Comparison of several glycerol reforming methods for hydrogen and syngas production using

thermodynamic analysis

Antonio C. D. Freitas, Reginaldo Guirardello*

School of Chemical Engineering, University of Campinas (UNICAMP), Av. Albert Einstein 500,
13083-852, Campinas-SP, Brazil.
*
Corresponding author. Tel.: +55 19 3521 3955; fax: +55 19 3521 3910.
e-mail address: guira@feq.unicamp.br

Abstract
This paper focuses on the comparison of different glycerol reforming technologies aimed to
hydrogen and syngas production. The reactions of steam reforming, partial oxidation, autothermal
reforming, dry reforming and supercritical water gasification were analyzed. For this, the Gibbs
energy minimization approach was used in combination with the virial Equation of state. The
validation of the model was made between the simulations of the proposed model and both,
simulated and experimental data obtained in the literature. The effects of modifications in the
operational temperature, operational pressure and reactants composition were analyzed with regard
to composition of the products. The effect of coke formation was discussed too. Generally, higher
temperatures and lower pressures resulted in higher hydrogen and syngas production. All reforming
technologies demonstrated to be feasible for use in the production of hydrogen or synthesis gas. The
proposed model showed good predictive ability and low computational time to perform the
calculation of the combined chemical and phase equilibrium in all systems analyzed here.
Keywords: Reforming processes, Gibbs energy minimization, Hydrogen, Syngas.

1. Introduction
In the past decades, biomass has received a great deal of attention as a new source for clean
energy because it is renewable, alternative and producing less carbon dioxide than fossil fuels [1]. In
the last decade, biodiesel has been becoming a major substitute for fossil diesel, as its environmental
benign characteristics can reduce carbon dioxide emission pollution and mitigate climate change
problems [2,3].
Glycerol is obtained as a by-product in biodiesel production using vegetable oils by a base-
catalyzed transesterification reaction. Glycerol of high purity is an important industrial feedstock for
applications in food, cosmetics, pharmaceutical and other industries [4]; however, it is costly to
refine crude glycerol, especially for medium and small-sized plants.
In any case, since glycerol production and utilization have a notable impact on the economics
and sustainability of biodiesel production, the development of novel processes for glycerol
valorization is essential. Among all possible routes, glycerol conversion into valuable gases like as
hydrogen and syngas as an energy carrier using different reforming processes are one of the most
attractive ways to aggregate value in the glycerol [4].
The reforming processes used to produce hydrogen and syngas from glycerol include some
methods, and in the literature, it is possible to find some articles about then. The articles principally
discussing the catalytic behavior of the systems, the main related methods for glycerol reforming are
steam reforming (SR) [5-11], partial oxidation (or oxidative reforming) (PO) [12-14], autothermal
reforming (ATR) [12, 14] and Supercritical Water Gasification (SCWG) [15-19].
Other methods like as dry reforming (or CO2 reforming) (DR) [20,21], dry autothermal (a
combination between dry, oxidative and steam reforming) [22] and pyrolysis [23], also, were
evaluated as possible routes for glycerol conversion.
 Normally catalysts were used to promote and accelerate these reforming processes. Ni, Co
and noble metals like as Pt, Pd and Rh based catalysts were related to promote the reforming process.
Among then, the Ni based catalysts are the most commonly used [5, 9].
In this paper, we focused on the comparison of several glycerol reforming technologies. The
SR, PO, ATR, DR and SCWG of glycerol were analyzed. For this, we used the Gibbs energy
minimization approach in combination with the virial EoS. Comparisons between experimental and
simulated data obtained from literature were performed. In addition, the effects of the process
variables (temperature, pressure and inlet composition) were evaluated with respect to the product
compositions. The coke formation was discussed too.

