =R k~cw (k - 1)VwCw
= Sm~(1 + VwSm/2~;~) -} ~ (7)
The partition coefficient can have values between zero and unity. For
dilute concentrations of small molecules in reasonably porous materials,
the partition coefficient has been found by Green et.al. [8] to be essentially
unity. However, the complete boundary condition will be evaluated to
determine the sensitivity of the solutions to the partition coefficient.
The other boundary conditions necessary to solve the mass transport
equation are,
~c
- - O, r=O
~r
(8)
C = Ci, Z = 0
v+r( r2 )
Vz =
[ 2Qi
-7~R 2
4VwZ
R
;(re)
1-
~T
(9)
c*(p, O) = 1
~C*
- O, p = 0 (10)
Op
. . . . ~0c*, p = 1
8p
k Sh w (k - 1)Pe w
+
1 4- Pew/4Sh W 2
oo
where R. are the eigenfunctions, 2. are the eigenvalues and A. are the
orthogonal coefficients all of which are determined for various values of
wall Sherwood number and wall P6cl6t number.
The fluid bulk concentration is expressed in terms of the fluid bulk
coefficient E. as follows,
co
c* = 2 E . ( 1 - 2Pew~);@2PEW (12)
n=0
where,
s(d.. "ew
2 R
)
E.= [dR" 1 d2R. 1 ' p= 1 (13)
~"(~" + 2eew) L dL, 4 ~ d~.dp-J
Numerical integration of the equations was carried out for the first five
eigenvalues and fluid bulk coefficients over a range of wall Sherwood
numbers and wall P6cl6t numbers including 0.1 _< Sh w ~ o e and - 10
<_ Pe w _< 10 for various values of the partition coefficient between 0.1 and
1.0. Selected values are shown in Table 1.
LAMINAR MASS TRANSFER 133
TABLE 1
k=l.O
Sh w 2, E, 2, E, 2, E,
k=0.8
Sh w 2, E, ~, E~ 2, E,
Selected eigenvalues and fluid bulk coefficients for two values of membrane partition
coefficient
134 H.R. DAVIS
2.0 , , t , ; ; I ,
Sh w = 0<:)
1.8
--- M
M--;= e x p /~t - Q1o / Q i I
i.6
1.4
Id
Mo 1.2
~=
I.O
0.8
I I I I 0./.~05.~
0.6 t , . ,
0.6 0,8 1.0 1.2 1.4
Q._~o
Oi
Fig. 1. Ratio of mass transfer with wall convection to mass transfer without wall convection
as a function of extraction or injection ratio for various dimensionless tube lengths,
Sh w = oo.
LAMINAR MASS TRANSFER 135
2e. (14)
.:o \
Mass transfer from a porous tube is given by,
M _ 1- Qo c* (15)
Qicl Qi
Therefore, the ratio of mass transfer with wall convection to mass transfer
without wall convection is
oo ( Qo ~{l +[A2~/(1-QO/Qi)]}
M 1-,=oZ E , \ ~ - j
o) (16)
iV/ 1- 52 Ez exp ( - 22~)
#?1=0
where, for a specific wall Sherwood number, the fluid bulk coefficients and
eigenvalues E, and 2,, or E,, and 2~, are obtained at some specified wall
P6cl6t number or Pe w = 0 respectively. Typical results for this mass trans-
fer ratio are plotted in Fig. 1 and Fig. 2 for wall Sherwood numbers of
infinity and unity for a partition coefficient of unity. There is very little
difference in results for other values of the partition coefficient.
Mass transfer from a porous tube without wall convection can also be
expressed in terms of an overall mass transfer coefficient, ho, which is
obtained from the overall Sherwood number as shown in [5]. Thus it
follows that,
M - 1 - e x p - hoA (17)
2.0
1.8
1.6
1.4
M
Mo
1.2
1.0
1.0
3.5~
0.8
I
0.6
0.6 0.8 1.0 1.2 1.4
Qo
Qi
Fig. 2. Ratio of mass transfer with wall convection to mass transfer without wall convection
as a function of extraction or injection ratio for various dimensionless tube lengths, S h w = 1 . 0 .
oo
hA - ln[ 2 Emexp(-22)] (19)
Qi m=0
the porous tube mass transfer can be plotted, for various values of the
extraction ratio, as a function of the variable hoA/Q i. These results give the
effect of wall convection on a mass exchanger which has a predetermined
overall mass transfer Coefficient for no wall convection. Even though the
overall mass transfer coefficient is strongly affected by the wall Sherwood
number, the effect of the wall Sherwood number on porous tube mass
transfer is very small making a 3 ~ variation at most for S h , > 0.5. Thus
the results shown in Fig. 3 give a close approximation to porous tube mass
transfer.
LAMINAR MASS TRANSFER 137
, . . . .
0.8 0.2 _. -
0.5 - -_ _
o,
M 0.5
0.4
I i F r I I I I I I
hob
~'= Qi
Fig. 3. Mass transfer from porous tubes or plates as a function of the overall mass transfer
coefficient without wall convection for various extraction or injection ratios. Empirical
equation is also shown.
