Separation and Purification Technology 173 (2017) 339–356

Contents lists available at ScienceDirect

Separation and Purification Technology

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

Cryogenic pressure temperature swing adsorption process for natural

gas upgrade
Mariana A. Moreira, Ana M. Ribeiro ⇑, Alexandre F.P. Ferreira ⇑, Alírio E. Rodrigues
Laboratory of Separation and Reaction Engineering, Associate Laboratory LSRE-LCM, Department of Chemical Engineering, Faculty of Engineering, University of Porto,
Rua Dr. Roberto Frias s/n, 4200-465 Porto, Portugal

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

Article history: Removal of carbon dioxide from natural gas for LNG production, can be carried out by cryogenic distilla-
Received 5 July 2016 tion using a multicolumn sequence to obtain a product stream with less than 50 ppm of CO2. Cryogenic
Received in revised form 22 September adsorption-based processes, particularly pressure thermal swing adsorption, arise as a reliable and inno-
2016
vative alternative to replace the distillation columns, in particular the last one. In this work, the imple-
Accepted 25 September 2016
Available online 28 September 2016
mentation of a new methodology to design and simulate an industrial scale pressure and temperature
swing adsorption process at low temperature (Cryo-PTSA) is presented, in order to produce an almost
pure methane stream, with a maximum of 50 ppm of CO2, maximizing the methane recovery.
Keywords:
Cryo-PTSA
Binderless 13X zeolite was selected as adsorbent and adsorption equilibrium isotherms of methane
Natural gas upgrade and carbon dioxide were determined experimentally, in the temperature range 183–473 K, up to
Binderless 13X zeolite 80 bar. The GERG2008 equation of state (EoS) was used during the dynamic simulations to account for
GERG-2008 EoS deviation from ideal gas law, through the use of REFPROP (REFerence fluid PROPerties) software package.
Cryogenic adsorption Since this method leads to high simulation time consumption, Surface Equations (SE) were used as alter-
native calculation method to determine the compressibility factor and the heat capacities at constant
pressure and volume. A methane recovery of 90.7% and a product stream with 41.8 ppm of CO2 in CH4
is achieved after cyclic steady state is reached. Cryo-PTSA simulation results were then compared with
an existing cryogenic distillation process. The results obtained by process simulation showed that it is
possible to replace the 3rd column of the cryogenic distillation process by a Cryo-PTSA achieving higher
CH4 recovery and lower power consumption (2.2 MW).
Ó 2016 Elsevier B.V. All rights reserved.

1. Introduction production, impurities must be removed, to enhance the calorific

value of the natural gas and avoid pipelines and equipment
Producing abundant quantity of energy with high quality, eco- corrosion.
nomic and environmental viability and sustainability is the main Carbon dioxide is one of the major contaminants in natural gas,
concern of the present times. One of such vital components of reducing its energy content and becoming acidic and corrosive in
the world’s energy supply that fulfills the abovementioned require- the presence of water, damaging pipelines and equipment. In
ments is natural gas [1]. LNG processing plants CO2 can solidify, blocking the pipeline sys-
Natural gas is generally considered as a clean fuel when com- tems and causing transportation drawbacks [1]. As a result, one
pared to other fossil fuels, since it emits less carbon dioxide than of the most challenging gas separation problems is the removal
other petroleum-derivate fuels. This makes it more favorable in of CO2 from the natural gas stream.
terms of the greenhouse effect, however, it is not necessarily free Among the several separation technologies available, such as
of impurities [1]. chemical absorption, physical absorption, cryogenic distillation
Natural gas consists of 85–95% methane, but it also contains and membrane processes, adsorption processes are considered a
considerable amounts of heavier hydrocarbons as well as other competitive solution [3,4]. Its major advantage is the simplicity
compounds (CO2, N2, Hg, He, H2S) [1,2]. In order to meet the pipe- of the adsorbent regeneration by thermal or pressure swing, as
line quality standard specifications or Liquefied Natural Gas (LNG) well as the production of high purity streams with lower energy
consumption [3].
⇑ Corresponding authors. Several studies were published focusing on the adsorption of
E-mail addresses: [email protected] (A.M. Ribeiro), [email protected] methane and carbon dioxide on various porous solids. However,
(A.F.P. Ferreira). most of the work was dedicated on the removal of CO2 for natural

http://dx.doi.org/10.1016/j.seppur.2016.09.044
1383-5866/Ó 2016 Elsevier B.V. All rights reserved.
340 M.A. Moreira et al. / Separation and Purification Technology 173 (2017) 339–356

Nomenclature

ap particle specific area (m
1) q0i adsorbed concentration of component i obtained by the
bi affinity constant (bar
1) pure-component isotherm (mol kg
1)
b0 affinity at infinite temperature (bar
1) qsat,i saturation capacity of component i (mol kg1s )
Cg,i bulk phase concentration of component i (mol m
3) Rg ideal gas constant (J mol
1 K
1)
Cg,T total bulk phase concentration (mol m
3 f ) t time (s)
Cp gas mixture molar specific heat at constant pressure T temperature of the mixture (K)
(J mol
1 K
1) T0 reference temperature (K)
b p;s
C particle specific heat at constant pressure (per mass Tb boiling point temperature (K)
unit) (J kg
1 K
1) Tc critical temperature (K)
b p;w
C wall specific heat at constant pressure (per mass unit) Tg bulk phase temperature (K)
(J kg
1 K
1) Tp solid temperature (K)
Cs,i concentration of component i at the particle surface Tw wall temperature (K)
(mol m
3) T1 ambient temperature (K)
Cv gas mixture molar specific heat at constant volume u0 superficial velocity (m3 m
2 s
1)
(J mol
1 K
1) Vads volume of the solid adsorbent (m3)
Cv,ads,i molar specific heat of component i in the adsorbed Vc volume of the balance inert parts (m3)
phase at constant volume (J mol
1 K
1) Vs volume of the solid adsorbent (m3)
Cv,i molar specific heat of component i at constant volume xi adsorbed phase mole fractions for component i (–)
(J mol
1 K
1) yi bulk mole fraction of component i (–)
Dax axial dispersion coefficient (m2/s) z axial position (m)
dp particle diameter (m) Z compressibility factor (–)
hf film heat transfer coefficient between the gas and parti-
cle (J s
1 K
1 m
2) Symbol
hw film heat transfer coefficient between the gas and wall aw ratio of the internal surface area to the volume of the
(J s
1 K
1 m
2) column wall (m
1)
kf film mass transfer coefficient (m s
1)
DH isosteric heat of adsorption of component i (J mol
1)
kg thermal conductivity (J mol
1 K
1) Dm mass difference (kg)
kh homogeneous mass transfer coefficient (s
1) e bed porosity (–)
L bed length (m) ep particle porosity (–)
Mi molecular weight of component i (g mol
1) k heat axial dispersion coefficient (J s
1 m
1 K
1)
P pressure (Pa) q bulk gas mixture density (kg m
3)
Pj partial pressure of component j (Pa) qap particle density (kg m
3)
Poi hypothetical pressure of a pure component (Pa) qb bed density (kg m
3)
Pref standard state pressure of the mixture (Pa) qg gas density (kg m
3)
i
qi adsorbed concentration of component i (mol kg
1) qwall wall density (kg m
3)
q0 saturation capacity at infinite temperature (mol kg
1) v molar volume (cm3 mol
1)
qi averaged adsorbed concentration of component i in the p spreading pressure of the adsorbed phase (Pa)
particle (mol kg
1) ci activity coefficient (1 for ideal solution) (–)
qi adsorbed concentration of component i in equilibrium
with Cs,i (mol kg
1)

gas and biogas upgrading at temperatures higher than 273 K. Nev- selectivity and thermal, chemical, and mechanical stability under
ertheless using temperatures lower than 273 K is of particular cycling conditions [5,6].
interest when the final product is LNG. Over the years a variety of solid adsorbents have been proposed
The novelty of this work is the implementation of a new taking into account their structures, compositions, and adsorption
methodology to design and simulate an adsorption process for pro- and regeneration mechanisms. Most of the adsorbents employed in
duction of a purified methane stream at low temperatures. This the commercial processes are carbon molecular sieves, activated
part of the study involves several steps, as defined in Fig. 1, which carbons, zeolites and other materials. However, the amount of
will be described in the following sections. The GERG2008 equa- information available in the literature at low temperature is not
tion of state (EoS) was used during the dynamic simulations in abundant and when available it is incomplete.
order to account for deviation from ideal gas behaviour and obtain Regarding the zeolites, natural chabazite, zeolite 4A, H-
more accurate results. Mordenite and a binderless zeolite 13X are the ones with more
Furthermore, for the first time are reported adsorption equilib- available information on the adsorption of methane and carbon
rium isotherms of CO2 and CH4 at a temperature range of 180– dioxide [7–10]. In Table 1 it is presented a list of zeolites found
473 K and up to 80 bar. in the literature with information on CH4 and CO2 adsorption at
low temperatures. Watson et al. [9] reported a very complete study
1.1. Literature review on adsorbents on the separation of CH4 and CO2 at low temperatures, concluding
that the chabazite has significant potential for removing CO2 from
The progress of separation technologies relies heavily on the natural gas. Grande et al. [11] published a study of an adsorption
development and selection of a proper adsorbent for the specific process for selective removal of CO2 from methane at low temper-
separation. An appropriate CO2 adsorbent should combine several atures to satisfy LNG specifications. In that work, adsorption equi-
attributes: fast kinetics, high CO2 adsorption capacity, high CO2 librium of CH4 and CO2 on commercial zeolite 4A and zeolite 13X
 M.A. Moreira et al. / Separation and Purification Technology 173 (2017) 339–356 341

highly tuneable crystalline structures and organic–inorganic

compositions with very important surface areas and pore volumes
[26,27].
The data found in the literature was obtained using materials in
the powder form, which is not suitable for industrial application.
Nevertheless, several studies were published in order to examine
the gas storage/adsorption capacity of each material (see Table 3).
Boutin et al. [28] published the adsorption equilibrium iso-
therms of CH4 and CO2 on MIL-53(Al), measured at several temper-
atures, from 183 to 343 K at pressures up to 10 bar. The authors
observed that at lower temperatures, adsorption and desorption
isotherms revealed steps and hysteresis loops, which can be linked
to adsorption-induced structural transitions (breathing).
Seo et al. [29] also reported a simple MOF based on Cu2+ [Cu
(pmc)2], that possesses a permanent porosity resulting from
straight 1-D channels of 5.5 Å. Unexpectedly, this porous MOF dis-
plays a highly selective sorption behaviour for CO2, while the sorp-
tion of CH4 is found to be almost negligible, at 195 K and 273 K.
Hayashi et al. [30] published a breakthrough experiment using a
CO2/CH4 gas mixture (about 50:50 v/v) on ZIF-20. The break-
through curves clearly show that ZIF-20 can separate CO2 from
CH4.

