Professional Documents
Culture Documents
h i g h l i g h t s
a r t i c l e i n f o a b s t r a c t
Article history: The water scrubbing of biogas is an efficient, cheap and environmental friendly process to remove CO2
Received 11 July 2016 from biogas in order to upgrade it into biomethane. This work deals with the modelling and simulations
Received in revised form 12 November 2016 of a scrubber-flash-stripper high pressure water scrubbing process, which is currently the most mature
Accepted 9 January 2017
technology. The model and its associated simulation procedure can be used to estimate appropriate
Available online 10 January 2017
device sizes as well as to assess the performance of a given plant. Thanks to this tool, an optimized con-
figuration is computed for a reference case. A sensitivity analysis is also realized in term of scrubbing effi-
Keywords:
ciency and CH4 slippage. The influence of parameters related to the water recycling as well as the
Biogas
Biomethane
operating parameters is investigated. Moreover, the sensitivity of the model with respect to the solubility
Water scrubbing and transfer coefficient parameters is analyzed, in order to gain insight on the parameter to focus in the
Scrubber-flash-stripper configuration framework of a model calibration by experimental results.
Process scaling Ó 2017 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.ces.2017.01.012
0009-2509/Ó 2017 Elsevier Ltd. All rights reserved.
640 C.E. Wylock, W.M. Budzianowski / Chemical Engineering Science 170 (2017) 639–652
Nomenclature
Greek
DE enthalpy variation, kJ/mol
g removal efficiency, %
2013). In this work we investigate the gas-liquid absorption by commonly consists of a packed column. Due to the small CO2 sol-
water, called hereafter water scrubbing. Indeed, water is a cheap ubility in water, the absorption rate is enhanced by setting its oper-
and environmental friendly solvent for removing CO2. It makes ating pressure between 0.8 and 1.2 MPa, in order to increase CO2
use of the higher solubility of CO2 (and other compounds like partial pressure in the biogas. This enables to reach high purity bio-
H2S) than the solubility of CH4 in water. The ratio of Henry con- methane at the top of the scrubber. The dissolved CH4 within the
stants between CO2 and CH4 is about 30 which makes water scrub- CO2-rich water leaving the scrubber is also increased under
bing of CO2 from biogas relatively efficient. elevated pressure operation. Therefore, the scrubbing solvent
The most mature technology is the scrubber-flash-stripper high needs to be subsequently separated in a flash tank operating at a
pressure water scrubbing (SFS-HPWS), which is sketched in Fig. 1. reduced pressure of about 0.11–0.2 MPa. The released gas, rich in
The CO2 capture from the biogas is performed in a scrubber, which CH4 and CO2 is then mixed with the raw biogas and reinjected into
the scrubber. To regenerate the scrubbing water, the CO2 has to be
released. To achieve this solvent regeneration a second low pres-
sure packed column is used as a stripper. In this case, CO2 is des-
orbed from water at ambient temperature using air as stripping
agent. This considerably reduces energy requirements for solvent
regeneration compared to solvents with chemical compounds that
strongly bind CO2 and requires consequently higher stripping
temperature.
Several scrubber-flash-stripper HPWS plants already exist at
the commercial scale. Some examples of such plants as well as
their advantages and drawbacks are presented in Budzianowski
et al. (2017). Despite, it seems that very few HPWS models are pro-
posed in the literature. In addition, the available experimental data
are generally sparse and often focused only on the scrubber
analysis.
To scale appropriately the gas-liquid devices and the associated
equipment’s (pump, compressor, etc.) and to identify optimum
operating conditions for the process, mathematical modelling
and numerical simulation are particularly valuable to reach high-
performance SFS-HPWS plant. The industrially validated models
may be also useful in process control, including on-line monitoring
of plant performance. This work deals with the mathematical mod-
elling and the numerical analysis of the SFS-HPWS plant. The
appropriate column dimensions are related to the characteristics
Fig. 1. Block-diagram of a scrubber-flash-stripper configuration of a pressurized of the biogas to upgrade (flow rate, concentration. . .), the targeted
water scrubbing process.
C.E. Wylock, W.M. Budzianowski / Chemical Engineering Science 170 (2017) 639–652 641
CO2 abatement and the effective transfer rates. The relevant trans- cules and hydroxide ion (OH ) to form bicarbonate (HCO
3 ) and
fer phenomena are therefore described at the local scale to reach carbonate (CO2
3 ) ions, according to the following scheme
reliable estimations of diameter, height and operating conditions (Danckwerts and Sharma, 1966; Cents et al., 2005; Danckwerts,
suitable for the plant scale. 1970; Vas Bhat et al., 2000):
Besides the model development, two procedures are proposed,
depending on the user’s goal: (1) scale a plant to ensure a given CO2 þ H2 O
HCO3 þ Hþ ð1Þ
performance or (2) evaluate the performance of a given plant oper-
ating with given conditions. This work notably analyses the influ- CO2 þ OH
HCO3 ð2Þ
ence of the parameters related to the scrubbing water recycling
on the scrubber and stripper sizes and the sensitivity of the perfor- HCO3 þ OH
CO2
3 þ H2 O ð3Þ
mance with respect to the operating conditions. It aims to get fur-
ther insight into the operations of SFS-HPWS plants by examining H2 O
Hþ þ OH ð4Þ
the trends and proposing some guidelines.
