Adsorption

https://doi.org/10.1007/s10450-020-00291-8

Process intensification of the high‑purity nitrogen production

in twin‑bed Pressure Swing Adsorption plants
A. Marcinek1 · J. Guderian1 · D. Bathen2

Received: 11 September 2020 / Revised: 16 November 2020 / Accepted: 12 December 2020

© The Author(s) 2021

Abstract
With increasing nitrogen purity, PSA plants require an over-proportional air demand with the consequence that high-purity
PSA systems engender a distinct interest in energy-saving measures. This paper presents process intensification strategies
with the focus on a reduced energy consumption. Therefore, the influence of PSA configuration and cycle organisation on
process performance was investigated. Results are presented at two product purity levels (10 ppm/1000 ppm O ­ 2) and two
operating temperatures (25 °C/45 °C) in a lab-scale twin bed PSA (2 × 2 L). It is shown that dedicated strategies are available
to intensify the PSA process; however, their effects are dependent on ambient conditions and product purity levels.

Keywords High-purity nitrogen · Nitrogen generation · Pressure swing adsorption · Carbon molecular sieve · Process
intensification

1 Introduction (at 0 °C, 1 bar abs) and product purity levels up to 10 ppm
of the residual oxygen concentration. The process allows the
At present, the production of nitrogen from the air is carried kinetic separation of oxygen from nitrogen which is possible
out mainly by three methods: cryogenic distillation, pressure by taking advantage of the remarkably faster sorption rate of
swing adsorption (PSA), and membrane separation [1]. The oxygen over nitrogen in PSA-plants equipped with carbon
selection of a suitable technique is primarily based on the molecular sieves (CMS) [2]. High selectivity is attainable
required production rate, load profiles, utilisation (e.g. oper- due to the sieving effect in intentionally narrowed micropore
ating hours per week), and purity level of the product gas. mouths [3]. The simplicity of the plant operation, together
While the operation of cryogenic air separation units (ASUs) with multiple process variables and cycle organisation
is the most efficient method if large amounts of high-purity strategies, makes the PSA a technology of choice due to an
nitrogen are demanded, the utilisation of membranes would opportunity for customising the system to individual require-
be in preference when the requirement for either quantity ments. However, any modification of the cycle organisation
and purity is lower. Currently, the PSA technology for nitro- or slight alteration of process conditions influences the dis-
gen generation is commercially established in the intermedi- tribution of driving forces as a function of time within the
ate area, for product flow rates up to several thousand ­Nm3/h production and regeneration steps. Any change affects in its
ultimate consequence the local gas interstitial velocity which
again impacts for instance the local column pressure drop,
Supplementary Information The online version of this article
(https​://doi.org/10.1007/s1045​0-020-00291​-8) contains the heat and mass transfer rate, and the axial dispersion of
supplementary material, which is available to authorized users. heat and mass [4].
The compression of air is the key aspect of operating
* A. Marcinek costs while the bed size is the dominating factor in capi-
marcinek.aleksandra@fh‑muenster.de
tal costs [5]. Similar to membrane plants, PSA systems
1
Department of Chemical Engineering, Muenster University require an over-proportional air demand with increasing
of Applied Sciences, Stegerwaldstrasse 39, 48565 Steinfurt, nitrogen purity. Consequently, the operation of high-purity
Germany PSA plants motivates a particular interest in process opti-
2
Department of Thermal Process Engineering, University misation with a focus on the reduction of air demand;
of Duisburg-Essen, Forsthausweg 2, 47057 Duisburg, however, without increasing the number or volume of
Germany

13
Vol.:(0123456789)
 Adsorption

adsorber columns. That is why research aimed to intensify particularly relevant to industrial-scale applications are
the process and thus increase the cost-effective applica- presented and discussed.
bility of the ­N2-PSA technology in the high-purity range
could reduce the impact on the natural environment since
it utilises significantly less energy, hence generates less 2 Experimental
greenhouse gas than cryogenic air distillation [6].
Moulijn and Stankiewicz defined process intensification The scheme of the experimental set-up is presented in Fig. 1.
as “any chemical engineering development that leads to a Three main sections can be recognised: the feed gas prepa-
substantially smaller, cleaner, safer, and more energy-effi- ration and pre-treatment section (a), the twin-bed PSA plant
cient technology” [7]. Thus, very different measures can section (b), and the product and tail-gas analytics section (c).
be considered for intensifying PSA processes, including Details of the unit as well as the properties of adsorbent and
the application of improved separation materials as well adsorber column are described in the supplementary informa-
as the configuration of the installation. Related thoughts tion and elsewhere [14]. The unit is fed with dry air under a
are e.g. multi-bed PSA systems [8], trim feed concepts [9], pressure of 8.1 bar abs. The adsorption process is performed
dual-reflux concepts [10], hybrid systems [11], or opting at a working pressure of 8 bar abs. To control the purity level
for vacuum-PSA (VPSA) instead of PSA systems. Addi- always the product stream flow rate is adjusted in the experi-
tionally, if the product gas is required on the low-pressure ments. This control strategy represents the common operation
level, the recovery of pressure energy from the product of commercial plants. The flow rate of counter-current purge
stream by microturbines coupled with electric power gen- stream equals 0.0485 ­Nm3/h, which corresponds to the 40%
erators is also a potential development. Also, the parti- fraction of the adsorber volume within the selected purge time.
cle diameter can play a significant role as a measure for The process comprises a six-step PSA cycle which consists of
intensifying processes—depending on the mass transfer (1) co- and counter-current bed pressure equalisation; (2) co-
controlling steps from bulk to particle internal adsorp- current pressurisation by feed with (3) counter-current back-
tion sites for the considered adsorbent [12]. However, in flow of productf; (4) production; (1) co- and counter-current
practice particularly the pressure drop is also influenced bed pressure equalisation; (5) counter-current blow-down; and
by the particle diameter, so that the minimal, respectively (6) counter-current purge by the product gas. Details of the
optimal particle diameter also depends on the possibility PSA process organisation are described elsewhere [14]. The
to achieve the striven adsorption pressure in the produc- scheme of the six-step cycle design is presented in the sup-
tion step which becomes the more difficult the bigger the plementary information.
adsorbers.
Unfortunately, many of the listed concepts are not fea-
sible in high-purity ­N2-PSA plants since the commercially 3 PSA performance indicators
established nitrogen productivity range of a maximum few
thousand ­Nm3/h makes those approaches not affordable so The most common performance indicators (PIs) for the
far. That is why in the presented work process intensifi- ­N2-PSA technology are the product purity, the productivity,
cation focuses on already available measures, optimising and the air demand. In this study, process PIs are derived from
primarily the plant configuration and the process cycle macroscopic material balances of the PSA-unit presented in
organisation. As a constraint the separation material is Eqs. 1 and 2. Therefore, productivity and air demand are cal-
fixed which means that the particle diameter is fixed in culated according to Eqs. 3 and 4, respectively, where Q is the
this study as well. stream flow rate ­(Nm3/h), ρ is the gas density [kg/Nm3], ­wO2 is
After all, a PSA set-up is considered, consisting of two the oxygen mass fraction in the gas [mass.-%], and V ­ CMS is the
single adsorber columns which are alternating between volume of adsorbent in the system [­ m3]. Detailed definitions,
adsorption and desorption modes, providing a semi-con- determination, and verification strategy of PIs are described
tinuous flow of a high-pressure nitrogen stream [13]. Effect elsewhere [14]. In the N ­ 2-PSA technology it is accepted that
of (1) the flow resistances in the piping system including the product purity comprises the content of both nitrogen and
the trim of equalisation streams, (2) the nitrogen receiver argon since many industrial applications do not require an
volume, (3) the system void volumes, (4) the purge stream additional separation of inert gas mixtures. Consequently, the
flow rate, (5) the half-cycle time, and (6) the adoption of determination of the product purity is performed by assuming
a cutting step on the separation process performance are a binary gas mixture where simply the difference to the oxygen
experimentally studied for the high-purity ­N2-PSA at dif- concentration results in the nitrogen purity.
ferent product purity levels and operating temperatures. Qfeed ⋅ 𝜌feed = Qproduct ⋅ 𝜌product + Qtail−gas ⋅ 𝜌tail−gas (1)
Recommendations for process intensification strategies

