International Journal of Innovative Technology and Exploring Engineering (IJITEE)

ISSN: 2278-3075, Volume-8 Issue-4, February 2019

Modelling and Optimization of PSA (Pressure

Swing Adsorption) Unit by using Aspen Plus®
and Design Expert ®
Pranta Sutradhar, Pritam Maity, Sayan Kar, Sourav Poddar

 separation unit (typically 75 ~ 105%). If a lower volume,

Abstract: Pressure swing adsorption (PSA) is a well-established gaseous oxygen or nitrogen product is required, then pressure
technique for separation of components from air, which is swing adsorption [2] and membrane separation [3] may be
commonly known as Air Separation Unit (ASU), drying of gas used or else larger volume of gaseous products, high purity
and nitrogen and hydrogen purification separation and etc. In
products or the recovery of argon, cryogenic air separation
PSA processes, the most important is adsorbent material
depending upon its properties. Generally, ASU is difficult to processes will be used [2], which is operated at extremely low
operate due to high degree of energy integration into itself. This temperature (-170oC ̴ - 195OC). Cryogenic air separation unit
research article represents the separation of nitrogen from air. As processes separate air components according to their different
separation of nitrogen is a very important in the field of chemical boiling temperatures [4]. Nowadays membrane processes like
engineering as it has wide applications in the various process molecular sieves are gaining importance. As membrane units
industries. There are various techniques for separation of are capable of producing nearly 600 tonnes of nitrogen per
nitrogen, amongst them the most common are reverse stirling day, having a purity of 90-99%. Commonly used molecular
cycle, LINDE-HAMPSON cycle, Joule Thompson effect and etc.
sieves are carbon molecular sieves [CMS] [15].
This article mainly focusses on the separation of nitrogen using
PSA unit only. The whole process was simulated using Aspen Plus This paper represents the process design of air separation
® and the simulated results were then optimized using Design unit process using Aspen Plus ®, computer based simulation,
Expert ®. Various flowrates ranging from 50 kg/h to 200 kg/h for different chemical engineering purposes. Our aim was to
were selected, depending upon the process conditions. The output calculate the production of N2. Various flowrates ranging
of the simulated results from Aspen Plus ® were then optimized from 50 kg/h to 200 kg/h were selected, depending upon the
using Box Behnken method, in order to obtain the optimized process conditions. The output of the simulated results from
flowrate of Nitrogen. The response pattern suggest that the Aspen Plus ® were then optimized, in order to obtain the
flowrates of nitrogen and other gases follows quadratic equation.
optimized flowrate of Nitrogen using Design Expert ®.
The significance of the coefficients of the equation and the
adequacy of the fit were determined using Student-t test and Though the design calculation does not give the real-life
Fischer F-test respectively. The final flowrates obtained are production environment but it can provide relief from making
interchanged in order to obtain the maximum conditions, except wide range of experiment without making the small-scale
for nitrogen production other production rates remain the same. reactor or plant.
Index Terms: Nitrogen, PSA (pressure swing Adsorption),
Aspen Plus®, Design Expert®. II. METHODOLOGY

I. INTRODUCTION The whole system was simulated using Aspen Plus ®,

which is shown in the figure 1. and then the sensitivity of the
Air is a composition of various gases, of which nitrogen final flowrates of N2 and other gases, from the
(N2) and oxygen (O2) are the main components and apart from computer-based simulation were optimized using Design
that there are other gases, which are present in the air in trace Expert ®.
amount. According to the previous researchers Liu. et. al.,
oxygen and nitrogen are used most extensively by various II. A. ASPEN PLUS MODELLING
industries [1]-[2]. Generally, nitrogen is used in chemical,
Aspen Plus ® has been used for the separation of Nitrogen
petroleum, food, pharmaceutical and nuclear and etc.,
from atmospheric air, as it provides accurate results compared
whereas oxygen is used by petroleum refineries, medical,
to the real life [5]. Depending on the comprehensive
concrete and wielding industries, chemical and gasification
thermodynamics’ properties, based on physical properties [8],
and etc. and other gases are used in steel making, heat
transport properties and phase behavior [6]. The present
treatment, manufacturing process for electronics and etc.
simulation used Ideal and Peng-Robinson models which fits
Because of the different demand for the gas purity, gas
best to equilibrium since components are gaseous and
amount and gas usage, there are two different types of air
non-polar. The components used are N2 (non-polar) and O2
(non-polar) for simplicity. Figure 1. shows the PSA of the
Revised Manuscript Received on 8 February 2019.
Pranta Sutradhar, Department of Chemical Engineering, Calcutta
process of separation of nitrogen from air.
Institute of Technology, Kolkata, India.
Pritam Maity, Department of Chemical Engineering, Calcutta Institute
of Technology, Kolkata, India.
Sayan Kar, Department of Chemical Engineering, Calcutta Institute of
Technology, Kolkata, India.
Dr. Sourav Poddar, Assistant Professor, Chartered Engineer and
Professional Engineer Department of Chemical Engineering, Calcutta
Institute of Technology, Kolkata, India.