Nevertheless, due to the absence of any counter-ion, the pH of
The developed model explicitly considers mass transfer
scrubbing water is always below 7. The contribution of the reaction
between gas and liquid and hence it is a so called non-
(2) is very small due to the negligible OH concentration and the
equilibrium rate based model which offers improved reliability in
reaction (3) is unlikely to occur. Only reaction (1) may have an
simulations of mass transfer in biogas upgrading plants. A new
effect on the absorption rate. Its reaction rate is given by
procedure for experimental model validation and calibration was
Danckwerts and Sharma (1966), Cents et al. (2005) by kH2 O ½CO2 ,
developed. Such models used with present-day computers will
with kH2 O 0.02 s1. This enables the estimation of its related
be suitable tools for improving resource and energy efficiency of
Hatta number (Trambouze and Euzen, 2002) by:
processing industries. Their use will contribute to reducing indus-
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
trial emissions and making existing industries economically viable. DL;CO2 kH2 O
Also such models will allow for facilitated introduction of innova- Ha ¼ ð5Þ
kL
tive processes to the industrial practice at a very low cost. There-
fore, the present study can be considered as cutting edge where DL;CO2 is the diffusive coefficient of CO2 in water and kL is the
research in process engineering and science. liquid-side mass transfer coefficient, as defined in Table 1. Ha is
then estimated to 0.05, which is significantly smaller than the
threshold value 0.3. The reaction is therefore much slower than
2. Mathematical modelling the mass transport and it does not enhance the transfer. The effect
of the reaction rate is therefore neglected for the mass transfer rate
2.1. Scrubber and stripper computation.
Despite the fact that the CH4 solubility in water is much smaller
It is well known that CO2 absorption in water is coupled with than CO2, the high pressure used in HPWS induces a substantial
chemical reactions: the dissolved CO2 may react with water mole- CH4 dissolution. The CH4 transfer may influence in the computa-
Table 1
Literature correlations used in the model.
Correlation References
Packing characteristics for 25 mm unglazed ceramic Intalox
saddle
dP = 25103 m Trambouze and Euzen (2002) (Tables 4.5 and 4.8)
AP = 250 m2/m3
F p = 320 m2/m3
rc = 61103 N/m
Henry coefficients
H CO2 ¼ 3:59 107 exp 2044
T
Versteeg and Van Swaaij (1988), in Vas Bhat et al. (2000), Wylock et al.
(2011)
H CH4 ¼ 1:3817 108 exp 1600 T1 298:15
1 Sander (1999)
Diffusive coefficients
DL;CO2 ¼ 2:35 106 exp 2119 Versteeg and Van Swaaij (1988), in Vas Bhat et al. (2000)
TL
h i1:8769 Guo et al. (2013)
DL;CH4 ¼ 1:95 109 229:8
TL
1
DG;CO2 ¼ 7:84
T 1:75
105 pG Reid et al. (1986), in Wylock et al. (2014)
G
DG;CH4 ¼ 4:29
T 1:5
104 pG Cowie and Watt (1971)
G
Equilibrium constants
K 1 ¼ exp 12092:1T 36:786lnðTÞ þ 235:482 qw Edwards et al. (1978), in Vas Bhat et al. (2000), Wylock et al. (2011)
K ¼ 10ð T þ22:4773logðTÞ61:2062Þ q2 Tsonopoulos (1976), in Vas Bhat et al. (2000), Wylock et al. (2011)
5839:5
w w
Column cross section area and pressure drop
qffiffiffiffiffi qffiffiffiffiffi
p2 ¼ uuGL qqL ¼ QQ GL qqG
G L
Trambouze and Euzen (2002) (Table 4.6)
0:75 uL qL 0:1 u2L qL 0:2 u2L APp 0:05 Onda et al. (1968), in Trambouze and Euzen (2002) (Table 4.7)
A ¼ AP 1 exp 1:45 rrc l AP rAP g
L
tion of the CO2 molar ratio and it is necessary to evaluate the dis- where the locale value of Q G and Q L are computed from the inert
solved concentration in scrubbing water to control the flash tank gas and liquid conservation.
efficiency. The gas-liquid transfer of both CO2 and CH4 is therefore It is considered that gas phase follows the Raoult’s law, there-
considered. Their mass transfer rate density, sCO2 and sCH4 respec- fore a species partial pressure pG;i is directly deduced from its
tively, are computed according to the two-film theory (Whitman, molar fraction yG;i , which is related to its mass fraction by
1923) by the classical following expressions (Trambouze and yG;i ¼
x
P
G;i
xG;j . The total pressure of the gas pG is considered as a lin-
Mi
Euzen, 2002; Coulson and Richardson, 1999): j Mj
dðQ G xG;CO2 Þ dxG;CO2 dQ G and the electroneutrality equation neglecting the OH concentra-
¼ QG þ xG;CO2 ¼ sCO2 ð8Þ tion, leading to:
dz dz dz
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi!
dðQ G xG;CH4 Þ dxG;CH4 dQ G qL 1 qL
¼ QG þ xG;CH4 ¼ sCH4 ð9Þ ½CO2 L ¼ xL;CO2 K 1 þ K 1 4xL;CO2 þ K1
dz dz dz MCO2 2 M CO2
ð20Þ
dðQ L xL;CO2 Þ dxL;CO2 dQ L
¼ QL þ xL;CO2 ¼ sCO2 ð10Þ
dz dz dz
2.2. Flash tank
dðQ L xL;CH4 Þ dxL;CH4 dQ L
¼ QL þ xL;CH4 ¼ sCH4 ð11Þ The flash tank is modelled as a perfectly mixed continuous
dz dz dz
device, with one inlet (the liquid leaving the scrubber) and two
z is vertical coordinate in the column, Q denotes the mass flow
outlets (the gas returning to the scrubber and the liquid
rate and x the mass fraction.
