Journal of Natural Gas Science and Engineering 26 (2015) 758e769

Contents lists available at ScienceDirect

Journal of Natural Gas Science and Engineering

journal homepage: www.elsevier.com/locate/jngse

Mercaptan removal from natural gas by the efcient cyclic adsorption

process; a simulation study
Zahra Tohidi, Shohreh Fatemi*, Omid Taheri Qazvini
School of Chemical Engineering, College of Engineering, University of Tehran, Tehran, Iran

a r t i c l e i n f o a b s t r a c t

Article history: A more efcient and economical cyclic adsorption process was proposed for mercaptan removal from
Received 26 May 2015 natural gas (NG) to reduce mercaptan content to less than 10 ppm and meet the environmental rules.
Received in revised form Continuous sulfur removal is studied for the NG feed stream, with pressure of 6.8 MPa, ow rate of
7 July 2015
2850 Nm3/hr and molar composition of 95.98% methane, 0.00182% water vapor, 1% carbon dioxide,
Accepted 8 July 2015
0.0134% mercaptan and 3% heavier hydrocarbons (C3). The proposed process of Pressure Vacuum Swing
Available online 13 July 2015
Adsorption (PVSA) was designed and simulated as a more efcient alternative process against the current
Industrial Pressure-Temperature Swing Adsorption (PTSA). In this work, an improved PVSA process was
Keywords:
Cyclic adsorption processes
simulated with sequences of bed pressurization, adsorption, equalization, blow down, bed regeneration
Mercaptan removal unit by vacuum and purge by product, in each process cycle. Vacuum condition of 10 KPa with the molar
Natural gas purge/feed ratio of 0.06 and temperature of 350 K was required for appropriate bed regeneration from
PVSA process adsorbed mercaptan to approach to the continuous cyclic steady condition. Comparison between PVSA
PTSA process and PTSA, at the same feed characteristics, same packed columns and adsorption operating conditions,
Economic study revealed that the PVSA process, with less cycle time than PTSA, could achieve the same product purity
with 94.8% recovery and 3.90 [mol/kg day] productivity, whereas PTSA has the recovery of 74.04% and
productivity of 2.79 [mol/kg day]. At the same time, operating with PVSA, instead of PTSA process, would
reduce the operating cost from 88 to 70 [Thousand $/year].
2015 Elsevier B.V. All rights reserved.

1. Introduction
The progress of the international energy demand shows an
average of 1.7% annual growth in the 2005 to 2020. This growth
concerns all energy sources and natural gas demands which would
be accounted for the highest growth rate in 2020 (Tagliabue et al.,
2009). Increasing concerns of the harmful effects of natural gas
contaminants on environment has led to the introduction of a
number of natural gas treatments. Out coming NG from the well
contains methane with impurities such as water vapor, carbon di-
oxide, nitrogen, hydrogen sulde, light mercaptans, ethane and
heavier hydrocarbons. Some part of the natural gas impurities must
be removed before commercial use. Today, according to the recent
environmental legislations, the sulfur emission in the atmosphere
should be considerably reduced to less than 20 ppm (Bellat et al.,
2008). Mercaptans, or more correctly thiols, are organic
* Corresponding author. School of Chemical Engineering, College of Engineering,
University of Tehran, Enghelab Avenue, Iran.
E-mail address: shfatemi@ut.ac.ir (S. Fatemi).
compounds in which the eSH groups are present in the molecular
structure of the hydrocarbons. The mercaptans must be removed
mainly for three reasons: (a) they have acidic property and can
cause serious corrosion problems, (b) they have offensive odor and
they are very inappropriate to be burnt, (c) most of them are highly
toxic and affect the subsequent catalytic reactions (Tamai et al.,
2006). In the conventional processes, the acid gases such as
hydrogen sulde and carbon dioxide are mostly removed in an
amine-wash unit. Since light mercaptans are not as acidic as
hydrogen sulde, they cannot be removed properly by amine
washing and an additional step is required to reduce the sulfur
concentration to an appropriate level. Gas purication by the
adsorption process could be as an alternative process in progress
for sulfur removal from NG (Bellat et al., 2008). Modeling and
simulation of the gas adsorption processes have been investigated
previously by some researchers. Some limited works have been
done for investigating mercaptan removal by adsorption using
different adsorbents (Bellat et al., 2008; Weber et al., 2008;
Cavenati et al., 2006; Dantasa et al., 2011; Mulgundmath et al.,
2012; Clausse et al., 2004; Zhang et al., 2009, 2008; Campo et al.,
2013). Shirani et al. (2010) also simulated mercaptan adsorption

