Syngas Production from Heavy Liquid
Fuel Reforming in Inert Porous Media
Cambridge University
Department of Engineering
This dissertation is submitted for the degree of
Doctor of Philosophy by:
Andrea Pastore
Fitzwilliam College
1st of July, 2010
Declaration
This dissertation is the result of my own work and includes noth-
ing which is the outcome of work done in collaboration except where
specifically indicated in the text. The dissertation contains approxi-
mately 35,000 words, 70 figures and three tables.
Andrea Pastore
Hopkinson Laboratory, Cambridge
July, 2010
I would like to dedicate this thesis to my family, to my girlfriend and
to all my friends.
Acknowledgements
Firstly, I would like to thank my supervisor Prof. Epaminondas Mas-
torakos for his support, invaluable help, trust and efforts to obtain
funding for my research and staying in Cambridge. I would also like
to thank my advisor Dr. Scott for helping with his valuable advice
and support on the experimental aspects of my project. A special
thank also to Dr. Balachandran for his support before the beginning
of this PhD and to Voller Energy plc for the financial sustainment
during these 3 years. I am also very grateful to the gentlemen of the
workshop: Michael, Robert, Peter, Sam, Ken and John for the ex-
cellent job and suggestions for my experimental rig. I also want to
thank all the colleagues and friends I met in the Engineering Depart-
ment who supported me during these years: Moncef, Christos, An-
tonios, Andy, Giorgio, Teresa, Alex, Giulio, Davide, Cheng, Camille,
Rob, Teresa, Frank, Sam, Simon, Thibault, Ronan, Johannes, Simon,
Marta, Larry, Chris, Jason, James, Christoph and Robin. I want
to acknowledge the invaluable moral support from Udine, from my
family and my old friends. I would also like to express my gratitude
to the Cambridge Volleyball Club, which allowed me to spend some
very exciting and sportful time during my staying in Cambridge. Fi-
nally, a very special thank to my Bettina, who made my last year in
Cambridge so fantastic and special.
Abstract
In the effort to introduce fuel cell technology in the field of decentral-
ized and mobile power generators, a hydrocarbon reformer to syngas
seems to be the way for the market uptake. In this thesis, a po-
tential technology is developed and investigated, in order to convert
commercial liquid fuel (diesel, kerosene and biodiesel) to syngas. The
fundamental concept is to oxidise the fuel in a oxygen depleted envi-
ronment, obtaining hydrogen and carbon monoxide as main products
of the reaction. In order to extend the flammability limit of hydro-
carbon/air mixtures, the rich combustion experiments have been car-
ried out in a two-layer porous medium combustor, which stabilises a
flame at the matrix interface and recirculates the enthalpy of the hot
products in order to enhance the reaction rates at ultra-rich equiva-
lence ratio. This thesis demonstrates the feasibility of the concept, by
exploring characteristic parameters for a compact, reliable and cost
effective device.
Specifically, a range of equivalence ratios, thermal loads and porous
materials have been examined. n-heptane was successfully reformed
up to an equivalence ratio of 3, reaching a conversion efficiency (based
on the lower heating value of H2 and CO over the fuel input) up to
75% for a packed bed of alumina beads. Thermal loads from P=2
to 12 kW at ϕ=2.0 demonstrated that heat losses can be reduced to
10%.
Similarly, diesel, kerosene and bio-diesel were reformed to syngas in
a Zirconia foam burner with conversion efficiency over 60%. The
effect of different burners, thermal loads and equivalence ratios have
also been assessed for these commercial fuels, leading to equivalent
conclusions.
A preliminary attempt to reduce the content of CO and hydrocarbons
in the reformate has been also performed using commercial steam re-
forming and water-gas shift reaction catalysts, obtaining encouraging
results. Finally, soot emission has been assessed, demonstrating par-
ticle formation for all the fuels above ϕ=2.0, with biodiesel showing
the lowest soot formation tendency among all the fuels tested.
List of Publications
Two journal papers and two conference papers have been published
about the results described in this thesis:
• A. Pastore and E. Mastorakos. “Rich n-heptane and Diesel Com-
bustion in Porous Media”, Experimental Thermal and Fluid Sci-
ence, 34:359-365, 2010.
• A. Pastore and E. Mastorakos. “Syngas Production from Liquid
Fuels in a Non-Catalytic Porous Burner”, Fuel, Article in Press,
2010.
• A. Pastore and E. Mastorakos. “Syngas Production from Diesel
Non-Catalytic Partial Oxidation in Inert Porous Media”, Pro-
ceedings of the 4th International Conference on Application of
Porous Media, Istanbul, 2009.
• A. Pastore and E. Mastorakos. “Non-Catalytic Multi-Fuel Re-
former for Syngas Production for Fuel Cell Applications”, AiChE
Spring Meeting, San Antonio, 2010.
Contents
Nomenclature xix
1 Introduction 1
1.1 Background and Motivation . . . . . . . . . . . . . . . . . . . . . 1
1.2 Fuel reforming . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.2.1 Basic concepts . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.2.2 Small-scale reforming . . . . . . . . . . . . . . . . . . . . . 5
1.2.3 Steam reforming . . . . . . . . . . . . . . . . . . . . . . . 6
1.2.4 Partial oxidation . . . . . . . . . . . . . . . . . . . . . . . 10
1.2.4.1 Catalytic POX . . . . . . . . . . . . . . . . . . . 10
1.2.4.2 Thermal POX . . . . . . . . . . . . . . . . . . . . 11
1.2.5 Autothermal reforming . . . . . . . . . . . . . . . . . . . . 12
1.2.6 Comparison of the three techniques . . . . . . . . . . . . . 13
1.2.7 Syngas clean-up and water-gas shift . . . . . . . . . . . . . 14
1.3 Combustion in porous media . . . . . . . . . . . . . . . . . . . . . 15
1.3.1 Rich operation . . . . . . . . . . . . . . . . . . . . . . . . 17
1.3.2 Lean operation . . . . . . . . . . . . . . . . . . . . . . . . 18
1.3.3 Kinetic modeling . . . . . . . . . . . . . . . . . . . . . . . 19
1.4 Scope of Thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
1.4.1 Specific objectives . . . . . . . . . . . . . . . . . . . . . . . 22
1.4.2 Structure of Thesis . . . . . . . . . . . . . . . . . . . . . . 22
2 Experimental methods 24
2.1 Apparatus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.1.1 Burners . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
vii
CONTENTS
2.1.2 Porous materials . . . . . . . . . . . . . . . . . . . . . . . 28
2.1.3 Flow control . . . . . . . . . . . . . . . . . . . . . . . . . . 32
2.1.4 Atomization and vaporization . . . . . . . . . . . . . . . . 32
2.1.5 Ignition . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
2.1.6 Temperature detection . . . . . . . . . . . . . . . . . . . . 36
2.1.7 Gas sample analysis . . . . . . . . . . . . . . . . . . . . . 37
2.1.8 Particulate detection . . . . . . . . . . . . . . . . . . . . . 38
2.2 Fuels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
2.3 Post reforming catalysis . . . . . . . . . . . . . . . . . . . . . . . 49
2.3.1 Steam reforming catalyst . . . . . . . . . . . . . . . . . . . 49
2.3.2 Water-gas shift catalyst . . . . . . . . . . . . . . . . . . . 49
2.4 Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
2.5 Equilibrium calculation . . . . . . . . . . . . . . . . . . . . . . . . 51
2.6 Carbon balance . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
3 n-heptane reforming 53
3.1 Preliminary investigations . . . . . . . . . . . . . . . . . . . . . . 53
3.1.1 Mixture homogeneity . . . . . . . . . . . . . . . . . . . . . 53
3.1.2 Reformate uniformity . . . . . . . . . . . . . . . . . . . . . 57
3.1.3 Operating envelope . . . . . . . . . . . . . . . . . . . . . . 59
3.2 Parametric investigations . . . . . . . . . . . . . . . . . . . . . . . 61
3.2.1 Thermal load effect . . . . . . . . . . . . . . . . . . . . . . 61
3.2.2 Equivalence ratio effect . . . . . . . . . . . . . . . . . . . . 66
3.2.3 Porous material effect . . . . . . . . . . . . . . . . . . . . 78
3.2.4 Burner size effect . . . . . . . . . . . . . . . . . . . . . . . 84
3.3 Heat losses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
3.4 Soot emission . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
3.5 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
3.6 Chapter summary . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
viii
CONTENTS
4 Diesel, Biodiesel and Kerosene reforming 92
4.1 Diesel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
4.1.1 Thermal load . . . . . . . . . . . . . . . . . . . . . . . . . 92
4.1.2 Equivalence ratio . . . . . . . . . . . . . . . . . . . . . . . 96
4.1.3 Porous material effect . . . . . . . . . . . . . . . . . . . . 102
4.1.4 Heat losses . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
4.1.5 Endurance test . . . . . . . . . . . . . . . . . . . . . . . . 106
4.2 Biodiesel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
4.3 Kerosene . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
4.4 Soot emission . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
4.5 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
4.6 Chapter summary . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
5 Syngas clean-up 121
5.1 n-heptane . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
5.1.1 Endurance test . . . . . . . . . . . . . . . . . . . . . . . . 129
5.2 Biodiesel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
5.3 Chapter summary . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
6 Conclusions 133
6.1 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
6.2 Suggestion for further research . . . . . . . . . . . . . . . . . . . . 135
References 149
ix
List of Tables
1.1 Examples of steam reformers integrated with fuel cells. . . . . . . 9
2.1 Porous medium specifications. . . . . . . . . . . . . . . . . . . . . 30
2.2 Fuel specifications. . . . . . . . . . . . . . . . . . . . . . . . . . . 41
2.3 Fig. 2.8 main peak description. . . . . . . . . . . . . . . . . . . . 44
2.4 Fig. 2.9 main peak description. . . . . . . . . . . . . . . . . . . . 45
x
List of Figures
1.1 U.S. Primary Energy Flow by Source and Sector [1]. . . . . . . . . 2
1.2 Steps in reforming process. Source [2]. . . . . . . . . . . . . . . . 6
1.3 Solid and gas temperature and volumetric heat release for inlet gas
velocity of 60 cm/s and ϕ=0.65. Source [3]. . . . . . . . . . . . . 16
1.4 Computational structure of the reaction zone for n-heptane/air
combustion at ϕ=2.5 and v=60 cm/s. Source [4]. . . . . . . . . . 20
1.5 Computed exhaust H2 : CO and C2H2 levels as a function of mix-
ture stoichiometry and for an air preheat temperature of 500◦C.
Source [5]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.1 Equipment setup for burner A, B and C. Burner D has alumina
beads as diffuser layer and burners E and F have a diameter of 100
mm. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.2 70 mm ID quartz tube equipped with a 25 mm layer of 2-3 mm
Al2O3 beads (flame holder) and a 80 mm layer of 6 mm Al2O3
beads (reforming layer). n-heptane/air flame at ϕ=2, P=5 kW
and preheating temperature T1=423 K. The flame on the rim of the
burner is due to the autoignition of the syngas with the laboratory
air. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.3 70 mm ID burner and 100 mm ID burner equipped with spark
plugs, coil stainless steel tube, thermocouples and top hoods. . . . 28
2.4 Porous foams used for the reforming layer. Specifications in Table
2.1 as burners A, B, C, E and F from left to right. . . . . . . . . . 30
xi
LIST OF FIGURES
2.5 Foam failures: (a) ZrO2 foam cracking after thermal cycles; (b)
SiC, Al2O3, SiO2 compound foam melted after high-temperature
operations. The alumina paper wrapping is evident at the side of
the cylindrical piece. . . . . . . . . . . . . . . . . . . . . . . . . . 31
2.6 Exploded view of the DELAVAN AL-15 nozzle. (a) original ver-
sion; (b) modified nozzle; (c) modified nozzle with water line. . . . 34
2.7 Front and top view of the fuel pre-vaporizer. The T1 thermocouple
on top of the cone, the air and fuel inlet and the nozzle are visible
in the two photographs. . . . . . . . . . . . . . . . . . . . . . . . 35
2.8 GC-MS output for Diesel oil. . . . . . . . . . . . . . . . . . . . . . 42
2.9 GC-MS output for Jet A-1. . . . . . . . . . . . . . . . . . . . . . 43
2.10 GC-MS output for RME biodiesel. . . . . . . . . . . . . . . . . . 43
2.11 Shell diesel specifications [6]. . . . . . . . . . . . . . . . . . . . . . 46
2.12 ADM biodiesel specifications [7]. . . . . . . . . . . . . . . . . . . . 47
2.13 CPS kerosene specifications [8]. . . . . . . . . . . . . . . . . . . . 48
2.14 (a) Su¨d-chemie G 90-EW steam reforming catalyst wheels; (b)
Fuelcellsmaterial.com ZDC/Pt catalyst pellets. . . . . . . . . . . . 50
3.1 Reactant mixture over the first porous layer (flame holder) mea-
sured by the FFID and expressed as normalized voltage over the
cross section of the two size burners (70 and 100 mm ID) at P=7
kW and ϕ=2.0. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
3.2 Reactant mixture over the first porous layer (flame holder) mea-
sured by the FFID and expressed as normalized voltage over the
cross section of the 70 mm ID burner at P=7 kW and ϕ=2.0, 2.5
and 3.0. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
3.3 Reactant mixture over the first porous layer (flame holder) mea-
sured by the FFID and expressed as normalized voltage over the
cross section of the 100 mm ID burner at P=7 kW and ϕ=2.0, 2.5
and 3.0. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
3.4 Syngas composition and exit temperature T3 over the reforming
layer of burner E at P= 7 kW and ϕ=2.0. . . . . . . . . . . . . . 58
xii
LIST OF FIGURES
3.5 Operating envelope for burner C for n-heptane reforming at a pre-
heat temperature of 423 K. Laminar burning velocity of free pre-
mixed flames calculated by the code COSILAB [9] and a detailed
chemical mechanism [10]. . . . . . . . . . . . . . . . . . . . . . . . 60
3.6 Calculated equilibrium at the adiabatic flame temperature and the
experimental mole fractions (on a dry basis) for a inlet air/fuel tem-
perature of 423 K at various thermal loads for n-heptane reforming
at ϕ=2.0 for burner A. (a) Hydrogen, carbon monoxide and carbon
dioxide. (b) Measured methane, acetylene, ethylene and ethane. . 63
3.7 Calculated equilibrium at the measured exit temperature T3 and
the experimental mole fractions (on a dry basis) of hydrogen, car-
bon monoxide and carbon dioxide for a inlet air/fuel temperature
of 423 K at various thermal loads for n-heptane reforming at ϕ=2.0
for burner A. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
3.8 Adiabatic and experimental exit temperature T3 for n-heptane re-
forming at conditions as in Fig. 3.6. . . . . . . . . . . . . . . . . . 64
3.9 H2 distribution in hydrogen, water and hydrocarbons present in
n-heptane reformate as a function of thermal load based on pro-
cessing of the measured dry molar fractions. Conditions as in Fig.
