Professional Documents
Culture Documents
1–10, 1998
© 1997 Elsevier Science Ltd
PII: S0038-092X(97)00080-7 All rights reserved. Printed in Great Britain
0038-092X/98 $19.00+0.00
Abstract—A composite desiccant dehumidifier made of mixed inert and desiccant materials is investigated.
A heat and mass transfer model that incorporates the composite nature of the structure is discussed. The
model includes both gas-side and solid-side resistances for heat and mass transport. The solid-side resis-
tance for the mass transport includes gas-phase diffusion and surface diffusion. The effectiveness for
moisture removal and heat transfer during adsorption and desorption processes in such a dehumidifier
during single blow operation is investigated. Results are presented for composite structures made of silica
gel and inert materials of different compositions and thermophysical properties. © 1997 Elsevier
Science Ltd.
(a)
(b)
Fig. 1. (a) The desiccant cooling system configuration. (b) The desiccant cooling system illustrated on a psychometric
diagram.
Fig. 2. Schematic of the flow channel and the sorbing felt made of composite structure.
increased capacity of the sensible heat exchang- cant and inert materials of varying thermophysi-
ers. One would like to achieve processes such cal properties.
as 1–2∞, 1–2◊ and ideally a process 1–2+ that In the present study, the performance of
gives dehumidification at constant temperature adsorption and desorption processes during a
in the dehumidifier and reduced sensible cooling single blow operation in a dehumidifier made
in the heat exchanger (see Fig. 1(b)). It seems of composite mixture of desiccant particles and
that such processes can be approached by using inert particles or fibers is investigated. The
a composite felt structure made of mixed desic- mathematical model used to simulate the tran-
Heat and mass transfer 3
sient coupled heat and mass transfer processes 2.1. Effective thermal conductivity
in a porous composite structure is presented. In order to model the dissipation of heat in
The model includes both gas-side and solid-side the wet composite desiccant felt, it is necessary
resistances for heat and mass transport. The to predict the effective thermal conductivity of
solid-side resistance for mass transfer includes the composite media. The effective thermal con-
gas-phase diffusion and surface diffusion in ductivity is primarily dependent on the thermal
intra-particle pores. The investigation will be conductivities of the composite phases, on that
based on using composite structures made of of a liquid phase of water, of the vapor phase,
silica gel particles and inert materials of different and the structure of the porous felt and par-
compositions and thermophysical properties, ticles. Sarwar and Majumdar (1995) developed
and assuming a negligible effect of the inert a model to predict the effective thermal conduc-
material on the moisture sorptive properties of tivity of sorbing desiccant felt with inert materi-
the silica gel. als. The effective thermal conductivity of the
desiccant felt is calculated in two steps. First,
the thermal conductivity of the particle is com-
2. MATHEMATICAL MODEL
puted for the given particle porosity, pore diam-
The sorbing felt composed of a composite eter, and thermal conductivities. The thermal
mixture of desiccant and inert material is shown conductivity of the felt is then computed based
in Fig. 2. The felt is assumed to be a particulate on the computed desiccant particle conductivity,
composite structure and is made of a mixture the inert material conductivity, the fraction of
of silica gel and inert particles. This composite inert material, the porosity of the felt, and the
felt structure could be prepared by mixing com- particle size.
mercially available silica gel particles with com-
2.2. Moisture transport in microporous felt
patible inert material particles that have high
thermal capacitance. In the case of a particle, Water vapor can diffuse through a porous
we have solid matter with a porosity e defined medium by either ordinary diffusion, Knudsen
p diffusion or surface diffusion or, sometimes,
as the volume fraction of dry gel occupied by
air. As water slowly gets adsorbed in the mate- combinations of them, depending on the pore
rial, a portion of the volume formerly occupied size and pressure. If the pores are much larger
by air is filled with water. If we designate a than the mean free path of the vapor, then the
volume fraction of water as w and assume molecules collide with each other more fre-
wp quently than with the pore walls, and ordinary
that pore size or the desiccant volume does not
molecular diffusion is assumed to be the con-
vary with sorption of moisture, then we have
trolling factor. The ordinary diffusion coefficient
w <e . Similarly, for the case of desiccant felt
wp p is calculated based on the kinetic theory of
made of desiccant and inert particles, and
gases and expressed as
having a felt porosity e , the volume fraction of
f
=1ul, m2 s−1.
