Вы находитесь на странице: 1из 12

Fuel 244 (2019) 184–195

Contents lists available at ScienceDirect

Fuel
journal homepage: www.elsevier.com/locate/fuel

Full Length Article

Hydrogen production through steam reforming of bio-oils derived from T


biomass pyrolysis: Thermodynamic analysis including in situ CO2 and/or H2
separation
M.A. Soria , Diogo Barros, L.M. Madeira

LEPABE, Department of Chemical Engineering, Faculty of Engineering, University of Porto, Rua Dr. Roberto Frias s/n, 4200-465 Porto, Portugal

GRAPHICAL ABSTRACT

ARTICLE INFO ABSTRACT

Keywords: A thermodynamic analysis based on the minimization of the Gibbs free energy method was performed in order to study the
Bio-oil steam reforming of real bio-oils (obtained from biomass fast pyrolysis) from different sources. For comparison purposes,
Steam reforming four types of reactors were used, namely: traditional reactor (TR), membrane reactor (MR) with H2 separation, sorption-
Hydrogen enhanced reactor (SER) with CO2 sorption, and sorption-enhanced membrane reactor (SEMR) with both H2 and CO2
Sorption-enhanced reactor
separation. The CO2 capture was studied varying the sorbent (CaO) to feed molar ratio (SFR) while the effect of the H2
Membrane reactor
Sorption-enhanced membrane reactor
removal was assessed with the use of a removal fraction (fH2) of 0 to 0.8 (typical recoveries for an H2-selective membrane).
The simulated results obtained for the TR showed a good agreement with those reported in literature ex-
perimental works at similar operation conditions. The optimum conditions for the production of hydrogen were
obtained in the SEMR in which it was possible to obtain 99% of the maximum theoretical yield for spruce bio-oil
and 97% of the maximum yield for wheat bio-oil. Furthermore, this reactor configuration minimizes CH4, H2,
CO, CO2 and coke yields. In addition, the effect of pressure in a range of 1 to 10 bar was also assessed for the
reactor configurations with H2 removal. It was shown that the increase of pressure decreases H2 yield; never-
theless, this can be offset by moderately increasing the WFR.

1. Introduction threats created by their exploitation, such use is a growing concern for
the society. This has led to an increasing demand for clean and re-
Fossil fuels are the main energy sources at present, but considering newable energy sources. Biomass is seen as an attractive primary en-
their non-sustainable consumption coupled with the environmental ergy source (renewable resource) if managed in a sustainable way.


Corresponding author.
E-mail address: masoria@fe.up.pt (M.A. Soria).

https://doi.org/10.1016/j.fuel.2019.01.156
Received 5 November 2018; Received in revised form 19 December 2018; Accepted 27 January 2019
0016-2361/ © 2019 Elsevier Ltd. All rights reserved.
M.A. Soria et al. Fuel 244 (2019) 184–195

Nomenclature T Temperature, [K]


V Volume, [m3]
List of variables Yieldi Yield of the species i, dimensionless
Hr298K Standard enthalpy of reaction, [kJ/mol]
Symbol Definition, Unit µi Chemical potential of species i, [kJ/mol]
fH2 H2 separation factor, dimensionless
G Gibbs free energy, [kJ] List of acronyms
k Represents each sub-separator (from 1 to N), dimension-
less Acronym Definition
mbio oil (feed) Molar flow rate of bio-oil in the feed of the R1 Gibbs MR Membrane Reactor
reactor, [mol/h] RWGS Reverse Water-Gas Shift
mi ,k Outlet molar flow of species i in the sub-separator k, [mol/ SEMR Sorption-Enhanced Membrane Reactor
h] SER Sorption-Enhanced Reactor
miout
,N +1 Molar flow rate of species i in the outlet stream of the SFR Sorbent to Feed Molar Ratio
(RN + 1) Gibbs Reactor, [mol/h] SR Steam Reforming
ni Number of moles of each component in the system, [mol] TR Traditional Reactor
P Pressure, [kPa] WFR Water to Feed Ratio
S Entropy, [kJ/K] WGS Water-Gas Shift

Furthermore, it can be converted in a large array of biofuels through viscosity and coking result in problems for the combustion equipment.
several processes, whose combustion has low (or overall null) carbon They can also be used to extract some chemicals like acetic acid or
emissions [1]. Biomass conversion into different kinds of fuels can be methanol, but these compounds can usually be obtained at a lower cost
done through thermal, biological or mechanical processes [2–5]. from other feedstock. Finally, they can be converted into transport fuels
Among the thermal processes, combustion is used in the production of like bio-diesel or hydrogen; even though there are still some challenges,
heat that can be used to generate electrical energy (e.g. via a steam this is an attractive application for bio-oils [11]. Hydrogen is a great
turbine). Gasification is mainly used to produce syngas that may have energy solution as it has a high calorific power and also acts as an
different downstream applications [6]. Pyrolysis is the thermal de- excellent energy vector. Nowadays, 95% of the world’s production of
composition of the biomass in the absence of oxygen, yielding charcoal, hydrogen uses fossil fuels as raw material [12], mainly through me-
fuel-gas and bio-oil, also called pyrolysis oil; the quantity in which thane steam reforming. However, the use of this technology for the
these products are formed as well as their content on oxygenated reforming of pyrolysis oils is particularly interesting because not only it
compounds depends on the residence time, heating rate and tempera- is an environmentally friendly alternative, but also, it is an application
ture of the process and on the composition of the biomass source. There that can turn a product that is otherwise viewed as waste (biomass) into
are different modes of pyrolysis: fast, intermediate and slow [5,7,8]. a useful fuel. Experimental studies performed in traditional reactors
The fast pyrolysis (which occurs at high heating rates, 450–600 °C and (TRs) with both raw bio-oil and bio-oil model compounds like acetic
less than 2 s of residence time) provides a higher yield in bio-oil, while acid, acetone and phenol showed that catalytic bio-oil steam reforming
the other sub-products (charcoal and fuel-gas) can be used to generate is a viable technology [13–18].
heat for the pyrolysis process itself [5]. The big advantage of producing The possible reactions involved in bio-oil steam reforming process
these bio-oils is that they have a volumetric energy density up to ten are presented in Table 1; these reactions were considered based on
times higher than that of biomass; consequently, they are much more previous studies carried out for steam reforming of oxygenated com-
suited (more cost-efficient) for transportation [9,10]. These oils can be pounds [13,15,19–21]. The bio-oil derived from biomass fast pyrolysis
used directly as combustion fuels, but their poor volatility and high and steam are converted into synthesis gas (mixture of carbon

Table 1
Reactions considered for the reforming process simulations.
Reaction Hr298K (kJ mol−1) No.

