2020 Adsorption N. Chouikhi Et Al.
2020 Adsorption N. Chouikhi Et Al.
2020 Adsorption N. Chouikhi Et Al.
To cite this article: Najib Chouikhi, Federico Brandani, Pluton Pullumbi, Patrick Perré, François Puel (2020)
Biomethane production by adsorption technology: new cycle development, adsorbent selection and process
optimization. Adsorption, https://doi.org/10.1007/s10450-020-00250-3
Abstract
Gas separation by adsorption processes such as pressure swing adsorption (PSA) presents an attractive
alternative for upgrading biogas to biomethane. A new vacuum pressure swing adsorption (VPSA) cycle is proposed
for a unit designed to purify pre-cleaned biogas (40% CO2 and 60% CH4) in industrial conditions (feed flow rate more
than 500Nm3/h and large-volume equipment). The process simulations performed to optimize the VPSA unit consider
the kinetic separation of the feed components by using an appropriate carbon molecular sieve (CMS) adsorbent having
a high kinetic separation selectivity for CO2 with respect to CH4. The designed VPSA unit is composed of five columns
that perform three equalization steps. Minimizing methane losses during the regeneration steps necessitates injecting
part of the off-gas rich in CO2 at the bottom of the column during the production step to push the CH4 forward. The
produced biomethane meets the specification (97% CH4) of grid injection purity. The developed cycle allows a CH4
recovery of 92% to be obtained with a specific energy consumption of 0.35kWh/Nm3, thus meeting the initial
requirements for industrial exploitation of VPSA technology for biomethane purification from biogas sources.
Keywords: Biomethane production - Biogas upgrade - Methane recovery - Vacuum Pressure - Swing Adsorption -
Carbon Molecular Sieve - Numerical process simulation
Declarations
Not applicable
1 Introduction
Biogas is a mixture of several gases produced by the anaerobic digestion or fermentation of organic matter (e
g., manure, sewage sludge, municipal solid waste biodegradable waste, biodegradable feedstock, etc.). It is characterized
by its chemical composition with methane (CH4) being the main component, making it a very promising source of
energy and considered a viable alternative to fossil sources. The composition of biogas varies with the type of organic
matter from which it is derived (Poulleau 2002). However, it is predominately composed of(CH4) and carbon dioxide
(CO2), with relatively small amounts of hydrogen sulfide (H2S), ammonia (NH3), hydrogen (H2), nitrogen (N2), carbon
monoxide (CO), oxygen (O2), water vapor, and/or siloxanes and volatile organic compounds (VOC)(Rasi et al. 2007,
Rasi 2009).
Affiliations
Francois Puel
francois.puel@centralesupelec.fr
1 LGPM, Laboratoire de Génie des Procédés et Matériaux, CentraleSupélec, Université Paris-Saclay, 3 rue Joliot-Curie, 91192 Gif-sur-
Yvette, France
2 Air Liquide, Paris Innovation Campus. 1, chemin de la Porte des Loges, 78350 Les Loges en Josas, France
3 LGPM, Laboratoire de Génie des Procédés et Matériaux, CentraleSupélec, Université Paris-Saclay, SFR Condorcet FR CNRS 3417, Centre
Européen de Biotechnologie et de Bioéconomie (CEBB), 3 rue des Rouges Terres 51110 Pomacle, France
1
The separation or removal of impurities (H2S, N2) and, in particular, the large amount of CO2, is crucial. Some
processes used to remove CO2 can also remove minor components. However, in some cases, it is necessary to install
pre-treatment units for the removal of these components.
Biomethane can be produced by several technologies already on the market such as adsorption separation (PSA,
VPSA processes), absorption separation (physical or chemical absorption), membrane separation (MEMS), membrane
contactors, and cryogenic separation (Fougerit et al. 2019; Poulleau 2002; Ryckebosch et al. 2011).
These commercially available technologies for biomethane purification are competitive, all producing high-
quality biomethane that satisfies the grid specifications (methane purity > 97% CH4). The competition between these
technologies is concerning features such as energy consumption, CH4 purity and loss during processing, and
capital/operational expenditures. The choice of the specific process, therefore, depends on the specifications of the unit,
such as biogas flow rate and quality, presence of impurities and biomethane use (Warren 2012). Gas separation by
adsorption technology is widely used for the separation and the purification of gases in a wide range of fields (Sircar
2002; Voss 2005). Processes such as pressure swing adsorption (PSA) are promising tools for biogas upgrading. The
adsorptive separation exploits the selective adsorption (thermodynamic or kinetic) of different gas components through
specific interactions between the surface of the adsorbent (porous medium) and the adsorbed molecules.
Gas separation by vacuum pressure swing adsorption (VPSA) technology is characterized by its transient and
cyclic operation since, after the adsorption phase, the adsorbents have to be partially or completely regenerated for
further cyclic use. VPSA processes are rather complex due to the rapidly oscillating dynamics imposed by each cycle.
Yet, after a certain number of cycles, the process converges to an established periodic regime ("cyclic steady state" or
CSS): In this regime, the state of the system at the beginning of two consecutive cycles is the same. The performance
of the VPSA process must be established once the CSS is reached. However, the performance is controlled by many
process parameters, such as cycle schedule, column length, fluid velocity, pressure ratio, recycle rate, and kinetics and
equilibrium parameters of the adsorbent-adsorbate system under consideration.
Many published studies (Knaebel 2012; Olajossy et al. 2003; Reinhold et al. 1996; Rocha et al. 2017) present a
variety of PSA cycle schedules that incorporate a recycling step. They report the positive impact of different possible
recycling varieties on the CH4 recovery from landfill gas of the (i) light stream recycling (LR), (ii) heavy stream
recycling (HR) , (iii) double stream recycling (LR and HR). The literature reports that the cycle with one stage of HR
recycling tends to be the most efficient. [Erden et al. 2018] have tested and compared three configurations of recycling
purge gas for methane and nitrogen separation with a cycle composed of four steps for two columns.