entering the stripper). The following simplifications are assumed:
Since the inert part of the gas and liquid flow rates,
(i) the separation between the phases is perfect (no liquid is
Q G ð1 xG;CO2 xG;CH4 Þ and Q L ð1 xL;CO2 xL;CH4 Þ respectively,
entrained by the gas), (ii) the thermodynamic equilibrium is
remains constant, the two following equations can be derived:
reached instantaneously between the gas and the liquid, (iii) the
dxG;CO2 dxG;CH4 dQ G pressure drop induced by the expansion is negligible (pFLS is con-
Q G þ þ ð1 xG;CO2 xG;CH4 Þ ¼0 ð12Þ
dz dz dz stant), (iv) the flash temperature is constant (isothermal), (v) the
water vapor content in the gas is negligible since the temperature
dxCH4 ;G sCH4 ð1 xCH4 ;G Þ þ sCO2 xCH4 ;G Q L;in ð1 xL;CO2 ;in xL;CH4 ;in Þ ¼ Q L;out ð1 xL;CO2 ;out xL;CH4 ;out Þ ð23Þ
¼ ð15Þ
dz QG
xG;CO2 ;out þ xG;CH4 ;out ¼ 1 ð24Þ
dxCO2 ;L sCO2 ð1 xCO2 ;L Þ þ sCH4 xCO2 ;L
¼ ð16Þ
dz QL ½CO2 L;out ¼ H CO2 pG;CO2 ;out ð25Þ
dxCH4 ;L sCH4 ð1 xCH4 ;L Þ þ sCO2 xCH4 ;L ½CH4 L;out ¼ H CH4 pG;CH4 ;out ð26Þ
¼ ð17Þ
dz QL
C.E. Wylock, W.M. Budzianowski / Chemical Engineering Science 170 (2017) 639–652 643
2.3. Model parameters the resulting liquid flow rate leaving the scrubber is then deduced by
1xCO
2 ;L;in;SCR
Q L;out;SCR ¼ Q L;in;SCR 1xCO xCH
.
2 ;L;out;SCR 4 ;L;out;SCR
The expressions correlating the packing characteristics and the
Eqs. (21)–(26) are solved numerically using Q L;in;FLS ¼ Q L;out;SCR ,
physico-chemical are found in the literature. They are presented in
Table 1. xCO2 ;L;in;FLS ¼ xCO2 ;L;out;SCR and xCH4 ;L;in;FLS ¼ xCH4 ;L;out;SCR , with the routine
It is worth to mention that the correlations used for the transfer fsolve, to compute Q G;out;FLS , Q L;out;FLS , xCO2 ;G;out;FLS , xCO2 ;L;out;FLS ,
coefficients are the correlations of Onda et al. (1968), which were xCH4 ;G;out;FLS and xCH4 ;L;out;FLS . The new values of Q G;out;FLS , xCO2 ;G;out;FLS
developed from experimental data for various liquids and operat- and xCH4 ;G;out;FLS are then used to realize a new iteration. This loop
ing conditions. Even though some recent works like Nock et al. is realized until the variations between two successive iterations
(2014) tend to show that the more recent correlations of Billet become slower than a given tolerance.
and Schultes (1993) are slightly better, the correlations of Onda Once the steady-state is reached, the eventually obtained
et al. remain the most commonly used for several industrial pro- Q L;in;SCR is used to calculate the requested scrubber diameter DSCR
cesses. The accuracy depend on the system studied and for exam- for the corresponding superficial velocities in the considered pack-
ple in the system ammonia-carbon dioxide-water-air similar to ing. The diameter is such that the superficial gas velocity corre-
that of the current study the correlation of Onda et al. was found sponds to 60% of the flooding one and it is calculated thank to
the most accurate from several correlations compared the correlations presented in Table 1.
(Budzianowski and Kozioł, 2006). Finally, the minimum scrubber height, HSCR , is estimated by
solving Eqs. (14)–(17) with the ode15s routine, from the bottom
to the top of the scrubber. A much larger height is used as an initial
3. Simulation procedure
guess. HSCR is identified as the height matching the outlet criterion,
i.e. xCO2 ;G;SCR ðHSCR Þ ¼ xCO2 ;G;out;SCR .
The simulations are performed using the computational soft-
ware Matlab. The model can be used to estimate appropriate
device size for reaching a given performance as well as to compute 3.1.2. Stripper
the performance of a given plant. Since the composition of the air injected at the stripper inlet are
known (xCO2 ;G;in;STR ¼ 6 104 , corresponding to the molar ratio of
3.1. Plant scaling 0.04% in the atmospheric air), it is required to impose only the
CO2 mass fraction to reach at the stripper outlet xCO2 ;G;out;STR . The
This procedure is used if the goal is to determine the required global CO2 stripping rate to achieve is calculated as
scrubber and stripper sizes, scrubbing water flow rate and strip- #CO2 ;STR ¼ Q L;in;SCR xCO2 ;L;in;SCR Q L;out;FLS xCO2 ;L;out;FLS . The needed inlet
ping air flow rate for reaching an imposed CO2 abatement at the 1xCO
2 ;G;out;STR
scrubber gas outlet. In this case, three regulation parameters air flow rate is then deduced by Q G;in;STR ¼ #CO2 ;STR xCO xCO
.
2 ;G;out;STR 2 ;G;in;STR
related to the scrubbing water recycling have to be specified: the Knowing the gas and liquid flow rates, the stripper diameter
CO2 mass fraction in the water entering and leaving the scrubber and height, DSTR and HSTR respectively, are estimated using the
and in the stripping air leaving the stripper: xCO2 ;L;in;SCR , xCO2 ;L;out;SCR same procedure than for the scrubber. The estimated HSTR is the
and xCO2 ;G;out;STR , respectively. height matching the condition xCO2 ;L;SCR ðHSCR Þ ¼ xCO2 ;L;out;SCR , with
The scaling is realized in two stages. The first one computes the xCO2 ;L;out;SCR ¼ xCO2 ;L;in;SCR .
characteristics of the scrubber in coupling with the flash tank.
