Ind. Eng. Chem. Res.

2001, 40, 885-891 885

Effects of Marangoni Convection on the Mass Transfer Performance

in a Packed-Bed Absorber
Honda Wu,† Tsair-Wang Chung,*,† and Ming-Hong Lai‡
Chemical Engineering Department, Chung-Yuan University, Chungli, 32023 Taiwan, Republic of China

The process of absorbing water vapor into triethylene glycol (TEG) solutions is achieved in two
steps. In the first step, water vapor is condensed into water droplets on the gas-liquid interface.
The second step involves the dissolution of water droplets into the TEG solution as a result of
the change in surface tension in the surface thin film. This study is focused on the effects of the
surface tension gradient on mass transfer performance of the absorption process. An investigation
of the disturbance induced by the surface tension gradient and an evaluation of the mass transfer
behavior were conducted. As noted, the phenomenon of fluid flow in the surface thin film due to
the surface tension gradient is termed the Marangoni effect, and the induced convection is usually
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

termed Marangoni convection. To discuss the relationship between the Marangoni effect and
the mass transfer performance, the surface tension of the liquid desiccant was measured under
different temperatures and concentrations during experimental runs. The conventional mass
Downloaded via UNIV DE COSTA RICA on April 3, 2020 at 23:16:26 (UTC).

transfer correlation was modified by using the term of the M index, which combines the surface
tension gradient and the driving force of mass transfer to explain Marangoni effects. This
improves the average error between predicted values and experimental data significantly.

Introduction the transfer of a component with lower surface tension

from the liquid to the gas phase results in an increase
Zarzycki and Chacuk1 found that surface flow was in the surface tension of the remaining liquid, which is
induced by the surface tension gradient in wetted-wall, termed the Marangoni positive, so that the packing
packed-bed, and plate columns. On the basis of their materials can be completely wetted by the liquid. On
descriptions, the Marangoni effect resulting from sur- the other hand, while the component evaporates with
face tension gradient was found to affect the mass relatively higher surface tension, the remaining liquid
transfer performance in those typical absorption col- can form spots, which is termed the Marangoni nega-
umns. As noted by them, one of the most significant tive, and as a consequence, the channeling phenomenon
physicochemical properties of the working solution can occur in the packed bed. Because the surface tension
related to the mass transfer mechanism could be the gradient increased the disturbance in the surface thin
surface tension. The conventional mass transfer theo- film, Smigelschi et al.4 and Ruckenstein et al.5 tried to
ries, such as film theory, penetration theory, and surface add the surfactant isobutanol to the air-water interface
renewal theory, did not consider the effects of the to enhance the efficiency of absorbing CO2 from air.
surface tension on the mass transfer performance. They also adopted the convection-diffusion equation to
Although the boundary layer theory incorporated the discuss the velocity contribution of the surface flow and
concept of a hydraulic boundary layer, it still did not showed that the derivation was consistent with experi-
clearly identify the effect of this physicochemical prop-
ment.
erty on the mass transfer performance. Discussions of
conventional gas-liquid contacting devices on the basis In addition, the local surface tension gradient result-
of their mass transfer performance incorporated with ing from the temperature difference of vaporizing the
Marangoni effects were limited. There are some inves- fluid layer at the gas-liquid interface is called an
tigators studying the Marangoni effect in distillation instability. Berg and Acrivos6 found that, when a
processes. However, similar studies on packed absorbers surfactant with lower surface tension was added to the
are rare so far. liquid surface, it would stabilize the fluid flow at the
Zuiderweg and Harmens2 demonstrated that the surface. The studies related to the Marangoni effect on
distillation efficiency of a bubble tower with a continu- fluid flow in surface thin films and on the distillation
ous liquid phase is higher than that of a spray tower column performance are shown in Table 1. A discussion
with a dispersive liquid phase at the same operating of the disturbance induced by the surface tension
conditions because the Marangoni effect occurred in the gradient on mass transfer performance in a packed-bed
bubble tower. Furthermore, Patberg et al.3 found that absorber is presented in this study. To make the liquid
flow in a continuous phase and avoid channeling effects
in the packed column, the liquid flow rates were
* To whom correspondence should be addressed. Dr. Tsair- operated beyond the minimum liquid flow rate. In
Wang Chung, Associate Professor, Department of Chemical general, the process variables of the packed-bed absorp-
Engineering, Chung-Yuan University, Chungli 32023, Taiwan,
ROC. Fax: 886-3-4563171 ext 4199. Phone: 886-3-4563171
tion systems include the gas and liquid flow rates, gas
ext 4125. E-mail: twchung@mbox.cycu.edu.tw. and liquid temperatures, concentrations of liquid desic-
† Chung-Yuan University. cant solutions, and type and size of the packings. The
‡ Present address: Taiwan Police College, Taipei, Taiwan, mass transfer performance of the packed-bed absorber
ROC. in this study is discussed in terms of the process
10.1021/ie000468a CCC: $20.00 © 2001 American Chemical Society
Published on Web 01/09/2001
886 Ind. Eng. Chem. Res., Vol. 40, No. 3, 2001