The most advanced evolutions of PSA and VPSA processes applied to biogas purification are presented for
comparison in Table 1, which summarizes the main configuration characteristics of PSA processes and their biomethane
purification performance, achieved quite recently. The new cycle configuration reported by (Santos et al. 2011) seems
promising for biomethane production. However, a clear limitation relates to the low biogas flow rates under
consideration (approximately 30Nm3/h). [Cavenati et al. 2006] have shown that it is useful to use several layers of
adsorbents, especially in terms of CH4 productivity. The use of several layers can also be used to remove minor
components such as water. Nevertheless, the main drawback relates to the insufficiently low level of CH4 recovery
compared to other biogas upgrade technologies such as membrane separation. The use of dual PSA (Augelletti et al.
2017; Grande and Blom 2012) presents an innovative solution for obtaining high-purity CH4 with high recovery for
flowrates less than 100Nm3/h. However, these performances are reached at an increased cost since the proposed
solutions use either a higher adsorption pressure (Grande and Blom 2012) or a lower regeneration pressure (0.1bara)
(Augelletti et al. 2017) leading to an increase of the specific energy.
Thanks to a simulation approach, the present study aims at designing a new cycle configuration for producing
biomethane from biogas meeting the specifications for industrial deployment. We have employed published data of
adsorption equilibrium and kinetics of a carbon molecular sieve 3K (Takeda) adsorbent reported in the literature by
(Cavenati et al, 2005) for performing a rapid study of the impact of the cycle configuration of the VPSA cycle for the
kinetic-based separation of CO2 methane mixture. It also considers realistic constraints for the target separation, such
as a methane purity of 97% and a methane recovery of at least 92% with the specific energy consumption lower than
0.35kWh/m3. In addition, feed flow rates ten times higher than the conditions presented in the literature (see Table 1)
are considered. Due to the high target methane purity, a VPSA process is necessary. To reduce the cost of purified
biomethane production, particular attention is made to decrease the required specific energy by using desorption pressure
levels greater than 0.5 bara. A lower desorption pressure would result in a substantial increase in energy consumption.
In order to simplify the numerical modeling part, we assumed as a first step that a constant LDF model for the
microporous diffusion resistance would give qualitative indications on the most impacting parameters of the cycle. The
consideration of the dependence of the kinetic selectivity on the concentration variation will be considered in a second
2
stage for completing the study integrating the experimental characterization of the adsorbent and lab-scale bench tests
together with new model developments after their validation.
Table 1 Review of published PSA cycles developed for the separation of CH4/CO2(Augelletti et al. 2017; Canevesi et
al. 2019; Cavenati et al. 2006; Grande and Blom 2012; Khunpolgrang et al. 2015; Santos et al. 2011)
70 2
(first (first
Equilibrium stage) stage)
Grande and Two-stage 10% CO2
CMS & 0,5 80 -100 98 99 --
Blom 2012 PSA 5 90% CH4
Kinetic
(second (second
stage) stage)
VPSA
Khunpolgr Equilibrium
Combined Zeolite 40% CO2
ang et al. & 4 0.08 2-12 99 93 0.24
with nitrogen 13X 60% CH4
2015 Kinetic
regeneration
Zeolite Equilibrium
Santos et 33% CO2
VPSA 13X/ & 4 0.2 20-30 98 88 --
al. 2011 37% CH4
CMS Kinetic
VPSA (With
Canevesi et 40% CO2
equalization CMS Kinetic 5 0.1 0.04 97 90 --
al. 2019 60% CH4
tank)
2 Description of the VPSA process
The reference cycle considered in this study is well studied in CO2 capture and is composed of four columns
and an 8-step cycle: two adsorptions, equalization-depressurization, purge provide, blowdown, purge, equalization-
pressurization, and feed pressurization (see detailed description in supplementary material). The sequence of the steps
of the cycle for a single bed is reported in Figure 1.
3
Figure 1: Reference cycle representation based on single bed sequence of 8 steps
A significant amount of CH4 is lost during the blowdown steps when the reference cycle is considered. Another
equalization step was added to improve CH4 recovery. Note that adding successive equalization steps would further
increase this recovery rate but at the expense of the increasing complexity of the cycle. Three equalization steps were
then selected. Consequently, it was necessary to add another column to the VPSA design. Three equalizations with five
columns seemed reasonable from a technological point of view, but the target CH4 recovery indicator remained low
compared to the ones obtained by other separation technologies, such as membrane separation. Recycling part of the
off-gas, as proposed in the literature, was then adopted as a complement to the three equalizations. However, the
recycling configuration was different from the one chosen by (Erden et al. 2018) since, in this study, CH4 is the light
component.
We propose a VPSA unit composed of 5 columns with 15 steps. The novelty lies in the configuration of the
cycle, which combines three equalizations and a purge gas recycling. The sequence of the cycle steps for a single bed is
shown in Figure 2.
4
The representation of the 5-columns and 15-step cycle, indicating the correlation between the columns and the
cycle step scheduling known as the cycle chart is reported in Table 2.