Since the gas entering the scrubber is a mixture between the raw
biogas to treat and the gas leaving the flash tank, which depends 3.2. Performance computation
on the composition of the liquid leaving the scrubber, several iter-
ations of the scrubber-flash loop are performed to compute the Whether the scrubber and stripper diameters and heights are
streams at steady-state. The second stage computes the stripper known and the scrubbing water and stripping air flow rates are
characteristics taking into the liquid stream coming from the flash imposed, or has been estimated by the plant scaling procedure,
tank and the CO2 depletion to reach for reinjection in the scrubber. the performance of each unit of the plant are computed. Due to
scrubbing water recycling, several iterations of the process are
3.1.1. Scrubber-flash realized to ensure the steady-state at a given tolerance.
The actual gas flow rate entering the scrubber is At each iteration, the outlet variables of each unit are computed
Q G;in;SCR ¼ Q G;RAW þ Q G;out;FLS and its composition is given by according to their inlet variables, with the following sequence.
xCO2 ;G;in;SCR ¼ ðQ G;RAW xCO2 ;G;RAW þ Q G;out;FLS xCO2 ;G;out;FLS Þ=Q G;in;SCR and The gas mixture at the scrubber gas inlet is given
xCH4 ;G;in;SCR ¼ ðQ G;RAW xCH4 ;G;RAW þ Q G;out;FLS xCH4 ;G;out;FLS Þ=Q G;in;SCR . At the by Q G;in;SCR ¼ Q G;RAW þ Q G;out;FLS , xCO2 ;G;in;SCR ¼ ðQ G;RAW xCO2 ;G;RAW þ
first iteration, Q G;out;FLS ¼ xCO2 ;G;out;FLS ¼ xCH4 ;G;out;FLS ¼ 0 can be used Q G;out;FLS xCO2 ;G;out;FLS Þ=Q G;in;SCR and xCH4 ;G;in;SCR ¼ ðQ G;RAW xCH4 ;G;RAW þ
as initial guesses. Q G;out;FLS xCH4 ;G;out;FLS Þ=Q G;in;SCR . The characteristics of the liquid inlet
Let xCO2 ;G;out;SCR be the targeted mass fraction to reach at the gas are given by the imposed Q L;in;SCR and by xCO2 ;L;in;SCR ¼ xCO2 ;L;out;STR
scrubber outlet. As an approximation, it can be considered that only and xCH4 ;L;in;SCR ¼ xCH4 ;L;out;STR . The evolution profiles of xCO2 ;G;SCR ,
CO2 is transferred from gas to liquid. Therefore, the global CO2 xCH4 ;G;SCR , xCO2 ;L;SCR , xCH4 ;L;SCR , Q G;SCR and Q L;SCR along the height HSCR
scrubbing rate can be estimated #CO2 ;SCR ¼ Q G;in;SCR are computed by solving Eqs. (14)–(17) with the bvp4c routine,
1xCO ;G;in;SCR
1 1xCO ;G;out;SCR . By imposing the CO2 mass fraction in the water
2 enabling the determination of the outlet variables xCO2 ;G;out;SCR ,
2 xCH4 ;G;out;SCR , xCO2 ;L;out;SCR , xCH4 ;L;out;SCR , Q G;out;SCR and Q L;out;SCR .
at the inlet and the outlet of the scrubber, xCO2 ;L;in;SCR and
At the first iteration, Q G;out;FLS ¼ xCO2 ;G;out;FLS ¼ xCH4 ;G;out;FLS ¼ 0 and
xCO2 ;L;out;SCR , respectively, the requested flow rate is estimated by
1xCO xCO2 ;L;in;SCR ¼ 7 107 and xCH4 ;L;in;SCR ¼ 4 1012 (corresponding to
2 ;L;out;SCR
Q L;in;SCR ¼ #CO2 ;SCR xCO ;L;out;SCR xCO ;L;in;SCR
. Considering that the dissolved the composition at equilibrium with the CO2 and CH4 content in
2 2
methane in the scrubber water is negligible at the inlet the atmosphere, respectively) can be used as initial guesses. In this
(xCH ;L;in;SCR 0) and at equilibrium with the entering gas phase case, a particular procedure is required to ensure the numerical
0 4 1
convergence of the collocation method used in the bvp4c routine.
@xCH ;L;out;SCR ¼ H CH pG;in;SCR x
x CH4 ;G;in;SCR
A, Indeed, the profile of xCH4 ;L;SCR with respect to height is non-
4 4 RT CO2 ;G;in;SCR xCH4 ;G;in;SCR ð1xCO2 ;G;in;SCR xCH4 ;G;in;SCR Þ
qL MCO
þ M þ Minert
2 CH4
644 C.E. Wylock, W.M. Budzianowski / Chemical Engineering Science 170 (2017) 639–652
Table 2
Experimental data on collected in the literature for the scrubber model validation.
N° pG;in;SCR DSCR HSCR Q G;in;SCR Q L;in;SCR xCO2 ;G;in;SCR xCO2 ;G;out;SCR pFLS Ref.