http://dx.doi.org/10.1016/j.jngse.2015.07.010
1875-5100/ 2015 Elsevier B.V. All rights reserved.
 Z. Tohidi et al. / Journal of Natural Gas Science and Engineering 26 (2015) 758e769
on 13X adsorbent in presence of water vapor at isothermal condi-
tion and presented the breakthrough curves from the packed bed.
In another work, Qazvini and Fatemi (2014) simulated a PTSA
(Pressure-temperature swing adsorption) unit for mercaptan
removal in South Pars region. However, the high cycle time taking
for cooling down and the high economic costs are the main dis-
advantages of the mentioned process. It believes that, by replacing
the current PTSA process with PVSA (Pressure-vacuum swing
adsorption), the efciency of the process would be improved. In
addition, in PTSA process, the temperature shock induced to the
solid adsorbents by temperature variation in adsorp-
tionedesorption steps would reduce the life time of the adsorbents,
whereas PVSA doesn't require energy induction to the adsorbents.
It should be noted that, there is no evidence in the literature about
mercaptan removal by PVSA from NG in presence of other impu-
rities such as CO2, water vapor and heavier hydrocarbons. In this
work, a systematic simulation of mercaptan removal from NG in
presence of other impurities is carried out by the PVSA process in
face of the industrial PTSA process simulated by Qazvini and Fatemi
(2014) with the same bed congurations and design. This com-
parison is carried out on the base of different regeneration condi-
tions and the results are presented in terms of concentration and
temperature proles, and performance parameters such as purity,
productivity and recovery as well as a brief economical study. In
this research, industrial PTSA refers to the real mercaptan removal
unit (MRU) in South Pars region of Iran that is already working, and
the process design parameters and the real outlet data are taken
from there (Qazvini and Fatemi, 2014).
2. Process description and design
2.1. Description of process
A systematic adsorption model has been presented and imple-
mented to simulate the PVSA process proposed for purication of
natural gas (NG) from light mercaptans in presence of other im-
purities such as water vapor, CO2 and heavier hydrocarbons (C3).
The industrial mercaptan removal unit (MRU) is currently working
in South Pars region of Iran and this process is designed in the mode
of PTSA. This unit consists of six insulated two-layer packed bed
columns in which the regeneration step is carried out by increasing
temperature of the bed by the hot gas NG. The pressure of the
upcoming feed is 6.8 MPa and the adsorption process has been
initially designed to work at this high pressure. The properties of
the inlet NG is specied at Table 1. In addition, the adsorption bed
dimensions in industrial scale and the type and amount of the
adsorbents are presented in this table.
Each column consists of two layers: 13X zeolite and activated
alumina. The 13X zeolite is used for major adsorption of mercaptan,
however because of high afnity of 13X to water vapor, water plays
competitive role in mercaptan adsorption. Therefore, a pre-layer of
Activated alumina is required to adsorb water vapor before entering
the 13X layer. Adsorption of mercaptan and C3 is ignorable
Table 1
Feed specications and the bed characteristics.
Parameter Value
Flow rate (Nm3/hr) 2850
Pressure (Mpa) 6.8
Temperature (k) 302
Water vapor fraction (ppmv) 18:2
Mercaptan fraction (ppmv) 133:8
CO2 fraction (ppmv) 1:0  104
Heavy hydrocarbons fraction (ppmv) 3:0  104
Methane fraction (ppmv) 95:9848  104
759
compared to H2O and CO2 adsorption by activated alumina.
(Ferreira and Magalhaees, 2011; Baumgarten et al., 1977). Physical
properties of the adsorbents are reported in Appendix A1.
In this work, the process of PVSA is designed on the base of
existing PTSA, but different according to the regeneration step and
its operating conditions. A simple PVSA with no equalization and an
improved PVSA with pressure equalization steps are planned to
study the impact of pressure equalization step on purity, recovery
and productivity of the process.
These results would be comparable with the established in-
dustrial PTSA that is currently working with no pressure equal-
ization. The current PTSA process consists regeneration step
including two heating stages; 1st stage working at 480 K and the
2nd one working at 590 K.
The proposed simple PVSA process is designed to include 10
steps as following:
I. Adsorption step; during 18 h, the NG feed is introduced from
top of the bed with a ow rate of 2.2 kmol/s at pressure of
6.8 MPa and temperature of 302 K.
II. 1st depressurizing step; the bed pressure is decreased for 6.8
to 4 MPa during 10 min. The gas content of the bed is
released to the atmosphere, from the top.
III. 2nd depressurizing step; the bed pressure is decreased from
4 to 1.5 MPa during 10 min by releasing from the top of the
bed to atmosphere.
IV. Blow down step; the rest of the bed pressure is reduced
down to the atmospheric pressure. 10 min time is considered
for this step.
V. Evacuation step; the bed pressure is evacuated by a vacuum
pump to reach to 0.01 MPa during 10 min.
VI. 1st purge step; the bed is purged counter-currently with 2%
of the product stream at the vacuum pressure of 0.01 MPa
during 10 min.
VII. 2nd purge step; the bed is purged counter-currently with 6%
of the product stream in atmospheric pressure and temper-
ature of 350 K during 15 h and 40 min.
VIII. 1st pressurizing step; the bed pressure is increased up to
1.5 MPa through feed introducing from top of the bed during
10 min.
IX. 2nd pressurizing step; the bed pressure is increased from 1.5
to 4 MPa during 10 min.
X. 3rd pressurizing step; the bed pressure is increased from 4 to
6.8 MPa during 10 min.
XI. Rest; the bed is put on the rest for one hour to be prepared for
the next cycle.
The improved PVSA process was designed as the alternative of
simple PVSA with the same conditions but considering pressure
equalization steps between the columns. Pressure equalization step
provides inter-connection between two columns in which one of
the columns is working at depressurizing mode and the other one is
at repressurizing mode. According to the above modes, step III is
Bed 13X AC
Bed length (m) 4.65 0.75
Bed diameter (m) 3.7 3.7
Bed void fraction 0.37 0.26
Bulk density (kg=dm3 ) 0.69 0.82
Adsorbent weight (ton) 21.723 4.89
760 Z. Tohidi et al. / Journal of Natural Gas Science and Engineering 26 (2015) 758e769

inter-connected to step VII, and the step II is inter-connected to step
IX. The improved PVSA can be used to reduce natural gas con-
sumption and to enhance process efciency. In Fig. 1(a), (b) and (c)
the steps' arrangement are exhibited for industrial PTSA, simple
PVSA and improved PVSA, respectively.
Fig. 2 presents the operating pressure of the manipulated steps
in the PTSA and proposed PVSA processes. In order to manipulate
and control the steps of PVSA process with equalization, the
adsorption time was set to 17.5 h and the time of desorption steps
(evacuation and purge) was set to 16.5 h. This is done because
pressurizing steps should be at the rst half time of the cycle in
order to equalize with the depressurizing steps at the second half
time of the cycle. In this situation, we could save more natural gas
and enhance the process performance. Fig. 3 shows the time-table
of the process with equalization steps.
2.2. Model and assumptions
A mathematical model with mass, energy and momentum bal-
ances that represents the dynamic behavior of the multicomponent
adsorption bed was developed based on the following
assumptions:
 Radial gradients for mass, velocity and temperature are
neglected in the bed.
 The bed uid is assumed dispersed plug ow.

Fig. 1. Schematic of the step cycle sequences, in three processes for mercaptan removing, a: PTSA, b: simple PVSA, c: improved PVSA.
 Z. Tohidi et al. / Journal of Natural Gas Science and Engineering 26 (2015) 758e769 761

Fig. 2. Operating bed pressure with elapsed time in (a): PTSA cycle, (b): PVSA cycle.