AO 3 :
water vapor in the desiccant felt w will be less D (6)
wf
than or equal to the felt porosity, e . With these
f For a water vapor–air mixture, the ordinary
assumptions the density, specific heat and con-
diffusion coefficient is given by
ductivity of the composite structures can be
expressed as follows: T1.685
D =1.735×10−9 , m2 s−1 (7)
AO P
density of the felt
where T is the temperature in kelvin and P is
r =(1−e )[(1−w )r +w r ]+e r (1) the pressure in atmospheres.
f f I p I i f aw
specific heat of the felt The mass diffusion flux with ordinary diffu-
sion can be expressed by
(1−e )r C +e r C
C= f p ps f aw aw (2) ∂Y*
f r m =−D . (8)
f z AO ∂z
where A
Knudsen diffusion is encountered in smaller
r =(1−w )r +w r (3) pores and low pressure. When water vapor is
aw wf a wf w
diffusing through a medium of smaller-size
C =(1−w )C +w C (4)
aw wf a wf w pores, and/or the vapor density is very low,
C =(1−w )C +w C . (5) molecules collide more frequently with the walls
ps I d I i
4 P. Majumdar
than with each other and the resistance to However, it requires considerable computa-
diffusion along the pore is primarily due to tional effort and is only valid for limited pore
molecular collisions with the walls. Assuming sizes at low concentrations. Lu et al. (1991)
the diffusional path is equivalent to a long obtained apparent solid-side diffusivity by
circular tube of radius a, and the two ends are matching the analytical solution to the experi-
maintained at different pressures, a Fick’s law mental data and showed the dependence of
type expression can be obtained for this type of surface diffusivity on the particle size and mois-
flow if a Knudsen diffusion coefficient is defined ture content. Sladek et al. (1974) proposed a
based on kinetic theory as simpler correlation for silica gel as
A B
T 1/2 D =D exp[−0.974×10−3(H /T+273.2)]
D =97a , m2 s−1. (9) s o ads
AK M (14)
1
The mass diffusion flux with Knudsen diffusion where
can be expressed as
D =1.6×10−10 m2 s−1. (15)
∂Y* o
m =−D . (10) A comparison of Knudsen and surface diffu-
z AK ∂z
A sion rates in pores shows that surface diffusion
In some intermediate range of pore sizes, dominates in pore sizes of 11 Å, whereas both
both ordinary and Knudsen diffusion contribute are important for gel with a pore size of 68 Å.
to the transfer of vapor in the media. In this If surface diffusion of the adsorbed layer is
range, the combined ordinary and Knudsen important, along with molecular and Knudsen
diffusion can be represented by assuming paral- diffusion, then the total diffusion flux can be
lel resistances and expressed as written as
A B
1 −1 ∂Y* ∂(r Y* )
A B
1 1 −1 1
p m
D = + (11) m =− + −D
A D D z D D ∂z s ∂z
AO AK AO AK A A
and the mass diffusion flux can be written as (16)
∂Y* and using eqn (11)
m =−D . (12)
z A ∂z ∂Y* ∂(r Y* )
A m =−D −D p m . (17)
Surface diffusion is the transport of adsorbed z A ∂z s ∂z
A A
molecules on the pore surface. A number of
models representing the possible mechanisms 2.3. Effective diffusion coefficient
for movement of adsorbed molecules on sur- Equations for diffusion coefficients are valid
faces are available. These are the hydrodynamic only for long, uniform radius capillaries, and
model, the mechanistic hopping model and should be modified for application to really
Fick’s law model. In the Fick’s law model, the porous media. In the case of diffusion through
surface flux is assumed to be proportional to a porous mass, the diffusion is reduced because
the concentration gradient and expressed as the free or open cross-section is a fraction of
the total. Also, the diffusion coefficient is
∂(r Y* )
m =−D p m (13) reduced owing to the presence of a tortuous
z s ∂z path and change in the cross-sectional area of
A
where D is the surface diffusion coefficient. pore channels. Using various available models,