Steam Reforming
>0 (1)
Cn Hm Ok + (n k ) H2 O (
nCO + n +
m
2 )
k H2
Water-gas shift
CO + H2 O CO2 + H2 −41 (2)
Overall steam reforming equation
>0 (3)
Cn Hm Ok + (2n k ) H2 O (
nCO2 + 2n +
m
2 )
k H2
Thermal decomposition of oxygenated compounds
Cn Hm Ok H2, CO, smaller oxygenated compounds hydrocarbons, coke >0 (4)
Methanation
CO + 3H2 CH4 + H2 O −206 (5)
Methane steam reforming
CH4 + H2 O 3H2 + CO a 206 (6)
Methane dry reforming
CH4 + CO2 2CO + 2H2 b 247 (7)
Carbon Formation
2CO CO2 + C −172 (8)
CH4 2H2 + C 75 (9)
CO + H2 H2 O + C −131 (10)
CO2 + 2H2 2H2 O + C c −90 (11)

a b c
Reverse of Eq. (5); Sum of reverse of Eq. (2) and Eq. (6); Sum of Eq.(10) and reverse of Eq.(2).

185
M.A. Soria et al. Fuel 244 (2019) 184–195

approach involves the minimization of the Gibbs free energy and allows
for an easy achievement of convergence in computation without the
need to select the chemical reaction(s) occurring and an accurate esti-
mation of the initial equilibrium composition, only needing to define
the chemical species involved according to the possible reaction(s)
[30].
The Gibbs free energy (G) depends on the temperature (T), pressure
(P) and molar quantities of the N components present in the system. Its
differential form can be written as follows:
N
dG = SdT + VdP + i=i
µi dni (12)
where S stands for the entropy, V is the volume, ni the number of moles
Fig. 1. Scheme of the sorption-enhanced membrane reactor for bio-oil steam of each component in the system and µi is the respective chemical po-
reforming. tential. When the system operates in isothermal and isobaric conditions,
the differential equation becomes:
monoxide and hydrogen) – Eq. (1). Following this step, carbon mon- dG =
N
µi dni
i=i (13)
oxide is converted into hydrogen and carbon dioxide through the water-
gas shift (WGS) reaction – Eq. (2). There are also secondary reactions The used approach implies that equation (12) is equal to zero since
associated with this process that form several by-products (see Table 1). at the equilibrium the system has the minimum Gibbs free energy. A
To minimize the formation of coke, CO and CH4 and thus enhance the more detailed description of the methodology is available elsewhere
H2 yield, different types of multifunctional reactors can be used, which [31,32].
allow to shift the equilibrium of the reactions involved by selectively With the aim of studying the influence of T, P and feed composition
removing the reaction products (CO2 and/or H2). Such multifunctional on the hydrogen yield, the thermodynamic analysis of the different bio-
reactors can be for instance a membrane reactor (MR), a sorption-en- oils in the steam reforming process was performed using Aspen Plus®
hanced reactor (SER) or a sorption-enhanced membrane reactor (SEMR; V8.8 software. To calculate the equilibrium composition of the system
Fig. 1) [22,23]. The MR allows the H2 produced to be selectively re- at certain T, P and feed composition, the Gibbs reactor (RGIBBS) block,
moved from the reaction zone as a permeate stream. Palladium-based which is a reactor model based on the minimization of the Gibbs free
membranes have been largely studied and applied in hydrogen se- energy method, was utilized. As mentioned, the method herein applied
paration/production processes, since they are very selective towards H2 does not require reaction stoichiometry input but it is only necessary to
[24–26]. In the case of the SER, the CO2 capture can be achieved by provide information about the chemical compounds present in the
filling the reactor with a mixture of steam reforming catalyst and CO2 system (reagents and products), which were defined according to the
sorbent. One of the most used CO2 sorbents is calcium oxide (CaO), reactions present in Table 1. The compounds common to all the simu-
which has been previously used in theoretical glycerol reforming stu- lations are H2, CO, CO2, H2O, CH4 and carbon, as well as smaller hy-
dies and in experimental acetic acid reforming, employed as a bio-oil drocarbons (C2H6 to C4H10) and oxygenated compounds (CH4O to
model compound [21,27]. To further shift the steam reforming reaction C4H10O) that can be formed due to reaction (4); however, these small
to the reaction products, the in-situ separation of both CO2 and H2 (that hydrocarbons and oxygenated compounds were not found in all the
are involved in most of the equilibrium-limited reactions reported in simulations presented herein.
Table 1) was carried out. The simultaneous removal of CO2 and H2 from In the cases where the type of reactor involves a hydrogen perme-
the reaction zone provides a hybrid new reactor concept (SEMR) that able membrane such as MR and SEMR, the use of only one RGIBBS
can originate a substantial improvement in the hydrogen yield and block is not enough to perform the simulation. Therefore, to simulate
minimize the undesired by-products formation [23]. the in-situ H2 removal a sequential modular approach is employed
Although there are some thermodynamic studies of bio-oil steam (Fig. 2). The MR (or SEMR) is divided into several (N + 1) sub-re-
reforming, these were carried out considering only model compounds formers (represented as R in Fig. 2) and (N) membrane sub-separators
[20,28,29]. The purpose of this work is to further explore biomass (represented as SEP). The latter is a block that allows to separate che-
valorisation through the steam reforming of pyrolysis oils, analysing the mical compounds which number (N) depends on the defined split
process attainable efficiency, i.e. on the thermodynamic equilibrium. fraction or flow (herein called hydrogen removal or recovery fraction).
For comparison reasons, several real fractions of bio-oil derived from This split fraction is in practice related to membrane factors like its
biomass fast pyrolysis, whose composition depends on the sources (e.g. selectivity, permeability, thickness, area and process conditions (e.g.
wood, sugar, palm oil, etc.), are considered. Moreover, the performance temperature, sweep-gas flow or pressure across the membrane). This
of different types of reactors, such as TR, MR, SER and SEMR, will be methodology has been also implemented in previous studies
compared with the aim of identifying the optimal operating conditions [21,31,33].
for high-purity hydrogen production. In order to assess the performance reached, for a given set of con-
In this work, and as far as we know, a thermodynamic study assessing ditions employed in the simulations, the yield of the species at equili-
bio-oil steam reforming with several real bio-oil mixtures and comparing brium was calculated according to the equation bellow:
different multifunctional hybrid reactors is performed for the first time. N
miout
, N +1 + k = 1 mi,k
Although the optimal operating conditions correspond to the equilibrium, Yieldi = mbio oil (feed ) (14)
they provide valuable information for operation under real conditions since
they indicate the limits that can be reached and the variables that can be where mbio oil (feed) stands for the molar flow rate of bio-oil in the feed of
modified in order to optimize the hydrogen yield and avoid (or minimize) the R1 unit, miout
, N + 1 is the molar flow rate of species i in the outlet stream
the coke formation, which is responsible for catalyst deactivation. N
of the (RN + 1) RGIBBS unit and k = 1 mi ,k is the sum of the outlet
molar flow of each chemical species i in the N sub-separators (please
2. Methodology refer to Fig. 2). This last sum will always be equal to zero, except for
hydrogen in the MR and in the SEMR since the N sub-separators act as
To determine the equilibrium composition of the chemical species in H2 perm-selective membranes (with infinite selectivity towards H2).
the system, a non-stoichiometric approach was employed. This The separation factor (fH2) is defined as shown in Eq. (15):

