Separation and Purification Technology 42 (2005) 123–127

The experimental study on the performance of a small-scale

oxygen concentration by PSA
Z. Yuwen∗ , W. Yuyuan, G. Jianying, Z. Jilin
Cryogenic Engineering Institute, Xi’an Jiaotong University, Xi’an 710049, China

Received in revised form 14 June 2004; accepted 3 July 2004

Abstract

A small-scale adjustable rich oxygen concentrator by pressure swing adsorption (PSA) was designed and manufactured to study influences
of nozzle size and purge quantity on characteristics of the production oxygen, such as its pressure, purity and volume flux. This device could
reach above 90% and up to 97% in volume. The flux of rich oxygen product could be adjusted intelligently in the range of 0.5 l/min–3.5 l/min.
The device can operate continuously and stably over two years and is quite usable in hospital, domestic application, oxygen bar and other
places in shortage of oxygen.
© 2004 Elsevier B.V. All rights reserved.

Keywords: Small-scale concentrator; Oxygen; Adsorption

1. Introduction adsorbed by adsorbent, while it is desorbed in the lower pres-

Originally, pressure swing adsorption (PSA) method was
based on so-called Skarstrome cycle (1960). The technology
was initially used in industrial application like air-drying and
purifying hydrogen. It was firstly used to produce oxygen
from air in 1970. There are many other methods of produc-
ing oxygen from air by industrial process such as cryogenic
methods, membrane permeation, chemical absorption and so
on. Compared with the above methods, PSA has a lot of mer-
its such as small investment, low energy consumption, safety,
convenience and automation. When very high purity of oxy-
gen is not required, PSA is especially suitable to be used. Such
situations wildly exist at home and in aquaculture where the
oxygen-rich air is required. With its great advantages, PSA is
also used in many fields like iron manufacture, refinery, glass
manufacture [1], medical industrial and so on.
The basic principle of producing oxygen from air by PSA
is based on the fact that the adsorbed nitrogen quantity of Qn
by adsorbents at higher pressure is larger than it at lower pres-
sure [2,3]. That is, in the higher pressure, nitrogen in air is
∗ Corresponding author.
E-mail address: ywzhang@mail.xjtu.edu.cn (Z. Yuwen).
sure with the absorber being regenerated. With pressurization
and depressurization operating alternately, oxygen in the air
will be separated from air. The operation can be performed as
the following processes [4,5]. Firstly, the pressurized, dry and
filtered feeding air flows into the adsorption column which is
filled with 5A zeolite molecular sieves (ZMS). Then most of
nitrogen will be adsorbed by 5A ZMS and most of oxygen
flows through the column. A small-scale oxygen concentra-
tor is constructed to investigate the effects of the operating
parameters on the performance of the device so as to optimize
them [4,5].
2. Experimental arrangements
2.1. The description of the experimental system and the
flow diagram
The experiments are performed to investigate how to get
either large flow flux of production oxygen or high purity of it
by adjusting HPAP and purge quantity of Qp . The schematic
drawing of the experimental system is shown in Fig. 1. Firstly,

1383-5866/$ – see front matter © 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.seppur.2004.07.001
124 Z. Yuwen et al. / Separation and Purification Technology 42 (2005) 123–127