Table 2 The cycle chart for the 5-column and 15-step cycle
o Two adsorption production steps (ADS1; ADS2) wherein the biogas is fed to the bottom of the column at
high pressure (8 bara), the adsorption occurs, and a CH4-rich product is obtained at the top of the column
(pure biomethane).
o One adsorption reflux step (ADS3_Ref) wherein a fraction of the off-gas (rich in CO2) leaving bottle number
3 during the BD2_Ref step pressurized to 8bara is fed to the bottom of the column. This fraction is called
the recycle ratio. The recycle ratio is the off-gas fraction leaving the column 3 during step 9 (BD2_Ref).
o Two equalization depressurization steps (Eq1_H & Eq2_H), wherein the high-pressure gas present in the
interstitial spaces of column 2 at the end of the adsorption phase is used to partially pressurize columns 4
and 5 which are respectively in the Equalization pressurization steps Eq2_L & Eq1_L.
o One purge provides step (PP) wherein column 2 (opened at the top and closed at the bottom) is connected
to column 3 (which is in the BD2_Ref) to depressurize the column and push the impurities.
o One equalization depressurization step (Eq3_H) wherein the gas at the intermediate pressure, present in the
interstitial spaces of column 3 is used to partially pressurize column number 4 from the desorption pressure
(under atmospheric pressure) to a pressure greater than the atmospheric pressure.
o One blowdown step (BD1) wherein the regeneration is started. Column 3 (closed in the top and opened in
the bottom) is depressurized until atmospheric pressure is reached to start the desorption of the CO2.
o One blowdown and reflux step (BD2_Ref), wherein the regeneration is completed by feeding the column
at low pressure using a vacuum pump. A fraction of the desorbed gas is recycled at the feed side of the first
column (which is in the ADS3_Ref step)
o Two equalization pressurization steps (Eq3_L, Eq2_L) wherein column 4 is partially pressurized with the
gas extracted from column 3 during the Eq3_H and column 2 during the Eq2_H).
o A time-out step (T-M) in which column 4 is isolated. The valves are closed at the bottom and top. During
this step, the column waits for the other bottles to finish their steps.
5
The Pressurization phase (PHASE 5) is composed of the following sequence steps:
o An equalization pressurization step (Eq1_L), wherein column 5 is partially pressurized with the high-
pressure gas extracted from column 2 during the Eq1_H step.
o Two Pressurization steps (F1, F2), wherein the column is pressurized with the biogas feed.
3 Physical modeling
3.1 Adsorption equilibrium and kinetics
Carbon molecular sieve (CMS) adsorbents (CMS 3K, Takeda) are known for their high kinetic selectivity
between CH4 and CO2, allowing for these mixtures to be successfully separated by kinetic PSA. CMS absorbents have
an average pore size of 3.7Å (kinetic diameter) (Fougerit et al. 2019). Methane with a pore size greater than CO2 (3.8
against 3.4Å) flows through the interstitial spaces of the CMS bed, while the CO2 is retained by adsorbent (Fougerit et
al. 2019).
Several published mathematical models are available to describe adsorption equilibrium. In this work, we have
selected the multicomponent extended Langmuir model (Ruthven 1984). The main advantage of this model compared
to the traditional Langmuir model lies in its ability to fit the isotherms better.
The quantity adsorbed for a gas i with 3 Langmuir sites is given by equation 1
𝑞 𝑏 𝑃 𝑞 𝑏 𝑃 𝑞 𝑏 𝑃
𝑞 (1)
1 𝑏 𝑃 1 𝑏 𝑃 1 𝑏 𝑃
Wherein 𝑞 , 𝑞 𝑞 Ncc/g are the saturation quantities of sites 1, 2, and 3 respectively, 𝑏 , 𝑏 𝑎𝑛𝑑 𝑏
are the curvature of each Langmuir site (bar-1), 𝐸 , 𝐸 𝑎𝑛𝑑 𝐸 (J/mol) are the activation energies of component i for
each Langmuir site and 𝑃 is the partial pressure of component i (bar).
The isotherm parameters of this model for each pure compound on a given adsorbent by determined by inverse analysis
of experimental data. The measured data are 𝑞 and 𝑃 and the parameters to be fitted are 𝑏 , 𝑏 . 𝑏 and 𝑞 , 𝑞
𝑞 for each site, acknowledging that the coefficients bki depend on the temperature according to Arrhenius' law given
by equation 2.
∗ ∗ (2)
𝑏 𝑏 𝑒
A good understanding of the kinetic limitations at different column scales is fundamental to design PSA kinetic
based separation. The CMS has a bidisperse structure resulting in both macropores and micropores. Regarding the
transport mechanism inside the micropores, several studies have been carried out and an extensive summary can be
found in (Qinglin et al. 2004). Three mechanisms have been identified, and they can be described by a pore model where
the resistance to diffusion is distributed inside the micropores, a barrier model where the diffusion resistance is at the
micropore mouth, and a dual resistance model where the first two mechanisms are combined. Regarding the diffusion
inside the micropores its dependency on the adsorbed phase concentration q^i can be described by the Darken’s equation
(Ruthven 1992):
d ln 𝑃
D , D ,, (3)
d ln 𝑞
wherein D , , is called the limiting diffusivity, which is the diffusivity in the Henry’s law region of the isotherm and 𝑃
the partial pressure.
[Qinglin et al. 2003] have shown that a modified dual resistance barrier gave a better representation of the
experimental data. This model has been used by (Effendy et al. 2017) to study the kinetic separation of N2 from CH4
via a PVSA process. The purpose of this study is to identify a new process cyle for the CH4/CO2 separation to provide
better CH4 recovery and use of specific energy compared to a given reference cycle (as described in paragraph 2). The
implementation of a more detailed kinetic model would require a further work of mathematical modeling and model
validation. For this reason for this screening study, we decided to use a simplified bilinear driving force (bi-LDF) model
6
used by(Cavenati et al. 2005; Grande and Rodrigues 2007)
The macroporous LDF constants (𝑘 ) were calculated using equation 4 which combines molecular diffusion
𝐷 , 𝑚 /𝑠 and Knudsen diffusion 𝐷 , 𝑚 /𝑠 and their values are reported in table 5 (Cavenati et al. 2005).
,
k . τ . D , 9700R (4)
, , ,
The LDF constants for the diffusion of micropores in the CMS adsorbent were calculated using equation 5
wherein 𝐷 , is the micropore diffusion, and 𝜏 is the tortuosity factor (2.2 for CMS). For CH4, a dual resistance
coefficient has been used, and kb corresponds to the barrier resistance at the micropore mouth coefficient.