1.2. Cryogenic separation processes in industry

Cyclic adsorption processes for gas separation are a promising

solution for carbon dioxide removal from a gas stream. In the liter-
ature several adsorption processes that can be adopted to achieve
CO2 separation were found, depending on the regeneration method
[45–48]. They include temperature swing adsorption (TSA) and
pressure swing adsorption (PSA) processes. In TSA method, the
adsorbent is regenerated by desorption at a higher temperature
than that used during the adsorption phase of the cycle. While in
PSA process the regeneration occurs when the pressure in the col-
umn is reduced [49]. Continuous TSA and PSA processes operate
with multiple beds containing a stationary adsorbent and use a
manifold of valves to switch gas flow to the beds corresponding
to adsorption and desorption steps [49].
A list of the main patents in CO2 removal by PSA is presented in
Table 4. The majority of the patents listed describe processes in
which the CO2 separation is done at room temperatures, with the
exception of two patents which have claims that refer tempera-
tures below 273 K [50,51]. In European Patent [50], an innovative
Fig. 1. Scheme of the methodology followed for the design and simulation of an process for separating gas mixtures by selective adsorption is pre-
industrial scale Cryo-PTSA process.
sented. According to the authors, a bench scale unit was run using
a feed gas comprising 42.5% CO2 and 57.5% CH4 at 5.8 bar pressure
was reported at low temperatures (198, 208, 223, 248, and 279 K) and 267 K. A CH4 enriched stream containing 99% CH4 at 5.7 bar
up to 1 bar [11]. Adsorption breakthrough curves of a mixture of was obtained with 99% recovery, using 13X zeolite as adsorbent.
1.5% CO2 and 98.5% CH4 were also presented using zeolite 4A at Some PSA-based technologies for CO2 and CH4 separation from
204 K. natural gas are already being operated by major companies (see
Some remarkable data on the adsorption and separation of CO2 Table 5). According to Xebec, Inc., M3100 PSA gas separation sys-
and CH4 on Activated Carbons (AC) and Carbon Molecular Sieves tem was designed to separate gas mixtures by PSA, ensuring a con-
(CMS) materials was also found (see Table 2). The work of Bae tinuous supply of product gas [46,58,59]. In 2010, this technology
et al. [18] is noteworthy since it presents a complete set of equilib- was commissioned at the Platform Gail in the Santa Barbara Chan-
rium and kinetic data of the adsorption of CO2 and CH4 on an CMS nel to process a 2950 Sm3/h gas feed containing 13% CO2 to a sales
T3A. Eight different molecules (O2, N2, Ar, CO, CO2, SO2, CH4 and H2) gas containing 1.42% CO2 [58]. PSA technology to extract nitrogen
were studied, using a volumetric method, in the temperature range and carbon dioxide from contaminated gas streams was also pre-
of 293–313 K, and pressures up to 15 bar. A detailed study of the sented by IACX Energy. IACX’s technology is interesting because
adsorption equilibrium of pure nitrogen, methane, and carbon of its scalability and high hydrocarbon recoveries, at lower pres-
dioxide gases on a commercial Shirasagi MSC 3K-161 sample was sures (from 2.4 bar to 12 bar). The adsorbent used is denominated
also published [19]. In that work, Watson et al. [19] extended IACX CO2 SpongeÒ [60].
the work developed by Bae et al. [18], to lower temperatures The Molecular GateTM technology, originally developed by Engel-
(115–323 K) and higher pressures (up to 50 bar). hard Corporation, now part of the BASF Group, is based on trade-
More recently some works were published on metal-organic mark adsorbents ETS-4 developed for size selective separations,
frameworks (MOFs). The main interest in these new materials is with which PSA process cycles are designed to maximize the yield
focused in adsorption and gas separation since they combine of the methane product [61]. According with the company reports
342 M.A. Moreira et al. / Separation and Purification Technology 173 (2017) 339–356

Table 1
Literature review on adsorption equilibrium at low temperatures on zeolites.

Zeolite Commercial provider Form Compound Equilibrium data range Ref.

a
T (K) P (bar) q (mmol/g)
Natural chabazite Zeox Corporation Granulates (1–5 mm) CO2; CH4 248–302 1.2 CO2 – 4.3 [8]
CH4 – 1.6
Natural chabazite Zeox Corporation Granulates (2 mm) CO2; CH4 244–305 30 CO2 – 3.3 [9]
CH4 – 0.7
4A (Si/Al = 1.1) Sigma-Aldrich Pellets (3 mm) CO2; CH4 248–302 1.2 CO2 – 4.2 [8]
CH4 – 1.6
4A Sigma-Aldrich Powder CO2; CH4 196–298 1 CO2 – 6.9 [10]
CH4 – 0.2
4A Fluka Extrudates CO2; CH4 198–279 1 CO2 – 5.3 [11]
CH4 – 2.7
5A Union Carbide Co. Pellets(1.59 mm) CO2; CH4 303–573 10 CO2 – 3.1 [12]
CH4 – 0.8
Na-Mordenite (Si/Al = 6) CU Chemie Uetikon AG Pellets CO2; CH4 279–308 22 CO2 – 2.6 [13]
CH4 – 1.1
H-Mordenite In house made – CO2; CH4 279–308 22 CO2 – 2.2 [13]
CH4 – 0.7
H-Mordenite (Si/Al = 1.8) TOSOH Corporation Pellets (3 mm) CO2; CH4 273–302 1.2 CO2 – 2.5 [8]
CH4 – 0.9
H-Mordenite (HSZ-640HOA) TOSOH Corporation Pellets (1  4 mm) CH4 244–303 9 CH4 – 1.5 [14]
13X Grace Extrudates CO2; CH4 196–279 1 CO2 – 7.2 [11]
CH4 – 4.1
13X (binder metakaolin) In house made Pellets (2  4 mm) CO2 273–298 1 CO2 – 4.7 [7]
13X (binderless) In house made Pellets (2  4 mm) CO2 273–298 1 CO2 – 5.4 [7]
13X Aldrich Co. Beads CO2 273–353 1 CO2 – 4.1 [15]
Zeocarbon Zeobuilder Co. Pellets CO2 273–353 1 CO2 – 4.2 [15]
Na-ZSM-5 (Si/Al = 30) Chemie Uetikon AG. – CO2; CH4 297 1 CO2 – 1.9 [16]
CH4 – 0.7
H-ZSM-5 In house made – CO2 297 1 CO2 – 1.8 [16]
ZSM-5 (Si/Al = 15, 30, 60) In house made Pellets (5  6 mm) CO2 281–309 1 CO2 – 2.1 [17]
a
At 1 bar and respective lowest temperature reported.

Table 2
Literature review on adsorption equilibrium at low temperatures on activated carbons and carbon molecular sieves.

Carbons Commercial provider Form Compound Equilibrium data range Ref.

T (K) P (bar) qa (mmol/g)
CMS – Shirasagi MSC 3K-161 Japan EnviroChemicals Pellets (1.8  4 mm) CO2; CH4 148–323 50 CO2 – 4.9 [19]
CH4 – 4.1
CMS – T3A Takeda Chemical Co. Pellets CO2; CH4 293–313 15 CO2 – 2.0 [18]
CH4 – 1.1
AC – 2GA-H2J Kuraray Chemical Co. Pellets (1.7  2.36 mm) CO2; CH4 293–323 10 CO2 – 2.8 [20]
CH4 – 1.2
AC – BPL Calgon Co. Granulates CO2; CH4 273–323 60 CO2 – 3.3 [21]
CH4 – 1.0
AC – BPL Pittsburgh Chemical Co. – CO2 213–301 40 CO2 – 7.9 [22]
AC – Coconut shell N/A Granulates CH4 158–298 90 CH4 – 19 [23]
AC – AU-1 N/A Particles (0.4–0.9 mm) CH4 178–273 70 CH4 – 10.2 [24]
AC – Norit R1 Extra Norit Co. Pellets CO2; CH4 273–323 60 CO2 – 4.3 [21]
CH4 – 1.7
AC – Norit RB3 IMCD Australia Ltd. Pellets (3  4 mm) CH4 246–303 1.2 CH4 – 1.9 [25]
AC – Maxsorb Kansai Netsu Kagaku Co. Powder CO2; CH4 273–323 50 CO2 – 5.9 [21]
CH4 – 3.3
AC – A10 A’dall Co. Fibers CO2; CH4 273–323 60 CO2 – 3.8 [21]
CH4 – 2.5
AC – A Osaka Gas Co. Crushed material CO2; CH4 273–323 50 CO2 – 4.3 [21]
CH4 – 2.2
a
At 1 bar and respective lowest temperature reported.

[61], in 2002 the first Molecular GateTM system for the removal of [62]. Also and according with a previous study published, it was
carbon dioxide and water was started-up at the Tidelands Oil possible to conclude that the classical 13X zeolite presents thermal
Production Company facility in Long Beach, California. This unit stability at low temperatures [11]. This feature is a good indication
processed 1652 Sm3/h gas feed containing 18.8% CO2 to a sales that the binderless 13X zeolite will also be stable at lower temper-
gas containing 1.9% CO2, operating at 4.5 bar and a temperature atures. In this way, binderless 13X zeolite was selected to be tested
range between 289 and 300 K [61]. as adsorbent to produce a purified methane stream.
Considering all the information gathered, the material selection The main objective of this work is to design and simulate an
was based on published literature [11,62]. A recent study stated adsorption process for production of a purified methane stream,
that binderless 13X zeolite material presents higher selectivity with a low carbon dioxide content, at low temperatures. In this
and capacity values, when compared to the classical 13X zeolite way, and since adsorption equilibrium data are essential for the
 M.A. Moreira et al. / Separation and Purification Technology 173 (2017) 339–356 343

Table 3
Literature review on adsorption equilibrium at low temperatures on MOFs.

MOF Form Compound Equilibrium data range Ref.

a
T (K) P (bar) q (mmol/g)
Cu(fma)(bpe)0.5 Powder CO2; CH4 195 1 CO2 – 4.3 [31]
CH4 – 1.5
MIL-53(Al) Powder CO2; CH4 183–343 10 CO2 – 11.2 [28]
CH4 – 8.1
Mn(HCO2)2 Powder CO2; CH4 195 1 CO2 – 4.9 [32]
CH4 – 0.4
Zn(dabco)2(dctp) Powder CO2 196–298 1 CO2 – 3.1 [33]
Cu(pmc)2 Powder CO2; CH4 195–273 1 CO2 – 3.7 [29]
CH4 – 0.1
CuL Powder CO2; CH4 195–298 1 CO2 – 8.1 [34]
CH4 – 4.9
IRMOF-1 [MOF-5] Powder CO2 195–273 1 CO2 – 3.5 [35]
Mg2(dhtp)(H2O)2 Powder CO2; CH4 179–473 100 CO2 – 11.4 [36]
[CPO27Mg, MOF74(Mg)] CH4 – 10.0
Ni2(dhtp)(H2O)2 Powder CO2; CH4 179–473 100 CO2 – 8.6 [36]
[CPO27Ni, MOF74(Ni)] CH4 – 3.2
H2[Ni3O(H2O)3(TATB)2] Powder CO2; CH4 195 1 CO2 – 4.8 [37]
[PCN-5] CH4 – 1.9
[Zn2(BPnDC)2(bpy)]n Powder CO2; CH4 195–298 1 CO2 – 9.9 [38]
[SNU-9] CH4 – 1.6
[Zn(pydc)(dma)] Powder CO2 195–273 1 CO2 – 6.2 [39]
MAMS-1 Powder CO2; CH4 113–195 1 CO2 – 4.1 [40]
CH4 – 0.7
MCF-19 Powder CO2; CH4 195 1 CO2 – 21.4 [41]
CH4 – 5.4
[Cu2Glu2(l-bpa)-(CH3CN)]n Powder CO2 195 1 CO2 – 4.5 [42]
[Cu2Glu2(l-bpp)-(C3H6O)]n Powder CO2 196–298 1 CO2 – 3.1 [42]
[C40(H40)-Cd2N4O9] Powder CO2 195–298 1 CO2 – 6.2 [43]
[MMCF-1]
Zn(Pur)2(DMF)0.75(H2O)1.5 Powder CO2; CH4 273 1 CO2 – 10 [30]
[ZIF-20] CH4 – 7.4
Co3(2,4-pdc)2(l3-OH)2]9H2O Powder CO2; CH4 196–298 1 CO2 – 7.1 [10]
[CUK-1] CH4 – 0.4
Co3(DPyP)34DMF Powder CO2; CH4 195 1 CO2 – 2.7 [44]
CH4 – 0.1
Zn3(DPyP)32DMF4H2O Powder CO2; CH4 195 1 CO2 – 2.8 [44]
CH4 – 0.1
a
At 1 bar and respective lowest temperature reported.

Table 4
Summary of the main breakthroughs in the CO2 removal by PSA.

Patent Year Assignee Adsorbent T (K) P (bar) Ref.