13
Adsorption

Fig. 1  Scheme of the PSA experimental set-up [14]

Qfeed ⋅ 𝜌feed ⋅ wO2 feed = Qproduct ⋅ 𝜌product ⋅ wO2 product the PSA cyclic operation, the issue of pressure drop in
the system comes to the fore when optimising the process
(2)
performance. Generally, there are three sources of pressure
+ Qtail−gas ⋅ 𝜌tail−gas ⋅ wO2 tail−gas
drop in the installation: (1) uncontrolled flow resistances
in the piping system, e.g. pipes, bends, in-line filters, con-
nectors, or adsorber column elements as perforated plates
( )
Qproduct Qfeed ⋅ 𝜌feed − Qtail−gas ⋅ 𝜌tail−gas
=
VCMS VCMS ⋅ 𝜌product or sieves for the packed-bed support; (2) controlled flow
resistances in the piping system, i.e. the control valves
( )
Qfeed ⋅ 𝜌feed ⋅ wO2 feed − Qtail−gas ⋅ 𝜌tail−gas ⋅ wO2 tail−gas
= installed for a flow rate regulation of particular streams;
VCMS ⋅ 𝜌product ⋅ wO2 product
and (3) the adsorbent fixed-bed. All of these factors must
(3)
be taken into account when aiming for the optimisation
Qfeed 𝜌product Qtail−gas ⋅ 𝜌tail−gas of process performance. Whereas the column dimensions
= +
Qproduct 𝜌feed Qproduct ⋅ 𝜌feed and armatures together with the size of the adsorbent pel-
𝜌product ⋅ wO2 product Qtail−gas ⋅ 𝜌tail−gas ⋅ wO2 tail−gas (4) lets are selected at the plant design stage, the performance
=
𝜌feed ⋅ wO2 feed
+
Qproduct ⋅ 𝜌feed ⋅ wO2 feed of already existing PSA plants can still be improved by
the proper adjustment of controlled flow resistances in the
piping system. Therefore, the effect of the regulation of
controlled flow resistances in the piping system on per-
4 Results and discussion formance indicators, i.e. productivity and air demand,
was investigated for a PSA process exhibiting cycle con-
4.1 Effect of flow resistances in the piping system ditions as presented in Table 1. Selected operating tem-
perature levels are in agreement with the vast majority of
Since the rates of pressure build-up and fall-off appoint the relevant industrial applications; however, lower tempera-
main driving forces of adsorption and desorption during tures are also common. Thus, certain deviations in PSA