,
𝑘 𝑘 (5)
,
𝜕 𝐶 𝜕 𝜕𝐶
𝐷𝑙 𝜀 𝑢𝐶 1 𝜀 𝑆 𝜀 (6)
𝜕𝑧 𝜕𝑧 𝜕𝑡
wherein ci (mol/m3) is the gas concentration of component i in the column, Dl m /s is the axial dispersion
coefficient, u m/s is the superficial velocity, ε is the voidage of the adsorbent bed, and 𝑆 is the source term describing
the mass transfer between the scale of the column and the scale of the bed.
The expression of 𝑆 is as follows:
𝑆 𝜀 𝜌 ; 𝜌 𝑘 𝑞∗ 𝑞 𝜀 𝑘 𝐶 𝐶𝑝, (7)
wherein εp is the porosity of the adsorbent particle, ρgr is the density of the grain 𝐶 is the average concentration
in the macropores, 𝑞 is the adsorbed quantity integrated both on the volume of the crystal and the particle of adsorbent,
and 𝑞 ∗ is the adsorbed quantity in the equilibrium.
Energy conservation equation
The thermal balance in the column is given by equation 8. This equation respectively considers the heat
accumulation in the solid, the heat effect due to compression, the heat convection in the gas phase, the heat conduction
in the solid, the heat effects due to adsorption, and the heat transfer rate through the wall.
𝜕𝑇 𝑑𝑃 𝜕𝑇 𝜕 𝑇 𝜕𝑞 2ℎ
𝜌𝐶 𝜀 𝜌 𝐶 𝑢 𝜀λ 𝜌 ∆𝐻 𝑇 𝑇 (8)
𝜕𝑡 𝑑𝑡 𝜕𝑧 𝜕𝑧 𝜕𝑡 𝑅
7
wherein λ (W/K.m) is the thermal conductivity of the solid, ρbed (kg/m3) is the packed density of the bed,
ℎ W/m²K is the convective heat transfer coefficient at the wall, 𝑇 𝐾 is the temperature of the wall, Rbed (m) is the
radius of the bed, ΔHi (kJ/mol) is the heat of adsorption of component i, Cpg is the calorific capacity of the gas, Cps is the
calorific capacity of the solid (kJ/mol K), and 𝜌𝐶 is the total thermal capacity.
Momentum conservation equation
The flow velocity of a gas in a porous medium generally depends on the pressure drop. For a laminar flow with
a Reynolds number less than 1, the velocity is proportional to the pressure gradient (Darcy's law). For faster flows, a
non-linear Darcy's law of order 2 can be used.
The linear pressure drop in a porous media of homogeneous and fixed spherical particles is expressed through
Ergun’s law:
150 𝜇 1 𝜀 1.75𝜌 1 𝜀
∆𝑃 𝑢 𝑢 (9)
𝑑 𝜀 𝑑 𝜀
wherein 𝜀 is the porosity of the bed, ρ is the density of the gas (kg/m3), 𝑑 is the diameter of the grains (m), 𝑢
is the superficial velocity (m/s), μ is the dynamical viscosity (Pa.s) of the fluid (kg/m/s), and ∆𝑃 is the linear pressure
drop (Pa/m).
𝜕𝐶𝑖 𝜕𝑇
0; 0 ∶ outgoing flow (11)
𝜕𝑧 𝜕𝑧
For the total pressure, two types of boundary conditions are used, either the pressure is imposed (for example,
P=8 bara during the adsorption step) or the pressure verifies the zero-derivative condition when the column is closed.
𝜕𝑃
0 ∶ Column closed (12)
𝜕𝑧
8
Table 3 Boundary conditions
∂Ci
y , P 0
Ci ∂z
RT
ADS1 & ADS2 ∂T
T T 0
∂z
P 8 bara
P valve cv Reg.
∂Ci
Ratio y , P 0
Ci ∂z
RT
ADS3_Ref ∂T
T T 0
∂z
P 8 bara
P valve cv Reg.
∂Ci
0 ∂Ci
∂z 0
∂z
∂T
Eq1_H & Eq2_H & Eq3_H &PP 0 ∂T
∂z 0
∂z
∂P
0 P valve cv Reg.
∂z
∂Ci
∂Ci 0
0 ∂z
∂z
∂T
BD1 ∂T 0
0 ∂z
∂z
∂P
P 1.013 bara 0
∂z
∂Ci y P
0 ,
∂z Ci
RT
BD2-ref ∂T
0 T T
∂z
P valve cv Reg.
P 0.5 bara
∂Ci ∂Ci
0 0
∂z ∂z
∂T ∂T
TM 0 0
∂z ∂z
∂P ∂P
0 0
∂z ∂z
∂Ci y _ , P
0 Ci
∂z RT
∂T T T
Eq1_H & Eq2_H & Eq3_H 0 _
∂z
P valve cv Reg.
∂P
0
∂z
∂Ci
y P 0
Ci
, ∂z
RT ∂T
F1&F2 0
T T ∂z
P 8 bara ∂P
0
∂z
9
Computational simulation
A non-isothermal dynamic model was developed in FORTRAN to simulate the VPSA cycle to evaluate the
performance of the process in terms of purity and biomethane recovery. The partial differential equations (PDE) of the
system, as described in the next section, are discretized by the finite difference method. The DLSODE (Double precision
Livermore Solver for Ordinary Differential Equation) FORTRAN subroutine from the package solver (Radhakrishnan
K. and Hindmarsh A. C. 1993) is used to solve the system.
Figure 3: Adsorption equilibrium (excess values) of CO2 (a) and CH4 (b) at 298K, 323K, and 343K (symbol : experimental data; solid line :
fitted data)
The results also show that the 3-site Langmuir model allows us to describe the adsorption equilibrium well, and
this model has the advantage of having an extension to predict the behavior of other components using single gas data.
The fitted values of the model parameters are reported in Table 4 (note that the reference temperature T* =293K).
However, the measured equilibria tell us that the adsorbent exhibits a low selectivity for CO2 regardless of the
temperature. For example, at 10bara and 343K, the equilibrium is 63 and 44Ncc/g adsorbent for CO2 and CH4,
respectively. Clearly, this adsorbent cannot be used in an equilibrium-controlled PSA to separate these gases due to the
poor selectivity (1.4), at 10 bar and 343 K.
