Accepted Manuscript

Pressure swing adsorption for biogas upgrading. A new process configuration for the
separation of biomethane and carbon dioxide

Rosaria Augelletti, Maria Conti, Maria Cristina Annesini

PII: S0959-6526(16)31611-0
DOI: 10.1016/j.jclepro.2016.10.013
Reference: JCLP 8204

To appear in: Journal of Cleaner Production

Received Date: 18 May 2016

Revised Date: 5 September 2016
Accepted Date: 5 October 2016

Please cite this article as: Augelletti R, Conti M, Annesini MC, Pressure swing adsorption for biogas
upgrading. A new process configuration for the separation of biomethane and carbon dioxide, Journal of
Cleaner Production (2016), doi: 10.1016/j.jclepro.2016.10.013.

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.
 ACCEPTED MANUSCRIPT

Number of words: about 4750

Pressure swing adsorption for biogas upgrading. A new process configu-

ration for the separation of biomethane and carbon dioxide

PT
Rosaria Augelletti, Maria Conti, Maria Cristina Annesini*

Department of Chemical Engineering, Materials & Environment, University of Rome “La Sapienza”, Via Eudossiana
18, I-00184 Roma

RI
*Corresponding author. E-mail address: annesini@uniroma1.it

SC
Abstract
Pressure swing adsorption (PSA) is an interesting technology for biogas upgrading, due to compactness of the
equipment, low energy requirements, low capital cost, and safety and simplicity of operation. Unfortunately, some

U
shortcomings penalize its diffusion in comparison with other technologies; in particular, conventional PSA has a
low methane recovery and cannot compete in this field with other processes such as amine scrubbing; furthermore,
AN
it produces an off gas stream with a significant methane content, which requires further treatment to avoid the
emission of residual methane into the atmosphere. In this framework, this study focuses on the feasibility of a PSA
based separation process able to obtain a biomethane stream suitable to be injected in the natural gas grid (CO2
<3% by volume) with a high methane recovery and an almost pure CO2 stream (CO2 > 99%). The proposed
process uses Zeolite 5A as adsorbent material in two PSA units; the biogas is fed to the first unit which produces
M
biomethane; the off gas of the first unit is sent to a second PSA unit which separates carbon dioxide from a residual
gas stream, recycled to the first to enhance methane recovery. A dynamic non-isothermal model, based on the linear
driving force approximation, is employed to demonstrate the technological feasibility of the separation units and to
D

assess the performance of the whole process. In particular a methane recovery greater than 99% can be obtained
with energy consumption of about 1250 kJ per kg of biomethane.
TE

Keywords
Biogas upgrading, pressure swing adsorption, biomethane, complete separation, methane recovery.
EP

Introduction
In the last decades biogas has achieved a significant importance in the field of renewable energy, especially as a heat
C

source; biogas is a gaseous mixture produced by methanogenic bacteria through anaerobic fermentation of organic
matter. Biogas from anaerobic digesters contains mainly methane and carbon dioxide, while other contaminants
(hydrogen sulphide, ammonia, oxygen, nitrogen, dragged solid particles, siloxane) almost always do not exceed the
AC

threshold of 4%; it is also saturated with water at the temperature at which it is produced. Biogas constitutes an
important methane resource, especially for those countries that have to import natural gas and other fossil fuel; in
2013 in Italy the production of biogas was around 1815 kteq, while in Europe it reached 14400 kteq (1kteq=1000
tonnes of oil equivalent) (Biogas barometer, 2014).

1
 ACCEPTED MANUSCRIPT

In addition to the purely economic benefit deriving from the use of biogas as a renewable fuel, the energy
recovery from biogas leads to environmental benefits since methane has a warming potential 21 times higher than
that of carbon dioxide (Grande and Rodrigues, 2007). Several paths can be undertaken in order to recover energy
from raw biogas (Goossens, 1996): heat production from combustion in actual flares, electric energy production
in internal combustion engines, combined heat and power production in cogeneration systems (CHP). A viable
alternative that is gaining more and more attention is biogas valorization (upgrading) to obtain a gas, biomethane,

PT
comparable to natural gas (CO2 content lower than 3% in volume), which can be used for automotive applications
or may be injected in the natural gas grid (Ravina and Genon, 2015). Upgrading requires processes to reduce the
carbon dioxide content (initially 35% - 55%) and to remove other contaminants from biogas; with regard the CO2
removal, several and well established processes are commonly carried out: physical absorption with water or organic

RI
solvents, amine scrubbing, membrane-based processes, cryogenic processes and pressure swing adsorption (De Hullu
et al., 2008; Kapdi et al., 2005; Petterson and Wellinger, 2009).
Pressure swing adsorption (PSA) is one of the most known and established industrial processes for gas separation

SC
because of the compactness of the equipment, low energy requirements, low capital investment cost, and safety and
simplicity of operation; all these advantages make the process suitable also for small installations (Delgrado et al.,
2006). PSA technology uses an adsorbent material which is subjected to pressure changes to selectively adsorb and
desorb the undesired gas components; the selective adsorption occurs because of different equilibrium capacities
(equilibrium adsorbent) or by differences in uptake rates (kinetic adsorbent) (Ruthven, Farooq and Knaebel, 1994).

