You are on page 1of 9

Ind. Eng. Chem. Res.

1996, 35, 4185-4193 4185

Dimensionless Mass-Transfer Correlations for Packed-Bed


Liquid-Desiccant Contactors
Shailesh V. Potnis† and Terry G. Lenz*
Chemical and Bioresource Engineering, Colorado State University, Fort Collins, Colorado 80523

Random and structured packings were studied with varying bed depths in the regenerator and
the dehumidifier of a solar-assisted liquid-desiccant system. The slopes of the log-log plots of
mass-transfer rate vs solution flow rate were found to be close to 0.8, which indicated that the
conditions for the liquid phase were turbulent for the operating conditions in both contactors.
The small intercepts obtained for the Wilson plots indicated that the gas-phase mass-transfer
resistance was negligible compared to the liquid-phase mass-transfer resistance. Liquid-phase
mass-transfer coefficients for the packed bed alone were obtained by separating the contributions
of the other mass-transfer regions in the contactors. The random packing mass transfer
coefficients varied from 0.48 to 2 mol/(s m2), while the double-layer, structured packing mass-
transfer coefficients varied from 0.018 to 0.035 mol/(s m2). These mass-transfer coefficients were
converted into a dimensionless form, utilizing experimentally obtained diffusivity values.

Introduction water. This indicates that mass transfer in air-solution


contacting is not the same as in air-water contacting.
Desiccant systems utilize the moisture affinity of a Gandhidasan et al. calculated (1986) the coefficients
desiccant material to control the humidity of the air and analyzed (1987) the processes of heat and mass
used for a process or for comfort purposes. Commonly transfer in a calcium chloride-based liquid-desiccant
used desiccants include aqueous lithium chloride or system. They assumed the heat-transfer resistance of
lithium bromide solution and zeolites. The moisture the liquid phase to be negligible and found that the
content of these materials approaches saturation as they desiccant-air system has a mass-transfer resistance in
remove moisture from the incoming air. They can then the liquid phase unlike that for the air-water system.
be regenerated to restore their capacity to remove Thus, it is necessary to calculate the liquid-phase mass-
moisture by heating them in the presence of relatively transfer resistance to obtain the overall mass-transfer
dry air. resistance, which ultimately determines the air and
Desiccant systems are particularly useful for air solution flow rates, auxiliary power input, contactor size,
conditioning applications where the latent heat load is solar collector area, and performance of the system. A
large compared to the sensible heat load. Energy better understanding of the process of mass transfer in
savings, relative-to-conventional vapor compression sys- these contactors can thus significantly improve the
tems, of up to 40% can be achieved by using a desiccant- contactor efficiency and suggest ways to reduce the
assisted air conditioning system (Griffiths, 1987). Also, initial and operating costs.
the regeneration temperatures required by desiccants, The present research probes the variation of mass-
such as lithium bromide solution, usually vary between transfer rate with solution flow rate to obtain the
60 and 75 °C, which can easily be obtained using flat relative magnitudes of mass-transfer resistances offered
plate solar collectors. Thus, fossil fuel energy savings by each phase. Packed-bed liquid-phase mass-transfer
can be increased further if solar energy is utilized for coefficients were obtained from the contactor liquid-
the regeneration of the desiccant. phase mass-transfer coefficients by separating contribu-
The major process components of interest regarding tions of other mass-transfer regions. These mass-
mass transfer for such a system are the regenerator and transfer coefficients were then converted into a dimen-
the dehumidifier. The mass-transfer rates achieved in sionless form using the experimentally measured dif-
these units determine the capacity of these systems. fusivity values.
However, the process of mass transfer in these contac-
tors is not well understood, and as a result, the
Experimental Apparatus
measured performance of many absorption chillers and
heat pumps, using similar components, is smaller than Figure 1 shows a schematic diagram of the experi-
expected (Wahlig, 1993). These contactors are usually mental system employed in these studies. The dehu-
designed assuming air-solution contact to be the same midifier is an 81-cm-diameter, 200-cm high fiberglass
as air-water contact (Khan and Ball, 1992). Löf et al. tower. The solution entering the dehumidifier is dis-
(1984) studied reconcentration of lithium chloride solu- tributed on top of a packed bed using a manifold of three
tion in an open-cycle absorption chiller by passing solar- spray nozzles. An electric steam humidifier permits
heated air through a packed column. They found that variable moisture addition to the air before it enters the
mass-transfer coefficients show considerable variability dehumidifier. A mist eliminator is used to minimize the
and generally lower values than would be predicted by liquid entrainment before the air exits the dehumidifier.
use of the heat/mass-transfer ratio for air and pure In a decoupled mode, the solution exiting the dehumidi-
fier is cooled in a cooling unit before it is sent back to
* Corresponding author. Phone: (970) 491-2540. Fax: (970) the top of the dehumidifier. In coupled mode, this
491-7369. solution is sent to the regenerator through the econo-
† Present address: Energy Concepts Co., 627 Ridgely Ave., mizer where it is heated by heat exchange with the hot
Annapolis, MD 21401. solution from the regenerator. The regenerator is
S0888-5885(96)00212-6 CCC: $12.00 © 1996 American Chemical Society
4186 Ind. Eng. Chem. Res., Vol. 35, No. 11, 1996

Figure 1. Schematic diagram of the solar-assisted liquid-desic-


cant system.
Figure 2. Comparison of mass-transfer rates obtained using
different packings in the dehumidifier.
similar in dimension and construction to the dehumidi-
fier unit. Hot air to the regenerator is provided by a
flat-plate solar collector array having a total area of 55.7
m2. Auxiliary heat is also supplied by three electric
strip heaters to increase the air temperature and
maintain a high regeneration rate under varying ambi-
ent conditions.
The hot solution exiting the regenerator is cooled in
the economizer by exchanging heat with the solution
from the dehumidifier. After passing through the
economizer, the warm solution is further cooled by water
in a four-pass shell-and-tube heat exchanger. A cooling
tower with a capacity of 28 kW supplies water to the
heat exchanger.

