Ind. Eng. Chem. Res.

2009, 48, 5439–5444 5439

Oxygen Separation from Air by Four-Bed Pressure Swing Adsorption

Masoud Mofarahi*
Chemical Engineering Department, Persian Gulf UniVersity, Bushehr, Iran

Jafar Towfighi and Leila Fathi

Chemical Engineering Department, Tarbiat Modares UniVersity, P.O. Box. 14115-111, Tehran, Iran

A study on a four-bed seven-step pressure swing adsorption (PSA) using zeolite 5A was performed
experimentally and theoretically for separation of oxygen from air. In this process the steps of feed
pressurization, production, blowdown, purge, pressure equalization (two steps), and product pressurization
are included in a cycle. The effects of various operating parameters such as adsorption pressure, cycle time,
production rate, and purge rate on the product purity and recovery were investigated experimentally. Oxygen
purity and recovery both increase when adsorption pressure increases. For most of the experiments at the
highest purge rate, a higher purity of oxygen is obtained, but the recovery of oxygen is reduced. Oxygen
recovery increases as production rate increases while the purge flow rate remains constant. It was observed
that increasing the cycle time increases the performance of the process. An equilibrium based isothermal
model in conjunction with an LDF (linear driving force) approximation was employed to simulate process
performance. Comparison of the results obtained from the experiments and simulation results shows reasonable
agreement.

1. Introduction 2. Process Description

Pressure swing adsorption (PSA) processes are widely used A four-bed process producing continuously enriched oxygen
in industries for air and other gas separations. Oxygen and over 5A zeolite was studied. With the objective of producing
nitrogen are produced from atmospheric air by either of two high purity oxygen, a seven-step PSA was designed for air
methods depending upon the volume of production. For high- separation as follows:5
volume production, cryogenic distillation of liquefied air is
(1) Production (adsorption), AD
employed, whereas, for low to medium volume production, air
(2) Pressure equalization for depressurization, ED
separations by methods such as pressure swing adsorption are
(3) Blowdown, BD
found to be more economical. This mature technology is
(4) Purge, PG
economical for plants producing up to 250 t/d O2 using zeolites
of type A (5A) or X (13X-NaX, LiX, or LiLSX). During the (5) Pressure equalization for pressurization, EP
last 30 years, commercial applications for adsorptive O2 (6) Pressurization with product, RP
generation from ambient air using several molecular sieve (7) Pressurization with feed, PF
zeolites have been expanded.1,2 However, because air contains The sequence time of each step is given in Table 1
a small amount of Ar, which has physical properties similar to Figure 1 shows the first 50 s of the whole cycle, which
those of O2, the product generated from the zeolite bed typically represents the period of producing product from bed A1. At the
contains a substantial amount of Ar impurity. Therefore, in the start of a cycle, feed air is fed to bed 1 from one end, a product of
equilibrium separation process using zeolites, O2 purity is limited enriched oxygen is drawn from the other end at a high bed pressure,
to about 94%. For this reason, O2 produced by adsorption and, then, some product is used to purge bed 3. At the same time,
technology typically has 90-93% O2, 4-5% Ar, and 2-6% beds 2 and 4 are connected to equalize the pressure of the two
N2 and is generally used in many chemical processes such as beds. This operation step lasts 20 s. For the next 20 s, bed 1 keeps
the biological treatment of wastewater, steel industries, paper being fed with air, producing product and purging bed 3, and at
and pulp industries, and glass-melting furnaces.3,4 the same time, bed 2 is pressurized byproduct. At this step, bed 4
Although multiple adsorption beds are usually involved in is depressurized by blowdown. For the next 10 s, bed A keeps
the applications of PSA processes in industry, the detailed producing product and purging bed 3, bed 2 is pressurized with
operation information is usually retained by each company and feed air, and bed 4 is still at blowdown. The first 50 s are the period
seldom revealed. Most of the published studies, therefore, are of producing product from bed 1. Then, the product is produced
limited to the processes of single- and dual-bed systems. from bed 2 during the second 50 s with a phase change. The roles
In the present study a four-bed PSA process using a of beds 2, 3, 4, and 1 in the second 50 s replace those of beds 1,
commercial 5A zeolite is evaluated by experiments and a 2, 3, and 4 in the first 50 s, respectively. After this phase change,
theoretical model. The PSA performances under the same feed
rate, adsorption pressure, and cycle time were compared by Table 1. Sequence Time of PSA for Air Separation
variation of the purge flow and production rate. The adsorption
time (s) 20 20 10 20 20 10 20 20 10 20 20 10
dynamics at each step and the experimental PSA results were
predicted by the mathematical model. adsorber 1 AD ED BD PG EP RP PF
adsorber 2 EP RP PF AD ED BD PG
* To whom correspondence should be addressed. E-mail: Mofarahi@ adsorber 3 PG EP RP PF AD ED BD
pgu.ac.ir. Fax: +98 771 4540376. adsorber 4 ED BD PG EP RP PF AD