13
 Adsorption

Table 1  Standardised cycle conditions of the experimental program valves were returned to the reference state. Regarding the
equalisation valves both regulators in the equalisation
Product purity [ppm ­O2] 10; 1000
pipelines (Top + Bottom) were operated symmetrically.
Operating pressure (bar abs) 8
All mentioned control valves assembled in the PSA test
Operating temperature (°C) 20; 45
unit are commercially available throttles or shut-off nee-
Half-cycle time (s) 60
dle valves. The favourable stem position of each control
Adsorption time/desorption time (s) 59
valve was found based on the PSA performance indicators
Equalisation time (s) 1
results at product purity of 1000 ppm ­O2 at 20 °C which
Equalisation strategy Top/
Top + Bot- are presented in Table 2 and Fig. 2. In accordance with a
tom/Bot- current global focus on the energy conservation issue, a
tom minimised air demand, rather than maximised productiv-
Purge time (s) 59 ity, designates the preferable process outcome.
Purge flow rate (­ Nm3/h) 0.0485 As demonstrated in Fig. 2, controlled flow resistances in
Backflow Yes the piping system have an evident impact on the dynamic
behaviour of the PSA system and therefore a significant
effect on the process performance. The control valve in the
feed pipeline mainly regulates the volume ratio of the co-
Table 2  PSA performance in various controlled flow resistances in
the piping system at 20 °C current feed and counter-current backflow streams in the
adsorber column during the pressurisation step. As shown in
Stem position Productivity Air demand Fig. 2a, the favourable stem position was found at an open-
Opening (%) (Nm3/h ­N2/ (Nm3/h air/
m3 CMS) Nm3/h ­N2) ing of 53%. This indicates an optimal amount of product
gas reversed to the system as a backflow—not excessive to
Feed pipeline 100 80.67 3.49 decrease the productivity, and not insufficient to slow down
72 82.60 3.40 the pressure build-up in the system or to cause a prema-
53* 83.04 3.39 ture adsorber breakthrough due to dispersion effects. When
34 82.60 3.42 adjusting the feed control valve to the favourable stem posi-
25 81.92 3.44 tion, the productivity increases by 2.94% and the air demand
15 81.72 3.45 decreases by 2.86% in relation to the reference process.
6 80.13 3.49 Considering the tail-gas pipeline in Fig. 2b, it is shown
Tail-gas pipeline 100* 80.67 3.49 that any increase of the flow resistance causes a declined
81 80.67 3.51 PSA performance due to the reduction of the driving force
53 80.03 3.53 for the desorption process, resulting in a deterioration of the
25 74.87 3.62 fixed-bed regeneration.
15 67.15 3.85 The pressure equalisation step is introduced to the cycle
Equalisation pipeline 100 80.67 3.49 mainly for the reduction of operating costs since already
75 83.50 3.41 compressed and purified air can be transferred partly to
50 89.65 3.22 the neighbour column. Control valves in pressure equalisa-
25* 94.71 3.19 tion pipelines (Top + Bottom) regulate the volume of gas
5 95.37 3.39 exchanged between the both columns during the equalisation
*Favourable stem position step, as well as the subsequent starting pressure for pres-
surisation and blow-down steps, respectively. Since it acts
as an idle period during the PSA operation, the time of the
performance results are expected at dissimilar operating pressure equalisation step should remain as short as possible.
temperature levels, mostly due to modified conditions of It is expected that the complete pressure alignment to the
a mass transfer in adsorbent micropores. arithmetic average level, by minimising the flow resistance,
As a reference, the process with minimised controlled would lead to the minimal air demand. However, during
flow resistances in the PSA unit was selected. In order the rapid pressure reduction in the high-pressure column
to do so, control valves in all investigated pipelines—the a simultaneous desorption of the fast-adsorbing gas takes
feed, the tail-gas, and the equalisation pipelines—were place. Thus, the desorbed oxygen is eventually transferred
fully opened, and the performance indicators were deter- into the low-pressure column, contaminating the previously
mined. Subsequently, the flow resistance in each men- regenerated adsorbent. Consequently, the adsorption capac-
tioned pipeline was examined individually by adjusting ity in the following production step is declined. For this
the control valve stem position, whilst all other control reason, the optimum position of stems was found at 25%

13
Adsorption

Fig. 2  PSA performance at various controlled flow resistances in the piping system: a feed pipeline, b tail-gas pipeline, c equalisation pipelines

opening, as presented in Fig. 2c. When adjusting equalisa-

tion control valves to the favourable stem position, the pro-
ductivity increases by 18.22% and the air demand decreases
by 8.60% in relation to the reference process. The pressure
measured at the top of the adsorber columns after pressure
equalisation is presented in Fig. 3. The diagram shows that
a remaining pressure difference of about 1.36 bar between
the columns after equalisation results in the optimal PSA
performance in this plant.
Finally, control valves in all investigated pipelines were
adjusted according to their individual favourable stem open-
ing. PSA performance indicators were determined at dif-
ferent product purity levels as well as in different operat-
ing temperatures and compared to those of the reference
process. The results are presented in Tables 3 and 4. Opti-
Fig. 3  Pressure in the adsorber columns after pressure equalisation at misation of controlled flow resistances performed by set-
different stem positions of the equalisation valves
ting the stem position of discussed control valves to their

Table 3  PSA performance Minimised con- Optimised controlled Increment (%)

with minimised and optimised trolled flow resist- flow resistances
controlled flow resistances in ances
the piping system at product
purity of 1000 ppm ­O2 Operating temperature (°C) 20 45 20 45 20 45
Productivity ­(Nm3/h ­N2/m3 CMS) 80.65 68.17 97.35 85.87 20.71 25.96
Air demand ­(Nm3/h air/Nm3/h ­N2) 3.49 4.27 3.18 3.77 − 8.88 − 11.71

Table 4  PSA performance Minimised con- Optimised con- Increment (%)

with minimised and optimised trolled flow resist- trolled flow resist-
controlled flow resistances in ances ances
the piping system at product
purity of 10 ppm ­O2 Operating temperature (°C) 20 45 20 45 20 45
Productivity ­(Nm3/h ­N2/m3 CMS) 12.67 11.25 29.64 19.80 133.94 76.00
Air demand ­(Nm3/h air/Nm3/h ­N2) 15.93 20.72 7.78 12.73 − 51.16 − 38.56

13
 Adsorption

Table 5  Effect of the Pressure equalisation Top/ Stem position opening Productivity ­(Nm3/h ­N2/ Air demand
asymmetric pressure Top + Bottom/Bottom (%) m3 CMS) ­(Nm3/h air/
equalisation on the PSA Nm3/h ­N2)
performance at product purity
of 1000 ppm ­O2 0+1 0 + 100 – –
1
+ 2 12.5 + 37.5 84.50 3.47
3 3
1
+ 1 25 + 25 97.35 3.18
2 2
2
+ 1
* 37.5 + 12.5 99.15 3.13
3 3
1+0 100 + 0 84.25 3.52