Mass-Transfer-Rate Studies
Initially, energy balance studies were carried out to
identify the reliably measured parameters and to mini- Figure 3. Comparison of mass-transfer rates obtained using
mize the errors. The experimental system employed different packings in the regenerator.
provided energy balance data having a maximum error
of 10% in the decoupled mode (Lenz and Potnis, 1991). bed. However, for the random packed bed, the evapora-
Precautions suggested by these studies were observed tion rates in the regenerator were found to be ap-
during subsequent experiments conducted to obtain the proximately 300% and 130% greater and the conden-
relative magnitudes of mass-transfer resistances. sation rates in the dehumidifier were found to be
The mass-transfer rate in the packed-bed contactors approximately 60% and 45% greater than the packed
(the dehumidifier and the regenerator) depends on beds of structured packing with heights 30 and 55 cm,
various parameters such as air flow rate, solution respectively. Considering the experimental error, there
concentration, height of the packed bed, air and solution was no significant difference between the condensation
inlet temperatures, and solution flow rate. All these rates offered by the structured packed beds of heights
parameters except the solution flow rate were main- 55 and 30 cm. Therefore, there was no advantage
tained constant (with a solution concentration of 51% associated with the additional height of structured
by weight) in a decoupled mode, for each contactor, to packing observed for the dehumidifier.
study the dependency of the mass-transfer rate on Figures 4 and 5 show Wilson plots constructed using
solution flow rate. Packed beds of random packing the mass-transfer-rate data. The small magnitudes of
(Polypropylene Tripack), with height 30 cm, and struc- the intercepts obtained for these plots indicate that the
tured packing (Munters CELDEK), with heights 30 and gas-phase mass-transfer resistance is negligible com-
55 cm, were used to study the variation of mass-transfer pared to the liquid-phase mass-transfer resistance for
rates with types and heights of packings. The mass- these contact operations. Therefore, a dimensionless
transfer rate between the two phases was obtained from mass-transfer correlation to estimate the liquid-phase
the rate of enthalpy change associated with the phase mass-transfer coefficient would be a very useful tool for
change of water. the design of these systems. However, to convert the
A log-log plot of Sherwood number vs Reynolds data obtained from the mass transfer experiments into
number is generally employed to correlate various flow dimensionless form, it is necessary to know the diffu-
parameters. However, since all parameters except the sivity of water in lithium bromide solution.
solution flow rate were kept constant, a plot of mass-
transfer rate vs solution flow rate was constructed to Diffusivity Studies
represent the conventional plot. Figures 2 and 3 show
that these plots are straight lines with slopes near 0.8. A diffusivity apparatus was developed by suitably
This indicates that the dehumidifier and the regenerator combining the twin-bulb method and the diaphragm cell
were operated under well-mixed, near-turbulent condi- method (Potnis et al., 1995). The diffusivity of water
tions for liquid flow for all the variations of the packed in aqueous lithium bromide solutions was measured,
Ind. Eng. Chem. Res., Vol. 35, No. 11, 1996 4187
contactor is small (0.2%),

Nw ) Kxa(x - x*av)SZ (3)

