Modeling and Testing of A Silica Gel Packed-Bed System
Modeling and Testing of A Silica Gel Packed-Bed System
Modeling and Testing of A Silica Gel Packed-Bed System
1994
Copyright Q 1994 Elsewer Science Ltd
Printedin Great Britain. All nghts reserved
0017-Y310;94 17.00+0.00
Pergamon
0017-9310(93)EOO90-4
of Mechanical
SAN7
and GWO-DONQ
Engineering,
National Chung-Hsing
Taichung, Taiwan, R.O.C.
JIANG
University,
250, Kuo-Kuang
Road,
1. INTRODUCTION
tigated. A comparison
between the result of experimental measurement
and that of theoretical analysis
for a thin silica gel packed bed was performed
by
Clark et al. [3]. The analysis was based on a gasside controlled heat and mass transfer model which
incorporates
the solid-side diffusion resistance in a
gas-side mass transfer coefficient. In the comparison a
high predicted initial rate of adsorption was obtained.
The sorption rate and apparent solid-side diffusivity
of water in silica gel was studied by Kruckels [4] and
Lu et al. [S]. In both works the surface diffusion is
verified to be the most important mechanism for water
vapor adsorbed by regular density silica gel. Thus
the ordinary diffusion and Knudsen diffusion can be
neglected in theoretical analysis.
Packed-bed systems can be applied to many industrial drying processes, such as the dehumidification
of
plastic beads in precise injection molding processes,
agricultural or industrial warehouse humidity control
and the dehumidification
of high pressure air supplied
kmperature
Zolumn
,
TAuthor to whom all the correspondence should be sent.
Humidity
CantroIIed
Column
A,
~~<;;zressor
FIG.
1173
and
1. Schematic
of a packed-bed
system.
1174
JIANC
NOMENCLATURE
l-9
A
A,
A,
measuring points
cross-sectionai area of column [m]
column free flow area [m]
column total heat or mass transfer area
b21
Bi,
c
cw
c*
L)+
D s,eff
f
G
h
kc
H ads
KG
L
Le
ml
Ntu
P
AP
P
r
r*
R
Re
t*
r, T,
T*
v
MODEL
Modeling
amI,
az* = Ntuh,s-md(l
L
h
&
L -+
I Fluid Friction
Diasipatioq
Gas Side
-m,J
(3)
aT*
2
=
az*
1175
system
Ntu[Le(T:-T,*+y,(T:-T,(m,,,-m,,,)l
in a pseudo-channel.
boundary
condition
where
pseudo-gas-side
controlled model (PGC) [l]. In the
model the overall mass transfer from the air stream to
the adsorbent is represented by a gas-side coefficient
which is reduced to account for solid-side resistance.
Recently, Pesaran and Mills [ 1] proposed a solid-side
resistance model (SSR) for predicting the transient
response of the heat and mass transfer in thin silica
gel packed beds. The model includes both solid and
gas-side resistances in the analysis of the diffusion of
water vapor. Thus it is more delicate and accurate
than the PGC model. In this work the solid-side resistance model is modified by adding a fluid friction effect.
The fluid friction effect may play an important role in
a long column packed bed or in the operation with a
high air mass flowrate.
In the solid-side resistance model [1] the equilibrium
condition on the surface of the silica gel particle is
initially evaluated. The result then provides the information in the analysis of the heat and mass transfer
in the pseudo-channels
of the column (Fig. 2). In
this work, a published isotherm equation [1, 61 for a
regular density silica gel is used and a corresponding
expression of the heat of adsorption is adopted [ 1,6].
Based on the above assumptions
and using several
defined nondimensional
variables and parameters, the
governing equations
can be rearranged
in the following forms [ 1, 61 : (i) Mass diffusion in the particle :
=~~(r*~D*g)
at*
initial condition
boundary
(1)
conditions
: g
= 0
,* = 0
-aw
dr*
(ii) Equilibrium
relationship
-Rn(ml,,-m,,e)
I* = 1
[ 1, 61 :
,s = P,,,,, -0.3784,
* P,,,(T,)
(2)
where
4. = 0.0078-0.05759~,+24.1655~:
- 124.478~:+204.226wf
(iii) Mass transfer
f [7-91 =
19.336 Re-0.66,
I l.478Re-.15,
~=~[Lc.ri(T:-TS+r3(m,.,-m,,)l
initial condition
(5)
: T,Yz*, t* = 0) = To/T,,.
J.-Y.
