Sei sulla pagina 1di 10

Renewable Energy 36 (2011) 779e788

Contents lists available at ScienceDirect

Renewable Energy
journal homepage: www.elsevier.com/locate/renene

A comparative study on hydrogen production from steam-glycerol reforming:


thermodynamics and experimental
Haisheng Chen a, b, *, Yulong Ding a, Ngoc T. Cong a, Binlin Dou a, Valerie Dupont a, Mojtaba Ghadiri a,
Paul T. Williams a
a
b

School of Process, Environmental and Material Engineering, University of Leeds, Leeds LS2 9JT, UK
Institute of Engineering Thermophysics, Chinese Academy of Science, Beijing 100190, China

a r t i c l e i n f o

a b s t r a c t

Article history:
Received 2 June 2010
Accepted 29 July 2010
Available online 23 August 2010

A detailed comparative study on thermodynamic and experimental analyses of glycerol reforming for
hydrogen production has been conducted in terms of the effects of temperature, pressure, water to
glycerol feed ratio, feeding reactants to inert gas ratio and feeding gas ow rate (residence time).
The thermodynamic analysis was conducted by using a non-stoichiometric methodology based on the
minimisation of Gibbs free energy. And the experiments were carried out with a pilot scale set-up.
The results show that the thermodynamic and experimental data agree fairly well with each other. The
measured hydrogen production is slightly lower than that predicted by the thermodynamic analysis,
which is mainly because the conversion of steam is incomplete. High temperature, low pressure, low
feeding reactants to inert gas ratio and low gas ow rate are favourable for steam reforming of glycerol
for hydrogen production. There is an optimal water to glycerol feed ratio for steam reforming of glycerol
for hydrogen production which is about 9.0. The glycerol conversion is a strong function of water to
glycerol ratio, whereas a weak function of other parameters over the conditions of this work. A novel
adsorption enhanced reaction process incorporating water and heat recovery is proposed for further
optimisation of hydrogen production from steam reforming of glycerol.
2010 Elsevier Ltd. All rights reserved.

Keywords:
Hydrogen production
Steam reforming of glycerol
Thermodynamic analysis
Experimental
Comparative study

1. Introduction
Due to the environmental concerns, the global demand for
hydrogen is expected to greatly increase in the future. One of the
solutions to this issue is to use biomass based raw materials to
produce hydrogen [1e3]. Currently, about 10 wt% of glycerol is
produced during the conversion of vegetable oils or animal fats into
biodiesel fuel through the catalytic transesterication process
[4e6]. Although a small amount of glycerol from biodiesel
production is puried for pharmaceutical and food applications, byproduct glycerol is now in surplus and the majority is taken as
a waste. This work is concerned about the use of glycerol, a major
by-product of biodiesel production, to produce hydrogen, a clean
energy carrier by steam reforming process.
A number of approaches, including aqueous phase reforming
[7], bioconversion using genetically engineered enzymes [8], auto-

* Corresponding author. Institute of Engineering Thermo physics, Chinese


Academy of Science, Beijing 100190, China. Tel.: 86 10 8254 3148; fax: 86 10
8254 3109.
E-mail address: chen_hs@mail.etp.ac.cn (H. Chen).
0960-1481/$ e see front matter 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.renene.2010.07.026

thermal reforming over Rh catalysts [9], thermal decomposition


[10,11] and catalytic steam reforming over Yttria/Zirconia supported Ru catalysts [4], have been proposed to convert glycerol to
hydrogen over the past few years. Among them, steam reforming of
glycerol has attracted much more attention [2,4e6,12e20] as it can
produce up to 7 mol of hydrogen per mole feed glycerol theoretically via the overall reaction:

C3 H8 O3 3H2 O 3CO2 7H2 ; DH298 128 kJ=mol

(1)

The studies conducted on steam-glycerol reforming for hydrogen


production can be classied into three categories: experiments on
the catalyst [4,7,13,17,20,21], experiments on process and parameter
optimisation [5,6,14,16,17,20,21] and numerical [16] and theoretical
analyses [12,15,18,19,22]. However, all of experimental studies were
conducted on a very small scale with a reactor diameter of mm,
catalyst mass of grams and gas ow rate of mL/min. A pilot scale
experimental study, with a reactor diameter of cm, catalyst mass of
kg and gas ow rate of l/min, will be conducted, for the rst time, in
this work. It could pave a way to the industrial application of steam
reforming of glycerol for hydrogen production. On the other hand,
most of the theoretical and experimental studies were conducted
separately [2,4e6,13e22]. There is clearly a mismatch between

780

H. Chen et al. / Renewable Energy 36 (2011) 779e788

the conditions in theoretical and experimental investigations.


Furthermore, in the few published comparative studies [12,23] on
theoretical and experimental analyses of steam-glycerol reforming
for hydrogen production, only few inuence factors were considered. This forms another motivation of this work to conduct
a detailed comparative study on thermodynamic and experimental
analyses of steam reforming of glycerol in terms of temperature,
pressure, water to glycerol feed ratio (WGFR), feeding reactants to
inert gas ratio (FI) and feeding gas ow rate (residence time). In fact,
there is also a lack of experimental data on the effect of parameters
except temperature in the whole community of steam reforming of
glycerol. This work can act as a useful reference for parametric
optimisation.
2. Theoretical approach
A non-stoichiometric thermodynamic analysis on the chemical
equilibrium is conducted based on the minimisation of Gibbs free
energy (G). It is known that the total Gibbs free energy of a chemical
system reaches a minimum at the equilibrium state and the Gibbs
energy of a chemical system is a function of temperature (T),
pressure (P) and species molar numbers (n). If the temperature and
the pressure of the system are constants, the equilibrium in the
system can be expressed as:

dG

N
X

mi dni 0

(2)

i1

where ni is the molar concentration of component i in the system


and mi is the chemical potential of component i. The question now is
to obtain G as a function of ni and to nd a set of ni that minimise
the value of G. In order to do this, the following constraint of the
elemental balances is introduced:
N
X

aji ni bj ; j 1; .; M

(3)

i1

where aji is the number of j type atoms in the i species and bj refers
to number of j type atoms in the feed. By introduction of the
Lagrangian multipliers, lj, a new function G is dened as:

N
X
i1

mi ni

M
X
j1

lj

N
X

!
aji ni  bj

(4)

i1

In gases chemical systems, the equation (4) can be further


expressed as:

b
DGfi =RT lnni =nT lnP ln f
i

M
X

lj aji =RT 0; i 1; .; N

j1

(5)
b is the fugacity coefcient in the gas mixture which can be
where f
obtained by the RedlicheKwong equation.DGfi is the standard Gibbs
free energy of formation which is only a function of temperature
and can be obtained from the handbooks of thermodynamic data
[24,25]. R is the universal gas constant. And nT is the total molar
number of species dened as:

nT

N
X

ni ; i 1; .; N

(6)

j1

Equations (3), (5) and (6) represent a (N M 1) non-linear


algebraic equation system that can be solved for the unknowns yi
and li hence ni values at equilibrium.

In the non-stoichiometric formulation, the species that coexist in


the equilibrium must be previously dened. The number of
compounds in the steam-glycerol system, resulting from the atomic
combination among C, H and O, could be very high. However, in
a homologous series the Gibbs free energy of formation increases
with the number of carbon atoms. Compounds with more than three
Carbon atoms are not likely to exist in the steam-glycerol system [26].
At the initial stage of this work, calculations with the typical methyl
(Carbon oxide: CO, Carbon dioxide: CO2, Methane: CH4, Methanal:
CH2O, Methanol: CH3OH), ethyl (Ethylene: C2H4, Ethane: C2H6,
Ethanal: CH3COH, Ethanol: C2H5OH) and propyl (Propane: C3H8,
Propene: C3H6, Propanal: CH3CH2CHO, Propanone: CH3COCH3)
compounds, plus glycerol (C3H5(OH)3), steam (H2O) and hydrogen
(H2), were rstly considered. It was found that the concentrations for
the compounds Methanal, Methanol, Ethylene, Ethane, Ethanal,
Ethanol, Propane, Propene, Propanal and Propanone (plus glycerol)
are all negligible (molar fractions are all less than 106). Carbon was
also considered in advance using a combined stoichiometric and nonstoichiometric method [15,26,27] due to the possible numerical
instability caused by the solid phase of carbon. It was found no carbon
formation occurs thermodynamically under the conditions of this
work. Therefore only methyl group, CO, CO2, CH4 plus glycerol, steam
and H2, are the compounds considered in the steam-glycerol system.
As you will see, CO2 removal is also considered in this paper. If
the equation (5) is applied for CO2, it takes the following form:

b lc 2lo =RT 0
DGfCO2 =RT ln nCO2 =nT lnP ln f
i

(7)

when the CO2 removal is utilised in the system, only a minor


modication is needed in Equation (7) as shown below:

b lc 2lo =RT 0
DGfCO2 =RT ln nCO2 $1  f=nT lnP ln f
i
(8)
where f is the fraction of CO2 produced in the steam-glycerol
system that is removed. Accordingly, the total molar number of
species is reduced by nCO2 $f and is expressed by:

nT

N
X

ni  nCO2 $f ; i 1; .; N

(9)

i1

As the inert gas (nitrogen) was used as the carrier gas in the
experiments which is also commonly used in the experimental
studies of steam reforming of glycerol [5,6,11,14,17,20,21]. Nitrogen
is also considered in the thermodynamic analysis. Two non-linear
algebraic equations are therefore added in the chemical equilibrium system. One accounts for the atomic balance (Equation 10),
and the other for Gibbs free energy (Equation 11):

2nN2 21 WGFR=FI


b 2l =RT 0
DGfN2 =RT ln nN2 =nT ln P ln f
N
i

(10)
(11)

where the FI refers to the molar ratio of the feed reactants


(H2O Glycerol) to the inert gas (Nitrogen), WGFR is the water to
glycerol feed ratio. Accordingly, the molar number of the nitrogen,
nN2, should be added to the total molar number of species, nT.
In total, 4 atoms (C, H, O, N) and 7 compounds (C3H8O3, H2O, H2,
CO, CO2, CH4, N2) are considered thermodynamically in the equilibrium system. The equations (3), (5), (7e11) form a system consisting of 12 non-linear algebraic equations with 12 unknown
variables (4 atoms, 7 compounds and nT). An in-house Fortran code
has been developed to solve the equations. The Fortran code solves
a system of non-linear equations using a modied Powell hybrid
algorithm and a nite-difference approximation to the Jacobian.

H. Chen et al. / Renewable Energy 36 (2011) 779e788

The program needs the input of the number of atoms, number of


compounds, temperature, pressure, value of Gibbs free energy and
guessed initial values of the variables at the beginning. All the
thermodynamic data were obtained from Refs. [24,25]. During the
calculations, the input of glycerol was kept constant as 1 mol and
the moles of H2O and N2 were adjusted depending on the ratios of
WGFR and FI respectively.
3. Experimental
The experimental system used in this work is schematically
shown in Fig. 1. It consists of 10 major components: (1) a feeding
tank containing the water and glycerol mixture; (2) a HPLC pump
(100 Flash pump, SMI-LabHut, UK); (3) an evaporator; (4) a catalyst
packed reactor; (5) a cooler; (6) a drier; (7) a gas composition
analyser (A02000 Continuous gas analyser, ABB, Germany); (8)
a Data Acquisition System (DAQ System, National Instrument, US);
(9) a heating and temperature controller (Watlow, UK); and (10)
a feeding gases supply unit.
The catalyst packed reactor was made of stainless steel and has
an internal diameter of 41 mm, a thickness of 3.5 mm and a length
of 1100 mm. The rst 300 mm is the so-called preheating zone
where 5 mm glass balls were randomly packed for further