U
The original PSA scheduling, designed by Skarstrom (Skarstrom, 1960), includes four steps (adsorption, blow-
down, purge and pressurization) of equal or different duration, implemented in two or more columns in order to
AN
make the gas treatment continuous. What makes PSA a very versatile technology is the possibility to act on several
variables, such as type of adsorbent material, type and sequence of cycle steps, steps duration, operative pressures,
column size, and use of single or multi-bed process.
PSA for biogas upgrading allows to obtain biomethane, a gas that has the same methane purity specifications of
M
natural gas (CO2 <3% by volume), with low energy consumption. Both equilibrium and kinetic adsorbents, already
commercially available for full scale applications, can be used for biogas upgrading; activated carbon, Zeolite 13X,
Zeolite 5A among the kinetic adsorbent and carbon molecular sieve (CMS) among the kinetic adsorbent have been
tested; furthermore, innovative materials like metal organic framework (MOF), silicalite or silicoaluminophosphate
D

sorbents (SAPOs) are also under investigation at laboratory scale (Cavenati, Grande and E., 2004; Himeno, Ko-
matsu and Fujitas, 2005; Jayaraman et al., 2002; Rivera-Ramos, Ruiz-Mercado and Hernandez-Maldonado, 2008).
TE

Unfortunately, if compared with other biogas upgrading technologies (such as amine scrubbing), PSA lacks in
methane recovery, since a part of methane fed with biogas is lost with the off gas, i.e. with the residual stream
mainly consisting of CO2 but with a significant methane content. Indeed, a methane recovery of 85-90% is ob-
tained and an off-gas with a methane content of about 15-20% is produced (Cavenati, Grande and Rodrigues, 2005;
EP

Grande and Rodrigues, 2007; Santos, Grande and Rodrigues, 2011). Because of this high methane content, the off
gas cannot be released into the atmosphere, but requires further treatment. Indeed, the off gas is often burned in
flare or sent to a cogeneration unit with raw biogas, but in many cases the heat produced in the combustion exceeds
the requirements of the anaerobic digester or of the heat consumers in the neighborhood of the upgrading plant,
since other wastes, such as the digested biomass or agriculture residues, are also available as fuels.
C

In order to overcome the problem of the off gas disposal and at the same time to increase the methane recovery,
in this work we have studied the feasibility of employing a double PSA unit to separate the biogas in a stream of
AC

biomethane and an almost pure CO2 stream (CO2 content greater than 99% by volume); assuming that water and
other contaminants were previously removed, we have considered a biogas mixture composed by CH4 and CO2 ,
with a methane content of 60%. The biogas is sent to a first PSA unit for the production of biomethane; the off
gas here produced is fed to a second PSA unit which separates carbon dioxide from a residual stream, recycled to
the first unit to enhance methane recovery. The high purity grade of carbon dioxide stream makes it suitable to be

2
 ACCEPTED MANUSCRIPT

used directly, taking advantage of its physico-chemical properties or as a carbon source in chemical and biochemical
processes; carbon dioxide can indeed be used as cooling fluid in refrigeration units for the transport of products
at low temperature or for environmental purpose, such as in the pH control in waste water treatment or even in
laboratories as supercritical fluid in extraction operations, etc.
Here, a Zeolite 5A has been employed as adsorbent material in both PSA units; a dynamic non isothermal
model based on the linear driving force approximation with a single lumped transport parameter to describe the

PT
adsorption/desorption kinetics has been developed; the model, implemented to assess the performance of the whole
process, has proved that biomethane can be obtained with a methane recovery greater than 99% and low energy
consumption.

RI
Process definition
Figure 1 shows the process diagram of the two PSA units studied in this work. A feed stream of 100 Nm3 /h

SC
of biogas (representative of a small-medium scale biogas plant) with a CH4 and CO2 content of 60% and 40%,
respectively, is considered. The biogas is assumed available at 1 atmosphere and to have been previously treated
to remove water (generally through refrigeration, adsorption on silica or allumina or absorption in triethylenglycol)
and other harmful contaminants, such as hydrogen sulphide (i.e. through biological aerobic oxidation in the digester

U
vessel or downstream the digester through adsorption on activated carbon or by means of membrane separation) or
siloxane (i.e. through adsorption on silica gel or activated carbon or by means of cryogenic separation) (Ryckebosch,
AN
Drouillon and Vervaeren, 2011). The biogas pretreatment has not been studied in this work, but must be carried
out in order to prevent the contaminants content from being harmful to the downstream separation units, natural
gas grid, motors or end-users. Biomethane is produced during the high pressure adsorption phase of the first PSA
unit; the off gas is instead sent to a second PSA unit to separate carbon dioxide from a residual stream which is
recycled to the first PSA unit to enhance methane recovery.
M
D
TE
C EP

diagram1.jpg
AC

Figure 1: Process diagram for the biogas upgrading.

