Académique Documents
Professionnel Documents
Culture Documents
A
R
t
[-]
[q
Literature
= distr.but.on
(Superscript)
S
= solid
coefficent
,,
,,
Cited
1) Abu Elamayen,
2)
Chlopin,
V.:
3) Doener,H.A.
4) Feibush,
1605
(1958).
M. A.: J. Inorg.
Z. Anorg.
Allgem.
andW. M. Hoskins:
(1925).
143,
97 (1925).
662
5) Kitamura,
M. and T. Nakai: Kagaku Kogaku Ronbunshu, 8,
442 (1982).
phase
[-]
= liquid
,,
(Presented
at
the
47th
Talanta,
Annual
12, 35 (1965).
Meeting
of The
Society
of
phase
AIR DRYING
BY PRESSURE
SWING
ADSORPTION
silica gel as adsorbent. Measurementswere also madeof batch adsorption of water vapor on silica gel to obtain
input data for the authors' PSAsimulation program. Experimental PSAresults were compared with this computer
simulation. Good agreement was obtained and the trend of experimental results was well explained by simulation
for both isothermal and non-isothermal cases. In the simulation, however, mass transfer coefficients two or three
times larger than those estimated conventionally from batch measurement were used to obtain good coincidence.
These larger masstransfer coefficients were determined by a recently developed methodfor short adsorption and
desorption cycle.5} The simulation method is useful in predicting the performance of air drying PSA.
Introduction
Pressure swing adsorption
(PSA) processes are
cyclic processes for the separation of gaseous mix-
removal of
during
change. The
16
NO.
1983
swing processes
are widely
used indus-
and pro-
tion for the removal of one adsorbing impure component, such as air drying, was simulated numerically
from start-up until steady cyclic modeis established.
293
though
equilibrium
Experimental
the
vapor
pressure,
Ps, at
308K,
323K
and
353K.
controlling.
=i-E
1+f
= Wsq0/V(P0/RT)
1.2 PSAoperation
The original pellets of silica gel were crushed and
screened to yield the desired size (14 to 20 Tyler mesh)
of small particles. This fraction was packed in the
Pyrex glass tube as the adsorbent columnof PSA.
Twotypes of column, thick and thin, were used. A
schematic diagram of the apparatus for PSAexperiments is shown in Fig. 1. Adjustable check valves keep
the columnpressure at a fixed value in the adsorption
cycle. A rotameter was used to measure both product
and reflux gas flow rates. Dewpoints offeed, product
and purge were measured by a Panametric hygrometer. The operation mode of PSA cycle is as described
in the article1)
by Chihara
and Suzuki.
Experimental
conditions are listed in Tables l(a) and
(b).
The dew point of product gas was continually
detected by hygrometer during each run to check
whether cyclic operation had reached steady state.
(1)
clic operation
were continued for more than ten
days until steady operation
was observed. Then
operational
conditions
were changed to reach another steady state. In the initial several runs, dew
Po
LMi.
1+1
Mm
tang
3+^n2
T = DJ/Rp2
294
(2)
boundary
location.
After
getting
steady
p_=1
pink-blue
6^+l)e--2t
Mt
and
nondimensional
0.75 and 1.0.
longitudinal
locations,
0, 0.25,
0.5,
vapor on original
JOURNAL
silica
gel pellets.
OF CHEMICAL
The solid
ENGINEERING
line in
OF JAPAN
Fig.
1.
PSA dryer,
experimental
apparatus.
1, Silica
gel
gel.
=
=
=
=
=
bed
0.127or0.15m
0.025m
0.40
0.73xl03kg/m3
1.22xl03kg/m3
Cps = 9.21xlO2J/kg-K
= 1.00xl03J/kg-K
Q = 6.3xl04J/mol
Feed
gas
vH = 0.058to0.061m/s
PH
c0
To
^
Purge
=
=
=
-
4.44x105Pa
0.48to0.71mol/m3
287to293K
600or1800s
gas
5x
bed
= 0.3m
= 0.012m
eter
Feed
gas
vH = 0.263m/s
Purge
gas
vL = a"vHm/s
a
?2
= 1.25to2.1
= 600or300s
of 2.2nm.
Experimental
diffusivity
as surface
To = 298K
tx
= 600or300s
= 1.01xl05Pa
104J/mol.
PH = 5.07x105Pa
c0 = 1.067mol/m3
i>L
lO4 to 6.3x
gel.
vL =ol-vhm/s
PL = 1.01xl05Pa
a
= 0.75to1.23
t2
= 600or1800s
Adsorbent
of H2O on silica
of
10-2(P/Ps)1/1
[mol/g]
(6)
NO.
at several
conditions
by.
changing volumetric purge to feed ratio, a, and halfcycle time, tx. These three kinds of information (dew
points, boundary locations and temperature changes)
were compared with three
simulated
profiles
(gas-
c=P/RT
VOL
were obtained
1983
ture) to check the suitability of the simulation method. Figures 4(a), (b), (c) and (d) show examples of
295
tally
obtained product concentration (dew point, D.P.
exp), boundary location (the arrow, BL exp, points
the location in the axial direction)
and two temperature profiles at the ends of both adsorption and
desorption cycles are shown. f=0 is the inlet at highpressure flow and the outlet at low-pressure purge
profile
(RUN A).
ratio,
t1/t0,
almost constant
as shown in
Figs. 4(a), (b) and (c) and other conditions were fixed
(Table l(a)).