(MPa) (m) (m) (m3/h) (kg/s) (MPa)
A1 1.0 0.15 3.0 20 1.00 0.40 0.070 – Nock et al. (2014), Läntelä and
A2 1.00 0.35 0.070 Luostarinen (2013)
A3 1.03 0.40 0.065
A4 1.03 0.35 0.065
A5 1.06 0.40 0.060
A6 1.06 0.35 0.060
A7 1.08 0.40 0.055
A8 1.08 0.35 0.055
A9 1.11 0.40 0.050
A10 1.11 0.35 0.050
A11 1.14 0.40 0.045
A12 1.14 0.35 0.045
A13 1.17 0.40 0.040
A14 1.17 0.35 0.040
A15 1.19 0.40 0.035
A16 1.19 0.35 0.035
A17 1.22 0.40 0.030
A18 1.22 0.35 0.030
B1 0.8 0.15 3.5 1.0 0.42 0.32 0.050 – Chandra et al. (2012)
B2 1.0 0.040
C1 0.3 0.10 1.0 0.3 0.3 0.40 0.108 – Huang et al. (2015)
D1 0.8 0.10 1.0 242.3 13.3 0.452 0.0047 0.3 Götz et al. (2011)
D2 12.06 0.0036 0.4
C.E. Wylock, W.M. Budzianowski / Chemical Engineering Science 170 (2017) 639–652 645
Fig. 3. Evolution of xCO2 ;G , xCH4 ;G , xCO2 ;L and xCH4 ;L with the height in the scrubber (a) and in the stripper (b).
646 C.E. Wylock, W.M. Budzianowski / Chemical Engineering Science 170 (2017) 639–652
The evolution of the CO2 and CH4 mass fraction in the gas and in scrubber, this assumption is not rigorous but seems nevertheless
the liquid of the scrubber and the stripper can also be examined. reasonable. Moreover, this estimation considers an adiabatic
They are presented in Fig. 3. column, whereas in a real system there would be heat transfer
with the ambient air through the column walls. The actual temper-
4.3. Thermal effect evaluation ature deviation should be therefore smaller than the estimated
one.
Thanks to the mass transfer rate evaluated in the reference case,
the heat release associated with the CO2 absorption and its reac- 4.4. Influence of regulation parameters
tion with water can be estimated. It is realized to assess the
assumption validity of isothermal transfer. According to The effect of the three regulation parameters xCO2 ;L;in;SCR ,
Taghizadeh et al. (2001), the enthalpy of CO2 absorption in water xCO2 ;L;out;SCR and xCO2 ;G;out;STR are examined in order to assess their
is DEabs ¼ 20:3 kJ/mol. The enthalpy of reaction of CO2 with water influence on the scaling. Indeed, these parameters have to be
to form HCO 3 is evaluated to DEr ¼ 46 kJ/mol thanks to the selected appropriately since they may influence the column sizes
enthalpy of formation provided in Planck (1903). The CO2 chemical (determining the capital expenditure) and flows to ensure (deter-
conversion rate is deduced from the transfer rate by mining the operating expenditure) for the same imposed CO2
sCO2 ½HCO þ
3 L ½H L
þ . Considering an adiabatic column and neglecting abatement.
M CO
2
K 1 þ½HCO3 L þ½H L
Variations are then applied to xCO2 ;L;in;SCR , xCO2 ;L;out;SCR and
the heat transfer to the gas phase, the local liquid temperature
xCO2 ;G;out;STR to run simulations, keeping the target of 2% molar CO2
variation due to the heat of absorption and reaction is therefore
at the scrubber outlet. The results are examined in terms of varia-
calculated by:
! tions on DSCR , HSCR , Q L;in;SCR , DSTR , HSTR and Q G;in;STR , which are pre-
dT L sCO2 ½HCO3 L ½Hþ L sented in Figs. 3–5.
C p;L Q L ¼ DEabs þ DEr ð28Þ
dz MCO2 K 1 þ ½HCO3 L þ ½Hþ L It is shown in Fig. 5 that xCO2 ;L;in;SCR has a slight influence on the
requested HSTR (below 3% for 10% variations of xCO2 ;L;in;SCR ). For the
where C p;L is the liquid specific heat capacity. other ones, the influence can be considered negligible. Therefore,
The liquid temperature evolution with the height in the scrub- CO2 ;L;in;SCR is not a critical parameter, at least in the investigated
ber and the stripper are presented in Fig. 4. It is observed that the range, and it is low enough to ensure good performance of the
temperature deviation between inlet and outlet is 2.7 K in the scrubber.
scrubber and 0.7 K in the stripper. It seems therefore relevant to It is observed in Fig. 6 that the values of xCO2 ;L;out;SCR strongly
consider that the transfer is isothermal in the stripper. In the influence the requested scrubber height to maintain the same
Fig. 5. Variations of scrubber and stripper characteristics with variations of xCO2 ;L;in;SCR .
Fig. 6. Variations of scrubber and stripper characteristics with variations of xCO2 ;L;out;SCR .
C.E. Wylock, W.M. Budzianowski / Chemical Engineering Science 170 (2017) 639–652 647
CO2 abatement performance, while it slightly affects the other kCH4 ;out;STR . They respectively decrease and increase while Q L;in;SCR
characteristics. The variations of Q G;in;STR reach about 1% for 10% increase, meaning that a high Q L;in;SCR is favorable for the CO2
variations of xCO2 ;L;out;SCR . They are below 1% for DSCR , Q L;in;SCR , DSTR scrubbing efficiency but it can be set off by an increase of the
and HSTR . methane loss. Q L;in;SCR is therefore an operating parameter that
Fig. 7 shows the characteristics variations with xCO2 ;G;out;STR . It has to be determined and controlled with accuracy in order to opti-
is observed that it is significant for HSTR and Q G;in;STR , while it mize and to maintain the performance of the plant.
has no influence on the scrubber characteristics. It shows that
decrease the targeted CO2 content in the air leaving the stripper 4.5.2. Stripping air flow rate
is interesting, in the investigated range, because it requires a The stripping air flow rate Q G;in;STR appears to be also an impor-
decrease of the stripper height but it is offset by the need of tant parameter, according to Fig. 9. As it could be expected, when it
increasing also the stripping air flow rate to keep constant the increases, the methane loss is slightly increased while xCO2 ;G;out;SCR is
CO2 abatement. decreased. It is therefore also a parameter to control but the sensi-
tivity of the system is weaker compared to the previous one.