10
Table 4 Fitted parameters of the 3-site Langmuir model. (T*=293K)
CO2 20.0 14.31 35750 40.0 1.08 30762 9.0 0.25 27436
CH4 20.0 15.32 33256 30.0 1.32 31593 19.0 0.015 29107
A good understanding of kinetic limitations at different column scales is fundamental to the design of PSA
kinetic-based separation, i.e., an analysis of mass transfer mechanisms in macropores and micropores is necessary. These
phenomena are complex because they are based on different mechanisms involving physical and geometric properties
of the system, such as pore size or adsorbate concentration. To that purpose, the macropore k and micropore k
diffusion constants (equations 3 and 4) were calculated and are reported in Table 5.
Table 5 Calculated kinetic parameters (original data from(Cavenati et al. 2005)
D .
𝑘 . 1/s k 1/𝑠 k 1/𝑠
R
𝐶𝐻 product
𝑀𝑒𝑡ℎ𝑎𝑛𝑒 𝑟𝑒𝑐𝑜𝑣𝑒𝑟𝑦 % (15)
𝐶𝐻 𝐹𝑒𝑒𝑑
- Methane purity is calculated as the ratio between the quantity of CH4 produced to the total quantity of the
product (considering CH4 and CO2).
Methane Purity CH %
CH4
(16)
- Methane productivity is the amount of CH4 produced per volume of adsorbent per unit of time.
11
- The specific energy consumption is a key parameter for evaluating the economic performance of a VPSA
process. It is composed of three terms: the compression energy, the energy consumed by the vacuum pump, and
the energy required to recycle pressurization during the regeneration steps.
γ 𝑄 𝑃ℎ
𝐷𝑢𝑟𝑎𝑡𝑖𝑜𝑛 ∗ γ 1 𝑅𝑇 𝛿 1
kWh 𝑃𝑙 (18)
Specific energy
Nm Methane CH product
wherein 𝛾 1.4, δ the mechanical efficiency of the machine (compressor/pump), Q the flow rate, and Ph
and Pl the high and low-pressure levels, respectively.
Among the four performance indicators listed above, it is necessary to identify which are used as controlled
parameters during the simulation and which are used for evaluating and optimizing the process. The feed flow rate is
monitored and adjusted during the simulation to reach the target CH4 purity (97%). Therefore, the purity is the (only)
controlled indicator. The three others, CH4 recovery, specific energy consumption and CH4 productivity, are considered
as output variables to be used for process optimization. The aim is to maximize the biomethane recovery, minimize the
specific energy consumption and obtain CH4 productivity suitable for an industrial application.
Figure 4 depicts the different column connections taken into consideration in the simulation. Valve V1 connects
the column to the feed, valve V2 (3-ways) ensures the purging of the column and the recycling of part of the off-gas,
and valve V4 allows the connection between columns during the balancing steps. Valve V3 is open during the production
step. It is important to point out that the valve flow coefficient (CV) of each valve, which defines its ability to pass the
gas flow, is controlled for each step.
The main operating parameters and column properties are shown in table 6. The dead volumes 1 and 2 present
the void part of the adsorber column, which are necessary for gas distribution. These void volumes don’t correspond to
the different porosities (bed, pellet, particle). They are used for representing real adsorber ; otherwise, the performances
of the unit are overestimated. These dead volumes are reported as a percentage of the adsorbent volume.
12
Table 6 Input simulation data
Adsorbent CMS
Number of columns 5
The parametric study of the effect of the recycle ratio on the CH4 recovery, as well as the criteria of performance
obtained with the reference cycle for a fixed purity (97%) are reported in Table 7. The simulation results of the reference
cycle showed that 26% of the methane was lost during the two blowdown steps. The first simulation with the new
configuration was performed without considering any recycling of purge gas in the adsorption step (ADS3_Ref). This
configuration allowed us to recover 6% more CH4 than in the reference cycle. Nevertheless, CH4 losses remained
important during the blowdown steps, especially the second one.
A 2% gain in CH4 recovery was obtained for each additional recycled 10% fraction. At the same time, the
specific energy consumption and CH4 productivity (and biogas flow rate) decreased. Recycling the purge gas at the end
of the adsorption phase leads to a decrease in productivity (see Figure 5) as the recycling gas replaces the feed flowrate.
The optimal recycle ratios offer good compromises between CH4 recovery and productivity while keeping a low enough
specific energy consumption (at the most 0.35kWh/Nm3) and a large biogas flow rate (at least 500Nm3/h) within the
range from 50 to 70%. Thus the CH4 recovery ranges from 89.4 to 93.5% and the CH4 productivity from 20.6 to
18.7Nm3/(h.m3).
13
Figure 5 Evolution of the CH4 recovery and productivity with the recycling ratio
Table 7 Effect of the recycling ratio on CH4 recovery for 97% CH4 purity with the optimized cycle, and comparison with the
reference cyle
Reference cycle
New Cycle
14
We then investigated the impact of this recycle ratio on different features of the cycle (pressure, temperature,
adsorbed amount of CH4 and CO2). A comparison is performed between cycles with and without recycle. The chosen
recycling ratio is 60% since it meets the desired performance. Visualizing the cycle features will help us to explain the
methane recovery improvement.
Figures 6a and 6b show the pressure profile and the mole fractions of CO2 and CH4 in the gas phase at the
bottom of the column without and with recycling, respectively. The pressure profile is the same for both cases, which is
consistent as the pressure is a controlled parameter. During the second purge step BD2_Ref, a vacuum pump is used to
evacuate the CO2. The pressure was reduced to 0.5 bara to create the condition for gas desorption. The molar fraction
of CH4 at the bottom of the column at the end of the ADS3 stage dropped by 40% due to the recycling of the purge gas.