The values of inlet humidities were obtained from a


knowledge of dew points. The values of the outlet
humidities were obtained using the mass transfer rates.
The values of x*inlet and x*outlet were obtained from these
humidities using an equilibrium chart for aqueous
lithium bromide solutions. The volumetric mass-
transfer coefficients (Kxa) were then evaluated using the
mass-transfer rates (Nw) obtained earlier.
Processes Influencing the Overall Mass Trans-
fer. The process of mass transfer in the contactors
occurs in three separate regions: a first region of small
droplets created by the spray nozzles, a second region
of thin films of solution flowing on the packing surface,
and a third region of the film flowing down the wall of
the contactor. To obtain the mass-transfer rate due to
Figure 4. Wilson plots for the dehumidifier. the packing alone, it was necessary to separate the
mass-transfer contributions due to the other two re-
gions.
In order to estimate the magnitude of the mass-
transfer contribution for the wall film, a thin plastic ring
was mounted at the bottom of the regenerator, just
above the pan. A tight fit was provided so that the ring
collected all the liquid flowing down the wall. The
solution flow rate on the wall was determined by
measuring the liquid collected by the ring in a known
time duration, using an external valve connected to the
ring. To allow for a maximum amount of liquid to flow
toward the wall, the measurements were done in the
absence of any packing. It was found that the solution
flow rates on the wall were only about 10% of the total
solution flow rates. This value was close to the other
experimental errors involved. Also, in the presence of
packing, the amount of solution flowing on the wall
would be further reduced due to the repeated redistri-
Figure 5. Wilson plots for the regenerator. bution of liquid over the packing surface at the contact
points of the wall and the packing. Therefore, the mass-
using tritiated water as the tracer, for concentrations
transfer contributions due to the films flowing down the
ranging from 0.5 to 11 M. The value of diffusivity was
walls of the contactors were neglected.
found to increase from 13.2 × 10-10 to 16.7 × 10-10 m2/s
To separate the contribution of the region of small
for the concentration variation of 0.5 to 4 M and then
droplets created by the spray nozzles, mass-transfer
to decrease to a steady value of approximately 6.5 ×
studies were conducted using only spray nozzles in both
10-10 m2/s from 8 to 11 M.
of the contactors (no packing). The experimental and
analytical procedure followed for these studies was
Packed-Bed Mass-Transfer Coefficients similar to the procedure followed for the mass transfer
Overall mass-transfer coefficients for the contactors studies with packings, described earlier.
were calculated using the data obtained from the mass- Packing Kxa. Since the mass-transfer contribution
transfer experiments. These mass-transfer coefficients of the solution film on the wall is negligible, the total
are composite coefficients for the packing, the spray mass-transfer rate in a contactor can be expressed as
droplets, and the film flowing on the wall of the
contactor. The coefficients for the packing alone were Nw ) (Kxa)contactor(x - x*av)contactorSZt
obtained from the composite coefficients for the contac- ) (Kxa)spray(x - x*av)spraySZs +
tors by separating the other contributions.
Contactor Overall Mass-Transfer Coefficients. (Kxa)packing(x - x*av)packingSZp (4)
The Wilson plots obtained from the mass-transfer
studies (Figures 4 and 5) have indicated that the gas- Since the concentration change of solution across the
phase mass-transfer resistance is negligible for these tower was very small (0.2 wt %), x was taken to be
contact operations. Therefore, constant. The change of x* due to spray alone was found
to be very small (<1%) (Potnis, 1994). Also, this change
kx ) Kx xi ) x* (1) would be further reduced in the presence of a packing,
as the height of the spray would be reduced. Therefore,
The height of a contactor can be expressed as the driving force term could be eliminated and the
volumetric mass transfer coefficient for the packing
could be obtained from eq 4.
Z) ∫K (x - x*)aS(1
L
- x)
dx (2)
x Zt Zs
(Kxa)packing ) (Kxa)contactor - (Kxa)spray (5)
Since the change of solution mole fraction across the Zp Zp
4188 Ind. Eng. Chem. Res., Vol. 35, No. 11, 1996

(Kxa)spray was obtained from the mass-transfer studies


conducted for spray alone, following the same method
employed to estimate (Kxa)overall. Using the values of
(Kxa)contactor, (Kxa)packing was then obtained from eq 5.
Random-Packing Mass-Transfer Coefficients.
The volumetric mass-transfer coefficient, Kxa, is a
combination of two independent variables, Kx and a.
Although these two variables depend on the same
parameters such as fluid flow rates, transport proper-
ties, and packing geometry, the impact of these param-
eters on these two variables is different. Therefore, it
is desirable to attempt to separate these two variables.
Onda et al. (1967) took into account the liquid surface
tension and surface energy of the packing materials and
derived an equation for the wetted surface area:
aw
at {
) 1 - exp -1.45
σc
σ () 0.75
Ret0.1Fr-0.05We0.2 } (6)
Figure 6. Comparison of mass-transfer coefficients obtained for
for different packings in the regenerator.