Table 1. Some Studies Related to the Marangoni Effect

in the Literature
discussion of the gas-liquid Smigolischi et al.4
interfacial phenomenon during Lu et al.13
the absorption process Lu et al.14
Vazquez et al.18
Wu et al.19
discussion of the Marangoni Zuiderweg and Harmens2
effect on distillation column Patberg et al.3
performance Zuiderweg16
Billet and Schultes20
Proctor et al.21 Figure 2. Water vapor condenses into water droplets.
comparison of distillers with Zuiderweg and Harmens2
and without the Marangoni Patberg et al.3
effect Moens and Bos22 In this absorption system, liquid desiccant solutions flow
Semkov and Kolev23 into the absorber through nozzles and leave from the
bottom of the absorber. Three full-cone nozzles were
Table 2. Some Studies of Mass Transfer Performance in used as the liquid distributor, and 2-in. polypropylene
the Literature
Flexi rings were packed in the absorber. The solution
discussion of the column Chung et al.10 flow rates were determined by calculations of the
or removal efficiency Ullah et al.24 minimum wetting rate.7 Because the solution flow rates
Chung et al.25
Sadasivan and Balakrishnan26
selected in this study are much larger than the mini-
discussion of the mass Gandhidasan et al.9 mum wetting rate and because the mass transfer is
transfer coefficient Chung et al.10 occurred in the gas-liquid interface, whose area is more
Chung et al.25 than 3 m2 in this packed tower, end effects are negli-
Sadasivan and Balakrishnan26 gible. The air inlet and outlet are on both sides of the
derivation and Gandhidasan et al.9 absorber. The setup of the stripper is similar to that of
regression of the mass Chung et al.10
transfer correlation Aroonwilas and Tontiwachwuthikul11 the absorber. This system can vary the air flow rates
Onda et al.17 from 0.047 to 0.063 kg/s and the liquid flow rates from
Nielsen et al.27 0.17 to 0.26 kg/s. The surface tensions of the TEG
solutions were measured by a surface tension meter
(CBVP A-3). The mass transfer coefficients were calcu-
lated from the absorption of water vapor by the TEG
solutions. The concentrations of the TEG solutions were
measured by a refractometer. A testo-400 hygrometer
with two humidity probes was used to measure the
absolute and relative humidity and temperature of the
air. The air flow rates were controlled by transistor
inverters on the 1 HP blowers. The liquid flow rates
were controlled by rotameters and global valves, and
the air velocities were measured with a hot-wire flow
meter.
Discussions of the relationship between the mass
transfer performance and the disturbance resulting
from the surface tension gradient are the purpose of this
Figure 1. Liquid desiccant dehumidifier of this study. study. A continuous and uniform fluid on the packing
surface was maintained, and channeling effects were
variables mentioned above. Discussions of mass transfer avoided. When the surface tension gradient resulting
performance are usually categorized into three portions, from absorbing water vapor is large enough to exhibit
including the column efficiency, the mass transfer the Marangoni effect, the disturbance in the surface thin
coefficient, and the derivation and regression of the film takes place. Furthermore, the surface tension of
mass transfer correlation (shown in Table 2). Discus- the liquid desiccant solution varies with the concentra-
sions of the Marangoni effect in packed-bed absorption tion and temperature of the solution. Therefore, the
systems are limited in the open literature. However, experiments in this study were conducted by means of
these discussions help us understand the mechanisms variations in the air and liquid flow rates as well as the
of water molecules dissolved in the liquid desiccant concentrations and temperatures of the TEG solution.
solution and more information can be acquired, such as
the mass transfer mechanisms of the packed-bed ab- Formula Derivation
sorption system, improvement of the column efficiency,
and reduction of the discrepancy in the correlation. In the process of the absorption of water vapor by the
TEG solutions, water vapor is condensed to water
Experimental Section droplets on the liquid surface. Because the density of
the TEG solution is larger than that of water droplets,
Experiments were conducted in an absorption-strip- the shape of the water droplets becomes a circular sheet
ping system. The configuration and the structure of the over the solution surface, as shown in Figure 2. Because
absorber are shown in Figure 1. The design of an inverse of the surface tension gradient induced by the leading
U-shaped air tunnel with an eliminator in the absorber edge of the circular sheet and the TEG solution, the
allow for countercurrent contact between the air and disturbance of a surface thin film causes a nonuniform
the solution, which reduces carryover of the solution. surface stress. The larger the surface tension gradient,
 Ind. Eng. Chem. Res., Vol. 40, No. 3, 2001 887