1176
SAN and
The values of cb and waveare integrated using a composite Simpsons scheme. In the analysis the value of
waveis used to evaluate the heat of adsorption and c,,.
3. EXPERIMENTAL
APPARATUS
AND PROCEDURE
In the experiment,
the length of the columns are
individually 9, 15 and 24 cm. The inner diameter of
the column is 5 cm. The air on the regeneration
side
is supplied by a one-horsepower
centrifugal blower
which provides the system with an air volumetric
flowrate of 6.6x lo-- m3 ss (Fig. 1). The air on
the adsorption
side is sucked by a two-horsepower
reciprocating
compressor which can be regulated to
result in a variable air mass flowrate. The cyclic switch
of the air is controlled by four electromagnetic
valves
with an inner diameter of 2 in. On the regeneration
side the air is heated by an electric heating coil with a
maximum output power of 1000 W. The heated air
temperature is set up at 65, 75 and 85C, respectively
by adjusting an on-off temperature controller. A programmable controller is used to set up the timing to
turn on or off electromagnetic
valves. Thus the cyclic
switch can be accurately executed.
In the work the inlet air for the adsorption
and
desorption
is directly extracted from the test room.
The air temperature
and humidity ratio in the room
is separately controlled by a heater and a humidifier.
A solid plate is placed between the two columns to
prevent the exhaust regenerative air being sucked into
the adsorption stream. Besides that, a fan slowly blows
the air to the system in order to achieve a good recirculation.
The relative humidities are measured by using five
thin film capacitance
type hygrometers
(I-100,
Rotronic Co.). The conversion of relative humidity to
humidity ratio is performed by using an ASHRAE
equation [lo]. One of the five hygrometers
is used
to measure the room temperature and humidity ; the
other four are individually installed at points 1, 2, 6,
9 in Fig 1. The hygrometers
have a response time
constant of 10 s and they are calibrated at a suitable
temperature
and relative humidity to ensure a 5%
error of the humidity ratio in the range of measurement. Several T-type thermocouples
are used to capture the transient response of the air temperature. The
diameter of the T-type thermocouples
is 0.1 mm and
the response time constant is measured to be less than
6 s. The error of the thermocouples
in the temperature
measurement is + 0.5C. The air volumetric flowrates
are measured by using a calibrated floating-ball tube
and a turbine flowmeter (EG & G Co.). The uncertainty of the former is 5% and the latter is within 2%.
The static pressure drops are measured by using a Utube filled with mercury and a less than 5% error is
assured. In the experiment a pressure drop range from
100 to 106 mm Hg on the desorption side and from
10 to 57 mm Hg on the adsorption side is measured.
The transient response of the temperature and relative
humidity are automatically
recorded by a HP-3852
G.-D.
JIANG
data acquisition system which assures a cyclic recording time period less than 5.5 s.
In this experiment
dyed silica gel particles with
diameters of 5 mm are used. The silica gel particles
have a density of 1008 kg mm3 which is classified as a
regular density silica gel. During the process of
adsorption the color of the particles will change from
blue to pink. Thus a good observation
of the water
vapor adsorption
in the transparent
column
is
achieved. However, for the sake of compactness it is
not shown in this paper.
In summary, the characteristic
factors affecting the
performance of the packed-bed system are : (i) regeneration temperature,
(ii) cyclic operating time period,
(iii) air mass flowrate ratio and inlet air velocity, (iv)
column length to diameter ratio, (v) inlet air humidity,
(vi) inlet air temperature,
(vii) characteristics
of silica
gel. In this work items (i)-(v) in the above list are
individually varied. The other factors are kept the
same in the measurement. Thus the time period setting
in the PLC, temperature controller of the heating coil,
inlet air humidity and air mass flowrate for adsorption
are required to be adjusted. The length to diameter
ratio is varied by replacing the two columns with
different lengths.
4. RESULTS
0.026
0.024
:r
0.022
;:
Eo.020 '2
2
I
aa Desorption
p-a
Adsorptmn
T= 4 min.
o
o
-0.016 c
0
:0.016 2
-0.014 >
S
:O.OlZ ~
dry air
Tc
75
65 "C
OC
65 'C
rO.O1O S
2
:0.006 r
-0.006
:0.004
*
a
*
*
1177
aa
0.3
9
a 0.2
L/d=
3
y= 0.52
T,,= 30 C
Y..= 0.0175
v= 1.9 m/s
0.1
CXXXX
n~an
MT
0.0,
0.0
,
1.0
TG= 65
Tc= 75
c = 65
2.b
?,
FIG.