3.6. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
3.10 Conversion efficiency for n-heptane reforming at ϕ=2.0 and at var-
ious thermal loads for a preheat temperature of 423 K. . . . . . . 65
3.11 Calculated equilibrium at the adiabatic flame temperature and the
experimental mole fractions (on a dry basis) for a inlet air/fuel
temperature of 423 K at various equivalence ratio for n-heptane
reforming at P=6 kW for burner A. (a) Hydrogen, carbon monox-
ide and carbon dioxide. (b) Measured methane, acetylene, ethylene
and ethane. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
xiii
LIST OF FIGURES
3.12 Calculated equilibrium at the adiabatic flame temperature and the
experimental mole fractions (on a dry basis) for a inlet air/fuel
temperature of 423 K at various equivalence ratio for n-heptane
reforming at P=7 kW for burner A. (a) Hydrogen, carbon monox-
ide and carbon dioxide. (b) Measured methane, acetylene, ethylene
and ethane. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
3.13 Calculated equilibrium at the adiabatic flame temperature and the
experimental mole fractions (on a dry basis) for a inlet air/fuel
temperature of 423 K at various equivalence ratio for n-heptane
reforming at P=8 kW for burner A. (a) Hydrogen, carbon monox-
ide and carbon dioxide. (b) Measured methane, acetylene, ethylene
and ethane. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
3.14 Calculated equilibrium at the adiabatic flame temperature and the
experimental mole fractions (on a dry basis) for a inlet air/fuel
temperature of 423 K at various equivalence ratio for n-heptane
reforming at P=6 kW for burner C. (a) Hydrogen, carbon monox-
ide and carbon dioxide. (b) Measured methane, acetylene, ethylene
and ethane. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
3.15 Calculated equilibrium at the adiabatic flame temperature and the
experimental mole fractions (on a dry basis) for a inlet air/fuel
temperature of 423 K at various equivalence ratio for n-heptane
reforming at P=7 kW for burner C. (a) Hydrogen, carbon monox-
ide and carbon dioxide. (b) Measured methane, acetylene, ethylene
and ethane. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
3.16 Calculated equilibrium at the adiabatic flame temperature and the
experimental mole fractions (on a dry basis) for a inlet air/fuel
temperature of 423 K at various equivalence ratio for n-heptane
reforming at P=8 kW for burner C. (a) Hydrogen, carbon monox-
ide and carbon dioxide. (b) Measured methane, acetylene, ethylene
and ethane. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
xiv
LIST OF FIGURES
3.17 Calculated equilibrium at the adiabatic flame temperature and the
experimental mole fractions (on a dry basis) for a inlet air/fuel
temperature of 423 K at various equivalence ratio for n-heptane
reforming at P=6 kW for burner D. (a) Hydrogen, carbon monox-
ide and carbon dioxide. (b) Measured methane, acetylene, ethylene
and ethane. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
3.18 Calculated equilibrium at the adiabatic flame temperature and the
experimental mole fractions (on a dry basis) for a inlet air/fuel
temperature of 423 K at various equivalence ratio for n-heptane
reforming at P=7 kW for burner D. (a) Hydrogen, carbon monox-
ide and carbon dioxide. (b) Measured methane, acetylene, ethylene
and ethane. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
3.19 Calculated equilibrium at the adiabatic flame temperature and the
experimental mole fractions (on a dry basis) for a inlet air/fuel
temperature of 423 K at various equivalence ratio for n-heptane
reforming at P=8 kW for burner D. (a) Hydrogen, carbon monox-
ide and carbon dioxide. (b) Measured methane, acetylene, ethylene
and ethane. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
3.20 (a) H2 distribution in hydrogen, water and hydrocarbons present
in n-heptane reformate as a function of equivalence ratio based on
processing of the measured dry molar fractions. (b) Conversion
efficiency defined by Eq. 3.1 against the equivalence ratio for n-
heptane. Conditions as in Fig. 3.12. . . . . . . . . . . . . . . . . . 77
3.21 Experimental mole fractions (on a dry basis) of (a) H2, (b) CO
and (c) total C1 (contributions from CH4, C2H2, C2H4 and C2H6)
for burners A, B, C and D at conditions as in Fig. 3.12. . . . . . . 80
3.22 Conversion efficiency, as defined by Eq. 3.1, for burners A, B, C
and D at conditions as in Fig. 3.12. . . . . . . . . . . . . . . . . . 81
3.23 Exit temperature (T3) and conversion efficiency, as defined by Eq.
3.1, for burners A and D at conditions as in Fig. 3.12. . . . . . . . 81
xv
LIST OF FIGURES
3.24 Experimental mole fractions (on a dry basis) for an inlet air/fuel
temperature of 423 K at various equivalence ratios for n-heptane
reforming at P=7 kW for burners E and F. (a) Hydrogen, carbon
monoxide and carbon dioxide. (b) total C1 (contributions from
CH4, C2H2, C2H4 and C2H6). . . . . . . . . . . . . . . . . . . . . 83
3.25 Experimental mole fractions (on a dry basis) for an inlet air/fuel
temperature of 423 K at various equivalence ratios for n-heptane
reforming at P=7 kW for burners A and E. (a) Hydrogen, carbon
monoxide and carbon dioxide. (b) total C1 (contributions from
CH4, C2H2, C2H4 and C2H6). . . . . . . . . . . . . . . . . . . . . 85
3.26 Exit temperature (T3) for burners A and E at condition as in Fig.
3.12. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
3.27 Heat losses as a percentage of the thermal load vs. thermal load.
Burner A and ϕ=2.0. . . . . . . . . . . . . . . . . . . . . . . . . . 87
3.28 Soot emissions measured by a filter-method machine (FSN defined
by Eq. 2.1) for n-heptane over a range of equivalence ratio for
burner A and D at P=7 kW. . . . . . . . . . . . . . . . . . . . . . 88
4.1 Experimental mole fractions (on a dry basis) for a inlet air/fuel
temperature of 573 K at various thermal loads for diesel reforming
at ϕ=2.0 for burner A. (a) Hydrogen, carbon monoxide and carbon
dioxide. (b) Methane, acetylene, ethylene and ethane. . . . . . . . 94
4.2 Experimental exit temperature T3 and conversion efficiency for
diesel reforming at conditions as in Fig. 4.1. . . . . . . . . . . . . 95
4.3 H2 distribution in hydrogen, water and hydrocarbons present in
diesel reformate as a function of thermal load based on processing
of the measured dry molar fractions. Conditions as in Fig. 4.1. . . 95
4.4 Experimental mole fractions (on a dry basis) for a inlet air/fuel
temperature of 573 K at various equivalence ratio for diesel re-
forming at P=7 kW for burner A. (a) Hydrogen, carbon monoxide
and carbon dioxide. (b) Methane, acetylene, ethylene and ethane. 98
xvi
LIST OF FIGURES
4.5 Experimental mole fractions (on a dry basis) for a inlet air/fuel
temperature of 573 K at various equivalence ratio for diesel re-
forming at P=7 kW for burner C. (a) Hydrogen, carbon monoxide
and carbon dioxide. (b) Methane, acetylene, ethylene and ethane. 99
4.6 (a) H2 distribution in hydrogen, water and hydrocarbons present
in diesel reformate as a function of equivalence ratio based on pro-
cessing of the measured dry molar fractions. (b) Conversion effi-
ciency defined by Eq. 3.1 against the equivalence ratio for diesel.
Conditions as in Fig. 4.4. . . . . . . . . . . . . . . . . . . . . . . . 100
4.7 Flame on the top of the reforming layer of burner A in diesel re-
forming at P=7 kW and ϕ=3.0. . . . . . . . . . . . . . . . . . . . 101
4.8 Experimental mole fractions (on a dry basis) of (a) H2, (b) CO
and (c) C1 (contributions from CH4, C2H2, C2H4 and C2H6) for
burner A and D at conditions as in Fig. 4.4. . . . . . . . . . . . . 104
4.9 Conversion efficiency, as defined by Eq. 3.1, for burners A and D
at condition as in Fig. 4.4. . . . . . . . . . . . . . . . . . . . . . . 104
4.10 Heat losses for burner A for a thermal load ranging from P=2 to
12 kW and ϕ=2.0 at condition as in Fig. 4.4. . . . . . . . . . . . 105
4.11 H2, CO and CO2 experimental dry molar fractions of diesel re-
forming at P=7 kW and ϕ=2.0 over time at condition as in Fig.
4.4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
4.12 Conversion efficiency of diesel reforming at P = 7 kW and ϕ=2.0
over time at condition as in Fig. 4.4. . . . . . . . . . . . . . . . . 107
4.13 Experimental mole fractions (on a dry basis) for a inlet air/fuel
temperature of 523 K at various equivalence ratio for RME biodiesel
reforming at P=7 kW for burner A. (a) Hydrogen, carbon monox-
ide and carbon dioxide. (b) Methane, acetylene, ethylene and ethane.109
4.14 Experimental mole fractions (on a dry basis) for a inlet air/fuel
temperature of 523 K at various equivalence ratio for RME biodiesel
reforming at P=7 kW for burner C. (a) Hydrogen, carbon monox-
ide and carbon dioxide. (b) Methane, acetylene, ethylene and ethane.110
4.15 Conversion efficiency, as defined by Eq. 3.1, for biodiesel reforming
in burners A and C at condition as in Fig. 4.13. . . . . . . . . . . 111
xvii
LIST OF FIGURES
4.16 Calculated equilibrium at the adiabatic flame temperature and the
experimental mole fractions (on a dry basis) for a inlet air/fuel tem-
perature of 523 K at various equivalence ratio for Jet A-1 kerosene
reforming at P=7 kW for burner A. (a) Hydrogen, carbon monox-
ide and carbon dioxide. (b) Methane, acetylene, ethylene and ethane.113
4.17 Soot emissions measured by a filter-method machine (FSN de-
fined by Eq. 2.1) for n-heptane, Diesel oil, Jet A-1 Kerosene and
Biodiesel over a range of equivalence ratio for burner A at P=7 kW.114
4.18 Experimental mole fractions (on a dry basis) of C1 (contributions
from CH4, C2H2, C2H4 and C2H6) for burner A at P=7 kW for
diesel, biodiesel and kerosene. . . . . . . . . . . . . . . . . . . . . 116
4.19 Conversion efficiencies of kerosene, diesel and RME biodiesel re-
forming, as defined by Eq. 3.1, at P=7 kW for burner A. . . . . . 116
5.1 Calculated equilibrium at the adiabatic flame temperature and the
experimental mole fractions (on a dry basis) for steam reforming
and water-gas shift reaction experimental results of n-heptane re-
formate obtained at P=7 kW and ϕ=2.0 in burner E. (a) ceramic
inert foam only; (b) water injection (m˙H2O/m˙C7H16=1); (c) steam
reforming with no extra water injected; (d) steam reforming with
water injection m˙H2O/m˙C7H16=1); (e) steam reforming and WGSR
with water injection (m˙H2O/m˙C7H16=1); (f) steam reforming and
WGSR with water injection (m˙H2O/m˙C7H16=2). . . . . . . . . . . 124
5.2 Calculated equilibrium at the adiabatic flame temperature and the
experimental mole fractions (on a dry basis) for steam reforming
and water-gas shift reaction experimental results of n-heptane re-
formate obtained at P=7 kW and ϕ=2.5 in burner E. (a) ceramic
inert foam only; (b) water injection (m˙H2O/m˙C7H16=1); (c) steam
reforming with no extra water injected; (d) steam reforming with
water injection m˙H2O/m˙C7H16=1); (e) steam reforming and WGSR
with water injection (m˙H2O/m˙C7H16=1); (f) steam reforming and
WGSR with water injection (m˙H2O/m˙C7H16=2). . . . . . . . . . . 125
xviii
LIST OF FIGURES
5.3 Off-gas flame at the burner’s outlet for n-heptane at P=7 kW and
ϕ=2.0 in burner E at configuration (a) and (b). . . . . . . . . . . 126
5.4 Off-gas flame at the burner’s outlet for n-heptane at P=7 kW and
ϕ=2.0 in burner E at configuration (c) and (d). . . . . . . . . . . 127
5.5 Off-gas flame at the burner’s outlet for n-heptane at P=7 kW and
ϕ=2.0 in burner E at configuration (e) and (f). . . . . . . . . . . 128
5.6 Experimental mole fractions (on a dry basis) over time for steam
reforming (configuration (c) in Fig. 5.1) of n-heptane at P=7 kW
and ϕ=2.5 in burner E. . . . . . . . . . . . . . . . . . . . . . . . . 129
5.7 Experimental mole fractions (on a dry basis) for steam reforming
and water-gas shift reaction experimental results of biodiesel refor-
mate obtained at P=7 kW and ϕ=2.5 in burner E. Configurations
(a), (c) and (f), as defined in Fig. 5.1. . . . . . . . . . . . . . . . 131
xix
Chapter 1
Introduction
1.1 Background and Motivation
Energy consumption has dramatically increased during the last 30 years, by al-
most doubling the energy demand estimated in million tonnes of oil equivalent
[11]. The electricity generation and the transportation system are mainly based
on the combustion of fossil fuels, generally defined as oil, coal and natural gas,
as shown in Fig. 1.1.
There are several issues to be considered about fossil fuel consumption. First
of all, the greenhouse gas emission, due to hydrocarbon combustion, has obviously
followed the rising trend of fossil fuel utilization since the current energy infras-
tructure is considered as the main source of anthropogenic emissions. It is clear
that CO2 is the main pollutant, being one of the main products, together with
water, in complete combustion processes. The concentration of carbon dioxide
present in the environment is regulated by the carbon flow, which is constituted
by the flux from the atmosphere to the land and oceans through natural pro-
cesses such as plant photosynthesis. Unfortunately, this process can absorb only
partially the CO2 emitted by human activities: of the 6.1 billion metric tons of
carbon monoxide absorbed per year, 3.2 billion metric tons are added annually
unbalancing the absorption/emission ratio [11]. Secondly, oil, coal and natural
gas are not renewable resources and, even though new extraction sites are often
discovered, an alternative solution has to be found [12]. Numerous alternative fu-
1
1.1 Background and Motivation
Figure 1.1: U.S. Primary Energy Flow by Source and Sector [1].
els have been investigated, such as biodiesel, methanol, ethanol, hydrogen, boron,
liquefied petroleum gas (LPG), Fischer-Tropsch fuel and solar fuels.
One of these is hydrogen, the most abundant of the chemical elements, con-
stituting roughly 75% of the universe’s elemental mass. Unfortunately, elemental
hydrogen is relatively rare on Earth, but it may be produced by using the process
of water electrolysis or by hydrocarbon reforming. In this sense, hydrogen cannot
be defined as a fuel, but it is an energy carrier, because it must be produced from
other naturally occurring species at significant cost in energy. The first process
seems to be the most promising for a clean environment solution, through the
utilization of nuclear, solar power or other renewable sources to provide the en-
ergy needed in water bond separation. Nowadays, water electrolysis is limited
by the cost and the emissions involved in upstream electricity generation. The
second option might be an intermediate step, characterised by centralised hy-
drogen generation with carbon capture and storage (CCS) [13]. If the Fuel Cell
Vehicle (FCV) system uses the H2 produced by this process, the CO2 emission
in the transportation field could be decreased to 70-80% of the traditional CO2
emission from the gasoline- or diesel-based IC engine vehicle systems [14]. Both
2
1.1 Background and Motivation
possibilities, resulting in hydrogen production and its subsequent utilization in
energy applications, result in reducing drastically the carbon footprint of the pro-
cesses. Hydrogen might be used for power generation or as a transport fuel, for
example in association with fuel cells. As an energy carrier, it could be also a
suitable substitute for electricity; examples of H2 as an alternative energy carrier
from solar or nuclear plant are given by [15], by claiming, for instance, that in
the case of fusion reactors of 30 GW or even 100 GW, hydrogen in pipelines is
the most economically viable solution. Moreover, unlike electricity, hydrogen can
be stored and accumulated.
In the future, hydrogen might also be used for power generation and in trans-
port by fuelling gas turbines and IC engines, thanks to its high LHV (120.1
MJ/kg) and its high flammability (from 4% to 75% of volume in air). Moreover,
it has a high flame speed (order of magnitude higher than gasoline), a high effec-
tive octane number, and no toxicity or ozone-forming potential [16]. The product
of hydrogen combustion with air is water vapor and negligible pollution when the
peak temperature is limited. Nevertheless, the heating value per unit volume is
lower compared to hydrocarbons, because of the low density of H2 (89.9×10
−6
kg/m3 STP). Another drawback linked with the low H2 density is the movement:
the power required to pump hydrogen is around 4.5 times higher than for natural
gas per unit of delivered energy [17]. Hydrogen can be stored on-board a vehicle
as a compressed gas, as a liquid in cryogenic containers, or as a gas bound with
certain metals in metal hydrides. Liquid hydrogen is not a cheap solution, since
the electricity required for hydrogen liquefaction at -253◦C and 1 atm is about
35-43 MJ/kg H2. Also refuelling stations are extremely expensive at the moment
and decentralised hydrogen production, even if it is the best choice for the market
uptake, is less efficient than large-scale centralised production and it makes CCS
impractical [17]. However, the estimation of the energy supply and demand in
2030 calls for an investment which is comparable to the estimated cost of the
power system replacement for the hydrogen production and distribution [17].
Currently, the cheapest and most popular sources of H2 are natural gas re-
forming (48% of the total hydrogen produced) and coal gasification (18% of the
total hydrogen produced) without CCS, and it is used primarily in the chemical
and refinery industries [16; 17; 18]. Since hydrogen production from renewable
3
1.2 Fuel reforming
and nuclear energy needs more research and development, the hydrocarbon de-
pendence will remain still predominant for a long period. In this context, this
research focuses on the reforming of logistic heavy liquid fuel to syngas, in or-
der to obtain hydrogen for a mobile power generator system, for example in an
integrated fuel processor/fuel cell device.
One of the most promising hydrogen applications is the one connected to
fuel cells [19; 20]. The major advantages of fuel cells can be summarized by the
following characteristics [21]:
• High energy conversion efficiency;
• No emissions, or extremely low emissions of pollutants;
• Low noise pollution;
• Process simplicity for conversion of chemical to electrical energy.
The most attractive fuel cell feature is its independence from the intrinsic
efficiency limitation of thermal processes recognised by Carnot [22] and the al-
most complete absence of emissions. Modern IGCC power stations operate at
efficiencies of about 50% [23], limited by the top temperature of the cycle and
the typical overall fuel efficiency of gasoline-powered cars is around 12% and 15%
for diesel-powered cars [24]. Fuel cells are electrochemical devices and in theory,
devices could be constructed with an efficiency of 80-90% [25]. It must be noted
that higher energy efficiency itself leads to a reduction in emissions. The absence
of mechanical steps also leads to other advantages such as the minimum require-
ment of rotating accessories pumps and motors demonstrating a low level of
noise and vibrations.
1.2 Fuel reforming
1.2.1 Basic concepts
While waiting for the hydrogen economy led by water electrolysis through re-
newable energy sources - as the ultimate hydrogen economy prediction for the
4
1.2 Fuel reforming
long term - the transitional approach seems to be the reformation of the avail-
able hydrocarbon fuels to fuel cell quality hydrogen [26]. Distributed hydrogen
production would be attractive in the early stages of a hydrogen economy. Thus,
small-scale reformers are the key technology for this first step [27]. In November
2000, DaimlerChrysler unveiled the NECAR 5 prototype automobile, which uti-
lizes methanol as a fuel with a fuel processor and a 75-kW fuel cell. Similarly, fuel
cells coupled to fuel reformers can be convenient generators for residential homes,
commercial buildings and building sites [2]. Nevertheless, the net efficiency of the
fuel cell system is reduced by using fuel processing for hydrogen production in
multiple stages (sulfur removal, fuel reforming, syngas clean-up), but it will still
show a significant efficiency advantage [2].