Published By:
Blue Eyes Intelligence Engineering
Retrieval Number: D262802841919©BEIESP 64 & Sciences Publication
Modelling and Optimization of PSA (Pressure Swing Adsorption) Unit by using Aspen Plus® and Design
Expert ®

Figure 1. Process block diagram of the process of separation of separation of nitrogen from air.
Process Description:
Generally atmospheric air contains dust. So for this reason It is assumed that the independent factors A and B are
air is needed to be purified, but in our simulation we had continuous and controllable by experiments with negligible
considered dust free air which is mainly composed of errors. The generalized second order polynomial, correlating
nitrogen, oxygen and small traces of water vapor. Initially, the responses with the independent factors, is of the following
for the simulation we had considered flowrates varying from form:
50kg/h to 200 kg/h and temperature variation from 298K to 2 2 2 2
313K and pressure of 7e-7bar and the mass fraction of air
constituted of nitrogen-77%, oxygen-20% and water
yi  i   ij X j   iuj X u X j   ijj X 2j
vapor-3%. j 1 j 1u 1 j 1
The gas stream passes through valve, (J-T valve), before u j
entering into compressor. The compressor type used during
The significance of the coefficients and the adequacy of the
simulation is Polytropic using GPSA method and with a
fit are determined using Student-t test and Fischer F-test
discharge pressure of 5 bar. After passing through the
respectively. The values of Nitrogen flowrates and other
compressor, the gas stream enters into cooler where the
gases flowrates (oxygen and water) respectively have been
temperature is maintained at 298K.
maximized and minimized. The development of model
The output of the cooler is spitted and enters into
equation and optimization has been done using Design
purification unit 1 and 2, where the number of stages,
-Expert Software 7.0 ® [9].
condenser, reboiler, valid phase, distillate rate and reflux ratio
are selected and the results are 33, total, kettle, vapor-liquid,
IV. RESULTS AND DISCUSSION
5.88e-7 kmol/sec and 0.75 respectively. The final output
gases product are nitrogen and other gases (oxygen and water After performing the simulation, we had observed that the
vapor). The detailed description of the process parameters is production of nitrogen is maximum, as compared to the other
provided in the table A5. gases, which were considered during the simulation. The final
flowrates obtained from Aspen Plus ® are shown in the figure
III. PARAMETRIC SENSITIVITY AND OPTIMIZATION 2 and figure 3. It is clearly evident from figure 2 that the
flowrates of nitrogen continuously increase as the flowrates of
The effects of parameters namely, temperature (T) and air
air increases. The detailed production rates are provided in
flowrate (a) of air are the major response variables namely, A
the table A4. Therefore, we can confirm that final
and B have been correlated mathematically. The model
composition of nitrogen varies from 76.8% to 77%. But
equations have been developed with the aid of response
surface methodology [10] simultaneously varying the values
of f and T. The values of (T) and (a) were fixed using Box
Behnken method [11,12].
The mathematical relationships between the responses (Yi)
and factors, air flowrates (X1) and temperature (X2) are given
by,
Y  f ( X , X ) where i 1, 2
i i 1 2
(1)
Y1 = air flowrates
Y2 = temperature

Published By:
Blue Eyes Intelligence Engineering
Retrieval Number: D262802841919©BEIESP 65 & Sciences Publication
 International Journal of Innovative Technology and Exploring Engineering (IJITEE)
ISSN: 2278-3075, Volume-8 Issue-4, February 2019

Figure 2. The flowrates of nitrogen with respect of inlet air.

this value is expected to be around 99%. The deviation in the 100 kg/h. After which, as the flowrates of air increases the
result happened due to the unavailability of process decrease in the production of oxygen and water decreases
conditions. The final flowrates of oxygen (figure 3.a) and minorly. Thus it is confirmed that the production of nitrogen
flowrates of water (figure 3.b.) are represented below. It is increases as the flowrates of air increases, whereas for other
evident from the figures that the flowrates of oxygen and gases as the air flowrates increases the production rates
water, which is the main composition of other gases, decreases.
decreases when the flowrates of air increases from 50 kg/h to

Figure 3. The flowrates of oxygen outlet with respect of inlet air. b. The flowrates of water outlet with respect of inlet
air

Optimized flowrates output and Parametric Sensitivity.