4.5. Influence of operating conditions
4.5.3. Raw biogas flow rate and composition
The sensitivity of the plant configuration with respect to the Several external parameters may affect a digester during oper-
operating parameters is investigated. Starting from the reference ation (temperature, feed composition . . .), therefore the character-
case identified in 4.2, QL;in;SCR , QG;in;STR , Q G;RAW , xCO2 ;G;RAW , T, pG;in;SCR , istics of the raw biogas to treat may variate more or less with time.
pG;in;STR , pG;FLS are varied and their influence on the gas streams at It is thus important to predict how it would affect the upgrading
the scrubber and stripper outlet are computed. The performance plant performance in order to adapt the operating parameters for
is evaluated on the basis of two criteria: the CO2 ratio at the scrub- keeping constant the performance. Fig. 10 presents influence of
ber outlet xCO2 ;G;out;SCR and the methane loss at the stripper outlet the raw biogas flow rate Q G;RAW . It is observed that an increase of
defined by kCH4 ;out;STR ¼ Q G;out;STR yCH4 ;G;out;STR . They are presented Q G;RAW increases considerably xCO2 ;G;out;SCR in relative variation,
in relative variations in Figs. 8–15, respectively. while it tends to decrease the methane loss. In such a situation,
an appropriate corrective action should be an increase of Q L;in;SCR .
4.5.1. Scrubbing water flow rate In addition, if the Q G;RAW decreases, the scrubbing is more efficient
It appears from Fig. 8 that the scrubbing water flow rate Q L;in;SCR and then the treated biogas is richer in CH4. This tends to increase
has a considerable influence on the scrubbing performance. Indeed, the CH4 partial pressure, leading to an enhanced absorption in the
even small variations can affect significantly xCO2 ;G;out;SCR and scrubbing water and inducing a higher methane loss. The raw bio-
Fig. 7. Variations of scrubber and stripper characteristics with variations of xCO2 ;G;out;STR .
Fig. 8. Variations of CO2 ratio at the scrubber outlet and CH4 loss at the stripper outlet with variations of scrubbing water flow rate.
648 C.E. Wylock, W.M. Budzianowski / Chemical Engineering Science 170 (2017) 639–652
Fig. 9. Variations of CO2 ratio at the scrubber outlet and CH4 loss at the stripper outlet with variations of stripping air flow rate.
Fig. 10. Variations of CO2 ratio at the scrubber outlet and CH4 loss at the stripper outlet with variations of raw biogas flow rate.
Fig. 11. Variations of CO2 ratio at the scrubber outlet and CH4 loss at the stripper outlet with variations of raw biogas CO2 fraction.
gas flow rate needs therefore to be monitored with accuracy in loss is also increased when the temperature decreases. It can
order to adapt efficiently the operating conditions. Similarly, if results from a combination of an increase of the CH4 absorption
the CO2 content of the raw biogas xCO2 ;G;RAW increases, the same in the scrubber (due to higher solubility and partial pressure)
trend is observed than for Q G;RAW , as it can be seen in Fig. 11. and a decrease of the its desorption in the flash tank.
Fig. 12. Variations of CO2 ratio at the scrubber outlet and CH4 loss at the stripper outlet with variations of temperature.
Fig. 13. Variations of CO2 ratio at the scrubber outlet and CH4 loss at the stripper outlet with variations of scrubber inlet pressure.
Fig. 14. Variations of CO2 ratio at the scrubber outlet and CH4 loss at the stripper outlet with variations of stripper inlet pressure.
for corrective actions to maintain the plant performance when the if pG;FLS because it hinders the CH4 desorption by the flash opera-
biogas characteristics are varying. tion. It is favorable to use the lowest possible pressure. However,
The trend is different for pG;in;STR , as it can be seen in Fig. 14. this tends also to increase xCO2 ;G;out;SCR , because more CO2 are rein-
When pG;in;STR is increased, xCO2 ;G;out;SCR is increased and kCH4 ;out;STR jected in the scrubber. Therefore, pG;FLS has to be regulated in cou-
is slightly decreased. It is therefore favorable to use the lowest pos- pling with other operating parameters like Q L;in;SCR and pG;in;SCR .
sible pressure at the stripper, since high pressure hinders the CO2
desorption from the scrubbing water and therefore decrease the 4.6. Influence of model parameters
treatment performance.
The opposite trend is observed for pG;FLS , as presented in Fig. 15, In order to calibrate the proposed model with experimental
with a more significant influence on kCH4 ;out;STR . This latter increases results, a particular attention should be paid to the accurate esti-
650 C.E. Wylock, W.M. Budzianowski / Chemical Engineering Science 170 (2017) 639–652
Fig. 15. Variations of CO2 ratio at the scrubber outlet and CH4 loss at the stripper outlet with variations of flash tank pressure.
mation of model parameters. On the one hand, the model is inher- umn independently. However, the need of accuracy is higher for
ently sensitive to the Henry coefficients (H CO2 and H CH4 ) and the the CO2 transfer coefficient than for the CH4 one. This need is also
gas-liquid transfer coefficients in both scrubber (aK L;CO2 ;SCR and higher for the scrubber than for the stripper. These trends are
aK L;CH4 ;SCR ) and stripper (aK L;CO2 ;STR and aK L;CH4 ;STR ). It can be shown quantified in Figs. 18–21. It is observed that the highest influence
that inaccuracy on any of these parameters may influence several
outlet variables. Their combined inaccuracies can therefore com-
plicate interpretation of experiments. For instance, they can skew
the study of the influence of a given regulation parameters by
enhancing or damping its effects. On the other hand, it is evaluated
that the model is rather insensitive to the reaction equilibrium
constant (K 1 ).