781

preheating and more uniform ow. The zone from 300 mm to


1100 mm is the so-called reaction zone which was randomly
packed with a commercial nickel/nickel oxide-based catalyst (1/800
cylindrical pellets; BET area of >70.00 m2/g) promoted with chromium oxides for steam reforming of hydrocarbons (KATALCO CRGLHR, Johnson Matthey Catalysts, UK) containing Ni (metal), NiO,
Cr2O3, MgO and amorphous silica (bulk density is 1.20e1.60 g/ml).
Totally about 1.7 kg catalyst was packed into the reactor. The reactor
was heated by a distributed ceramic heater system with the
programmable temperature controller (Watlow, UK). The ceramic
heater was insulated with w50 mm thickness porous insulation
materials. The reactor was tted with inlet and outlet lines for
introducing the feed gases (gaseous glycerol and steam), purge gas
(H2/N2 and steam), catalyst reduction gas (H2) and efuent gas. The
pipelines of the whole system were heat-traced by heating tape
(HT9, Electrothermal, UK) to ensure no condensation.
Before each run, the catalyst was reduced with a continuous
ow of 2.0 SLPM H2 mixed with 3.0 SLPM N2 at 600  C for 4 h under
atmosphere pressure. During the experiments, the reaction system
was rst heated up to the preset temperature at atmospheric
pressure (w2 h) and then the preset pressure for 20 min by using
the carrier gas N2. After the system reaches the thermal and ow
steady state, liquid glycerol-water mixture was supplied to the
reactor via the evaporator where the mixture was gasied. Most the
efuent ow was exhausted to atmosphere directly except a small
amount was sampled to the on-line ABB gas analyser after
removing the water via the condenser and drier.
The ow rate of the liquid mixture was controlled by the HPLC
pump and those of gases were metered and controlled by mass ow
controllers. Various thermocouples (J Type, TC direct, UK), a pressure transducer (RS, UK) and a pressure regulator (Swagelok, UK)
were installed to measure and control temperature and pressure
respectively, See Ref. [28] for details. All the signals of temperature,
pressure, ow rate and gas compositions were recorded by the DAQ
system. The thermocouples were calibrated using a thermal bath
and found to have an accuracy of 0.5  C. The accuracy of the gas
ow controller was calibrated by a modelled ow meter (RS, UK)
and the uncertainty was found below 2.0%. The ABB analyser was
calibrated by the gas mixture of methane, hydrogen, carbon dioxide
and nitrogen (16%/16%/16%/balance, v/v) before each experiment
and the uncertainties of the gas concentrations were estimated
within 0.5%.
During the data processing, the glycerol conversion cgly
calculation is based on carbon molar outow of the gas products
and the glycerol molar inow of the reactor:

cgly

P
n_ gly;in  n_ gly;out
n_ dry;out yC;out

n_ gly;in
3n_ gly;in

(12)

_ and y refer to the relevant molar ows and gas mol


where n
fractions, respectively, and n_ dry;out is the dry total molar ow of gas
products leaving the reactor calculated through an elemental
balance of nitrogen:

n_ dry;out

Fig. 1. Experimental system.

n_ N2 ;in
n_ N2 ;in


P
yN2 ;out
1  yi;out

(13)

where yi,out refers to the dry gas products (except for water and
nitrogen) at outlet. Due to the limit of experimental apparatus, only
four gas compositions N2, H2, CO2 and CH4 were recorded by the
ABB analyser. This assumes no other signicant dry gases than N2,
H2, CO2 and CH4 evolved from the reactor. This has been veried by
both experimental [5] and theoretical results [15] over the experimental conditions of this work. In Ref. [5] Only negligible C2H6
except for N2, H2, CO, CO2 and CH4 was detected in the experiments.

782

H. Chen et al. / Renewable Energy 36 (2011) 779e788

It was also found that CO concentration is very low under the


conditions of this work and the maximum relative error of glycerol
conversion led by not recording CO is about 2%. The conversion of
steam cH2 O is given by:

n_ H2 O;conv
n_ H2 O;in

(14)

which is based on the hydrogen elemental balance. H2 selectivity in


steam reforming of glycerol is dened as:

H2;Sel

2n_ H2 ;out

2cH2 O n_ H2 O;in 8cgly n_ gly;in

(15)

H2 purity was calculated by the ratio of H2 volume fraction to the


sum of the volume fractions of gases products (except for water and
nitrogen) at the outlet.

yH
H2;pur P 2;out
yi;out

(16)

4. Results and discussion


4.1. Transient prole of product gases
A typical transient prole of dry product compositions at outlet
is shown in Fig. 2. The reaction conditions for Fig. 2 are 0.5 SLPM
glycerol, 4.5 SLPM steam, 20 SLPM nitrogen, 450  C temperature
and 1 bar pressure. The corresponding WGFR is 9:1 and FI is1:4
respectively. One can see in Fig. 2 that it takes about 700 s for the
system to reach a steady state. Similar results are observed under
other conditions. The averaged experimental results at the steady
state are then used for the comparisons with equilibrium results
from thermodynamic analyses.
4.2. Effect of temperature
The inuence of temperature on the product compositions is
shown in Fig. 3. Due to the heating ability of the experimental
system, only three temperature points around 450  C were investigated. This work is more relevant to conditions at low temperature. The other conditions are kept constants as P 1 bar,
WGFR 9:1, FI 1:4.
In Fig. 3, the same trend is clearly indicated in the theoretical
and measured results. With increasing temperature, the molar
numbers of hydrogen and carbon dioxide increase up to the optimal

10
CH4
CO2
H2

Concentration (%)

8
6
4
2
0

500

1000

1500

2000

2500

3000

Reaction Time (s)


Fig. 2. Transient prole of dry product compositions: T 450
WGFR 9:1 and FI 1:4.