M / 1 - Qo/Q,
- e x p ~--2h-oA~i )
(20)
fitted the results predicted by (16) within 2 percent for all wall Sherwood
numbers for the limitations of 0.5 _ M / M o _< 2 and Qo/Qi < 2. This fit is
also shown in Fig. 1 and Fig. 2 for all values of the partition coefficient.
Thus (17) and (20) give an analytical approximation to the porous tube
mass transfer results of Fig. 3 which is valid within the range of limitations
given above. The fact that Fig. 3 contains the calculated mass transfer
results for a wide range of wall Sherwood number, wall Peclet numbers,
and modified Graetz numbers and that these results are nearly inde-
pendent of the wall Sherwood number indicates that mass transfer due to
wall convection is only weakly coupled to mass transfer by diffusion when
138 H.R. DAVIS
the wall Reynolds number is small. Therefore mass transfer from porous
tubes with wall resistance can be expressed simply in terms of mass transfer
from identical non-porous tubes.
5. Experiments
Some experiments were conducted to determine the effect of wall con-
vection on mass transfer from porous tubes. These experiments used
dialyzers of the tube-in-shell type in which small diameter hollow fibers
were fixed at their ends within a tubular chamber. The fibers were con-
nected to manifolds to complete the feed solution circuit while the tubular
chamber provided a fluid path around the exterior surfaces of the hollow
fibers for the dialyzing solution. The hollow fibers were fabricated by a
proprietary method to yield both high permeability to low molecular
weight solutes transferred by diffusion and high permeability to pressure
induced flow. These fibers had a nominal inside diameter of 0.02 cm with a
wall thickness of 0.003 cm and a 15 cm length. Each dialyzer contained 300
fibers and had an active membrane area of approximately 300 cm 2.
The procedure consisted of pumping a dilute saline solution (0.01
Molal) through the dialyzer while measuring the conductivity of the
outflow continuously to determine the ratio of the outlet-to-inlet saline
concentration. Calibrated syringe pumps on both the fluid outlet and inlet
controlled the volumetric flow rate through the dialyzer and allowed the
desired amount of wall convection to be established without reference to
the system fluid pressures. The range of inlet flow rates included 1 cm3/min
to 16 cm3/min while the outlet flow rates were set to achieve extraction
ratios of 0.4 _< Qo/Qi -< 1.0. Distilled water circulated through the tubular
dialyzer chamber with suffcient flow rate to give turbulent mixing over the
fibers. The entire system was maintained at 25C.
The experimental results are presented in Fig. 4 for three different
extraction ratios. The method used was to observe the outlet concentration
at one inlet flow rate without extraction, then set an extraction ratio and
observe the resulting change in outlet concentration. Equations (15) and
(17) were used to calculate the mass transfer rate and overall mass transfer
coefficient for each data point. Also presented in Fig. 4 are the theoretical
mass transfer curves for the three extraction ratios. The discrepancy in
these results can be attributed to two main effects.
The first effect is concerned with the membrane area available for mass
transfer by diffusion and mass transfer by wall convection. Pressure in-
duced flow through the hollow fiber wall will occur over the entire mem-
LAMINAR MASS TRANSFER 139
I.O
r~
0.8-. co~ []
QT.~i - o
0.4
0,2 o
brane area exposed to the feed solution whereas mass transfer by diffusion
only occurs with efficiency where the flushing solution circulating in the
dialyzer chamber reaches the exterior surfaces of the fibers. In the ap-
paratus used for these tests it was estimated that between 2 ~ and 5~o of the
active membrane area was not exposed to the flushing solution. Thus it can
be expected that the observed effect of the extraction ratio on mass transfer
will be greater than if all the active membrane area had been utilized for
diffusional mass transfer.
The second effect is concerned with the actual efficiency of the dialyzer
for mass transfer. The experimental dialyzers operated a* a wall Sherwood
number of approximately 0.2. The theoretical mass transfer curves pre-
sented in Fig. 3 are for S h w >_ 0.5. The effect of very small wall Sherwood
numbers is greatest when the dimensionless mass transfer coefficient is
y = 0.3 and for S h w = 0.2 amounts to approximately 5 ~ error. The theoreti-
cal mass transfer curve for each extraction ratio would then be very close to
the experimental results.
The accuracy of the empirical equation for mass transfer from porous
tubes can be appreciated from the data in Table 2 where the experimentally
observed mass transfer rates for various extraction ratios are given for
several values of the dimensionless mass transfer coefficient, 7. The calcu-
lated mass transfer from the empirical equation and exact theory, corrected
for wall Sherwood number, are also given for comparison.