0.04 < Ret < 500

1.2 × 10-7 < We < 0.27

2.5 × 10-9 < Fr < 1.8 × 10-2


σc
0.3 < <2 (7)
σ
By using eq 6 and by assuming a wetted area equal
to the effective area, Onda et al. correlated a large
amount of liquid-phase mass-transfer data for organic
liquids as well as for water (Sherwood and Holloway,
1940; Vivian and King, 1964; Yoshida and Koyanagi,
1958; Onda et al., 1959; Hikita et al., 1954) to within
(20%. Packings studied included a variety of sizes of
Raschig rings, Berl saddles, pall rings, spheres, and
rods, made of ceramic, glass, poly(vinyl chloride), and
paraffin-coated material. Figure 7. Comparison of mass-transfer coefficients obtained for
Due to the success of eq 6, it is being widely used as the dehumidifier using different packings.
a method of predicting the wetted area of packings
(Ponter and Au-Yeung, 1986; Nawrocki et al., 1991). coefficients. Figures 6 and 7 show the variation of mass-
However, this equation is developed for the wetted transfer coefficients obtained with the variation of
areas, and the effective area in a packed column is not solution flow rate for the random packing used, and the
always the same as the wetted area. The wetted area following are the corresponding correlations:
comprises the static area, due to static holdup of liquid
in the column, and the dynamic area, due to dynamic Kx ) 0.5(L*)1.2 for the regenerator (9)
holdup. For absorption without chemical reaction, the
stagnant and semistagnant pockets of fluids, which Kx ) 0.4L* for the dehumidifier (10)
contribute to static holdup, tend to become saturated,
rendering the static area ineffective for mass transfer. where Kx is the overall liquid-phase mass-transfer
Therefore, the effective interfacial area is nearly the coefficient (mol/(m2 s) (mole fraction)) and L* is the
same as the dynamic area. Puranik and Vogelpohl solution flow rate per unit cross-sectional area (kg/(s
(1974) carried out a statistical analysis of the available m2)).
data of effective interfacial areas for absorption with and Structured-Packing Mass-Transfer Coefficients.
without chemical reaction and found that the static area The coating material used for the structured packing
equals zero for was carbon-impregnated cellulose, which is a strong
adsorbent of water. This feature combined with a cross-
We fluted surface provides near complete wettability of
g 12.5 (8) these packings. Also, Erickson (1968) found that even
Fr
a deliberate distribution of water on only 50% of the
Thus, if this criterion is satisfied, then eq 6 can be cross-sectional area provides an effective area of more
used for predicting effective areas, provided other than 80% of the dry area. As the nozzles employed in
parameters satisfy the criteria given in eq 7. Therefore, the present studies are arranged to give near complete
after determining that the range of parameters used in coverage of the cross-sectional area of the packing, the
the present studies satisfied these criteria, eq 6 was effective area was assumed to be same as the dry area.
used to obtain effective areas. Using these areas, mass- This assumption was also found to yield satisfactory
transfer coefficients for the random packing were esti- results for studies conducted by Griffiths (1993) for
mated from the corresponding volumetric mass-transfer lithium chloride solution and by Dowdy and Karabash
Ind. Eng. Chem. Res., Vol. 35, No. 11, 1996 4189
(1987) for evaporative cooling. Using this assumption, which in turn change the thickness of the boundary
the mass-transfer coefficients were calculated from the layer and the local mass-transfer coefficient.
volumetric coefficients for structured packing, obtained The solution sprayed over the randomly packed bed
earlier. The variation of these coefficients with solution flows over a large number of small packing units
flow rate is shown graphically in Figures 6 and 7, and (Tripack spheres), which are divided into small separate
the following are the corresponding correlations: surfaces. The solution sprayed over structured packing
enters the flutes or the corrugations of the structured
for the regenerator packing and flows almost undisturbed over the fluted
surface until it comes out of the layer of structured
Kx ) 0.01(L*)1.1 for structured packing (30 cm) packing. Therefore, the distance traveled by the solu-
(11) tion on a structured packing surface is larger than that
on a random packing surface. Thus, from eq 15, the
Kx ) 0.015L* for structured packing (55 cm) (12) boundary layer thickness for the structured packing is
greater than that for the random packing. It is thus
reasonable that the random-packing mass-transfer coef-
for the dehumidifier ficient is significantly higher than that for the struc-
tured packing in both contactors, as observed in this
Kx ) 0.02(L*)0.9 for structured packing (30 cm) research.
(13) When two layers of structured packing are stacked,
due to crosswise arrangement, the solution undergoes
Kx ) 0.015L* for structured packing (55 cm) (14) some mixing and the films flowing down the flutes of
the structured packing are disturbed near the junction
where Kx is the overall liquid-phase mass-transfer of the two layers. This causes a reduction in the average
coefficient and L* is the solution flow rate per unit cross distance traveled on the packing surface by the solution
sectional area. films, which in turn tends to reduce the thickness of
the boundary layer. However, when solution flows from
Comparison of Mass-Transfer Coefficients one layer of packing to another, it undergoes a pressure
drop. This causes a reduction in the approach velocity
Data obtained from the mass-transfer-rate studies for (ν∞) of the solution for the second layer of packing, which
both contactors were analyzed for the presence of pinch tends to increase the boundary layer thickness. There-
or crossover points, and it was determined that no such fore, these processes which are in opposite directions
points were reached during these studies. Since the determine the change of boundary layer thickness with
solution concentration across each contactor remained the addition of another layer of packing.
almost constant and since there were no pinch or Figure 7 shows that, considering experimental error,
crossover points reached in any contactor, the average the mass-transfer coefficients for single and double
overall mass-transfer coefficient obtained can be con- layers of packing are almost the same. This indicates
sidered to be the local mass-transfer coefficient. Also, that there is no major change in the boundary layer
since the mass-transfer studies for both contactors were thickness with the addition of another layer of struc-
carried out at the same solution concentration, a tured packing in the dehumidifier. However, in the
comparison of the mass-transfer coefficients obtained regenerator, the mass-transfer coefficients for the double
for the regenerator and the dehumidifier is justified. layer of structured packing are 50% higher than those
Therefore, it is conceivable that the transport properties obtained for the single layer of structured packing. Also,
of the liquid phase may have played a significant role the ratio of the mass-transfer coefficients obtained for
in determining the trends of the mass-transfer coef- the random packing to those obtained for the double
ficients. layer of structured packing is near 40 in the regenerator
The transfer of the water molecules from the bulk and near 30 in the dehumidifier. This implies that the
solution to the air involves transport from the turbulent change of boundary layer thickness with change of
bulk to the laminar boundary layer and transport across packing depends on the type of contactor.
the laminar boundary layer. Due to the slow rate of If the ratios of the boundary layer thicknesses ob-
transport, the mass-transfer resistance resides prima- tained for different packings in each contactor are
rily in the laminar liquid boundary layer at the gas- considered, then the transport property terms will
solution interface. This mass-transfer resistance is cancel because all packings are operated at the same
directly proportional to the thickness of the boundary average temperature in each contactor. Thus, the
layer (Treybal, 1981). Therefore, the thickness of the values of the ratios will be primarily determined by the
boundary layer determines the relative influence of each distances traveled by the solution on the packing
region on the net rate of mass transfer and the local surfaces and the approach velocities of the solution. The
mass-transfer coefficient. changes in these two quantities can be linked to the
The thickness of the laminar boundary layer depends changes in surface tension and pressure drop with
on various parameters such as the transport properties temperature.
of the solution, the distance traveled by the solution film The regenerator solution temperature (41 °C) is
on the packing surface (X), and the velocity with which higher than the dehumidifier solution temperature (25
the solution approaches the packing surface (ν∞). This °C). This higher temperature reduces the viscosity and
dependence can be represented by the equation (Bird surface tension of the solution in the regenerator. The
et al., 1960) reduction in viscosity causes a reduction in the pressure
drop, which in turn increases the approach velocity for