 C,

P 1 bar,

6
Number of Moles

cH2 O



n_ dry;out 2yH2 4yCH4 8xgly n_ gly;in

2n_ H2 O;in

Theorectical data, H2
Theorectical data, CO2
Theorectical data, CH4
Theorectical data, CO
Measured data, H2
Measured data, CO2
Measured data, CH4

5
4
3
2
1
0
350

400

450

500

550

600

650

700

Temperature ( C)
Fig. 3. Effect of temperature: number of moles.

values and decrease thereafter, whereas the major competitive


product e methane decreases eventually. The temperature has
a strong effect on hydrogen production. The maximum hydrogen
production occurs at w580  C with the optimal number of
hydrogen moles of 6.2. The optimal number of hydrogen moles is
higher than the corresponding thermodynamic results in Ref. [22]
because of the use of inert gas (N2 in this work). See more discussion in Section 4.5. The production of carbon oxide is negligible at
temperature of 350  C and then increase eventually.
The hydrogen and carbon dioxide productions increase and, in
the meanwhile, the methane production decreases with increasing
temperature, suggesting the steam reforming of methane
CH4 2H2 O CO2 4H2 has occurred signicantly. This indicates the catalyst is not only active for steam reforming of glycerol
but also for steam reforming of methane [6].
On the other hand, in Fig. 3, it is also shown that the experimental results of hydrogen moles are slightly lower than those of
the equilibrium calculation. The experimental value, for example,
for the typical case T 450  C, P 1 bar, WGFR 9:1 and Fi 1:4, is
3.86 mol H2/mol glycerol which achieves 91.4% of equilibrium.
However, the measured carbon dioxide and methane productions
are very close to the corresponding theoretical values. This indicates the conversion of glycerol is almost complete whereas the
conversion of steam is not. To show this explicitly, Fig. 4 is
presented.
Fig. 4 shows the results of glycerol conversion, steam conversion, hydrogen selectivity and hydrogen purity as a function of
temperature. Considering the experimental error and the purity of
glycerol, the glycerol conversions at all the three temperatures are
regarded as completion (>96%). The results are consistent with the
theoretical results and similar to experimentally observed results
by [10,20]. Ref. [20] observed a complete glycerol conversion from
400  C with Ir/CeO2 catalyst and Ref. [10] achieved an over 98%
conversion in a thermal decomposition of glycerol above 260  C.
However, the steam conversions are lower than the theoretical
predictions. This indicates the major reactions of steam reforming
of glycerol can be regarded as steam reforming of methane and
water gas shift reaction after glycerol is completely converted into
C1 products, water and hydrogen. The catalyst for steam reforming
of glycerol is required to have a sufcient capacity for steam
reforming of methane for hydrogen simultaneously. There is still
a room for improving the catalyst used in this work.
Fig. 4 also shows that the hydrogen selectivity, hydrogen purity
and steam conversion all increase with increasing temperature. A
combination of Figs. 3 and 4 suggests high temperatures are
favourable for steam reforming of glycerol.

H. Chen et al. / Renewable Energy 36 (2011) 779e788

by the theoretical analysis over the pressures considered. The


possible reason is the formations of a small amount of carbon, CxHy
and OCHs, which were experimentally observed by Refs. [5], [17]
and [14] respectively.

Theorectical data, Glycerol Conversion


Theorectical data, H2O Conversion
Theorectical data, H2 Selectivity
Theorectical data, H2 Purity
Measured data, Glycerol Conversion
Measured data, H2O Conversion
Measured data, H2 Selectivity
Measured data, H2 Purity

4.4. Effect of WGFR

1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
350

400

450

500

550

600

650

700

Temperature ( C)
Fig. 4. Effect of temperature: glycerol conversion, steam conversion, hydrogen selectivity and hydrogen purity.

4.3. Effect of pressure


Fig. 5 shows the effect of pressure on the product composition.
The other conditions are kept constants as T 450  C, WGFR 9:1,
FI 1:4. Again, the same trend is obtained in the theoretical and
measured results. With increasing pressure, the molar numbers of
hydrogen and carbon dioxide decrease, whereas the major
competitive product e methane increases eventually. The production of carbon oxide is very small over the pressures considered and
the effect of pressure on the carbon oxide production is negligible.
Results of glycerol conversion, steam conversion, hydrogen
selectivity and hydrogen purity as a function of pressure are shown
in Fig. 6. One can see that, increase of pressure is a negative factor
for all the four parameters. A combination of Figs. 5 and 6 indicates
low pressures are favourable for steam reforming of glycerol. It is
noticed that the glycerol conversion deceases slightly with
increasing the pressure although a complete conversion is achieved