be opened and valve V2 be closed with V3 closed and V4

Nomenclature opened. The high-pressure air flows into column A via V1
and the component of nitrogen in the air is adsorbed in col-
Qo the flow flux of production oxygen (L/min) (t umn A. Then, one part of the oxygen-rich air is supplied to
= 25 ◦ C, P = 0.05 MPa) the consumer as product oxygen of Qo via 15. Another part
Qp the flow flux of purge quantity (L/min) (t = of the oxygen-rich air will act as the purge gas of Qp to enter
25 ◦ C, P = 0.05 MPa) the column B. At last, the purge gas is discharged into at-
Qe The flow flux of oxygen-enriched air exiting mosphere via valve V4 and via muffler 16. In next step, the
out of adsorption column (L/min) (t = 25 ◦ C; column B will be in adsorbing and the column A will be in
P, corresponding to exiting pressure out of ad- desorbing with the similar process. Thus, the product oxy-
sorption column, MPa) gen of Qo will be supplied continuously by the small device.
Qa The flow flux of air feeding into adsorption The electric magnetic valve of V1–V4 is controlled by the
column (L/min) (t = 25 ◦ C; P, corresponding controller 11. There are three nozzles in the system. The noz-
to exiting pressure out of adsorption column, zle P1 is located at exit port of adsorption column. One part
MPa) of the discharge oxygen gas flows through P1 and flow-rate
Qn The flow flux of adsorbed nitrogen in adsorp- meter and enters into column B through nozzle P2. Another
tion column (L/min) (t = 25 ◦ C; P, correspond- part of the discharge oxygen gas flows through nozzle P3
ing to exiting pressure out of adsorption col- and single direct valve of 9 as production oxygen gas. All
umn, MPa) of the nozzles have same diameter. The flow flux and pu-
P The high plateau adsorption pressure (HPAP) rity of the production oxygen are measured by 14 and 15,
(MPa) respectively.

2.2. The experimental procedures

the feed air (atmospheric pressure) is passed through the air
filter 1 to remove dust and other solid substances, then, en-
ters the compressor 2 to be pressed to the required HPAP. It
is further cooled to the ambient temperature in the air cooler
3 and gets rid of the water, oil and CO2 in the filter 4. Finally,
it flows into the adsorption columns 10 to produce oxygen-
rich air. Two columns of A and B operate alternately to be
controlled by the controlling system 11. While A is adsorb-
ing, B is desorbing at the same time, then the valve V1 will
The alternate frequency (adsorption period) of V1–V4 and
the nozzle size are kept constant all the time in each turn of
experiment. In this way, the HPAP of the turn of experiment
is determined. In each turn of experiment, the volume flux of
the purge gas and the purity of the production oxygen will
be changed by adjusting the volume flux of the production
oxygen. In the next turn of the experiment, the nozzle size will
be changed, so as to increase or decrease HPAP in adsorption

Fig. 1. Experimental system: (1) dust filter; (2) compressor; (3) air cooler; (4) oil–water and CO2 filter; (5) pressure meter for air; (6) single direct valve for
oxygen; (7) pressure meter for oxygen; (8) nozzle components (P1, P2 and P3 with flow-rate meter); (9) single direct valve for oxygen; (10) adsorption column;
(11) control system; (12) filter for oxygen; (13) adjusting valve for oxygen pressure; (14) flow-rate meter for oxygen production; (15) concentration meter and
(16) muffler.
 Z. Yuwen et al. / Separation and Purification Technology 42 (2005) 123–127 125

column. The above experimental procedure will be repeated 2.5 mm and 2 mm, respectively. By this new packing ar-
at another HPAP of the column. rangement, the flow velocity distribution has been improved
greatly.
2.3. The testing condition