2. Methodology
2.1. Model formulation
2.1.1. Gibbs energy minimization model
The thermodynamic equilibrium condition for reactive multicomponent closed system, at
constant pressure and temperature, with given initial composition, can be obtained by the
minimization of Gibbs energy (G) of the system, with respect of the number of moles of each
component in each phase, given by:
NC NC NC
min G = ∑ nig µ ig + ∑ nil µ il + ∑ nis µ is (1)
i =1 i =1 i =1

While satisfying the restrictions of non-negative number of moles of each component in each
phase:
nig , nil , nis ≥ 0 (2)

In addition, the restriction of mole balances, given by atom balance for reactive systems (non-
stoichiometric formulation):
NC NC

∑ ami .(nig + nil + nis ) = ∑ ami .ni0 m = 1,..., NE (3)

i =1 i =1

Smith and Missen [24] demonstrated that the stoichiometric formulation is equivalent to the
g
non-stoichiometric one, if all independent reactions are considered. The values of µ i can be
calculated from the formation values given at some reference conditions, using the following
thermodynamic conditions:
⎛ ∂H i ⎞ i = 1,..., NC
⎜ ⎟ = Cp i (4)
⎝ ∂T ⎠ P
∂ ⎛ µ i ⎞ Hi
⎜ ⎟ = − i = 1,..., NC (5)
∂T ⎝ RT ⎠ P R.T 2

In this work, which considers only a gas phase and a possible solid phase as carbon for coke
formation, the Gibbs energy can be expressed as:
NC
 NC

i=1
( (
G = ∑ nig µi0 + RT ln P + ln γ i + ln φi )) + ∑ nCs ( s ) µC0( s )
i=1
(6)

Since, some of the systems analyzed by the present work was at high pressure, the virial
equations of state (EoS) truncated at second virial coefficient, were utilized to determine the fugacity
coefficient. The second virial coefficient was calculated by the correlation of Pitzer and Curl [25],
which was modified by Tsonopoulos [26]. The following relation determined the fugacity
coefficient:
 # m & P
ln φi = %2∑ y j Bij − B( (7)
%$ j (' RT

The Gibbs energy minimization method is widely studied in the literature, being applied in a
wide range of systems [27-33]. Some of then related to glycerol reforming processes [19, 27], but
none of them presents a complete thermodynamic study about these reactions.

2.2. Model implementation

During the process of optimization, utilizing the Gibbs energy minimization method the
number of moles of the gaseous ( n ig ) and solid ( nis ) phase are considered decision variables, while
T, P and the chemical potential of the pure component in the reference state ( µ i0 ) are considered
parameters. The software GAMS® 23.2.1, (General Algebraic Modeling System) with the CONOPT
solver was used in the resolution of the combined chemical and phase equilibrium problem.
A description of GAMS software can be found in Brooke et al. [34]. The solid phase formed
was considered as solid carbon (pure component). These methodologies and considerations were
applied in previous works of our research group with good predictive results for similar systems with
methane and biomass [29-33].
A Total of 12 output compounds were selected as representative of the main compounds,
which can be found in the output stream of these reactive systems. The thermodynamic data used in
the calculations were obtained from literature [35-37]. In almost all the thermodynamic analyses of
glycerol reforming processes, the systems were represented considering primary species, such as
hydrogen, methane, carbon monoxide, carbon dioxide, water and glycerol, and in some papers the
solid carbon (C(s)) was considered too. The consideration of the formation of solid carbon is very
important because the C(s) can poison the catalyst in reforming reactions [38].
However, some experimental works in the literature related the presences of secondary
components in these reactions, some of the secondary compounds are methanol, ethanol,
acetaldehyde, acetone and ethane [39]. In this work, a complete thermodynamic analysis of these
reactions was performed and all species mentioned above and some others have been take into
account. The complete list of output compounds, along with their critical parameters, is reported in
Table 1.

Table 1. Chemical compounds considered in the simulations and your thermodynamic properties.
Compound Chemical formula TC (K)1 PC (bar)1 VC (m3/kmol)1 ω (-)1
Water H2 O 647.3 220.0 0.056 0.348
Glycerol C3H8O3 850.0 75.0 0.260 0.513
Hydrogen H2 33.0 13.0 0.064 0.000
Methane CH4 191.1 45.8 0.099 0.013
Carbon dioxide CO2 304.2 73.9 0.094 0.420
Carbon monoxide CO 133.0 35.0 0.093 0.041
Oxygen O2 154.6 50.4 0.073 0.022
Ethane C2H6 305.4 48.2 0.148 0.105
Propane C3H8 369.9 42.0 0.200 0.152
Ethylene C2H4 283.1 50.5 0.124 0.073
Propylene C3H6 369.9 45.4 0.182 0.143
Methanol CH3OH 512.6 80.13 0.116 0.559
1
Source: [35-37].
3. Results and discussion
3.1.Model validation
The validation was performed using simulated and experimental data obtained in literature in
comparison with the simulations performed using the proposed virial model. The objective of the
validating the model is to verify the predictive ability of the model against experimental data
obtained in the literature and verify the ability of the virial equation in represent the non-ideality’s of
the studied systems.
In order to compare the results calculated using the proposed approach with the values found
in literature, the mean relative error (MRE) was used, according to equation 8.
NPE NCE x lit − x sim
1 j ,i j ,i
MRE = .∑ ∑ lit
(8)
NPE . NCE j i x j ,i