As a decreases,
profile
(RUNB).
product
concentration
becomes
worse from D.P.=201 K to 253K, and the temperature peak at the end of adsorption cycle and the
boundary location shift to the outlet at high-pressure
flow. Figure 4(d) (RUN D) can be compared with
Fig. 4(a) (RUN A), since only t1 was changed (to
1800s) in RUND from RUNA. As a result, dew
point rises, and temperature peak and boundary
location shift to the outlet at PH in RUND.
Figure 5 shows an example of the temperature
changes at various positions in the axial direction, as
obtained
in
RUNA.
Dew points of product gas and locations of
boundary between pink zone and blue zone at
steady operation were experimentally obtained at
several conditions by changing volumetric purge to
feed ratio, a, and half-cycle time, tl9 for the thin column experiments also. Figures 6 and 7 show the de-
pendence of product dew points and boundary loFig. 4(c). Example of temperature
profile
(RUN C).
half-cycle
time lowers
the boundary
temperature
profiles,
location
boundary
loca-
8cal)wt%respectively)
and product concentrations
(D.P.
are shown in corresponding
figures (Figs. 4(a),
profile
(RUND).
Input
data for
OF CHEMICAL
ENGINEERING
OF JAPAN
Ksav
= \5yDjRp2
(7)
Temperature
change at various
positions
in axial
thus
obtained
(Ksav=0.$2kg/m3-s
for 600s
and product dew point. Also, the trend of the experimental data with change of operational
condition,
ratio,
tjto,
is well
explained
by the
simulation.
column experi-
(Eq.
(6))
and overall
mass transfer
coefficient, Ksav. Equation (6) was used for computation just as shown. As for mass transfer coefficient, Ksav determined by Eq. (7) was first used for
(Ksav=0.96kg/m3-s
for 300s)
determined
for
isotherm
(Eq.
(6))
16
NO.4
1983
are two to three times larger than Ksav by Eq. (7). The
simulation
performance of air drying PSAwith the new estimation method for Ksav.5) Further improvement of
the simulation might be possible by improving the
treatment of heat transfer and the estimation of heat
transfer
Acknowledgment
The authors express their gratitude to Mr. Ryuichi Shimoyoshi,
on leave from Kanagawa University, for his assistance in the
experimental part of this work.
Nomenclature
In these simulations,
isosteric
heat of adsorption,
g, was assumed to be zero for thin column experi-
by inputting
the exact heat of adsorption and estimated complex heat transfer coefficient through the
wall, h0, with column diameter to the computer.
three
circles)
(a=0.17,
^=600s),
which
shows
Mt
M^
p
Po
PH
PL
Ps
P^
with
estimated
which corat
PH).4)
air drying
experiments
by pressure
diameter
column) and non-isothermal (large-diameter
column) cases. In the simulation,
however, mass
transfer
than
good coincidence.
coefficient.
These larger
q0
qn
q*
q0*
R
Rp
r
T
To
t
r0
tt
t2
= isosteric
heat of adsorption
[J/mol]
[kg/kg]
[-]
[kg/kg]
[kg/kg]
[J/mol"K]
[m]
[m]
[K]
[K]
[s]
[s]
[s]
[s]
V
- volume of vessel
[m3]
vH = interstitial
velocity at high pressure flow [m/s]
vL = interstitial
velocity at low pressure flow [m/s]
Ws = weight of adsorbent
[kg]
z
= position in bed
[m]
2
= z/L
[-]
a
y
8pg
ps
X
z
OF CHEMICAL
ENGINEERING
OF JAPAN
=to/{(pgCpg+yCps)r/2ho}
= defined
by Eq. (5)
(1978).
5) Nakao,
S. and M. Suzuki: J. Chem. Eng. Japan,
(1983).
Literature Cited
1) Chihara,
(1983).
K. and M. Suzuki:
134,
25 (1973).
6)
Skarstrom,
Science,"
COAGULATION
IN A VISCOUS
Developments
(1972).
7) Shendalman,
1449
TURBULENT
DISPERSED
C. W.: "Recent
(1972).
L. H. and J. E. Mitchell:
16, 114
in Separation
Cleveland,
Ohio
27,
OF PARTICLES
FLUID
Introduction
The coagulation
of colloidal
particles
is found in
theory developed by Saffman and Turner11} is somewhat more rigorous. However, it is assumed in their
analysis that particles encounter each other without
interacting
hydrodynamically;
that
medium. If there
or the interparticle
attractive force acts such that it
cancels the effect of the hydrodynamic interaction.
But it is unlikely that the theory is applicable
to the
A theoretical
exists
no repulsive
treatment
surface force
of turbulent
coagulation
September
27, 1982.
Correspondence
to a diffusion problem
diffusion coefficient. A
concerning
this
article
should
be