L yAin - yAi
kxa ) ln (7)
V yAout - yAi

From Dalton’s partial pressure law

PyA ) PA (8)

the mole fraction yA in eq 7 can be represented by the

ratio of the partial pressure and the total pressure, i.e.

PAin PAi
-
L P P
kxa ) ln (9)
Figure 3. (a) Continuous fluid flow in the packed-bed column V PAout PAi
and (b) fluid flow on a packing surface (scale-up). -
P P
the more intensive the disturbance of the thin film that
Because water molecules are very soluble in the TEG
occurs.
solution, the process of dehumidification is regarded as
A schematic diagram of a packed-bed absorber is
a gas-phase-resistance-controlling system, i.e.
shown in Figure 3a. The continuous liquid film on the
packing surface, as shown in Figure 3b, is assumed to 1 1
be a stagnant film. The mass balance can be written as . (10)
ky kx
NA dA ) kx(xAi - xA)a dV ) L dxA ) G dyA (1)
and
where kx is the mass transfer coefficient of diffusing yA - yAi . xAi - xA (11)
water vapor into the stagnant film of TEG solution on
the packing surface. Compared to the mass transfer In a gas-liquid contact system, the concentration profile
coefficient, kx′, in the different case of equimolar coun- is represented as in Figure 4. Because water molecules
terdiffusion, kx is equal to kx′ divided by a (1 - xA)lm are very soluble in the TEG solution, once the water
term corresponding to a bulk flow concentration factor vapor touches the TEG solution surface, the water vapor
for transfer through a stagnant film (Geankoplis,8 1993). is changed to the liquid phase right away. Therefore,
For dilute systems, the log-mean term approaches 1, the inequality PA/P . PAi/P holds, and PAi approaches
and kx is similar to kx′. L and G are the molar flow rates 0 at the interface. Thus, eq 9 is simplified to the
of liquid and gas, respectively. By considering mass following formula:
transfer in the liquid phase, eq 1 becomes
PAin
kxa(xAi - xA) dV ) L dxA (2)
L P
kxa ) ln (12)
where kxa represents the volumetric mass transfer V PAout
coefficient. Therefore, eq 2 becomes P
kxa(xAi - xA) dV ) L dxA (3) From Chapter 9 of the book by Geankoplis,8 the partial
pressure of water vapor in air can be represented by
Integration of eq 3 with respect to the packed-bed the humidity of the air.
volume and the concentration of water molecules yields
PA 28.97 × H
) (13)
L xAout dxA P 18.02 + 28.97H
kxa ) ∫
V xAin xAi - xA
(4)
The term PA/P in eq 12 is replaced by eq 13. Equation
12 then becomes
The concentration of water molecule at the gas-liquid
interface can be assumed to be a constant in the 28.97Hin
equilibrium state during absorption process. Let mole
fraction of xAi be a constant. Equation 4 is integrated L 18.02 + 28.97Hin
kxa ) ln (14)
and rewritten as V 28.97Hout
xAi - xAin 18.02 + 28.97Hout
L
kxa ) ln (5)
V xAi - xAout The above equation was used to estimate the mass
transfer coefficients in this study.
As noted, the molar fluxes in the gas and liquid films
are the same and ky(yA - yAi) is equal to kx(xAi - xA).
Therefore Results and Discussions