where NPE is the number of experimental points, NCE is the number of components for each
experimental point, x sim
j ,i
is the value calculated by the present work using the proposed virial model
and x litj ,i is the experimental or simulated value obtained in the literature.
Figure 1 shows the validation of the virial proposed model in comparison with the
experimental results of Byrd et al. [15] and the simulated results from the work of Castello and Fiori
[27]. The experiments of Byrd et al. [15] were performed at the following conditions: temperature of
800 °C (1073.15 K), pressure of 241 bar and with feed concentration in the range of 5 to 40 wt%.
The simulations of the present work and that obtained in the work of Castello and Fiori [27]
were performed under the same above mentioned conditions. However, Castello and Fiori [27] used
the Peng-Robinson EoS in the Gibbs energy minimization model.

80

70
H2
Molar fraction (mol %)

60

50

40

CO2
30

20
CH4
10
CO
0

0
5
10
15
20
25
30
35
40
45

Feed Concentration (wt%)

Figure 1. Virial model validation with the experimental data from Byrd et al. [15] and the simulated
data from Castello and Fiori [27] for the SCWG of glycerol. Symbols: solid line: virial proposed
model predictions; black symbols: experimental data from Byrd et al. [15]; dashed line + white
symbols: simulated data from Castello and Fiori [27].

Analyzing the Figure 1 it is possible to verify that the virial proposed model predictions are in
good agreement with the experimental behavior reported by Byrd et al. [15]. The highest deviations
were observed in the higher initial feed concentration region (above 25 wt%). It can also be seen that
the model predictions for H2 production is better than that observed for the other compounds (CO2,
CH4 and CO). The MRE between the experimental and simulated data was 0.692.
Still in the Figure 1, it is also interesting to note that the use of the Virial equation instead of
Peng-Robinson EoS does not result in changes in the predictive ability of the Gibbs energy model,
although it leads to a significant mathematical simplification of the Gibbs energy minimization
problem.

3.2.Comparison between the technologies

Table 1 shows the better conditions for H2 production observed in all reaction analyzed with
the operational temperature fixed at 1073.15 K. The SR, PO, ATR and DR were analyzed at 1 bar
and the SCWG was analyzed at 250 bar.

Table 1. Better conditions for H2 production for the different reforming technologies analyzed.
Molar fraction
Process Pressure (bar) Temperature (K) Molar ratio
H2 Syngas
SR 1 1073.15 1/3 (Gly/H2O) 61.9 88.4
PO 1 1073.15 1/0.5 (Gly/O2) 52.9 92.8
ATR 1 1073.15 1/1/1 (Gly/H2O/O2) 51.0 81.5
DR 1 1073.15 1/1 (Gly/CO2) 46.2 92.7
SCWG 250 1073.15 1/9 (Gly/H2O) 40.5 48.2

Analyzing the Table 1 it’s interesting to emphasize that higher molar fraction of H2 was
observed in the SR reaction (with a molar ratio glycerol/water = 1.0/3.0 in the feed stream). The
lower molar fraction of H2 was observed for the SCWG reaction. However this process presented the
lowest production of CO between all processes analyzed, with that, the subsequent purification
processes of H2 become simpler for the SCWG.
The better conditions observed for syngas production (with a molar ratio more close to 2.0 as
possible) was presented in the Table 2. The SR, PO, ATR and DR were analyzed at 1 bar and the
SCWG was analyzed at 250 bar again. Each case presented a different temperature of reaction to the
better condition for syngas production.
Analyzing the Table 2 it is interesting to emphasize that the highest molar fraction of H2 was
observed to the SR reaction (91.4% of the dry gas), with 1273.15 K of operational temperature and
with a glycerol/water molar ratio of 1.0/3.0, in that conditions the H2/CO molar ratio was 1.99.
The SCWG of glycerol presented the highest H2/CO molar ratio between all analyzed
processes, 3.21 at 1273.15 K and with a glycerol/H2O molar ratio of 1.0/9.0.