10.1021/ie801805k CCC: $40.75  2009 American Chemical Society

Published on Web 05/11/2009
5440 Ind. Eng. Chem. Res., Vol. 48, No. 11, 2009

intervals with a recorder (Logoscreen 50, Jumo Co.). During

measurement, the adsorption cell was immersed in a water bath
(MC 12, Julabo Tech.) maintained at (0.02 K via a refrigeration
circulator. Prior to each isotherm measurement, zelolite 5A particles
were regenerated in an oven overnight (300 °C for 6 h) under a
vacuum of less than 0.05 mbar.
3.3. PSA Setup. The layout of the PSA pilot system is shown
in Figure 2. It contains four main sections: (a) The feed section
is equipped with a mass flow controller which is connected to
a compressor. The gaseous feed is pressurized with a two-stage
compressor C1. (b) The PSA columns have an ID of 3 cm and
a length of 100 cm. They are filled with about 730 gr of 5A
molecular sieve adsorbent.
A calibrated three resistance temperature detector (RTD; Pt 100
Ω) is installed at one of the four columns (A1) in order to track
the temperature front. (c) The bed pressure and pressure drop were
measured by the pressure transducers and pressure gauge located
at each bed, feed, and product lines. The product and purge flow
rates were controlled by mass flow controllers. (d) The analytical
section is dedicated to controlling the cycle operation of the system
and online analysis of the composition of the mixture exiting the
system. The gas chromatograph system, a CP 3800 Varian with a
TCD detector, was used for the oxygen/nitrogen mixtures. The GC
is coupled with an automatic valve sample system including a
4-valve with an electric actuator and switching valve. The
chromatographic analysis is performed with the software provided
by Varian. The system was fully automated by a personal computer
with a developed control program, and all measurements including
pressure, temperature, and mass flow rate were indicated and
recorded on the computer. When the cyclic steady state was
reached, the collected data were integrated to obtain product
recovery, purity, etc.

4. Mathematical Model
The mathematical models describing the PSA process are
Figure 1. First 50 s of the operation schedule of the four-bed PSA.
developed considering the following assumptions: local equi-
Table 2. Physical Properties of Zeolite 5A5 librium is established, temperature is constant, axial dispersion
intrusion volume 0.24 cm3/g is negligible, plug flow conditions are assumed, pressure drop
specific surface area 572 m2/g along the column is negligible, and gas behavior is ideal. Mass
particle diameter 2-3 mm transfer is represented by the linear driving force (LDF)
bed density 670 kg/m3 approximation.6,7
particle porosity, εp 0.4 On the basis of these assumptions, the mass balance for each
bed inner diameter 3 cm
bed length 100 cm component of the mixture and the total mass balance are written
as follows:
beds 3 and 4 are used to produce enriched oxygen during the third
and fourth 50 s periods, respectively, and the overall cycle is
Overall mass balance
completed. n
∂qji
3. Experimental Section
∂C
∂t
+
∂(uC)
∂t
+ ∑ F ( 1 -ε ε ) ∂t
P )0 (1)
i