The dimension of the adsorption column (two): 3. Experimental results and discussions
82 mm (diameter), 650 mm (height); adsorption period,
10 s–30 s (adjustably); void fraction: 0.32–0.47; temper- 3.1. Attainable parameters of the device
ature: ambient room temperature; experimental pressure,
0.11 MPa–0.28 MPa; adsorbent, 5A molecular sieve (d = From the results of the experiment, in Fig. 3, the highest
1.6 mm–2.5 mm). volume flow flux of the production oxygen is 3.0 l/min with
the HPAP of 0.17 MPa. The flux of feed air at the outlet of
2.4. New packing arrangement for molecular sieves the compressor is not more than 3.2 m3 /h with the HPAP of
0.3 MPa. It is found that the purity of oxygen can be above
An important improvement has been made in packing ar- 95% with the small device. All the results shown in Fig. 3
rangement for molecular sieve in two adsorption columns. are obtained in the condition that the purity of oxygen is
The traditional packing arrangements are uniform packing maintained above 90%.
by using only one kind of molecular sieve. Namely, the same
size of diameter of one kind of molecular sieve is uniformly 3.2. The influence of HPAP on Qe
distributed in the adsorption column. The disadvantage of
this uniform packing is insufficient usage of molecular sieve The discharge flux out of adsorption column of Qe , that is
for whole column. It is caused by the mal-distribution of total flux of high purity oxygen, is equal to the flux of pro-
flow velocity in the column. The maximum velocity lies in duction oxygen of Qo plus the flux of purge quantity of Qp .
the central core of the column. While the velocity outside The Qe is related to the HPAP as shown in Fig. 3. The rela-
the central core, it is much smaller. Therefore, the molecular tionship between the discharge flux of Qe and HPAP shows
sieve in central core is firstly saturated by adsorbing enough
nitrogen and thus firstly penetrated by feed air. However, the
molecular sieve outside the central core is still not saturated,
and is not sufficiently used.
In order to use the molecular sieve sufficiently, the flow
velocity in the central core must be reduced in comparison
with the traditional packing. The new packing arrangement
is called multi-layer arrangement is called multi-layer and
multi-section packing method. The small diameter of molec-
ular sieves is packed in the central core while the larger di-
ameter of sieve lies in the outer layers outside of the central
core.
According to previous experimental results, the arrange-
ment of layers will be adjusted so as to obtain appropriate
flow resistance and flow velocity distribution at last.
In our experiment, there are three layers in radial direction
and two sections in axial direction of the column, which is
shown in Fig. 2. The size of molecular sieve in the central
core, in outside layer and in middle layer are of 1.6 mm, Fig. 3. Adsorption quantity of Qe related to P.

Fig. 2. New packing arrangement in radial and axial direction.

126 Z. Yuwen et al. / Separation and Purification Technology 42 (2005) 123–127

the same tendency as the typical Langmuir curve. The results

also indicate that the higher the volume flux of the produc-
tion oxygen of Qo , the lower the Qe in the condition that
the HPAP was kept constant, and the flow flux of feed air of
Qa was also kept constant. When the flux of the production
oxygen is higher, the purge quantity of Qp will drop. The
quantity of the purge gas of Qp has decisive effect on the
process of regeneration of adsorbent. Thus the regeneration
of the adsorbent in the adsorption column is not sufficient
due to the insufficient quantity of the purge gas of Qp . The
adsorption ability will be decreased. It will directly lead to
the decrease of the effective adsorption, that is, the decrease
of Qe . Therefore, the Qe decreases with the increase of the
volume flow flux of the product oxygen when the purity of
production oxygen and the HPAP were kept constant. Fig. 5. Pressure related to nozzle size.

3.3. The effect of HPAP on the flux of production oxygen

is found to be 0.17 MPa–0.23 MPa according to the Fig. 4.
Qo
The corresponding nozzle diameter is of 0.7 mm–1.2 mm.
Fig. 4 also shows the relationship between the flow flux
3.4. The effect of nozzle size on HPAP
of the production oxygen and the purity of the production
oxygen with different adsorption pressure. Only in case the
A series of nozzles are designed and manufactured to en-
pressures were kept in medium range, both the purity of pro-
sure the stable and effective operation of the small device.
duction oxygen and the flux of production oxygen can ob-
The nozzle size is related to the adsorption pressure as shown
tain optimum performance. Neither the higher nor the lower
in the Fig. 5. The smaller the nozzle diameter is, the higher
HPAP would make the performance of the device better ac-
the HPAP is. The higher HPAP can be attained in short ad-
cording to the Fig. 4. The main reason is as follows. When
sorption time by the small diameter nozzle so that the energy
the HPAP is below a minimum value, the adsorbent will have
consumption will be reduced. According to the experimental
a little drop of the purity of the production oxygen according
results in Fig. 5, the right size of nozzle can be selected in
to basic theory of PSA. Inversely, adsorption ability, it will
accordance with the required HPAP.
directly lead to the drastic when the HPAP is too high, the
purge quantity will decrease based on above analysis. When
3.5. The effect of the purge quantity Qp on the purity of
the purge quantity is so small to be less than the quantity
oxygen
required by the regeneration of the adsorbent, the adsorp-
tion ability will be decreased. It will cause the decrease of
With the nozzle size unchanged the purity of the produc-
the effective adsorption so as to decrease the purity of the
tion oxygen is related to the purge quantity as shown in the
production oxygen. Therefore, the HPAP should be in an ap-
Fig. 6. When the purge quantity of Qp is moderate, the high
propriate range to ensure both the high flow flux and purity of
purity of the production oxygen of Qo can be obtained. Too
the product oxygen at the same time. The appropriate HPAP