While the biomethane is produced continuously, the off gas is periodically extracted during the desorption phases
of the first PSA. Due to the different cycle timing of the two PSA units and in order to guarantee a continuous feed
and a homogeneous composition to the second unit, a storage system is included between the two units. For the

3
 ACCEPTED MANUSCRIPT

same reason another storage is placed before the first unit in order to mix the raw biogas with the recycle stream
produced discontinuously. Both PSA units use zeolite 5A as adsorbent material.

Modeling and Simulation

PT
In order to simulate the PSA separation units and thus to evaluate the process performance in terms of purity,
recovery and energy requirement, a dynamic non-isothermal model based on the linear driving force (LDF) approx-
imation is used (Glueckauf and Coates, 1947); in this model, the adsorption/desorption kinetics is described by a
single lumped transport parameter for each component, kLDF,i . Axially dispersed plug flow, perfect gas behavior,

RI
no radial concentration and temperature gradients, thermal equilibrium between gas and solid phase and no heat
exchange with the external environment have been assumed in the development of the mathematical model. In
particular, the assumption of adiabatic behavior becomes reasonable for large, industrial columns, for which heat
exchange with the external environment is negligible. The equations of the model are summarized in table 1 together

SC
with the correlations used for estimations of mass and heat transfer parameters.
In order to solve the system of partial differential equation, boundary and initial conditions are required. Since a
PSA process is a sequence of different elementary steps, specific boundary conditions for each of these steps should
be set. Table 2 reports the boundary conditions used in the in the various step constituting the PSA cycle, which

U
will be described further below; in the same table the pressure trend adopted in each step of the cycle is reported.
As for the initial conditions, the column filled with biogas at minimum pressure and with clean adsorbent solid has
been considered at the beginning of the simulation.
AN
After a screening between several adsorbent materials, based on the analysis of the breakthrough curves, we have
chosen zeolite 5A for both the PSA units (Ahn et al., 2012); zeolite 5A is an equilibrium-based adsorbent material,
with selective adsorption towards CO2 . Adsorption equilibrium is described by the multicomponent Langmuir
M
isotherm:
qi bi pi
= (1)
D

P
qmax,i 1 + j bj pj
where qi and pi are the adsorbed phase concentration and the partial pressure of each component, respectively and
qmax,i and bi are the single component temperature-dependent parameters:
TE

qmax,i (T ) = qmax,i (T0 ) + kq,i (T − T0 ) (2)

EP

1 1
bi (T ) = bi (T0 )exp(kb,i ( − )) (3)
T T0

Adsorption isotherm parameters and the linear driving force mass transfer coefficient for CH4 and CO2 adsorption
C

on zeolite 5A are reported in Table 3; in the same table the geometrical and physical-chemical properties of the
zeolite beads are reported. Figure 2 shows the adsorption isotherm of the two pure components on the zeolite
AC

3A at 25°C and at 50°C. It is clear that that zeolite 5A has a greater adsorption capacity for carbon dioxide, as
shown in Figure 2, but the comparison of KLDF,i values, reported in Table 3, indicates that kinetics favor methane
adsorption. It is worth noting that the nonlinearity of the adsorption isotherm of CO2 suggests to operate the solid
regeneration under vacuum, thus realizing a vacuum PSA (VPSA).

The performance of the PSA separation process has been evaluated in terms of:

4
 ACCEPTED MANUSCRIPT

PT
Table 1: Model equations used in PSA simulations.
Mass balance equation for component i-th in the bed

RI
2 ∂(uci )
∂ q¯i
εb ∂c ∂ ci
∂t + (1 − εb ) ∂t = εb Dz ∂z 2 −
i
∂z

Total mass balance equation

SC
( ∂t ) = − ∂(uC)
P ∂ q¯i
εb ∂C
∂t + (1 − εb ) ∂z
nc

Energy balance equation for solid and gas phases

U
2 ∂(uC) ∂cp,g ∂cp,g
εb
−εb kz ∂∂zT2 + cp,g uC ∂T
∂z + cp,g T ∂z + uCT ∂z + ∂P
R (cp,g ∂t +P ∂t )+
AN
(1 − εb )cp,s ρs ∂T
∂t − (1 − εb )
P
(−∆Hi ) ∂∂tq¯i = 0
nc

Linear driving force equation for the adsorption rate

M
∂ q¯i ∗
∂t = KLDF,i (qi − q̄i )
D

Langmuir adsorption isotherm

qi bP
i pi
=
TE

qmax,i 1+ bj pj
nc

Axial mass dispersion coefficient Ruthven D. M. (1984)

EP

εb Dz ρg udp µg
Dm = 20 + 0.5ScRe Re = µg Sc = ρg Dm

Axial heat dispersion coefficient Wakao and Funazkri (1978)