Eventually, when fuel cells and hydrogen demand will build up, a switch
can be made to central hydrogen production, by using fossil sources with CO2
sequestration and finally by the use of low carbon primary sources. Since the
infrastructure for producing and distributing fossil fuels is well established, the
research and development to introduce fuel cell and hydrogen utilisation in the
transport and residential power generation market is focused on this area [28].
This is the main reason driving the research described in this document, which
describes the development and performance of a liquid fuel processor.
1.2.2 Small-scale reforming
An on-board/on-site fuel processor should meet the following requirements:
• Hydrogen production levels smaller than those in chemical plants;
• Severe constraints on size and weight;
• Ability to cycle through frequent start-ups and shutdowns;
• Hydrogen production rate should be responsive to changes in demand;
• Strict cost target.
The reformate from a fuel processor generally contains 40-80% H2 in volume on
a dry basis [26] depending on the reforming technique and fuel adopted. From
5
1.2 Fuel reforming
Figure 1.2: Steps in reforming process. Source [2].
the same reference, the authors underline that the fuel cell anode operates with
80-85% H2 utilisation. They propose that the balance of the unconverted H2 can
be oxidized and its heat can be recovered. For example, they suggest that this
recovered heat can be used to preheat the incoming water to produce steam or
to preheat fuel and air prior to injection into the reactor. The main steps and
options for on-site and on-board processing of liquid and gaseous hydrocarbons
are shown in Fig. 1.2.
The potential candidates for on-site and on-board reforming are gasoline,
diesel and jet fuels together with natural gas and carbon-neutral resources like
bio-fuels, since they all have existing infrastructure of manufacturing and distri-
bution. However, the problem associated with heavy fuel processing is mainly
the low H/C ratio, high energy consumption in fuel evaporation, the high tem-
perature at fuel reforming and the ease of carbon deposition and soot emission
[29]. In the next sub-sections, the three main different fossil fuel reforming meth-
ods will be covered, namely, steam reforming, partial oxidation and autothermal
reforming.
1.2.3 Steam reforming
Steam reforming (SR) is the most common technique to produce hydrogen in
the chemical process industry. One of the earliest steam reforming processes was
developed in Germany by IG Farbenindustrie in 1926 [30]. An example of an early
commercial steam reformer is a laboratory-scale device from Johnson-Matthey,
developed in the 1970s, which could produce high-purity hydrogen using a Pd
diffuser to purify the product gas [31].
6
1.2 Fuel reforming
In catalytic steam reforming, the fuel reacts with steam in the presence of
a catalyst, producing hydrogen and carbon monoxide. A global reaction can be
written as:
CmHn +mH2O = mCO + (n/2 +m)H2 (1.1)
The steam reforming reaction is strongly endothermic and reactors are limited
by heat transfer, rather than by reaction kinetics [32]. Therefore steam reactors
are in general designed to promote heat exchange and hence they tend to be
large and heavy, which is an acceptable drawback since the heat inertia of this
system, rapid start and dynamic response are not usually part of the prerogatives
of steam reforming. Size and weight are not desirable for a reformer involving
transient duty cycles. Consequently, steam reformers are well suited for long
periods of steady state operation, showing high concentration of hydrogen in the
reformate. Another advantage for SR is that high pressure reformate can be
generated by pumping liquid fuel and water without the need to compress air.
Active catalysts should maximize hydrogen selectivity and inhibit coke formation
as well as CO production [26]. The catalyst utilization in steam reforming is
summarized by [33]. The authors report the wide use of nickel based catalysts,
which are prone to coke formation and sulfur poisoning. The production of coke
accelerates the degradation of the catalyst, until the catalyst is disintegrated and
reforming becomes impossible. Similarly, it is reported that sulfur poisons nickel-
based catalysts by forming metal sulfides and preventing reactants from adsorbing
at the surface. Consequently, the alternative is the implementation of noble metal
catalyst i.e. ruthenium, rhodium-based, which improves the tolerance to sulfur
and limiting coke formation but also increases dramatically the cost of the reactor
[34].
The steam to carbon ratio (S/C) plays an important role. Steam prevents solid
carbon formation on the catalyst, it pushes the equilibrium towards the products,
leading to a higher H2 production and finally it acts as diluent reducing the local
gradients in the reforming catalyst. However, high S/C ratios involve an increase
in reactor size and the production and superheating of steam is at the expense of
the fossil fuel to be reformed, lowering the global efficiency of the process [35].
As discussed previously, most of the hydrogen produced comes from natural
gas steam reforming [17], but alcohol fuels are becoming an attractive alternative
7
1.2 Fuel reforming
because of the possible production of ethanol and methanol from biomass. Specifi-
cally, methanol steam reforming can be carried out at 250◦C [36], whereas ethanol
must be reformed above 300◦C [37], by using the proper catalyst for each feed-
stock. Ethanol is the most attractive, since it is less toxic than methanol. How-
ever, even methanol steam reforming is tremendously attractive, as the increasing
number of publications and extensive research demonstrate [38; 39; 40; 41] and
[31]. The advantage of methanol SR over the other reforming techniques consists
in the low CO production, a feature which cannot be translated to other hydro-
carbon fuels. Compared to heavier fuels, methanol does not contain carbon to
carbon bonds, it has a low boiling point (65◦C), it has a high H/C ratio (4:1) and
it can be converted to hydrogen at lower temperature than other fuels with the
aid of a sufficiently active catalyst. Nevertheless, the main drawbacks of these
two alcohol fuels are the low energy density (about half of the non-oxygenate
hydrocarbons) and the lack of infrastructure for their distribution.
The definition of thermal efficiency for the steam reforming process is not triv-
ial and it mainly depends on the heat source for driving the endothermic reaction.
Specific choices must be taken: the heat can be provided by using supplemental
fuel or recycled product gases. It seems quite clear that reformers do not com-
plete the global reaction, but make some CO and CH4 as products. Moreover,
practical systems cannot separate all the H2 from the reformate stream, so there
is some residual that can be used as fuel. It makes sense to burn the reformate
mixture instead of wasting its heating value, as experimentally demonstrated by
[42]. Some examples of steam reformers coupled to fuel cells from the literature
are shown in Table 1.1.
In conclusion, the challenges in steam reforming are mainly the design of the
heat transfer system and, above all, the choice of the catalyst for the specific
hydrocarbon to be reformed.
8
1.2 Fuel reforming
Table 1.1: Examples of steam reformers integrated with fuel cells.
Firing
rate
Fuel Efficiency a
Gas cleaning
system
Fuel cell Ref.
1 kW NG, LPG 77% (full load) PROX 2.5 kW PEMFC [43]
2-8 kW NG, LPG 67% Pd membrane 5 kW PEMFC [44]
- FT diesel 58% WGS and PSA 4 kW PEMFC [45]
300 W Methanol 56.7% PROX 150 W PEMFC [46]
- NG 64.9% Pd membrane 5 kW PEMFC [47]
abased on H2 LHV
9
1.2 Fuel reforming
1.2.4 Partial oxidation
Partial oxidation (POX) of hydrocarbons, characterised by a limited supply of
oxidiser, is an exothermic reaction. Thus, compared to steam reformers, sim-
pler reactors can be developed, since internal heat exchangers are not necessary.
The first studies about partial oxidation started in 1946, where experiments were
undertaken with Ni catalyst. The authors assumed a complete combustion of
methane with oxygen and a subsequent reaction of CO2 and H2O with the unre-
acted methane yielding eventually syngas [48]. On the contrary, other authors [49]
concluded that carbon monoxide and hydrogen are formed as primary reaction
products. Another early significant work on partial oxidation is [50]. Working
with reactors at high temperatures (750◦C) and short residence time (0.1 s), a
methane and oxygen conversion to syngas above 90% over precious metal cata-
lyst was achieved. Hydrocarbon feedstocks are oxidized to produce hydrogen and
carbon monoxide according to:
CmHn + (0.5)mO2 = mCO + (0.5)nH2 (1.2)
The main advantage of the partial oxidation technique is the small size of the
reactor and the rapid response to changes. However, the main disadvantages lie
with the low H2 : CO ratio and the danger of handling premixed air/fuel mixtures
[34]. Moreover, if the fuel is oxidised by air, a high fraction of nitrogen dilutes the
hydrogen content of the reformate. The partial oxidation reforming technique can
be divided in two subcategories: CPOX (catalytic partial oxidation) and TPOX
(thermal partial oxidation) since catalysts are not vital for POX, because of the
high temperature involved. The partial oxidation reforming technique is the only
one of the three that can be performed without the aid of a catalyst.
1.2.4.1 Catalytic POX
In the CPOX the hydrogen yield can be enhanced compared to the TPOX by
using a properly designed catalyst [51]. Moreover, the use of catalyst offers the
advantage of lower temperatures and no soot formation or unwanted by-products.
Lower temperatures favour the water-gas shift reaction downstream the reforming
process; it will mean also less fuel consumed during start-up and a wider choice
10
1.2 Fuel reforming
of reactor materials, which can lower the device cost. Higher temperatures re-
quire more thermal integration and more insulation adding to the size and costs.
Catalytic partial oxidation is more difficult to control and the main issues include
over-heating or hot spots due to the exothermic nature of the reactions and cok-
ing problem [2]. Gaseous fuels CPOX has been demonstrated by processing on
a Rh catalyst n-butane at T=450◦C [52], propane at T=700◦C [53] and many
works focused on methane reforming with this technique [54; 55; 56]. Catalytic
partial oxidation of liquid higher hydrocarbons such as diesel and gasoline surro-
gates and desulfurized heavy liquid fuels has been demonstrated by [57; 58] and
[59], suggesting that reforming into hydrogen will be possible in the future. The
results and details from these papers are discussed and compared to this project
in Section 4.5.
1.2.4.2 Thermal POX
TPOX shows more flexibility regarding the fuel to be reformed and more robust
and cheaper reactors compared to CPOX. Thermal partial oxidation reformers are
characterized by flame temperatures of about 1300-1500◦C [2] but the optimal op-
erating range for syngas production lies in the ultra-rich regime close to or beyond
the conventional rich flammability limits. This limit has been overcome through
several experimental works using filtration wave, parallel channel, spouted bed
combustor and two-layer porous burner concepts [60; 61; 62; 63; 64; 65]. These
non-catalytic reforming techniques enhance reaction rates by increasing reaction
temperatures through internal heat transfer. Cool flame oxidation for heavy
hydrocarbon breakdown techniques [27; 66] were used in fossil fuels partial ox-
idation reforming, especially for heavy liquid fuels such as diesel. In [65], the
authors reformed methane/air mixtures with a proportion of fuel of 40% above
the normal rich flammability limit, thanks to the heat recirculation due to the
spouted bed. They also suggest the technique as a way to reform coal and heavy
fuel oil. The parallel channel concept has been developed by Ellzey et al. [63],
who successfully reformed propane/air mixtures in a meso-scale reformer. The
reformer consists of parallel channels separated by conducting walls. By alternat-
ing the flow directions between adjacent channels, counter-flow heat exchange is
11
1.2 Fuel reforming
exploited to preheat the unreacted fuel/air mixture to a temperature level where
a self-sustained reaction zone is viable. Nevertheless, all the works produced on
hydrocarbon reforming to syngas in a non-catalytic partial oxidation reformer are
on the development stage and at laboratory scale. Combustion in porous media
will be discussed later in this Chapter, being one of the main features of this
experimental work and specific details are given in Section 1.3.
1.2.5 Autothermal reforming
Also known as indirect partial oxidation or oxidative steam reforming, autother-
mal reforming (ATR) was firstly introduced by Haldor Topsoe in the late 1950s
[67]. In essence, this process may be considered as a combination of partial oxi-
dation and steam reforming:
CmHn + (0.5)mH2O + 0.25mO2 = mCO + (0.5m+ 0.5n)H2 (1.3)
It involves the introduction of both air and steam into a reforming reactor: the
partial oxidation occurs in an inlet zone of the reactor, providing heat for the
steam reforming reaction occurring in a second zone of the vessel that is packed
with catalyst. The catalyst controls the reaction pathways and determines the
relative extents of the oxidation and steam reforming reactions. No external
heating is required, since the exothermic oxidation reaction provides the heat
necessary for the endothermic steam reforming reaction. ATR can overcome the
load limitations of SR, because the oxidation process produces the heat required
for the endothermic reactions, leading to a rapid response to power demand and
faster start-up. The heat necessary to rapidly heat the reactor can be generated
by operating at a higher than normal oxygen-fuel ratio. A practical example of an
ATR processor is given by [68], describing a reactor capable of reforming gasoline
to produce a stream containing 40% H2 on a dry basis with an efficiency around
70% based on the lower heating value.
Thermodynamic calculations show that autothermal reforming theoretically
yields higher reforming efficiencies than partial oxidation even at lower preheat
temperatures [69]. Within an autothermal reformer heat can be integrated, for
example, to vaporize the water and preheat the steam. In the case of partial
12
1.2 Fuel reforming
oxidation, this heat is waste energy and has to be removed by a cooling sys-
tem. Intuitively, it results that the maximum theoretical efficiency is achieved at
the thermoneutral point, where enough heat is provided to the steam reforming
process to drive the reaction and no energy produced in the exothermic partial
oxidation reaction is wasted [32]. The maximum achievable reforming efficiency
is around 80% [69].
Three advantages are particularly important:
• Less complicated reactor design and lower reactor weights because of the
less heat integration required;
• Wider choice of reactor materials;
• Lower fuel consumption during start-up.
An automotive application has been developed by a work sponsored by Re-
nault and Nuvera Fuel Cells [70], showing hydrogen concentrations around 40%
and CO level below 100 ppm using ethanol or gasoline as hydrocarbon feed stock.
The reaction can in theory be completely balanced, but in practice a little excess
of air is added to compensate the heat losses and to obtain a rapid response.
The difference between the efficiency of the ATR and SR consists in the loss
represented by the latent heat of vaporization of the water that escapes with the
combustion products in the steam reforming system [32], when part of the fuel
feed is used to combusted to generate heat for the reforming process. In au-
tothermal reforming operations, the H2O/CmHnOp and O2/CmHnOp ratios play
a fundamental role in thermal integration (global exothermicity or endothermic-
ity), hydrogen and carbon monoxide yield, coke deposition and soot emission.
In recent years, ATR reformers gained interest as application for APU (aux-
iliary power unit) systems, in order to replace the idle functioning of IC engines
with SOFC or PEMFC, due to stricter regulations, coupled with gasoline or diesel
fuel processors [71; 72; 73; 74].
1.2.6 Comparison of the three techniques
Partial oxidation is much faster than any of the other two reactions - by roughly
2 orders of magnitude compared to steam reforming - and response time in fuel
13
1.2 Fuel reforming
supply is shorter [67]. This feature makes POX attractive for automotive appli-
cations. Moreover, it presents higher simplicity and scalability [32]. The main
drawbacks of POX compared to steam reforming are the lower hydrogen yield and
the dilution of the produced hydrogen by the inert nitrogen, which lowers the hy-
drogen concentration significantly when using air as the oxidiser. Catalytic steam
reforming offers higher hydrogen concentrations: 70-80% hydrogen is achieved for
steam reforming versus 40-50% for partial oxidation and autothermal reforming
on a dry basis in the crude reformate gas [33], but it operates at high tempera-
ture and low space velocity owing to slow kinetics. Even though steam reforming
yields more hydrogen than the other two techniques, POX results superior to the
other systems in terms of the energy cost to produce the same amount of hydro-
gen [51]. The difference in hydrocarbon consumption among the three reforming
systems depends strongly upon the efficiency of the heat exchanger [75]. Reform-
ing through partial oxidation is less expensive than with steam reforming because
of the compactness of the reactors, which do not need heat exchangers [76]. For
practical applications, partial oxidation and autothermal reforming are more at-
tractive because of their higher energy efficiency and small and light hardware
compared to SR. Moreover, POX is the only of the three techniques that allows
the reforming of fossil fuel in absence of catalyst. For this reason, the thermal
POX technique has been chosen as the concept to design the liquid fuel reformer
presented in this thesis.
1.2.7 Syngas clean-up and water-gas shift
Carbon monoxide must be kept under 10 ppm for the requirements of PEM
fuel cell electrodes [2]. At low temperature CO can deactivate the anode catalyst
made of Pt/C. This requirement is initially approached by using a water gas shift
reaction (WGSR), a process discovered in 1780 by the physicist Felice Fontana
[77]:
CO +H2O → CO2 +H2 (1.4)
This reaction is exothermic and independent of pressure, thus it is thermodynam-
ically favoured at low temperatures. For this reason, a general rule is that the
higher the temperature is, the higher the carbon monoxide concentration will be.
14
1.3 Combustion in porous media
In the shift reactor, the hydrogen yield is increased and the content of carbon
monoxide in the reformed gas is reduced to 1-2% approximately with two different
temperature stages (HT-WGSR and LT-WGSR) [2; 78]. The catalyst involved
in driving the reaction must be carefully handled, regarding sulphur and oxygen
content in the reactants, temperature, phyroporicity and hot spot, especially for
exothermic processes (partial oxidation).
Other reactions can also be employed to further reduce the CO content in the
feed:
• Preferential or selective oxidation (through a precious catalyst only CO
is adsorbed and converted in CO2 through a small amount of extra air -
PROX) [79].