3.1. Materials. The zeolite 5A adsorbant was provided by

Linde in the form of pellets (1.6 mm extrudates). Some physical Component mass balance
2
properties of zeolite 5A are shown in Table 2. Adsorbents were ∂ Ci ∂(uci) ∂ci 1 - ε ∂qi
regenerated overnight at 300 °C before being used in measurements. -DL 2 +
∂z ∂z
+
∂t
+ Fp
ε ∂t (
)0 ) (2)
3.2. Equilibrium Measurement. The adsorption isotherms
of pure oxygen and nitrogen on zeolite 5A were measured at The initial conditions for fluid flow are the following:
four different temperatures (273, 283, 303, and 343 K) using a
high pressure volumetric apparatus up to 9.5 bar. u(z, 0) ) 0, q(z, 0) ) 0 (3)
This apparatus consists of an adsorption cell and a loading cell.
The total amount of gas introduced and recovered in the system The boundary conditions of each step are given in Table 3. The
was determined by appropriate pressure, temperature, and volume sorption rate into an adsorbent is described by the LDF model
measurements. The temperature in each cell was measured by with a single lumped mass transfer parameter.
K-type thermocouple operated within (0.01 K accuracy); the
pressure was measured with a pressure transducer (JUMO MIDAS ∂qji
) Ki(qj*i - qji) (4)
1001). Temperatures and pressures were recorded at constant time ∂t
 Ind. Eng. Chem. Res., Vol. 48, No. 11, 2009 5441

Figure 2. PSA pilot setup.

Table 3. Boundary Conditions for Different Steps

overall
step component balance balance
adsorption -DL ∂Ci/∂z|z)0+ ) u|z)0(Ci|z)0 - Ci|z)0+) u|z)0 ) uOH
∂Ci/∂z|z)L ) 0
equalization to ∂Ci/∂z|z)0 ) 0, ∂Ci/∂z|z)L ) 0 u|z)0 ) 0
depressurization
blowdown ∂Ci/∂z|z)0 ) 0, ∂Ci/∂z|z)L ) 0 u|z)L ) 0
purge ∂Ci/∂z|z)0 ) 0 u|z)L ) uOL
-DL ∂Ci/∂z|z)L- ) u|z)L(Ci|z)L - Ci|z)L-)
equalization to -DL ∂Ci/∂z|z)0+ ) u|z)0(Ci|z)0 - Ci|z)0+) u|z)L ) 0
pressurization ∂Ci/∂z|z)L ) 0
product pressurization -DL ∂Ci/∂z|z)0+ ) u|z)0(Ci|z)0 - Ci|z)0+) u|z)L ) 0
∂Ci/∂z|z)L ) 0
feed pressurization -DL ∂Ci/∂z|z)0+ ) u|z)0(Ci|z)0 - Ci|z)0+) u|z)L ) 0
∂Ci/∂z|z)L ) 0

The individual mass transfer resistance for a biporous adsorbent,

in the case of negligible micropore resistance,8 is related to the
LDF mass transfer coefficient by the following equation:

( )( )
Figure 3. Isotherms of pure oxygen on zeolite 5A.
1 RP RP2 ∂qj*i
) + (5) All the variables and parameters were made dimensionless. The
Ki 3kfi 15εPDei ∂ci
above set of partial differential equations were then converted
Using the values kfi and Dei that were determined for the to first-order ordinary differential equations (ODEs) of dimen-
breakthrough and desorption runs, the mass transfer coefficient sionless time by discretizing all the spatial variables (dimension-
(Ki) was determined from eq 5. less forms) using the orthogonal collocation method. A 12-point
The adsorption equilibrium of oxygen and nitrogen was well- collocation was found to be a suitable compromise between
represented by a six-parameter Langmuir-Freuendlich isotherm: minimizing the magnitude of oscillations in the solution of the
derivative equations and the overall computation time. Starting
qmibiPin from initial bed conditions, normally in equilibrium with the
qi ) n
(6) feed mixture at either the high or the low operating pressure,
1+ ∑bP j i
nj the system of ODEs were solved using Matlab software.
j
5. Results and Discussion
Where, the isotherm parameters are functions of temperature:
5.1. Adsorption Isotherms and Heat of Adsorption. Fi-
qmi ) ki + k2T, bi ) k3 exp(k4 /T), ni ) k5 + k6 /T gures 3 and 4 show the adsorption isotherms of oxygen and
(7) nitrogen on zeolite 5A at 273, 283, 303, and 343 K and pressure
5442 Ind. Eng. Chem. Res., Vol. 48, No. 11, 2009

Figure 4. Isotherms of pure nitrogen on zeolite 5A. Figure 6. Steady-state experimental temperature profiles at the top of the
column (PH ) 5 bar and cycle time ) 200 s).
Table 4. Parameters of Extended Langmuir-Freundlich Isotherm
and Heats of Adsorption
parameter oxygen nitrogen unit
k1 14.31 10.95 mol/kg
k2 -0.031 -0.028 mol/(kg K)
k3 0.0005 0.0016 1/bar
k4 1126.2 1304.3 K
k5 0.52 -0.74
k6 148.2 521.4 K
LDF coefficient Ki 0.0098 0.0032 1/s