ky
xAi - xA ) (y - yAi) (6) 1. Discussion of kx and kxa. Because the mass
kx A transfer performance of the absorber is affected by the
gas-liquid contacting area, it is reasonable to use the
Substituting eq 6 into eq 5, we find values of kxa9-11 to assess the performance of the
888 Ind. Eng. Chem. Res., Vol. 40, No. 3, 2001

Figure 4. Gas- and liquid-phase concentration gradients at the

gas-liquid interface.

Figure 6. Effect of air flow rate on liquid-phase mass transfer

coefficient.

Figure 5. Effect of liquid flow rate on liquid-phase mass transfer

coefficient.

absorber. However, to calculate the Marangoni num-

ber,12 Ma ) σi - σB/(µLk*x), estimations of k12-15 are
also necessary. The M index is a popular index used to
describe the Marangoni effect. Not only the rate change
of surface tension, but also the driving force for mass
transfer, is taken into account through adoption of the
M index. Therefore, the M index was used to discuss
the effects of surface tension on the mass transfer Figure 7. Effect of air inlet humidity on liquid-phase mass
performance in this study. transfer coefficient.

2. kxa Influenced by Process Variables. On the of the disturbance induced by the surface tension
basis of the disturbance resulting from the surface gradient, this result can be clearly explained. The
tension gradient, the different temperatures and con- definition of each local position of water vapor absorbed
centrations of the liquid were controlled and the gas and by TEG solution is termed an absorption site in this
liquid flow rates were fixed at the same operating study. The number of absorption sites increases as the
condition for studying the effect of the disturbance level air humidity increases, thereby increasing the distur-
on the mass transfer coefficients. However, the conven- bance area. Therefore, the value of kxa is increased by
tional interpretations were usually used to explain the increasing the air humidity.
effect of the gas and liquid flow rates on kxa. In addition, studies of the effects on the liquid-phase
As shown in Figure 5, when the air flow rate is kept mass transfer coefficient of variations in the tempera-
constant, the value of kxa increases as the liquid flow ture and concentration of the liquid desiccant solution
rate increases. Because the amount of treated air is were also conducted by other researchers. Because the
fixed and the amount of absorbent (TEG) is increased, surface tension of the liquid desiccant solution is a
the mass transfer coefficients should increase. However, function of the temperature and concentration, the
Figure 6 shows that the mass transfer coefficient parameters of temperature and concentration of liquid
decreases with an air flow rate increase when the liquid desiccant solution were replaced by surface tension to
flow rate is kept constant. Similarly, when the amount discuss the mass transfer coefficient for the liquid phase.
of treated air is increased and the amount of absorbent Not only the relationship between the mass transfer
(TEG) is fixed, the value of the mass transfer coefficient performance and the surface tension gradient but also
should decrease. the changes in the surface tension of the TEG solution
When the air and liquid flow rates are kept constant, due to different temperatures and concentrations were
the mass transfer performance in Figure 7 increases as acquired in this study. As shown in Table 3, the surface
the inlet humidity increases. On the basis of the concept tension of the triethylene glycol solutions increases as
 Ind. Eng. Chem. Res., Vol. 40, No. 3, 2001 889

Table 3. Surface Tension of Triethylene Glycol Solution

at Different Concentrations and Temperatures
surface
TEG conc TEG temp tension
(wt %) (°C) (dyne/cm)
99.96 35 44.5
99.96 30 44.7
99.96 25 45.0
99.96 20 45.4
97.60 35 44.8
97.60 30 45.1
97.60 25 45.3
97.60 20 45.7
97.60 15 46.1
94.50 35 45.3
94.50 30 45.5
94.50 25 45.8
94.50 20 46.2
94.50 15 46.6
93.00 35 45.5
93.00 30 45.7
93.00 25 45.9 Figure 8. Relationship between concentration of desiccant solu-
93.00 20 46.3 tion and the ratio of the surface tensions of the TEG solution and
93.00 15 46.8 water.
90.00 35 45.9
90.00 30 46.1
90.00 25 46.4
90.00 20 46.8
90.00 15 47.2
88.00 35 46.2
88.00 30 46.5
88.00 25 46.8
88.00 20 47.1
88.00 15 47.5