Fig. 4. The purity of oxygen related to Qo . Fig. 6. Oxygen purity related to purge quantity Qp .
 Z. Yuwen et al. / Separation and Purification Technology 42 (2005) 123–127
Fig. 7. Qp /Qo related to Qo .
small purge quantity leads to the decrease of the regener-
ation of the adsorbent in the adsorption column and cause
the remarkable drop of the purity of the production oxygen
of Qo . On the other hand, when the purge quantity of Qp is
too high to exceed the quantity the adsorbent, the excessive
required by quantity of pressurized oxygen-rich air is filled
into the adsorption column. It will cause a higher pressure in
the adsorption column to reduce the adsorbing ability of the
adsorbent. Therefore, the purity of the production oxygen is
also reduced.
3.6. The effect of the purge quantity Qp on the flux of
production oxygen Qo
Fig. 7 shows the relationship between the flow of the purge
quantity Qp and the flow flux of the product oxygen Qo for
different nozzle sizes when the purity of the production oxy-
gen is kept above 93% in volume. The higher the flow of the
production oxygen Qo , the lower the purge quantity Qp is,
when the nozzle size is kept constant. As is known, the flow
of the product oxygen Qo subtracted from Qe gives the purge
quantity Qp . Therefore, the purge quantity Qp decreases with
the increase of the flow flux of the production oxygen Qo
when the HPAP is kept constant. Only when the ratio of the
127
purge quantity Qp to the flow of the product oxygen Qo is
maintained above 2.5 atm least, the device can then be en-
sured to operate properly.
4. Conclusions
1. It is found that the small-scale concentrator could operate
in optimum performance, with the purity of oxygen being
above 90%. The relationship between the HPAP and the
Qe is a typical curve of Langmuir.
2. The optimal operating pressure of the experiment device
is 0.18 MPa–0.23 MPa.
3. A series of nozzles were constructed to ensure the steady
and effective operation of the small-scale concentrator.
The smaller the nozzle diameter, the higher the HPAP.
The optimal nozzle diameter is from 0.8 mm to 1.2 mm
while the adsorption time is 10 s.
4. The highest purity of the production oxygen is about 97%
in volume by using this small-scale concentrator while the
flow flux of the production oxygen is of 2.0 l/min. When
the flow flux of the production oxygen of Qo is as high as
3.5 l/min, the purity of the production oxygen produced
by this small device can also be ensured above 80% in
volume. The small-scale concentrator is quite suitable for
hospital, domestic usage and oxygen bar.
References
[1] H. Hui, The study of pressure swing adsorption, Phys. Exp. 4 (2000)
16–18 (in Chinese).
[2] Z. Xuejun, G. Tong, The mathematic simulation to medical oxygen-
enriched by pressure swing adsorption, Iatrical Device (1998)
616–618 (in Chinese).
[3] Y. Chunyu, The mathematic simulation to the pressurization and de-
pressurization step during separating air by PSA, Pet. Chem. Eng. 5
(1998) 335–340 (in Chinese).
[4] F. Jinzhe, The semi-experiential formulation of the relationship be-
tween the pressure and oxygen concentration in the rich oxygen de-
vice by using PSA method, Cryog. Eng. 5 (1997) 12–17 (in Chinese).
[5] C. Guanj, W.N. Ng, W.J. Thomas, Dynamics of pressurization and
depressurization during pressure swing adsorption, Chem. Eng. Sci
49 (1994) 2657.