The effect of WGFR on the product composition is shown in


Fig. 7 at T 450 C, P 1 bar, FI 1:4. Similarly, the theoretical and
measured results have the same trend as a function of WGFR. The
molar numbers of hydrogen and carbon dioxide increase with
increasing WGFR, whereas the major competitive product e
methane decreases eventually. The production of carbon oxide is
very small over the WGFR considered and the effect of WGFR on the
carbon oxide production is negligible. A high WGFR is favourable
for hydrogen production.
A similar conclusion can also be drawn from the results of
glycerol conversion, hydrogen selectivity and hydrogen purity as
shown in Fig. 8. One can see that, an increase of WGFR results in the
increase of glycerol conversion, hydrogen selectivity and hydrogen
purity, which indicates again a high WGFR is favourable for
hydrogen production. However, it is also noticed that the steam
conversions of both experimental and theoretical results have
optimal values. This implies that there is an optimised WGFR for
steam reforming of glycerol. A combination of Figs. 6 and 7 implies
WGFR 9.0 is the optimised WGFR based on both the experimental
results and theoretical analysis. Another important reason for
WGFR 9.0 being the optimised condition is that the measured
glycerol conversion is blow 100% if the WGFR is lower than 9.0. In
Fig. 8, the measured glycerol conversion decreases with decreasing
WGFR signicantly indicating a stronger effect of WGFR in
comparison with the results in Figs. 4 and 6 respectively. Similarly,
the possible reason is the formations of carbon, CxHy and OCHs
[5,14,17]. In fact, during the current experiments with WGFR 5.0,
black particles did be visibly detected in the exhausting gas. This
indicates WGFR also plays an important role in inhibitions of
carbon deposition.

Theorectical data, Glycerol Conversion


Theorectical data, H2O Conversion
Theorectical data, H2 Selectivity
Theorectical data, H2 Purity
Measured data, Glycerol Conversion
Measured data, H2O Conversion
Measured data, H2 Selectivity
Measured data, H2 Purity

1.0
Theorectical data, H2
Theorectical data, CO2
Theorectical data, CH4
Theorectical data, CO
Measured data, H2
Measured data, CO2
Measured data, CH4

6
Number of Moles

783

5
4

0.9
0.8
0.7
0.6
0.5
0.4

0.3

0.2
0.1

0.0
1

0
1

Pressure (bar)

Pressure (bar)
Fig. 5. Effect of pressure: number of moles.

Fig. 6. Effect of pressure: glycerol conversion, steam conversion, hydrogen selectivity


and hydrogen purity.

784

H. Chen et al. / Renewable Energy 36 (2011) 779e788

7
6
5

Number of Moles

FIs are favourable for steam reforming of glycerol. This can be


explained by the so-called dilution effect and the equation (1)
moves forward if additional inert gas applied. According to the Le
Chateliers principle, if a chemical system at equilibrium experiences a change in concentration, temperature, volume, or partial
pressure, then the equilibrium shifts to counteract the imposed
change. When an inert gas is added into the equilibrium system at
a constant pressure, the partial pressures of reactive gases will
decrease (dilution effect) hence resulting in a shift towards the
direction with the greater number of moles of gas.

Theorectical data, H2
Theorectical data, CO2
Theorectical data, CH4
Theorectical data, CO
Measured data, H2
Measured data, CO2
Measured data, CH4

4
3
2

4.6. Effect of feeding gas ow rate (residence time)

The effect of feeding gas ow rate on the product composition is


shown in Fig. 11 at T 450  C, P 1 bar, WGFR 9:1. During the
experiments FI was kept a constant of 1:4. The feeding reactants
ow rates were taken as 2.5 SLPM, 3.75 SLPM, 5.0 SLPM, 6.25 SLPM
and 7.5 SLPM respectively. Accordingly, the corresponding inert gas
ow rates are 10 SLPM, 15 SLPM, 20 SLPM, 25 SLPM and 30 SLPM
respectively.
One can see that, in Fig. 11, although the variations are not
signicant, the molar numbers of hydrogen and carbon dioxide
decrease with increasing feeding gas ow rate, whereas the molar
number of methane increases eventually. This implies a high residence time results in a higher hydrogen production. Fig. 12 also
supports the conclusion that a high residence time is favourable for
hydrogen production showing the hydrogen selectivity, hydrogen
purity and steam conversion (the variation of glycerol conversion is
marginal) all decrease with increasing feeding gas ow rate. The
results of feeding gas ow rate (residence time) effects are
consistent with those obtained by Ref. [29] for steam reforming of
methane, are opposite to those obtained Ref. [14] for reforming of
glycerol using supercritical water. The large pressure (241 bar) used
by Ref. [14] may be the mechanism behind the controversy as the
machination of carbon dioxide CO2 4H2 CH4 2H2 O becomes
highly favourable at such a high pressure [14,15,22].

0
4

10

11

12

WGFR
Fig. 7. Effect of WGFR: number of moles.

4.5. Effect of feeding reactants to inert gas ratio


The effect of feeding reactants to inert gas ratio (FI) on the
product composition is shown in Fig. 9 at T 450  C, P 1 bar,
WGFR 9:1. During the experiments the inert gas (N2) was kept
a constant of 20 SLPM. The feeding reactants ow rates were taken
as 2.5 SLPM, 5.0 SPLM and 7.5 SLPM respectively, which correspond
to the FIs of 1:8, 2:8 (1:4) and 3:8 respectively. One can see that
Fig. 9 is similar to Fig. 5. The molar numbers of hydrogen and
carbon dioxide decrease with increasing FI, whereas the molar
number of methane increases eventually. A low FI is favourable for
hydrogen production. Fig. 10 is also similar to Fig. 6.The hydrogen
selectivity, hydrogen purity and steam conversion (the variation of
glycerol conversion is marginal) all decrease with increasing FI. Low

Theorectical data, Glycerol Conversion


Theorectical data, H2O Conversion
Theorectical data, H2 Selectivity
Theorectical data, H2 Purity
Measured data, Glycerol Conversion
Measured data, H2O Conversion
Measured data, H2 Selectivity
Measured data, H2 Purity

5. Further discussion on optimisation


A combination of the Figs. 2e8 suggests that the optimised
conditions for hydrogen production are T 580  C, P 1 bar,
WGFR 9:1 with a maximum hydrogen production 6.16 mol H2/
mol glycerol. According to equation (1) the ideal hydrogen
production is 7.0 mol H2/mol glycerol. There still is a room for
further optimisation.