C

kz cp,g µg
kg = (7.0 + 0.5P rRe) Pr = kg
AC

5
 ACCEPTED MANUSCRIPT

PT
Table 2: Boundary conditions and pressure time profile implemented in the PSA simulation model.
z=0 z=L
Feed dP
dt
=0

RI
uC = uF E CF E
∂ci
∂z
=0
ucF E,i = uci − εb Dz ∂c
∂z
i
∂T
∂z
=0
εb kz ∂T = −ucp,g (TF E − T )

SC
∂z
α∆P0 e−t/τ
Equalization depressurization dP
dt
= τ (α+1)
u=0
∂ci
∂z
=0
∂ci
∂z
=0

U
∂T
∂z
=0
∂T
∂z
=0
.
Blow-down dP
= P BD (t)
dt
AN ∂ci
∂z
=0
u=0

∂ci
∂z
=0
∂T
∂z
=0
∂T
=0
M
∂z
Purge dP
dt
=0
uC = uP U CP U
∂ci
∂z
=0
ucP U,i = uci − εb Dz ∂ci
D

∂z
∂T
∂z
=0
εb kz ∂T
∂z
= −ucp,g (TP U − T )
∆P0 e−t/τ
Equalization pressurization dP
=
TE

dt τ (α+1)
u=0
ucEQ,i = uci − εb Dz ∂c
∂z
i
∂ci
∂z
=0
εb kz ∂T
∂z
= −ucp,g (TEQ − T )
∂T
EP

∂z
=0
.
Pressurization dP
dt
= P P R (t)
u=0
ucP R,i = uci − εb Dz ∂c
∂z
i
∂ci
∂z
=0
C

εb kz ∂T
∂z
= −ucp,g (TP R − T )
∂T
∂z
=0
AC

6
 ACCEPTED MANUSCRIPT

PT
RI
U SC
Figure 2: Equilibrium isotherm on zeolite 5A at 25°C and 60°C.

AN
M
D

Table 3: Langmuir parameters, LDF coefficients and physical-chemical properties of the zeolite (Ahn et al., 2012)
TE

CH4 CO2
qmax (298 K) mol kg −1
2.28 4.49
kq mol kg −1 K −1 -0.01192 -0.01858
b(298 K) bar−1 0.2 3.12
EP

kb K 1731 207
−4H kcal mol−1 5.4 9.33
kLDF s−1 0.147 0.0135
Average pellet radius m 6 · 10 −4
C

Pellet density kg m−3 1160

Pellet porosity - 0.64
AC

Specific heat capacity cal kg −1 K −1 220

7
 ACCEPTED MANUSCRIPT

• purity of the biomethane stream

NCH4
CH4 purity =
NCH4 + NCO2

where NCH4 and NCO2 are the molar flow rates of methane and carbon dioxide, respectively in the biomethane

PT
stream;
• CH4 recovery

RI
MCH4
CH4 recovery =
FCH4

where MCH4 is the molar flow rate of methane recovered with the biomethane stream (equal to NCH4 ) exiting

SC
the first PSA unit or recovered with the recycled stream exiting the second PSA unit; FCH4 is the molar flow rate
of methane fed to each PSA unit.
• purity in CO2 of the CO2 -rich stream

U
LCO2
CO2 purity =
LCH4 + LCO2
AN
where LCO2 and LCH4 are the molar flow rates of carbon dioxide and methane respectively in the CO2 -rich
stream exiting the second PSA unit.
• energy consumption, calculated by
M
ˆ tcycle
= P dt
0
D

where tcycle is the total cycle time and P is the power consumption, evaluated assuming that the devices operate
under adiabatic regime with a single stage compression
TE

 
k RT . Pout k−1
P= B ( ) k −1
k−1 η Pin

where k = cp /cv (assumed equal to 1.31), η is the mechanical efficiency, which typically assumes the value of
EP

.
0.8, R is the universal constant of gases, T is the gas temperature, B is the molar flow rate to be compressed, Pout
and Pin are the pressures in the outlet and inlet section of compressor (or vacuum pump) respectively.
As it will be shown in more detail later, we have considered four columns for both PSA units, which are
interconnected with each other according to a proper timing cycle. In order to simplify the simulation and to
C

reduce the CPU time required, simulations have been carried out considering a single column for each PSA unit
and taking into account the recycles between the columns in the subsequent steps of the cycle; modeling a PSA
AC

unit with a single column, instead of the complete array of four columns, leads to the same results once the cyclic
steady state is reached (for each unit an average of 40 cycles were needed to reach the steady state, corresponding
to almost 2 h of simulation time with a i7 intel core processor). After a first simulation of the two units placed
in succession, the loop has been closed and the simulation was repeated until convergence, which was fixed when
two successive iterations differed in about 0.1%, calculated on the variation of the methane mole fraction in the
biomethane stream. In the following, results obtained at the periodic steady state are discussed.