140 H. R. DAVIS
TABLE 2
E Et Ee ? E Et Ee ? E Et Ee
1.552 0.841 0.840 0.841 1.552 0.887 0.894 0.897 1.552 0.934 0.943 0.956
1.006 0.715 0.716 0.701 0.996 0.787 0.788 0.771 0.922 0.853 0.862 0.834
0.808 0.644 0.644 0.627 0.808 0.734 0.737 0.710 0.808 0.837 0.836 0.804
0.526 0.524 0.512 0.495 0.573 0.655 0.648 0.618 0.559 0.767 0.766 0.732
0.452 0.476 0.472 0.453 0.452 0.597 0.595 0.566 0.452 0.730 0.732 0.706
0.282 0.367 0.363 0.350 0.289 0.508 0.509 0.501 0.289 0.665 0.670 (0.709)
0.168 0.291 0.283 0.281 0.168 0.444 0.445 (0.508) 0.144 0.623 0.621 (1.07)
Experimental data from porous tubular mass exchanger showing the fit of the exact and
approximate theoretical mass transfer efficiencies
6. Conclusions
T h e velocity profiles for fluid flow in p o r o u s tubes or fiat plates can be
stated in terms of the wall P6cl~t n u m b e r or e x t r a c t i o n ratio alone when the
wall R e y n o l d s n u m b e r is low. W i t h these velocity profiles an d the mass
t r a n s p o r t equations, mass transfer f r o m p o r o u s tubes or plates with wall
resistance was calculated for v a r i a t i o n s in wall c o n v e c t i o n including in-
j e c t i o n and suction. T h e results were expressed as a c o m p a r i s o n of mass
transfer with wall c o n v e c t i o n to mass transfer w i t h o u t wall convection.
W h e n the overall mass transfer coefficient was defined for n o wall
c o n v e c t i o n as a function of the wall S h e r w o o d n u m b e r an d modified
G r a e t z number, mass transfer f r o m p o r o u s tubes or plates was d e t e r m i n e d
in terms of this overall mass transfer coefficient for specified values of wall
convection. A result of this d e t e r m i n a t i o n was that mass transfer with wall
co n v ect i o n , when expressed as a function of the dimensionless mass trans-
fer coefficient, a p p e a r e d to be only weakly d e p e n d e n t on the wall Sher-
w o o d n u m b e r or on the m e m b r a n e p a r t i t i o n coefficient.
O v e r the ra n g e of interest in any practical mass exchanger, the mass
transfer p e r f o r m a n c e was also expressed in terms of a c o n v e n i e n t empirical
LAMINARMASSTRANSFER 141
relationship which involved only the overall mass transfer coefficient and
extraction ratio. This empirical equation is valid when the contributions to
mass transfer by diffusion and wall convection are of the same order of
magnitude. The results apply equally for tubular and flat plate mass
exchangers.
A series of experiments conducted on porous tubular mass exchangers
for several values of extraction ratio verified the theory and showed that
the effect of wall convection on mass transfer can be expressed as a function
of mass transfer without wall convection.
APPENDIX I
The analysis for mass transfer from porous flat plates follows essentially in
the manner stated for porous tubes. For flat plates which have fully
developed flow and constant concentration upstream of the porous regio n
and in which diffusion in the flow direction is negligible in comparison to
diffusion towards the plates and where the flow channel width is much
greater than its height, the mass transfer transport equation becomes,
~C ~C 02C
+ v ~ - - = ~ @~y2
-- (A-l)
b / ~x- oy
c(0, y) = c i
~C
-- =0, y=0 (A-2)
#y
Dc k~c (k - 1)VwC
- + , y=h
Dy @Sin(1 + VwSrn/2~ )
The fluid velocity profiles for flow between parallel flat plates are,
(A-3)
142 H.R. DAVIS
c*(~/, 0) = 1
~C*
-0, ~ =0
@
~C*
- - = - - (PC*, r/ = 1 (A-5)
@
q~ = k Sh w + (k - 1)Pe w
1 + Pew/4Sh w 2
0o
and where the fluid bulk concentration is expressed in terms of a fluid bulk
coefficient E. as follows,
0o
where
[.dr. Pew ]
3Ida/ 2 Y"
E.= ( 3Pew']FdY . 1 dzY. ] ' r/=l
2. 2.2 + ~---7 [-d~-.+ ~ d2.d~/J
The eigenvalues and fluid bulk coefficients were calculated for a wide range
of wall Sherwood numbers and wall P6cl6t numbers.
Mass transfer from porous plates without wall convection can be ex-
pressed in terms of an overall mass transfer coefficient, h 0, as follows.
LAMINAR MASS TRANSFER 143
oo
hA - ln[ Z E~ exp ( - ~ 24m 2~ ) ] (a-9)
Qi m=O
with the mass transfer from p o r o u s flat plates with wall c o n v e c t i o n at this
dimensionless length given by,
Thus, for various values of the extraction ratio, the p o r o u s flat plate mass
transfer can be plotted as a f u n c t i o n of the variable hoA/Q i. The results for
flat plate mass transfer are s h o w n in Fig. 3 a n d are n o t distinguishable from
the results for p o r o u s tube mass transfer.
REFERENCES