Fig. 3. The time-table of the improved PVSA process, at each column for a six-bed unit.

 Porosity along the bed is constant with uniform packing.
 Mass transfer to the particles is described by linear driving force
(LDF) model with solid phase concentration.
 Diffusion and adsorption into the particles are assumed as lump
kinetic transfer model.
 There are no concentration and temperature gradients inside
the particles.
 Pressure drop along the bed is predicted by Ergun equation.
 Gas behavior follows Peng Robinson equation of state
throughout the process.
 Adsorption equilibrium is described by Extended Langmuir
Isotherm model for multicomponent adsorption, in which any
isotherm parameter is temperature dependent.
 The ow velocity varies along the bed and is calculated by the
total mass balance equation.
 The column operates in non-isothermal and non-adiabatic
conditions in which the heat is exchanged with surrounding.
 Temperature of uid and solid phase is not identical and heat
can be transferred between solid and uid.
 The solid temperature is homogeneous and a function of time
during adsorption and desorption process.
 In heat transfer to the environment, the wall resistance follows
thin wall assumption.
 Heat capacities of the adsorbents and adsorbed phase are
assumed constant.
 Heat of adsorption is assumed constant and derived from the
isotherm equation for each component.
According to the assumptions, mass and energy balances of the
gas and solid phases are shown in Table 2.
The isotherm parameters of the components were found from
literature (Weber et al., 2008; Ferreira and Magalhaees, 2011; Desai
et al., 1992; Ruthven, 1974) and they are reported in Table 3.
The correlations for physical properties, mass transfer and en-
ergy transfer coefcient are reported in Appendices A2 and A3,
respectively.
2.3. Numerical calculations
To obtain temperature and concentration proles along the bed
versus time at each cycle, model equations should be solved,
simultaneously. Because of complexity, these nonlinear equations
cannot be solved in analytical form and numerical solutions should
be used. The implemented numerical algorithm was nite differ-
ence scheme with the rst order Upwind Differencing Scheme
(UDS1) method to convert spatial derivatives of the PDEs to the
discrete form and convert the PDEs to the sets of ODEs. Then by
using implicit Euler method, all the ODEs were solved simulta-
neously for the next time step (Da Silva and Rodrigues, 1999;
Vining, 1997).
762 Z. Tohidi et al. / Journal of Natural Gas Science and Engineering 26 (2015) 758e769

Table 2
Mathematical model of the dynamic adsorption including, total mass, mole and energy balances.

Gas components' mass balance v Ci 2

vqi vuCi
vC
vt Dax;i v z2 e1  rs vt  vz
i

 
Overall mass balance in gas phase vuC P n vq
vC
vt  vz  1  rs 1
c
vt
i

Solid components' mass balance vqi

vt ki q*i  qi
Gas phase energy balance 2
Kax vvzT2  urCpg vTvz  rCpg vt  ap 1  hi T  Ts  hsurr T  Tsurr 0
vT
Pnc
Energy balance of the solid rs Cps vT s
r s DH vqi
i dt ap hi T  Ts

i1
vT
!
Ergun equation 37:5 12 mu
0:875rg 1u
2
vP  r 43
vz rp 42 3 p

 
Extended Langmuir isotherm DHi
b0;i exp RT
pi
q*i
 
qmi Pnc DHi
1 i1
b0;i exp RT pi

The bed was discretized by 20 nodes for each packed bed of
activated alumina and 13X zeolites. The sensitivity of the simulation
results to the node numbers was evaluated and it was found that
the results are independent of the node numbers for node numbers
greater than 20. The time step for implicit Euler method was also
chosen to be 100 s. The dependence of the results on the time step
was also examined and it was found that the results were inde-
pendent of the time step for time step values smaller than 100 s.
Each cycle consists of different steps in which different bound-
ary and initial conditions are governed. The initial and boundary
conditions for each step are introduced in Appendix A4.
3. Results and discussions
To choose the best regeneration method for mercaptan
desorption step, different conditions were performed and
compared with each other on the base of mercaptan loading. Fig. 4
shows mercaptan loading in different processes of PTSA, PVSA at
302 K and PVSA at 350 K in the regeneration step. The PTSA results
have been taken from the simulation of Qazvini and Fatemi (2014)).
Obviously in Fig. 4, PVSA with regeneration temperature of
350 K exhibits reduction of mercaptan loading to the lower level
during a shorter time, whereas PTSA and PVSA without bed
warming could not reach to this fast and short regeneration time.
Therefore, PVSA with warming up to 350 K, was considered as the
regenerating operation in the cyclic process.
The mass and heat transfer coefcients were inserted to the
balance equations by the algebric correlation equations that are
dependent to temperature and physical properties at each step, and
they are exhibited in Appendix A5.
dependent variables approach to a constant value. In the present
work, total adsorbed amount in adsorption stage between two
subsequent cycles were checked to approach to the CSS as
following:
Z L  ZL  
   
 qi dz  qi dz <d (1)
  
0 n1th cycle 0 nth cycle
In the above equation, convergence parameter d was considered
to be 105 and this condition was selected as the stop criteria of the
dynamic simulation.
It was revealed that, after about twelve sequential running cy-
cles of simple PVSA process, the steady state condition was ach-
ieved. Fig. 5 illustrates methane purity out of evacuation step,
versus the cycle number. The purity of methane is approaching to a
constant value after 12 cycles. Fig. 6a presents mercaptan impurity
at the outlet stream from the adsorption step of simple PVSA versus
time for 12 cycles and Fig. 6b shows the solid temperature at the
feeding section of the bed during one cycle. Theses results approve
that the CSS condition is achieved almost after 12 cycles from the
start up of the process. Regeneration of the adsorbent is a critical
point in cyclic adsorption processes. Appropriate procedure is
required for evacuation of the impurities from the bed. In the
present case, mercaptan is the strongest adsorptive material on 13X
zeolite, therefore properly bed evacuation from mercaptan at each
cycle and approaching to a constant temperature is required for a
continuous cyclic process.