the temperature decreases and decreases as the con-

centration increases. The values of the M index16 in this
study of the Marangoni effect are calculated via

dσ′ dσ′
M ) ∆x ) (x - x*) (15)
dx dx

where ∆x represents the driving force for mass transfer

and dσ′/dx is the change in the dimensionless surface
tension with mole fraction. The dimensionless surface
tension is defined as
Figure 9. Comparison between experimental and predicted mass
transfer coefficients using the correlation without the M index
σ′ ) σsoln/σH2O (16) term.

The above variables are arranged into pertinent dimen-

As shown in Figure 8, the change in the surface tension
sionless groups by using the Buckinghan pi method. The
increased with increasing mole fraction of water. Be-
mass transfer correlation obtained from the dimensional
cause the M index is the combination of the surface
analysis is given below as
tension gradient and the driving force, the relationship

( )( ) ( )
between these two variables is not linear. The larger
kxaMdp2 FLLdp c µL d
L e
the value of the M index, the more intensive the surface )b (18)
disturbance. Therefore, it can be seen from Figure 10 µL µL DL F L G
that kxa increases with increasing M index.
3. Mass Transfer Correlation Incorporated with where b, c, d, and e are constants that are obtained by
Surface Tension Difference. The mass transfer coef- a nonlinear regression of the experimental data. The
ficients calculated from eq 14 were correlated in terms experimental data of this study were used to evaluate
of the process variables by employing a dimensional the constants in eq 16. The correlation was obtained as
analysis. Variables that affect the mass transfer coef-
kxaMdp2 FLLdp
( )( )()
ficient for the liquid phase include the air and liquid 0.5 µL 1/3 0.81
L
flow rates, the physical properties of the air and liquid, ) 1 × 10-7 (19)
µL µL DLFL G
the packing volume and size, and the diffusion coef-
ficient of water in the TEG solution. In functional form, The above correlation was applied to predict the mass
the mass transfer coefficient can be expressed as transfer coefficients under different temperatures and
concentrations in the experimental runs. The average
f(kxa, DL, dp, dc, FL, G, L, µL, M) ) 0 (17) error was 15%, as shown in Figure 9. As mentioned
890 Ind. Eng. Chem. Res., Vol. 40, No. 3, 2001

enhances the mass transfer coefficient but also practi-

cally affects the mass transfer mechanism.

Conclusion
Generally speaking, discussions of the effects of
process variables on absorption processes sometimes
neglect the physical properties of the working solutions,
such as their surface tension. However, the surface
tension affects the mass transfer mechanism signifi-
cantly. The surface traction resulting from the surface
tension gradient of absorbing water vapor into a TEG
solution is one of the most important reasons causing
the disturbance phenomenon in the surface thin film.
The concept of a surface fluid disturbance induced by
the surface tension gradient was suggested in this
paper. Through a consideration of the disturbance
occurring in surface thin film, the mass transfer per-
formance in the air dehumidification process can be
explained successfully. The mass transfer correlation
Figure 10. Effect of the M index on the liquid-phase mass obtained by using dimensional analysis, which involves
transfer coefficient. the ratio of the liquid and gas flow rates and the M
index, reduces the average error between the predicted
values and the experimental data to about 7%. Fur-
thermore, to understand the mass transfer mechanism
exactly, additional studies related to surface flow will
be conducted in the future.

Acknowledgment
This work is supported by the National Science
Council of the Republic of China under Grant NSC88-
2214-E-033-004.