186
M.A. Soria et al. Fuel 244 (2019) 184–195

Fig. 2. Diagram of modular approach used for simulation of bio-oil stem reforming in the different reactor configurations.

Whatever the original biomass, the presence of hydrocarbons and oxy-


Table 2 genated compounds were not observed, meaning that there is total con-
Reactions involved in the carbon dioxide capture with calcium oxide.
version in all the simulations carried out, even bellow the stoichiometric
Reaction Hr298K (kJ mol−1) No. WFR, not only because of the steam reforming process itself, but also due to
the thermal degradation of the chemical compounds present in the feed – cf.
CaO + H2 O Ca (OH ) 2 −65 (16) Eq. 4. Such equations are thus indicated as being irreversible in Table 1.
Ca (OH )2 + CO2 CaCO3 + H2 O −113 (17)

3.1.1. Effect of water to feed molar ratio


N
k=1
m H2, k The variation of the products yield as a function of the WFR for
fH2 = N
m H2, k + m Hout (15) different bio-oils are represented in Fig. 3. In order to compare the
k=1 2, N + 1
results obtained for a conventional reactor with those obtained for a
where k represents each sub-separator, m H2, k represents the sum of
N
k=1 membrane reactor and a sorption-enhanced (membrane) reactor (in the
the hydrogen molar flow rate in the permeate stream of each sup-se- next sections), a temperature of 400 °C was chosen; such temperature is
parator and m Hout
2, N + 1
the outlet hydrogen molar flow rate in the last sub- suitable to operate with both Pd-Ag membrane and CaO sorbent.
reformer. The number of sub-separators needed in the simulation de- As shown in Fig. 3, all the bio-oils exhibit a similar trend. As it can
pends on the separation factor set. The higher the separation factor be seen, below the stoichiometric WFR there is coke formation, which
(fH2), the more sub-separators will need to operate. To simulate the probably occurs from Eqs. 8, 9 and 10. As the WFR increases, coke
variations of temperature, water to feed molar ratio and sorbent to feed production is inhibited, meaning that the reverse of Eq. 10 is promoted;
molar ratio, the Sensitivity Analysis Tool from Aspen was employed, simultaneously, Eqs. 8 and 9 are inhibited (so CH4 and CO yields in-
where the analysed variable and its range of variation is introduced. crease) because the WGS reaction is favoured (producing CO2 and H2)
To simulate the sorption-enhanced reactor for CO2 capture, three with the increases of WFR. As a consequence, there is more CO and H2
more components need to be defined, namely the sorbent (CaO), and and, for not excessive WFR, the methanation reaction (reverse of me-
the products resulting from the reactions: calcium hydroxide (Ca(OH)2) thane steam reforming) is favoured, showing an increase of CH4 yield.
and calcium carbonate (CaCO3) – cf. Table 2). Calcium oxide is there- The H2 yield increases with the WFR as it shifts the equilibrium of
fore added as an inlet stream to the RGIBBS unit (represented as R1 in both WGS and methane steam reforming reaction towards the right side
Fig. 2). The CaO-to-feed molar ratio (SFR) can be varied in order to find (Eqs. 2 and 6, respectively, in Table 1). Accordingly, the excess of
the optimal value in terms of hydrogen yield and total CO2 capture. water, beyond the stoichiometric value, also causes a slight decrease in
methane production since the methanation reaction (Eq. 5) is inhibited.
3. Results and discussion Carbon monoxide yield has a similar behaviour to that of CH4 since it
decreases slightly when reaching the WFR stoichiometric value, al-
3.1. Comparison of different bio-oils in a conventional steam reforming though only very few CO is formed when compared to the other pro-
reactor ducts, very probably due to the fact that it is consumed by the WGS.
Regarding coke production, it is inhibited with excess water as can be
As it was previously stated, bio-oil is originated from the pyrolysis of elucidated by the carbon formation reactions where water is present
biomass, which can come from a large diversity of sources. Thus, de- (Eq. 10); thus, there is no coke above the WFR stoichiometric value,
pending on the biomass origin, the bio-oil will have different oxygenated which varies from one bio-oil to another (see also Table 3).
compounds in its composition (see Supporting information; Tables SI.1 to Fig. 3 also shows that hydrogen yields are still far from the theo-
SI.11). In this section, several bio-oils were used as feedstock, for the retical (maximum) yields predicted, as reported in Table 3. In order to
steam reforming process, in order to compare the yield of the products, better compare the performance of the different bio-oils, the variation
varying either the water to feed molar ratio (WFR) or the temperature. of the relative yield (i.e., H2 yield/maximum H2 yield) with the nor-
Considering only the overall steam reforming equation (3) and the ele- malized WFR (related to the stoichiometric WFR) was plotted in Fig. 4.
mental composition (C, H and O) of each bio-oil, it is possible to estimate It is possible to observe that the higher the WFR/WFRsq ratio, the higher
the stoichiometric WFR required and the maximum H2 yield that is the relative hydrogen yield difference between the different bio-oils
possible to be produced (assuming that no parallel reactions occur) for used. Bio-oil obtained from rice produces the highest relative hydrogen
each bio-oil considered (see values in Table 3). yield while oak (and maple) or eucalyptus the lowest ones, depending