The sensitivity of the performance computed by the models
with respect to H CO2 and H CH4 are presented in Figs. 16 and 17,
respectively. The model predict a considerable influence of H CO2
both xCO2 ;G;out;SCR and kCH4 ;out;STR . As it could be expected, it is
observed in Fig. 16 increasing values of H CO2 decrease significantly
xCO2 ;G;out;SCR . However, it increases also kCH4 ;out;STR because the scrub-
bing water in the scrubber is in contact with a CH4 richer gas
stream. The effect of H CH4 is more moderate, as it can be seen in
Fig. 17. Increasing values of H CH4 increases obviously kCH4 ;out;STR
and increases also slightly xCO2 ;G;out;SCR .
The Henry coefficients can be estimated thanks to various cor-
relations available in the literature. It is nevertheless recom-
mended to assess the accuracy of these expressions, since it is Fig. 17. Variations of CO2 ratio at the scrubber outlet and CH4 loss at the stripper
shown that it can significantly affect the process performance. outlet with variations of the CH4 Henry coefficient H CH4 .
Concerning the transfer coefficient, it is recommended to esti-
mate the CO2 and CH4 gas-liquid transfer coefficient on each col-
Fig. 16. Variations of CO2 ratio at the scrubber outlet and CH4 loss at the stripper Fig. 18. Variations of CO2 ratio at the scrubber outlet and CH4 loss at the stripper
outlet with variations of the CO2 Henry coefficient H CO2 . outlet with variations of the CO2 transfer coefficient in the scrubber aK L;CO2 ;SCR .
C.E. Wylock, W.M. Budzianowski / Chemical Engineering Science 170 (2017) 639–652 651
is for aK L;CO2 ;SCR . It can be seen in Fig. 18 that both xCO2 ;G;out;SCR and
kCH4 ;out;STR decrease when aK L;CO2 ;SCR . It means that maximize the
transfer efficiency of the scrubber has a considerable favorable
effect on the global performance of the plant. By contrast, the influ-
ence of aK L;CH4 ;SCR is slight. Surprisingly, it is observed in Fig. 19 that
kCH4 ;out;STR decreases slightly for increasing values of aK L;CH4 ;SCR ,
while there is no clear trend for xCO2 ;G;out;SCR . The influence of
aK L;CO2 ;STR is illustrated in Fig. 20. It is observed that increasing val-
ues of aK L;CO2 ;STR decrease xCO2 ;G;out;SCR . The maximization of the
transfer efficiency in the stripper has a significant considerable
favorable effect on the global performance, at least regarding to
the CO2 abatement. It is explained by the fact that a high transfer
coefficient in the stripper favors the CO2 depletion in the scrubbing
water, which thus enhances the CO2 transfer driving force in the
scrubber. However, it is also observed that it induces a slight
increase of kCH4 ;out;STR . Finally, it is observed in Fig. 21 that
aK L;CH4 ;STR has a very slight influence of kCH4 ;out;STR , which increases
for increasing values of aK L;CH4 ;STR and almost no influence on
xCO2 ;G;out;SCR .
Fig. 19. Variations of CO2 ratio at the scrubber outlet and CH4 loss at the stripper
outlet with variations of the CH4 transfer coefficient in the scrubber aK L;CH4 ;SCR . These results shows therefore that experimental investigation
needs to focus on the CO2 and CH4 Henry coefficients and on the
CO2 gas-liquid transfer coefficient in the scrubber and in the
stripper.
5. Conclusions
A first improvement of the current model would be realized by Edwards, T.J., Maurer, G., Newman, J., Prausnitz, J.M., 1978. Vapor-liquid equilibria
in multicomponent aqueous solutions of volatile weak electrolytes. A.I.Ch.E. J.
including the temperature dependence of the physico-chemical
24, 966–976.
parameters at the local scale and by taking into account the water Götz, M., Köppel, W., Reimert, R., Graf, F., 2011. Optimierungspotenzial von
evaporation/condensation in the mass transfer equations. A deeper Wäschen zur Biogasaufbereitung. Teil 1-Physikalische Wäschen. Chem. Ing.
improvement would be provided by integrating a more realistic Tech. 83 (6), 858–866.
Guo, H., Chen, Y., Lu, W., Li, L., Wang, M., 2013. In situ Raman spectroscopic study of
description of the hydrodynamic in the columns, for instance by diffusion coefficients of methane in liquid water under high pressure and wide
coupling the Matlab code with a 2-D dimensional model of the col- temperatures. Fluid Phase Equilibria 60, 274–278.
umn considered as porous medium thanks to the COMSOL Multi- Huang, L., Qing, C., Tao, H., Wang, Y., Liu, X., 2015. Experimental study for biogas
upgrading by water scrubbing under low pressure. Open Chem. Eng. J. 9, 34–38.
physics software. Similarly, it would be advantageous to take Läntelä, J., Luostarinen, J., 2013. D5.4 Results of design, construction and testing of
into account the non-equilibrium effect in the flash tank modelling. lowcost modular biogas upgrading systems. Deliverable of EU FP7VALORGAS
Indeed, it is considered in this work as instantaneously at equilib- project.