8
 ACCEPTED MANUSCRIPT

First PSA unit for biomethane production

The first PSA unit has been designed to produce the biomethane stream, that is a stream composed by methane
and carbon dioxide with a methane content of at least 97%. This unit is composed by four interconnected column,
each of them is subjected to the same phases sequence, as shown in Figure 3; in the same figure the sequence of
the phases in bed 1 is also reported, with the details of the flux direction and the opening and closing conditions

PT
of the column sections during each step of the cycle. It is worth noting that this column is fed with the mixture of
raw biogas and the gas recycled from the second PSA unit (see Figure 1); once periodic steady state conditions are
achieved, a feed flow rate of 113.47 Nm3 /h with a CH4 content of 62.6% is obtained.

RI
U SC
AN
M
D
TE
EP

Figure 3: Scheduling of four-columns PSA cycle and sequence of the phases in bed 1. FE: feed, EQ1 and EQ2:
equalization, BD: blow-down, PU: purge, PR: pressurization.
C

The phase sequence is composed by a) a feed step (FE) where the biogas is fed to the column at high pressure
and the adsorption occurs; a CH4 -rich product is obtained from the column (biomethane); b) an equalization step
AC

(EQ2), where the high pressure gas present in the interstitial spaces of the column at the end of the feed phase
is used to partially pressurize another column; c) a blow-down step (BD), where the column is depressurized until
vacuum conditions in order to desorb carbon dioxide; d) a purge step (PU), where the regeneration is completed
by feeding the column at low pressure with part of the product obtained during the feed phase; e) an equalization
step (EQ1), where the column is partially pressurized with the high pressure gas extracted from another column; f)

9
 ACCEPTED MANUSCRIPT

Table 4: Column characteristics, pressure operating conditions and cycle timing of the two PSA units.
First PSA Second PSA
Column length m 1.5 1
Column diameter m 0.4 0.31
Column density kg m−3 742.4 742.4

PT
Column porosity - 0.36 0.36
Pressure feed bar 6 6
Pressure purge bar 0.2 0.2
Feed Time s 190 130

RI
Purge time s 190 130
Pressurization time s 170 110
Blow down time s 170 110

SC
Equalization time s 20 20

a pressurization phase (PR), where the column is pressurized with the biogas. As can be seen, the cycle scheduling
ensures the continuous production of biomethane (during the feed step), while the off gas, withdrawn during BD

U
and PU, is produced in a discontinuous manner.
The cycle simulation has been carried out by implementing the model described in Table 1, with the boundary
AN
conditions and the pressure profiles reported in Table 2. In particular, a constant pressure value has been set
during the feed and the purge steps at the maximum and the minimum pressure value of the cycle respectively,
while linear trends of pressure vs. time have been assumed during pressurization and blow-down steps. In the
equalization phases, instead, simultaneous exponential pressure variations have been set (pressure increase during
M
EQ1 and pressure decrease during EQ2) with a time constant τ , as reported in Table 2. It is worth noting that
the parameter α in the pressure variation equation accounts for the different capacities of the two columns that are
connected during pressure equalization; its value is determined by ensuring that the amount of gas released during
the equalization depressurization is equal to that fed into the column during the equalization pressurization.
D

The column characteristics, pressure conditions, and the cycle timings used in the first PSA unit are summarized
TE

in Table 4, while in Figure 4 the simulation results are reported; in more detail, figure 4a) shows the pressure history
during the PSA cycle, while figure 4b) and 4c) report the molar flow rate of the two component during each step
of the cycle. For the convention adopted, the flows are positive if directed as the z axis: therefore a positive flow
indicates that it is entering the column at z=0 and is exiting the column at z/L=1; flows equal to zero during
EQ1, PR, EQ2 and BD indicate the closure of the column during these phases. As can be seen, the flow entering
EP

the column during the pressure equalization pressurization step (EQ1) corresponds to that exiting another column
that is simultaneously in the pressure equalization depressurization step (EQ2); the connection between the two
columns is reflected in the equivalence of the areas under the two equalization steps. Moreover it can be noted that
the column regeneration occurs mainly at the end of the blow down step and the beginning of the purge step, as
C

indicated by the carbon dioxide flow.

Figure 4d) and 4e) report the adsorbed amount of the two components as a function of the axial position along
AC

the column; these plots show that methane adsorption occurs in the pressure equalization pressurization phase and
in the pressurization phase, at the end of which methane is almost uniformly distributed along the column, except
for the inlet section, where on the contrary carbon dioxide adsorption is greater. In the feed phase, methane is
desorbed from the solid and replaced by carbon dioxide, as can be noted by the methane desorption front, shifted
toward the outlet section. It is also evident that, as mentioned before, the CO2 removal takes place mainly during

10
 ACCEPTED MANUSCRIPT

the blow down step, as is clear by the area between the EQ2 and BD curves. The comparison between the values of
adsorbed amounts of the two components, together with the replacement of methane by the carbon dioxide, reflect
the adsorption selectivity of zeolite 5 A towards carbon dioxide.
Figure 4 f) shows the temperature profile during a cycle at three different positions: the figure shows that during the
feed phase a significant temperature increase occurs along the column, especially around the position at z/L=0.35
at which corresponds the maximum adsorbed amount of carbon dioxide, with a temperature increase of about