187
M.A. Soria et al. Fuel 244 (2019) 184–195

Table 3 Table 3). All the bio-oils exhibit a similar behaviour in terms of H2, CO,
Stoichiometric water to feed molar ratio (WFR) and maximum H2 yield for the CO2, CH4 and coke yield with the rise of the temperature (Fig. 5).
different bio-oils. Obtained hydrogen yields for each bio-oil are in line with the sequence
Original Biomass Stoichiometric Maximum H2 Yield for the maximum expected yields reported in Table 3. As stated above,
WFR* and whatever the original biomass, total conversion was reached in all
the simulations, meaning that steam reforming of the bio-oils is com-
Spruce wood 3 4.75
plete. So, the effect of temperature is related with other reactions. In-
Sugar Bagasse 3 4.66
Rice husk 7 8.36 creasing temperature promotes the formation of more hydrogen mainly
Maize stalk 7 9.53 due to the progressive inhibition of the methanation reaction (3 mol of
Pine sawdust 6 8.73 H2 for each mole of CH4), which is highly exothermic (Eq. (5) – reverse
Mesquite sawdust 7 9.92
of methane steam reforming, Eq. 6); this happens until a maximum
Wheat shell 9 11.94
Oil palm shells 8 10.35
around 700 °C and then the hydrogen yield decreases. This maximum
Oak 2 3.71 occurs because the WGS is slightly exothermic and for temperatures
Maple 2 3.76 above 700 °C, reverse water-gas shift (RWGS) is more favoured. Carbon
Eucalyptus wood 5 7.94 dioxide, which is also involved in RWGS and coke formation (Eq. 8; for
T less than 550 °C), has a similar behaviour to hydrogen but its max-
* Rounded up values.
imum happens at lower temperatures (close to 550 °C). This difference
of maximum temperatures happens because while H2 is being con-
on the amount of steam fed. While for low values of the WFR/WFRsq
sumed through RWGS and saved through the inhibition of methanation
ratio all bio-oils exhibit a very similar behaviour, it is possible to ob-
(the maximum yield of H2 matches with the almost complete inhibition
serve that even when the WFR is twice the WFRsq, H2 yield is still far
of methane formation), CO2 is no longer produced from Eq. 8 at high
from the maximum yield possible. This means that there is a possibility
temperatures and is only being consumed through RWGS. Concerning
for improvement, either changing the reaction conditions or the reactor
carbon monoxide, until 500 °C its yield is very close to zero because it is
type, as proposed in the following sections.
consumed by WGS, methanation and coke formation (Eqs. 8 and 10) –
all exothermic reactions. However, it increases from around 500 °C,
since higher temperatures favour either the RWGS reactions or the in-
3.1.2. Effect of temperature
The temperature effect on the products yield was studied for the hibition of methanation and coke formation. As it was already stated,
for high temperatures the methanation reaction is inhibited, so CH4 is
different bio-oils at the stoichiometric WFR of each of them (reported in

Fig. 3. Effect of water to feed molar ratio (WFR) on the yield of the products for several bio-oils in a conventional steam reforming reactor at 400 °C and 1 bar.

188
M.A. Soria et al. Fuel 244 (2019) 184–195

Fig. 4. Normalized H2 yield as a function of the normalized water to feed molar ratio for several bio-oils in a conventional reactor at 400 °C and 1 bar.

Fig. 5. Effect of temperature on the yield of the products for several bio-oils in a conventional steam reforming reactor at stoichiometric WFR conditions and 1 bar.

no longer produced for temperatures above 700 °C. Besides that, for the reverse of methanation and methane dry reforming is a linear
high temperatures coke formation also does not occur as most carbon combination of the methanation and RWGS as indicated in the foot-
formation reactions are exothermic. notes “a” and “b” in Table 1, respectively.
In the discussions above (and below), the species yields are ex- The ratio between the H2 yield and the maximum theoretical yield
plained considering mostly methanation and WGS (or RWGS) reactions; as a function of the temperature is represented in Fig. 6. It can be seen
actually, the reader should consider that methane steam reforming is that the different bio-oils behave in a similar way, being the highest