• CO methanation (it consists in the hydrogenation of CO using the H2
already present in the gas stream) [80].
• Molecular oxygen bleeding (by injecting low amounts of O2 into a H2 anode
feed stream contaminated by CO, the severe effects of carbon monoxide
could be completely eliminated for fuel cells operating at 80◦C) [81].
• Membranes (Palladium membrane allows only hydrogen to pass through its
crystal lattice when it is heated above 300◦C) [82].
1.3 Combustion in porous media
Combustion in porous media offers a way to establish non-catalytic POX and
is discussed here, as it forms the topic of this thesis. In traditional combustion
applications, the peak temperatures inside a combustion zone are directly related
to the energy content of the unburned air/fuel mixture. The theoretical maximum
temperature is predicted by the chemical equilibrium of an adiabatic process and
is thus called the adiabatic equilibrium temperature. The insertion of a porous
medium in the reaction zone allows the transfer of enthalpy from the burned
gases to the unburned gases. The recirculation of the sensible enthalpy of the
flame products occurs through convection from the hot gas to the inert matrix,
which, in turn, transfers the heat through conduction and radiation to the matrix
15
1.3 Combustion in porous media
Figure 1.3: Solid and gas temperature and volumetric heat release for inlet gas
velocity of 60 cm/s and ϕ=0.65. Source [3].
upstream of the flame, which then heats the incoming unburned mixture [83]. The
higher thermal conductivity of the solid matrix compared to the conductivity of
the gas enhances the effective heat transfer from the products to the reactants.
Through this mechanism, it is possible to achieve a local temperature higher than
the adiabatic flame temperature, while at the exit the temperature is limited by
the first law of thermodynamics (Fig. 1.3). This mechanism leads to reaction
zones effectively operating with substantial amounts of preheat, which results
in faster kinetics and hence higher laminar flame speed (up to 20 times higher
than a corresponding free flame [84]) and in the extension of the flammability
limit of the mixture [85]. At ultra-rich or ultra-lean equivalence ratios the flame
temperature is higher than in free flame conditions, which allows initiation and
chain branching mechanisms to still occur. The superadiabatic effect in porous
media has been observed in 1912 [86] and extensively studied by several authors
[83; 87; 88; 89; 90].
16
1.3 Combustion in porous media
1.3.1 Rich operation
The temperatures achieved in superadiabatic combustion and the fuel-rich con-
ditions provide an environment necessary for thermal cracking and/or partial
oxidation of various hydrocarbons. The extended rich flammability limit and the
higher flame temperatures allow partial oxidation conditions to be established,
thus allowing the stripping of the hydrogen out of the original fuel molecule.
Consequently, the main products of the reactions are molecular hydrogen and
carbon monoxide (syngas), along with nitrogen (since air provides the oxidiser)
and some other species in lower concentration (carbon dioxide and light hydro-
carbons). The main parameters affecting the oxidation behaviour are the thermal
load, the equivalence ratio, the thermal conductivity of the porous matrix (ma-
terial, porosity). Because of the heat recirculation, a balance must be achieved
between heat feedback, heat release and heat losses, such that the effective flame
speed equals the incoming fresh gas velocity. In this case a stationary flame can
be developed. Otherwise, the reforming in porous media can assume an unsteady
(filtration wave) condition. This operating mode has been experimentally demon-
strated for gaseous [62], and liquid fuels [4] and the concept is to alternate the
direction of the flame front by switching alternatively the reactant inlet with the
product outlet. Depending on the gas flow velocity and the equivalence ratio,
wave propagates either in the downstream direction producing super-adiabatic
temperature or in the upstream direction (counter-flow) [91]. Babkin et al. clas-
sified the filtration wave mechanism in five regimes based on the wave velocity
[92], showing that the low velocity regime (10−4 m/s) leads to strong thermal
coupling between gaseous and solid phase. Filtration wave operation may have
the disadvantage of a more complex fluid handling system, whereas stabilised
flames, on the other hand, are easier to control and hence this latter option has
been adopted in this research. A rich flame in porous media can be stabilised
through different techniques [92; 93]:
• cooling of the porous matrix at the desired flame position;
• equalization of flame velocity with the flow speed with ad hoc reactor design;
• heat radiation from a specific position which lowers the flame speed;
17
1.3 Combustion in porous media
• two-layer porous structure with a small pore region (flame holder) and a
large pore region (reforming layer).
The last option has been chosen for this project because of the wider range of
velocity and equivalence ratio flame conditions that can be stabilised compared
to the other techniques. Flame propagation is possible only when the rate of heat
release from the reaction is higher than the heat loss to the surroundings. For
small pore structures, the flame propagation is limited to the upstream direction,
because, for some conditions, the heat losses from the flame in the small-pores
layer are higher than the heat released from the reaction. Moreover, the higher
optical thickness limits the radiative heat transfer in the upstream direction.
Therefore, the flame is quenched and cannot propagate upstream and sits at the
interface of the two porous layers. This concept has been introduced for first
by Babkin et al. [92], and it has been successfully demonstrated for gaseous
[60; 94; 95] and for liquid fuel reforming to syngas [60; 61]. This research focused
on the same principle, but demonstrated for the first time the applicability to
heavy commercial fuels.
1.3.2 Lean operation
As the superadiabatic effect in porous media allows to extend the flammability
limit, there are advantages also in the ultra-lean zone. Porous medium burners
working at ultra-lean equivalence ratios have demonstrated low NOx and CO
emissions compared to traditional burners and hence they obtained wide interest
in the field of radiant heating applications [96]. The subsequent decrease in tem-
perature due to the recirculation of the heat from the hot combustion products
inhibits the formation of nitrogen oxides. NOx may be formed either by the
thermal (Zeldovich) or prompt (Fenimore) mechanism. Because NOx formation
is dependent on the peak temperature, NOx emissions are observed to decrease
with decreasing equivalence ratio [97; 98]. Reduced temperatures in the combus-
tion zone might be predicted to lead to increased emissions of CO and unburned
hydrocarbons (UHC). However, the use of lean mixtures means that this is not
the case. A number of investigations find that CO emissions tend to be very low
(less than 40 ppm) and that they decrease with decreasing equivalence ratio [97].
18
1.3 Combustion in porous media
It is in any case fundamental to avoid hot and cold spots to prevent formation of
NOx, CO and UHC. For these reasons commercial products are already available
in this area, such as in space and water heating, industrial furnaces for glass pro-
duction, food industries, metal heat-treating, coating and paint drying, chemical
processing, paper and wood drying, and food processing [99; 100]. Further lean-
burn applications include the reburn of exhaust gases from existing combustion
systems, and the mitigation of fugitive methane emissions.
1.3.3 Kinetic modeling
Some numerical works have been performed on partial oxidation of hydrocarbons
in porous media. Ellzey et al. studied numerically the conversion of n-heptane in
a porous medium reactor consisting of alumina pellets [4]. The model used treats
the combustion process as one-dimensional and transient conservation equations
are solved for mass, gas-phase energy, solid-phase energy, and gas species. Heat
transfer processes include solid conduction within the porous matrix, convective
heat transfer between the solid and gas, solid radiation and gas-phase conduc-
tion. Transport associated with species diffusion is also included. Chemical
kinetics and transport processes are modeled using CHEMKIN and TRANFIT
with an intermediate complexity mechanism for n-heptane [10]. The significant
finding in this study is the path by which hydrogen is produced in a porous re-
actor, which consists in initial partial oxidation reactions and steam reforming
occurring downstream of the main reaction zone. Small hydrocarbons play a key
role in the steam reforming reaction, because they provide a valuable pathway
to increase hydrogen production. The computational results shown in Fig. 1.4
illustrate the prediction for the structure of a heptane/air reaction zone within
porous media. After the n-heptane is thermally broken down, there is formation
and consumption of small hydrocarbon compounds as the hydrogen concentration
continues to increase. The initial oxidation reactions result in a sharp increase in
hydrogen in the main reaction zone. However, the hydrogen concentration con-
tinues to increase downstream of the main reaction zone due to steam reforming;
evidence of which is illustrated by the peak and subsequent decrease of the water
concentration in the gas. In the process of steam reforming, water is consumed to
19
1.3 Combustion in porous media
Figure 1.4: Computational structure of the reaction zone for n-heptane/air com-
bustion at ϕ=2.5 and v=60 cm/s. Source [4].
form CO, CO2, and H2. The numerical analysis predicted that steam reforming
contributed approximately 18% of the total hydrogen produced.
Similar behavior also appeared for the case of methane partial oxidations using
a batch reactor model, the SENKIN code of CHEMKIN 3.6.2 [64]. This finding
indicates that a long post-oxidation zone is necessary for the oxidation of the
soot precursors and for reactions to complete. It can be also concluded that the
injection of steam in the post oxidation zone will enhance the WGS and oxidized
the soot that could be formed.
In ref. [101], the oxidation of n-hexadecane at ϕ=0.5, 1.0 and 1.5 has been
modeled with a reaction mechanism consisting of 238 species and 1756 reversible
reactions. At ϕ=1.5 and T=1000 K, n-hexadecane preferentially reacts with H
(41.8%), and to a lesser extent with OH (17.1%) and CH3 (16.6%); thermal
decompositions represent 18.3% of the fuel consumption. This value increases to
83.2% at 1200 K.
Trimis et al. [5] performed experimental and numerical investigation of a
TPOX reforming process for methane/air mixtures. The model showed the mo-
lar H2 : CO ratio increasing with decreasing air/fuel ratio (Fig. 1.5) and it was
explained by considering the respective carbon monoxide and hydrogen formation
pathways. Carbon monoxide is formed from reactions involving either the formyl
20
1.3 Combustion in porous media
radical (HCO) or ketene (CH2CO), the former being dominant for leaner mix-
tures while the latter for richer. Both HCO and CH2CO are oxygenated species
and their levels are expected to drop as the mixture becomes richer. At the same
time, rich conditions favour both hydrogen radical and unburned hydrocarbons
formation, thus promoting H abstraction reactions, such as:
CH4 +H → CH3 +H2 (1.5)
and
C2H4 +H → C2H3 +H2 (1.6)
which lead directly to H2 formation. Also the acetylene fraction is predicted to
increase with the equivalence ratio.
Figure 1.5: Computed exhaust H2 : CO and C2H2 levels as a function of mixture
stoichiometry and for an air preheat temperature of 500◦C. Source [5].
21
1.4 Scope of Thesis
1.4 Scope of Thesis
1.4.1 Specific objectives
The main target of this thesis is to demonstrate experimentally the feasibility of
a compact and flexible liquid fuel reformer for mobile applications. These two
requirements led to consideration of the non-catalytic partial oxidation as the
reforming technique. In this way it has been possible to reform common com-
mercial fuels (diesel, kerosene and biodiesel), which, because of their poisoning
compounds, cannot be reformed in any reformer involving catalysts. The par-
tial oxidation technique allows reduction of the dimension of the reactor and the
two-layer porous layer concept led to a sharp decrease in the device costs. The
specific objectives of this thesis are to:
• design a compact reactor including a vaporization stage, an ignition strategy
and a reforming layer;
• study the behaviour of the reformer varying the characteristic parameters
of the system (equivalence ratio, thermal load, materials);
• demonstrate the flexibility of the system with different heavy liquid fuels;
• test the possibility of reducing the carbon monoxide content and improve
the hydrogen fraction of the reformate.
1.4.2 Structure of Thesis
These targets were achieved and demonstrated with a set of experiments de-
scribed in the following chapters. The second chapter describes the experimental
methods, materials, fuels and measuring techniques that allowed optimisation
and investigation of burner functionality. The third chapter concerns n-heptane
reforming and presents a range of preliminary investigations, which lead to the
burner characterisation in terms of mixture preparation and operating envelope
of the burner. Moreover, parametric investigations have been performed to assess
the sensitivity of the reformate composition to the characteristic parameters. The
22
1.4 Scope of Thesis
use of a standard fuel allowed comparison of the results with chemical equilib-
rium data. The commercial fuel reforming experiments, described in the fourth
chapter, demonstrate the flexibility of the reformer in processing commercial and
carbon neutral fuels such as diesel, kerosene and biodiesel in the same burner.
The last chapter describes the investigations and the preliminary tests on steam
reforming and WGS in order to clean the reformate from poisoning species (CO)
and approach the requirements of a fuel cell.
23
Chapter 2
Experimental methods
This chapter describes the experimental methods used in this project. The dif-
ferent stages to achieve fuel reforming are presented, including fuel and material
specifications and the measurement techniques. At the end of the chapter, the
procedure adopted to run the tests is also presented and the equilibrium and
carbon balance calculation tools are explained.
2.1 Apparatus
The reformer developed for these experiments is illustrated in Fig. 2.1. Its
distinctive characteristic is the two-layer porous structure, the atomizer section,
and the capability to use various types of porous media. Different stages are
implemented to achieve fuel reforming to syngas and they are explained in details
in the following sections.
In principle, the liquid fuel needs to be atomized, vaporized and mixed with
the oxidizer. Then the mixture must be ignited and eventually reforming takes
place by means of the superadiabatic effect in the porous medium. In the ex-
periments, the flow rates of air (oxidizer) and fuel are accurately delivered by
means of two mass-flow controllers. Subsequently, the fuel needs to be intimately
mixed with the oxidizer through liquid break-up and evaporation. This target is
achieved using a modified commercial air-assisted atomizer and providing a hot
environment where the mixture is formed. The unreacted mixture is then ignited
by means of four spark plugs and stabilized over a flame holder. Thereafter, the
24
2.1 Apparatus
flame is switched to rich condition and the fuel is then reformed in the porous
material that allows the superadiabatic condition to be established leading to the
extension of the flammability limit. Eventually, gas samples have been analyzed
by means of a Gas Chromatograph and soot emission was also investigated by
means of a filter-paper-method machine.
Moreover, it has been necessary to monitor the temperature of the different
stages. For this purpose, one K-type and two R-type thermocouples are inserted
in the reactor in the vaporization stage, at the ignition level and at the exit of
the burner (T1, T2 and T3 respectively in Fig. 2.1).
Figure 2.1: Equipment setup for burner A, B and C. Burner D has alumina beads
as diffuser layer and burners E and F have a diameter of 100 mm.
2.1.1 Burners
Initially, 70 mm ID quartz tubes were used as burners in order to optically detect
the flame inside the porous material, as shown in Fig. 2.2 where the tube is filled
with two layers of alumina beads of different size. At the bottom of the tube a
25
2.1 Apparatus
brass plug was fixed and sealed with high-temperature O-rings and the nozzle was
centrally attached. The complete vaporization of the fuels could also be visually
checked, by denoting the absence of condensation on the tube’s walls.
Later on in the project, stainless steel tubes have been adopted to approach
the target of a practical device. A standard 316 SS grade was used due to its
good machining and welding characteristics and for the broad availability of the
material. Two different sizes were chosen in order to assess the scalability of
the device. A 70 mm ID x 300 mm burner was the most tested version, but a
100 mm ID modular burner was also investigated for some steam reforming and
water-gas shift reaction experiments (Chapter 5). Both the burners are flanged
at the bottom, to fit and seal a plate with the nozzle, the T1 thermocouple and
a containing stainless steel cone fitted in it. The plate is sealed to the rest of
the burner by a copper washer, since no plastic materials can withstand the high
temperatures of the process, and checked periodically for liquid and gas leakage.
The stainless steel burners are displayed in Fig. 2.3.
A 6 mm ID stainless steel tube coil has been wrapped around the burner to
integrate the heat produced by the exothermic reforming reaction. This heat has
been used to preheat the vaporizing air used to prepare the fuel/air mixture and
hence the burner was independent from external heat sources after the start-up
of the system.
The burners have been insulated with alumina wool (ZIRCAR Alumina mat)
in order to minimize heat losses to the environment. Specifically, only the re-
forming layer was externally insulated, whereas the vaporization stage was left
without any covering over the stainless steel tube. In this way the heat losses will
not allow this section to overheat and hence will oppose the flame propagation
to the upstream direction, preventing the ignition of the spray on the top of the
nozzle.
26
2.1 Apparatus
Figure 2.2: 70 mm ID quartz tube equipped with a 25 mm layer of 2-3 mm Al2O3
beads (flame holder) and a 80 mm layer of 6 mm Al2O3 beads (reforming layer).
n-heptane/air flame at ϕ=2, P=5 kW and preheating temperature T1=423 K.
The flame on the rim of the burner is due to the autoignition of the syngas with
the laboratory air.
27
2.1 Apparatus
Figure 2.3: 70 mm ID burner and 100 mm ID burner equipped with spark plugs,
coil stainless steel tube, thermocouples and top hoods.
2.1.2 Porous materials
The porous matrices used for the reforming layer are in the form of ceramic foam
or beads. Table 2.1 summarizes the main features of each material. The fine-pore
layer assists evaporation and prevents the flame to propagate upstream in most
of the conditions investigated, while the second (downstream) layer is where the
reforming occurs.
The first layer is a flame holder and it consists of a 50 mm long packed bed of
2-4 mm Al2O3 beads. The beads are held in place by a stainless steel wire mesh
placed over the stainless steel containing cone. The bead material is commercially
used for drying purposes (Sigma-Aldrich) and it has a melting point of 2313 K
[102]. This layer also enhances oxidizer/fuel mixing before ignition because of the
convoluted path that the mixture has to cover before reaching the layer interface.