3.1. Cyclic steady state condition

The cyclic adsorption process as every dynamic process should

pass several cycles to reach the steady state condition. Therefore at
the next cycles, the extract and rafnate compositions become
constant and do not change with time. A cyclic steady state (CSS) is
dened for the same bed position in each cycle in which the

Table 3
Parameters of Extended Langmuir Isotherm for Activated alumina and 13X.
     
Activated alumina kmol 1 kj
qm;i kg
B0;i Mpa DHi mol

Water vapor 0.0151 3:22  107 60.014

Carbon dioxide 0.0011 1:94  104 29.805
Zeolite 13X
Mercaptan 0.003580 4:32  108 65.1624
Water vapor 0.016842 4:82  107 62.5258
Carbon dioxide 0.005115 3:61  104 34.0031
Propane 0.002821 3:36  104 32.5412
Fig. 4. Mercaptan loading on the top of 13X layer at the mentioned processes.
 Z. Tohidi et al. / Journal of Natural Gas Science and Engineering 26 (2015) 758e769
Fig. 5. Methane purity after evacuation stage versus the cycle number in a simple
PVSA process, with the inlet mole fraction of 95.9848% methane.
Fig. 6. a: Mercaptan mole fraction at the outlet stream of adsorption step, b: solid
temperature prole as a function of cycle number, for the simple PVSA process.
3.2. Validation
Average outlet mercaptan and water vapor mole fractions which
are resulted from PVSA simulation at steady conditions are well
validated with industrial unit data. The comparison of simulated
and real values for mercaptan and water vapor mole fractions are
shown in Fig. 7. According to the reported values, mercaptan and
water vapor relative errors are 1.5% and 7.8%, respectively.
In order to investigate the validity of the simulation results,
mercaptan breakthrough curves from adsorption step in the real
and simulated conditions are plotted and compared with each
other, as shown in Fig. 8. Mercaptan outlet from 30 h adsorption
time shows gradually increasing mercaptan at the exit section of
763
Fig. 7. Comparison of simulated and real values of mercaptan and water vapor mole
fractions at the outlet stream.
Fig. 8. Simulated mercaptan breakthrough curve in comparison with real data.
the bed. The breakthrough results show that real plant data and
simulated results are close together and there is high precision
between real and model data during 18 h adsorption which is the
adsorption time of the process.
3.3. Breakthrough curves
The breakthrough curves of the species are shown in Fig. 9
which present the mole fraction of each species in gas phase at
different length of the bed. Feed gas enters from the top of bed and
passes through the alumina pre-layer. In this layer, most of water
vapor and some of CO2 content of feed is adsorbed before entering
the 13X layer. As shown in Fig. 9a, the bottom of the bed (13X,
bottom) is approximately free from mercaptan whereas the top of
13X is rapidly reached to the feed mercaptan content. As shown in
Fig. 9b, the majority of water vapor has been removed by alumina
pre-layer and the residue was adsorbed in 13X layer. Fig. 9c and 9d
show breakthrough curves of carbon dioxide and propane at
different points of the bed, respectively. It can be seen that the
fractions of carbon dioxide and heavy hydrocarbons reach to the
feed content between 1 and 2 h after passing through the column.
However, the afnity of 13X towards CO2 is higher than that of
propane because of the rolling up occurred in C3 breakthrough.
3.4. PVSA process in comparison with simulated PTSA process
3.4.1. Loading curves
One important factor in every cyclic adsorption process is the
adsorbed amount of solid phase in adsorption and desorption
stages. The loading curves show the maximum and minimum
adsorbed amount and the time required for regeneration. Due to
764 Z. Tohidi et al. / Journal of Natural Gas Science and Engineering 26 (2015) 758e769

Fig. 9. Breakthrough curves in the simple PVSA process, (a) Mercaptan, (b) water vapor, (c) carbon dioxide and (d) propane.

Fig. 10. Loading values versus cycle time at PVSA and PTSA cycles at the top of 13X layer, (a) Mercaptan, (b) water vapor.

Table 4
Comparison of the cyclic processes from the point of mercaptan and water desorption time and amount.

Simple PVSA Improved PVSA PTSA

Mercaptan time for complete desorption (hr) 6 6 12

Water vapor time for complete desorption (hr) 4 4 7
Mercaptan loading at top of the bed (Kmol/g) 0.000033 0.000033 0.019
Water vapor loading at top of the bed (Kmol/g) 0.00028 0.00028 0.015

the fact that purging gas enters through the bed from bottom to top,
the components accumulate at the top of the bed and this point of
the bed needs more time for desorption. Therefore the time
required for mercaptan desorption from top of the bed can be an
important factor for design of the cycle time that affects the re-
covery of the process. Fig. 10 shows mercaptan and water vapor
loading in PVSA and PTSA processes at top of the bed (Qazvini and
Fatemi, 2014). According to Fig. 10a, it is found that the time needed
for complete mercaptan desorption at top of the bed in the PVSA
process is 24 h while in the PTSA process, it requires 30 h to be
totally desorbed.
Water vapor acts as the dominant rival for mercaptan species in
the adsorption by 13X. As it was seen previously in breakthrough
curves, water vapor is a competitive adsorbing component with
 Z. Tohidi et al. / Journal of Natural Gas Science and Engineering 26 (2015) 758e769 765

Fig. 11. Temperature prole along the bed at (a) simple PVSA process, (b) PTSA process.

mercaptan and slows down bed regeneration. Therefore it is

necessary to investigate water vapor competitive behavior and the
impact of regeneration method on water desorption in both PVSA
and PTSA process. Fig. 10b indicates that loaded water vapor is
evacuated from the top of 13X after 22 h in PVSA process while in
the PTSA, loaded water vapor has reached to 0.015 mol/kg after 25 h
(at the end of rst heating step). The results reveal that it is possible
to work with lower cycle time in the PVSA process rather than PTSA
because lower time is required for the regeneration and this means
saving recovery and productivity.
Table 4 reports the regeneration step time required for different
cyclic processes to desorb mercaptan and water vapor and prepare
the bed for the next adsorption cycle. The results show that using
PVSA process has shortened the cyclic time compared with PTSA
and it can improve the productivity of the process.
Fig. 12. Basic performance parameters for PTSA, simple PVSA and improved PVSA.