189
M.A. Soria et al. Fuel 244 (2019) 184–195

relative H2 yield of about 0.7 reached at 700 °C. This means that the H2 range of temperature the H2 yield remained constant, in spite of the SFR
yield can be further improved, namely using other types of reactors, increases, because the sorption process needs H2O to occur (Eq. (16))
which are able to shift the equilibrium of the involved reactions, such as and for the stoichiometric WFR of bio-oil steam reforming (which are
a membrane reactor, a sorption-enhanced reactor or a sorption-en- the conditions employed in Fig. 7) there is not enough water to make
hanced membrane reactor. Since the bio-oils have similar behaviours, the sorption reaction fully happen, leaving unreacted CaO in the reac-
the thermodynamic analysis for these multifunctional reactors was tion zone.
performed only for two of them. The bio-oils chosen were the one from As reported in the Supporting Information section (Figures SI.1 and
wheat shell, as it has the highest H2 production potential (see Table 3), SI.2), with the increase of the SFR the CO2 yield decreases, in the
and the one from spruce wood, as it is one with the lowest H2 pro- temperature range in which the sorption reaction operates
duction potential and is an abundant natural resource in Portugal. The (450–700 °C), reaching zero at SFR = 2 for spruce and SFR = 5 for
other bio-oils seem to have a performance in between these two ex- wheat (i.e. complete CO2 sorption). The removal of CO2 promotes the
tremes. WGS (Eq. 2) consuming CO and, consequently, in the range above
The results obtained herein by simulation were compared with ex- mentioned, the CH4 yield is much lower than that obtained in the
perimental results from the literature [16,17]. Experiments of steam conventional reactor as methanation (reverse of methane steam re-
reforming of raw bio-oil, whose composition is very similar to that used forming) is inhibited.
in this work for pine sawdust, were performed in a conventional reactor The optimum values of hydrogen yield obtained varying the tem-
with Ni-based catalysts at atmospheric pressure, temperatures between peratures for SFR = 2 were 3.87 mol H2/mol of bio-oil (81% of the
550 and 750 °C, and different WFRs. Simulated results performed at theoretical maximum – Table 3) in the case of spruce, and for SFR = 5
stoichiometric WFR and similar P and T show full conversion and the maximum yield was 10.17 mol H2/mol of bio-oil (85% of the
normalized H2 yield between 0.6 and 0.75, which are very close to the maximum theoretical) in the case of wheat, both at 450 °C. Analysing
experimental values (∼0.6 for 600 °C and ∼0.75 for 750 °C) obtained the whole range of temperatures, it is worth mentioning that the re-
in a thermodynamic regime i.e., for high values of space-time. It was moval of CO2 shifts the WGS to produce more H2, thus consuming CO;
also observed that the increase in both temperature and WFR lead to an therefore, between 450 and 600 °C (where carbon monoxide was pre-
attenuation of coke deposition. Therefore, the predictions obtained at sent in the TR), this product is nearly inexistent in the SER, as can be
similar conditions in a TR show a very good consistency regarding the seen in Figures SI.1 and SI.2. With the temperature increase, CO2
experimental observations. sorption is inhibited because the sorption reaction is exothermic and
RWGS becomes more favoured, so CO formation increases. Carbon di-
oxide yield increases with the increase of temperature until it reaches a
3.2. Sorption-enhanced reactor (SER)
maximum, then decreases due to the inhibition of the sorption reaction
and the promotion of RWGS; when there is no CaO (conventional re-
The objective of the SER is to remove CO2 from the reaction zone
actor – SFR = 0) this maximum is reached at a lower temperature since
using CaO as a CO2 sorbent (Eqs. (16) and (17)), which allows to shift
only RWGS happens and not the sorption reactions. Due to the reduc-
the equilibrium of the steam reforming reaction, enhancing the pro-
tion of CO and the removal of CO2, coke formation in the SER only
duction of H2. For this type of reactor, a parametric analysis varying the
occurs at very low temperatures with low SFR, as its formation depends
temperature, the water-to-feed molar ratio (WFR) and the sorbent-to-
on these two compounds (Eqs. 8–11).
feed molar ratio (SFR) was addressed.
As in the conventional reformer, for the sorption-enhanced reactor
The effect of the temperature and SFR on the H2 yield for bio-oil
the H2 yield increases with the WFR (Fig. 8a and b) as it favours WGS
derived from the pyrolysis of spruce and wheat is shown in Fig. 7a and
and also the sorption reaction(s), while inhibiting the methanation (see
b, respectively. For a given temperature (in the range of 450 °C to
reactions in Tables 1 and 2). Consequently, it allows a higher CO2 re-
700 °C), the H2 yield rises with SFR until reaching a maximum and then
moval, so that for high SFR values CO2 is nearly inexistent (see Figures
remains nearly constant; this happens from SFR = 2 for spruce and
SI.3 and SI.4 in the Supporting Information). Moreover, for high SFRs
from SFR = 5 for wheat (see the projections in the bottom plane of
there is no carbon monoxide because the CO2 removal by the use of the
Fig. 7). The fact that the maximum H2 yield is reached between 450 and
sorbent promotes the WGS reaction to completion. As stated above, the
700 °C shows that this is the range of temperature where the sorption
increase in the water to feed molar ratio also promotes the WGS
reactions are more favoured (when coupled with the steam reforming
(consuming CO), therefore, this indirectly contributes to the inhibition
process) as the sorption reactions are exothermic. However, in such

Fig. 6. Normalized H2 yield as a function of temperature for several bio-oils in a conventional reactor at stoichiometric WFR and 1 bar.

190
M.A. Soria et al. Fuel 244 (2019) 184–195

Fig. 7. Hydrogen yield as a function of temperature and sorbent-to-feed ratio for a) spruce bio-oil and b) wheat bio-oil in the sorption-enhanced reactor at stoi-
chiometric water-to-feed ratio and 1 bar.

of methanation (reverse of methane steam reforming). Thus, for high hydrogen (Eq. 5), and it shifts the WGS reaction to the hydrogen pro-
values of SFR and WFR methane is not present (Figures SI.3 and SI.4). duction side – Eq. 2 (see Fig. 9a and b, and specifically the projections
In addition, both the increase of SFR and WFR inhibit the formation of in the bottom plane). This increase also happens with the increase of
coke, which is only present for low values of both variables. The hy- temperature (from 300 to 500 °C), mostly due to the progressive in-
drogen yield will keep increasing with the increase of WFR until it al- hibition of the exothermic methanation reaction, (Eq. (5)). It is note-
most reaches the theoretical maximum. The best H2 yield values ob- worthy that it is possible to obtain a high hydrogen yield when using
tained were 4.71 mol H2/mol of bio-oil for spruce at SFR = 3 and the MR compared to a conventional reactor. Actually, the maximum
WFR = 15 and 11.32 mol H2/mol of bio-oil for wheat at SFR = 5 and value occurred for fH2 = 0.8 at 500 °C and stoichiometric WFR and was
WFR = 20. Although these values correspond to 99% (for spruce) and 3.44 mol of H2/mol of bio-oil for spruce and 9.31 mol of H2/mol of bio-
95% (for wheat) of the theoretical maxima reported in Table 3, WFR oil for wheat; in the same conditions, yields in the conventional re-
values well above the stoichiometric are required, implying high op- former were 1.52 and 3.70, respectively.
erating costs that can be reduced with other reactor configurations. As stated above, with the removal of hydrogen the methanation
reaction is inhibited. Thus, the yield of CO should increase when fH2
increases, however, it remains nearly constant because CO is consumed
3.3. Membrane reactor (MR) mainly by WGS (Figures SI.5 and SI.6). With the increase of tempera-
ture CO formation increases due to the RWGS and inhibition of coke
The employment of a hydrogen permselective membrane in the (reverse of Eqs. (2) and (8), both being endothermic). Likewise, because
multifunctional reactor is expected to improve the hydrogen yield. The of the inhibition of methanation, for high removal fractions of H2 the
study of the MR was made analysing the behaviour of the bio-oils while CH4 yield decreases (particularly at higher temperatures). CO2 shows
varying the temperature (Fig. 9a and b) and water-to-feed ratio the same behaviour of hydrogen because of the WGS reaction, although
(Fig. 10a and b) for different hydrogen removal fractions ( fH2 ). A re- H2 yield increases in a higher degree due to the methanation inhibition.
moval fraction superior to 0.8 was not simulated since it is not expected Inherently, the increase of the yield of CO2 and CO for higher H2 re-
to be experimentally feasible; moreover, the temperature range used moval fractions implies a higher coke yield (although its value is very
was between 300 °C and 500 °C due to the commonly reported metallic low), since they are involved in the coke formation reactions (Eqs. 8, 10
(e.g. Pd-based) membrane limitations. and 11). Because prevailing coke formation reactions are exothermic,
With the increase of the hydrogen removal fraction, for a given its formation is reduced for high temperatures.
temperature, the hydrogen yield increases because it inhibits the me- The increase of both the H2 removal fraction and water-to-feed
thanation (reverse of methane steam reforming), which consumes

Fig. 8. Hydrogen yield as a function of water-to-feed ratio and sorbent-to-feed ratio for a) spruce bio-oil and b) wheat bio-oil in the sorption-enhanced reactor at
400 °C and 1 bar.