*Favourable pressure equalisation strategy

Table 6  Effect of the asymmetric pressure equalisation on the PSA purity of 10 ppm O ­ 2, the favourable pressure equalisation
performance at product purity of 10 ppm ­O2 strategy maintained to be a symmetric setting of equalisa-
Pressure equalisa- Stem position Productivity Air tion control valves.
tion Opening (%) (Nm3/h ­N2/m3 demand The observed effect is directly related to the development
Top/Top + Bot- CMS) (Nm3/h of the mass transfer zone (MTZ) within the adsorbent fixed-
tom/Bottom air/Nm3/h
­N2)
bed in the cyclic steady state (CSS) conditions which can
be tracked via the measurement of temperature alteration
0+1 0 + 100 – – during the production step; assuming that the propagation
1
+ 2 12.5 + 37.5 22.40 9.83 velocity of the temperature front is not higher than of the
3 3
1
+ 1
* 25 + 25 29.64 7.78 concentration front. Because the PSA experimental unit can
2 2
2
+ 1 37.5 + 12.5 25.70 8.65 be considered as a non-adiabatic system, the formation of a
3
1+0
3
100 + 0 – – pure thermal wave front as described by Pan and Basmadjian
[15] will not occur. The experimental temperature profiles
*Favourable pressure equalisation strategy at investigated product purity levels are presented in Fig. 4.
At low product purity the gas velocity in the packed-bed
individually-established favourable position enhanced the is relatively high so that the MTZ becomes elongated due
PSA performance at every considered process condition; to mass dispersion effects. In that case, an increase of the
however, the effect becomes more significant as the product flow resistance in the bottom equalisation pipeline reduces
purity increases. In spite of that, to demonstrate the over- the driving force for the desorption process which results in
all optimum of PSA performance indicators in the function the minor transfer of the highly oxygen-enriched gas into the
of the controlled flow resistances regulation in the piping bottom part of the neighbour column, and thus, the adsorp-
system, a multi-parameter optimisation should be executed. tion capacity in the subsequent production step is not harm-
Additionally, the trim of the both top and bottom equali- fully affected. Moreover, the appropriate adjustment of the
sation control valves has been investigated, whilst other flow resistance in the top equalisation pipeline allows the
control valves have remained in their individual optimum straightforward transfer of the highly purified nitrogen gas
position of the stem. The effect on the PSA performance into the top part of the neighbour column which supports the
of the asymmetric settings was considered at 20 °C. The subsequent backflow stream.
results are presented in Tables 5 and 6. In the case of pres- On the other hand, at high product purity, the gas veloc-
sure equalisation exclusively through the bottom equalisa- ity in the packed-bed is relatively low and the mass transfer
tion pipeline the targeted oxygen residual concentration in occurs mainly in the bottom part of the adsorbent fixed-bed.
the product gas could not be accomplished at both inves- In that case, the top part of the bed serves as the nitrogen
tigated product purity levels; likewise, whilst the pressure safeguard bulk that ensures nearly complete oxygen rejec-
equalisation exclusively through the top pipeline at high tion from the compressed air feed stream. The symmetric
product purity level. setting of equalisation control valves results in the highest
At the product purity of 1000 ppm O ­ 2 the favourable performance, since the top part of both adsorber columns is
pressure equalisation strategy proved to be an asymmet- not contacted with oxygen-enriched gas—neither in co- or
ric setting with a 2/3 of the gas transferred through the counter-current pressure equalisation stream, so the nitro-
top equalisation pipeline and a 1/3 through the bottom gen safeguard bulk is not harmfully affected. The graphical
one. By means of this strategy the productivity increases representation of the MTZ with related favourable pressure
by 22.94% and the air demand decreases by 10.32% in equalisation strategies at investigated product purity levels is
relation to the reference process. However, at the product introduced in Fig. 5. These results motivate a more realistic

13
Adsorption

Fig. 4  Experimental temperature profiles along the adsorbent fixed-bed at 20 °C: a at product purity of 1000 ppm ­O2, b at product purity of
10 ppm ­O2

simulation of mass transfer effects in PSA adsorbers as being

state-of-the-art today and the subject of further studies by
the authors in the near future.
The subsequent experiments presented in this paper were
performed in the condition of optimised controlled flow
resistances with the symmetric operation of equalisation
control valves.

4.2 Effect of the volume of the ­N2‑receiver tank

The ­N2-receiver tank is implemented in the PSA system

in order to operate the cyclic process with a constant flow
rate of a high-pressure nitrogen stream that is requested in
Fig. 5  Visual demonstration of the oxygen MTZ development with the majority of industrial applications. Moreover, a mixing
related favourable pressure equalisation strategy: a at product purity space for the product gas is provided which is generated
of 1000 ppm ­O2, b at product purity of 10 ppm O­ 2
during the production step at inconstant composition. In this

Table 7  PSA performance at Vreceiver/Vadsorber 6 3.5 2.5

different volume ratios of the
­N2-receiver tank to the adsorber Operating temperature (°C) 20 45 20 45 20 45
column at product purity of
Productivity ­(Nm3/h ­N2/m3 CMS) 97.35 85.87 95.00 84.75 92.80 83.45
1000 ppm ­O2
Air demand ­(Nm3/h air/Nm3/h ­N2) 3.18 3.77 3.22 3.82 3.27 3.88

Table 8  PSA performance at Vreceiver/Vadsorber 6 3.5 2.5

different volume ratios of the
­N2-receiver tank to the adsorber Operating temperature (°C) 20 45 20 45 20 45
column at product purity of
Productivity ­(Nm3/h ­N2/m3 CMS) 29.64 19.80 25.47 15.90 22.82 12.80
10 ppm ­O2
Air demand ­(Nm3/h air/Nm3/h ­N2) 7.78 12.73 8.84 15.45 9.65 18.85

13
 Adsorption

Fig. 6  PSA performance in different volume ratio of the N­ 2-receiver tank to the adsorber column: a, b at product purity of 1000 ppm O
­ 2, c, d at
­ 2, a, c in operating temperature of 20 °C, b, d in operating temperature of 45 °C
product purity of 10 ppm O

Table 9  Alteration of the PSA

performance at reduced volume Product purity level [ppm O ­ 2] 10 1000
ratios of N
­ 2-receiver tank to Operating temperature (°C) 20 45 20 45
adsorber column ­ receiver/Vadsorber (%)
Reduction of V 42 58 42 58 42 58 42 58
Increment of productivity (%) − 14.07 − 23.01 − 19.70 − 35.35 − 2.41 − 4.67 − 1.30 − 2.82
Increment of air demand (%) 13.62 24.04 21.37 48.08 1.26 2.83 1.33 2.92

test rig also the purge and the backflow streams directed was experimentally studied by replacing the existing tank by
to the adsorber columns are served from the N ­ 2-receiver smaller tanks. The results are presented in Tables 7 and 8
tank. The selection of the smallest possible buffer vessel and Fig. 6. The increments of performance indicators associ-
is of essential importance while aiming for a reduction of ated with the change of the volume ratio of the ­N2-receiver
the unit footprint. However, an insufficient volume of the to the adsorber column are displayed in Table 9.
tank would give rise to enlarged pressure differences in the During the PSA operation, the pressure maintained in
system and could be a reason for reduced unit performance. the ­N2-receiver directly affects the gas flow rate through
Thus, the effect of the volume ratio of the product buffer ves- the adsorbent fixed-bed which consequently influences the
sel to the adsorber column on PSA performance indicators mass transfer conditions especially in the case of kinetically