1.0
Theorectical data, H2
Theorectical data, CO2
Theorectical data, CH4
Theorectical data, CO
Measured data, H2
Measured data, CO2
Measured data, CH4

0.9
0.8

0.7

0.6
Number of Moles

0.5
0.4
0.3
0.2
0.1
0.0
4

10

11

12

WGFR

5
4
3
2
1
0
0.250

Fig. 8. Effect of WGFR: glycerol conversion, steam conversion, hydrogen selectivity and
hydrogen purity.

0.375

0.500

0.625

Fig. 9. Effect of FI: number of moles.

0.750

H. Chen et al. / Renewable Energy 36 (2011) 779e788

785

Theorectical data, Glycerol Conversion


Theorectical data, H2O Conversion
Theorectical data, H2 Selectivity
Theorectical data, H2 Purity
Measured data, Glycerol Conversion
Measured data, H2O Conversion
Measured data, H2 Selectivity
Measured data, H2 Purity

Theorectical data, Glycerol Conversion


Theorectical data, H2O Conversion
Theorectical data, H2 Selectivity
Theorectical data, H2 Purity
Measured data, Glycerol Conversion
Measured data, H2O Conversion
Measured data, H2 Selectivity
Measured data, H2 Purity

1.0

1.0

0.9

0.9

0.8

0.8

0.7

0.7

0.6
0.5

0.6

0.4

0.5

0.3

0.4

0.2

0.3

0.1

0.2

0.0
0.250

0.1

0.500

0.625

0.750

feeding ratio

0.0
0.250

0.375

0.500

0.625

0.750

feeding ratio
Fig. 10. Effect of FI: glycerol conversion, steam conversion, hydrogen selectivity and
hydrogen purity.

5.1. Decrease of pressure


In Fig. 5 one can see that the decrease of pressure below 1 bar
could further enhance the hydrogen production. However, an
operating pressure below the ambient pressure is difcult to achieve in practice. An alternative way is use of the carrier gas, as
shown in Figs. 3 and 9, to decrease the partial pressure of reactants.
In Fig. 3, the optimised conditions are similar to those obtained by
Ref. [22] using a thermodynamic analysis in absence of the carrier
gas, but with a higher hydrogen production and a lower optimal
temperature (the optimal conditions in Ref. [22] are T w660  C,
P 1 bar, WGFR 9:1 with a maximum hydrogen production

Theorectical data, H2
Theorectical data, CO2
Theorectical data, CH4
Theorectical data, CO
Measured data, H2
Measured data, CO2
Measured data, CH4

7
6
Number of Moles

0.375

3
2
1
0
0.375

0.500

5.95 mol H2/mol glycerol). A lower operating temperature is also


critical important for industrial application enabling the reductions
of energy usage, catalyst sintering and material expense [5,28,29]. A
comparison between the working conditions of this work and
Ref. [22] indicates the use of carrier gas is the key factor for such an
improvement.
Fig. 13 shows the hydrogen production at different feeding
reactants to inert gas ratio. One can see that, with increasing carrier
gas ratio, not only the maximum hydrogen production increases
but also the optimal temperature (at which the maximum
hydrogen production occurs) drops eventually. For example, when
the FI increases from 1:0 to 1:32, the maximum hydrogen
production increases from 5.95 mol H2/mol glycerol to 6.43 mol H2/
mol glycerol. Correspondingly, the optimal temperature at which
the maximum hydrogen production occurs drops from w660  C to
w500  C. The addition of the carrier gas can be an effective way for
enhancement of steam-glycerol reforming. However, the addition
of carrier gas seems not desirable in the industrial applications,
which requires complex afterward separation processes. The
alternative way is to replace the inert gas using steam, see more
discussion in Section 5.2.
5.2. Increase of feeding steam ratio

0.250

Fig. 12. Effect of Effect of feeding gas ow rate: glycerol conversion, steam conversion,
hydrogen selectivity and hydrogen purity.

0.625

0.750

nf
Fig. 11. Effect of feeding gas ow rate: number of moles.

If the inert gas is replaced by steam, Fig. 14 is obtained. One


can see that the hydrogen production is enhanced considerably in
comparison with that in Fig. 13. For example, the maximum
hydrogen production at FI 1:4 (WGFR 49:1) is very close to
7.0 mol H2/mol glycerol e the ideal molar number of hydrogen
production. At the same time, the optimal temperature, at which
the maximum hydrogen production occurs, drops to w500  C.
The mechanism lies in the replacement of steam for the carrier
gas not only decreases the partial pressure of glycerol but also,
more importantly, increases the reactant e water. According to
equation (1), both the two effects could enhance the forward
reaction.

786

H. Chen et al. / Renewable Energy 36 (2011) 779e788

7
6
Number of Moles

5
4

FI=1:0
FI=1:1
FI=1:2
FI=1:4
FI=1:8
FI=1:16
FI=1:32

3
2
1
0
350

400

450

500

550

600

650

700

Temperature ( C)
Fig. 15. Process with the heat regenerator.

Fig. 13. Hydrogen production at different feeding reactants: P 1 bar, WGFR 9.0.