PT
35°C. During the regeneration steps the temperature decreases, with the maximum decrease during the BD phase,
to which a maximum in the desorbed amount corresponds. As can be noted, during a cycle the bed experiences
an appreciable temperature excursion, during which temperature reaches the maximum value of about 70° C corre-
sponding to z/L=0.35. The significant temperature increase is due to the high heat of adsorption, as shown in Table

RI
3; on the other hand it is worth recalling that the simulations have been carried out in adiabatic conditions and
therefore overestimate the temperature changes and consequently underestimate the column performances, which
are negatively affected by the heat effects connected with the adsorption process.

SC
The table in Figure 6 reports the mass balance of the first PSA and the cycle performances in terms of methane
purity and recovery with the biomethane stream, together with the power consumption required for the separation.
A product (biomethane) with a methane content of 98.9 % is obtained, but the methane recovery is rather low.
The power consumption has been evaluated by dividing the energy required in a cycle by the time of the cycle; it

U
is important to point out that here the energy is referred only to the energy required for biogas compression up to
the maximum pressure of the cycle (6 atm) and for vacuum pump during the regeneration steps, without taking
AN
into account the additional energy required for the further compression of biomethane before its injection into the
natural gas distribution grid.

Second PSA unit for off gas separation

M
The off gas produced in the first PSA unit is sent to a second PSA unit for carbon dioxide separation; a storage
at atmospheric pressure separates the two units in order to guarantee a constant feed to the second unit, which
D

moreover works with a different cycle timing. The cycle scheduling is the same adopted for the first unit (see Figure
3) while the column characteristics,pressure conditions, and cycle timings are summarized in Table 4. Figure 5
shows the simulation results of this second PSA cycle.
TE

As can be seen from the adsorbed amount profiles, methane is adsorbed appreciably only during the pressure
equalization step (EQ1) and especially during the pressurization step, at the end of which it only occupies the
second half of the column, towards the outlet section. During the remaining steps of the cycle, the amount of
EP

methane adsorbed is practically close to zero; in particular during the feed phase, carbon dioxide replaces almost
completely the methane that was adsorbed during the pressurization phases and now recovered with the product
stream. The molar flow rates at the outlet section of the column (see figure 5c)) reflect the displacement of methane
by carbon dioxide during the feed phase, since for about the first 90 seconds, the gas exiting the column is mainly
C

composed by methane.
It is worth noting that the column is now subjected to high temperature variations (see figure 5f), with a
AC

maximum of about 80 °C corresponding to z/L= 0.3; in this position the lower temperature value of about -13 °C
is reached at the end of the purge step; the higher temperature value of about 97°C is instead reached at the end
of the feed phase at z/L=1, where carbon dioxide is more adsorbed.
The process performances of the second PSA unit are reported in the table of Figure 6. As can be seen, this
second unit completes the separation by producing a CH4 -rich stream recycled to the first unit and an almost
pure carbon dioxide stream (off-gas) withdrawn during the discharge phases; the high methane recovery with the

11
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
C EP

Figure 4: Simulation of the first PSA unit: pressure history (a); molar flow rate of the two components at the top
(b) and the bottom (c) of the column; adsorbed amount of CH4 (d) and CO2 (e) along the bed at the end of each
step; temperature profile during a cycle at three different positions (f).
AC

12
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP

Figure 5: Simulation of the second PSA unit: pressure history (a); molar flow rate of the two components at the
C

top (b) and the bottom (c) of the column; adsorbed amount of CH4 (d) and CO2 (e) along the bed at the end of
each step; temperature profile during a cycle at three different positions (f).
AC

13
 ACCEPTED MANUSCRIPT

recycle stream is a positive consequence of the displacement of methane by carbon dioxide during the feed phase,
as observed from the simulation profiles. The power consumption is now referred to the power required to compress
the off gas of the first unit from 1 (pressure of the storage tank) to 6 atmospheres (higher pressure of the second
PSA unit) and to power the vacuum pump that operates during the regeneration steps. The further power required
to store the carbon dioxide (for example in cylindrical tanks) was not taken into account in the power consumption
assessment.