s such as the cylindrical pore model, parallel pore
The surface diffusion coefficient was esti-
mated experimentally by various investigators model and random pore model, it can be shown
(Sladek et al., 1974; Kruckles, 1976; Lu et al., that an effective ordinary or Knudsen diffusion
1991). Kruckles (1976) considered the sub- coefficient can be expressed as
monolayer concentration range and proposed e
that the mechanism of surface diffusion was D = m D (18)
A,e t A
that of the activated hopping of molecules in a
random walk process. His proposed formula where e is the felt porosity that accounts for
m
was based on a combined mathematical model the reduction of free area for diffusion due to
and experimental data. This formula showed the presence of a solid phase and t is the
a dependence on initial moisture content. tortuosity factor that accounts for the increase
Heat and mass transfer 5
K
Similarly, the surface diffusion coefficient ∂Y*
=0 (24)
given by eqn (14) is valid for smooth surfaces, ∂z
and should be modified to account for the rough z=0
K A K B
walls of the porous media. Since the surface ∂Y* Y*
diffusion is a surface phenomenon, the reduc- =l Y− (25)
∂z 12 r
tion of the area normal to the direction of flux z=1 A z=1
due to the presence of the solid phase does not for temperature
affect the effective surface diffusivity. The effec-
K
tive surface diffusion coefficient is expressed as ∂T* ∂T
=l al (26)
D ∂z 15 ∂t
D = s (19) z=0
s,e t
K
s ∂T*
where t is the tortuosity factor for surface =l (T −T* ), (27)
s ∂z 16 1 z=1
diffusion. z=1
where
2.4. Mathematical model of sorption processes
The diffusion of mass and the conduction of eD a2
l = f , l = 1 , l =l l
heat that occurs in desiccant felt can be 1 (1−w )D 2 a2 4 1 2
described in terms of conservation equations, g A,e w
transfer rate equations, boundary conditions, (28)
and equilibrium relations for the desiccant felt.
The model is based on the following assump- c r k r
l = 1 a, l = w s ,
tions: (i) the axial conduction and diffusion in 5 c 6 (1−w )r QD
the channels are small compared with the con- 2 g p AO
vective flow of heat and mass, and are negligible; CAP
l = −1 (29)
(ii) the heat and mass transfer from the air in 7 (1−w )r Qc
g p 1
the process and cooling channels to the surface
of the desiccant felt can be approximated by 1 Cr
l =l l l , l = , CAP= f f
constant transfer coefficients; (iii) the axial con- 9 2 6 10 10 l l 1−w
duction and diffusion within the desiccant felts 7 8 f
are negligible; (iv) the rate of diffusion in pores (30)
of the particles is fast compared with that in
2h K a a a
the intra-particle channels. The set of governing l = 1 , l = YC w , l = al w
equations and boundary conditions describing 11 K (∂H /∂T ) 12 eD 14 a2
YC 1 1 f 1
this model will now be given.
(31)
2.5. Governing equations
r D ∂Y* ∂Y*
For the desiccant felt l = a AO , c = m , c = m (32)
13 r c k 1 ∂T* 2 ∂Y*
∂Y* ∂2Y* ∂T* s 2 e
=l −l (20) h a
∂t 4 ∂z2 5 ∂t
l =c l l , l = 1 w , D=D +D c .
15 al 13 14 16 k A,e s,e 2
∂T* ∂2T* ∂Y* e
=l +l . (21) (33)
∂t 9 ∂z2 10 ∂t
A K B
∂Y Y* Y*(x, 0)=Y* (34)
=2 −Y (22) I
∂x r
A z=1 T*(x, 0)=T* . (35)
∂T I
1 =l (T*| −T ) (23)
∂x 11 z=1 1 The inlet conditions are
The boundary conditions are as follows: T (0, t)=T , Y(0, t)=Y (36)
1 o o
6 P. Majumdar
Fig. 3. Cooling capacity with varying fraction of inert mate- Fig. 4. Instantaneous exit states of the adsorbing air stream
rial and different thermal capacitance ratio. for w =0.2 and varying thermal capacitance ratio.
I
Fig. 5. Adsorbing air exit temperature. Fig. 6. Average felt temperature during the adsorption
process.
where
D =0.0329, D =−0.4113×10−5,
1 2
D =0.0105×10−3,
3
D =0.6586×10−6,
4
D =0.7894×10−10,
5
D =0.6747×10−12
6
The composite structure under consideration
is assumed to have a negligible effect on the
sorption characteristics of silica gel.