28
2.1 Apparatus
On top of this first layer, different porous materials were implemented, foams
(burner A, B, C, E and F) and beads (burner D). Five pieces of foams, different
in terms of their porosity and material, have been tested (Fig. 2.4). The ceramic
foams tested in this work come from the foundry industry for metal filtering
and thus are high temperature resistant while the ceramic beads are designed
as heat exchange media. Beads are more robust to thermal cycles compared
to ceramic foams, which are brittle and crack after few (around 10) cycles of
high temperature reforming and cooling down (Fig. 2.5a). Furthermore, beads
have the advantage of ease of fitting inside the burner. The main drawback
of the utilization of beads in place of foams is the higher thermal inertia due
to the matrix geometry (thin solid bridges for ceramic foams compared to ∼6
mm diameter solid beads) and hence the lower response time of configuration D.
Moreover, the ignition system used for burner A, B, C, E and F failed for burner
D, because the heat released by the flame kernels does not offset the heat losses
to the alumina beads. The flame had to be ignited from the top of the burner
and the reaction layer slowly penetrated into the packed bed while heating the
beads.
The foam pieces had to be wrapped in an alumina paper in order to seal
them to the burner’s walls, to minimize leaking around the porous media and
to protect them from the different thermal expansion compared to the stainless
steel tube. Zirconia foam is MgO stabilised and it exhibits higher resistance to
high temperatures than the SiC, Al2O3, SiO2 compound foam [103]. Fig. 2.5b
shows the damage to the foam of burner C due to the high temperatures reached
at equivalence ratios close to stoichiometry and the high thermal loads.
29
2.1 Apparatus
Table 2.1: Porous medium specifications.
Burner A B C D E F
Material
ZrO2
foam
ZrO2
foam
SiC,
Al2O3,
SiO2
foam
D≃6mm
α−Al2O3
beads
ZrO2
foam
ZrO2
foam
Porosity 30 ppi 10 ppi 30 ppi 40%a 30 ppi 30 ppi
Length [mm] 100 100 100 100 50 25
Max
Temperature
[K]
1973
[103]
1973
[103]
1723
[103]
1923 [105]
1973
[103]
1973
[103]
Thermal
conductivityb
[Wm−1K−1]
2 2 -
12-30
[106]
2 2
Supplier Lanik Lanik Lanik Almatis Lanik Lanik
aPacked bed porosity ε estimated by [104], with dP particle diameter and D reactor diameter:
ε = 0.375 + 0.34
dp
D
bof the raw material
Figure 2.4: Porous foams used for the reforming layer. Specifications in Table 2.1
as burners A, B, C, E and F from left to right.
30
2.1 Apparatus
Figure 2.5: Foam failures: (a) ZrO2 foam cracking after thermal cycles; (b) SiC,
Al2O3, SiO2 compound foam melted after high-temperature operations. The
alumina paper wrapping is evident at the side of the cylindrical piece.
31
2.1 Apparatus
2.1.3 Flow control
The air flow is supplied by the laboratory compressed air line and it is controlled
by a Bronkhorst Mass Flow Controller. The controller works on the thermal mass
flow measuring principle (the temperature drift between no flow condition and gas
flowing through a heated channel is directly proportional to mass flow through
the sensor). The maximum pressure supplied is 2 bar (relative) before the nozzle.
The air flow can be directed to a 750 W electric heater, able to preheat the air
up to 573 K, or it can be switched to the stainless steel tube wrapped around the
burner. During the experiment it may become necessary to mix the air preheated
by the coil with some fresh air to keep the temperature at the desired value.
The liquid flow is delivered by a magnetic driven gear pump. The amount
of liquid supplied is controlled by a mini CORI-FLOW Bronkhorst Mass Flow
Meter which gives the voltage feedback to the pump according to the flow reading,
adjusting the speed of the pump.
The mini CORI-FLOW is able to measure the density and the temperature
of the liquid passing through and then it infers the flow rate, independently from
the liquid characteristics and hence no calibration is required.
For safety reasons, only a small amount of liquid is kept next to the feeding
system and the flow controllers shut their valves if a power cut occurs. Moreover,
for better accuracy the fuel line is flushed with acetone before a different fuel is
tested.
2.1.4 Atomization and vaporization
The atomization turned out to be one of the key and most delicate stages of the
reforming process. Several commercial and custom-made nozzles have been tested
before finding a suitable design. Eventually, a commercial air-assisted nozzle has
been used for the experiments. The main issue was the processing of a very low
liquid flow rate and the achievement at the same time of a good atomization qual-
ity. This goal has been achieved using a DELAVAN AL-15 nozzle, commercially
used for heavy oil burners. The nozzle is designed around the external mix of
air and fuel concept. In this nozzle, the liquid phase is ejected in a jet form and
the air is impinging on it externally to the nozzle, causing the break-up and the
32
2.1 Apparatus
atomization. Therefore, this design allows independent control of the fuel and the
air flow rate, giving high turndown ratios. The supplier certifies droplets with a
Sauter Mean Diameter (SMD) of 50 µm and a spray cone of around 45 ◦.
However, the commercial nozzle had to be modified to withstand the high
temperatures of the experiments. In particular, all the plastic seals have been
removed and the fuel channel has been replaced with a hypodermic tube in order
to reduce the residence time of the liquid inside the hot nozzle housing. The
nozzle is heated by the hot air flowing through and this can cause boiling of the
liquid inside the fuel channel. Consequently, flow oscillations may occur with low
boiling point liquids (i.e. n-heptane). An extra hypodermic tube has been added
next to the fuel line in order to inject water for the steam reforming and water-
gas shift reaction experiments. The different steps in the nozzle modification are
illustrated in Fig. 2.6.
The spray flows in a 100 mm long stainless steel conical section and the
evaporated fuel-air mixture reaches the first porous layer. A K-type thermocouple
placed just before the upstream side of the first porous section (i.e. at the end
of the atomiser-vapouriser conical section) gives the voltage feedback to the air
heater through a temperature controller (T1). This configuration is displayed in
Fig. 2.7. Different temperature set points have been fixed, according to the fuel
specifications, as explained in Section 2.2.
The uniformity of the air/fuel mixture has been assessed and the results are
reported in Section 3.1.1. For that purpose, a Fast Flame Ionization detector
(Cambustion HFR500 FFID) has been used. A 1.5 mm probe has been used to
sample the carbon content of the mixture over the burner cross section imme-
diately downstream the fine porous layer (flame holder). The detector gives the
total C1 as contributions from all the hydrocarbons present in the gas sample.
The sample line has been kept at 423 K when sampling n-heptane/air mixtures,
in order to avoid fuel condensation in the probe. Only n-heptane could be sam-
pled, since the other fuels have higher boiling points. The machine probe head
would suffer damages to the plastic sealings if its temperature is raised above 423
K and the condensation of fuel (especially from heavy fuel like the one used in
this research) in case of lower temperatures would flood the machine and damage
the sensors.
33
2.1 Apparatus
The FFID machine has a linearity of the output (voltage output over carbon
concentration) until 2.0-2.5% C1. After this threshold, the reading is lower than
the actual value. Since the mixture sampled contained higher concentration of
hydrocarbons (ϕ=2.0 leads to a molar fraction of 3.7% of C7H16 in air), the
reading is normalized over the average value, in order to detect variation from
the mean value, rather than obtaining the exact proportion of fuel in air.
Figure 2.6: Exploded view of the DELAVAN AL-15 nozzle. (a) original version;
(b) modified nozzle; (c) modified nozzle with water line.
34
2.1 Apparatus
Figure 2.7: Front and top view of the fuel pre-vaporizer. The T1 thermocouple
on top of the cone, the air and fuel inlet and the nozzle are visible in the two
photographs.
35
2.1 Apparatus
2.1.5 Ignition
The main problem associated with this stage was the development of a stable
flame in between the two “cold” porous layers. Tests with a quartz tube demon-
strated the ease of quenching of a flame immediately after being generated by
two electrodes with a gap of 2 mm on the peripheral area of the burner. The
high heat losses to the porous matrix overcomes the heat generated by the flame
kernel.
Eventually, the ignition has been achieved by fixing four automotive spark
plugs evenly distributed on the burner walls at the porous layer interface. The
peripheral position prevents the spark plugs to undergo the maximum tempera-
ture of the burner, improving the lifetime of the electrodes. Additionally, after
few seconds of voltage applied to the plugs, four flame kernels quickly develop
and cover the whole cross section of the tube. Thus, a long emission of unreacted
mixture is avoided.
At the porous layer interface a R-type thermocouple is placed (T2) and it
gives the indication of successful ignitions. The actual temperature reading of
this thermocouple is not taken into consideration as it is strongly affected by
radiation from the porous material, the temperature of which cannot be measured
in order to be used for a correction.
2.1.6 Temperature detection
The temperatures of the different stages have been measured by three thermo-
couples. T1 is a K-type thermocouple with 1.5 mm bead diameter. K-type ther-
mocouples have an accuracy within ±2.5 K from 233 to 606 K which is suitable
for the measurement of the evaporation stage temperature (max. 573 K) [107].
Two more thermocouples were located inside the burner. T2 is a R-type thermo-
couple for the ignition detection at the porous layers interface and T3 is a R-type
thermocouple placed at the exit of the burner, where the gas samples are ex-
tracted. Both have a bead diameter of 0.5 mm and hence they are quick response
detectors (0.025 seconds from 293 K to 373 K [107]). T3 gives the temperature of
the syngas that is eventually analyzed by the Gas Chromatograph and provides
the indication of the achievement of a thermal steady state, once the reading
36
2.1 Apparatus
becomes constant. This temperature is also used to calculate the energy balance
between the input species and the output species, to assess the heat losses (Sec-
tion 3.27). R-type thermocouples’ accuracy is within ±0.0025·|T| K from 873 to
1873 K [107].
T1 is directly plugged into the temperature controller for the electric air heater,
whereas T2 and T3 are connected to an amplifier (Omega Omni-Amp IV), which
provides also the cold junction compensation, and thereafter are plugged into a
digital acquisition card (NI-PCI-6034E). Eventually, a computer equipped with
NI Labview converts the voltage measured by these two detectors into a temper-
ature reading.
2.1.7 Gas sample analysis
In order to collect a gas sample, it was necessary to quench the ongoing reactions
and to remove the water vapour to avoid any damage to the gas chromatograph.
A water-cooled probe attached to a vacuum pump allows the quenching and
extraction of a sample of the exhaust gas at the burner exit. Subsequently, the
sample is passed through a calcium chloride cartridge for water removal. Between
the extraction and the drying stage, the sample line has been kept over 373 K to
avoid water condensation, by using a heating cord wrapped around the extraction
tube.
To obtain a consistent sample of the reformate, a hood with a 25 mm outlet
has been attached to the top of the burners. Therefore, the sample analyzed
represents a mixture of the gas coming from the very hot part (central) and from
the cold parts of the burner (peripheral — close to the burner walls). This hood
has a conical shape in order to avoid strong pressure drops which may affect the
spray and the flame.
The gas samples have been injected into a gas chromatograph (GC). The GC
is equipped with a Thermal Conductivity Detector (TCD) and with a Flame Ion-
ization Detector (FID) and two columns for the species separation. The Molecular
Sieve is used to separate H2, O2, N2, CH4 and CO. It is too retentive for C2
hydrocarbons and it can be deactivated by CO2 and H2O. The Haysep N porous
polymer column do not retain permanent gases and then H2, air, CO and CH4
37
2.1 Apparatus
pass straight through without separation. CO2 is separated along with C2H2,
C2H4 and C2H6. Therefore, the gas sample passes first through the Haysep N
(porous polymer bead column) which retains carbon dioxide and C2 hydrocar-
bons while the other species elute quickly to the Molecular Sieve column. A
3-way valve allows the bypass of the Molecular Sieve before CO2 comes off the
porous polymer bead column. Finally, the sample passes through the TCD (non-
destructive detector) and afterwards through the FID (sample is combusted).
The FID detector is about 100 times more sensitive than the TCD. This detector
ionizes carbon atoms, in a hydrogen flame. Normally a hydrogen flame produces
a negligible amount of ions, so only hydrogen carbon bounded compounds carried
into the flame result in the formation of ions, whose quantity is measured in the
FID and the carbon amount deducted.
The carrier gas used is high-purity Argon and pure air and 99.995% purity
hydrogen were used for the FID flame. The machine is calibrated by a gas mixture
with species concentrations similar to the ones expected in the reformate before
every set of tests. The first calibration gas used was composed of 19.94% H2,
1.00% O2, 15.00% CO, 4.04% CO2, 4.87% CH4, 3.01% C2H2, 1.99% C2H4, 1.00%
C2H6, balance nitrogen. Later on in the research, a new calibration gas was
used, to match better the new syngas composition obtained through the system
optimization: 19.83%H2, 0.993%O2, 18.09% CO, 2.95% CO2, 4.00% CH4, 2.98%
C2H2, 2.01% C2H4, 0.514% C2H6, balance nitrogen. The sample line has been
periodically checked for leaks and the samples have been injected at the same
temperature and pressure. Given the partial oxidation process, no air should be
present in the gas samples, therefore the absence of oxygen in the GC reading
may insure that no leaks are present in the sampling line.
2.1.8 Particulate detection
Soot emission was investigated by means of a filter-paper-method machine (AVL
415S). A gas sample is taken from the exit of the burner with a probe and passed
through a filter paper. The blackening of the filter paper is measured with a
reflectometer and this indicates the soot content in the exhaust gas. The output
38
2.2 Fuels
value (FSN) is the Filter Smoke Number and it corresponds to the Paper Black-
ening (PB), as defined by Eq. 2.1, for a sampling length of 405 mm at 298 K
and 1 bar:
PB =
100−RR
10
(2.1)
with RR=
RP
RF
·100% and RP the reflectometer value of sample, RF the reflectome-
ter value of the unblackened paper and RR the relative brightness of the sample.
The soot detector extracted five gas samples for each condition and the average
value is reported in Section 3.4 and 4.4.
2.2 Fuels
The fuels investigated in this work are n-heptane, Jet A-1 kerosene, commercial
diesel oil and Rapeseed-Oil Methyl Ester (RME) biodiesel. Laboratory analyses
were performed for Jet A-1, Diesel and biodiesel by the Chemistry Department
Labs of Cambridge University. From the carbon-to-hydrogen ratio it is possible
to calculate the approximate molecular formulas of the fuel which, together with
the fuel specifications, are summarised in Table 2.2. n-heptane is high-purity
laboratory grade fuel (99+%). The diesel oil used for the reforming experiments
is commercial fuel from a local gas station and kerosene is from a supplier for the
local airport. The RME biodiesel is rapeseed-oil based and follows the EN 14214
standards and requirements.
The choice of reforming n-heptane has been dictated by the possibility of com-
paring experimental data with numerical simulations and computational models,
given the availability of chemical mechanisms [10] for such a reference fuel. Diesel,
kerosene and biodiesel experiments represent practical applications of the re-
former with commercial fuels. Different evaporation temperatures (T1) have been
set for ensuring full evaporation for each fuel. n-heptane has been vaporised in a
423 K environment. The boiling point range of Diesel is 457-648 K [108; 109] due
to the different properties of the different component contained; consequently,
the evaporation temperature was set to 573 K, being the maximum achievable
39
2.2 Fuels
with the configuration adopted. It has to be noted that the first layer improves
the vaporization process, as demonstrated by [110]. This effect is due to the
high temperature of the first layer for the heat conduction and radiation coming
from the downstream flame front and for the complex flow pattern covered by
the mixture inside the ceramic bead layer, which enhances the heat transfer to
the residual liquid spray phase. Similarly, kerosene boiling range is 463-523 K
[108] and hence T1 has been set at 523 K. The biodiesel/air mixture preparation
temperature has been set at 523 K, limited by the risk of cracking at high tem-
peratures, caused by the thermal instability of this fuel due to the presence of
oxygen atoms in the fuel molecule, as explained by [111; 112]. Figs. 2.8, 2.9 and
2.10 give the GC-MS reading of those fuels. The spectra clearly show the pres-
ence of heavier hydrocarbon components in diesel compared to kerosene, as listed
in Table dieselgcms and kerosenegcms. The biodiesel spectrum is consistent with
similar analyses published for the same type of fuel [113]. The fuel specifications
given by the suppliers are listed in Figs. 2.11, 2.12 and 2.13.
40
2.2 Fuels
Table 2.2: Fuel specifications.
Fuel n-heptane Jet A-1 Diesel oil RME biodiesel
Chemical for-
mula
C7H16 C12H24 C12H23 C19H36O2
LHV [MJ/kg] 44.926 42.800 43.000 37.600
Kinematic
Viscosity
[mm2/s]
@ 40 ◦C
0.51 1.0-2.0 2.0-4.5 3.50-5.0
Vapour Pres-
sure [kPa]
0.05 0.5 0.5 <0.267
Density
[kg/m3]
@ 15 ◦C
0.682 0.820 820-835 0.875-0.900
Heat of
vaporization
[kJ/kg]
318 251 250 254
Supplier Acros CPS fuels Shell ADM
41
2.2 Fuels
Figure 2.8: GC-MS output for Diesel oil.
42
2.2 Fuels
Figure 2.9: GC-MS output for Jet A-1.
Figure 2.10: GC-MS output for RME biodiesel.
43
2.2 Fuels
Table 2.3: Fig. 2.8 main peak description.