The flowrates of nitrogen obtained varying flowrates and
f nitrogen  11.74090 - 1.28473e -3  (a ) 
temperature are considered for calculating the optimization 0.076555  (T ) 
condition using response surface methodology. Design (2)
Expert® software had been used for this purpose. The 6.25927e 6  (a)  (T )
quadratic equations obtained, which is predicted by the  2.34918e 6  (a) 2 
statistical modelling are as follows
2 2 2 2 1.24563e  4  (T ) 2
yi   i   ij X j   iuj X u X j   ijj X 2
j
j 1 j 1 u 1 j 1 (1)
u j
Figure 4 and figure 5 shows the flowrates of nitrogen and
other gases (oxygen and water vapor) as a function of
temperature and air inlet. From the ANOVA tables provided
in the Appendix A1.-A3. The model equation obtained
represents a surface quadratic type, indicating flowrates of
nitrogen and other gases are dependent variable. The model
equation for optimum nitrogen flowrates is

Published By:
Blue Eyes Intelligence Engineering
Retrieval Number: D262802841919©BEIESP 66 & Sciences Publication
 Modelling and Optimization of PSA (Pressure Swing Adsorption) Unit by using Aspen Plus® and Design
Expert ®

The model equations for optimum other gases (oxygen and The maximum results for the flowrates of nitrogen (87.2136
water vapor) flowrates are kg/h), oxygen (140.608 kg/h) and water vapor (107.62 kg/h)
foxygen   21.39390- 0.02383 ( a ) (3) have been obtained at T = 301.75K, 312.34K and 299.34K
and a = 124.55 kg/h, 183.25 kg/h and 82.36 kg/h respectively.
 0.15209 (T ) As predicted by the model equations 2-4, the optimum
 8.4443e 6  ( a )  (T ) simulated results of nitrogen, oxygen and water vapor were
interchangeable. Then nitrogen flow rate remained at 104.19
 8.11448e 6  ( a ) 2 kg/h when the optimum operating condition for the maximum
 2.46624e  4  (T ) 2 oxygen production was maintained. Therefore, interchanging
the conditions in the above equations, except for nitrogen
production others remains almost constant, which proves that
f water vapor   4.92826- 2.36966e-3  (a ) at 104.19 kg/h the process is sensitive.
 0.033509 (T )
 4.857785e 6  (a )  (T ) (4)
5
 1.18665e  (a) 2

5
 5.36769e  (T ) 2

Figure 4. The flowrates of nitrogen as a function of temperature and air flowrates.

Figure 5. a. The flowrates of oxygen as a function of temperature and air flowrates. b. The flowrates of water vapor as
a function of temperature and air flowrates.

IV. RECOMMENDATION FOR FUTURE SCOPE This paper demonstrates the optimized flowrates of
Demand of nitrogen and oxygen are increasing from time nitrogen using various process conditions. The process
to time. To meet this demand many industries especially simulation software used for simulating the results is Aspen
chemical industries, refrigeration, air conditioning, nuclear Plus® and the optimization of the process is carried out by
and etc., established their own separate unit. In order to Design Expert ®. The maximum optimized results for the
separate nitrogen from common source like ambient air. flowrates of nitrogen (87.2136 kg/h),
Simply PSA (pressure swing adsorption), which is a
membrane based operation, is the most common and basic
process of separation of nitrogen from ambient air.

Published By:
Blue Eyes Intelligence Engineering
Retrieval Number: D262802841919©BEIESP 67 & Sciences Publication
 Please Enter Title Name of Your Paper