x ( )
µ X DF 1/3 the next unit of packing. The reduction in the surface
δm ) C1 (15)
F ν∞ µ tension helps the solution film detach from the packing
surface, thus reducing the distance traveled by the film
Different types of packing and contactors change the (X in eq 15) on a single packing surface. Thus, both
parameters involved on the right-hand side of eq 15, processes tend to reduce the boundary layer thickness
4190 Ind. Eng. Chem. Res., Vol. 35, No. 11, 1996

with an increase in temperature. It is thus reasonable


that the mass-transfer coefficients undergo greater
changes in the regenerator than in the dehumidifier.
The mass-transfer coefficients obtained for a single
layer of structured packing in the regenerator are lower
than those obtained in the dehumidifier. After entering
the flutes of the structured packing, solution flows
almost undisturbed over the packing surface. There-
fore, the distance traveled by the solution on the surface
of the single layer of structured packing is the same in
both contactors. Also, since the mass-transfer coef-
ficients are compared at the same flow rate, the solution
approach velocities are the same in both contactors for
this case. The density change is also negligible for the
change of temperature considered here. Therefore,
according to eq 15, this change of mass-transfer coef-
ficients can be attributed to the change of viscosity and
the change of diffusivity with temperature. Since the
change of absolute temperature between the dehumidi- Figure 8. Variation of Sherwood number with Reynolds number
for different packings in the dehumidifier.
fier and the regenerator is small (≈5%), the diffusivity
can be considered to be inversely proportional to the
viscosity (Cussler, 1984). As the temperature increases, The characteristic length of packing (Dp), required for
the viscosity of the solution decreases. Therefore, it calculation of the Sherwood number and Reynolds
follows from eq 15 that the boundary layer thicknesses number, was defined as the diameter of a sphere
are higher for a single layer of structured packing in possessing the same surface area as a unit of packing
the regenerator than those in the dehumidifier. Also, (Shulman et al., 1955). The characteristic length for one
this increased boundary layer thickness in the regen- sphere (one unit) of the random packing was found to
erator will reduce the mixing effects at the entrance and be 0.027 m. The diameter of the polypropylene sphere
exit of the packing. Thus, the mass-transfer coefficients determined from the terminal velocity of the sphere in
obtained for the single layer of structured packing in water was found to be 0.028 m. The actual diameter of
the regenerator are lower than those obtained in the the polypropylene sphere was 0.025 m. These values,
dehumidifier. determined using different methods, confirmed that the
characteristic dimension of the random packing was the
diameter of the polypropylene tripack sphere. There-
Representation of Mass-Transfer Coefficients in
fore, the diameter obtained from the conventional
Dimensionless Form
definition used by Shulman et al. was used to obtain
The results of mass-transfer experiments are usually the dimensionless numbers.
correlated using three dimensionless groups: Sherwood One flute of the corrugated surface was considered
number , Reynolds number, and Schmidt number to be one unit of the structured packing. The dimen-
sions of the fluted surface were determined, and the
Sh ) CReaScb (16) surface area of the flute was equated to the surface area
of a regular sphere. The characteristic length thus
In the present studies, the dependence of the mass-
obtained was found to be 0.087 m.
transfer coefficient on the solution flow rate was studied,
Although the bulk solution is turbulent, the boundary
while maintaining all other parameters constant. Thus,
layer near the gas-liquid interface is in laminar motion.
the Schmidt number was kept constant. Therefore,
Also, due to high thermal conductivity of the solution,
according to eq 16, log-log plots of Sherwood number
the temperature gradient across this layer is small, and
vs Reynolds number should be straight lines.
due to a small concentration difference, the density
Film theory suggests that mass-transfer coefficients
difference across this layer is small. Therefore, the
are directly proportional to diffusivity, indicating that
water transport across this boundary layer is primarily
the exponent for the Schmidt number in eq 16 is zero.
due to molecular diffusion (Treybal, 1981; Cussler, 1984;
However, film theory assumes the presence of a stag-
Bird et al., 1960). Thus, the Sherwood number can be
nant film at the interface and is applicable in only a
defined as
few cases, such as for the presence of high mass-transfer
flux and mass transfer associated with chemical reac-
tion. Since in the present studies the gas-liquid KxDp
Sh ) (17)
interface is not stagnant, film theory is not considered cD
to be applicable.
Penetration theory (Higbie, 1935) and surface-re- The log-log plots of Sherwood number vs Reynolds
newal theory (Danckwerts, 1951, 1955) assume a non- number for different packings in the dehumidifier are
zero velocity at the gas-liquid interface for predicting shown in Figure 8.
the dependence of the mass-transfer coefficient (kl) on To obtain Sherwood numbers for the regenerator, it
diffusivity (DAB). Both theories predict that kl is pro- is necessary to have the value of diffusivity of water in
portional to DAB0.5, indicating that the exponent of the lithium bromide solution at the regenerator solution
Schmidt number is 0.5 in eq 16. This prediction is temperature (41 °C). The diffusivity data obtained
verified through the correlations proposed by Sherwood earlier were at 25 °C (Potnis et al., 1995). The value of
and Holloway (1940) and Shulman et al. (1955) for diffusivity at 41 °C was thus obtained from these values
random packings and by Lamourelle and Sandall (1972) following the procedure discussed in the following
for a wetted wall column. Therefore, the exponent of paragraphs.
Schmidt number was assumed to be 0.5 for the present Since the solution used in the present studies is a
studies. concentrated solution, commonly used equations for
Ind. Eng. Chem. Res., Vol. 35, No. 11, 1996 4191
dilute solutions cannot be used for these calculations.
Using the Stokes-Einstein equation for concentrated
solution (Batchelor, 1972,1976; Reed and Anderson,
1980),