Peak
number
Retention
Time
Component
Type
1 2.254 Aromatic
2 2.480 Unknown
3 3.354 Aromatic
4 4.007 Aromatic
5 4.047 Alkane
6 6.748 Alkane
7 9.068 Alkane
8 11.002 Alkane
9 12.669 Alkane
10 14.142 Alkane
11 15.482 Alkane
12 16.743 Alkane
13 16.769 Alkane
14 17.910 Alkane
15 17.976 Alkane
16 19.023 Alkane
17 19.276 Methyl Ester
18 20.077 Alkane
19 21.037 Unknown
20 21.097 Unknown
21 22.037 Alkane
22 22.950 Alkane
23 23.824 Alkane
44
2.2 Fuels
Table 2.4: Fig. 2.9 main peak description.
Peak
number
Retention
Time
Component
Type
1 1.740 Alkane
2 2.007 Alkane
3 2.147 Unknown
4 2.214 Alkane
5 2.260 Aromatic
6 2.487 Alkane
7 4.054 Alkane
8 6.741 Alkane
9 9.041 Alkane
10 10.962 Alkane
11 12.609 Alkane
12 14.069 Alkane
13 15.409 Alkane
45
2.2 Fuels
Figure 2.11: Shell diesel specifications [6].
46
2.2 Fuels
Figure 2.12: ADM biodiesel specifications [7].
47
2.2 Fuels
Figure 2.13: CPS kerosene specifications [8].
48
2.3 Post reforming catalysis
2.3 Post reforming catalysis
In order to explore the possibility of reducing the carbon monoxide and hydrocar-
bon content of the syngas, two further stages have been implemented and tested.
These two stages consists of a steam reforming stage to process the remaining
light hydrocarbons present in the off gas and a water-gas shift reaction to trans-
form the carbon monoxide to carbon dioxide and hydrogen. Both these stages
need steam and a catalytic bed where the reaction can take place. Burner E was
used for these experiments, due to the larger section area and therefore the longer
residence time of the syngas inside the catalytic bed. It has to be noted that both
the catalysts are poisoned by sulphur and therefore out of the fuels used in this
project, only n-heptane and biodiesel could be tested.
2.3.1 Steam reforming catalyst
The steam reforming catalyst used for this investigation is the G 90-EW from
Su¨d-chemie which is nickel-based on an aluminium oxide and calcium aluminate
carrier in “spoked wheel” shape, as shown in Fig. 2.14a. Specifically, the com-
ponent proportion on a weight basis is 11% Ni, 80% Al2O3, 9% CaO (bound as
CaAl12O19) and less than 0.02% of Na and SiO2. This catalyst is designed to
process natural gas, but it will be shown later that it has excellent performance
with the reformate produced by the partial oxidation of n-heptane, which con-
tains (apart from hydrogen and carbon monoxide) mainly methane and acetylene.
The operating temperature range for this catalyst is 773-1173 K and for this rea-
son it was simply poured on top of the ZrO2 foam, in order to use all the heat
produced by the reforming process without the aid of any external heat source.
A modular section of 100 mm in length was filled with this material.
2.3.2 Water-gas shift catalyst
The catalyst used for the water-gas shift reaction is the NextCatA from the
supplier Fuelcellsmaterial.com. It comes in pellet shapes (Fig. 2.14b) composed
of Zirconia doped with Ceria and 2% of Pt. The catalyst is for high-temperature
49
2.4 Procedure
operations and the working range is 673-1073 K. A cartridge filled with pellets
was attached to the top of the hood of burner E.
Figure 2.14: (a) Su¨d-chemie G 90-EW steam reforming catalyst wheels; (b) Fu-
elcellsmaterial.com ZDC/Pt catalyst pellets.
2.4 Procedure
A typical test procedure starts with the warming up of the reactor by a flow of
only the oxidiser (i.e. the air) through the reactor to bring the temperatures up
to the appropriate value corresponding to the fuel characteristics so as to avoid
fuel condensation. This first stage takes approximately from 10 to 20 minutes,
depending on the fuel that is processed. At this point, the fuel is injected. An
air/fuel ratio close to stoichiometry and a thermal power of 2 kW is used first
to achieve a prompt ignition. At this condition, the unreacted air/fuel mixture
flow velocity nearly matches a typical hydrocarbon laminar flame speed at ϕ=1.0
(approx. 40 cm/s [114] at normal atmospheric temperature and pressure) for the
size of the burners used in this experiment. The four automotive spark plugs are
activated and provide the means of ignition. After the ignition, the reaction zone
quickly extends inside the porous matrix and the equivalence ratio is switched
to rich values. When the thermocouple on the top of the burner (T3) gives a
constant reading, the system has reached thermal equilibrium and samples can
50
2.5 Equilibrium calculation
be extracted through the sampling line. From cold to fully stabilised operation,
the system takes about 20-30 minutes, depending on the fuel.
2.5 Equilibrium calculation
The experimental molar fractions of H2, O2, N2, CO and CO2 and hydrocar-
bons up to C2 (methane, acetylene, ethylene and ethane) have been compared
to thermodynamic equilibrium calculations. The results were obtained by means
of a computational tool which elaborates the NASA database thermodynamic
coefficients [115]. The inputs to this code are the pressure (atmospheric in our
case) and the temperature of the mixture whose equilibrium composition is to
be calculated. Two options were used: (a) the adiabatic flame temperature cor-
responding to the ϕ and initial T (T1) used, calculated by [116], and (b) the
experimentally-determined sample temperature (T3).
2.6 Carbon balance
An algorithm has been written and used to calculate the water content of the
syngas. An iterative water molar fraction value has been introduced in the calcu-
lation in order to minimize the O:N ratio of the dry species minus the O:N ratio
of the wet species, since the O:N ratio must be conserved. In this way the water
mole fraction is estimated and hence the percentage of H2 bound by water and
light hydrocarbons present in the off-gas can be estimated.
The sum of the three components (molecular hydrogen, hydrogen in hydro-
carbons and hydrogen in water) ranges from around 90% down to 75% of the
hydrogen contained in the fuel molecule over the equivalence ratio. The discrep-
ancy at rich equivalence ratios may be due to the soot formation or to deposits
which have been detected on the first porous layer (flame holder). Moreover,
potential formation of hydrocarbons not detectable by the Gas Chromatograph
or condensed in the drier cartridge together with water may contain the missing
hydrogen. Hydrocarbons not detectable are those higher than C2 which, even if in
small fractions, can “hide” large quantities of hydrogen. The Gas Chromatograph
columns can tolerate a maximum temperature of 160 ◦C and hence high-boiling
51
2.6 Carbon balance
point hydrocarbons cannot be analyzed by the described configuration in any
case.
52
Chapter 3
n-heptane reforming
This chapter deals with the experimental investigations carried out on n-heptane
reforming. Several parameters have been explored such as the thermal load, the
equivalence ratio, the different material characteristics and the burner size effect
on the syngas composition. Some preliminary studies have also been performed
concerning the homogeneity of the unreacted air/fuel mixture before the layer
interface and the syngas composition uniformity over the cross section of the
burner. Also the operating range of one of the configurations has been investi-
gated, demonstrating the superadiabatic effect on the flame speed.
3.1 Preliminary investigations
In this section the mixture uniformity, the reformate uniformity over the cross
section of the burner and the operating range of the reformer have been investi-
gated.
3.1.1 Mixture homogeneity
The functionality of the atomisation and evaporation stage of the burner has
been investigated in terms of mixture homogeneity by means of the HFR500
FFID and with the methodology described in Section 2.1.4. The tests have been
conducted on the 70 and 100 mm ID burner with n-heptane/air mixtures at P=7
kW and ϕ=2.0, 2.5 and 3.0 with a preheat temperature of 423 K. The mixture
53
3.1 Preliminary investigations
was sampled at 100 Hz and 1000 readings were recorded and averaged at 10 mm
displacement steps of the probe above the flame holder layer. Fig. 3.1 shows the
results of this investigations with the plot of the local voltage output normalized
by the average value over the cross section of the fuel/air mixture at ϕ=2.0 for
the two burners. It can be noted that the mixture is richer than the average value
in the central part of the burners, whereas on the sides it results leaner. However,
the deviation is limited to ±2% for the 70 mm ID burner and to ±5% for the 100
mm ID burner.
The same investigation has been carried out at ϕ=2.5 and 3.0 at P=7 kW.
Fig. 3.2 and Fig. 3.3 display the results from the two burners. A mixture inho-
mogeneity is more marked at higher equivalence ratios. At ϕ=2.5 the deviation
from the average value is around ±4% and ±10% and at ϕ=3.0 it reaches ±7%
and ±15% for the small and big burner respectively. Probably, the modification
of the nozzle design and the reduced air flow rate at high equivalence ratios im-
pede the initial breakdown of the liquid jet and hence may affect the uniform
distribution of the air/fuel mixture over the burner cross section. This disparity
is very mild for the small burner and confined to the peripheral area of the cross
section, whereas it is stronger for the 100 mm ID burner. This is may also be due
to the narrow spray cone angle (45 ◦) that prevents the mixture to cover uniformly
the cross section of the tube. Nevertheless, in view of the relatively small mixture
inhomogeneity and the further mixing achieved inside the reforming section, it is
expected that the reformate composition will be uniform across the burner. This
is demonstrated in more detail in the next Section.
54
3.1 Preliminary investigations
probe position [cm]
n
o
rm
al
iz
ed
vo
lta
ge
[-]
-4 -3 -2 -1 0 1 2 3 40.96
0.98
1
1.02
1.04
1.06
100mm ID burner
70mm ID burner
Figure 3.1: Reactant mixture over the first porous layer (flame holder) measured
by the FFID and expressed as normalized voltage over the cross section of the
two size burners (70 and 100 mm ID) at P=7 kW and ϕ=2.0.
55
3.1 Preliminary investigations
probe position [cm]
n
o
rm
al
iz
ed
vo
lta
ge
[-]
-3 -2 -1 0 1 2 30.9
0.95
1
1.05
1.1
φ=2.0
φ=2.5
φ=3.0
Figure 3.2: Reactant mixture over the first porous layer (flame holder) measured
by the FFID and expressed as normalized voltage over the cross section of the 70
mm ID burner at P=7 kW and ϕ=2.0, 2.5 and 3.0.
probe position [cm]
n
o
rm
al
iz
ed
vo
lta
ge
[-]
-5 -4 -3 -2 -1 0 1 2 3 4 50.85
0.9
0.95
1
1.05
1.1
1.15
1.2
φ=2.0
φ=2.5
φ=3.0
Figure 3.3: Reactant mixture over the first porous layer (flame holder) measured
by the FFID and expressed as normalized voltage over the cross section of the
100 mm ID burner at P=7 kW and ϕ=2.0, 2.5 and 3.0.
56
3.1 Preliminary investigations
3.1.2 Reformate uniformity
After the investigation of the air/fuel mixture before ignition, a similar test has
been carried out for the uniformity of the syngas composition after the porous
reforming layer. For this purpose, burner E has been used (100 mm ID tube with
a 30 ppi ZrO2 foam), since it developed the largest divergence from a uniform
distribution of air/fuel mixture. Five different locations were sampled, going from
the burner walls to the centre of the tube. The extraction of the gas sample has
been carried out right above the foam, before diffusion and mixing could occur.
Also the temperature has been recorded at the different sampling points.
The results are shown in Fig. 3.4. The syngas composition results to be uni-
form over the radial positions, whereas the temperature trend follows a parabolic
curve over the flame holder layer, due to the heat losses at the walls. Hydrogen
and carbon monoxide peaks at the centre of the reformer and they are lower in
molar fraction at the peripheral area. Similarly, the temperature peaks at the
centre because of the heat losses at the tube’s walls. The syngas composition
difference between the five sampling points is anyway very limited and it is con-
tained within 10% variation of the average value.
57
3.1 Preliminary investigations
probe radial position [cm]
dr
y
m
o
la
r
fra
ct
io
n
[%
]
T
[K
]
-4 -2 0 2 4
5
10
15
20
900
1000
1100
1200
1300
T3
H2
CO
CO2
Figure 3.4: Syngas composition and exit temperature T3 over the reforming layer
of burner E at P= 7 kW and ϕ=2.0.
58
3.1 Preliminary investigations
3.1.3 Operating envelope
The operating points of the burner have been explored in order to detect sta-
tionary flame conditions. Burner C was used for this purpose. The thermal load
has been fixed and flames with equivalence ratios ranging from 1.5 to richer val-
ues were investigated. Then further sets of tests were carried out with a higher
thermal load. Four different flame behaviours have been observed: (a) quenching
– the heat produced by the flame does not offset the heat losses and the flame
cannot be sustained; (b) flash-back – the flame speed is high enough (or the flow
rate low enough) to penetrate the flame holder layer in the upstream direction
and ignite the mixture in the vaporization area; (c) filtration wave – a reaction
wave travels from the layer interface up to the top of the burner; (d) steady flame
stabilized at the interface or immediately downstream. This condition is the one
of interest in this work. The difference between conditions (c) and (d) was de-
tected by the temperature measurements. When the thermocouple placed at the
interface (T2) gave a constant and higher reading than the exit gas temperature
(T3), a filtration wave did not initiate and the flame would stabilise. On the con-
trary, a filtration wave gives a decreasing reading of the interface temperature T2
with time. A steady condition is associated with the stabilisation of the reaction
zone in the first half of the large pore layer, as demonstrated by [90]. Moreover,
the flow and fuel/air proportion conditions (P and ϕ) have been kept constant
for at least 20 mins, as a filtration wave travels with a speed around 10−4 m/s
[62; 63; 92] , which means that the flame would reach the top of the porous matrix
in approximately 16 mins, for this specific configuration. Therefore, if the flame
did not appear on the top of the porous media after 20 mins, the working point
was included in condition (d).
The operating envelope is described in Fig. 3.5, where the working points of
the reactor are plotted at different flow rates and equivalence ratios. Since the
flame is stabilised, the flow velocity in the burner is equivalent to the flame speed.
High equivalence ratios (up to ϕ=5.0) were reached and a maximum flame speed
of 88.6 cm/s at ϕ=2.0 has been achieved, limited by the equipment maximum
scale and the hazard of the great heat produced. The broadest steady operating
zone occurs at ϕ=2.0, consistent with [117].
59
3.1 Preliminary investigations
The experimental results have been compared to numerical calculations of free
premixed flames (i.e. in the absence of porous media) carried out by means of
a commercial code (COSILAB) [9]. The calculated flame speeds are for a free
n-heptane flame at the corresponding preheat temperature and equivalence ratio
of the experimental results. A detailed mechanism for n-heptane have been used
[10]. The comparison of the two sets of data demonstrates that the flame in the
porous media achieves much higher flame speeds than the the free flame at the
same ϕ, showing the establishment of superadiabatic conditions.
Figure 3.5: Operating envelope for burner C for n-heptane reforming at a pre-
heat temperature of 423 K. Laminar burning velocity of free premixed flames
calculated by the code COSILAB [9] and a detailed chemical mechanism [10].
60
3.2 Parametric investigations
3.2 Parametric investigations
This section deals with the sensitivity of the syngas composition on different
parameters, namely equivalence ratio, thermal load, porous material and burner
size.
3.2.1 Thermal load effect
The effect of the thermal load on the syngas quality for a fixed equivalence ratio
has been investigated in burner A. The thermal load is defined in this work as
the mass flow rate of fuel multiplied by the lower heating value (P = m˙ · LHV )
and thus it is strictly related to the bulk velocity of the flow inside the burner.
The equivalence ratio has been kept constant at ϕ=2.0, since, as demonstrated
by Section 3.1.3, at this condition the burner showed stationary flame conditions
for a broad range of flow speed. The thermal load range has been varied from 2
to 12 kW, corresponding to a flame speed ranging from 11.3 to 67.6 cm/s. The
results are reported in Fig. 3.6. The dry molar fractions of hydrogen and carbon
monoxide increase and approach equilibrium as the thermal load increases. Sim-
ilarly, the dry carbon dioxide molar fraction decreases with P. In Fig. 3.7, the
experimental dry molar fraction of hydrogen, carbon monoxide and carbon diox-
ide is compared to the calculated equilibrium at the measured exit temperature
T3. The similar trends and the similarity of the values demonstrate consistency
of the experimental results.
The increase of hydrogen and carbon monoxide concentration may be at-
tributed to the higher reaction temperatures which are measured with increasing
thermal loads. Fig. 3.8 demonstrates this increase with the plot of the mea-
sured exit temperature (T3) compared to the adiabatic flame temperature of a
n-heptane/air mixtures at ϕ=2.0 over the thermal load. Also, the light hydrocar-
bons present in the syngas show a decrease in their concentration passing from
a thermal load of 2 kW to 7 kW and afterwards the molar fractions assume con-
stant values. This enhancement is highlighted also by the conversion efficiency,
defined by Eq. 3.1:
61
3.2 Parametric investigations
ηLHV =
m˙H2LHVH2 + m˙COLHVCO
m˙C7H16LHVC7H16
(3.1)
ηLHV is 46.6% at P=2 kW and increases up to 70% at P=7 kW and then it
improves slowly up to almost 72% at P=12 kW, as shown in Fig. 3.10. Analo-
gously, the molar hydrogen conversion, given by Eq. 3.2, can be introduced and
it represents the percentage of H2 present in the reformate over the H2 content
of n-heptane, in terms of molar fraction X:
ηH2 =
XH2,out
8XC7H16,in
(3.2)
ηH2 is 27.8 % at P=2 kW and increases up to 48.1% at P=7 kW and it reaches
49% at P=12 kW. Fig. 3.9 displays the hydrogen distribution in the products.