oxygen (140.608 kg/h) and water vapor (107.62 kg/h) R2= 0.9872, Adj R2=0.9781, Pred R2=0.7204, Adeq Adeq
had been obtained at T = 301.75K, 312.34K and 299.34K Precision=25.318
and a = 124.55 kg/h, 183.25 kg/h and 82.36 kg/h
respectively. As, the optimum simulated results of A4. Values of flowrates of nitrogen, oxygen and water
nitrogen, oxygen and water vapor were interchangeable, vapor against flowrates of air.
then nitrogen flow rate remained at 104.19 kg/h when the
optimum operating condition for the maximum oxygen Flowrate Flowrate Flowrate of Flowrate of
production was maintained. Therefore, interchanging the air (kg/h) of Oxygen water-vapor
conditions, except for nitrogen production others remains Nitrogen (kg/h) (kg/h)
almost constant, which proves that at 104.19 kg/h the (kg/h)
process is sensitive. So, there is a huge scope lies in the 50 38.499 0.98008 1.3400012
improvement and simulation of the process, but the process 100 76.99 0.06250248 0.01665252
lacks energy exchange with surroundings. So advancement 150 115.5 0.09375336 0.002497878
must be taken in the forward direction for the betterment of 200 153.99 0.06250046 0.033305076
the process.
A5. Unit wise specification of process parameters and
APPENDIX reactions of PSA UNIT
A1. ANOVA for response surface quadratic model for
the flowrate of nitrogen as a function of temperature Unit Aspen Parameters Value
and flowrate of air Process
Code
Source Sum of df Mean F p-value Air Stream Temperature 250C
Squares Square Value Prob < F Pressure 0.07 N/m2
Model 2.867E-004 5 5.734E-005 1.01 <0.4746 significant
A-AIRINLET 4.376E-005 1 4.376E-005 0.77 0.4081
Total flow 50 kg/h
B-temp 2.534E-005 1 2.534E-005 0.45 0.5246 Composition
AB 4.959E-005 1 4.959E-005 0.88 0.3801 Mass
A2 1.537E-004 1 1.537E-004 2.72 0.1431 fraction 0.77
B2 4.145E-005 1 4.145E-005 0.73 0.4201
N2 0.2
Residual 3.956E-004 7 5.652E-005
O2 0.03
H2O
R2= 0.7202, Adj R2=0.8061, Pred R2=0.8456, Adeq Adeq
Precision=26.19 J-T valve Valve Pressure 1 Bar
Temperature 25 K
A2. ANOVA. for response surface quadratic model for estimation
the flowrate of oxygen as a function of temperature and Compressor Compressor Compressor Polytropic
flowrate of air model using GPSA
Outlet method
Source Sum of df Mean F p-value discharge 5 Bar
Squares Square Value Prob < F pressure
Model 1.73 5 0.35 158.69 < 0.0001 significant
Cooler Cooler Temperature 250C
A-AIRINLET 1.22 1 1.22 557.60 < 0.0001 0.07 kg/cm2
B-temp 5.512E-005 1 5.512E-005 0.025 0.8783
AB 9.025E-005 1 9.025E-005 0.041 0.8447
Pressure
A2 0.63 1 0.63 288.71 < 0.0001 Splitter Splitter Stream 5
B2 1.350E-003 1 1.350E-003 0.62 0.4575 Stream 6
Residual 0.015 7 2.184E-003
Flash Option
Pressure 0.07 kg/cm2
R2= 0.9913, Adj R2=0.9850, Pred R2=0.9468, Adeq Adeq
Valid phase Vapor-Liquid
Precision=29.631
Absorber 1 Column Number of 33
A3. ANOVA for response surface quadratic model for RadFrac stages Total
the flowrate of water vapor as a function of Condenser Kettle
temperature and flowrate of air Reboiler Vapor
Valid phases –Liquid
Source Sum of df Mean F p-value Convergence Standard
Squares Square Value Prob < Distillate 5.88 E-07
F
Reflux Ratio kmol/sec
Model 0.036 5 7.258E-003 108.10 < 0.0001 significant
0.75 mole
A-AIRINLET 0.026 1 0.026 379.82 < 0.0001
B-temp 2.774E-006 1 2.774E-006 0.041 0.8447
AB 2.987E-005 1 2.987E-005 0.044 0.5262
A2 0.011 1 0.011 162.50 < 0.0001
B2 3.275E-005 1 3275E-005 0.49 0.5074

Residual 4.700E-004 7 6.714E-005

Published By:
Blue Eyes Intelligence Engineering
Retrieval Number: D262802841919©BEIESP 68 & Sciences Publication
 International Journal of Innovative Technology and Exploring Engineering (IJITEE)
ISSN: 2278-3075, Volume-X, Issue-X, don’t delete Top & Bottom Header, & Fill up Manuscript details (Ist Page, Bottom, Left Side)