US 4,077,779 1978 Air Products and Chemicals, Inc. BPL Activated carbon 293 28.2 [52]
EP 0,193,716 A2 1986 Air Products and Chemicals, Inc Zeolite 13X, Activated carbon 267 10.2 [50]
a
US 4,770,676 1987 Air Products and Chemicals, Inc. Zeolite 13X, Activated carbon 394–755 [53]
US 4,775,396 1988 Union Carbide Corporation Zeolite of the faujasite type 223–373 0.015–69 [54]
a
US 5,411,721 A 1992 UOP Zeolite, 3.5–17.2 [55]
a
WO 2003/020674 A1 2003 Engelhard Corporation Silica gel 10–55 [56]
a
US 7,828,877 B2 2005 Xebec Adsorption, Inc. Alumina, silica gel, activated carbon, zeolites 5A or 13X 8–18 [57]
WO 2008/072215 A2 2008 Porto University Dual bed containing: zeolite 4A, CMS, ETS + Zeolite 13X, 263–373 2–150 [51]
silica gel, alumina, activated carbon
a
No value range is given in the claims.

Table 5
PSA processes for CO2/CH4 separation, operating or/in patent claim.

Licensor Adsorbent Flow rate (Sm3/h) Ref.

Xebec M3100 PSA Xebec Metal-based <2950 [58]
CO2 SpongeÒ IACX Energy Activated carbon 236–5900 [60]
Molecular GateTM BASF Titanosilicate 590–59000 [61]

process design, adsorption equilibrium isotherms of methane and Additionally, an industrial scale pressure and temperature
carbon dioxide were measured experimentally on 13X zeolite swing adsorption process at low temperature (Cryo-PTSA) was
binderless beads, at temperature between 183 K and 473 K up to designed to produce an almost pure methane stream, with a max-
80 bar. imum of 50 ppm of CO2, maximizing the methane recovery.
344 M.A. Moreira et al. / Separation and Purification Technology 173 (2017) 339–356

2. Process description to obtain an enriched CH4 stream with a maximum of 50 ppm of

CO2, maximizing the CH4 recovery. It was considered that the col-
As mentioned before, in order to upgrade natural gas by remov- umn regeneration should be done with part of the product stream
ing carbon dioxide, different technologies are available [4]. When (<50 ppm CO2), at 5 bar and 473 K.
the final product is LNG, one of the possibilities to remove the
CO2 is cryogenic distillation. In such a case, the bulk removal of 3. Experimental section
CO2 will happen in one column, while one or more columns will
be used to polish the gas to LNG grade. Since LNG represents sen- 3.1. Materials
sibly one quarter of the natural gas transported all over the world,
for economic reasons, the existing processes are preferably applied Binderless 13X zeolite was used as adsorbent in order to study
to treat large volumes of natural gas. Going to smaller scales of gas the CO2/CH4 separation. The material was synthesized and shaped
processing, alternatives with lower operating costs must be pur- into beads by Chemiewerk Bad Köstritz GmbH. The shaping of this
sued. Therefore, cryogenic adsorption-based processes, particularly improved material, was performed with metakaolin, a non-zeolitic
pressure thermal swing adsorption (Cryo-PTSA), arise as a reliable component that was later converted into zeolite. More detailed
and innovative alternative to replace the distillation columns, in information about synthesis and characterization of the material
particular the last one. can be found elsewhere [64]. The main characteristics of the adsor-
Berstad et al. [63] reported a cryogenic distillation process, bent are summarized in Table 6.
where a natural gas feed was treated to meet the LNG CO2 content The gases used in this study were all pure components, supplied
specification (<50 ppm). According to the authors, a feed stream at by Air Liquide, with purities greater than 99.99%.
70 bar, 313 K and with a molar flow rate of approximately
24,800 kmol/h is treated in a three distillation columns sequence 3.2. Magnetic suspension balance and cryogenic cell set-up
process.
In the first column, also denominated as Bulk column, the con- Pure gas adsorption measurements were performed on a
centration of CO2 is reduced from 50.6–10.6%. In the liquid bottom Rubotherm magnetic suspension balance (MSB, Rubotherm,
product of the Bulk column, CO2 with a purity of approximately Bochum, Germany), with a precision of 10 lg and operated in
85% is obtained which is then fed to the CO2 recovery column closed system.
where purification to 94.35% is carried out. As can be seen in the schematic diagram of the experimental
The top product is further separated in the 2nd column, where a set-up (see Fig. 3), the system comprised a basket, where the
top product with a CO2 content of 0.9% is produced. The 3rd low- adsorbent is placed; a permanent magnet, placed in the adsorption
temperature column is fed with the top product of the 2nd column, chamber and linked to the basket; an electromagnet, that works
which presented a molar flow rate of 8280 kmol/h at 40 bar and together with the permanent magnet; and an analytical balance,
185 K. where the weight variations of the system are measured.
In this column, a top product with a 50 ppm content of CO2 is The MSB presents the possibility of measuring the mass change
produced at 40 bar and 185 K. The authors reported for the 3rd col- of two samples at the same time, by decoupling the measuring
umn a methane and carbon dioxide recovery of 88.5% and 99.5%, load. In this way, MSB has three different vertical positions of the
respectively. In order to obtain these values, the authors indicated suspension magnet, which correspond to three different measuring
an power consumption of 22.3 MW. More detailed information positions.
about this cryogenic distillation process can be found somewhere The first position corresponds to the zero point, allowing the
else [63]. balance to be tared and calibrated. In the second position, the bas-
Therefore, and based in the above information, the present ket is lifted up and its mass is weighed. In the third position of the
work presents a pressure and temperature swing adsorption pro- balance, the sinker is raised together with the basket and both
cess at low temperatures (Cryo-PTSA) process that allows the masses are weighed. By subtracting the first measuring point value
replacement of the 3rd column of the cryogenic distillation process, from the second, the mass of the second sample is given. This fea-
reported by Berstad et al. [63]. ture is important as it allows the measurement of the density of
In Fig. 2, a scheme of the proposed global Cryo-PTSA process is the fluid phase in the adsorption chamber by using an inert sinker
presented. The separation process should treat 8280 kmol/h of feed with known volume. The density of the fluid phase is then mea-
at 190 and 40 bar, containing 99.1% of CH4 and 0.9% of CO2, in order sured with high accuracy using Archimede’s buoyancy principle
[65].
A main advantage of the MSB is its wide working temperature
range, since it can be operated in a temperature range between
123 K and 673 K, by the use of a cold gas cooling jacket or an elec-
trical heater.

3.3. Adsorption equilibrium measurements

The adsorption equilibrium measurements of pure components

were performed using a gravimetric method (Rubotherm magnetic

Table 6
Properties of the binderless 13X Zeolite.

Adsorbent properties Units

Shape Beads –
Particle diameter 1.2–2.0 mm
Average crystal diameter 2 lm
Apparent particle density 1072 kg m
3
Particle porosity 0.39 –
Fig. 2. Global process representative of the case study.
 M.A. Moreira et al. / Separation and Purification Technology 173 (2017) 339–356 345

Fig. 3. Schematic diagram of the experimental set-up for adsorption equilibrium measurements (Rubotherm magnetic suspension balance).

suspension balance). The isotherms were determined at 180, 220, T
Tb

v ¼ vb þ ðb
v b Þ ð3Þ
269, 323, 373 and 473 K, up to 80 bar. Tc
Tb
After cooling down or heating up the system to the measuring
In this equation, TC and Tb (K) are the critical and boiling point
temperatures, pure feed gas was introduced in the MSB, step by
temperatures respectively, mb (cm3 mol
1) and b(cm3 mol
1) are
step, waiting for equilibrium to be achieved in between steps.
the molar volume at Tb and the van der Waal’s volume, respectively
The adsorbent weight differences, the pressure increments and
[68]. However, above the critical temperature, m is assumed equal
the temperature inside the adsorption chamber were recorded
to the van der Waal’s volume (see Eq. (4)) [68]:
continuously by a computer.
Prior to any adsorption experiments, the adsorbent sample was v¼b ð4Þ
placed in the basket to be activated, under vacuum at 583 K.
The buoyancy effect was taken in consideration and corrected,
assuming that the density of the adsorbed phase is equal to the
4. Theoretical section
density of the liquid. In this way, the adsorption capacity can be
calculated from Eq. (1):
4.1. Adsorption equilibrium model
Dm þ ð V s þ V c Þ  q g ql
q¼ ð1Þ
ms M ql
qg Adsorption equilibrium data is one of the most important infor-
mation for modelling and simulation of adsorption based separa-
where Dm represents the mass difference between the mass given tion processes. In this way adsorption equilibrium of CH4 and
by the microbalance and the initial mass of the basket loaded with CO2 was measured at several temperatures, in a pressure range
the activated sample; Vs is the volume of the solid adsorbent, Vads is of 0–80 bars, using 13X zeolite binderless beads as adsorbent.
the volume of adsorbed gas, Vc is the volume of inert parts and qg is There are several adsorption equilibrium models that can be
the density of the gas at measuring conditions. More details about used to represent the experimental adsorption equilibrium results.
the buoyancy effect correction methodology can be found else- However, the model selected in this study, was Dual Site Langmuir
where [66]. (DSL) model, since it was the one that best represented the exper-
The adsorbed phase density (ql) can be calculated from the imental data (see Eq. (5)) [69].
molar mass and the molar volume (see Eq. (2)).
b1 P b2 P
M qi ¼ qsat;1 þ qsat;2 ð5Þ
ql ¼ ð2Þ 1 þ b1 P 1 þ b2 P
v
Several semi-empirical methods for estimating the molar vol- where qi is the adsorbed phase concentration, qsat is the saturation
ume, which are temperature dependent, have been proposed. capacity, b is the affinity constant, P is the partial pressure. This
One of such methods is given by Eq. (3) which was adopted in this model assumes that the adsorbent surface is heterogeneous and
work [67,68]. takes into account two different types of adsorption sites [69].
346 M.A. Moreira et al. / Separation and Purification Technology 173 (2017) 339–356

Since the experimental adsorption equilibrium results were This theory assumes that the spreading pressures calculated
obtained at a large temperature range, a temperature dependence from the Gibbs adsorption isotherms are equal for the two compo-
of the equilibrium parameters is also required. Each parameter b nents at equilibrium [70,71,73]. The basic equation for IAST can be
can be obtained by [70]: written as:
 

DH yi P ¼ xi Poi ðpÞ ð10Þ
b ¼ b0 exp ð6Þ
Rg  T
where yi and xi are the gas phase and the adsorbed phase mole frac-
where b0 is the affinity at infinite temperature, Rg is the ideal gas tions for component i, respectively, p is the spreading pressure of
constant and
DH is the heat of adsorption. Each temperature the adsorbed phase, and Pi0 is the hypothetical pressure of a pure
dependency of the maximum adsorption capacity (qsat) can be component that gives the same spreading pressure than the
described by the empirical correlation shown in Eq. (7), where T0 adsorbed mixture on the surface. On the other hand, spreading
is a reference temperature [70]. pressure can be calculated by the following equation:
   Z
T pA P01
q0i ðPÞ
qsat ¼ q0 exp v 1
ð7Þ ¼ dP ð11Þ
T0 Rg T 0 P
In order to fit the experimental data to the DSL model, it was where qi0 is obtained by the pure-component isotherm. This theory
necessary to adjust eight parameters (b0,1; b0,2; q0,1; q0,2; DH1; also assumes that the adsorbed solution is defined by a pure-
DH2; v1; v2), for each component. component standard state (see Eq. (12)).
The isosteric heat of adsorption (
DH) represents the interac-
tion strength between the adsorbate molecules and the adsorbent P ¼ ci xi Pref
i ð12Þ
surface, giving valuable information about the energetic hetero- where cI is the activity coefficient (1 for ideal solution) and xi
geneity of the adsorbent surface. This parameter can be calculated is the mole fraction of component i. Pref is the standard state, at
i
from the Classius–Clapeyron Eq [70]: the temperature and pressure of the mixture. It is important to
  note that the reduced spreading pressure for each component
DH @ ln P