13
Adsorption

­ receiver/Vadsorber: a 6, b 3.5, c 2.5

Fig. 7  Experimental pressure profiles of adsorber columns and ­N2-receiver tank at different V

controlled separations. The experimental pressure profiles of ­N2-receiver to adsorber column causes a decrease of the
adsorbers and product buffer tank at different volume ratios productivity and an increase in the air demand by nearly
of the ­N2-receiver to the adsorber column are presented in 35% and 50%, respectively, which is unacceptable for an
Fig. 7. As a consequence of the co-current pressurisation of industrial PSA operation.
the adsorbent fixed-bed with a simultaneous counter-current
backflow of product, the pressure in the ­N2-receiver tank 4.3 Effect of the void volumes in the PSA system
declines at the beginning of the PSA cycle. That pressure
downfall becomes more extensive as the volume of the prod- The PSA void volumes can be generally classified into three
uct buffer tank is reduced. An increased pressure difference categories: (1) packed-bed void volume, characterised by
between the adsorber column and the N ­ 2-receiver is created the parameter of bed porosity which usually includes voids
which causes an enlarged gas velocity in the adsorbent fixed- between adsorbent particles in the bulk of the fixed-bed as
bed and therefore a decreased contact time of gas and solid well as voids between adsorbent particles and column walls
phases. Moreover, the final production pressure is achieved alongside the adsorber; (2) adsorber void volume which is
in the adsorber column with a significant delay, diminishing the column volume not occupied by the adsorbent, com-
the adsorption process efficiency. Therefore, at every investi- monly the armature or fittings components supporting the
gated process condition the reduction of the volume ratio of packed-bed e.g. perforated plates, sieves, stabilising rings,
the ­N2-receiver tank to the adsorber column results in lower ceramic balls, coconut mats etc.; and (3) non-adsorber void
nitrogen productivity and higher air demand. volume, associated with assembled elements exterior to the
However, the reduction of the product buffer vessel vol- adsorber column which is essentially volume of the piping,
ume is recommended when low purity nitrogen is required. but also other elements, e.g. installed in-line filters. Since the
In this case, the gas velocity in the packed column is already packed-bed void has the most significant impact on the prod-
relatively high so that its further increase caused by the uct purity level and thus productivity [16], the adsorber and
enlarged pressure difference between the adsorber column non-adsorber voids could notably influence the air demand
and the N­ 2-receiver tank has not a great impact on the mass since the specific amount of feed gas must be devoted to fill
transfer dynamics. At both investigated operating tempera- up these volumes during the pressurisation step. Therefore,
ture levels, the modification between highest and lowest considering the already mentioned focus on the energy con-
volume ratio of N ­ 2-receiver to adsorber column causes servation issue, the effect of the adsorber and non-adsorber
a decrease in productivity and an increase in air demand void volumes on PSA performance indicators was experi-
merely by 5% and 3%, respectively. On the other hand, the mentally studied.
generation of nitrogen at a high purity level demands a prod- The experimental PSA test rig consists of: (1) the bottom
uct buffer vessel of large volume. In this case every accelera- adsorber void of app. 0.0062 L, (2) the top adsorber void
tion of the gas within the adsorber causes an enlargement of app. 0.0068 L, (3) the bottom non-adsorber void of app.
of the mass transfer zone; thus, the premature breakthrough 0.016 L, and (4) the top non-adsorber void of app. 0.018 L.
of oxygen. Especially at high operating temperature, the As the single adsorber packed-bed volume equals app. 2 L,
modification between highest and lowest volume ratio of the primary ratio of the listed system voids to the packed-bed

13
 Adsorption

Fig. 8  PSA performance at different purge stream flow rates: a, b at product purity of 1000 ppm ­O2, c, d at product purity of 10 ppm ­O2, a, c at
operating temperature of 20 °C, b, d at operating temperature of 45 °C

Table 10  PSA performance at Vvoids/Vadsorber 2.35% 4.35% 6.35%

different volume ratios of the
system voids to the adsorber Operating temperature (°C) 20 45 20 45 20 45
packed-bed volume at product
Productivity ­(Nm3/h ­N2/m3 CMS) 97.35 85.87 97.35 85.20 97.35 84.75
purity of 1000 ppm ­O2
Air demand ­(Nm3/h air/Nm3/h ­N2) 3.18 3.77 3.20 3.82 3.22 3.87

Table 11  PSA performance in Vvoids/Vadsorber 2.35% 4.35% 6.35%

different volume ratios of the
system voids to the adsorber Operating temperature (°C) 20 45 20 45 20 45
packed-bed volume at product
Productivity ­(Nm3/h ­N2/m3 CMS) 29.64 19.80 29.64 19.40 29.64 19.20
purity of 10 ppm O­ 2
Air demand ­(Nm3/h air/Nm3/h ­N2) 7.78 12.73 7.80 12.89 7.82 13.05

volume in the investigated PSA unit equals app. 2.35%. the non-adsorber void volume. The adsorber void is strictly
The different volume ratios of the voids to the adsorber associated with a column design and cannot be varied easily
packed-bed were accomplished by increasing exclusively in the already existing set-up. The piping extensions of the