temperature arises to the purge temperature of 350 K and the

3.4.2. Temperature prole temperature varies from bottom to top of the bed, because the gas
Since adsorption is exothermic, every process with adsorption stream loses its energy and becomes cooler along the bed.
and desorption steps includes temperature variations. In such
processes, temperature increases in adsorption stage and decreases 3.4.3. Process efciency
in desorption stage. As shown in Fig. 11, temperature is changed in The performance of a cyclic adsorption process is commonly
both PVSA and PTSA process. As can be seen in Fig. 11a and b, at the evaluated according to basic parameters of product purity, pro-
beginning of the adsorption step, the gas temperature rises to about ductivity and recovery dened, respectively by equations (2)e(4) as
320 K because of adsorption of high amounts of carbon dioxide and below.
heavy hydrocarbons, however after 2 h, none of these components Z tfeed
are adsorbing. Therefore, the temperature reaches to the feed
cCH4 u0 jzL dt
stream temperature. In Fig. 11a, a sharp temperature decrease is 0
PURITY Z (2)
observed after the end of adsorption time (18 h). This is because of Pn tfeed

extreme pressure drop in blow down and evacuation step i1 ci u0 jzL dt

0
(6.8 MPae0.1 MPa) in the PVSA process. According to this gure, at
the top of the bed, this decrement is more, since in these two steps, The values of purity, productivity and recovery for three pro-
the gas stream ows from down to top. After that the bed cesses of PTSA, simple and improved PVSA are showed in Fig. 12

0 1
Z tfeed Z tdepress Z tpurge
B C
@ cCH4 u0 jzL dt  cCH4 u0 jzL dt  cCH4 u0 jzL dtAAbed
0 0 0
PRODUCTIVITY (3)
tcycle Wads

Z tfeed Z tdepress Z tpurge

cCH4 u0 jzL dt  cCH4 u0 jzL dt  cCH4 u0 jzL dt
0 0 0
RECOVERY Z tfeed Z tpress (4)
cCH4 u0 jz0 dt cCH4 u0 jz0 dt
0 0
766 Z. Tohidi et al. / Journal of Natural Gas Science and Engineering 26 (2015) 758e769

Fig. 13. Operating cost for PTSA, simple PVSA and improved PVSA processes.
(Qazvini and Fatemi, 2014). According to the data reported in
Fig. 12, the purity of methane has increased by a value of 0.0299%
using the PVSA processes. The incremental change of purity is not
signicant, whereas there are signicant enhancement in the
values of productivity and recovery of the PVSA compared with
PTSA. The value of productivity has been increased from
h i h i
mol
2.79 kg$day mol
in PTSA to 3.90 kg$day in simple PVSA and to a less
h i
mol
greater value of 3.94 kg$day in improved PVSA. The value of re-
covery has increased from 74.04% in PTSA to 94.80% in simple PVSA
and to a greater value of 98.27% in improved PVSA.
The much less amount of PTSA recovery is due to the purge step
in which excess amount of the natural gas is released to the at-
mosphere at high pressure, whereas in the PVSA process the purge
step is carried out at low pressure. The impact of pressure equal-
ization has been studied in improved PVSA on the NG recovery and
this impact can be seen in comparison of the simple and improved
PVSA processes. Pressure equalization has improved the recovery of
the process by 3.47% from simple PVSA to improved PVSA. This
enhancement is due to avoiding gas release during depressuriza-
tion steps to the atmosphere.
3.4.4. Economic study
In order to compare three processes of simple PVSA, improved
PVSA and PTSA from the operating cost point of view, some cal-
culations have been done.
The effective parameters of the operating cost for the PVSA
process is included purge gas cost, gas stream cost for pressuriza-
tion step, vacuum pump cost and purge temperature cost. Oper-
ating cost of PTSA process is inuenced by the purge gas cost, gas
stream cost for pressurization step and temperature change cost in
heating and cooling steps.
In order to calculate the operating costs, gas ow rate in each
step, gas heat value (HV), vacuum pump power and efciency, and
price of the feed gas, product gas, electricity and fuel gas are
required. In Table 5 and Fig. 13, operating cost calculations for three
processes are exhibited (Exclusive Price List of Oil, Gas and Energy
Facilities, 2014; Peters et al., 2002).
As can be seen in Fig. 13, the operating cost of PTSA process is
more than PVSA processes. According to Table 5, the main reason is
the energy cost in heating steps (480 K in 1st heating step and
590 K in 2nd heating step) of PTSA process. Because of increasing
the operating temperature (303 K) to the high temperature and
thereafter cooling down the gas consumes large amounts of energy.
The difference of operating cost between simple and improved
PVSA processes comes from pressure equalization steps in which
the required gas for pressurizing the bed is supplied with the
released gas in depressurizing steps, using improved PVSA cycle.
Therefore, a large amount of valuable gas can be saved and the
operating cost can be decreased, signicantly.
4. Conclusions
Process of cyclic adsorption based on PVSA was proposed as a
more efcient process and it was compared with the PTSA process
established in South Pars Region of Iran. The most difference of the
proposed PVSA process was according to the regeneration method
which was based on evacuation at low pressure rather than heating
at high pressure. The simple PVSA was improved by adding
equalization steps and the results were compared from the point of
recovery, productivity and economic study.
It was concluded that improved PVSA would successfully remove
mercaptan to lower than 10 ppm with higher recovery and pro-
ductivity than those in existing PTSA and simple PVSA. At the same

Table 5
Operating cost calculations.