Figure. 6 Pressure profile and gas molar fraction at the bottom of the column: a) without recycling b) with 60% recycling
The benefits of recycling the purge gas can be observed in Figure 7. At the beginning of the blowdown step
(BD1), the fraction of CH4 at the bottom of the column is 40% lower when the recycling is implemented. This difference
declines progressively down to 12% at the beginning of BD2_Ref. By recycling 60% of the off-gas (rich in CO2), the
CO2 pushes the CH4 already adsorbed upwards during the ADS3_Ref stage. The total loss of CH4 occurs during these
two blowdown steps. It is clear that decreasing the CH4 composition in these two streams improves its recovery.
15
Figure. 7 Methane molar fraction at the bottom of the column during purge steps (BD1 and BD2_Ref)
The temperature profiles at three different axial positions in the column (z=0.5, 1, and 2m) without and with
recycling are shown in Figures 8a and 8b, respectively. The temperature variations are similar at the top of the column
regardless of the configuration. During the adsorption steps, the temperature increased by 15K from the initial value,
especially around z=1m (z/l=0.4), which corresponds to the maximum adsorbed amount of CO2. As expected, the
temperature decreased during the regeneration steps. The highest temperature level is higher when the recycler is
operating since the CO2 adsorbed amount is more important.
16
Figure. 8 Temperature profile at three different axial positions a) without recycling and b) with 60% recycling
Figures 9a and 9b present the adsorbed amount of CH4 at the end of each step as a function of the axial position
without or with recycling, respectively. The increase of CH4 adsorbed in the CMS in the axial position reflects the
composition of the gas in the column, which is increasingly purer. The CH4 profiles are quite similar, depending on the
configuration. The maximum amount of CH4 absorbed is higher with recycling (20.4Ncc/g instead of 119.5Ncc/g). The
main difference between the adsorbed amount of CH4 is observed in the first lower section of the column. The
unexpected result is the absence of variation of CH4 adsorbed between steps. At cyclic steady state, the CH4 already
adsorbed on the CMS is not desorbed during the generation step. Additional simulations (see supplementary material)
were performed using a lower desorption pressure level (0.1 bara). The impact of this operating parameter was not
significant. As we will see in the next section, only the amount of adsorbed CO2 varies from one step to the other. This
result was already reported by (Santos et al. 2011) under similar situation (VPSA process and use of CMS for biogas
upgrading). However, the adsorbed quantity is 10 times higher in our study since the feed velocity level is 20 times
lower (0.1cm/s instead of 2.3cm/s).
17
Figure. 9 Adsorbed amount of CH4 and CO2 at the end of each step as a function of the axial position without or with recycling: a) CH4
adsorbed amount without recycling, b) CH4 adsorbed amount with 60% recycling, c) CO2 adsorbed amount without recycling, and d) CO2
adsorbed amount with 60% recycling
Figures 9.c and 9.d show the amount of CO2 adsorbed at the end of each step without and with recycling,
respectively. It is clear that the adsorption profile of CO2 is reverse with that of CH4. The highest values of adsorbed
CO2 are found at the bottom of the column. Unlike CH4, the variations in the amount of CO2 adsorbed between steps
are significant, reflecting the greater kinetic ability of this component to adsorb and desorb with this adsorbent. We can
also note that the implementation of recycling increases the variations of adsorbed quantities along the column and from
one stage to another. It is during the ADS3_Ref step that the CO2 adsorbed amount at the bottom of the column is at its
maximum. This value is 12Ncc/g higher with recycling, which justifies the lower quantity of CH4 adsorbed at the bottom
of the column (see Figure 9a). On the other hand, it can be seen that at the end of the purge stages (BD2_ref), the quantity
of CO2 adsorbed is slightly higher with recycling (difference of + 2.2Ncc/g), which means that the CMS desorbs less
CO2. With the same desorption capacity, a slightly longer purge time is required to desorb the additional amount of
adsorbed CO2 (12Ncc/g).
18
5 Conclusion
In this work, we have considered the effect of VPSA cycle arrangement for biomethane production. The
resulting VPSA unit composed of five columns is proposed as the best solution. It includes one recycle and three
equalization steps for improving the biogas upgrade with a CH4 purity target of 97%. Their performances are better than
those obtained with a reference cycle composed of four columns and eight steps. However, the assumption of constant
micropore diffusivities used for this simulation study as a first approximation for the cycle screening study needs to be
revisited at a further stage with the more detailed model as described in paragraph 3.1.
The initial analysis of the simulation results of the cycle without recycling showed that 20% of the CH4 was lost
during the purge steps, especially the second one. To minimize loss, a partial modification of the cycle configuration,
by including a recycle step, was adopted. Part of the purge gas (rich in CO2) flowing from the second purge step
BD2_Ref was pressurized and injected at the bottom of the column in the last production period ADS3_Ref.
A series of simulations with different recycle ratio values were conducted to evaluate the impact of the recycling.
Results indicate that it is indeed possible to increase the VPSA unit performance. This modification was very profitable
since a 2% gain in CH4 recovery was obtained for each additional 10% recycled fraction. However, this recycling led
to a decrease in productivity since the recycled gas substitutes the feed flow rate.
The trade-off between sufficient CH4 recovery (> 91.5%), specific energy consumption lower than
0.35kWh/Nm-3 and acceptable productivity is reached for a 60% recycling of the second purge of the cycle. The main
configuration characteristics of the optimized cycle (5 columns, 3 equalization steps, recycling of at least half of the
second purge) are a priori acceptable in an industrial application.
Acknowledgements
The authors are grateful to Air Liquide for financial support.