¼ ð8Þ should be equal at equilibrium, and must obey the following
RT 2 @T q
constrains.
X
xi ¼ 1 and ci ¼ 1 ð13Þ
4.2. Multicomponent adsorption equilibrium model
The total amount of adsorbed gases qi depends on the amount
In order to design an industrial adsorptive separation process, it adsorbed in the standard state qi0 and can be expressed according
is necessary to have a reliable multicomponent adsorption equilib- to the following relationship:
rium model. The adsorption equilibrium data of gas mixtures can
1 X xi
be determined experimentally. However, as the number of compo- ¼ ð14Þ
nents increases, the required experimental effort becomes more q q0i ðPÞ
demanding. Therefore, one needs to be capable of predicting how It should be mentioned that the set of equations presented pre-
each species would interact with other species and the adsorbent viously were solved by numerical integration using an algorithm
surface, from pure-component isotherm data of the individual [70]. The success of the calculation of IAST depends on how well
adsorbates [71]. the single component data are fitted, especially in low pressure
The multicomponent adsorption equilibrium data for CO2 and region [70].
CH4 was predicted by the extended Dual Site Langmuir (DSL)
model (see Eq. (9)): 4.3. Mathematical model
b1;i Pi b2;i Pi
qi ¼ qsat;1;i P þ qsat;2;i P ð9Þ A cyclic process in which carbon dioxide is adsorbed from a gas
1 þ nj¼1 b1;j Pj 1 þ nj¼1 b2;j Pj
stream on to the surface of a solid, is based in two simple steps:
where qsat,1,i and qsat,2,i is the adsorption saturation capacity of each adsorption, in which the more adsorbable species are selectively
site for component i; b1,i and b2,i are the affinity constants of com- removed from the feed gas; and regeneration (desorption), when
ponent i for each adsorption site; and Pi is the partial pressure of these species are removed from the adsorbent so that it can be
component i. ready for the next cycle [46,69].
As seen in the equation above, in order to fit the adsorption Several adsorption processes can be adopted to achieve CO2
equilibrium data in the wide range of temperatures and pressures separation depending on the regeneration methods. The most
reported in this work, a significant number of parameters (8 for common regeneration methods are vacuum or pressure swing
each compound) was required. For this reason, in order to have adsorption (VPSA), temperature swing adsorption (TSA) and elec-
some validation of the multicomponent adsorption equilibrium tric swing adsorption (ESA) [45,47,48].
prediction, the extended DSL model was compared with the Ideal The prediction of the dynamic behaviour of an adsorption sep-
Adsorbed Solution Theory (IAST) prediction (assuming this model aration process requires always a definition of a mathematical
represents well the system behaviour). model. The mathematical model to be used in this study was
The IAST was developed in 1965 by Myers and Prausnitz [71– developed taking into consideration the following assumptions:
73]. It is a general theory capable of predicting multicomponent
adsorption equilibrium from the pure-component adsorption iso-  real gas behaviour;
therm data. Adsorption equilibrium is represented not by a specific  axial dispersed flow;
equation, but by a system of equations [71–73]. IAST approach is  no mass, heat or velocity gradients in the radial direction;
based upon three assumptions: all adsorbates have the same avail-  external mass and heat transfer resistances expressed with the
able surface area; the adsorbent is inert; and, multicomponent film model;
mixture behaves as an ideal solution at constant pressure and tem-  particle mass transfer resistance, expressed with the Linear
perature [73]. Driving Force (LDF) model;
 M.A. Moreira et al. / Separation and Purification Technology 173 (2017) 339–356 347

 no temperature gradients inside the particles as the heat trans-  the Ergun equation is valid locally: in the momentum balance,
fer in the solid particles is much higher than in the gas phase; only the terms of pressure drop and velocity change are
 the column wall interchanges energy with the gas phase inside considered.
the column but not with the external environment;
 constant porosity along the bed; The simulation of the adsorption separation process was per-
formed considering a large variation in the bed temperature, as
Table 7 well as, high pressure operation. In this way, it cannot be assumed
Mass, energy and momentum balance equations of the mathematical model of a fixed that the gas phase will behave as an ideal gas. In order to obtain
bed adsorption system.
more accurate and realistic results, the GERG2008 Equation of
Gas Phase: State (EoS) will be used for the calculation of the thermodynamic
Material balance properties.
  
@ @yi @ @C g;i
@z eDax C g;T @z
@z ðu0 C g;i Þ
e @t
ð1
eÞap kf C g;i
C s;i ¼ 0
The model equations include material, momentum and energy
balances, and are presented in Table 7. In order to solve the system
Energy balance
  of partial differential equations, boundary and initial conditions are
@T @T @C
@
k @zg
u0 C g;T C p @zg þ eZRg T g @tg;T
ð1
eÞap hf ðT g
T p Þ
4hw
ðT g
T w Þ
@z dwi needed. Considering a fixed bed operation, where pressure is con-
@T g

eC g;T C V @t ¼0 trolled at the end of the column, starting with a column filled with
Momentum balance a given composition, the initial and boundary conditions are pre-
150 lð1
eÞ 2
eÞq sented in Table 8.

@P
@z ¼ u0 þ 1:75ð1

e3 d p ju0 ju0
3 2 e dp
In the case of cyclic operation, the modelling of each step,
Real gas equation involves the same conservation equations coupled with the corre-
P ¼ ZC g;T Rg T g
spondent boundary conditions. The appropriate boundary condi-
Solid Phase: tions for each step are given in Table 9.
Material balance The transport parameters required within the model were cal-
ap kf 

qap ðC g;i
C s;i Þ ¼ kh qi
qi
culated by correlations. In this way, axial dispersion coefficients
and mass and heat transfer coefficients were obtained using the
Linear driving force model (LDF)
@qi Wakao and Funazkri correlations [49,74]. The convective heat
@t ¼ kh ðqi
qi Þ
transfer coefficient between the gas and the column wall was cal-
Energy balance culated with the Wasch and Froment correlation [75]. The molec-
h P i
b p;s @T p ¼ q Pn ð
DH Þ @qi
ð1
eÞ qap ni¼1 qi C v ;ads;i þ qap C @t b i¼1 ads i @t ular diffusivities were calculated with the Chapman–Enskog
þð1
eÞap hf ðT g
T p Þ equation [76]. While, the molar specific heat of the adsorbed gas
was assumed to be equal to the molar specific heat in the gas phase
Column wall:
[77]. General properties of the gases, like density, viscosity, and
Energy balance
molar specific heat were calculated by the REFPROP (REFerence
qw Cb p;w @T@tw ¼ aw hw ðT g
T w Þ
fluid PROPerties) software package [78] using GERG-2008 EoS.
The homogeneous mass transfer coefficients (kh) for the binder-
Table 8 less 13X zeolite were estimated from the uptake curves measured
Boundary and initial conditions of the mathematical model of an adsorption fixed in the MSB. The following equations, one for each compound, were
bed. used in the mathematical model (kh in s
1).
Boundary conditions:
2
kh;CO2 ¼ 1:4  10
4 e1:710 T
ð15Þ
Inlet (z = 0) Outlet (z = L)
u0inlet C inlet;i ¼ u0 C g;i
eDax C g;T @yi
@z
@C g;i
@z ¼0
2
u0inlet C inlet;T ¼ u0 C g;T P = Pexit kh;CH4 ¼ 2:7  10
4 e1:410 T
ð16Þ
@T @T g
u0inlet C inlet;T C p T inlet ¼ u0 C g;T C p T g
k @zg @z ¼0

Initial conditions:
yi ¼ initial composition
4.3.1. Thermodynamic equation of state (EoS)
@qi
¼0
Equations of state (EoS) can be used to calculate thermody-
@t
T g ¼ T p ¼ T w ¼ T inlet namic properties, from the development of empirical equations
of state for pure substances. The models allow the accurate

Table 9
Boundary conditions of the mathematical model for each step of the Cryo-PTSA cycle.

Step z=0 z=L

Co-current pressurization u0inlet C inlet;T ¼ u0 C g;T u0 ¼ 0

eDax C g;T @yi
@z þ u0 C g;i ¼ u0inlet C inlet;i @C g;i
@z ¼ 0
@T @T g

k @zg þ u0 C g;T C p T g ¼ u0inlet C inlet;T C P T inlet @z ¼ 0

Adsorption u0inlet C inlet;T ¼ u0 C g;T P ¼ P exit

eDax C g;T @yi
@z þ u0 C g;i ¼ u0inlet C inlet;i @C g;i
@z ¼ 0
@T @T g

k @zg þ u0 C g;T C p T g ¼ u0inlet C inlet;T C P T inlet @z ¼ 0

Blowdown P ¼ P exit u0 ¼ 0
@C g;i @C g;i
@z ¼ 0 @z ¼ 0
@T g @T g
@z ¼ 0 @z ¼ 0

Heating P ¼ P exit u0inlet C inlet;T ¼ u0 C g;T

@C g;i
@z ¼ 0

eDax C g;T @yi
@z þ u0 C g;i ¼ u0inlet C inlet;i
@T g @T g
@z ¼ 0
k @z þ u0 C g;T C p T g ¼ u0inlet C inlet;T C P T inlet
348 M.A. Moreira et al. / Separation and Purification Technology 173 (2017) 339–356