Table 2. Better conditions for syngas production for the different reforming technologies analyzed.
Pressure molar fraction ratio moles
Process Temperature (K) molar ratio
(bar) H2 Syngas H2/CO C(s)
SR 1 1273.15 1/3 (Gly/H2O) 60.8 91.4 1.99 0.00
PO 1 873.15 1/0.5 (Gly/O2) 46.5 65.4 2.44 0.65
ATR 1 973.15 1/1/1 (Gly/H2O/O2) 51.6 78.0 1.95 0.00
DR 1 873.15 1/1 (Gly/CO2) 41.7 62.5 1.99 1.16
SCWG 250 1273.15 1/9 (Gly./H2O) 57.7 75.7 3.21 0.00

The lower molar fraction of syngas was observed for the dry reforming of glycerol (62.5%) at
873.15 K and with a glycerol/CO2 molar ratio of 1.0/1.0, in that conditions the H2/CO molar ratio
was 1.99, however was observed solid carbon formation (1.16 moles of C(s)). Solid carbon formation
was observed in the PO too (0.65 moles of C(s) at 873.15 K and with a glycerol/O2 molar ratio of
1.0/0.5).
Table 3 shows the comparison between all reforming processes with respect to H2 and CO
production and the results obtained for the H2/CO molar ratio. The same glycerol/X (where X are
H2O, O2, CO2 or combinations between then) molar ratio of 1.0/3.0, was used in all simulations
performed.
The SR, PO, ATR and DR were simulated at 1 bar and the SCWG was simulated at 250 bar
and three operational temperatures (873.15, 973.15 and 1073.15 K) were evaluated.
Analyzing the Table 3 it is possible to verify that the highest concentrations of H2 are
observed in the steam reforming reaction. Still in the Table 3 it’s possible to verify that the SR and
ATR presented H2/CO molar ratio close to 2.0 in the conditions in which the reactions were
analyzed. The SCWG presented the highest H2/CO molar ratio between all reforming processes
analyzed.
Another interesting effect is that the elevation of operational temperature results in reductions
of the H2/CO molar ratio in the product, basically by inducing the production of CO in the system.
This behavior was observed in all reforming processes analyzed.

Table 3. Comparison between all reforming technologies of glycerol.

Reforming Molar fraction (mol%) Molar ratio
Temperature (K) Pressure (bar)
technology H2 CO Syngas H2/CO
873.15 250 8.5 1.6 10.1 5.34
SCWG 973.15 250 15.8 4.7 20.6 3.34
1073.15 250 25.3 10.5 35.7 2.42
873.15 1 55.0 14.8 69.8 3.72
SR 973.15 1 61.7 23.0 84.8 2.68
1073.15 1 61.9 26.5 88.4 2.34
873.15 1 46.1 14.5 60.6 3.17
ATR 973.15 1 47.8 21.0 68.8 2.28
1073.15 1 46.3 25.3 71.5 1.84
873.15 1 19.8 6.8 26.6 2.92
OR 973.15 1 17.9 9.5 27.4 1.88
1073.15 1 16.0 12.0 28 1.34
873.15 1 29.0 24.7 53.7 1.17
DR 973.15 1 32.1 43.7 75.8 0.73
1073.15 1 31.8 47.7 79.4 0.67