Nock, W., Walker, M., Kapoor, R., Heaven, S., 2014. Modeling the water scrubbing
rium, eliminating its volume in equations. Such an improved process and energy requirements for CO2 capture to upgrade biogas to
model would enable an accurate estimation of the required biomethane. Ind. Eng. Chem. Res. 53 (32), 12783–12792.
volume. Onda, K., Takeuchi, H., Okumoto, Y., 1968. Mass transfer coefficients between gas
and liquid phases in packed columns. J. Chem. Eng. Jpn.
Planck, M., 1903. Treatise on Thermodynamics. Longmans, Green, London.
Acknowledgments Reid, R.C., Prausnitz, J.M., Poling, B.E., 1986. The Properties of Gases and Liquids.
MacGraw-Hill, New York.
Ryckebosch, E., Drouillon, M., Vervaeren, H., 2011. Techniques for transformation of
The support obtained from the members of the Renewable
biogas to biomethane. Biomass Bioenergy 35, 1633–1645.
Energy and Sustainable Development (RESD) Group (Poland) under Sander, P., 1999. Compilation of Henry’s Law constants for inorganic and organic
the grant no. RESD-RDG04/2016 is greatly acknowledged. species of potential importance in environmental chemistry. Available: http://
www.mpch-mainz.mpg.d/~sander/res/henry.html.
Sun, Q., Li, H., Yan, J., Liu, L., Yu, Z., Yu, X., 2015. Selection of appropriate biogas
References upgrading technology - a review of biogas cleaning, upgrading and utilisation.
Renew. Sustain. Energy Rev. 51, 521–532.
Abatzoglou, N., Boivin, S., 2009. A review of biogas purification processes. Biofuels Taghizadeh, M., Jallut, C., Tayakout-Fayolle, M., Lieto, J., 2001. Non-isothermal gas–
Bioprod. Biorefin. 3, 42–71. liquid absorption with chemical reaction studies: temperature measurements
Beil, M., Beyrich, W., 2013. Biogas upgrading to biomethane. In: Wellinger, A., of a spherical laminar film surface and comparison with a model for the CO2/
Murphy, J., Baxter, D. (Eds.), The Biogas Handbook. Woodhead Publishing, pp. NaOH system. Chem. Eng. J. 82 (1–3), 143–148.
342–377. Thrän, D., Billig, E., Persson, T., Svensson, M., Daniel-Gromke, J., Ponitka, J., Seiffert,
Billet, R., Schultes, M., 1993. Predicting mass transfer in packed columns. Chem. Eng. M., Baldwin, J., Kranzl, L., 2016. Biomethane - status and factors affecting
Technol. 77, 1–9. market development and trade. In: IEA Task 40 and Task 37 Joint Study.
Budzianowski, W.M., 2012. Benefits of biogas upgrading to biomethane by high- Trambouze, P., Euzen, J.-P., 2002. Chemical Reactors: From Design to Operation.
pressure reactive solvent scrubbing. Biofuels Bioprod. Biorefin. 6 (1), 12–20. Editions Technip, Paris.
Budzianowski, W.M., Kozioł, A., 2006. Stripping of ammonia from aqueous solutions Tsonopoulos, C., 1976. Ionization constants of water pollutants. J. Chem. Eng. Data
in the presence of carbon dioxide. Effect of negative enhancement of mass 21, 190–193.
transfer. Chem. Eng. Res. Des. 83 (A2), 196–204. Vas Bhat, R., Kuipers, J., Versteeg, G., 2000. Mass transfer with complex chemical
Budzianowski, W.M., Wylock, C.E., Marciniak, P.A., 2017. Power requirements of reactions in gas-liquid systems: two-step reversible reactions with unit
biogas upgrading by water scrubbing and biomethane compression: stoichiometric and kinetic orders. Chem. Eng. J. 76 (2), 127–152.
comparative analysis of various plant configurations. Energy Convers. Manage. Versteeg, G.F., Van Swaaij, W.P.M., 1988. Solubility and diffusivity of acid cases
Cents, A.H.G., Brilman, D.W.F., Versteeg, G.F., 2005. CO2 absorption in carbonate/ (CO2, N2O) in aqueous alkanolamine solutions. J. Chem. Eng. Data 33, 29–34.
bicarbonate solutions: the Danckwerts-criterion revisited. Chem. Eng. Sci. 60, Whitman, W.G., 1923. The two-film theory of absorption. Chem. Metal. Eng. 29 (4),
5830–5835. 146–148.
Chandra, R., Vijay, V., Subbarao, P., 2012. Vehicular quality biomethane production Wylock, C., Dehaeck, S., Cartage, T., Colinet, P., Haut, B., 2011. Experimental study of
from biogas by using an automated water scrubbing system. ISRN Renew. gas-liquid mass transfer coupled with chemical reactions by digital holographic
Energy. no. 2012:904167. interferometry. Chem. Eng. Sci. 66, 3400–3412.
Coulson, J.M., Richardson, J.F., 1999. . Chemical Engineering, 6 ed., vol. 1. Wylock, C., Antuñano, N., Arias, P.L., Haut, B., 2014. Analysis of the simultaneous
Butterworth Heinemann. gas-liquid CO2 absorption and liquid-gas NH3 desorption in a
Cowie, Watt, 1971. Diffusion of methane and chloromethanes in air. Can. J. Chem. hydrometallurgical Waelz oxides purification process. Int. J. Chem. Reactor
49, 74. Eng. 12, 1–14.
Danckwerts, P.V., 1970. Gas-Liquid Reactions. McGraw-Hill.
Danckwerts, P., Sharma, M., 1966. Absorption of carbon dioxide into solutions of
alkalis and amines. Chem. Eng., 244–280