PT
Overall process

RI
The overall process scheme is reported in Figure 6, together with the mass balances and the overall process perfor-
mances.
As can be seen, the complete separation of biomethane and almost pure carbon dioxide (CO2 > 99%) is obtained
with a complete methane recovery (> 99%) in the biomethane. An energy consumption of about 1250 kJ per kg of

SC
biomethane, corresponding to 550 kJ per Nm3 of biogas, was evaluated; it includes the electrical energy required
for biogas compression as well as blowdown and purge withdrown at low pressure; no recompression is required to
recycle the F2 stream to the first PSA unit, because the two PSA columns operate at the same maximal pressure.
No energy for biomethane compression for injection in the gas grid is considered (biomethane is then available at

U
4 atm). It is important to note that the specific energy consumption here evaluated seems to be very favorable
if compared to values reported in other works, in which a single PSA unit is used for biogas upgrading, without
AN
operating a further separation step for the production of the carbon dioxide stream. Indeed, energy consumption
between 700 and 1100 kJ per Nm3 of biogas are reported for a single PSA unit(Persson, Jonsson and Wellinger,
2006; Bauer et al., 2013). More specifically, Grande and Rodrigues (2007) studied the upgrading of a biogas stream
with a composition very close to that considered in this work (55% CH4 and 45% CO2 ) in a single PSA unit with a
working cycle including two depressurization steps in order to improve the methane recovery and evaluated energy
M
consumption of about 4500 and 2680 kJ for kg of biomethane with Zeolite 13X or CMS-3K as adsorbent material,
respectively. Wu et al. (2015) considered the upgrading of a biogas with a higher methane content (67%) in a PSA
unit working with the four standard steps (PR,FE, BD, PU) and an equalization step, with a maximal pressure of
D

8 atm; these authors compared the performance of PSA process with the same adsorbent materials of Grande and
Rodrigues (2007) and an innovative MOF and report energy consumption of about 2375, 1200 and 1040 kJ for kg of
biomethane for Zeolite 13X, CMS-3K and MOF, respectively. The reason for the low specific energy consumption
TE

evaluated in this work with respect to other works is mainly due on the one hand to the better separation capability
of the adsorbent material chosen for the separation process and, on the other, to the almost total methane recovery
obtained from the whole process.
C EP
AC

14
 ACCEPTED MANUSCRIPT

PT
RI
SC
diagram2.jpg

U
Flow rate Composition

F1
F2
AN
kg/h
133.96
12.61
Nm3 /h
100
13.47
CO2
40%
17.8% 82.2%
CH4
60%

F3 146.57 113.47 37.4% 62.6%

43.98 60.37 1.2% 98.8%
M
F4
F5 89.98 53.1 78.5% 21.5%
F6 77.37 39.63 99.14% 0.86%
First PSA Second PSA
D

CH4 recovery 83.9 97

Power consumption 10.2 kW 5.1 kW
Total CH4 recovery 99.4%
TE

Energy consumption 1250 kJ/kgbio

Figure 6: Overall process scheme of biogas upgrading for the separation of biomethane and carbon dioxide.
EP

Conclusion
The use of double PSA units for biogas upgrading seems to be a very interesting and feasible process to obtain an
C

almost complete separation of biogas components. The configuration studied in this work, indeed, allows to obtain,
on one side, biomethane with a CO2 content lower than 3%, that makes it suitable to be injected in the natural gas
AC

grid and, in addition, an almost pure carbon dioxide stream (>99%) that may be used for several applications as a
carbon source in chemical and biochemical processes. The whole process yields an almost total methane recovery
(>99%), which is reflected in the very low specific energy consumption (1250 kJ per kg of biomethane) if compared
to other biogas upgrading processes based on a single PSA unit.

15
 ACCEPTED MANUSCRIPT

Nomenclature
bi equilibrium parameter of Langmuir model for component i, atm −1
ci gas phase concentration of component i, mol m−3
C total gas phase concentration, mol m−3
cp,i molar constant pressure specific heat of component i, cal mol−1 K −1

PT
cp,g molar constant pressure specific heat of the gas mixture, cal mol−1 K −1
cp,s constant pressure specific heat of the adsorbent, cal kg −1 K −1
dp particle diameter, m
Dz axial dispersion coefficient, m2 s−1

RI
Dm molecular diffusivity of component i, m2 s−1
KLDF,i LDF coefficient of component i, s−1
kb equilibrium parameter of Langmuir model for component i, K
kq equilibrium parameter of Langmuir model for component i, mol kg −1 K −1

SC
kg gas thermal conductivity, W m−1 K
kz axial gas phase thermal conductivity, cal m−1 s−1 K −1
nc number of components
P pressure, atm

U
pi partial pressure of component i, atm
P r Prandtl number
∗
AN
q i volume-averaged adsorbed phase concentration of component i, mol m−3
qi equilibrium adsorbed phase concentration of component i, mol m−3
qmax,i equilibrium parameter of Langmuir model for component i, mol m−3
Re Reynolds number
M
Sc Schmidt number
R universal gas constant
t time, s
tcycle total cycle time, s
D

T temperature, K
u superficial gas velocity, m s−1
TE

yi molar fraction of component i in the gas phase

z axial position in the bed, m
εb bed porosity
∆Hi heat of adsorption of component i, cal mol−1
EP

µg gas viscosity, P a s
ρs gas density, kg m−3
ρs pellet density, kg m−3
C
AC

16
 ACCEPTED MANUSCRIPT

References
Ahn, S., Y. W. You, D. G. Lee, K. H. Kim, M. Oh and C. H. Lee. 2012. “Layered two- and four-bed PSA processes
for H2 recovery from coal gas.” Chemical Engineering Science 68:413–423.
Bauer, F., T. Persson, C. Hulteberg and D. Tamm. 2013. “Biogas upgrading technology overview, comparison and
perspectives for the future.” Biofuels, Bioproducts and Biorefining 7:499–511.