description of the thermodynamic properties of mixtures for a gas cannot be considered an ideal gas. In this way, is possible to
range of temperature of 60–700 K and pressure up to 70 MPa claim that the thermodynamic properties, with more influence in
[79,80]. the simulation process, are the compressibility factor and the heat
Peng-Robinson (PR), Patel-Teja (PT), Redlich-Kwong (RK) and capacities at constant volume and pressure.
Soave-Redlich-Kwong (SRK) EoS are widely used in the petroleum The simulation time is very high and demanding when the sim-
and gas industries, especially for refinery and reservoir simulation. ulation program interacts with the property and thermodynamic
The major advantage of these models is the little input information tool program; therefore, the behaviour of the simulation program
requirement, leading to little computer time consumption and will be tested under different scenarios. Breakthrough curves were
good phase equilibrium predictions [80]. However, some of the simulated, based on the case study mentioned previously and
thermodynamic properties are not well predicted leading to inac- using the mathematical model described before, considering five
curate predictions [80]. different scenarios to calculate the thermodynamic parameters:
Recently the Groupe Européen de Recherches Gazières (GERG)
presented a new EoS, adopted as an ISO Standard (ISO 20765- 1. all thermodynamic parameters were considered constant
2/3) for natural gases calculations in 2013, and called GERG-2008 throughout the simulation and equal to the values obtained
equation of state [79,81]. for the feed mixture;
The GERG-2008 equation of state (EoS) is an upgraded version 2. the compressibility factor and the heat capacities at constant
of the GERG-2004 equation, which is explicit in the Helmholtz free volume and pressure were considered variables and calculated
energy as a function of density, temperature and composition. In locally through the use of polynomial equations.
addition to the 18 components covered by GERG-2004, the 3. all thermodynamic parameters were considered constant
GERG-2008 EoS considers three more components: n-nonane, n- except the compressibility factor, that was calculated locally
decane, and hydrogen sulphide [81]. by the REFEPROP using the GERG-2008 EoS;
The basis for the development and evaluation of the empirical 4. all thermodynamic parameters were considered constant
equations of state for mixtures are experimental data. The normal except the compressibility factor and the heat capacities at con-
range of validity of GERG-2008 is temperatures from 90 K to 450 K stant volume and pressure, that were calculated locally by the
and pressures up to 35 MPa, however the extended validity range REFEPROP using the GERG-2008 EoS;
reaches from temperatures of 60 K to 700 K and up to 70 MPa in 5. all thermodynamic parameters (compressibility factor, viscos-
pressure. In this way, GERG-2008 EoS will be used for the calcula- ity, density, thermal conductivity and heat capacities at con-
tions of mixtures properties of natural gas fluids, in order to per- stant volume and pressure) were considered variables and
form the process simulation. More detailed information about calculated locally by the REFEPROP using the GERG-2008 EoS.
this thermodynamic EoS can be found somewhere else [79,81].
In order to employ the GERG-2008 EoS, REFPROP (REFerence Polynomial equations offer a compromise between generality
fluid PROPerties) software package was used. This property pack- and simplicity that is suitable to many purposes [84]. Therefore,
age was developed by the National Institute of Standards and Tech- compressibility factor and heat capacities at constant volume and
nology (NIST) and allows the calculation of thermodynamic and pressure values were obtained with GERG-2008 EoS and then fitted
transport properties of natural gas fluids and their mixture. The to polynomial equations (from this point forward named surface
equations are valid over the entire vapour and liquid regions of equations - SE). These SE were used to calculate these thermody-
the fluid, including supercritical states [78]. namic properties for CO2 and CH4 (pure components) and are pre-
The mathematical model described previously involves a sys- sented in Table 10.
tem of partial differential and algebraic equations (PDAEs), which where P and T are the pressure and temperature of operation,
can be solved using modelling software. The modelling software respectively, and the parameters a to o are the surface equation
chosen was gPROMSÒ ModelBuilder (PSE) [82], which is a reliable parameters.
simulation tool. gPROMSÒ provides a general interface that can Z, CP and CV values for pure components were obtained using
incorporate other external property and thermodynamic tools. In REFPROP software defining the GERG-2008 as the equation of state.
this way, REFPROP was integrated in the gPROMSÒ simulation tool From these values, the parameters of the equations were adjusted
by the use of REFPROP CAPE-OPEN (Computer Aided Process Engi- in order to minimize the errors between the EoS values and the
neering) physical properties socket [83]. ones obtained by the surface equation. In Table 11 the parameters
Consequently the process simulation will be a dynamic and cyc- values to be used in the surface equations are presented, for pure
lic procedure. The information provided by gPROMSÒ (pressure, CH4 and CO2 at a pressure up to 80 bar and from 180 to 480 K.
temperature and mixture composition) will be sent to REFPROP The values calculated for pure CH4 and pure CO2 with the GERG-
in order to calculate the physical properties, at a specific time 2008 EoS and surface equation are compared in Fig. 4 in function of
and position in the fixed bed column. On the other hand, the phys- the temperature and pressure. As can be seen, the SE can estimate
ical properties calculated by REFPROP will be sent to gPROMSÒ that fairly well the compressibility factor and heat capacities at con-
uses them to solve once more the mathematical model, giving new stant volume and pressure values obtained in the REFPROP pro-
pressure, temperature and composition values that will be used in
REFPROP.
The cyclic separation process considered is a nonisothermal Table 10
Compressibility factor and heat capacities at constant volume and pressure surface
adsorption process, which will originate a temperature wave as
equations.
well as concentration waves that propagates through the bed.
The relative velocities of the two waves are determined by the Compressibility factor
adsorption and physical properties as well as the carrier gas flow Z ¼ a þ Tb þ cP þ d PT þ e TP2
Heat capacity at constant pressure
rate. Depending on the magnitude of these values, different shapes 2 2 2 3 4
C P ¼ a þ bT þ cP þ dT þ eTP þ fP þ gT 3 þ hT P þ jTP 2 þ kP þ lT þ mT 3 P
and combinations of concentration and temperature waves are
þnT 2 P 2 þ oTP 3
possible. Since the velocity of the temperature wave is derived
Heat capacity at constant volume
from the heat balance, this one is severely influenced by the heat 2 2 2 3 4
C V ¼ a þ bT þ cP þ dT þ eTP þ fP þ gT 3 þ hT P þ jTP 2 þ kP þ lT þ mT 3 P
capacity of the fluid [49]. On the other hand, for the operating con-
þnT 2 P 2 þ oTP 3
ditions considered in the proposed cyclic separation process, the
 M.A. Moreira et al. / Separation and Purification Technology 173 (2017) 339–356 349

Table 11
Parameters values of the surface equations for the compressibility factor and heat capacities at constant volume and pressure.

SE parameter Z Cp CV
CH4 CO2 Units CH4 CO2 Units CH4 CO2 Units
a 0.96 0.99 – 247.7 157.9 J mol
1 K
1 50.58 40.7 J mol
1 K
1
b 13.1 1.71 K
3.04
1.82 J mol
1 K
2
0.34
0.30 J mol
1 K
2
c 0.003
0.005 bar
1 8.53 23.2 J mol
1 K
1 bar
1 0.90 4.90 J mol
1 K
1 bar
1
d
0.50 5.61 K bar
1 0.015 0.009 J mol
1 K
3 0.002 0.002 J mol
1 K
3
e
260.2
1782 K2 bar
1
0.077
0.19 J mol
1 K
2 bar
1
0.007
0.037 J mol
1 K
2 bar
1
f – – 0.041 0.18 J mol
1 K
1 bar
2 2.83  10
4 0.014 J mol
1 K
1 bar
2
g – – –
3.32  10
5
1.98  10
5 J mol
1 K
4
2.77  10
6
3.43  10
6 J mol
1 K
4
h – – – 2.23  10
4 5.22  10
4 J mol
1 K
3 bar
1 1.88  10
5 9.11  10
5 J mol
1 K
3 bar
1
j – – –
1.53  10
4
8.68  10
4 J mol
1 K
2 bar
2 3.11  10
6
5.92  10
5 J mol
1 K
2 bar
2
k – – –
2.02  10
4
1.08  10
5 J mol
1 K
1 bar
3
1.10  10
5
2.15  10
5 J mol
1 K
1 bar
3
l – – – 2.58  10
8 1.51  10
8 J mol
1 K
5 1.95  10
9 2.48  10
9 J mol
1 K
5
m – – –
2.05  10
7
4.64  10
7 J mol
1 K
4 bar
1
1.59  10
8
7.50  10
8 J mol
1 K
4 bar
1
n – – – 1.06  10
7 1.01  10
6 J mol
1 K
3 bar
2
9.09  10
9 6.12  10
8 J mol
1 K
3 bar
2
o – – – 5.69  10
7 7.60  10
8 J mol
1 K
2 bar
3 2.69  10
8 5.39  10
8 J mol
1 K
2 bar
3

gram using the GERG-2008 EoS. For Cp values, it was found a max- In order to verify if the sample structure and capacity was not
imum error of 7.0% at 80 bar and 480 K for CO2, and 13.2% at 55 bar compromised during the adsorption measurements, four points
and 210 K for CH4. In the case of CV, a maximum error of 4.5%, at of the first of all measured adsorption isotherm (CO2 at 473 K)
30 bar and 270 K, for CO2 and 2.7%, at 25 bar and 180 K, for CH4 were repeated at the end of all measurements. As can be seen in
was found. For the Z values, a maximum error of 4.6% for CO2, at Fig. 5(b), the repetition points (Rep (473 K) - dark blue) are in good
70 bar and 320 K, and 6.7% for CH4, at 50 bar and 210 K, was found. agreement with the measured isotherm. In this way it is possible to
According to Dalton’s law, each component acts independently conclude that the capacity of the adsorbent was not compromised
in its contribution to the total pressure. Likewise, compressibility during the adsorption measurements.
factor and heat capacities at constant volume and pressure for a The adsorption isotherms for CH4 and CO2 are reversible since
mixture can be described as the sum of each pure component value no hysteresis is observed. In Fig. 5 it is possible to observe that
weighted by the respective mole fraction [84,85]. In this way, the the closed symbols, measured during the adsorption steps, are
compressibility factor and the heat capacities at constant volume aligned with the open symbols, which correspond to the points
and pressure for mixtures of CO2 in CH4 were calculated using measured during the desorption steps. It is also noted that the
the adjusted surface equations (see Table 10) and considering the adsorbent presents higher adsorption capacity for carbon dioxide
Eqs. (17)–(19), for mixtures containing up to 40% of CO2 in CH4 than for methane.
at a pressure up to 80 bar and from 180 to 480 K. The isosteric heat of adsorption for each component was deter-
mined from the experimental adsorption data of CH4 and CO2 over
X
n
Z¼ yi Z i ð17Þ binderless zeolite 13X as function of the loading (see Fig. 6). The
i¼1 obtained values for the isosteric heats of adsorption are in agree-
ment with the values founded in the literature, 17.2 kJ/mol and
X
n 35.4 kJ/mol for CH4 and CO2 respectively [62].
CP ¼ yi C P;i ð18Þ
i¼1
5.2. Multicomponent adsorption equilibrium isotherms

X
n
CV ¼ yi C V;i ð19Þ In real separation processes, there are multiple gas species
i¼1 involved. Therefore, it is important to have adsorption equilibrium
models capable of predicting how each species would interact with
The use of these much simpler mixing rules introduces some other species and the adsorbent surface. To accomplish this, it is
errors in the calculation, being of 6.6%, 14.8% and 4.5% for Z, Cp necessary to predict multicomponent adsorption equilibria for
and Cv, respectively at the lowest temperature and highest pres- CH4 and CO2.
sure, when compared with the GERG-2008 EoS, for which errors Based on the single component isotherm DSL fitting parameters
of the order of 1–3% are reported. However, the advantages from (see Table 12), extended DSL model was used to predict the multi-
mathematical/simulation point of view are very significant as is component adsorption equilibrium isotherms. This model was
shown in the following section. then compared with the IAST prediction in order to conclude if a
good prediction was achieved.
5. Results and discussion As example, in Fig. 7 the comparison between the multicompo-
nent extension of the DSL model with the correspondent IAST pre-
5.1. Adsorption equilibrium isotherms diction is presented, considering the feed conditions of the case
study proposed (190 K and 40 bar). As can be observed, the multi-
The adsorption equilibrium isotherms of carbon dioxide and component DSL extension is in good agreement with the IAST pre-
methane on binderless zeolite 13X beads, obtained at 180, 220, diction. In this way, multicomponent DSL extension will be used to
269, 323, 373 and 473 K, and up to 80 bar are presented in Fig. 5. perform all simulations.
The adsorption equilibrium data is given in table in the Support
Information. The experimental data was fitted with the DSL model. 5.3. Breakthrough curves simulation
The fitting parameters can be found in Table 12. As can be seen the
experimental data was well described by the DSL model shown as In the Cryo-PTSA cycle, it was considered a cooling step and a
lines in Fig. 5. heating step. In this way, breakthrough simulations based in these
350 M.A. Moreira et al. / Separation and Purification Technology 173 (2017) 339–356

Fig. 4. Comparison between the values calculated with the GERG-2008 EoS and Surface Equation, in function of the temperature and pressure, for: (a) compressibility factor
for pure CH4; (b) compressibility factor for pure CO2; (c) heat capacity at constant pressure for pure CH4; (d) heat capacity at constant pressure for pure CO2; (e) heat capacity
at constant volume for pure CH4; (f) heat capacity at constant volume for pure CO2.
 M.A. Moreira et al. / Separation and Purification Technology 173 (2017) 339–356 351

Fig. 6. Isosteric heats of adsorption for methane and carbon dioxide on binderless
zeolite 13X. Symbols represent the values calculated using the experimental data
and the lines are for the literature values [62].

Fig. 5. (a) Adsorption equilibrium isotherms on binderless zeolite 13X beads for (a)
methane and (b) carbon dioxide at 180, 220, 269, 323, 373 and 473 K. Symbols
represent the experimental data and solid lines represent the Dual-Site Langmuir Fig. 7. Example of the comparison between the multicomponent extension of the
isotherm fit. Dual-Site Langmuir model with correspondent IAST prediction, at 190 K and 40 bar,
on binderless zeolite 13X.