13
Adsorption

original internal diameter were installed symmetrically at gas which should be replaced by nitrogen-enriched product
the top and the bottom column sections, therefore no addi- gas. However, this coarse approximation of X not always
tional effect of changing the gas velocity is introduced. The results in the improved process performance since the suit-
results are presented in Tables 10 and 11. able amount of the purge gas significantly influences the
As expected, at every investigated process condition concentration profiles along the adsorber length. Therefore,
an increase of the system void volume results in a slightly the effect of the purge stream flow rate on the PSA perfor-
higher air demand. The effect is a little more pronounced mance indicators was experimentally studied. The results are
at elevated operating temperature probably due to thermal presented in Tables 12 and 13 and Fig. 8.
expansion of the gas. Furthermore, the air demand increases At a low operating temperature level (20 °C) an increased
slightly with an enlargement of the system voids, following a purge flow rate into the system ensures elevated nitrogen
linear function. This finding indicates that the non-adsorber productivity values regardless of the investigated purity
voids in the PSA system have no impact on the mass transfer level. The efficiency of the desorption process is diminished
behaviour or column dynamics since the gas velocity is not as the operating temperature decreases so that an additional
influenced. Thus, an increase of the system void volume reinforcement in the form of the oxygen expulsion from
shows no or just a negligible effect on the nitrogen produc- the packed-bed void volume by the nitrogen purge stream
tivity at every investigated process condition. is required in order to regenerate the adsorber bed prop-
erly. The effect is more pronounced at a low product purity
4.4 Effect of the purge stream flow rate level (1000 ppm ­O2) rather than at a high product purity
level (10 ppm ­O2) due to the extended oxygen MTZ in the
The counter-current purge is a step of the PSA cycle applied adsorber column which is caused by the high gas super-
simultaneously to the counter-current blow-down, during ficial velocity and therefore intensified axial mass disper-
which a specified volume of product gas is reversed into the sion. Since the air demand remains rather insensitive to the
adsorber column. The step is an element of the regeneration change of the purge flow rate, it is recommended to operate
process of the packed-bed to reduce the amount of oxygen the PSA unit with an increased purge stream flow rate wher-
in the column before starting the following production step. ever the process conditions are corresponding.
The flow rate of the purge stream is adjusted according to On the contrary, at a high operating temperature level
Eq. 5: (45 °C) an increased purge flow rate into the system causes
lower nitrogen productivity values when the high-purity
X ⋅ Vadsorber
Qpurge = (5) product is generated. It indicates that the proper regenera-
tpurge tion of the adsorbent packed-bed with a narrow mass trans-
fer zone is accomplished already at the lowest investigated
where: ­Qpurge is the flow rate of the purge stream (­ Nm3/h),
value of the purge flow rate. Here, the thermodynamic effect
­Vadsorber is the volume of the packed-bed (­ m3), ­tpurge is time
comes to the fore as the desorption process is promoted at
of the purge step (h), and X is a proportionality factor [-].
elevated operating temperatures. Therefore, any enlarge-
The magnitude of the proportionality factor X can be associ-
ment of the purge stream flow rate results in the dissipa-
ated with the packed-bed porosity as it exhibits the adsorber
tion of the product gas; consequently, the air demand boosts
volume fraction occupied by oxygen-enriched desorption
up largely. When the low-purity product is generated, a

Table 12  PSA performance at

different purge stream flow rates X 30% 40% 50% 60%
at product purity of 1000 ppm Purge flow rate ­(Nm3/h) 0.0364 0.0485 0.0606 0.0727
­ 2
O Operating temperature (°C) 20 45 20 45 20 45 20 45
Productivity ­(Nm3/h ­N2/m3 CMS) 94.35 85.20 97.35 85.87 98.70 86.10 100.4 85.42
Air demand ­(Nm3/h air/Nm3/h ­N2) 3.19 3.77 3.18 3.77 3.17 3.79 3.18 3.85

Table 13  PSA performance at

different purge stream flow rates X 30% 40% 50% 60%
at product purity of 10 ppm O­ 2 Purge flow rate ­(Nm3/h) 0.0364 0.0485 0.0606 0.0727
Operating temperature (°C) 20 45 20 45 20 45 20 45
Productivity ­(Nm3/h ­N2/m3 CMS) 29.40 20.50 29.64 19.80 30.12 18.30 30.55 16.33
Air demand ­(Nm3/h air/Nm3/h ­N2) 7.70 12.04 7.78 12.73 7.73 13.72 7.78 15.45

13
 Adsorption

Fig. 9  PSA performance at different half-cycle times: a, b at product purity of 1000 ppm ­O2, c, d at product purity of 10 ppm ­O2, a, c in operat-
ing temperature of 20 °C, b, d in operating temperature of 45 °C

competition between the requirement of intensified purging 4.5 Effect of the half‑cycle time
of the packed-bed with an extended oxygen MTZ on one
side, and the saving of the product gas on the other side, The proper estimation of the contact time of gas and solid
can be found. At the proportionality factor value of 40% the phases is of essential importance in the case of kinetically
highest productivity is achieved with the lowest air demand controlled separations. Excessively prolonged duration of
value at the same time. adsorption diminishes separation selectivity by approaching
the thermodynamic equilibrium state. On the other hand,
too short adsorption is not beneficial either since diffusivity

Table 14  PSA performance

at different half-cycle times at Half-cycle time (s) 60 45 40
product purity of 1000 ppm O ­ 2 Purge time (s) 59 44 39
Purge flow rate ­(Nm3/h) 0.0485 0.0650 0.0733
Operating temperature (°C) 20 45 20 45 20 45
Productivity ­(Nm3/h ­N2/m3 CMS) 97.35 85.87 106.55 101.05 108.54 106.39
Air demand ­(Nm3/h air/Nm3/h ­N2) 3.18 3.77 3.38 3.81 3.50 3.86

13
Adsorption

Table 15  PSA performance

at different half-cycle times at Half-cycle time (s) 60 45 40
product purity of 10 ppm O ­ 2 Purge time (s) 59 44 39
Purge flow rate ­(Nm3/h) 0.0485 0.0650 0.0733
Operating temperature (°C) 20 45 20 45 20 45
Productivity ­(Nm3/h ­N2/m3 CMS) 29.64 19.80 25.46 19.82 22.38 19.81
Air demand ­(Nm3/h air/Nm3/h ­N2) 7.78 12.73 10.42 14.88 12.37 15.92

Fig. 10  Experimental pressure

profiles of adsorber columns at
different cutting times: a 0 s, b
5 s, c 10 s, d 15 s

limitations occur. Therefore, the existence of an optimal con- reduced working pressure are more pronounced as the half-
tact time of phases is expected which greatly depends on cycle time is shortened. However, in industrial practice brief
applied process conditions i.e. operating temperature and gas cycle times are aimed in order to increase the number of
pressure. In the PSA systems, additional effects of the mass cycles per hour of PSA operation and thus boost-up nitrogen
axial dispersion and the pressure drop along the fixed-bed productivity. For that reason, the effect of the half-cycle time
should be taken into consideration since the modification of on PSA performance indicators was experimentally studied.
a half-cycle time changes the time span proportion of pres- The results are presented in Tables 14, 15 and Fig. 9. Experi-
surisation and production steps. Therefore, the consequences ments were performed with the value of the proportionality
of those effects e.g. premature column breakthrough or factor X (see Eq. 1) equals 40%.