D1 T1
(18)
D 2 T2
)

Using this equation, the diffusivity of water in the


regenerator was found to be 0.97 × 10-9 m2/s.
However, this equation sometimes predicts smaller
variations of diffusivity with changes of concentrations
for small solutes than those experimentally observed
(Cussler, 1984). Rosevaere et al. (1941) studied a
nonideal mixture of chloroform-ethyl ether and ob-
tained the product D0µ as


D 0µ ) (19)
d log γ
1+ Figure 9. Variation of Sherwood number with Reynolds number
d log c
for different packings in the regenerator.
They plotted these values as a function of concentra-
tion and found that the product D0µ has a very small solution flow rates. Mangers and Ponter (1980) have
linear dependence on concentration. Therefore, found that the exponent for the solution flow rate is 1.44
for highly viscous solutions using random packings, a
D0µ ≈ constant (20) case similar to that studied in the present research.
Although this exponent is for kLa data, the packings
Applying limd log Cf0, d log γf0 to eq 19 and using the were completely wet. Thus, the exponent for the solu-
relation between the activity coefficient and the dif- tion flow rate should be the same for variation of the
ferential heat of dilution (Smith and Van Ness, 1975), mass-transfer coefficient alone. Also, Cornell et al.
the diffusivity in the regenerator was found to be 0.83 (1960) have found that the exponent for the solution flow
× 10-9 m2/s. However, this prediction, again, is not rate is 1 for a large amount of mass-transfer data
completely reliable, as D0µ shows some dependence on including Sherwood-Holloway data for random pack-
concentration. Therefore, the average of the values ings. Again, this exponent is for kLa data. However,
predicted by both equations (0.9 × 10-9 m2/s) was used using the corresponding correlation for the effective
for the diffusivity value at the average regenerator area, the exponent for solution flow rate is found to be
temperature. Using this value of diffusivity, Sherwood 0.93 for variation of the mass-transfer coefficient alone.
numbers for the regenerator were calculated. The log- Thus, the exponents obtained in the present studies are
log plots of these Sherwood numbers vs Reynolds slightly higher than this exponent. However, the liquid
numbers are shown in Figure 9. As discussed earlier diffusivity values involved in Cornell et al. studies are
the exponent for the Schmidt number is assumed to be significantly higher than those involved in the present
0.5 to obtain correlations to predict the Sherwood studies. Although the liquid viscosity values are smaller,
numbers. The equations thus obtained for a liquid considering the inverse proportionality of the diffusivity
desiccant system using structured and random packings and the viscosity, the boundary layer thicknesses are
in the regenerator and the dehumidifier are higher for the Cornell et al. studies than those for the
present studies. Thus, for the Cornell et al. work, the
regenerator influence of the laminar layer should be greater and this
is consistent with a smaller exponent for the solution
Sh ) 0.46Re1.2Sc0.5 for random packing (21) flow rate relative to the present studies. Therefore, the
exponents for the Reynolds numbers obtained in the
Sh ) 0.01ReSc0.5 for structured packing (30 cm) present studies appear to be consistent with the data
(22) obtained by other researchers.

Sh ) 0.02ReSc0.5 for structured packing (55 cm) Comparison with Another Dimensionless
(23) Correlation

dehumidifier The experimentally obtained mass-transfer coeffi-


cients were compared with those predicted by a correla-
tion proposed by Onda et al. (1968) (Figures 10 and 11).
Sh ) 0.8ReSc0.5 for random packing (24)

Sh ) 0.04Re0.9Sc0.5 for structured packing (30 cm)


(25)
k1( )Fl
µg
1/3
( )( )
) 0.0051
L*
awµ
2/3
µ
FlD
-1/2
(atDp)0.4 (27)

Sh ) 0.03ReSc0.5 for structured packing (55 cm) The mass-transfer coefficients predicted by this cor-
(26) relation for the random packings were found to be
within (20% of those obtained at lower flow rates in
The exponents for the Reynolds number obtained in the present studies. However, the difference was found
these correlations vary from 0.9 to 1.2. These exponents to increase with an increase in flow rate. Again, this
result from the exponents obtained for the solution flow behavior can be explained through the types of packings
rates (L) from the plots of mass-transfer coefficients vs and fluids studied. Due to a variety of fluids and
4192 Ind. Eng. Chem. Res., Vol. 35, No. 11, 1996

were found to be close to 0.8, which indicated that the


conditions for the liquid phase were near turbulent for
the contact operations in the regenerator and the
dehumidifier. The small intercepts obtained for the
Wilson plots indicate that the gas-phase mass-transfer
resistance was negligible compared to the liquid-phase
mass-transfer resistance for these contact operations.
Mass-transfer rates obtained in the regenerator for the
randomly packed bed were found to be approximately
300% and 130% greater and those obtained in the
dehumidifier were found to be approximately 60% and
45% greater than those for the structured packed beds
with heights 30 and 55 cm, respectively.
The mass-transfer coefficients obtained for the ran-
dom packing varied from 0.48 to 2 mol/(s m2), while
those for the double layer of structured packing varied
from 0.018 to 0.035 mol/(s m2). The significant differ-
ence in these values can be explained by the difference
in the geometry between the two packings. In addition,
the structured-packing mass-transfer area is signifi-
Figure 10. Comparison of the random-packing mass-transfer
coefficients in the regenerator with those obtained using Onda’s cantly larger than that of the random packing. Thus,
correlation. the random-packing mass-transfer rates are only about
50% higher (not 30 times higher) than the structured-
packing mass-transfer rates in the dehumidifier.
These mass-transfer coefficients and the experimen-
tally determined diffusivity values were then utilized
to obtain dimensionless correlations between Sherwood
number (Sh), Reynolds number (Re), and Schmidt
number (Sc). Since the mass-transfer contributions of
the regions other than the packed beds were separated,
the dimensionless equations developed in these studies
should be useful for design of such liquid-desiccant
contactors with different geometries and liquid-distribu-
tion devices, employing similar packings and desiccants.