Table 12
Parameters of the Dual-Site Langmuir isotherm for binderless zeolite 13X beads. Table 13
Parameter values used in the fixed bed simulations.
Parameter CH4 CO2 Units
Feed conditions Cooling step Heating step Units
b1,0 2.9  10
4 4.1  10
4 bar
1
b2,0 1.4  10
5 1.3  10
5 bar
1 P 40 5 bar

D H1 15.8 23.0 kJ/mol T 190 473 K

D H2 1.9 19.0 kJ/mol Fin 2300 200 mol/s
q1,0 3.47 5.78 mmol/g yin CH4 0.991 0.99995 –
q2,0 23.6 1.16 mmol/g yin CO2 0.009 50 ppm –
v1 0.90 0.51 – Fixed bed characteristics
v2 0.002 0.23 – Lcol 6 6 m
T0 473 K dwi 3.5 3.5 m
e 0.40 0.40 –
Thermodynamic parameters at feed conditions
two steps were run, using the binderless zeolite 13X as adsorbent Z 0.5932 0.99949 –
and changing the calculation method of the thermodynamic l 9.22  10
5 6.29  10
4 Pa s
parameters, as described in the previous Section. q 73.3 2.04 kg/m3
During the cooling step a feed stream containing 0.9% of CO2 in Cp 118.5 45.0 J/mol K
Cv 33.3 36.6 J/mol K
CH4, at 190 K and 40 bar, is fed to the bed. It was considered that
kg 0.036 0.032 J/s m K
initially the bed is filled with a gas mixture with 50 ppm of CO2
Transport parameters at feed conditions
at 5 bar and 473 K. When the heating step is simulated, it was con- Dax 5.94  10
5 6.29  10
4 m2/s
sidered that the bed is initially filled with a gas mixture containing k 2.04 0.95 J/s m K
0.9% of CO2 at 5 bar and 190 K. And the column will be fed with a kf 1.16  10
3 4.56  10
2 m/s
heated stream containing 50 ppm of CO2 at 5 bar and 473 K. hf 340.1 309.6 J/s m2 K
J/s m2 K
The mathematical model was solved using the gPROMSÒ Mod- hw 996.3 482.8

elBuilder (PSE), using the orthogonal collocation on finite elements

(OCFEM) as numerical method. The values of the parameters used
in the model are presented in Table 13. GERG-2008 EoS. In this case each thermodynamic parameter is cal-
The scenario that presents more realistic results, is the one culated locally and the simulations provide more accurate results,
where all thermodynamic parameters are calculated with the being the reference case when all cases are compared.
352 M.A. Moreira et al. / Separation and Purification Technology 173 (2017) 339–356

In Fig. 8 the breakthrough curve simulation results for the cool- Fig. 8(a) and(c)). Furthermore, it is also possible to observe, that
ing step, considering the five scenarios for the thermodynamic Z, Cp and Cv had a great influence in the simulation of the gas tem-
properties calculations are presented. During the cooling step the perature (see Fig. 8(b)). When all thermodynamic parameters are
temperature wave travels faster than the concentration wave. As considered constant or when only Z is obtained by the GERG-
can be seen in Fig. 8(a) and (b), at the time of 310 s, the tempera- 2008 EoS, the velocity of temperature wave is higher. However,
ture wave has reached the middle of the column while the concen- when Z, CP and CV are considered variable, either through the SE
tration wave is still at approximately 0.5 m. However, it is possible or GERG-2008 EoS, the temperature waves assume a similar beha-
to observe that all concentration waves are quite similar indepen- viour to the case where all thermodynamic parameters are calcu-
dently of the thermodynamic scenario that was considered (see lated with the GERG-2008 EoS.
During the heating step, the concentration and temperature
waves propagate together through the column (see Fig. 9). Once
more, when the thermodynamic parameters were considered all
constants or just Z is variable, it is possible to see some differences
between in temperature and concentration waves velocity when
compared to the other three scenarios.
Regarding all these results it is possible to conclude that, when
all thermodynamic parameters are considered constants or only Z
is variable, leads to simulations results that are deviating from the
more realistic case (all thermodynamic parameters calculated
through the use of GERG-2008 EoS). It is also possible to assess that
the SE are quite accurate, leading to similar results of those
obtained with simulations performed considering the GERG-2008
EoS.
Despite the similarity between the results, the computational
times necessary to perform the simulations are quite different. In
Table 14 is presented a summary of the breakthrough simulations

Fig. 8. Results of the breakthrough curve simulation of the adsorption of 0.9% Fig. 9. Results of the breakthrough curve simulation of the adsorption of 50 ppm of
carbon dioxide in methane at 190 K and 40 bar over a bed initially filled with carbon dioxide in methane at 473 K and 5 bar over a bed initially filled with 0.9% of
50 ppm of carbon dioxide in methane, using binderless 13X zeolite: (a) CO2 molar carbon dioxide in methane, using binderless 13X zeolite: (a) CO2 molar fraction
fraction along the bed at time of 310 s, (b) temperature of the gas along the bed at along the bed at time of 4000 s, and (b) temperature of the gas along the bed at time
time of 310 s, and (c) CO2 molar fraction along the bed at time of 4000 s. of 4000 s.
 M.A. Moreira et al. / Separation and Purification Technology 173 (2017) 339–356 353

Table 14 5.4. Design and simulation of an industrial scale Cryo-PTSA process

Computational time necessary to complete the breakthrough curve simulations.

Thermodynamic parameter calculation method Simulation time, s The first step towards the design and optimization of a cyclic
All constant 22 adsorptive process is the sizing of the adsorption column to be
Z, CP and CV by SE 111 used. Due to limitations in the transport of large PSA vessels, the
Z by GERG-2008 EoS 2770 diameter of these vessels is usually limited to 4 m [86] and so a
Z, CP and CV by GERG-2008 EoS 74,033 3.5 m bed diameter was assumed.
All variable by GERG-2008 EoS 69,391
Based in a proposed design procedure described in the litera-
Simulations performed with an IntelÒ CoreTM 7-4770 CPU @ 3.40 GHz Processor. ture [87], and according with the feed molar flow rate specified
in Section 2 (Fin = 8280 kmol/h), a superficial velocity (uo) of
0.05 m s
1 was found. This value is within the range of typical feed
performed considering the cooling step with the different thermo- superficial velocities which is 0.01–0.05 m/s. In order to obtain the
dynamic properties calculation method, as well as the necessary column length, it was necessary to have in consideration that, due
time to complete them. In all cases, the numerical method used to limitations in the mass transfer and temperature effects, the
was the orthogonal collocation on finite elements with fourth effective use of the adsorbent will not be 100%. In this way, an
order polynomials and 39 intervals. effective use of the adsorbent of 65% and a bed porosity of 0.4
When all the properties are kept constant, or when the Z, CP and was assumed. Setting all this information with an adsorption time
CV are calculated by the surface equation, the simulations are quite (tads) of 8500 s, a carbon dioxide feed molar flow rate of
fast. However, when all properties are calculated by the GERG- 74.52 kmol/h, a working capacity value of the bed of 8.41 mol/kg
2008 EoS a much higher computational time is required to com- and an adsorbent apparent density of 1200 kg m
3, the volume of
plete the simulations. Even if only the Z parameter is calculated the adsorbent obtained was 54.2 m3. For a column with 3.5 m
by the GERG-2008 EoS the simulation times are quite considerable diameter, this gives a column length of approximately 6 m.
(2770 s). A similar tendency is observed when the breakthrough In order to test and observe the behaviour of the simulation
simulations were performed considering the heating step. program, breakthrough curves were simulated using the mathe-
Therefore, we can conclude that when Z, Cp and CV are calcu- matical model describe previously in the Section 4.3. It should be
lated with SE it is possible to obtain accurate simulations results noted that the mathematical model used in the present study
with a much lower computational time. In this way, SE were used has already been used in the simulation of fixed bed behaviour
to perform the Cryo-PTSA process simulation. and PSA applications of different mixtures under ideal conditions,

Fig. 10. Cryo-PTSA cycle proposed for the production of a purified CH4 stream, with a CO2 content lower than 50 ppm.
354 M.A. Moreira et al. / Separation and Purification Technology 173 (2017) 339–356

presenting accurate results and good agreement between predic- Table 15

tions and experimental data. Furthermore, the range of validity Parameter values used in the Cryo-PTSA simulations.

presented by GERG-2008 EoS covers the range of temperature Feed conditions Units
and pressure used in the present study. In this way, it was consid- Pin 40 bar
ered that the most realistic results are obtained when ‘‘all GERG” Tin 190 K
reference state is used. Fin 2300 mol/s
The operation of a PTSA process involves the definition of a yin CH4 0.991 –
yin CO2 0.009 –
cycle, that is, a sequence of elementary steps through which the
column undergoes. Based in the process requirements and in order Purge conditions
Pheat 5 bar
to obtain a light product of a purified methane stream with a low Theat 473 K
carbon dioxide content, a cycle with four steps was designed (see Fheat 200 mol/s
Fig. 10). The proposed cycle comprises a co-current pressurization Blowdown conditions
step with feed stream and an adsorption step (P + Ads), followed by Pbld 5 bar
a counter-current blowdown (bld) and a counter-current purge Transport parameters at feed conditions
step with a heated stream (heat). Dax 5.94  10
5 m2/s
During the pressurization and adsorption step, the feed stream k 2.04 J/s m K
is fed to the bed, producing an enriched methane stream with a kf 1.16  10
3 m/s
hf 340.1 J/s m2 K
CO2 content lower than 50 ppm at the column outlet. This step is
hw 996.3 J/s m2 K
followed by a blowdown step, where the system pressure is
decreased from 40 to 5 bar counter-currently. During the purge
step, part of the product stream is fed counter-currently to the used to calculated locally the Z, Cp and Cv values. Transport param-
bed at 5 bar and 473 K, allowing the additional desorption of CO2 eters, as well as stream conditions required in the mathematical
regenerating the bed. model are presented in Table 15.
The duration of the pressurization plus the adsorption step was Since the process described demands a high purity of the pro-
set equal to the blowdown plus the purge step. It was also consid- duct stream and the regeneration of the column is performed by
ered that the product stream could not have a CO2 content higher temperature change, at the end of the regeneration step the col-
than 50 ppm, and the time of the blowdown plus heating step umn is practically in the same conditions of the previous cycle,
allowed the column regeneration. In this way, it was considered and therefore the CSS is reached after only a few cycles.
a pressurization plus adsorption step of 8450 s, followed by a The CO2 molar fraction history obtained at the column outlet
blowdown step of 300 s and purge/heating step of 8150 s. In this (z = 0 m during the pressurization and adsorption steps and
way, the total cycle time is around 4.7 h, which is within the typ- z = 6 m during the blowdown and purge steps) and the gas temper-
ical length of a TSA cycle that ranges from several hours to over a ature history at the middle of the column (z = 3 m), during one
day [49]. cycle at CSS, are presented in Fig. 11. As can be seen in the molar
The mathematical model, mentioned in Section 4.3, was solved
in gPROMSÒ ModelBuilder (PSE), using the orthogonal collocation
on finite elements (OCFEM) as numerical method, with second
order polynomials and two hundred equal intervals. The simula-
tions were carried out until cyclic steady state (CSS) was reached.
The performance of cyclic adsorption processes is commonly
evaluated according to product purity and recovery. In this study,
Cryo-PTSA cycle was evaluated according to methane recovery
(Eq. (20)) and purity of the methane product. This last one was
described in terms of CO2 content (in ppm), as given by Eq. (21).
Methane productivity was also calculated according to Eq. (22).
Power consumption was obtained using Eq. (23).