4. Conclusion
A thermodynamic analysis for hydrogen and syngas production from glycerol using different
reforming technologies has been performed. The SR, PO, ATR, DR and SCWG were evaluated. The
composition of the products was calculated using the Gibbs energy minimization approach in
combination with the virial equation of state.
The model was validated with simulated and experimental data obtained in literature with
good agreement between then.
Hydrogen yields were found to increase directly with temperature in all reforming processes
analyzed. The increase of the operating pressure results in large amounts of solid carbon in the
product stream in the partial oxidation and dry reforming processes. In the other processes the
formation of solid carbon was not observed. The syngas production with H2/CO molar ratio close to
2.0 was found without solid carbon formation only in the steam reforming of glycerol.
The proposed model implemented in the GAMS software and solved with the CONOPT
solver show to be robust, fast and reliable to perform the thermodynamic analysis of the different
glycerol reforming technologies analyzed.
References
[1] Saxena RC, Seal D, Kumar S, Goyal HB. (2008), “Thermo-chemical routes for hydrogen rich gas
from biomass: A review”. Renewable and Sustainable Energy Reviews 12: 1909-27.
[2] Ma F, Hanna MA. (1999), “Biodiesel production: a review”. Bioresource Technology 70: 1-15.
[3] Pagliaro M, Ciriminna R, Kimura H, Rossi M, Della Pina C. (2007), “From glycerol to value
added products”. Angewandte Chemie International Edition 46: 4434-40.
[4] Vaidya PD, Rodrigues AE. (2009), “Glycerol reforming for hydrogen production: a review”.
Chemical Engineering & Technology 32: 1463-69.
[5] Zhang B, Tang X, Li Y, Xu Y, Shen W. (2007), “Hydrogen production from steam reforming of
ethanol and glycerol over ceria supported metal catalysts”. International Journal of Hydrogen Energy
32: 2367-73.
[6] Buffoni IN, Pompeo F, Santori GF, Nichio NN. (2009), “Nickel catalysts applied in steam
reforming of glycerol for hydrogen production”. Catalysis communications 10: 1656-60.
[7] Adhikari S, Fernando SD, Haryanto A. (2008), “Hydrogen production from glicerin by steam
reforming over nickel catalysts”. Renewable Energy 33: 1097-1100.
[8] Thyssen VV, Maia TA, Assaf EM. (2013), “Ni supported on La2O3-SiO2 used to catalyze
glycerol steam reforming”. Fuel 105: 358-63.
[9] Iriondo A, Barrio VL, Cambra JF, Arias PL, Guemez MB, Sanchez MC,Navarro RM, Fierro
JLG. (2010), “Glycerol steam reforming over Ni catalysts supported on ceria and ceria promoted
alumina”. International Journal of Hydrogen Energy 35: 11622-33.
[10] Pompeo F, Santori GF, Nichio NN. (2011), “Hydrogen production by glycerol steam reforming
with Pt/SiO2 and Ni/SiO2 catalysts”. Catalysis Today 172: 183-88.
[11] Pompeo F, Santori GF, Nichio NN. (2010), “Hydrogen and/or syngas from steam reforming of
glycerol. Study of platinum catalysts”. International Journal of Hydrogen Energy 35:8912-20.
[12] Wang H, Wang X, Li M, Li S, Wang S, Ma X. (2009), “Thermodynamic analysis of hydrogen
production from glycerol autothermal reforming”. International Journal of Hydrogen Energy 34:
5683-90.
[13] Wang W. (2010), “Thermodynamic analysis of glycerol partial oxidation for hydrogen
production”. Fuel Processing Technology 91: 1401-08.
[14] Yang G, Yu H, Peng F, Wang H, Yang J, Xie D. (2011), “Thermodynamic analysis of hydrogen
generation via oxidative steam reforming of glycerol”. Renewable Energy 36: 2120-27.
[15] Byrd AJ, Pant KK,Gupta RB. (2008), “Hydrogen production from glycerol by reforming in
supercritical water over Ru/Al2O3 Catalyst”. Fuel 87: 2956-60.
[16] Chakinala AG, Brilman DWF, van Swaaij WPM, Kersten SRA. (2009), “Catalytic and non
catalytic supercritical water gasification of microalgae and glycerol”. Industral & Engineering
Chemistry Research 49: 1113-22.