References
Augelletti, R., Conti, M., Annesini, M.C.: Pressure swing adsorption for biogas upgrading. A new process configuration
for the separation of biomethane and carbon dioxide. Journal of Cleaner Production. 140, 1390–1398 (2017).
https://doi.org/10.1016/j.jclepro.2016.10.013
Canevesi, R.L.S., Andreassen, K.A., Silva, E.A., Borba, C.E., Grande, C.A.: Evaluation of simplified pressure swing
adsorption cycles for bio-methane production. Adsorption. 25, 783–793 (2019). https://doi.org/10.1007/s10450-
019-00049-x
Cavenati, S., Grande, C., Rodrigues, A.: Separation of CH 4 / CO 2 / N 2 mixtures by layered pressure swing adsorption
for upgrade of natural gas. Chemical Engineering Science - CHEM ENG SCI. 61, 3893–3906 (2006).
https://doi.org/10.1016/j.ces.2006.01.023
Cavenati, S., Grande, C.A., Rodrigues, A.E.: Upgrade of Methane from Landfill Gas by Pressure Swing Adsorption.
Energy Fuels. 19, 2545–2555 (2005). https://doi.org/10.1021/ef050072h
Da Silva, F.A., Silva, J.A., Rodrigues, A.E.: A General Package for the Simulation of Cyclic Adsorption Processes.
Adsorption. 5, 229–244 (1999). https://doi.org/10.1023/A:1008974908427
Effendy, S., Xu, C., Farooq, S.: Optimization of a Pressure Swing Adsorption Process for Nitrogen Rejection from
Natural Gas. Ind. Eng. Chem. Res. 56, 5417–5431 (2017). https://doi.org/10.1021/acs.iecr.7b00513
Erden, L., Ebner, A.D., Ritter, J.A.: Separation of Landfill Gas CH4 from N2 Using Pressure Vacuum Swing
Adsorption Cycles with Heavy Reflux. Energy Fuels. 32, 3488–3498 (2018).
https://doi.org/10.1021/acs.energyfuels.7b03534
Fougerit, V., Pozzobon, V., Pareau, D., Théoleyre, M.-A., Stambouli, M.: Experimental and numerical investigation
binary mixture mass transfer in a gas – Liquid membrane contactor. Journal of Membrane Science. 572, 1–11
(2019). https://doi.org/10.1016/j.memsci.2018.10.042
Grande, C.A., Blom, R.: Utilization of Dual-PSA Technology for Natural Gas Upgrading and Integrated CO2 Capture.
Energy Procedia. 26, 2–14 (2012). https://doi.org/10.1016/j.egypro.2012.06.004
Grande, C.A., Rodrigues, A.E.: Biogas to Fuel by Vacuum Pressure Swing Adsorption I. Behavior of Equilibrium and
Kinetic-Based Adsorbents. Ind. Eng. Chem. Res. 46, 4595–4605 (2007). https://doi.org/10.1021/ie061341+
19
Khunpolgrang, J., Yosantea, S., Kongnoo, A., Phalakornkule, C.: Alternative PSA process cycle with combined vacuum
regeneration and nitrogen purging for CH4/CO2 separation. Fuel. 140, 171–177 (2015).
https://doi.org/10.1016/j.fuel.2014.09.100
Knaebel, K.S.: Multi-stage adsorption system for gas mixture separation,
https://patents.google.com/patent/US8211211B1/en?oq=Knaebel%2c+K.+S.+Multi-
Stage+Adsorption+System+for+Gas+Mixture+Separation.+Patent+US8211211+B1%2c+2012., (2012)
Olajossy, A., Gawdzik, A., Budner, Z., Dula, J.: Methane Separation from Coal Mine Methane Gas by Vacuum Pressure
Swing Adsorption. Chemical Engineering Research and Design. 81, 474–482 (2003).
https://doi.org/10.1205/026387603765173736
Poulleau, J.: Caractérisation des BIOGAZ Bibliographie Mesures sur sites. 82 (2002)
Qinglin, H., Farooq, S., Karimi, I.: Binary and Ternary Adsorption Kinetics of Gases in Carbon Molecular Sieves.
Langmuir. 19, (2003). https://doi.org/10.1021/la0270791
Qinglin, H., Farooq, S., Karimi, I.A.: Prediction of binary gas diffusion in carbon molecular sieves at high pressure.
AIChE Journal. 50, 351–367 (2004). https://doi.org/10.1002/aic.10032
Radhakrishnan, K.: Description and Use of LSODE, the Livermore Solver for Ordinary Differential Equations. 124
Rasi, S.: Biogas composition and upgrading to biomethane. (2009)
Rasi, S., Veijanen, A., Rintala, J.: Trace compounds of biogas from different biogas production plants. Energy. 32,
1375–1380 (2007). https://doi.org/10.1016/j.energy.2006.10.018
Reinhold, H.E., D’Amico, J.S., Knaebel, K.S.: Natural gas enrichment process,
https://patents.google.com/patent/US5536300A/en, (1996)
Rocha, L.A.M., Andreassen, K.A., Grande, C.A.: Separation of CO2/CH4 using carbon molecular sieve (CMS) at low
and high pressure. Chemical Engineering Science. 164, 148–157 (2017).
https://doi.org/10.1016/j.ces.2017.01.071
Ruthven, D.M.: Principles of Adsorption and Adsorption Processes. John Wiley & Sons (1984)
Ruthven, D.M.: Diffusion of oxygen and nitrogen in carbon molecular sieve. Chemical Engineering Science. 47, 4305–
4308 (1992). https://doi.org/10.1016/0009-2509(92)85108-N
Ryckebosch, E., Drouillon, M., Vervaeren, H.: Techniques for transformation of biogas to biomethane. BIOMASS &
BIOENERGY. 35, 1633–1645 (2011). http://dx.doi.org/10.1016/j.biombioe.2011.02.033
Santos, M.S., Grande, C.A., Rodrigues, A.E.: New cycle configuration to enhance performance of kinetic PSA
processes. Chemical Engineering Science. 66, 1590–1599 (2011). https://doi.org/10.1016/j.ces.2010.12.032
Shafeeyan, M.S., Wan Daud, W.M.A., Shamiri, A.: A review of mathematical modeling of fixed-bed columns for
carbon dioxide adsorption. Chemical Engineering Research and Design. 92, 961–988 (2014).