Acknowledgment
We appreciate the support for this work provided by
the Active Heating and Cooling Division, Office of Solar
Heat Technologies, U.S. Department of Energy.

Nomenclature
a ) effective area of the packing (m2/m3)
Figure 11. Comparison of the random-packing mass-transfer at ) total surface area of the packing (m2/m3)
coefficients obtained in the dehumidifier with those predicted by aw ) wetted surface area of the packing (m2/m3)
Onda’s correlation. c ) molar density (mol/m3)
C1 ) constant determined by the velocity profile in the
packings studied, a large range of at/a ratio is covered laminar boundary layer
by this correlation. However, the diffusivities encoun- D ) diffusivity of water in the lithium bromide solution
tered in the present studies are lower than those studied (m2/s)
by Onda et al., and the values for the at/a ratio obtained D1 ) diffusivity value at the dehumidifier temperature (m2/
for the present studies appear to be on the lower side s)
of the range covered by Onda et al. Thus, the average D2 ) diffusivity value at the regenerator temperature (m2/
boundary layer thickness is likely higher for the studies s)
by Onda et al. Therefore, the mass-transfer coefficients D0 ) transport coefficient (m2/s)
obtained by Onda et al. are more influenced by the Dp ) characteristic length of packing (m)
Fr ) (L2at)/(Fl2g) Froude number
laminar region. This results in the dependence of kl on
g ) gravitational constant (m/s2)
(L*)2/3, which is typical of the transition region. There-
G ) gas flow rate per unit cross sectional area (kg/(s m2))
fore, although this correlation predicts the mass-transfer kl ) liquid-phase mass-transfer coefficient (mol/(m2 s)
coefficients at lower flow rates to within (20%, as the (concentration difference))
effect of turbulence increases with the increase in flow kx ) local liquid-phase mass-transfer coefficient (mol/(m2
rate, the difference between the predicted coefficients s) (mole fraction))
and those obtained in the present studies increases. Kx ) local overall liquid mass-transfer coefficient (mol/(m2
s) (mole fraction)
Conclusions L ) solution flow rate (mol/s)
L* ) solution flow rate per unit cross-sectional area (kg/(s
Mass-transfer-rate studies were conducted and di- m2)))
mensionless mass-transfer coefficients were obtained for Nw ) local rate of mass transfer (mol/s)
a packed-bed liquid-desiccant system. The slopes of the Ret ) L/(atµl) ) Reynolds number based on total dry
log-log plots of mass-transfer rate vs solution flow rate packing area
Ind. Eng. Chem. Res., Vol. 35, No. 11, 1996 4193
S ) cross-sectional area of the packing (m2) midifiers and Regenerators. ASHRAE Trans. 1992, RP-298,
Sc ) Schmidt number ) µ/DF 525.
Sh ) Sherwood number ) KxDp/cD Lamourelle, A. P.; Sandall, O. C. Gas Absorption into a Turbulent
T ) absolute temperature (K) Liquid. Chem. Eng. Sci. 1972, 27, 1035.
Lenz, T. G.; Potnis, S. V. Energy balance and mass transfer studies
T1 ) average temperature for the dehumidifier (K) for a liquid desiccant based solar cooling system. In Solar World
T2 ) average temperature for the regenerator (K) Congress; Arden M. E., Burley, M. A., Coleman, M., Eds.;
v∞ ) approach velocity of the solution (m/s) Pergamon: New York, 1991; Vol. 2 Part 1, p 1613.
We ) L2/(Flσat) ) Weber number Löf, G. O. G.; Lenz, T. G.; Rao, S. Coefficient of Heat and Mass
x ) mole fraction of water in solution Transfer in a Packed Bed Suitable for Solar Regeneration of
X ) distance traveled by the solution on a packing surface Aqueous Lithium Chloride Solutions. J. Solar Energy Eng.
(m) 1984, 106, 387.
xi ) mole fraction of water at the interface Mangers, R. J.; Ponter, A. B. Effect of Viscosity on Liquid Film
x* ) mole fraction of water in equilibrium with bulk Resistance to Mass Transfer in a Packed Column. Ind. Eng.
Chem. Process Des. Dev. 1980, 19, 530.
concentration in air Nawrocki, P. A.; Xu, Z. P.; Chuang, K. T. Mass Transfer in
Zp ) packing height (m) Structured Corrugated Packing. Can. J. Chem. Eng. 1991, 69,
Zs ) Zt - Zp (m) 1336.
Zt ) distance between the spray nozzles and the pan used Onda, K. H.; Sada, E.; Murase, Y. Liquid-Side Mass Transfer
to collect the solution (m) Coefficients in Packed Towers. AIChE J. 1959, 5, 235.
Onda, K. H.; Takeuchi, H.; Koyama, Y. Kagaku Kojgaku 1967, 31,
Greek Symbols 126.