Cost Calculations PTSA Simple PVSA Improved PVSA

       
730 470 235
Costpressurize y$ 3
pressurize flow rate my  HV BTU  pricefeed BTU$
m3
       
15,270 15,270 15,270
Costpurge y$ 3
purge flow rate my  HV BTU $
 priceproduct BTU
m3
  !    
e 44,000 44,000
Costvaccum y$
Wpump
 priceelectricity $  24  366 H
hpump KWH y
     
$ m3 $
72,000 13,200 13,200
Costthermal y fuel flow rate y  pricefuel gas m3
 
$ 88,000 72,940 72,705
Total Cost y

pricefeed 0.2.
priceproduct 0.3.
priceelectricity 0.14.
pricefuel gas 0.41.
PTSA pressurize ow rate 106,000.
Simple PVSA pressurize ow rate 68,000.
Improved PVSA pressurize ow rate 34,000.
Purge ow rate 1,476,000.
PTSA consumed fuel ow rate 175,000.
PVSA consumed fuel ow rate 32,200.
Wpump: calculated by thermodynamic correlation for isentropic work 250 KW.
HV: Heat Value of gas 1/29.
hpump :pump efciency 0.7.
 Z. Tohidi et al. / Journal of Natural Gas Science and Engineering 26 (2015) 758e769 767

time, operating cost of the PVSA that is using relative vacuum for the Re Reynolds number and
bed regeneration is lower than the cost of PTSA that is using high gas Sc Schmidt number
temperature for regenerating adsorbents. At the same time, cooling Sh Sherwood number
down the column at PTSA process takes much more cycle time and t time (s)
therefore lower productivity than that of PVSA would be resulted. T gas temperature (K)
Investigation of the basic performance parameters showed that Ts adsorbent temperature (K)
the recovery and productivity of improved PVSA process has been Tr reduced temperature
increased by 3.47% and 1%, respectively, rather than the simple PVSA Tc critical temperature (K)
process. In addition, the improved PVSA process could increase the Tsurr surroundings temperature
recovery and productivity by 24.23% and 44.8% respectively versus u supercial gas velocity (m/s)
PTSA process. The economic study of the PTSA, simple PVSA and Vc critical molar volume (m3 /Kmol)
improved PVSA indicated that by using the PVSA process instead of y0 initial mole fraction
PTSA process, the operating cost would be decreased by 17%. yi mole fraction of species i
z axial coordinate in the bed (m)

Nomenclature
Greek letters
bed porosity
ap specic area of adsorbent (1/m)
p adsorbent porosity
b heat capacity equation parameter (J/mol/K)
t tortuosity factor
B0;i afnity constant at innite temperature (1/KPa) 
rs adsorbent density (kg 3 )
c heat capacity equation parameter (K)  m
r gas density (mol m3 )
Ci concentration of adsorption of species i (mol/m3 )
m gas viscosity (kg/m s)
CiA concentration of adsorption of species i in adsorption step
4 shape factor
(mol/m3 )
d convergence parameter
CiD concentration of adsorption of species i in
h pump efciency
depressurization step (mol/m3 )
CiE concentration of adsorption of species i in evacuation step
(mol/m3 ) Subscripts
CiR concentration of adsorption of species i in regeneration F feed
step (mol/m3 ) H high pressure
Cip concentration of adsorption of species i in pressurization L low pressure
step (mol/m3 ) A adsorption
Cpg gas heat capacity (J/mol/K) R regeneration
Cps average heat capacity of adsorbent (J/kg/K) D depressurization
d heat capacity equation parameter (J/mol/K) E evacuation
 P pressurization
Dax;i effective axial dispersion coefcient of species i (m2 s)

Dm;i molecular diffusivity of species i in mixture (m2 s)

DK;i Knudsen diffusivity of species i (m2 s) Appendix

Di;j Molecular diffusivity of species i in j (m2 s)
e heat capacity equation parameter (K) A1. The physical properties of the adsorbents
E characteristic energy of the adsorbent (J/mol)
hi lm heat transfer coefcient between gas and solid (W/ The physical properties of adsorbents and the bed Characteris-
m2 /K) tics are given in Table A1.
HV gas heat value (BTU/m3)
DHi heat of adsorption of species i (J/mol)
A2. Molecular diffusivity and physical properties prediction and
Kg gas thermal conductivity (W/m/K)
correlation
kf;i lm mass transfer coefcient of species i (m=s)

Kp;i pore diffusivity of species i (m2 s)
Molecular diffusivity of each component in the gas mixture is
ki overall mass transfer coefcient (s1 )
calculated by the equation (A.1) (Green and Perry, 2007):
Kax effective axial thermal conductivity (W/m/K)
k0g vapor thermal conductivity at atmospheric pressure (W/ 1y
m K) Dm;i Pn yi i (A.1)
xj Di;x
M molecular weight (g/mol)
Nu Nusselt number In this equation, yi is the mole fraction of the ith component and
nc number of the adsorbed components in the mixture n is the number total components in the gas mixture. For the
ncycle number of cycle mixture of natural gas, Di;j can be assumed as Di;CH4 . Di;j is the
n number of all component in gas mixture molecular diffusivity of binary gas mixture and is calculated by the
P pressure (KPa) Wilke-Lee equation (Green and Perry, 2007).
pi partial pressure (KPa) The amount of molecular diffusivity of each component through
qi adsorbed phase concentration of species i (mol/kg) the mixture at the adsorption, evacuation and purge is brought in
q*i equilibrium adsorbed phase concentration (mol/kg) Table A2.
qm;i maximum adsorbed phase concentration (mol/kg) All physical properties are considered as methane property.
R ideal gas law constant (J/mol/K) Because methane is the dominate component of the gas phase.
rp adsorbent radius (m) Pure gas specic heat capacity Cpg has been calculated as the
rpore pore radius (m) equation (A.2) (Mehdipour and Fatemi, 2012):
768 Z. Tohidi et al. / Journal of Natural Gas Science and Engineering 26 (2015) 758e769

! Table A4
c 2 Mass and heat transfer correlation equations to be used in the model equations.
Cpg a b     
T exp T eexp  Tc
c
Kax
The effective axial bed thermal conductivity 7 0:5PrRe
Kg
! (A.2) The effective mass transfer coefcient r r2 rc2
1 3kp 15 pKp;i
e 2 ki
d      (Ruthven, 1974) f ;i 15K K2 Dc;i
T exp e exp  e External lm resistance (Wakao and Funazkri, 1978) shi Dm;i
T T kf ;i
2rp
The coefcients of the above correlation are listed in Table A3. 1=3
Shi 2 1:1Sci Re0:6
The axial dispersion coefcient (Welty et al., 2000) Dax;i
Dm;i 20 0:5Sci Re
A3. Mass transfer and heat transfer correlations !
The macropore diffusion coefcient
1 t 1 D1m;i
Kp;i DK;i
The transport parameters needed in the model were calculated
Tortuosity Factor (Zhang et al., 2008) t p 1:51  p
employing frequently used correlations. The correlations used for  0:5
Knudsen diffusion coefcient (Yang, 1987)
the mass and energy balances in gas and solid phase are shown in DK;i 97rpore MT i
Table A4. ChiltoneColburn analogy (Byron Bird et al.) 1=3
Nui 2:0 1:1Pri Re0:6