Absorber 2 Column Number of 33

Pritam Maity p is a B. Tech (Chemical
RadFrac stages Total
Engineering) student of Calcutta Institute of
Condenser Kettle Technology. He is working under Dr. Sourav
Reboiler Vapor Poddar in the department of Chemical Engineering
Valid phases –Liquid of Calcutta Institute of Technology. He has attended
one conference and published one conference
Convergence Standard
proceedings.
Distillate 5.88 E-07
Reflux Ratio kmol/sec
0.75 mole Sayan Kar is a B. Tech (Chemical Engineering)
student of Calcutta Institute of Technology. He is
Mixer 1 Mixer Pressure 0.07 kg/cm2
working under Dr. Sourav Poddar in the department
Valid phases Vapor-Liquid of Chemical Engineering of Calcutta Institute of
Mixer 2 Mixer Pressure 0.07 kg/cm2 Technology. He has attended one conference and
Valid phases Vapor-Liquid published one conference proceedings.

REFERENCES
1. Ming-Lung Li, Hao-Yeh Lee, Ming-Wei Lee and I-lung Chien,“ Dr. Sourav Poddar has completed PhD (Engg)
Simulation and Formula Regressionof an Air Separation Unit in from Jadavpur University, West Bengal, India in
China Steel Corporation“ , ADCONP, 2014, pp. 213-218. 2017, Chartered Engineer and Professional
2. D.R.Vinson,“ Air separation control technology“, Computers and Engineer from Institute of Engineer in 2018 and
Chemical Engineering, 30, 2006, pp. 1436-1446 2019 respectively. Presently, he is working as an
3. S.Ivanova, R. Lewis,“ Producing Nitrogen via Pressure swing Assistant Professor at Calcutta Institute of
Adsorption“, Chemical Engineering Progress,108(6), 2012, pp. Technology, Uluberia. He has published 7
38 -42.. research articles and has 9 conference papers. His
4. Z. Xu, J. Zhao, X. Chen, Z. Shao, J. Qian, L. Zhu, Z. Zhou, H. research interest is on Renewable Energy and waste
Qin,“ Automatic load change system of cryogenic air separation management, cryogenics and process control.
process“, Separation and Purification Technology, 81, 2011, pp.
451-465.
5. Aspen Plus Tutorial #1: Aspen Basic. Available:
https://www.aspentech.com
6. Aspen PlusTutorial #2: Thermodynamic Method. Available:
https://www.aspentech.com
7. Stoecker W.F., “Design of Thermal stress”, Toronto, Tata
McGraw Hill, 1986.
8. Aspen Tech, Aspen Physical Property System 11.1. Aspen
Technology, Inc ,Cambridge, MA, USA, 2001, Available:
https://www.aspentech.com
9. http://www.statease.com/training.html (Stat-Ease Webinars)
10. Marcos Almeida Bezerra, Ricardo Erthal Santelli, Eliane Padua
Oliveira, Leonardo SilveiraVillar, Luciane Amélia Escaleira,
“Response surface methodology (RSM) as a tool for optimization
in analytical chemistry“, Talanta, 75(5), 2008, pp. 965 -977.
11. http://www.weibull.com/hotwire/issue130/hottopics130.htm
(Box-Behnken Designs for optimizing Product Performance
Designs for optimizing Product Performance)
12. Box, G. and Behnken, D., “Some New Three. Level Designs for
the Study of Quantitative. Variables“, Technometrics, 2, 1960, pp.
455 – 475.
13. http://www.weibull.com/hotwire/issue130/hottopics130.htm
(Box-Behnken Designs for optimizing Product Performance)
14. Chatterjee, S., B. Price, Regression Analysis by Example. 2nd
Edition, John Wiley & Sons, New York, 1991, xvii, 278 pp.,
ISBN: 0‐471‐88479‐0, Available: https://onlinelibrary.wiley.com
15. W.F. Castle, “Air separation and liquefaction: recent
developments and prospects for the beginning of the new
millennium”, International Journal of Refrigeration, 25, 2002, pp.
158-172.
16. Randall F. Barron, Cryogenic systems, 2nd edition, Oxford
University Press, 1985, ISBN-13: 978-0195035674, Available:
https://www.amazon.com.

AUTHORS PROFILE

Pranta Sutradhar is a Diploma in Chemical

Engineering from Hooghly Institute of Technology
and B. Tech (Chemical Engineering) student of
Calcutta Institute of Technology. He is working
under Dr. Sourav Poddar in the department of
Chemical Engineering of Calcutta Institute of
Technology. He has attended one conference and
published one conference proceedings.

Published By:
Blue Eyes Intelligence Engineering
Retrieval Number: D262802841919©BEIESP 69 & Sciences Publication