It can be noted that molecular hydrogen percentage increases with the thermal
load reaching a peak at P=7-8 kW and it assumes a constant value afterwards.
Higher thermal loads have not been investigated due to equipment limitations,
the hazard of the high temperatures and the high hydrogen and carbon monoxide
flow rates produced. Similarly, lower thermal loads than 2 kW have not been
investigated for the risk of damaging the air heater at low air flow rates and for
the little heat produced by the flame which could not offset the heat losses to the
sides of the burner and by radiation downstream.
62
3.2 Parametric investigations
thermal load [kW]
dr
y
m
o
la
r
fra
ct
io
n
[%
]
2 4 6 8 10 120
5
10
15
20
H2 exp
CO exp
CO2 exp
H2 num,TAD
CO num,TAD
CO2 num,TAD
(a)
thermal load [kW]
dr
y
m
o
la
r
fra
ct
io
n
[%
]
2 4 6 8 10 12
0
0.5
1
1.5
CH4 exp
C2H2 exp
C2H4 exp
C2H6 exp
(b)
Figure 3.6: Calculated equilibrium at the adiabatic flame temperature and the
experimental mole fractions (on a dry basis) for a inlet air/fuel temperature of
423 K at various thermal loads for n-heptane reforming at ϕ=2.0 for burner A.
(a) Hydrogen, carbon monoxide and carbon dioxide. (b) Measured methane,
acetylene, ethylene and ethane.
63
3.2 Parametric investigations
thermal load [kW]
dr
y
m
o
la
r
fra
ct
io
n
[%
]
2 4 6 8 10 120
5
10
15
20
H2 exp
CO exp
CO2 exp
H2 num,T3
CO num,T3
CO2 num,T3
Figure 3.7: Calculated equilibrium at the measured exit temperature T3 and the
experimental mole fractions (on a dry basis) of hydrogen, carbon monoxide and
carbon dioxide for a inlet air/fuel temperature of 423 K at various thermal loads
for n-heptane reforming at ϕ=2.0 for burner A.
Thermal load [kW]
Te
m
pe
ra
tu
re
[K
]
2 4 6 8 10 12600
800
1000
1200
1400
1600
T3 exp
Tadiabatic
Figure 3.8: Adiabatic and experimental exit temperature T3 for n-heptane re-
forming at conditions as in Fig. 3.6.
64
3.2 Parametric investigations
Figure 3.9: H2 distribution in hydrogen, water and hydrocarbons present in n-
heptane reformate as a function of thermal load based on processing of the mea-
sured dry molar fractions. Conditions as in Fig. 3.6.
thermal load [kW]
co
n
ve
rs
io
n
ef
fic
ie
n
cy
[%
]
2 4 6 8 10 1240
50
60
70
80
Figure 3.10: Conversion efficiency for n-heptane reforming at ϕ=2.0 and at vari-
ous thermal loads for a preheat temperature of 423 K.
65
3.2 Parametric investigations
3.2.2 Equivalence ratio effect
Different porous materials for the reforming layer (burners A, C and D) at thermal
loads ranging from 6 to 8 kW have been tested at equivalence ratios from ϕ=1.1
to ϕ=3.0 with steps of 0.1. The results from the equivalence ratio effect on the
reformate are shown in Figs. 3.11, 3.12, 3.13, 3.14, 3.15, 3.16, 3.17, 3.18 and 3.19
where the experimental molar fractions on a dry basis of H2, CO, CO2 and of
hydrocarbons up to C2 are displayed, along with the calculated equilibrium data
for burner A, C and D at P= 6, 7 and 8 kW. The thermodynamic equilibrium
calculations predict negligible concentrations of hydrocarbons.
The general trend is a monotonic increase of the hydrogen and carbon monox-
ide molar fraction and decrease of carbon dioxide molar fraction against the
equivalence ratio. Experimental results and calculated thermodynamic equilib-
rium data are very close up to ϕ=1.7 for P=6 kW and up to ϕ=1.9 for P=7 and
8 kW for burner A. For burner C at ϕ=1.7, H2, CO, CO2 dry molar fractions
are: 10.0% and 10.2%, 14.2% and 15.7%, 4.8% and 4.3% for the experimental and
equilibrium results respectively. At P=7 kW and ϕ=1.9 the dry molar fractions
of H2, CO, CO2 for burner A are 13.4% and 13.8%, 15.9% and 17.3%, 4.2% and
3.6% for the experimental and equilibrium results respectively. For burner D,
experimental data coincide well with equilibrium calculated molar fractions up
to an equivalence ratio of ϕ=2.0-2.1. At P=8 kW and ϕ=2.0, burner D presents
the following H2, CO, CO2 molar fractions: 15.4% and 15.5%, 17.9% and 18%,
3.4% and 3.3% for the experimental and equilibrium results respectively. For all
the conditions displayed, at ϕ richer than 2.0, hydrogen and carbon monoxide
molar fractions stay constant, whereas the hydrocarbon (HC) presence increases
considerably. For instance, passing from ϕ=2.1 to 2.9 for burner A, H2, CO
and CO2 molar fractions pass from 15.28% to 15.29%, 16.7% to 15.8% and from
3.8% to 2.9% respectively, whereas methane increases from 1.18% to 3.3%, acety-
lene from 0.7% to 1.8%, ethylene from 600 ppm to 1.3% and ethane from below
detectable limit to 140 ppm. Hence, the disparity of the hydrogen and carbon
monoxide values between the experimental and equilibrium data is probably due
to the formation of light hydrocarbons at high equivalence ratios.
66
3.2 Parametric investigations
Rich flames lead to a large presence of hydrogen and carbon monoxide in
the samples but at the same time the reaction temperatures drop because for
the technique used in this study the heat provided is solely produced by the
oxidation process. Consequently, after an increase of H2 and CO molar fractions,
the fuel is reformed to light hydrocarbons rather than to syngas, because of the
lower temperature of the process. Ideally, by providing more heat (i.e. the steam
reforming concept relies on external heat sources [118]), the syngas molar fraction
curve over the equivalence ratio may increase monotonically.
In general, the operating point around ϕ=2.0 has demonstrated better perfor-
mance in terms of high stability and repeatability, high conversion efficiency and
low soot formation for the foam burners. This aspect is underlined by Fig. 3.20a,
where the H2 content is maximised at ϕ=2.0 for burner A at P= 7 kW. High
water content is exhibited at lower ϕ and high HC content is observed at higher
ϕ. Both water and light hydrocarbons act as hydrogen sinks. Also the conversion
efficiency, defined by Eq. 3.1, peaked around the same condition, as illustrated
by Fig. 3.20b. It is also interesting to note the consistent anomaly in the hydro-
carbon molar fraction trend at around ϕ=2.5, especially for Fig. 3.11b, 3.14b,
3.16b and 3.19b. Such behaviour may need further investigations and analysis.
67
3.2 Parametric investigations
equivalence ratio [-]
dr
y
m
o
la
r
fra
ct
io
n
[%
]
1 1.5 2 2.5 30
5
10
15
20
25
30
H2 exp
CO exp
CO2 exp
H2 num,TAD
CO num,TAD
CO2 num,TAD
(a)
equivalence ratio [-]
dr
y
m
o
la
r
fra
ct
io
n
[%
]
1 1.5 2 2.5 30
0.5
1
1.5
2
2.5
3
3.5
CH4 exp
C2H2 exp
C2H4 exp
C2H6 exp
(b)
Figure 3.11: Calculated equilibrium at the adiabatic flame temperature and the
experimental mole fractions (on a dry basis) for a inlet air/fuel temperature of
423 K at various equivalence ratio for n-heptane reforming at P=6 kW for burner
A. (a) Hydrogen, carbon monoxide and carbon dioxide. (b) Measured methane,
acetylene, ethylene and ethane.
68
3.2 Parametric investigations
equivalence ratio [-]
dr
y
m
o
la
r
fra
ct
io
n
[%
]
1 1.5 2 2.5 30
5
10
15
20
25
30
H2 exp
CO exp
CO2 exp
H2 num,TAD
CO num,TAD
CO2 num,TAD
(a)
equivalence ratio [-]
dr
y
m
o
la
r
fra
ct
io
n
[%
]
1 1.5 2 2.5 30
0.5
1
1.5
2
2.5
3
3.5
CH4 exp
C2H2 exp
C2H4 exp
C2H6 exp
(b)
Figure 3.12: Calculated equilibrium at the adiabatic flame temperature and the
experimental mole fractions (on a dry basis) for a inlet air/fuel temperature of
423 K at various equivalence ratio for n-heptane reforming at P=7 kW for burner
A. (a) Hydrogen, carbon monoxide and carbon dioxide. (b) Measured methane,
acetylene, ethylene and ethane.
69
3.2 Parametric investigations
equivalence ratio [-]
dr
y
m
o
la
r
fra
ct
io
n
[%
]
1 1.5 2 2.5 30
5
10
15
20
25
30
H2 exp
CO exp
CO2 exp
H2 num,TAD
CO num,TAD
CO2 num,TAD
(a)
equivalence ratio [-]
dr
y
m
o
la
r
fra
ct
io
n
[%
]
1 1.5 2 2.5 30
0.5
1
1.5
2
2.5
3
3.5
4
CH4 exp
C2H2 exp
C2H4 exp
C2H6 exp
(b)
Figure 3.13: Calculated equilibrium at the adiabatic flame temperature and the
experimental mole fractions (on a dry basis) for a inlet air/fuel temperature of
423 K at various equivalence ratio for n-heptane reforming at P=8 kW for burner
A. (a) Hydrogen, carbon monoxide and carbon dioxide. (b) Measured methane,
acetylene, ethylene and ethane.
70
3.2 Parametric investigations
equivalence ratio [-]
dr
y
m
o
la
r
fra
ct
io
n
[%
]
1 1.5 2 2.5 30
5
10
15
20
25
30
H2 exp
CO exp
CO2 exp
H2 num,TAD
CO num,TAD
CO2 num,TAD
(a)
equivalence ratio [-]
dr
y
m
o
la
r
fra
ct
io
n
[%
]
1 1.5 2 2.5 30
0.5
1
1.5
2
2.5
3
3.5
4
CH4 exp
C2H2 exp
C2H4 exp
C2H6 exp
(b)
Figure 3.14: Calculated equilibrium at the adiabatic flame temperature and the
experimental mole fractions (on a dry basis) for a inlet air/fuel temperature of
423 K at various equivalence ratio for n-heptane reforming at P=6 kW for burner
C. (a) Hydrogen, carbon monoxide and carbon dioxide. (b) Measured methane,
acetylene, ethylene and ethane.
71
3.2 Parametric investigations
equivalence ratio [-]
dr
y
m
o
la
r
fra
ct
io
n
[%
]
1 1.5 2 2.5 30
5
10
15
20
25
30
H2 exp
CO exp
CO2 exp
H2 num,TAD
CO num,TAD
CO2 num,TAD
(a)
equivalence ratio [-]
dr
y
m
o
la
r
fra
ct
io
n
[%
]
1 1.5 2 2.5 30
0.5
1
1.5
2
2.5
3
3.5
CH4 exp
C2H2 exp
C2H4 exp
C2H6 exp
(b)
Figure 3.15: Calculated equilibrium at the adiabatic flame temperature and the
experimental mole fractions (on a dry basis) for a inlet air/fuel temperature of
423 K at various equivalence ratio for n-heptane reforming at P=7 kW for burner
C. (a) Hydrogen, carbon monoxide and carbon dioxide. (b) Measured methane,
acetylene, ethylene and ethane.
72
3.2 Parametric investigations
equivalence ratio [-]
dr
y
m
o
la
r
fra
ct
io
n
[%
]
1 1.5 2 2.5 30
5
10
15
20
25
30
H2 exp
CO exp
CO2 exp
H2 num,TAD
CO num,TAD
CO2 num,TAD
(a)
equivalence ratio [-]
dr
y
m
o
la
r
fra
ct
io
n
[%
]
1 1.5 2 2.5 30
0.5
1
1.5
2
2.5
3
3.5
CH4 exp
C2H2 exp
C2H4 exp
C2H6 exp
(b)
Figure 3.16: Calculated equilibrium at the adiabatic flame temperature and the
experimental mole fractions (on a dry basis) for a inlet air/fuel temperature of
423 K at various equivalence ratio for n-heptane reforming at P=8 kW for burner
C. (a) Hydrogen, carbon monoxide and carbon dioxide. (b) Measured methane,
acetylene, ethylene and ethane.
73
3.2 Parametric investigations
equivalence ratio [-]
dr
y
m
o
la
r
fra
ct
io
n
[%
]
1 1.5 2 2.5 30
5
10
15
20
25
30
H2 exp
CO exp
CO2 exp
H2 num,TAD
CO num,TAD
CO2 num,TAD
(a)
equivalence ratio [-]
dr
y
m
o
la
r
fra
ct
io
n
[%
]
1 1.5 2 2.5 30
0.5
1
1.5
2
2.5
3
3.5
4
CH4 exp
C2H2 exp
C2H4 exp
C2H6 exp
(b)
Figure 3.17: Calculated equilibrium at the adiabatic flame temperature and the
experimental mole fractions (on a dry basis) for a inlet air/fuel temperature of
423 K at various equivalence ratio for n-heptane reforming at P=6 kW for burner
D. (a) Hydrogen, carbon monoxide and carbon dioxide. (b) Measured methane,
acetylene, ethylene and ethane.
74
3.2 Parametric investigations
equivalence ratio [-]
dr
y
m
o
la
r
fra
ct
io
n
[%
]
1 1.5 2 2.5 30
5
10
15
20
25
30
H2 exp
CO exp
CO2 exp
H2 num,TAD
CO num,TAD
CO2 num,TAD
(a)
equivalence ratio [-]
dr
y
m
o
la
r
fra
ct
io
n
[%
]
1 1.5 2 2.5 30
0.5
1
1.5
2
2.5
3
3.5
CH4 exp
C2H2 exp
C2H4 exp
C2H6 exp
(b)
Figure 3.18: Calculated equilibrium at the adiabatic flame temperature and the
experimental mole fractions (on a dry basis) for a inlet air/fuel temperature of
423 K at various equivalence ratio for n-heptane reforming at P=7 kW for burner
D. (a) Hydrogen, carbon monoxide and carbon dioxide. (b) Measured methane,
acetylene, ethylene and ethane.
75
3.2 Parametric investigations
equivalence ratio [-]
dr
y
m
o
la
r
fra
ct
io
n
[%
]
1 1.5 2 2.5 30
5
10
15
20
25
30
H2 exp
CO exp
CO2 exp
H2 num,TAD
CO num,TAD
CO2 num,TAD
(a)
equivalence ratio [-]
dr
y
m
o
la
r
fra
ct
io
n
[%
]
1 1.5 2 2.5 30
0.5
1
1.5
2
2.5
3
3.5
CH4 exp
C2H2 exp
C2H4 exp
C2H6 exp
(b)
Figure 3.19: Calculated equilibrium at the adiabatic flame temperature and the
experimental mole fractions (on a dry basis) for a inlet air/fuel temperature of
423 K at various equivalence ratio for n-heptane reforming at P=8 kW for burner
D. (a) Hydrogen, carbon monoxide and carbon dioxide. (b) Measured methane,
acetylene, ethylene and ethane.
76
3.2 Parametric investigations
(a)
equivalence ratio [-]
co
n
ve
rs
io
n
ef
fic
ie
n
cy
[%
]
1 1.5 2 2.5 3
20
30
40
50
60
70
(b)
Figure 3.20: (a) H2 distribution in hydrogen, water and hydrocarbons present
in n-heptane reformate as a function of equivalence ratio based on processing of
the measured dry molar fractions. (b) Conversion efficiency defined by Eq. 3.1
against the equivalence ratio for n-heptane. Conditions as in Fig. 3.12.
77
3.2 Parametric investigations
3.2.3 Porous material effect
This section focus on different porous materials for the reforming stage. Fig.
3.21 compares the experimental results obtained from a similar investigation as
the one in Section 3.2.2 (ϕ ranging from 1.1 to 3.0 at P=7 kW), but carried
out in burners A, B, C and D. The only significant enhancement in the reformer
performance has been demonstrated by burner D, which uses a packed bed of
alumina beads, as demonstrated by Fig. 3.22. The maximum deviation, calcu-
lated as the standard deviation of three tests carried out for some conditions,
resulted to be ±0.5% vol. from the value reported in the graphs. Therefore, this
enhancement is especially meaningful at equivalence ratios higher than ϕ=2.2.
Fig. 3.21 shows a higher molar fraction of H2 and CO for very rich equivalence
ratios, along with a lower hydrocarbon content compared to the foam burners.
Specifically, the conversion efficiency peaks at 75% at ϕ=2.5 with H2, CO and
CO2 experimental dry molar fractions of 19.3%, 19.9% and 2.3% respectively for
burner D, compared to a conversion efficiency of 57.9% and 15.0%, 17.1% and
2.9% molar fractions for burner B, 56.7% and 54.7% for burner A and C respec-
tively (Fig. 3.22) at the same equivalence ratio. The temperatures shown in Fig.