R tads Rt
F CH4 ;out dt
0heat F CH4 ;in dt
CH4 Recov ery ¼ 0
R tads ð20Þ
0
F CH4 ;in dt

R tads
F CO2 ;out dt
CO2 Content ¼ 0
ð21Þ
X
n
R tads
0
F i;out dt
i¼1

R tads
F CH4 ;in dt
CH4 Productiv ity ¼ 0
 CH4 Recov ery ð22Þ
mads t cycle

2  t heat
Power consumption ¼ F heat  C p  DT  ð23Þ
t cycle
In order to perform the Cryo-PTSA simulations, it was consid-
ered that at the start of the simulation the column was filled with Fig. 11. Cryo-PTSA cycle simulation results, with blowdown step at 5 bar, using
a gas mixture of 50 ppm of CO2 in CH4 at the heating temperature binderless 13X zeolite: (a) carbon dioxide molar fraction history at the column
at low pressure (473 K and 5 bar). As mentioned before SE were outlet and (b) gas temperature history at the middle of the column (z = 3 m).
 M.A. Moreira et al. / Separation and Purification Technology 173 (2017) 339–356 355

fraction history during the adsorption step, all CO2 fed to the col- Competitividade e Internacionalização (POCI) – and by national
umn is adsorbed, resulting in a product stream with a CO2 content funds through FCT – Fundação para a Ciência e a Tecnologia. The
below 50 ppm. Then during the blowdown and heating step, the authors thank to Chemiewerk Bad Köstritz GmbH (CWK) for
column is regenerated, being possible to obtain a maximum CO2 providing of the binderless 13X zeolite beads; special thanks to
content of 14% during the heating step. The average CO2 content Kristin Gleichmann.
of the CO2 enriched stream is around 8.8%.
Observing the gas temperature history, it is possible to realize
Appendix A. Supplementary material
that, during the adsorption step, the bed achieves rapidly the feed
temperature value (190 K). Furthermore, during the blowdown
Supplementary data associated with this article can be found, in
step the temperature of the bed decreases, which is related with
the online version, at http://dx.doi.org/10.1016/j.seppur.2016.09.
the CO2 desorption.
044.
The simulation results show that the recovery of CH4 is high,
around 90.7%, being possible to obtain a product stream with
41.8 ppm of CO2 in CH4. It was also found a CH4 productivity of References
100.1 mol kg
1ads h

1
. When Cryo-PTSA simulation is compared with
the cryogenic distillation process described before, it is possible to [1] B. Shimekit, H. Mukhtar, Natural Gas Purification Technologies - Major
Advances for CO2 Separation and Future Directions, in: D.H. Al-Megren (Ed.),
conclude that higher methane recovery value is obtained (90.7% InTech, 2012.
against 88.5%). Comparing the power consumption of the two pro- [2] Union Gas Limited. Natural Gas. 2016 [cited 2016 January]; Available from:
cesses, a value of 2.2 MW is obtained for Cryo-PTSA process against <https://www.uniongas.com/about-us/about-natural-gas>.
[3] M. Songolzadeh, M.T. Ravanchi, M. Soleimani, Carbon dioxide capture and
22.3 MW necessary for cryogenic distillation process, which corre- storage: a general review on adsorbents, Int. J. Chem., Mol., Nucl., Mater.
spond to a decrease of the total power consumption of 14.7%. The Metall. Eng. 6 (10) (2012) 900–907.
results obtained by process simulation showed that it is possible to [4] P. Riemer, Greenhouse gas mitigation technologies, an overview of the CO2
capture, storage and future activities of the IEA Greenhouse Gas R&D
replace the 3rd column of the cryogenic distillation process by a programme, Energy Convers. Manage. 37 (6–8) (1996) 665–670.
Cryo-PTSA achieving higher CH4 recovery and lower energy [5] J.-R. Li, Y. Ma, M.C. McCarthy, J. Sculley, J. Yu, H.-K. Jeong, P.B. Balbuena, H.-C.
consumption. Zhou, Carbon dioxide capture-related gas adsorption and separation in metal-
organic frameworks, Coord. Chem. Rev. 255 (15–16) (2011) 1791–1823.
[6] C.-H. Yu, C.-H. Huang, C.-S. Tan, A review of CO2 capture by absorption and
adsorption, Aerosol Air Qual. Res. 12 (2012) 745–769.
6. Conclusions [7] Z. Zhao, X. Cui, J. Ma, R. Li, Adsorption of carbon dioxide on alkali-modified
zeolite 13X adsorbents, Int. J. Greenhouse Gas Control 1 (3) (2007) 355–359.
Single adsorption equilibrium isotherms of methane and carbon [8] N.K. Jensen, T.E. Rufford, G. Watson, D.K. Zhang, K.I. Chan, E.F. May, Screening
zeolites for gas separation applications involving methane, nitrogen, and
dioxide on binderless 13X zeolite beads were measured by gravi- carbon dioxide, J. Chem. Eng. Data 57 (1) (2012) 106–113.
metric method, at temperature range between 183 K and 473 K [9] G.C. Watson, N.K. Jensen, T.E. Rufford, K.I. Chan, E.F. May, Volumetric
up to 80 bar. The adsorption equilibrium model that best repre- adsorption measurements of N2, CO2, CH4, and a CO2 + CH4 mixture on a
natural chabazite from (5 to 3000) kPa, J. Chem. Eng. Data 57 (1) (2012) 93–
sented the experimental data is the Dual Site Langmuir (DSL)
101.
model. After comparing the multicomponent DSL extension with [10] J.W. Yoon, S.H. Jhung, Y.K. Hwang, S.M. Humphrey, P.T. Wood, J.S. Chang,
the IAST prediction, it is possible to conclude that they are in good Gas-sorption selectivity of CUK-1: a porous coordination solid made of
agreement. In this way, the extended DSL model was used to pre- cobalt(II) and pyridine-2,4- dicarboxylic acid, Adv. Mater. 19 (14) (2007)
1830–1834.
dict the multicomponent adsorption equilibrium. [11] C.A. Grande, R. Blom, Cryogenic adsorption of methane and carbon dioxide on
A pressure and temperature swing adsorption process at low zeolites 4A and 13X, Energy Fuels 28 (10) (2014) 6688–6693.
temperatures (Cryo-PTSA) was used in order to produce a product [12] S. Pakseresht, M. Kazemeini, M.M. Akbarnejad, Equilibrium isotherms for CO,
CO2, CH4 and C2H4 on the 5A molecular sieve by a simple volumetric
stream with a CO2 content lower than 50 ppm in CH4. In order to apparatus, Sep. Purif. Technol. 28 (1) (2002) 53–60.
obtain accurate simulations results with a lower computational [13] J.A. Delgado, M.A. Uguina, J.M. Gómez, L. Ortega, Adsorption equilibrium of
time, compressibility factor and heat capacities at constant volume carbon dioxide, methane and nitrogen onto Na- and H-mordenite at high
pressures, Sep. Purif. Technol. 48 (3) (2006) 223–228.
and pressure values were calculated by polynomial equations that [14] T.H. Saleman, G.Y. Watson, T. Rufford, P. Hofman, K.I. Chan, E. May, Capacity
were obtained by adjusting the values given by GERG-2008 EoS. It and kinetic measurements of methane and nitrogen adsorption on H+-
was possible to conclude that the values obtained with these equa- mordenite at 243–303 K and pressures to 900 kPa using a dynamic column
breakthrough apparatus, Adsorption 19 (6) (2013) 1165–1180.
tions were in good agreement with the ones obtained using the [15] J.-S. Lee, J.-H. Kim, J.-T. Kim, J.-K. Suh, J.-M. Lee, C.-H. Lee, Adsorption equilibria
GERG-2008 EoS presenting a maximum error of 6.6%, 14.8% and of CO2 on Zeolite 13X and Zeolite X/activated carbon composite, J. Chem. Eng.
4.5% for Z, Cp and Cv, respectively. A methane recovery of 90.7% Data 47 (5) (2002) 1237–1242.
[16] J.A. Dunne, M. Rao, S. Sircar, R.J. Gorte, A.L. Myers, Calorimetric heats of
and a product stream with 41.8 ppm of CO2 in CH4 is achieved after
adsorption and adsorption isotherms. 2. O2, N2, Ar, CO2, CH4, C2H6, and SF6 on
cyclic steady state is reached. It was also found a CH4 productivity NaX, H-ZSM-5, and Na-ZSM-5 Zeolites, Langmuir 12 (24) (1996) 5896–5904.
of 100.1 mol kg
1 ads h

1
. When Cryo-PTSA simulation is compared [17] G. Calleja, J. Pau, J.A. Calles, Pure and multicomponent adsorption equilibrium
with the cryogenic distillation process, it is possible to conclude of carbon dioxide, ethylene, and propane on ZSM-5 Zeolites with different Si/
Al ratios, J. Chem. Eng. Data 43 (6) (1998) 994–1003.
that higher methane recovery value is obtained. A lower power [18] Y.S. Bae, C.H. Lee, Sorption kinetics of eight gases on a carbon molecular sieve
consumption value is also found when Cryo-PTSA process at elevated pressure, Carbon 43 (1) (2005) 95–107.
(2.2 MW) is compared with cryogenic distillation process [19] G. Watson, E.F. May, B.F. Graham, M.A. Trebble, R.D. Trengove, K.I. Chan,
Equilibrium adsorption measurements of pure nitrogen, carbon dioxide, and
(22.3 MW). The results obtained by process simulation showed methane on a carbon molecular sieve at cryogenic temperatures and high
that it is possible to replace the 3rd column of the cryogenic distil- pressuresy, J. Chem. Eng. Data 54 (9) (2009) 2701–2707.
lation process by a Cryo-PTSA achieving higher CH4 recovery and [20] Y. Park, D.-K. Moon, Y.-H. Kim, H. Ahn, C.-H. Lee, Adsorption isotherms of CO2,
CO, N2, CH4, Ar and H2 on activated carbon and zeolite LiX up to 1.0 MPa,
lower energy consumption. Adsorption (2014) 1–17.
[21] S. Himeno, T. Komatsu, S. Fujita, High-pressure adsorption equilibria of
methane and carbon dioxide on several activated carbons, J. Chem. Eng. Data
Acknowledgments 50 (2) (2005) 369–376.
[22] D.P. Valenzuela, A.L. Myers, Adsorption Equilibrium Data Handbook, III ed.,
This work was financially supported by: Project POCI-01-0145- Prentice-Hall Inc., 1989.
[23] S.Y. Zhou Li, Zhou Yaping, An experimental study on the adsorption behavior
FEDER-006984 – Associate Laboratory LSRE-LCM funded by of gases crossing the critical temperature, Chin. J. Chem. Eng. 10 (3) (2002)
FEDER funds through COMPETE2020 – Programa Operacional 259–262.
356 M.A. Moreira et al. / Separation and Purification Technology 173 (2017) 339–356