Fig. 8. Instantaneous exit states of the desorbing air stream for Y =2.0 and varying fraction of inert material.
c
ratios the peaks shift towards the left with a shown in this figure, and approach ideal thermo-
lower fraction of inert material. A 19.5% dynamic processes such as 1–2∞ and 1–2◊, as
increase in cooling capacity is achieved with a previously indicated and shown in Fig. 1(b).
fraction of inert material w =0.2 and thermal Lower exit temperature of the adsorbing air
I
capacitance ratio Y =24.0. Instantaneous exit stream, as shown in Fig. 5, will also lead to a
c
states of the adsorbing air stream are shown on reduced requirement for sensible cooling in the
a temperature/humidity space domain in Fig. 4 heat exchanger to meet the space load.
for the case with w =0.2 and varying thermal Figure 6 shows the variation of the average
I
capacitance ratios. It can be observed that the temperature of the felt during adsorption for
adsorbing air stream exits at a lower temper- w =0.2 and varying thermal capacitance ratios.
I
ature and with slightly higher moisture content. Note that the adsorption process takes place
This caused a dehumidification process with less with a much lower temperature rise in the felt
sensible heating and resulted in a higher cooling with an increase in thermal capacitance ratios.
capacity. This is evident by observing that the For the case with Y =24.0, the average temper-
c
locations of exit states move towards the left at ature of the felt was increased only by 4–5°C
a lower enthalpy in the psychometric chart, as compared with 10°C for the case with Y =0.2.
c
A lower average temperature of the felt during capacity. It is also observed that the rate of
adsorption maintains a lower exit temperature desorption increases with an increase in the
of the air stream and a higher moisture removal fraction of inert material up to a value of
rate, as shown in Figs 5 and 7, respectively. w =0.4. Further increase in the fraction of inert
I
The effect of the fraction of inert material on material causes a deterioration of the desorption
the adsorption process is shown in Figs 8–10. rate, which causes a higher exit temperature
Figure 8 shows instantaneous exit states of the and a lower exit humidity ratio.
desorbing air on a temperature/humidity ratio
space domain for the case with Y =2.0 and
c
varying fraction of inert material. Results show
that the rate of desorption increases with an
REFERENCES
increase in the fraction of inert material up to
a value w =0.4. Further increase in the fraction Anderson D. R. and Pesaran A. A. (1991) Innovative solid
I desiccant substrates for desiccant dehumidifiers.
of inert material causes a deterioration of the ASHRAE Transactions: Symposia, pp. 578–586.
desorption rate, which causes a higher exit Belding W. A., Worek W. M., Novosel D. and Holeman
temperature and a lower exit humidity ratio, as (1991) Desiccant development for gas-fired desiccant
cooling systems. ASHRAE Transactions: Symposia,
shown in Figs 9 and 10, respectively. pp. 587–594.
Collier R. K., Cale T. S. and Lavan Z. (1986) Advanced
desiccant materials assessment. Gas Research Institute
4. CONCLUSIONS Report GRI-86/0182.
Dini S. and Worek W. M., Heat Recovery Sys 6, (1986) 151.
The performance of adsorption and desorp- Kruckles W. W. (1976) On gradient dependent diffusivity.
tion processes during a single blow operation Chemical Engineering Science 18, 1565–1576.
in a dehumidifier made of a composite mixture Lu L. T., Charoensupaya D. and Lavan Z. (1991) Determi-
nation of sorption rate and apparent solid-state diffusiv-
of silica gel particles and inert particles is ity of pure H O in silica gel using constant volume/
investigated. 2
variable pressure apparatus. J. Solar Energy 113,
It is observed that with higher values of 257–263.
capacitance ratio, such as 8 and above, the Majumdar P. and Worek W. M. (1989) Combined heat and
mass transfer in a porous adsorbent. Energy 14, 161–175.
cooling capacity increases to a maximum peak Sarwar M. K. and Majumdar P. (1995) Thermal conductiv-
value at around a fraction of inert material of ity of wet composite porous media. Heat Recovery Sys-
w =0.2. A 19.5% increase in cooling capacity is tems and CHP 15, 369–381.
I Sladek, K. J., Gilliland, E. R. and Buddour, R. F. (1974)
achieved with a thermal capacitance ratio Diffusion on surface II, correlation of diffusivities of
Y =24.0. Under these conditions, the adsorb- physically and chemically adsorbed specie. Industrial
c
ing air stream exits at a lower temperature and Chemical Fundamental, 13(2).
with a slightly higher moisture content. This Worek W. M., Zheng W., Belding W. A., Novosel D. and
Holeman W. D. (1991) Simulation of advanced gas-
caused a dehumidification process with less sen- fired desiccant cooling systems. ASHRAE Transactions
sible heating and resulted in a higher cooling 97, 609–614.
10 P. Majumdar