Nomenclature
a ) surface area of liquid film per unit volume of packed
bed, m2/m3
D ) diffusion coefficient, m2/s
dc ) column diameter, m
dp ) packing diameter, m
G ) gas flow rate, kmol/s
H ) humidity, kg of H2O/kg of dry air
kx ) mass transfer coefficient in the case of diffusion to a
Figure 11. Comparison between experimental and predicted
stagnant film, kmol m-2 s-1 (mole fraction)
mass transfer coefficients using the correlation with the M index
term. kx′ ) mass transfer coefficient in the case of equimolar
counterdiffusion, kmol m-2 s-1 (mole fraction)
earlier, the M index is a key variable affecting the mass ky ) mass transfer coefficient in the case of equimolar
transfer coefficient, as shown in Figure 10. The M index counterdiffusion for the gas phase, kmol m-2 s-1 (mole
should be taken into account in the mass transfer fraction)
correlation. The mass transfer correlation obtained by L ) liquid flow rate, kmol/s
M ) molecular weight of air, kg/kmol
using the dimensional analysis and nonlinear regression
P ) total pressure, mmHg
is given as
PA ) partial pressure of H2O, mmHg

( )( )()
T ) liquid temperature, °C
kxaMdp2 FLLdp 0.5 µL 1/3
L 1.55
V ) packing volume, m3
) 1 × 10-7 M0.25 (1 - xA)lm ) bulk flow concentration factor, [(1 - xA) -
µL µL DLFL G
(1 - xAi)]/ln[(1 - xA)/(1 - xAi)]
(20) xA ) mole fraction of H2O in TEG solution, mol/mol
xA* ) xA in equilibrium with yA
This mass transfer correlation is different from that of yA ) mole fraction of H2O in gas phase, mol/mol
Onda et al.17 The ratios of liquid to air flow rate and M µL ) liquid viscosity, kg m/s
index, which involve physical properties of the liquid FL ) liquid density, kg/m3
desiccant solution, were considered in this correlation σ′ ) dimensionless surface tension, σsoln/σH2O
to reduce the average error. The values of the mass σsoln ) surface tension of solution, N/m
transfer coefficient calculated from eq 14 and predicted σH2O ) surface tension of water molecules, N/m
from eq 18 were plotted in Figure 11. About 90% of the
data points are within the (15% error, and the average Subscripts
error was 7%. The results demonstrate that the distur- A ) H2O component
bance induced by the surface tension gradient not only x ) liquid phase
 Ind. Eng. Chem. Res., Vol. 40, No. 3, 2001 891