Although the separation of water after reforming has no technical difculties, excess steam may be a hammer requiring a large
amount of water consumption and energy for heating and evaporating water. A possible methodology to tackle the problem is
shown in Fig. 15 by using a heat regenerator. In Fig. 15, the
exhausting gas is introduced into the heat regenerator, inside
which the steam is cooled and condensed and then is separated in
the liquidegas separator. Gases hydrogen and carbon dioxide are
obtained for further processing. At the same time, the liquid water
is recycled and pumped back through the heat regenerator to
extract the heat from exhausting gas and then is injected and mixed
with feeding glycerol for reforming reaction. One can see that the
process in Fig. 15 combines the heat regeneration and water
separation within one heat regenerator.
It should be noticed that although the increase in feeding steam
ratio would increase the production of hydrogen, this effect
becomes limited at high feeding steam ratio, especially at higher
temperature [30]. Additionally, the regenerator could be very big
comparing with the reactor and may be not so cost effective for the
steam reforming process at very high feeding steam ratio conditions. Further work is obviously needed for optimising the process
in Fig. 15.

carbon dioxide as soon as it is formed from the reaction zone. This is


the so-called in-situ adsorption enhanced steam reforming for
hydrogen production and has been widely accepted as an effective
method to improve the reforming process for hydrogen production
[5,15,29,31,32]. In this way, the glycerol conversion and rate of
forward reaction of equation (1) can be enhanced. Fig. 16 shows the
thermodynamic results as a function of CO2 removal fraction. One
can see that, with increasing CO2 removal ratio, the productions of
H2 and CO2 increase whereas those of CH4 and CO decease. The
productions of H2 and CO2 are enhanced and those of CH4 and CO
are inhibited due to the CO2 adsorption. This is desirable for
improving steam-glycerol reforming for hydrogen production. The
ideal case with 100% CO2 removal, in Fig. 16, yields 7 mol of
hydrogen, 3 mol of CO2, 0 mol of CH4, 0 mol of CO. Fig. 17 shows
a new process to realize the removal of CO2 and, at the same time,
incorporate the process of water and heat recycling. The process
uses a solid adsorbent for CO2, which circulates between the reactor
(for the steam reforming with in-situ adsorption of CO2) and the
desorber (for ex-situ CO2 desorption), thus allowing a continuous
operation of the reactor.

5.3. Carbon dioxide removal

H2, FI=1:0
CH4, FI=1:0
CO, FI=1:0
CO2, FI=1:0

According to Le Chateliers principle, besides use of excess steam,


another possible way to promote the steam reforming is to remove
8

7
6

5
4

FI=1:0 (WGFR=9:1)
FI=1:1 (WGFR=19:1)
FI=1:2 (WGFR=29:1)
FI=1:4 (WGFR=49:1)
FI=1:8 (WGFR=89:1)
FI=1:16(WGFR=169:1)
FI=1:32(WGFR=329:1)

3
2
1
400

450

500

550

600

650

700

Molar number

Number of Moles

0
350

H2, FI=1:4
CH4, FI=1:4
CO, FI=1:4
CO2, FI=1:4

5
4
3
2
1
0
0.0

0.2

0.4
0.6
CO2 removal fraction

0.8

1.0

Temperature ( C)
Fig. 14. Hydrogen production at different feeding steam ratio: P 1 bar.

Fig. 16. Moles of equilibrium compositions as a function of CO2 removal fraction


(WGFR 9:1, P 1 bar, T 450  C).

H. Chen et al. / Renewable Energy 36 (2011) 779e788

f
G
H
DH298
i
K
M
N
n
nf
O
P
R
RC
S
S/C
WGFR
T
V
y
Fig. 17. Process with CO2 adsorption.

6. Conclusions
A detailed comparative study on thermodynamic and pilot-scale
experimental analyses of glycerol reforming for hydrogen production has been conducted in terms of temperature, pressure, water
to glycerol feed ratio, feeding reactants to inert gas ratio and
feeding gas ow rate (residence time). Following conclusions can
be drawn:
(1) Thermodynamic and experimental data agree fairly well with
each other and they have same trend in terms of all the 5
parameters. But the measured hydrogen production is slightly
lower than that predicted by the thermodynamic analysis. The
major reason lies in the conversion of steam is incomplete.
(2) High temperatures, low pressure, low feeding reactants to inert
gas ratio and low gas ow rate are favourable for steam
reforming of glycerol for hydrogen production. However, there
is an optimal water to glycerol feed ratio for steam reforming of
glycerol for hydrogen production which is about 9.0. The
glycerol conversion is a strong function of water to glycerol
ratio, but a weak function of other parameters over the
conditions of this work.
(3) Methodologies for further optimisation of the steam reforming
of glycerol for hydrogen production have been discussed
including a lower feeding reactants to inert gas ratio, a higher
water to glycerol feed ratio and carbon dioxide removal. A new
adsorption enhanced reaction process incorporating water and
heat recovery is proposed which clearly requires further
investigation.
Acknowledgement
The authors would like to extend their thanks to UK EPSRC for
nancial support under Grant EP/F027389/1.
Nomenclature

AC
aji
bj
C

Activity of carbon formation


number of j-type atoms in the i-species
number of j-type atoms
carbon atom

m
l
b
f

787

CO2 removal fraction


Gibbs fee energy
hydrogen atom
formation enthalpy
number of component
equilibrium constants for reaction
total number of atoms
total number of species
molar number
molar ratio of carrier gas (nitrogen) to the feed reactants
oxygen atom
pressure
universal gas constant
relative change of product production
entropy
molar ratio of steam to carbon feed
molar ratio of water to glycerol feed
temperature
volume
molar fraction
chemical potential
Lagrangian multipliers
fugacity coefcient of the gas mixture