[17] Guo S, Guo L, Cao C, Yin J, Lu Y, Zhang X. (2012), “Hydrogen production from glycerol by
supercritical water gasification in a continuos flow tubular reactor”. International Journal of
Hydrogen Energy 37: 5559-68.
[18] van Bennekon JG, Venderbosch RH, Assink D, Heeres HJ. (2011), “ Reforming of methanol
and glycerol in supercritical water”. The Journal of Supercritical Fluids 58: 99-113.
[19] Voll FAP, Rossi CCRS, Silva C, Guirardello R, Souza ROMA, Cabral VF, Cardozo-Filho L.
(2009), “Thermodynamic analysis of supercritical water gasification of methanol, ethanol, glycerol,
glucose and cellulose”. International Journal of Hydrogen Energy 34: 9737-44.
[20] Wang X, Li M, Wang M, Wang H, Li S, Wang S, Ma X. (2009), “Thermodynamic analysis of
glycerol dry reforming for hydrogen and synthesis gas production”. Fuel 88: 2148-53.
[21] Fernández Y, Arennilas A, Bermúdez JM, Menéndez JA. (2010), “Comparative study of
conventional and microwave assisted pyrolysis, steam and dry reforming of glycerol for syngas
production, using a carbonaceous catalyst”. Journal of Analytical and Applied Pyrolysis 88: 155-59.
[22] Kale GR, Kulkarni BD. (2010), “Thermodynamic analysis of dry autothermal reforming of
glycerol”. Fuel Processing Technology 91: 520-30.
[23] Valliyappan T, Bakhshi NN, Dalai AK. (2008), “Pyrolysis of glycerol for the production of
hydrogen or syngas”. Bioresource Technology 99: 4476-83.
[24] Smith WRM, Missen RW. (1982), “Chemical reaction equilibrium analysis: Theory and
algorithms”. Original Edition by JohnWiley & Sons.
[25] Pitzer KS, Curl RF. (1957), “The volumetric and thermodynamic properties of fluids III.
Empirical equation of the second virial coeficient”. J. Am. Chem. Soc. 20: 263-72.
[26] Tsonopoulos C. (1974), “An empirical correlation of second virial coeficients”. AIChE Journal
20: 263-72.
[27] Castello D, Fiori L. (2011), “Supercritical water gasification of biomass: Thermodynamic
constraints”. Bioresource Technologie 102: 7574-82.
[28] Castillo J, Grossmann IE. (1981), “Computation of phase and chemical equilibria”. Computers
& Chemical Engineering 5: 99-108.
[29] Freitas ACD, Guirardello R. (2012), “Oxidative reforming of methane for hydrogen and
synthesis gas production: Thermodynamic equilibrium analysis”. Journal of Natural Gas Chemistry
21: 571-80.
[30] Lu Y, Guo X, Zhang X, Yan Q. (2007), “Thermodynamic modeling and analysis of biomass
gasification for hydrogen production in supercritical water”. Chemical Engineering Journal 131: 233-
44.
[31] Nichita DV, Gomez S, Luna E. (2002), “Multiphase equilibria calculation by direct
minimization of Gibbs free energy with a global optimization method”. Computers & Chemical
Engineering 26: 1703-24.
[32] Rossi CCRS, Berezuk ME, Cardozo-Filho L, Guirardello R. (2011), “Simultaneous calculation
of chemical and phase equilibria using convexity analysis”. Computers & Chemical Engineering 35:
1226-37.
[33] Freitas ACD, Guirardello R. (2012), “Supercritical water gasification of glucose and cellulose
for hydrogen and syngas production”. Chemical Engineering Transactions 27: 361-66.
[34] Brooke K, Meeraus D, Raman R. (1998), GAMS – “A user’s manual”.
[35] Polling BP, Prausnitz JM, O’Connel PJ. (2000), “The properties of gases and liquids”. 5th ed.
New York: McGraw Hill.
[36] Reid RP, Prausnitz JM, Sherwood TK. (1987), “The properties of gases and liquids”. 4th ed.
New York: McGraw Hill.
[37] DIPPR, DIADWM Public v. 1.2. (2000), “Design Institute for physical property data”.
Information and data evaluation manager, 2000.
[38] Edwards JH, Maitra AM. (1995) “The chemistry of methane reforming with carbon dioxide and
its current and potential applications”. Fuel Processing Technologie 42: 269-89.
[39] Adhikari S, Fernando SD, To SDF, Bricka RM, Steele PH, Haryanto A. (2008) “Conversion of
glycerol to hydrogen via a steam reforming process over nickel catalysts”. Energy & Fuels 22: 1220-
26.