191
M.A. Soria et al. Fuel 244 (2019) 184–195

Fig. 9. Hydrogen yield as a function of temperature and hydrogen removal fraction for a) spruce bio-oil and b) wheat bio-oil at stoichiometric WFR and 1 bar in the
membrane reactor.

Fig. 10. Hydrogen yield as a function of the water-to-feed molar ration and hydrogen removal fraction for a) spruce bio-oil and b) wheat bio-oil at 400 °C and 1 bar in
the membrane reactor.

molar ratio promote the improvement of the hydrogen yield for both WFR = 15 and 8.83 mol H2/mol of bio-oil for wheat (74% of the the-
bio-oils considered (Fig. 10a and b). It is worth noting that when WFR oretical maximum) at WFR = 20, all with a removal fraction of 0.8.
increases beyond the stoichiometric value (3 for spruce and 9 for Once again, very high values of WFR are required meaning high op-
wheat), the coke formation is completely avoided (Figures SI.7 and erating costs that can be reduced with other reactor configurations, as
SI.8). Therefore, increased WFR promotes the WGS (increasing H2 and detailed bellow.
CO2 yield) and inhibits the methanation reaction, particularly at high
fH2 (Fig. 10a and b). The increase of both WFR and fH2 inhibits me- 3.4. Sorption-enhanced membrane reactor (SEMR)
thanation; so, CH4 yield decreases with the increase of these variables.
In this reactor configuration, for the temperatures and WFR ranges This hybrid multifunctional reactor configuration combines both in-
tested, the best values obtained for H2 yield were 4.66 mol H2/mol of situ carbon dioxide and hydrogen removal, which could further im-
bio-oil for spruce (98% of the theoretical maximum – Table 3) at prove the hydrogen yield. In order to compare the SEMR with other

Fig. 11. Comparison of hydrogen yield as a function of temperature for a) spruce bio-oil in TR, SER (SFR = 2), MR (fH2 = 0.8) and SEMR (SFR = 2, fH2 = 0.8) and b)
for wheat bio-oil in TR, SER (SFR = 5), MR (fH2 = 0.8) and SEMR (SFR = 5, fH2 = 0.8) both at stoichiometric WFR and 1 bar.

192
M.A. Soria et al. Fuel 244 (2019) 184–195

reactor types, a combination of the optimal conditions (that provided inherent grow of cost). For this reason, for the SEMR the effect of the
the highest H2 yield) of each reactor previously studied was selected, pressure was assessed for the conditions in which these reactors showed
which were as follows: for the SER, a SFR of 2 and 5 for bio-oil derived the highest hydrogen yields. The results obtained are presented in
from spruce and wheat respectively; for the MR, a hydrogen removal Fig. 15a and b for bio-oil derived from pyrolysis of spruce and wheat,
fraction of 0.8. This comparison was addressed at different tempera- respectively. Since the bio-oil conversion is complete and there is no
tures and water to feed molar ratios. coke present, the variation of the yields of the products with the total
Regarding the effect of temperature, the simulations were run for pressure is only ascribed to the methanation (the WGS reaction is not
the respective stoichiometric WFR of each stream. It is observed from affected by pressure changes). At low WFR, whatever the bio-oil used,
Fig. 11 that for the SEMR it is possible to achieve the highest hydrogen the pressure meaningfully affects the H2 yield. Nevertheless, the ne-
yield (although below the theoretical maximum) at 360 °C for spruce gative effect of pressure can be offset by the increase of WFR as it in-
and at 420 °C for wheat, while for the other type of reactors the max- hibits methanation, which leads to a more stable H2 yield as a function
imum H2 yield (always below that for the SEMR) is achieved at higher of pressure. Therefore, at WFR of 6 for spruce and 10 for wheat, the H2
temperatures. It is also noteworthy that whatever the temperature, the yield decreases 12% and 26% respectively when the total pressure is
removal of CO2 (given by the used SFR) has a more significant effect on increased from 1 bar to 8 bar. However, such decreases is only of 4%
the hydrogen yield improvement than the removal of H2 (fH2 = 0.8). and 6% at WFR of 8 and 14 for spruce and wheat, respectively.
Also, when capture of carbon dioxide occurs, there is a very significant
reduction of the other by-products like CO, CO2 and coke. Methanation
is also almost fully inhibited in the SEMR (Figures SI.9 and SI.10). 4. Conclusions
To further compare the different type of reactors studied and to
demonstrate how the performance is improved with the use of the An equilibrium thermodynamic analysis of the steam reforming
hybrid reactors, the hydrogen yield was plotted at SEMR optimal process was carried out for several pyrolysis bio-oils in order to access
temperature conditions (Fig. 12a and b). It is observed that all the the effects of different important operating parameters on the hydrogen
hybrid reactors improve the performance as compared to the conven- (and by-products) yield. This analysis was performed in different re-
tional reactor and the best configuration is clearly the SEMR. As com- actor types and the effect of the main operating conditions was assessed
pared to the TR, impressive improvements of over 1000% and 330% as well.
can be reached for spruce and wheat bio-oil, respectively. In the case of the traditional reformer, temperature and water to
Concerning the variation of water to feed molar ratio (Fig. 13), the feed ratio (WFR) showed a positive effect on the production of hy-
study was carried out at 400 °C. The increase of the WFR increases the drogen. However, the H2 yields observed by such traditional reactor
hydrogen yield in all reactor configurations; the maximum yield was (TR) are still far from the maximum theoretical yields as considerable
achieved at a WFR higher than 6 for spruce bio-oil (stoichiometric by-products are formed. Compared to the TR, the in-situ CO2 removal
WFR = 3) and 10 for wheat bio-oil (stoichiometric WFR = 9) for the using calcium oxide as sorbent in a sorption-enhanced reactor (SER)
SEMR, which is the only reactor able to achieve the maximum (theo- and the in-situ removal of H2 through a selective permeable membrane
retical) H2 yield possible for spruce bio-oil (Fig. 13a), although this is (membrane reactor - MR) allowed to increase the hydrogen production
not verified for wheat bio-oil (Fig. 13b) as methanation is still occur- while also inhibiting the formation of carbon monoxide, methane and
ring. The increase of steam promotes the WGS, whereas methanation is coke. Both the MR and the SER allow to reach high hydrogen yields at
inhibited, and so the only by-product whose formation increases is lower temperatures than those employed in the conventional steam
carbon dioxide, which will be removed in the reactor configurations reforming reactor.
with CO2 sorption, while carbon monoxide, methane and coke are re- On the other hand, the sorption-enhanced membrane reactor
duced (Figures SI.11 and SI.12). (SEMR) allows to attain the highest hydrogen yield, namely: 4.72 for
Fig. 14a and b represent the comparison between the different types spruce (at 400° C, WFR = 6, SFR = 2, fH2 = 0.8 and 1 bar) and 11.62
of reactor at optimum WFR for the SEMR. In that figures are also de- for wheat (at 400° C, WFR = 10, SFR = 5, fH2 = 0.8 and 1 bar) derived
noted the improvement percentage of the SEMR regarding to the other bio-oils. Therefore, this multifunctional reactor is the one that provides
types of reactor. It is again observed the same performance trends than the best hydrogen yield (even reaching, in certain cases, the maximum
those of the comparison at optimal temperature conditions, namely: theoretical value at not very severe operating conditions), from a
SEMR > SER > MR > TR. thermodynamic standpoint. Since pressures higher than atmospheric
The total pressure favours the CO2 sorption capacity of the sorbent are needed to obtain high hydrogen removal fractions in a membrane
and the driving force for H2 permeation through a H2 perm-selective reactor, and increase the CO2 breakthrough time in a SER, the effect of
membrane [34]. Therefore, when working with membranes it is not pressure on the H2 yield was addressed for the SEMR. Results showed
realistic to expect to achieve hydrogen removal fractions as high as 0.8 that hydrogen yield decreased with pressure, however, this can be
at 1 bar; otherwise, a huge membrane area would be required (with the offset by increasing the WFR.