0-9.5 bar. As expected, the results showed high selectivity for

nitrogen over oxygen at the range of experimental conditions. Figure 7. Steady-state simulated velocity profile at the exit end, for one
The adsorption isotherm parameters of extended Langmuir- column and one entire cycle.
Freundlich model and LDF coefficients of oxygen and nitrogen
are also shown in Table 4. Table 5. Operating Conditions of Experimental Runs for Oxygen
Separation
5.2. Experimental and Simulation Results of Sample
Runs. On the basis of the experimental and simulation results PH/PL (bar) cycle time (s) P/F ratio product (L/min)
for a cycle time of 200 s and pressure of 5 bar as a sample of 4/1 100/140/200 0.1 0.25/0.5/0.75/1/1.5
test runs, the general characteristics of the process are discussed. 0.16 0.25/0.5/0.75/1/1.5
Figure 5 shows the pressure history of the first column during 0.2 0.25/0.5/0.75/1/1.5
one cycle. As shown in this figure, the adsorption and equaliza- 0.24 0.25/0.5/0.75/1/1.5
0.3 0.25/0.5/0.75/1/1.5
tion pressure are 5 and 2.3 bar, respectively. At all steps except 5/1 100/140/200 0.1 0.25/0.5/0.75/1/1.5
the BD step, the pressure changes almost linearly. Temperature 0.16 0.25/0.5/0.75/1/1.5
profiles in the top of the first column for the cycle time of 200 s 0.2 0.25/0.5/0.75/1/1.5
and pressure of 5 bar are given in Figure 6. The temperature in 0.24 0.25/0.5/0.75/1/1.5
the column has an increase in pressurization and adsorption, 0.3 0.25/0.5/0.75/1/1.5
6/1 100/140/200 0.1 0.25/0.5/0.75/1/1.5
and a decrease in the depressurization and blowdown steps is 0.16 0.25/0.5/0.75/1/1.5
observed. This figure shows that the temperature variation in 0.2 0.25/0.5/0.75/1/1.5
the top of the column is less than 4 °C during whole cycle time. 0.24 0.25/0.5/0.75/1/1.5
0.3 0.25/0.5/0.75/1/1.5
Considering the pressure drop and temperature variations to
be negligible, a typical simulated velocity profile, based on
boundary conditions of Table 4, will be similar to Figure 7.
This figure shows a simulated velocity profile for the cycle time
of 200 s and pressure of 5 bar. As mentioned in this figure,
velocity values at the end of the column (Z ) L) are drawn. As
shown in Figure 7, the velocity changes from positive to
negative values according to flow direction. The first step is
the adsorption step, and the other steps are shown in the order
as indicated in Figure 7.
5.3. Pressure Swing Adsorption Experiments. A group of
PSA experiments were carried out following the adsorption
sequence shown in Figure 1. The experimental operating
conditions are summarized in Table 5. According to the design
Figure 5. Steady-state simulated and experimental pressure history at the specification, the feed rate was fixed at 18-20 L/min, and the
exit end, for one column and one whole cycle. cycle time was considered to be a variable parameter to obtain