13
 Adsorption

When the low-purity (1000 ppm ­O2) product is generated one side and premature column breakthrough on the other
the productivity increases as the half-cycle time is shortened. side.
This finding confirms that the multiplication of the number At every investigated process condition, the reduction of
of cycles per hour brings benefits for the process perfor- the half-cycle time results in an increased air demand due
mance as the kinetic selectivity is more exploited. Thus, to the aforementioned change of time proportion for the dif-
the premature breakthrough of adsorbent fixed-bed does not ferent steps in the cycle. The adjustment of short half-cycle
occur even at relatively high gas superficial velocity and times makes the pressurisation step more significant in rela-
thus extended oxygen MTZ. By the reduction of half-cycle tion to the production step. Since the specific volume of
time to 40 s, productivity is increased by 11.5%. The effect compressed air is required to achieve the operating pressure
is more pronounced at a high operating temperature level level, a smaller amount of compressed air is utilised for the
(45 °C) due to a higher rate of mass transfer—in this case, production purpose at a short half-cycle time in comparison
productivity is increased by 23.9%. to a long half-cycle time. Consequently, at shortened half-
However, the opposite effect is found when the high- cycle time the air demand is higher in order to maintain
purity (10 ppm ­O2) product is generated. In that case, the the targeted purity of the product. The greatest air demand
productivity decreases as the half-cycle time is shortened enhancement is observed when the high-purity (10 ppm ­O2)
at a low operating temperature level (20 °C). It means that product is generated at a low operating temperature level
either the time provided for a mass transfer is insufficient and (20 °C).
thus the diffusivity limit was encountered; or a premature
column breakthrough occurs due to axial dispersion effects 4.6 Effect of the cutting time
during the pressurisation step. Nevertheless, the observation
could also be explained as a combination of both effects. The The cutting step is a supplementary cycle design element
productivity is insensitive to manipulation of half-cycle time implemented at the end of the PSA cycle terminating the
at a high operating temperature level (45 °C) which indicates blow-down and purge steps before finishing the adsorption
the compensated effects of faster mass transfer kinetics on step in the neighbour column. Cutting can be adapted in

Fig. 11  PSA performance at

different cutting times: a, b at
product purity of 1000 ppm O ­ 2,
c, d at product purity of 10 ppm
­O2, a, c in operating tempera-
ture of 20 °C, b, d in operating
temperature of 45 °C

13
Adsorption

Table 16  PSA performance

at different cutting times at Cutting time (s) 0 5 10 15
product purity of 1000 ppm O ­ 2 Purge time (s) 59 54 49 44
Purge flow rate ­(Nm3/h) 0.0485 0.0530 0.0584 0.0650
Operating temperature (°C) 20 45 20 45 20 45 20 45
Productivity ­(Nm3/h ­N2/m3 CMS) 97.35 85.87 95.85 86.75 93.90 85.85 91.50 86.10
Air demand ­(Nm3/h air/Nm3/h ­N2) 3.18 3.77 3.15 3.72 3.17 3.70 3.18 3.60
Tail-gas ­O2 concentration [vol.-%] 30.57 28.35 30.54 28.58 30.50 28.61 30.53 28.93

Table 17  PSA performance

at different cutting times at Cutting time (s) 0 5 10 15
product purity of 10 ppm O ­ 2 Purge time (s) 59 54 49 44
Purge flow rate ­(Nm3/h) 0.0485 0.0530 0.0584 0.0650
Operating temperature (°C) 20 45 20 45 20 45 20 45
Productivity ­(Nm3/h ­N2/m3 CMS) 29.64 19.80 30.10 22.05 30.10 22.46 30.10 24.30
Air demand ­(Nm3/h air/Nm3/h ­N2) 7.78 12.73 7.49 11.25 7.42 10.96 7.20 9.82
Tail-gas ­O2 concentration [vol.-%] 24.11 22.71 24.18 22.94 24.26 22.97 24.28 23.26

order to improve the PSA process performance of kinetic- When the high-purity (10 ppm ­O2) product is generated at
based separation since the difference in the sorption rates of a high operating temperature level (45 °C) implementation
molecules in the feed gas mixture can be exploited. Whereas of the cutting step brings the greatest benefit for the process
the duration of the blow-down step is reduced, the desorp- performance. The productivity increases and the air demand
tion of gases from the CMS adsorbent could still progress decreases significantly as the cutting time is prolonged. By
while the adsorber column is not opened to the ambient. means of implementing a 15 s cutting step the productiv-
The effect is more pronounced for components with slow ity increases by 22.73% and the air demand decreases by
mass transfer kinetics due to insufficient time provided for 22.86% in relation to the process without cutting. It indi-
its complete desorption from the adsorbent surface. Since cates that the oxygen desorption occurs in the initial phase
nitrogen is the component with slow kinetics, it is expected of the blow-down and the purge steps, respectively, so the
that losses of adsorbed nitrogen to the wasted tail-gas can be packed-bed with a narrow mass transfer zone is regenerated,
avoided. Consequently, the system pressure at the end of the especially at elevated operating temperatures. Moreover, the
PSA cycle increases, as presented in Fig. 10, which could mean oxygen concentration in the tail-gas escalates when
possibly diminish the air demand values without a premature the cutting time is prolonged so that the tail-gas becomes
breakthrough of the fast-adsorbing component. less diluted by nitrogen. It signifies that the desorption of
In practice, the cutting time is set by finding analytically nitrogen progresses during the cutting step which ensures
that time point at which the oxygen concentration under- lower oxygen concentration in the adsorber column at the
cuts the concentration level of ambient air at the adsorber beginning of the following production step and also a higher
exit. This procedure prevents avoidable nitrogen losses initial pressure in the system.
and helps to get a low air demand. Generally, cutting is not On the contrary, at a low operating temperature level
considered for the PSA reference process, since the opti- (20 °C) an implementation of the cutting step causes lower
mal placement of this step depends on the CMS proper- nitrogen productivity values when the low-purity (1000 ppm
ties and cannot be fixed independently from the selected ­O2) product is generated. It indicates that an insufficient
CMS-type or even CMS-lot [14]. Besides, analogously to amount of time was provided for a desorption of oxygen
the previously discussed case of changing the purge stream which forms a wide MTZ along the adsorber. Consequently,
flow rate in the PSA system, also the implementation of a the packed-bed is not properly regenerated before the follow-
cutting step impacts the oxygen mass transfer zone along the ing production step, and the premature oxygen breakthrough
adsorber during production and regeneration; thus, its adop- occurs. Since the air demand remains rather insensitive to
tion greatly depends on the process conditions. Therefore, the cutting time, it is recommended to operate the PSA unit
the effect of the duration of the cutting step on the PSA per- without the adoption of cutting wherever the process condi-
formance indicators was experimentally studied. The results tions are corresponding.
are presented in Tables 16, 17 and Fig. 11. Experiments were For other investigated process conditions it can be noticed
performed with the value of the proportionality factor X (see that the mean oxygen concentration in the tail-gas increases
Eq. 1) equals 40%. as the cutting time is prolonged which consequently results