Fig. 12. Comparison of hydrogen yield as a function of the reactor type for a) spruce bio-oil in TR, SER (SFR = 2), MR (fH2 = 0.8) and SEMR (SFR = 2, fH2 = 0.8) at
360 °C and b) for wheat bio-oil in TR, SER (SFR = 5), MR (fH2 = 0.8) and SEMR (SFR = 5, fH2 = 0.8) at 420 °C, in both cases at stoichiometric WFR and 1 bar.

193
M.A. Soria et al. Fuel 244 (2019) 184–195

Fig. 13. Comparison of hydrogen yield as a function of the water to feed molar ratio for a) spruce bio-oil in TR, SER (SFR = 2), MR(fH2 = 0.8) and SEMR (SFR = 2,
fH2 = 0.8) and b) for wheat bio-oil in TR, SER (SFR = 5), MR(fH2 = 0.8) and SEMR (SFR = 5, fH2 = 0.8) both at 400 °C and 1 bar.

Fig. 14. Comparison of hydrogen yield as a function of the reactor type for a) spruce bio-oil in TR, SER (SFR = 2), MR (fH2 = 0.8) and SEMR (SFR = 2, fH2 = 0.8) at
WFR = 6 and b) for wheat bio-oil in TR, SER (SFR = 5), MR (fH2 = 0.8) and SEMR (SFR = 5, fH2 = 0.8) at WFR = 10, in both cases at 400 °C, and 1 bar.

Fig. 15. Hydrogen yield as a function of pressure and water-to-feed ratio (WFR) in the SEMR for a) spruce bio-oil at 400 °C, fH2 = 0.8 and SFR = 2 and b) for wheat
bio-oil at 400 °C, fH2 = 0.8 and SFR = 5.

Acknowledgments Appendix A. Supplementary data

The authors are grateful to LEPABE – Laboratory for Process Supplementary data to this article can be found online at https://
Engineering, Environment, Biotechnology and Energy, and the project doi.org/10.1016/j.fuel.2019.01.156.
NORTE‐01‐0145‐FEDER‐000005 – LEPABE-2-ECO-INNOVATION, sup-
ported by North Portugal Regional Operational Programme (NORTE References
2020), under the Portugal 2020 Partnership Agreement, through the
European Regional Development Fund (ERDF), for his research grant. [1] Bridgwater AV. Review of fast pyrolysis of biomass and product upgrading. Biomass
M.A. Soria is grateful to the FCT for the postdoctoral grant (SFRH/BPD/ Bioenergy 2012;38:68–94.
[2] Faba L, Díaz E, Ordóñez S. Recent developments on the catalytic technologies for
88444/2012), with financing from the European Social Fund (ESF) and the transformation of biomass into biofuels: a patent survey. Renew Sustain Energy
the Human Potential Operational Programme (POPH). Rev 2015;51:273–87.