https://doi.org/10.1016/j.cherd.2013.08.018
Sircar, S.: Pressure Swing Adsorption. Applied Catalysis. 46, (2002). https://doi.org/10.1021/ie0109758
Voss, C.: Applications of Pressure Swing Adsorption Technology. Adsorption. 11, 527–529 (2005).
https://doi.org/10.1007/s10450-005-5979-3
Warren, K.: A techno-economic comparison of biogas upgrading technologies in Europe. (2012)
20
Biomethane production by adsorption technology: New cycle development,
adsorbent selection and process optimization
Najib Chouikhi1, 2 ∙ Federico Brandani2 ∙ Pluton Pullumbi2 ∙Patrick Perre1,3 ∙ François Puel1
Supplementary Material
Reference cycle description
The reference cycle considered in this study is well studied in CO2 capture and is a four column and an 8-step cycle.
The sequence of the steps of the cycle for a single bed is reported in Figure 1.
Affiliations
Francois Puel
francois.puel@centralesupelec.fr
1 LGPM, Laboratoire de Génie des Procédés et Matériaux, CentraleSupélec, Université Paris-Saclay, 3 rue Joliot-Curie, 91192 Gif-sur-
Yvette, France
2 Air Liquide, Paris Innovation Campus. 1, chemin de la Porte des Loges, 78350 Les Loges en Josas, France
3 LGPM, Laboratoire de Génie des Procédés et Matériaux, CentraleSupélec, Université Paris-Saclay, SFR Condorcet FR CNRS 3417, Centre
Européen de Biotechnologie et de Bioéconomie (CEBB), 3 rue des Rouges Terres 51110 Pomacle, France
21
The Depressurization phase (PHASE 2) is composed of the following sequence steps:
o One equalization depressurization step (Eq1_H), wherein the high-pressure gas present in the interstitial
spaces of column 2 at the end of the adsorption phase is used to partially pressurize columns 4 which is in
the Equalization pressurization step Eq1_L.
o One purge provides step (PP) wherein column 2 (opened at the top and closed at the bottom) is connected
to column 3 (which is in the BD2) to depressurize the column and push the impurities.
The Regeneration phase (PHASE 3) is composed of the following sequence steps:
o One blowdown step (BD1) wherein the regeneration is started. Column 3 (closed in the top and opened in
the bottom) is depressurized until atmospheric pressure to start the desorption of the CO2.
o One blowdown step (BD2), wherein the regeneration is completed by feeding the column at low pressure
using a vacuum pump.
The Pressurization phase (PHASE 4) is composed of the following sequence steps:
o An equalization pressurization step (Eq1_L), wherein column 4 is partially pressurized with the high-
pressure gas extracted from column 2 during the Eq1_H step.
o One pressurization step (F1), wherein the column is pressurized with the biogas feed.
The simulation results of the reference (Table 1) cycle showed that 26% of the methane was lost during the two
blowdown steps.
Table 8 Simulation results of the reference cycle
Figure 2 shows the pressure profile and the mole fractions of CO2 and CH4 in the gas phase at the bottom of the column.
The adsorption pressure remains constant at 8 bara during the adsorption steps (ADS1 and ADS2) which is consistent
as the pressure is a controlled parameter. During the second purge step BD2, a vacuum pump is used to evacuate the
CO2. The pressure was reduced to 0.5 bara to create the condition for gas desorption. The molar fraction of methane at
the bottom of the column remains constant during the first four steps. This fraction drops from 60% to 15% at the end
of BD1. During the BD2 step, the methane fraction at the bottom of the column increases to 0.42, which means that at
the end of the BD2, 42% of the purge gas is methane. This explains the low CH4 recovery level of the process.
Figure 11 Pressure profile and gas molar fraction at the bottom of the column
Figures 3a and 3b presents respectively the adsorbed amount of CH4 and CO2 at the end of each step as a function of
the axial position. The increase of CH4 adsorbed in the CMS in the axial position reflects the composition of the gas in
the column, which is increasingly purer. The adsorbed quantities of methane at the end of each step are similar over the
22
entire height of the column. This unexpected profile means that once the cyclic equilibrium state is reached, the already
adsorbed methane does not desorb during the purge phase. The adsorption profile of CO2 is reverse with that of CH4.
The highest values of adsorbed CO2 are found at the bottom of the column. Unlike CH4, the variations in the amount of
CO2 adsorbed between steps are significant, reflecting the greater kinetic ability of this component to adsorb and desorb
with this adsorbent.
Figure 12 Adsorbed amount of CH4 and CO2 at the end of each step as a function of the axial position with the reference cycle a) CH4 adsorbed
amount, b) CO2 adsorbed amount
Case with low desorption pressure ( 𝑷𝒅𝒆𝒔 𝟎. 𝟏 𝒃𝒂𝒓𝒂 ) for the new cycle
A new cycle with 5 columns and 15 steps is discussed in the article. Additional simulations (see Table 2) were performed
using a lower desorption pressure level (0.1bara) for the BD2_ref step to see its impact on the adsorbed methane quantity
profile.
The impact of this operating parameter was not significant on the adsorbed amount of methane profile at the end of
each step (see Figure 4), only the maximum amount of methane adsorbed was decreased to 17.3 Ncc/g. This means that
even with a desorption pressure of 0.1 bara the desorption of methane from the CMS is not greatly improved. Finally, it
should be noted that the diminution of the desorption pressure significantly increases the methane recovery rate to 97%.
This is also accompanied by an increase in the specific energy of biomethane production (0.56 kWh/Nm3).
Table 9 Simulation results of the new cycle (case with P_des= 0.1 bara)
New Cycle
23
Figure 13 Adsorbed amount of CH4 and CO2 at the end of each step as a function of the axial position with 60% recycling and P_des=0.1bara
a) CH4 adsorbed amount with 60% recycling, b) CO2 adsorbed amount with 60% recycling
24