13
 Adsorption

in lower air demand together with rather stable productivity References

values.
1. Castle, W.F.: Air separation and liquefaction: recent developments
and prospects for the beginning of the new millennium. Int. J.
Refrig 25, 158–172 (2002)
5 Conclusions 2. Shirley, A.I., Lemcoff, N.O.: Air Separation by Carbon Molecular
Sieves. Adsorption 8, 147–155 (2002)
Process intensification strategies for high-purity twin-bed 3. Patel, S.V., Patel, J.M.: Separation of High Purity Nitrogen from
PSA with a focus on the PSA unit configuration and cycle Air by Pressure Swing Adsorption on Carbon Molecular Sieves.
International Journal of Engineering Research & Technology
organisation were introduced and discussed. Effects of the (IJERT) 3, 450–454 (2014)
flow resistances in the piping system including equalisation 4. Vemula, R.R., Kothare, M.V., Sircar, S.: Anatomy of a rapid
trim, the nitrogen receiver volume, the system void volumes, pressure swing adsorption process performance. AIChE J. 61,
the purge stream flow rate, the half-cycle time, and the adop- 2008–2015 (2015)
5. Smith, A.R., Klosek, J.: A review of air separation technologies
tion of cutting step on the separation process performance and their integration with energy conversion processes. Fuel Pro-
were studied. Recommendations for improving the process cess. Technol. 70, 115–134 (2001)
performance with a particular emphasis on the reduction of 6. P. Froehlich, A Sustainable Approach to the Supply of Nitrogen,
energy consumption were given based on the experimental 2013.
7. Moulijn, J.A., Stankiewicz, A., Grievink, J., Górak, A.: Process
research at two product purity levels and two operating tem- intensification and process systems engineering: A friendly sym-
peratures. On one hand, it was clearly shown that measures biosis. Comput. Chem. Eng. 32, 3–11 (2008)
associated with the PSA configuration, which properly man- 8. Luberti, M., Friedrich, D., Brandani, S., Hyungwoong, A.: Design
age the flow resistances in the piping system, the volume of a H2 PSA for cogeneration of ultrapure hydrogen and power
at an advanced integrated gasification combined cycle with pre-
of the product receiver tank, and the system void volumes combustion capture. Adsorption 20, 511–524 (2014)
result in increased productivity and reduced air demand 9. Santos, M.S., Grande, C.A., Rodrigues, A.E.: New cycle configu-
numbers regardless of the required purity level or operating ration to enhance performance of kinetic PSA processes. Chem.
temperature. On the other hand, the optimal cycle organisa- Eng. Sci. 66, 1590–1599 (2011)
10. Bhatt, T.S., Storti, G., Denayer, J.F.M., Rota, R.: Equilibrium The-
tion is strictly related to the ambient temperature and the ory-Based Assessment of Dual-Reflux Pressure Swing Adsorp-
required purity level—their effect on the performance indi- tion Cycles That Utilize Light Gas for Pressure Swing. Ind. Eng.
cators are much more pronounced in high-purity rather than Chem. Res. 58, 350–365 (2019)
in low-purity PSA plants. 11. Esteves, I.A.A.C., Mota, J.P.B.: Simulation of a new hybrid
membrane/pressure swing adsorption process for gas separation.
Desalination 148, 275–280 (2002)
Acknowledgements This experimental research was generously sup- 12. Rota, R., Wankat, P.C.: Intensification of Pressure Swing Adsorp-
ported by CarboTech AC GmbH, Essen, Germany. We gratefully thank tion Processes. AIChE J. 36, 1299–1312 (1990)
Mr. Daniel Böke and Mr. Carsten Schledorn for their espousal. 13. Jasra, R.V., Choudary, N.V., Bhat, S.G.T.: Separation of Gases
by Pressure Swing Adsorption. Sep. Sci. Technol. 26, 885–930
Funding Open Access funding enabled and organized by Projekt (1991)
DEAL. 14. A. Marcinek, J. Guderian, D. Bathen, Performance determination
of high-purity N2-PSA-plants, Adsorption (2020).
Open Access This article is licensed under a Creative Commons Attri- 15. Pan, C.Y., Basmadjian, D.: An analysis of adiabatic sorption of
bution 4.0 International License, which permits use, sharing, adapta- single solutes in fixed beds: pure thermal wave formation and its
tion, distribution and reproduction in any medium or format, as long practical implications. Chem. Eng. Sci. 25, 1653–1664 (1970)
as you give appropriate credit to the original author(s) and the source, 16. Kim, Y.H., Lee, D.G., Moon, D.K., Byeon, S.-H., Ahn, H.W.,
provide a link to the Creative Commons licence, and indicate if changes Lee, C.H.: Effect of bed void volume on pressure vacuum swing
were made. The images or other third party material in this article are adsorption for air separation. Korean J. Chem. Eng. 31, 132–141
included in the article’s Creative Commons licence, unless indicated (2014)
otherwise in a credit line to the material. If material is not included in
the article’s Creative Commons licence and your intended use is not Publisher’s Note Springer Nature remains neutral with regard to
permitted by statutory regulation or exceeds the permitted use, you will jurisdictional claims in published maps and institutional affiliations.
need to obtain permission directly from the copyright holder. To view a
copy of this licence, visit http://creat​iveco​mmons​.org/licen​ses/by/4.0/.

13