194
M.A. Soria et al. Fuel 244 (2019) 184–195

[3] Akhtar J, Amin NAS. A review on process conditions for optimum bio-oil yield in [20] Goicoechea S, Ehrich H, Arias PL, Kockmann N. Thermodynamic analysis of acetic
hydrothermal liquefaction of biomass. Renew Sustain Energy Rev acid steam reforming for hydrogen production. J Power Sources 2015;279:312–22.
2011;15:1615–24. [21] Leal AL, Soria MA, Madeira LM. Autothermal reforming of impure glycerol for H2
[4] Sadhukhan J, Martinez-Hernandez E, Murphy RJ, Ng DKS, Hassim MH, Siew Ng K, production: thermodynamic study including in situ CO2 and/or H2 separation. Int J
et al. Role of bioenergy, biorefinery and bioeconomy in sustainable development: Hydrogen Energy 2016;41:2607–20.
Strategic pathways for Malaysia. Renew Sustain Energy Rev 2018;81:1966–87. [22] Alírio E. Rodrigues, Luís M. Madeira, Y.-J. Wu, R.P.V. Faria, Sorption Enhanced
[5] Bridgwater T. Biomass for energy. J Sci Food Agric 2006;86:1755–68. Reaction Processes, World Scientific. Sustainable Chemistry Series vol. 1, London.
[6] Mondal P, Dang GS, Garg MO. Syngas production through gasification and cleanup 2017.
for downstream applications — recent developments. Fuel Process Technol [23] Soria MA, Tosti S, Mendes A, Madeira LM. Enhancing the low temperature wa-
2011;92:1395–410. ter–gas shift reaction through a hybrid sorption-enhanced membrane reactor for
[7] Oasmaa A, Meier D. Norms and standards for fast pyrolysis liquids. J Anal Appl high-purity hydrogen production. Fuel 2015;159:854–63.
Pyrol 2005;73:323–34. [24] Basile A, Iulianelli A, Longo T, Liguori S, De Falco M. Chapter 2 - Pd-based Selective
[8] Dhyani V, Bhaskar T. A comprehensive review on the pyrolysis of lignocellulosic Membrane: State-of-the-Art. In: Springer (Ed.) Membrane Reactors for Hydrogen
biomass. Renewable Energy 2018;129:695–716. Production Processes. 2011.
[9] Raffelt K, Henrich E, Koegel A, Stahl R, Steinhardt J, Weirich F. The BTL2 process of [25] Silva JM, Soria MA, Madeira LM. Challenges and strategies for optimization of
biomass utilization entrained-flow gasification of pyrolyzed biomass slurries. Appl glycerol steam reforming process. Renew Sustain Energy Rev 2015;42:1187–213.
Biochem Biotechnol 2006;129:153–64. [26] Okazaki J, Ikeda T, Tanaka DAP, Sato K, Suzuki TM, Mizukami F. An investigation
[10] Trane R, Dahl S, Skjøth-Rasmussen MS, Jensen AD. Catalytic steam reforming of of thermal stability of thin palladium–silver alloy membranes for high temperature
bio-oil. Int J Hydrogen Energy 2012;37:6447–72. hydrogen separation. J Membr Sci 2011;366:212–9.
[11] Czernik S, Bridgwater AV. Overview of Applications of Biomass Fast Pyrolysis Oil. [27] Gil MV, Fermoso J, Pevida C, Chen D, Rubiera F. Production of fuel-cell grade H2 by
Energy Fuels 2004;18:590–8. sorption enhanced steam reforming of acetic acid as a model compound of biomass-
[12] Adhikari S, Fernando SD, Haryanto A. Hydrogen production from glycerol: an up- derived bio-oil. Appl Catal B 2016;184:64–76.
date. Energy Convers Manage 2009;50:2600–4. [28] Sahebdelfar S. Steam reforming of propionic acid: Thermodynamic analysis of a
[13] Remiro A, Valle B, Aguayo AT, Bilbao J, Gayubo AG. Operating conditions for at- model compound for hydrogen production from bio-oil. Int J Hydrogen Energy
tenuating Ni/La2O3–αAl2O3 catalyst deactivation in the steam reforming of bio-oil 2017;42:16386–95.
aqueous fraction. Fuel Process Technol 2013;115:222–32. [29] Yao X, Yu Q, Xie H, Duan W, Han Z, Liu S, et al. The production of hydrogen
[14] Kechagiopoulos Panagiotis N, Voutetakis Spyros S, Lemonidou Angeliki A, Vasalos through steam reforming of bio-oil model compounds recovering waste heat from
IA. Hydrogen Production via Steam Reforming of the Aqueous Phase of Bio-Oil in a blast furnace slag. J Therm Anal Calorim 2018;131:2951–62.
Fixed Bed Reactor. Energy Fuels 2006;20:2155–63. [30] Chen H, Zhang T, Dou B, Dupont V, Williams P, Ghadiri M, et al. Thermodynamic
[15] Xie H, Yu Q, Wei M, Duan W, Yao X, Qin Q, et al. Hydrogen production from steam analyses of adsorption-enhanced steam reforming of glycerol for hydrogen pro-
reforming of simulated bio-oil over Ce–Ni/Co catalyst with in continuous CO2 duction. Int J Hydrogen Energy 2009;34:7208–22.
capture. Int J Hydrogen Energy 2015;40:1420–8. [31] Rocha C, Soria MA, Madeira LM. Steam reforming of olive oil mill wastewater with
[16] Valle B, Aramburu B, Olazar M, Bilbao J, Gayubo AG. Steam reforming of raw bio- in situ hydrogen and carbon dioxide separation – Thermodynamic analysis. Fuel
oil over Ni/La2O3-αAl2O3: influence of temperature on product yields and catalyst 2017;207:449–60.
deactivation. Fuel 2018;216:463–74. [32] Soria MA, Mateos-Pedrero C, Guerrero-Ruiz A, Rodríguez-Ramos I. Thermodynamic
[17] Bizkarra K, Bermudez JM, Arcelus-Arrillaga P, Barrio VL, Cambra JF, Millan M. and experimental study of combined dry and steam reforming of methane on Ru/
Nickel based monometallic and bimetallic catalysts for synthetic and real bio-oil ZrO2-La2O3 catalyst at low temperature. Int J Hydrogen Energy
steam reforming. Int J Hydrogen Energy 2018;43:11706–18. 2011;36:15212–20.
[18] Remiro A, Arandia A, Oar-Arteta L, Bilbao J, Gayubo AG. Regeneration of NiAl2O4 [33] Silva JM, Soria MA, Madeira LM. Thermodynamic analysis of Glycerol Steam
spinel type catalysts used in the reforming of raw bio-oil. Appl Catal B Reforming for hydrogen production with in situ hydrogen and carbon dioxide se-
2018;237:353–65. paration. J Power Sources 2015;273:423–30.
[19] Wang D, Czernik S, Montane D, Mann M, Chornet E. Biomass to Hydrogen via Fast [34] Iulianelli A, Longo T, Basile A. CO-free hydrogen production by steam reforming of
Pyrolysis and Catalytic Steam Reforming of the Pyrolysis Oil or Its Fractions. Ind acetic acid carried out in a Pd–Ag membrane reactor: the effect of co-current and
Eng Chem Res 1997;36:1507–18. counter-current mode. Int J Hydrogen Energy 2008;33:4091–6.

195

Вам также может понравиться