δm ) boundary layer thickness for mass transfer (m) Onda, K. H.; Takeuchi, H.; Okamoto, Y. Mass Transfer Coefficients
φ1 ) volume fraction of the solute between Gas and Liquid Phases in Packed Columns. J. Chem.
Eng. Jpn. 1968, 1, 56.
γ ) activity coefficient Potnis, S. V. Development of Dimensionless Mass Transfer Cor-
µ ) viscosity of the solution (Pa s) relations for Packed Bed Liquid Desiccant Contactors. Ph.D.
µ1 ) viscosity of the liquid (Pa s) Dissertation, Colorado State University, Fort Collins, 1994.
F1 ) density of the solution (kg/m3) Potnis, S. V.; Lenz, T. G.; Dunlop, E. H. Measurement of Water
σ ) surface tension of the liquid (103N/m) Diffusivity in Aqueous Lithium Bromide and Lithium Chloride
σc ) critical surface tension of the packing material (103N/ Solutions. Chem. Eng. Commun. 1995, 139, 41.
m) Ponter, A. B.; Au-Yeung, P. H. Estimating Liquid Film Mass
Transfer Coefficients in Randomly Packed Towers. In Handbook
of Heat and Mass Transfer; Cheremisinoff, N. P., Ed.; Gulf:
Literature Cited 1986; Vol. 2, p 938.
Bird, R. B.; Stewart, W. E.; Lightfoot, E. N. Transport Phenomena; Puranik; Vagelpohl. Unpublished results, 1974.
John Wiley and Sons: New York, 1960. Reed, C. C. Anderson J. L. Hindered Settling of a Suspension at
Batchelor, G. K. Sedimentation in a dilute dispersion of spheres. Low Reynolds Number. AIChE J. 1980, 26, 816.
J. Fluid. Mech. 1972, 52, 245. Batchelor, G. K. Brownian Rosevaere, W. E.; Powell, R. E.; Eyring, H. The Structure and
diffusion of particles with hydrodynamic interaction. J. Fluid. Dynamics of Liquids. J Appl. Phys. 1941, 12, 669.
Mech. 1976, 74, 1. Sherwood, T. K.; Holloway, F. A. L. Performance of Packed Towers-
Cornell, D.; Knapp, W. G.; Fair, J. R. Mass Transfer Efficiency- Liquid Film Data for Several Packings. Trans. Am. Inst. Chem.
Packed Columns Part 1. Chem. Eng. Prog. 1960, 56 (8), 68. Eng. 1940, 36, 39.
Cussler, E. L. Diffusion: Mass transfer in Fluid Systems; Cam- Shulman, H. L.; Ullrich, C. F.; Proulx, A. Z.; Zimmerman, J. O.
bridge University: Cambridge, MA, 1984. Performance of Packed Columns II. Wetted and Effective
Danckwerts, P. V. Gas Absorption Accompanied by Chemical Interfacial Areas, Gas- and Liquid-Phase Mass Transfer Rates.
Reaction. AIChE J. 1955, 1, 456. Danckwerts, P. V. Significance AIChE J. 1955, 1, 253.
of Liquid-Film Coefficients in Gas Absorption. Ind. Eng. Chem. Smith, J. M.; Van Ness, H. C. Introduction to Chemical Engineer-
1951, 43, 1460. ing Thermodynamics, 3rd ed.; McGraw-Hill: New York, 1975;
Dowdy, J. A.; Karabash, N. S. Experimental Determination of Heat p 613.
and Mass Transfer Coefficients in Rigid Impregnated Cellulose Treybal, R. E. Mass Transfer Operations; McGraw-Hill: New York,
Evaporative Media. ASHRAE Trans. 1987, 93 (3, No. 3078). 1981.
Erickson, B. Munters’ Gas-To-Liquid Contact Surfaces. Company Vivian, J. E.; King, C. J. The Mechanism of Liquid Phase
Report, 1968. Resistance to Gas Absorption in a Packed Column. AIChE J.
Gandhidasan, P; Kettleborough, C. F.; Ullah, M. R. Calculation 1964, 10, 221.
of Heat and Mass Transfer Coefficients in a Packed Tower Wahlig, M. Absorption Systems and Components, In Active Solar
Operating with a Desiccant-Air Contact System. J. Solar Systems; Löf, G. O. G., Ed.; MIT: Cambridge, MA, 1993; p 811.
Energy Eng. 1986, 108, 123. Yoshida, F.; Koyanagi, J. Liquid Phase Mass Transfer Rates and
Gandhidasan, P; Ullah, M. R.; Kettleborough, C. F. Analysis of Effective Interfacial Area in Packed Absorption Columns. Ind.
Heat and Mass Transfer between a Desiccant-Air System in a Eng. Chem. 1958, 50, 365.
Packed Tower. J. Solar Energy Eng. 1987, 109, 89.
Griffiths, W. C. Desiccant-Based Environmental Control Systems. Received for review April 15, 1996
Strategies for energy efficient plants and intelligent buildings; Revised manuscript received July 8, 1996
Fairmont: 1987. Accepted July 9, 1996X
Griffiths, W. C. Personal communication, 1993.
Higbie, R. The Rate of Absorption of a Pure Gas into a Still Liquid IE960212Y
During Short Periods of Exposure. Trans. AIChE 1935, 31, 365.
Hikita, H.; Sugata, E.; Kamo, K. Kagaku Kojgaku. 1954, 18, 454.
X Abstract published in Advance ACS Abstracts, September
Khan, A. Y.; Ball, H. D. Development of Generalized Model for
Performance Evaluation of Packed Type Liquid Sorbent Dehu- 15, 1996.

You might also like