high-purity oxygen and determine the performance of the
process. Three different pressures for adsorption steps are
explored in this study. They include three cycle times each of
five purge to feed ratios and five production rates.
5.3.1. Effect of Adsorption Pressure. The study of adsorp-
tion pressure influence on oxygen purity and recovery was
carried out by varying the adsorption pressure at the same
cycle time and production rate. The general conclusion is
that, within our experimental range, oxygen purity and
recovery both increase when adsorption pressure increases.
Figure 8 shows the effect of adsorption pressure on process
performance for a cycle time of 150 s, production rate of
0.75 L/min, and different purge to feed ratios. For each
pressure, the oxygen purity increases but oxygen recovery
decreases as purge to feed ratio increases. Also both oxygen
purity and oxygen recovery increases as pressure increases.
The reason is that when pressure increases, more nitrogen
will be adsorbed by columns. Therefore, more oxygen is
produced in the adsorption step and more nitrogen is
exhausted during desorption steps. The reasonable agreement
between experimental date and simulated results is shown
in Figure 8. At a pressure of 6 bar the oxygen purity increases
to about 95%, while the oxygen recovery decreases to around
23%. Chou et al.,6 Mendes et al.,9 Farooq et al.,10 and
Ruthven et al.11 observed the same trend for both purity and
recovery.
5.3.2. Effect of Purge Flow. The purging ratio, P/F, was
defined as the ratio of the purge flow to feed flow rate of a
cycle. Purging is an important step of PSA operation. Insufficient
purging leads to insufficient regeneration of adsorbents and
further leads to a decrease of product purity. Overpurging does
not help improve the product purity but rather leads to a decrease
Figure 8. Effect of pressure on oxygen purity and recovery.
Figure 9. Effect of purge to feed ratio on oxygen recovery and purity.
Ind. Eng. Chem. Res., Vol. 48, No. 11, 2009 5443
Figure 10. Effect of production rate on oxygen recovery and purity.
in product recovery. The recovery and purity of oxygen at a
certain feed flow rate depends on the amount of gas used to
purge the column. In this study for all groups of test runs, the
effect of the purge flow was considered. For most of them at
the highest purge rate, a higher purity of oxygen is obtained,
but the recovery of the oxygen is reduced. The effect of the
P/F ratio on the purity and recovery of oxygen at a pressure of
5 bar, cycle time of 200 s, and production rate of 0.25 L/min is
presented in Figure 9. The simulated values of oxygen purity
and recovery as a function of P/F ratio also corresponds well
with the experimental data (Figure 9). The function of the purge
step is to clean the bed by purging the column with part of the
product stream produced during adsorption step. The cleaner
the bed becomes, the higher the purity of the product. The bed
can be more thoroughly cleaned by simply using more purge
gas. On the other hand, loss of oxygen increases as purge flow
increases. Hence, oxygen recovery is reduced if more purge flow
is used. Chou et al.6 and Mendes et al.8 observed the same trend
for both purity and recovery.
5.3.3. Effect of Production Rate. The influence of the
product flow rate on product purity and recovery was discovered
by varying the production rate at a certain feed rate and purge
to feed ratio. Figure 10 shows the effect of production rate on
process performance for a cycle time of 200 s, 6 bar pressure,
and purge to feed ratio of 0.24. When the purge flow remains
constant, oxygen recovery increases as the production rate
increases, because in this case loss of oxygen in tail gas is
reduced. On the other hand, increasing production rate increases
the nitrogen amount in purging gas and therefore oxygen purity
will be reduced.
4.3.4. Effect of Cycle Time. In the PSA process, adsorption
takes place in both the pressurization and adsorption steps, and
also, desorption and cleaning take place in the depressurization
and purge steps. The choice of each step time has a great
effective on process performance. Sufficient cycle time should
be provided, so that the desired product purity and recovery
can be achieved. The effect of cycle time on process purity,
recovery, and productivity are shown in Figures 11 and 12. The
main feature of these plots is that, at low cycle time (100 s), no
high oxygen purity, recovery, and productivity can be obtained.
Among the three cycle times (100, 150, and 200 s) tested, Figure
11 shows that increasing the cycle time increases the perfor-
mance of the process. The duration of the cycle time in Figure
11 for a short cycle is not large enough to equilibrate the particle
with its environment, and thus, depressurization and pressuriza-
tion after the first cycle time are initialized from a nonequili-