PT
Biogas barometer. 2014. Technical report Eurobserv’er www.eurobserv-er.org: .
Cavenati, S., C.A. Grande and A.E. Rodrigues. 2005. “Upgrade of Methane from Landfill Gas by Pressure Swing

RI
Adsorption.” Energy & Fuels 19:2545–2555.
Cavenati, S., C.A. Grande and Rodrigues A. E. 2004. “Adsorption Equilibrium of Methane, Carbon Dioxide, and
Nitrogen on Zeolite 13X at High Pressures.” Journal of Chemical Engineering Data 49:1095–1101.

SC
De Hullu, J., J. I. W Maasen, P. A. Van Meel, S. Shazad and J. M. P. Vaessen. 2008. Comparing Different Biogas
Upgrading Techniques. Technical report DMT Dirkse Milieutechniek.
Delgrado, J. A., M. A. Uguina, J. Sotelo, B. Ruiz and J. Gomez. 2006. “Fixed bed adsorption of carbon diox-

U
ide/methane mixtures on silicalite pellets.” Adsorption 12:5–18.
Glueckauf, E. and J. Coates. 1947. “Theory of chromatography. Part 3.” Journal of the Chemical Society 0:1315–
1321. AN
Goossens, M. A. 1996. “Landfill Gas Power Plants.” Renewable Energy 9:1015.
Grande, C. A. and A. E. Rodrigues. 2007. “Biogas to Fuel by Vacuum Pressure Swing Adsorption I. Behavior of
M
Equilibrium and Kinetic-Based Adsorbents.” Industrial and Engineering Chemistry Research 46:4595–4605.
Himeno, S., T. Komatsu and S. Fujitas. 2005. “High-Pressure Adsorption Equilibria of Methane and Carbon Dioxide
on Several Activated Carbons.” Journal of Chemical Engineering Data 50:369–376.
D

Jayaraman, A., A. S. Chiao, J. Padin, R. T. Yang and C. L. Munson. 2002. “Kinetic Separation of Methane/Carbon
Dioxide by Molecular Sieve Carbons.” Separation Science and Technology 37:2505–2528.
TE

Kapdi, S.S., V.K. Vijay, S.K. Rajesh and R. Prasad. 2005. “Biogas Scrubbing, Compression and Storage: Perspective
and Prospectus in Indian Context.” Renewable Energy 30:1195–1202.
EP

Persson, M., O. Jonsson and A. Wellinger. 2006. “Biogas upgrading to vehicle fuel standards and grid injection.”
Energy from biogas and landfill gas .
Petterson, A. and A. Wellinger. 2009. Biogas upgrading technologies-developments and innovations. Technical
report IEA Bioenergy.
C

Ravina, M. and G. Genon. 2015. “Global and local emissions of a biogas plant considering the production of
biomethane as an alternative end-use solution.” Journal of Cleaner Production 102:115–126.
AC

Rivera-Ramos, M. E., G. J. Ruiz-Mercado and A. J. Hernandez-Maldonado. 2008. “Separation of CO2 from Light
Gas Mixtures using Ion-Exchanged Silicoaluminophosphate Nanoporous Sorbents.” Industrial and Engineering
Chemistry Research 47:5602–5610.
Ruthven D. M. 1984. Principles of adsorption and adsorption processes. John Wiley & Sons.

17
 ACCEPTED MANUSCRIPT

Ruthven, D.M., S. Farooq and K. S. Knaebel. 1994. Pressure Swing Adsorption. VCH Publ.
Ryckebosch, E., M. Drouillon and H. Vervaeren. 2011. “Techniques for transformation of biogas to biomethane.”
Biomass and Bioenergy 35:1633–1645.
Santos, M. P. S., C. A. Grande and A. E. Rodrigues. 2011. “Pressure Swing Adsorption for biogas upgrading.
Effect or recycling streams in Pressure Swing Adsorption design.” Industrial and Engineering Chemistry Research

PT
50:974–985.
Skarstrom, C.W. 1960. “Method and apparatus for fractionating gaseous mixtures by adsorption.” Esso Research
and Engineering Company. US Patent 2,944,627.

RI
Wakao, N. and T. Funazkri. 1978. “Effect of fluid dispersion coefficients on particle-to-fluid mass transfer coefficients
in packed bed.” Chemical Engineering Science 33:1375–1384.

SC
Wu, B., X. Zhang, Y. Xu, D. Bao and S. Zhang. 2015. “Assessment of the energy consumption of the biogas upgrading
process with pressure swing adsorption using novel adsorbents.” Journal of Cleaner Production 101:251–261.

U
AN
M
D
TE
C EP
AC

18