References
[1] Cortright RD, Davda RR, Dumesic JA. Hydrogen from catalystic reforming of
biomass-derived hydrogcarbons in liquid water. Nature 2002;418:964e7.
[2] Hu X, Lu G. Investigation of the steam reforming of a series of model
compounds derived from bio-oil for hydrogen production. Appl Catal B
Environ 2009;88:376e85.
[3] Saxena RC, Adhikari DK, Goyal HB. Biomass-based energy fuel through
biochemical routes: a review. Renewable Sustainable Energy Rev
2009;13:167e78.
[4] Hirai T, N-o Ikenanga, Miyake T, Suzuki T. Production of hydrogen by steam
reforming of glycerin on Ruthenium catalyst. Energy & Fuels 2005;19:1761e2.
[5] Dou B, Dupont V, Rickett G, Blakeman N, Williams PT, Chen H, et al. Hydrogen
production by sorption-enhanced steam reforming of glycerol. Bioresour
Technol 2009;100:3540e7.
[6] Douette AMD, Turn SQ, Wang W, Keffer VI. Experimental investigation
of hydrogen production from glycerin reforming. Energy & Fuels 2007;
21:3499e504.
[7] Huber GW, Shabaker JW, Dumesic JA. Raney Ni-Sn Catalyst for H2 production
from biomass-derived hydrocarbons. Science 2003;300:2075e7.
[8] Cameron DC, Koutsky JA. Final report to national biodiesel development
board. Internet site: http://www.biodiesel.org/resources/reportsdatabase/
reports/gen/19941001_gen-243.pdf; 1994.
[9] Dauenhauer PJ, Salge JR, Schmidt LD. Renewable hydrogen by autothermal
steam reforming of volatile carbohydrates. J Catal 2006;244:238e47.
[10] Dou B, Dupont V, Williams PT, Chen H, Ding Y. Thermogravimetric kinetics of
crude glycerol. Bioresour Technol 2008;100:2613e20.
[11] Valliyappan T, Bakhshi NN, Dalai AK. Pyrolysis of glycerol for the production of
hydrogen or syn gas. Bioresour Technol 2008;99:4476e83.
[12] Adhikari S, Fernando S, Haryanto A. A comparative thermodynamic and
experimental analysis on hydrogen production by steam reforming of glycerin. Energy & Fuels 2007;21:2306e10.
[13] Adhikari S, Fernando SD, Haryanto A. Hydrogen production from glycerin by
steam reforming over nickel catalysts. Renewable Energy 2008;33:1097e100.
[14] Byrd AJ, Pant KK, Cupta RB. Hydrogen production from glycerol by reforming
in supercritical water over Ru/Al2O3 catalyst. Fuel 2008;87:2956e60.
[15] Chen H, Zhang T, Dou B, Dupont V, Williams PT, Ghadiri M, et al. Thermodynamic analyses of adsorption enhanced steam reforming of glycerol for
hydrogen production. Int J Hydrogen Energy 2009;34:7208e22.
[16] Dou B, Dupont V, Williams PT. Computational uid dynamics simulation of
gas-solid ow during steam reforming of glycerol in a uidized bed reactor.
Energy & Fuels 2008;22:4102e8.
[17] Iriondo A, Barrio VL, Cambra JF, Arias PL, Guemez MB, Navarro RM, et al.
Inuence of La2O3 modied support and Ni and Pt active phase on glycerol
steam reforming to produce hydrogen. Catal Commun 2009;10:1275e8.
[18] Rossi CCRS, Alonso CG, Antunes OAC, R.Guirardello, Cardozo-Filho L. Thermodynamic analysis of steam reforming of ethanol and glycerine for
hydrogen production. Int J Hydrogen Energy 2009;34:323e32.
[19] Wang X, Li S, Wang H, Liu B, Ma X. Thermodynamic analysis of glycerin steam
reforming. Energy & Fuels 2008;22:4285e91.
[20] Zhang B, Tang X, Li Y, Xu Y, Shen W. Hydrogen production from steam
reforming of ethanol and glycerol over ceria-supported metal catalysts. Int J
Hydrogen Energy 2007;32:2367e73.

788

H. Chen et al. / Renewable Energy 36 (2011) 779e788

[21] Murata K, Wang L, Saito M, Inaba M, Takahara I, Mimura N. Hydrogen


production from steam reforming of hydrocarbons over alkaline-earth metalmodied Fe- or Ni-Based catatlysts. Energy & Fuels 2004;18:122e6.
[22] Adhikari S, Fernando S, Gwaltney S, Filip SD, Bricka RM, Steele PH, et al. A
thermodynamic analysis of hydrogen production by steam reforming of
glycerol. Int J Hydrogen Energy 2007;32:2875e80.
[23] Slinn M, Kendall K, Mallon C, Andrews J. Steam reforming of biodiesel byproduct to make renewable hydrogen. Bioresour Technol 2008;99:8.
[24] Lide DR. CRC handbook of chemistry and physics. Florida, USA: CRC Press;
2005.
[25] Yaws CL. Handbook of Thermodynamic and Physical Properties of Chemical
Compounds. Online version available at. Knovel, http://knovel.com/web/
portal/browse/display?_EXT_KNOVEL_DISPLAY_bookid667&VerticalID0;
2003.
[26] Garcia EY, Larorde MA. Hydrogen production by the steam reforming of
ethanol: thermodynamic anaylsis. Int J Hydrogen Energy 1991;16:307e12.

[27] Mas V, Kipreos R, Amadeo N, Laborde M. Thermodynamic analysis of ethalnol/


water system with the stoichiometric method. Int J Hydrogen Energy;
2006:31.
[28] Cong TN, He Y, Chen H, Ding Y, Wen D. Heat transfer of gasesolid two-phase
mixtures owing through a packed bed under constant wall heat ux
conditions. Chem Eng J 2007;130:1e10.
[29] Cong TN. Adsorption enhanced steam-methane reforming for low temperature hydrogen production using solids circulation. PhD thesis, University of
Leeds, Leeds, UK, 2009.
[30] Comas J, Laborde M, Amadeo N. Thermodynamic analysis of hydrogen
production from ethanol using CaO as a CO2 sobent. J Power Sources
2004;138:7.
[31] Huffton JR, Mayorga S, Sircar S. Sorption-enhanced reaction process for
hydrogen production. Am Inst Chem Engineers J 1999;45:9.
[32] Ding Y, Alpay E. Adsorption-enhanced steam-methane reforming. Chem Eng
Sci 2000;55:12.

Potrebbero piacerti anche