4.
qoDnT,=
kg
H.O/kg
dry
Operating
L/d=
3
Regenerative
Air Mass
= 440 liters/min.
TB= 30 C
Y,.= 0.0175
Kg H,O/Kg
0.1
i
I
4.0
5.0
Time
6.0
air
C
C
C
3.0
65
7.0
6.0
9.0
o.oj
FIG.
air
0.5
Ratio)
0.4
0.3
y (MassFlonrate
Period(min.)
dry
r,,Irmr~IN1,7
I,,,,/,,
10.0
Flowrate
5.
i
0.45
(0.40
b
_j
y--o.43
T,=
30 OC
Tc= 65 C
V&=
1.6 m/s
0.35
3
0.30
FIG.
id6.0
9.0
10.0
16.0
17.0
1178
J.-Y.
0.40
SAN
and G.-D.
0.S
12 min.
1.1 m/s
cI
T.+=
V=
JIANG
i
0.36
0.4
lop,= 4 min.
V= 1.6 m/s
(2
----
0.36
0.3
_~_
*
~.__
*
a 0.34
a 0.2
i+=
v=
0.32
L/d=
3
y= 0.52
30 C
Y,,= 0.0175
kg
V&=
1.9 m/s
Tc= 65 C
r,.=
4 min.
1.9 m/s
y=o.43
T,=
30 C
Y,,=
0.0175
TG.= 65 C
H&I/kg
dry
c
_
air
0.30
0.0
0.0
-T?7-1
I.0
r ,
FIG. 7. Effect of column
length
__
MrhM
Numerical
Experimental
Data
Data
70
FIG. 8. Comparison
60
90
dry
air
0.1
* +h%z
kg
H&)/kg
Experimcnlnl
Numerical
2.0
3.0
Operating
FIG. 9. Comparison
Dab
Data
4.0
Time
I
5.0
I
6.0
Period
of humidity removal
ting time periods.
7.0
.o
6.09.0
(min.)
for various
opera-
Figure 8 represents a comparison between the measured data and analytical result. Since the operation
is with a moderate column length and air velocity, the
analytical result is obtained without consideration
of
the fluid friction effect. As can be seen the computer
modeling provides an acceptable agreement with the
experimental
data. The minor discrepancy might be
due to the inaccuracy in the measurement
and the
selected expression of the isotherm and heat of adsorption. Figure 9 represents a comparison of the humidity
removal between the measured data and analytical
result. As shown in the diagram
the predicted
humidity removal is higher than the measured data.
Figure 10 shows the prediction of the heat and mass
transfer in the packed bed with/without consideration
of the fluid friction effect. It is found that the fluid
friction effect plays an important role in the case with
a high Reynolds number or a long column. As can be
observed from Fig. 10 for the case with the Reynolds
number of 500, (r of 2.77 x lO--, C* of 0.2 and Ntu of
0XxX
Re=500;
Ntu=13.50;C=0.20;n=2.77xiO-~
s
Re=2000;Ntu=7.54;
C=O.O5;0=2.77~10~*
MRe=500;
Ntu=53.99;C=0.22;u=l.llxlO-
Iv,
I I ,
,
I
60
70
60
90
10
20
30
40
T (0
FIG.
10. Predictions
with/without
consideration
friction effect.
0.006
0.006
L
;
;&0.002
of the fluid
Modeling
system
1179
5. CONCLUSIONS
A simulation model for the heat and mass transfer
in the silica gel packed-bed system with the consideration of the fluid friction effect is developed. The
result is consistent with the measured data. The fluid
friction effect is related to the Reynolds number and
column length. For the operation with a high Reynolds number or a long column the Auid friction effect
becomes crucial. Without considering the fluid friction
effect the predicted value of the humidity removal
might be overestimated.
The packed-bed system is a device for the dehumidification of humid air without following the traditional dewpoint principle. In the experiment the
inlet air velocity of the column is relatively high, thus
the measured value of the humidity removal is low.
However, the general trend of the humidity removal is
obtained from the results. The higher the regeneration
temperature, the higher the humidity removal. An
optimum cycle time is obtained for the system with a
specific regeneration temperature and air mass flowrate ratio. The optimum cycle time corresponds to the
operation with a maximum humidity removal. The
higher the regeneration temperature, the shorter the
optimum cycle time. But the variation of the optimum
cycle time is very minor.
For the considered range of regeneration tem-