brated state. This leads to a decreased process performance. In
Figure 12, productivity lines for cycle time of 150 and 200 s
cross each other at about 90% purity. This means that at low
5444 Ind. Eng. Chem. Res., Vol. 48, No. 11, 2009
Figure 11. Effect of cycle time on oxygen purity and recovery.
Figure 12. Effect of cycle time on oxygen purity and productivity.
purity (less than 90%) process productivity for a cycle time of
150 s is higher than that for cycle time of 200 s. The comparison
between experimental data with results obtained from simulation
can be seen in Figures 11 and 12. Model predictions in terms
of cycle time are quite far from experimental results especially
for cycles of 150 and 200 s.
6. Conclusions
Oxygen production from air in a four-bed pressure swing
adsorption setup on a commercial type zeolite 5A has been
studied both experimentally and numerically over a wide range
of operating conditions. A PSA unit using seven steps allowed
the production of high-purity oxygen directly from air. Influence
of adsorption pressure, purge to feed ratio, production rate, and
cycle time on process performance were investigated experi-
mentally and were compared with simulation results. For our
experimental range, oxygen recovery and purity increases when
adsorption pressure increases from 4 to 6 bar. For a cycle time
of 200 s, the P/F ratio between 0.2 and 0.3 guarantees purities
higher than 90%, while pressure and production rate are kept
at 5 bar and 0.25 L/min, respectively. The cycle time has a
considerable effect on the oxygen purity and recovery. The
performance of the PSA process investigated was better at a
higher cycle time. Furthermore, the oxygen recovery increases
and oxygen purity decreases when the production rate increases.
Although a considerable agreement has been found between the
experimental data and simulated results of oxygen purity and
recovery for various operating parameters, results of experiments
and prediction of process performance have shown a slightly
different behavior especially for different cycle times.
Acknowledgment
The authors are thankful to Linde AG, Linde Engineering
Division, Germany, and Gostaresh Co. Ltd., Iran, for supplying
zeolite particles.
Notation
bi ) equilibrium parameter for Langmuir-Freuendlich model
C ) gas phase concentration (mol/m3)
ci ) concentration of ith component in the gas phase (mol/m3)
Dei ) effective macropore diffiusivity (cm2/s)
dp ) particle diameter (cm)
k ) parameter for Langmuir-Freuendlich model
K ) proportionality parameter for LDF model (1/s)
kfi ) effective mass transfer coefficient of ith component
n ) parameter for Langmuir-Freuendlich model
P ) pressure (bar)
q*i ) equilibrium amount adsorbed (mol/kg)
qi ) amount adsorbed of ith component (mol/kg)
qmi ) saturation loading (mol/kg)
Rp ) radius of the particle (cm)
t ) time (s)
u ) interstitial velocity (m/s)
z ) distance along the length of the column (m)
Greek Letter
ε ) bed void fraction
εp ) particle porosity
Fp ) density of particles (kg/m3)
Literature Cited
(1) Rege, S. U.; Yang, R. T.; Buzanowski, M. A. Sorbents for air
prepurification in air separation. Chem. Eng. Sci. 2000, 55 (21), 4827.
(2) Jee, J. G.; Lee, S. J.; Kim, M. B.; Lee, C, H. Three-Bed PVSA
Process for High-Purity O2 Generation from Ambient Air. AIChE J. 2005,
51 (11), 2988.
(3) Yang, R. T. Gas Separation by Adsorption Processes; Butterworth:
Boston, MA, 1987.
(4) Ruthven, D. M.; Farooq, S.; Knaebel, K. S. Pressure Swing
Adsorption; VCH Publishers: New York, 1994.
(5) Silva, J. A. C.; Rodrigues, A. E. Sorption and Diffusion of n-Pentane
in Pellets of 5A Zeolite. Ind. Eng. Chem. Res. 1997, 36, 493.
(6) Chou, C. T.; Huang, W. C. Simulation of a Four Bed Pressure Swing
Adsorption Process for Oxygen Enrichment. Ind. Chem. Eng. Res. 1994,
33, 1250.
(7) Jain, S.; Moharir, A. S.; Li, P.; Wozny, G. Heuristic Design of
Pressure Swing Adsorption: Preliminary Study. Sep. Purif. Technol. 2003,
33, 25.
(8) Iyuke, S. E.; Daud, W. R. W.; Mohamad, A. B.; Kadhum, A. A. H.;
Fisal, Z.; Shariff, A. M. Application of Sn-activated carbon in pressure
swing adsorption for purification of H2. Chem. Eng. Sci. 2000, 55 (20),
4745.
(9) Raghavan, N. S.; Ruthven, D. M. Numerical simulation of a fixed-
bed adsorption column by the method of orthogonal collocation. AIChE J.
1983, 29 (6), 922.
(10) Mendes, A. M. M.; Costa, C. A. V.; Rodrigues, A. E. Oxygen
Separation from Air by PSA: Modelling and Experimental Results Part I:
Isothermal Operation. Sep. Purif. Technol. 2001, 24, 173.
(11) Farooq, S.; Ruthven, D. M.; Boniface, H. A. Numerical simulation
of a pressure swing adsorption oxygen unit. Chem. Eng. Sci. 1989, 44, 2809.
(12) Ruthven, D. M.; Farooq, S. Air separation by pressure swing
adsorption. Gas Sep. Purif. 1990, 4, 141.
ReceiVed for reView November 25, 2008
ReVised manuscript receiVed April 5, 2009
Accepted April 29, 2009
IE801805K