A4. The initial and boundary conditions for numerical calculation

As stated in numerical calculation section, to obtain the results of Table A5

PVSA simulation, several non-linear equations should be solved, The initial conditions for numerical calculation employed at simple PVSA simulation.

simultaneously. The initial and boundary conditions for the nu- Adsorption stage:
merical solution of simple PVSA process are reported in Table A5. The t0:

initial and boundary condition of PTSA process is described in the Ci CiP;ncycle qi qiP;ncycle ;
P PF 6:8 MPa ; T TF 302K
paper of Taheri and Fatemi (Qazvini and Fatemi, 2014).Where the
z0:
subscripts F, H, L, A, R, D, E and P represent the feed, high pressure, vCi
vz 0 ; vz 0
vT
low pressure, adsorption, regeneration, depressurization, evacuation z L 5:4 m :
and pressurization, respectively, and L is the length of bed. 8
>
> vCi u
< vz D CF;i  Ci ;
>
ax;i

A5. Mass transfer and heat transfer parameters for kinetic >
> vT ucpg P
>
: T  TF ; P PF ; u uF
adsorption vz Kax RT
Depressurization stage:
t0: 1st depressurize step :
The average transfer parameters of the PVSA simulation were 8
> C CiAncycle ; qi qiAncycle ; PH 6:8MPa;
< i
found based on the correlations listed in Table A4. The parameters PH  PL PL 4:0MPa
>
: T TAncycle ; P  t PL
in different steps of the process are reported in Table A6. The timedepress 2nd depressurize step :
amounts of hi;avg that are reported in the Table A6, are the overall z0: PH 4:0MPa;
8 PL 1:5MPa
heat transfer coefcient which is dependent to hi and hsurr by > vCi
>
< 0;
  vz 3rd depressurize step :
equation (A.3). The amount of hsurr is assumed 10 mKw . > vT
> PH 1:5MPa;
2 $K : 0; u 0
vz PL 0:1MPa
1 1 1 z L 5:4 m : timedepress 10min
(A.3)
hi;avg hi hsurr vCi
0; vT
vz vz 0; P P zL
Countercurrent evacuation stage:
Table A1 t0:

Physical properties of the adsorbents. Ci CiDncycle ; qi qiDncycle ;
T TDncycle P PE 0:01 MPa
Physical properties of adsorbents Zeolite 13X Activated alumina z0:
vCi
vz 0; vz 0; u 0
vT
Particle radius (mm) 2.1 2.3
Average macropore radius (nm) 2.44 4.6 z L 5:4 :
vCi
BET surface area (m2 =g) 445.6 441.3
vz 0; vT
vz 0
Particle void fraction 0.24 0.28 Countercurrent regeneration stage:
Particle density (g=m3 ) 1.10 1.24 t0:
Heat capacity (kJ/kg K) 1.07 3.85 Ci CiE;ncycle ; qi qiE;ncycle ; T TE;ncycle
Tortuosity 1.38 1.36 z0:
8
< vCi  u Ci  0; vT ucpg P T  TF ;
>
Table A2 vz Dax;i vz Kax RT
>
:
Average molecular diffusivity of gases in mixtureDm;i cm2 =s. u uR ; P PR 0:1 MPa
z L 5:4 m :
Stage Adsorption Evacuation Purge
vCi
vz 0; vT
vz 0
Carbon dioxide 2.238  103 1.6 0.16 Co-current pressurization:
Heavy hydrocarbons 3  103 0.95 9.5  102  102 t0: 1st pressurize step :
Water vapor 2.59  103 3.4 0.34 8
> C CiR;ncycle 1 ; qi qiR;ncycle 1 PH 1:5 MPa;
Mercaptan 1.1988  103  103 0.79 7.9  102  102 < i
P  PL PL 0:1 MPa
>
: T TR;ncycle 1 ; P H t PL
timepress 2nd pressurize step :
Table A3 z 0: PH 4:0 MPa;
Heat capacity coefcients of methane. vCi PL 1:5 MPa
vz 0; vz 0; u 0
vT
3rd Depressurize step :
a  104 b  105 c  103 d  104 e  104 z L 5.4 m
PH 6:8 MPa;
(j mol1 K1) (j mol1 K1) (K) (j mol1 K1) (K) vCi
0; vT 0
vz vz PL 4:0 MPa
CH4 3.33 0.7993 2.0869 4.16 9.92 timepress 10 min
 Z. Tohidi et al. / Journal of Natural Gas Science and Engineering 26 (2015) 758e769 769

Table A6 R. Byron Bird, W. E. Stewart, E. N. Lightfoot. Transport Phenomena. John Wiley &
Mass and heat transfer parameters of the adsorption kinetics. Sons, Ing. New York Chichester Weinheim, Brisbane Singapore Toronto.
Campo, M.C., Ribeiro, A.M., Ferreira, A., Santos, J.C., Lutz, C., Loureiro, J.M.,
Step Zeolite 13X Activated alumina Rodrigues, A.E., 2013. New 13X zeolite for propylene/propane separation by
vacuum swing adsorption. Sep. Purif. Technol. 103, 60e70.
Adsorption kavg(1/s)
Cavenati, S., Grande, Carlos A., Rodrigues, Alrio E., 2006. Separation of CH4/CO2/N2
Water vapor 0.00098 0.00014
mixtures by layered pressure swing adsorption for upgrade of natural gas.
Mercaptan 0.00093 e Chem. Eng. Sci. 61, 3893e3906.
Carbon dioxide 0.00141 0.008205 Clausse, M., Bonjour1, J., Meunier, F., 2004. Adsorption of gas mixtures in TSA ab-
Heavy hydrocarbons 0.00731 e sorbers under various heat removal conditions. Chem. Eng. Sci. 59, 3657e3670.
Dax,avg(cm2/s) Da Silva, F.A., Rodrigues, A.E., 1999. Adsorption equilibria and kinetics for propylene
and propane over 13X and 4A zeolite pellets. Ind. Eng. Chem. Res. 38,
Water vapor 1.7028 2.4177
2051e2057.
Mercaptan 1.7031 2.4182
Dantasa, T.L.P., Lunab, F.M.T., Silva Jr., I.J., Torres, A.E.B., 2011. Carbon dioxideeni-
Carbon dioxide 1.7021 2.4171 trogen separation through pressure swing adsorption. Chem. Eng. J. 172,
Heavy hydrocarbons 1.7025 2.4172 698e704.
 