y ) gas phase (15) Fujinawa, K.; Hozawa, M.; Imaishi, N. Effects of desorption
L ) liquid and absorption of surface tension-lowering solutes on liquid-phase
G ) gas mass transfer coefficients at a turbulent gas-liquid interface. J.
i ) interface Chem. Eng. Jpn. 1978, 11 (2), 107-111.
B ) bulk solution (16) Zuiderweg, F. J. Marangoni effect in distillation of alcohol-
water mixtures. Chem. Eng. Res. Des. 1983, 61, 388-390.
(17) Onda, K.; Takeuchi H.; Okumoto, Y. Mass transfer coef-
Literature Cited ficients between gas and liquid phases in packed columns. J.
Chem. Eng. Jpn. 1968, 1 (1), 56-62.
(1) Zarzycki, R.; Chauck, A. Absorption Fundamentals &
Applications; Pergamon Press: New York, 1993. (18) Vazquez, G.; Antorrena, G.; Navaza, J. M.; Santos, V.
(2) Zuiderweg, F. J.; Harmens, A. The influence of surface Absorption of CO2 by water and surfactant solutions in the
phenomena on the performance of distillation columns. Chem. Eng. presence of induced Marangoni effect. Int. Chem. Eng. 1994, 34
Sci. 1958, 9, 89-108. (2), 247-254.
(3) Patberg, W. B.; Koers, A.; Steenge, W. D. E.; Drinkeburg, (19) Wu, T. C.; Lu, H. H.; Yang, Y. M.; Maa, J. R. Absorption
A. A. H. Effectiveness of mass transfer in a packed distillation Enhancement by the Marangoni effect - pool absorption of steam
column in relation to surface tension gradients. Chem. Eng. Sci. by aqueous lithium bromide solutions. J. Chin. Inst. Chem. Eng.
1983, 38 (6), 917-923. 1994, 25 (5).
(4) Smigelschi, O.; Suciu, D. G.; Ruckenstein, E. Absorption (20) Billet, R.; Schultes, M. Predicted mass transfer in packed
under the action of artificially provoked Marangoni effect. Chem. columns. Chem. Eng. Technol. 1993, 16, 1-9.
Eng. Sci. 1969, 24, 1227-1234. (21) Proctor, S. J.; Biddulph, M. W.; Krishnamurthy, K. R.
(5) Ruckenstein, E.; Smigelschi, O.; Suciu, D. G. A steady Effect of Marangoni surface tension forces on modern distillation
dissolving drop method for studying the pure Marangoni effect. packings. AIChE J. 1998, 44 (4), 831-835.
Chem. Eng. Sci. 1970, 25, 1249-1254. (22) Moens, F. P.; Bos, R. G. Surface renewal effects in
(6) Berg, J. C.; Acrivos, A. The effect of surface active agents distillation. Chem. Eng. Sci. 1972, 27, 403-408.
on convection cells induced by surface tension. Chem. Eng. Sci. (23) Semkov, KR.; Kolev, N. On the evaluation of the interfacial
1965, 20, 737-745. turbulence (the Marangoni effect) in gas (vapor)-liquid mass
(7) Sawistowski, H.; Smith, W. Mass Transfer Process Calcula- transfer: Part I. A method for estimating the interfacial turbulence
tion; Imperial College of Science and Technology: London, 1963. effect. Chem. Eng. Process. 1991, 29, 77-82.
(8) Geankoplis, C. J. Transport Processes and Unit Operations; (24) Ullah, M. R.; Kettleborough, C. F.; Gandhidasan, P.;
Prentice Hall: Englewood Cliffs, NJ, 1993. Effectiveness of moisture removal for an adiabatic counterflow
(9) Gandhidasan P.; Kettleborough, C. F.; Ullah, M. R. Calcula- packed tower absorber operating with CaCl2-air contact system.
tion of heat and mass transfer coefficients in a packed tower Trans. ASME 1988, 110, 98-101.
operating with a desiccant-air contact system. J. Sol. Energy Eng.
(25) Chung, T. W.; Ghosh, T. K.; Hines, A. L. Dehumidification
1986, 108, 123-128.
of moist air with simultaneous removal of selected indoor pollut-
(10) Chung, T. W.; Ghosh, T. K.; Hines, A. L. Comparison
ants by triethylene glycol solutions in a packed-bed absorber. Sep.
between random and structured packings for dehumidification of
Sci.Technol. 1995, 30, 1807-1832.
air by lithium chloride solutions in a packed column and their heat
and mass transfer correlations. Ind. Eng. Chem. Res. 1996, 35, (26) Sadasivan, M.; Balakrishnan, A. R. Experimental inves-
192-198. tigations on the thermal effects in packed-bed liquid desiccant
(11) Aroonwilas, A.; Tontiwachwuthikul, P. Mass transfer dehumidifiers. Ind. Eng. Chem. Res. 1994, 33, 1636-1640.
coefficients and correlation for CO2 absorption into 2-amino-2- (27) Nielsen, C. H. E.; Kiil, S.; Thomsen, H. W.; Dam-Johansen,
methyl-1-propanol (AMP) using structured packing. Ind. Eng. K. Mass transfer in wetted-wall columns: correlations at high
Chem. Res. 1998, 37, 569-575. Reynolds numbers. Chem. Eng. Sci. 1998, 53 (3), 495-503.
(12) Brian, P. L. T.; Vivian, J. E.; Mayr, S. T.; Cellular (28) Hines, A. L.; Maddox, R. N. Mass Transfer Fundamentals
convection in desorbing surface tension-lowering solutes from and Applications; Prentice Hall: Englewood Cliffs, NJ, 1985;
water. Ind. Eng. Chem. Fundam. 1971, 10 (1), 75. Chapter 5.
(13) Lu, H. H.; Yang, Y. M.; Maa, J. R. Effect of artificially
provoked Marangoni convection at a gas/liquid interface on Received for review May 5, 2000
absorption. Ind. Eng. Chem. Res. 1996, 35, 1921-1928. Revised manuscript received October 18, 2000
(14) Lu, H. H.; Yang, Y. M.; Maa, J. R. On the induction Accepted November 8, 2000
criterion of the Marangoni convection at the gas/liquid interface.
Ind. Eng. Chem. Res. 1997, 36, 474-482. IE000468A