[24] E.M. Strizhenov, A.V. Shkolin, A.A. Fomkin, V.A. Sinitsyn, A.A. Zherdev, I.A. [54] H. Rastelli, C.C. Chao, D.R. Garg, Selective adsorption of CO2 on zeolites, US
Smirnov, A.L. Pulin, Low-temperature adsorption of methane on microporous 4,775,396, 1988.
AU-1 carbon adsorbent, Prot. Metals Phys. Chem. Surfaces 50 (1) (2014) 15–21. [55] R.T. Maurer, Methane purification by pressure swing adsorption, US 5,171,333
[25] T.E. Rufford, G.C.Y. Watson, T.L. Saleman, P.S. Hofman, N.K. Jensen, E.F. May, A, 1992.
Adsorption equilibria and kinetics of methane + nitrogen mixtures on the [56] W.B. Dolan, M.J. Mitariten, CO2 rejection from natural gas, WO 2003/020674
activated carbon Norit RB3, Ind. Eng. Chem. Res. 52 (39) (2013) 14270–14281. A1, 2003.
[26] G. Férey, Hybrid porous solids: past, present, future, Chem. Soc. Rev. 37 (1) [57] J.A. Sawada, M.L. Babicki, A. Chiu, A. Boulet, S. Roy, E.J. Rode, Separation of
(2008) 191–214. carbon dioxide from other gases, US 7,828,877 B2, 2010.
[27] J. Long, O. Yaghi (Eds.), Themed issue: Metal–organic frameworks. Chemical [58] J. Toreja, B. VanNostrand, N. Chan, J.P. Dickinson, Rotary-Valve, Fast-Cycle
Society Reviews, vol. 38, The Royal Society of Chemistry, 2009, pp. 1201–1202. Pressure-Swing Adsorption Technology Allows West Coast Platform To Meet
[28] A. Boutin, F.o.-X. Coudert, M.A. Springuel-Huet, A.V. Neimark, G.r. Férey, A.H. Tight California Specifications And Recover Stranded Gas, 2011.
Fuchs, The behavior of flexible MIL-53(Al) upon CH4 and CO2 adsorption, J. f [59] N. Chan, M. Zhou, Xebec – A World powered by clean energy, 2011.
Phys. Chem. C 114 (50) (2010) 22237–22244. [60] IACX Energy LLC, IACX Energy Formed to Provide Small Scale Nitrogen and CO2
[29] J. Seo, N. Jin, H. Chun, Topologies of metal–organic frameworks based on Removal Solutions to Natural Gas Producers, Oil & Energy, 2007 [cited 2016
pyrimidine-5-carboxylate and unexpected gas-sorption selectivity for CO2, January]; Available from: www.iacx.com.
Inorg. Chem. 49 (23) (2010) 10833–10839. [61] J. Wills, M. Shemaria, M.J. Mitariten, Production of Pipeline Quality Natural Gas
[30] H. Hayashi, A.P. Cote, H. Furukawa, M. O’Keeffe, O.M. Yaghi, Zeolite A With the Molecular GateTM CO2 Removal Process 2009, Tidelands Oil
imidazolate frameworks, Nat. Mater. 6 (7) (2007) 501–506. Production Company & Guild Associates Inc.
[31] B. Chen, S. Ma, F. Zapata, F.R. Fronczek, E.B. Lobkovsky, H.-C. Zhou, Rationally [62] M.C. Campo, A.M. Ribeiro, A.F.P. Ferreira, J.C. Santos, C. Lutz, J.M. Loureiro, A.E.
designed micropores within a metal–organic framework for selective sorption Rodrigues, Carbon dioxide removal for methane upgrade by a VSA process
of gas molecules, Inorg. Chem. 46 (4) (2007) 1233–1236. using an improved 13X zeolite, Fuel Process. Technol. 143 (2016) 185–194.
[32] D.N. Dybtsev, H. Chun, S.H. Yoon, D. Kim, K. Kim, Microporous manganese [63] D. Berstad, P. Nekså, R. Anantharaman, Low-temperature CO2 Removal from
formate: a simple metal–organic porous material with high framework Natural Gas, Energy Procedia, 26(2nd Trondheim Gas Technology Conference),
stability and highly selective gas sorption properties, J. Am. Chem. Soc. 126 2012, pp. 41–48.
(1) (2003) 32–33. [64] K. Schumann, B. Unger, A. Brandt, F. Scheffler, Investigation on the pore
[33] J.-M. Gu, T.-H. Kwon, J.-H. Park, S. Huh, DABCO-functionalized metal-organic structure of binderless zeolite 13 shapes, Microporous Mesoporous Mater.
framework bearing a C2h-symmetric terphenyl dicarboxylate linker, Dalton 154 (2012) 119–123.
Trans. 39 (24) (2010) 5608–5610. [65] Rubotherm GmbH, Magnetic Suspension Balances, [cited 2016 January];
[34] S.-M. Zhang, Z. Chang, T.-L. Hu, X.-H. Bu, New three-dimensional porous metal Available from: <http://www.rubotherm.com/magnetic-suspension-balances.
organic framework with tetrazole functionalized aromatic carboxylic acid: html>.
synthesis, structure, and gas adsorption properties, Inorg. Chem. 49 (24) [66] A.F.P. Ferreira, J.C. Santos, M.G. Plaza, N. Lamia, J.M. Loureiro, A.E. Rodrigues,
(2010) 11581–11586. Suitability of Cu-BTC extrudates for propane–propylene separation by
[35] K.S. Walton, A.R. Millward, D. Dubbeldam, H. Frost, J.J. Low, O.M. Yaghi, R.Q. adsorption processes, Chem. Eng. J. 167 (1) (2011) 1–12.
Snurr, Understanding inflections and steps in carbon dioxide adsorption [67] M.M. Dubinin, M.M. Vishnyakova, E.D. Zaverina, E.G. Zhukovskaya, A.I.
isotherms in metal-organic frameworks, J. Am. Chem. Soc. 130 (2) (2007) 406– Sarakhov, Study of the adsorptional properties and secondary porous
407. structure of adsorbents with a molecular sieve action, Bull. Acad. Sci. USSR,
[36] P.D.C. Dietzel, V. Besikiotis, R. Blom, Application of metal-organic frameworks Div. Chem. Sci. 10 (8) (1961) 1293–1299.
with coordinatively unsaturated metal sites in storage and separation of [68] N.A. Al-Baghli, K.F. Loughlin, Adsorption of methane, ethane, and ethylene on
methane and carbon dioxide, J. Mater. Chem. 19 (39) (2009) 7362–7370. titanosilicate ETS-10 Zeolite, J. Chem. Eng. Data 50 (3) (2005) 843–848.
[37] S. Ma, X.-S. Wang, E.S. Manis, C.D. Collier, H.-C. Zhou, Metal–organic [69] D.M. Ruthven, Principles of Adsorption and Adsorption Processes, Jonh Wiley &
framework based on a trinickel secondary building unit exhibiting gas- Sons, Inc., New York, 1984, pp. 401–405.
sorption hysteresis, Inorg. Chem. 46 (9) (2007) 3432–3434. [70] D.D. Do, Adsorption analysis: equilibria and kinetics, in: R.T. Yang (Ed.),
[38] H.J. Park, M.P. Suh, Stepwise and hysteretic sorption of N2, O2, CO2, and H2 Chemical Engeneering, Imperial College Press, London, 1998.
gases in a porous metal-organic framework [Zn2(BPnDC)2(bpy)], Chem. [71] H. Brenner, Adsorption Calculations and Modelling, Elsevier Science, 2013.
Commun. 46 (4) (2010) 610–612. [72] J. Chen, L.S. Loo, K. Wang, An Ideal absorbed solution theory (IAST) study of
[39] H. Chun, J. Seo, Discrimination of small gas molecules through adsorption: adsorption equilibria of binary mixtures of methane and ethane on a
reverse selectivity for hydrogen in a flexible metal–organic framework, Inorg. templated carbon, J. Chem. Eng. Data 56 (4) (2011) 1209–1212.
Chem. 48 (21) (2009) 9980–9982. [73] A.L. Myers, J.M. Prausnitz, Thermodynamics of mixed-gas adsorption, AIChE J.
[40] S. Ma, D. Sun, X.-S. Wang, H.-C. Zhou, A mesh-adjustable molecular sieve for 11 (1) (1965) 121–127.
general use in gas separation, Angew. Chem. Int. Ed. 46 (14) (2007) 2458– [74] N. Wakao, T. Funazkri, Effect of fluid dispersion coefficients on particle-to-fluid
2462. mass transfer coefficients in packed beds. Correlation of Sherwood numbers,
[41] Y.-B. Zhang, W.-X. Zhang, F.-Y. Feng, J.-P. Zhang, X.-M. Chen, A highly Chem. Eng. Sci. 33 (10) (1978) 1375–1384.
connected porous coordination polymer with unusual channel structure and [75] A.P.D. Wasch, G.F. Froment, Heat transfer in packed beds, Chem. Eng. Sci. 27
sorption properties, Angew. Chem. Int. Ed. 48 (29) (2009) 5287–5290. (1972) 567–576.
[42] I.H. Hwang, J.M. Bae, W.-S. Kim, Y.D. Jo, C. Kim, Y. Kim, S.-J. Kim, S. Huh, [76] R.B. Bird, W.E. Stewart, E.N. Lightfoot, Transport Phenomena., Wiley
Bifunctional 3D Cu-MOFs containing glutarates and bipyridyl ligands: International, Singapore, 2002.
selective CO2 sorption and heterogeneous catalysis, Dalton Trans. 41 (41) [77] S. Sircar, Excess properties and thermodynamics of multicomponent gas
(2012) 12759–12765. adsorption, J. Chem. Soc., Faraday Trans. 1: Phys. Chem. Condens. Phases 81 (7)
[43] W.-Y. Gao, Y. Niu, Y. Chen, L. Wojtas, J. Cai, Y.-S. Chen, S. Ma, Porous metal- (1985) 1527–1540.
organic framework based on a macrocyclic tetracarboxylate ligand exhibiting [78] E.W. Lemmon, M.L. Huber, M.O. McLinden, NIST Standard Reference Database
selective CO2 uptake, CrystEngComm 14 (19) (2012) 6115–6117. 23: Reference Fluid Thermodynamic and Transport Properties-REFPROP, in:
[44] H.-C. Kim, Y.S. Lee, S. Huh, S.J. Lee, Y. Kim, CO2 selective 1D double chain Standard Reference Data Program, National Institute of Standards and
dipyridyl-porphyrin based porous coordination polymers, Dalton Trans. 43 Technology, Gaithersburg, 2013.
(2014) 5680–5686. [79] O. Kunz, R. Klimeck, W. Wagner, The GERG-2004 Wide-Range Equation of
[45] H. An, B. Feng, S. Su, CO2 capture by electrothermal swing adsorption with State for Natural Gases and Other Mixtures, in GERG Technical Monograph 15.
activated carbon fibre materials, Int. J. Greenhouse Gas Control 5 (1) (2011) 2007, GERG – Groupe Européen de Recherches Gazières.
16–25. [80] S. Sandler, Models for Thermodynamic and Phase Equilibria Calculations,
[46] T.E. Rufford, S. Smart, G.C.Y. Watson, B.F. Graham, J. Boxall, J.C. Diniz da Costa, Marcel Dekker Inc., New York, 1994.
E.F. May, The removal of CO2 and N2 from natural gas: a review of conventional [81] O. Kunz, W. Wagner, The GERG-2008 wide-range equation of state for natural
and emerging process technologies, J. Petrol. Sci. Eng. 94–95 (2012) 123–154. gases and other mixtures: an expansion of GERG-2004, J. Chem. Eng. Data 57
[47] S. Cavenati, C.A. Grande, A.E. Rodrigues, Removal of carbon dioxide from (11) (2012) 3032–3091.
natural gas by vacuum pressure swing adsorption, Energy Fuels 20 (6) (2006) [82] Process Systems Enterprise, gPROMS. 1997–2016; Available from: <www.
2648–2659. psenterprise.com/gproms>.
[48] C.A. Grande, A.E. Rodrigues, Electric swing adsorption for CO2 removal from [83] J.v. Baten, REFPROP CAPE-OPEN Property Package Manager. 2016; Available
flue gases, Int. J. Greenhouse Gas Control 2 (2) (2008) 194–202. from: <http://www.amsterchem.com/>.
[49] R.T. Yang, Gas Separation by Adsorption Processes, Series on Chemical [84] J.M. Smith, H.V.V. Ness, M.M. Abbott, Introduction to Chemical Engineering
Engineering, Imperial College Press, 1987. Thermodynamics, sixth ed., McGraw-Hill, 2001.
[50] S. Sircar, W.R. Koch, Adsorptive separation of methane and carbon dioxide gas [85] Y.V.C. Rao, Chemical Engineering Thermodynamics, Universities Press, India,
mixtures, EP 0193716 A2, 1986. 1997.
[51] C.A. Grande, S. Cavenati, A.E. Rodrigues, Separation column and pressure swing [86] M. Whysall, L.J.M. Wagemans, Very Large-Scale Pressure Swing Adsorption
adsorption process for gas purification WO 2008/072215 A2, 2008. Processes, U.S. Patent 6,210,466, 2001.
[52] S. Sircar, J.W. Zondlo, Hydrogen purification by selective adsorption, US [87] A.M. Ribeiro, J.C. Santos, A.E. Rodrigues, PSA design for stoichiometric
4,077,779 A, 1978. adjustment of bio-syngas for methanol production and co-capture of carbon
[53] S. Sircar, R. Kumar, W.R. Koch, J. Vansloun, Recovery of methane from land fill dioxide, Chem. Eng. J. 163 (3) (2010) 355–363.
gas, US 4,770,676 A, 1988.