K ax;avg m$Kw 5.848 5.941 Desai, R., Hussain, M., Ruthven, M., 1992. Adsorption of water vapour on activated
  alumina. I e Equilib. Behav. Can. J. Chem. Eng. 699e706.
3.1621 4.1246
hi;avg mkw2 $K
Exclusive Price List of Oil, Gas and Energy Facilities, 2014. Ministry of Petroleum,
Engineering Department, 058.
Evacuation kavg (1/s) Ferreira, D., Magalhaees, R., 2011. Effective Adsorption Equilibrium Isotherms and
Water vapor 0.0118 0.00142 Breakthroughs of Water Vapor and Carbon Dioxide on Different Adsorbents,
Mercaptan 0.0043 e Industrial & Engineering Chemistry Research, pp. 10201e10210.
Carbon dioxide 0.00084 0.0072 Green, D., Perry, R., 2007. Perry's Chemical Engineers' Handbook, eighth ed.
Heavy hydrocarbons 0.0008 McGraw-Hill, New York.
e Mehdipour, M., Fatemi, S., 2012. Modeling of a PSA-TSA Process for Separation of
 
CH4 from C2 Products of OCM Reaction, Separation Science and Technology,
Dax;avg cm
2
s
pp. 1199e1212.
Water vapor 2.148 3.493 Mulgundmath, V.P., Jones, R.A., Tezel, F.H., Thibault, J., 2012. Fixed bed adsorption
Mercaptan 2.186 3.528 for the removal of carbon dioxide from nitrogen: breakthrough behavior and
Carbon dioxide 2.175 3.537 modeling for heat and mass transfer. Sep. Purif. Technol. 85, 17e27.
Heavy hydrocarbons 2.166 3.529 Peters, M., Timmerhaus, K., West, R., 2002. Plant Design and Economics for
  Chemical Engineers, fth ed.
K ax;avg m$Kw 0.35 0.35
Qazvini, O.T., Fatemi, S., 2014. Modeling and Simulation Pressureetemperature
 
0.2135 0.3931 Swing Adsorption Process to Remove Mercaptan from Humid Natural Gas; a
hi;avg mkw2 $K Commercial Case Study, Separation and Purication Technology.
Purge kavg (1/s) Ruthven, .D.M., 1974. Principles of Adsorption and Adsorption Processes. John
Wiley, New York (.
Water vapor 0.0311 0.051
Shirani, B., Kaghazchi, T., Beheshti, M., 2010. Water and mercaptan adsorption on
Mercaptan 0.0592 e 13X zeolite in natural gas purication process. Korean J. Chem. Eng. 27 (1),
Carbon dioxide 0.0057 1.1369 253e260.
Heavy hydrocarbons 0.0126 e Tagliabue, M., Farrusseng, D., Valencia, S., Aguado, S., Ravon, U., Rizzo, C., Corma, A.,
  Mirodato, C., 2009. Natural gas treating by selective adsorption: material sci-
Dax;avg ms
2
ence and chemical engineering interplay. Chem. Eng. J. 155, 553e566.
Tamai, H., Nagoya, H., Shiono, T., 2006. Adsorption of methyl mercaptan on surface
Water vapor 8.843 13.217
modied activated carbon. J. Colloid Interface Sci. 300, 814e817.
Mercaptan 8.859 13.223 Vining, G.G., 1997. Statistical Methods for Engineers. Duxbury Press, Pacic Grove.
Carbon dioxide 8.854 13.219 Wakao, N., Funazkri, T., 1978. Effect of uid dispersion coefcients on particle-to-
Heavy hydrocarbons 8.820 13.216 uid mass transfer coefcients in packed beds. Chem. Eng. Sci. 33, 1375e1384.
 
K ax;avg m$Kw 0.4548 0.4564 Weber, G., Benoit, F., Bellat, J.-P., Paulin, C., Mougin, P., Thomas, M., 2008. Selective
  adsorption of ethyl mercaptan on NaX zeolite. Microporous Mesoporous Mat.
0.5256 0.6499 109, 184e192.
hi;avg mkw2 $K
Welty, R., Wicks, C.E., Wilson, R.E., Rorrer, G., 2000. Fundamentals of Momentum,
Heat and Mass Transfer. John Wiley and Sons, New York.
Yang, R.T., 1987. Gas separation by Adsorption Processes. World Scientic Publishing
Company, Boston.
References Zhang, J., Webley, P.A., Xiao, P., 2008. Effect of process parameters on power re-
quirements of vacuum swing adsorption technology for CO2 capture from ue
Baumgarten, E., Weinstrauch, F., Hoffkes, H., 1977. Adsorption isotherms of hydro- gas. Energy Convers. Manag. 49, 346e356.
carbons on U-Alumina. J. Chromatogr. 138, 347e354. Zhang, J., Xiao, P., Li, G., Webley, P.A., 2009. Effect of ue Gas impurities on CO2
Bellat, J.P., Benoit, F., Weber, G., Paulin, C., Mougin, P., Thomas, M., 2008. Adsorption capture performance from ue Gas at Coal-red power Stations by vacuum
equilibria of binary ethylmercaptan/hydrocarbon mixtures on a NaX zeolite. swing adsorption. Energy Procedia 1, 1115e1122.
Adsorption 14, 501e507.