3.23 may aid the understanding of this trend; it appears that the beads lead to
higher reforming temperature (by 100 K) compared to the foam matrices. The
thermal conductivity of Al2O3 is higher compared to ZrO2, as listed in Table
2.1, and this may enhance the heat recirculation and the superadiabatic effect.
Therefore, as shown by the experimental results, higher reforming temperatures
lead to a lower presence of HCs in the reformate, resulting in higher H2 and CO
molar fraction. The other parameters such as porosity and foam material (burner
A, B and C) seem to not affect appreciably the reformate composition.
78
3.2 Parametric investigations
equivalence ratio [-]
H
2
dr
y
m
o
la
r
fra
ct
io
n
[%
]
1 1.5 2 2.5 3
2
4
6
8
10
12
14
16
18
20
22
burner A
burner B
burner C
burner D
(a)
equivalence ratio [-]
CO
dr
y
m
o
la
r
fra
ct
io
n
[%
]
1 1.5 2 2.5 3
2
4
6
8
10
12
14
16
18
20
22
burner A
burner B
burner C
burner D
(b)
79
3.2 Parametric investigations
equivalence ratio [-]
to
ta
lC
1
dr
y
m
o
la
r
fra
ct
io
n
[%
]
1 1.5 2 2.5 30
2
4
6
8
10 burner A
burner B
burner C
burner D
(c)
Figure 3.21: Experimental mole fractions (on a dry basis) of (a) H2, (b) CO and
(c) total C1 (contributions from CH4, C2H2, C2H4 and C2H6) for burners A, B,
C and D at conditions as in Fig. 3.12.
80
3.2 Parametric investigations
equivalence ratio [-]
Co
n
ve
rs
io
n
ef
fic
ie
n
cy
[%
]
1 1.5 2 2.5 3
20
30
40
50
60
70
80
burner A
burner B
burner C
burner D
Figure 3.22: Conversion efficiency, as defined by Eq. 3.1, for burners A, B, C and
D at conditions as in Fig. 3.12.
equivalence ratio [-]
co
n
ve
rs
io
n
ef
fic
ie
n
cy
[%
]
T
[K
]
1 1.5 2 2.5 3
20
40
60
80
100
120
600
800
1000
1200
Burner A - ηLHV
Burner D - ηLHV
Burner A - T3
Burner D - T3
Figure 3.23: Exit temperature (T3) and conversion efficiency, as defined by Eq.
3.1, for burners A and D at conditions as in Fig. 3.12.
81
3.2 Parametric investigations
A further investigation on the reforming layer with n-heptane concerns the
length of the porous medium. Burners E and F were set for this purpose. The
reformate component molar fraction graphs (Fig. 3.24) demonstrate higher values
of CO and H2 and lower molar fraction of C1 of burner E compared to burner F.
This may be due to the larger amount of heat captured and therefore recirculated
by a longer matrix, but also to the longer residence time in a hot environment
for the partial oxidation products. However, the limited improvement of the
performance observed, suggested that any further tests with longer pieces of foam
would not be very productive.
82
3.2 Parametric investigations
equivalence ratio [-]
dr
y
m
o
la
r
fra
ct
io
n
[%
]
1.5 2 2.5
5
10
15
20
25
H2 burner E
H2 burner F
CO burner E
CO burner F
CO2 burner E
CO2 burner F
(a)
equivalence ratio [-]
dr
y
m
o
la
r
fra
ct
io
n
[%
]
1.5 2 2.5
0
2
4
6
8
10
C1 burner E
C1 burner F
(b)
Figure 3.24: Experimental mole fractions (on a dry basis) for an inlet air/fuel
temperature of 423 K at various equivalence ratios for n-heptane reforming at
P=7 kW for burners E and F. (a) Hydrogen, carbon monoxide and carbon dioxide.
(b) total C1 (contributions from CH4, C2H2, C2H4 and C2H6).
83
3.2 Parametric investigations
3.2.4 Burner size effect
The results from the two burners of different diameter have been compared. The
same porous material (30 ppi ZrO2 foam) has been used for the reforming layer.
The results are displayed in Fig. 3.25. Hydrogen molar fractions are very similar
up to an equivalence ratio of 1.7. For richer equivalence ratios, burner A develops
a higher hydrogen presence in the syngas. For CO and CO2, the trend is the
opposite. Burner E shows a higher concentration of carbon monoxide and carbon
dioxide compared to burner A up to an equivalence ratio of 1.8. From this point
onwards the two curves overlap. This behaviour is mirrored by the C1 graph,
which shows a slight higher hydrocarbon presence in burner E syngas. Since
the exit temperatures recorded by the themocouple T3 are very similar for both
burners (Fig. 3.26), a possible reason is a better mixture uniformity in the smaller
burner, demonstrated in Section 3.1.1.
84
3.2 Parametric investigations
equivalence ratio [-]
dr
y
m
o
la
r
fra
ct
io
n
[%
]
1 1.5 2 2.5 30
5
10
15
20
25
30
H2 exp burner A
CO exp burner A
CO2 exp burner A
H2 exp burner E
CO exp burner E
CO2 exp burner E
(a)
equivalence ratio [-]
dr
y
m
o
la
r
fra
ct
io
n
[%
]
1 1.5 2 2.5 30
2
4
6
8
10
12
C1 burner A
C1 burner E
(b)
Figure 3.25: Experimental mole fractions (on a dry basis) for an inlet air/fuel
temperature of 423 K at various equivalence ratios for n-heptane reforming at
P=7 kW for burners A and E. (a) Hydrogen, carbon monoxide and carbon dioxide.
(b) total C1 (contributions from CH4, C2H2, C2H4 and C2H6).
85
3.2 Parametric investigations
equivalence ratio [-]
T 3
[K
]
1 1.5 2 2.5 3800
1000
1200
burner A
burner E
Figure 3.26: Exit temperature (T3) for burners A and E at condition as in Fig.
3.12.
86
3.3 Heat losses
3.3 Heat losses
Heat losses were assessed by means of a energy balance between the inlet species
and the products given by the Gas Chromatograph analyses. For this purpose,
the NASA database of species enthalpies has been used. The polynomial fits,
calculated at the measured exit temperature, have been used to infer the mixture
enthalpy, which has been then multiplied by the molar flow rate of the products.
The difference between the energy stored in the reactants and the energy stored
in the products gives the energy lost through radiation from the porous matrix
at the top of the burner in the downstream direction and through conduction
and convection at the outer wall of the reactor. Fig. 3.27 shows the percentage
of heat losses versus the thermal power input. At low firing rate the heat losses
are around 34% of the energy of the fuel, whereas at higher firing rates this ratio
drops to 10%. It is straightforward to devise ways to reduce these heat losses, for
example by preheating the atomising air (coil tube wrapped around the burner)
or the fuel or to produce steam for a potential syngas clean-up stage.
thermal load [kW]
he
at
lo
ss
[%
]
2 4 6 8 10 12
10
20
30
40
Figure 3.27: Heat losses as a percentage of the thermal load vs. thermal load.
Burner A and ϕ=2.0.
87
3.4 Soot emission
3.4 Soot emission
The soot emission has been measured with a filter-method machine, as described
by Section 2.1.8. The measurement gave the value of the FSN , defined by Eq.
2.1, which is plotted vs. the equivalence ratio in Fig. 3.28. The measurement
gives only a value which can be used as a qualitative trend of the particulate for-
mation. No indications about the size and the number of particles can be obtained
with this configuration. The trend shows that soot appears at an equivalence ra-
tio of ϕ=2.0-2.1 and it constantly increases with richer flames. Moreover, burner
D presents less particulate emission compared to burner A. The main reason for
this difference may be the lower presence of C2 hydrocarbons for n-heptane re-
forming in burner D compared to burner A, as highlighted by the comparison
between Fig. 3.18b and Fig. 3.12b, since C2 hydrocarbons are soot precursors
[119].
equivalence ratio [-]
FS
N
1.5 2 2.5 30
0.5
1
1.5
2
burner A
burner D
Figure 3.28: Soot emissions measured by a filter-method machine (FSN defined
by Eq. 2.1) for n-heptane over a range of equivalence ratio for burner A and D
at P=7 kW.
88
3.5 Discussion
3.5 Discussion
N -heptane reforming to syngas has been investigated by Ellzey et al. [4] through
the filtration wave mode and the experimental results may be compared with this
work. The porous layer used by those authors was a packed bed of alumina pellets
and the comparison with the packed bed of alumina beads (burner D) used in this
paper leads to significant similarities. In [4], both the equivalence ratio and the
flow velocity (thermal load) effect on the reformer efficiency have been examined.
The thermal load impact assessed by [4] matches the trend observed in a previous
paper [61]. Pertinent to this work, the equivalence ratio effect described in the
previous section is also compatible with the filtration-wave reformer [4] trends. In
both investigations, H2 and CO molar fractions increase steeply with ϕ, followed
by a moderate increase at high equivalence ratios. In [4] CO molar fraction at
ϕ=3.0 was 22% for a flow velocity of 60 cm/s, while in this paper it reaches 20%
for a velocity of about 30 cm/s. Ellzey et al. [4] also introduced the hydrogen and
carbon monoxide conversion efficiency, defined as the molar flow rate of H2 and
CO over the molar flow rate of the hydrogen and carbon bound in the reactants.
Results are very similar with ηH2=75% and ηCO=90% and 88% for [4] and this
work, respectively. A qualitative soot inspection in [4] revealed a very low soot
formation for 1.5<ϕ<2.0, as stated in the soot emission section of this paper.
Therefore the present stable-flame n-heptane reforming at a two-layer porous
medium yields similar syngas quality to the filtration wave technique.
As demonstrated in Section 3.2.3, the porous material length affects the syngas
characteristics in n-heptane reforming. Numerical simulations [120; 121] suggest
that in the partial oxidation reforming the fuel goes through combination of
partial and complete oxidation reactions. The water produced downstream drives
steam reforming and produces H2. Also, in [62] this two-stage process is noticed
by observing a first quick thermal decomposition of methane, characterized by
rapid consumption of oxygen and hydrogen and water formation, followed by the
steam reforming of methane reacting with water producing hydrogen and carbon
monoxide in an endothermic reaction. The analysis carried out [62] shows that
the first stage characteristic time is around 20 ms whereas the steam reformation
stage is highly sensitive to temperature and a difference of 300 K leads to a
89
3.6 Chapter summary
reaction characteristic time shifting from 102 to 10-1 s. Being an endothermic
process, the presence of a hot and thicker material may enhance this second stage
of the reforming process.
3.6 Chapter summary
Various burner sizes, materials, equivalence ratios and thermal loads have been
explored for evaluating reforming of n-heptane. Two burners have been tested in
order to assess the homogeneity of the n-heptane/air distribution. The 70 mm
ID burner demonstrated a good vapour distribution at P=7 kW at ϕ=2.0, 2.5
and 3.0. The 100 mm ID burner showed a higher difference between the centre
and the peripheral area of the cross section, reaching a divergence up to 15% at
ϕ=3.0. However, the syngas composition over the reforming layer for the same
burner resulted to be uniform within 10% from the mean value.
The operating points of one foam burner (burner C) has been investigated
and it verified the establishment of flame speed higher than the free-flame at the
corresponding condition. Moreover, the equivalence ratio of 2.0 resulted to be
the broadest steady-flame operating zone.
The parametric investigation (thermal load, equivalence ratio, porous material
and burner size) highlighted the close dependency of the conversion efficiency to
the process temperature. In all cases, the parameter providing a better heat recir-
culation and therefore a stronger superadiabatic effect, lead to a condition close
to thermodynamic equilibrium. Specifically, a high thermal load and the bead
burner (burner D) demonstrated the best performance. Furthermore, the equiva-
lence ratio effect investigation showed a peak in the conversion efficiency around
ϕ=2.0 for the foam burners. From this value onwards, the syngas composition
moves away from equilibrium and light hydrocarbons start to be at significant
molar fractions. Eventually, the burner size did not affect heavily the reformer
output; a slight lower performance of the 100 mm ID burner seems to be due to
the lower homogeneity of the mixture distribution.
Finally, the heat losses and the soot emission have been also assessed. The 70
mm ID burner demonstrated a minimum of 10% of the input fuel energy lost by
90
3.6 Chapter summary
conduction and radiation and the filter-method machine pointed, qualitatively,
to ϕ=2.0 as the soot limit condition for burner A and D at P=7 kW.
91
Chapter 4
Diesel, Biodiesel and Kerosene
reforming
This chapter deals with the reforming of “real” fuels. The scope of this research
was the reforming of commercially-available fuels and the achievement of this
target is described in the following sections. Diesel has been reformed at various
thermal loads, equivalence ratios and for different burners. Also Jet A-1 Kerosene
and RME Biodiesel has been successfully processed in the same reformer as the
other fuels.
4.1 Diesel
This section contains the diesel processing results. Commercial grade has been
chosen to better demonstrate the feasibility and the performances with the target
fuel set for this research.
4.1.1 Thermal load
A thermal load ranging from P=2 to 12 kW has been tested in burner A at a fixed
equivalence ratio of ϕ=2.0. The flame speed corresponded to a range from 15.2
cm/s to 91.3 cm/s and it is higher compared to the n-heptane experiment because
of the higher preheat temperature T1 used (lower density of air). The results are
reported in Fig. 4.1. As for the n-heptane experiment in Section 3.2.1, the dry
92
4.1 Diesel
molar fraction of hydrogen and carbon monoxide increases constantly starting
from 5.3% and 10.1% and reaching 14.1% and 18.4% respectively. At the same
time carbon dioxide passed from 8.3% at P=2 kW and to 3.8% at P=12 kW.
Also the hydrocarbons detected by the FID embedded in the gas chromatograph
show a steep drop in their molar fraction. Specifically, as illustrated in Fig. 4.1b,
the main components, methane and ethylene, decreased from 1.2% to 0.4% and
from 1.1% to 90 ppm respectively. Acetylene and ethane demonstrated a very
scarce presence over all the tests.
The conversion efficiency, as defined by Eq. 3.1 and graphically shown in Fig.
4.2, starts from 31.1% at P= 2 kW and reaches 71.1% at P=12 kW. The exit
temperature T3 matches this trend, by peaking at 1300 K for P=12 kW. Similar
to the investigation carried out in Section 3.2.1, the distribution of hydrogen in
the products has been assessed. As for Eq. 3.2, the percentage of molecular
hydrogen present in the products passes from 17.8% to 49.6% at P=12 kW, as
depicted in Fig. 4.3. The main issue with diesel reformate analyses is that the
sum of the three components (molecular hydrogen, hydrogen in hydrocarbons
and hydrogen in water) is around 75% of the total hydrogen contained in the fuel
molecule, compared to the 80% obtained with the n-heptane experiments. The
possible explanation is the complexity of the diesel molecule, which may lead to
the formation of a larger quantity of heavy hydrocarbons compared to n-heptane
during the reforming process, which are not detectable by the gas chromatograph.
This explanation finds support also from the difference between the total hydrogen
contained in the products at low and at high temperatures (i.e. thermal loads).
At P=2 kW the sum of the hydrogen spread over the three different components
is 71.8%, whereas at P=12 kW is 75.8%. Therefore, it appears that the high
temperatures achieved at the maximum thermal load of this test enhance the
reaction rates and lead to diesel reforming to “light” components which can be
detected by the current configuration.
93
4.1 Diesel
thermal load [kW]
dr
y
m
o
la
r
fra
ct
io
n
[%
]
2 4 6 8 10 120
5
10
15
20
H2 exp
CO exp
CO2 exp
(a)
thermal load [kW]
dr
y
m
o
la
r
fra
ct
io
n
[%
]
2 4 6 8 10 12
0
0.5
1
1.5
CH4 exp
C2H2 exp
C2H4 exp
C2H6 exp
(b)
Figure 4.1: Experimental mole fractions (on a dry basis) for a inlet air/fuel
temperature of 573 K at various thermal loads for diesel reforming at ϕ=2.0 for
burner A. (a) Hydrogen, carbon monoxide and carbon dioxide. (b) Methane,
acetylene, ethylene and ethane.
94
4.1 Diesel
thermal load [kW]
co
n
ve
rs
io
n
ef
fic
ie
n
cy
[%
]
T
[K
]
2 4 6 8 10 1220
40
60
80
800
1000
1200
Burner A - ηLHV
Burner A - T3
Figure 4.2: Experimental exit temperature T3 and conversion efficiency for diesel
reforming at conditions as in Fig. 4.1.
Figure 4.3: H2 distribution in hydrogen, water and hydrocarbons present in diesel
reformate as a function of thermal load based on processing of the measured dry
molar fractions. Conditions as in Fig. 4.1.
95
4.1 Diesel
4.1.2 Equivalence ratio
Burner A and C have been used to test the equivalence ratio effect on the syngas
obtained from diesel oil. Burner C has been tested from ϕ=1.1 to 2.5, while
the investigation on burner A has been extended up to ϕ=3.0. Steps of 0.1 were
taken and the analyses of the gas samples gave the results depicted in Fig. 4.4 and
Fig. 4.5. Both burners presented a similar trend, also equivalent to the n-heptane
reforming results. However, diesel reformate presents higher molar fraction values
of CO and lower molar fraction values ofH2. This difference is due to the different
H/C ratio of the two fuels (2.28 for C7H16 and 1.92 for C12H23). Specifically to
diesel reformate, H2 and CO peak at ϕ=2.5 with molar fraction values of 14.5%
and 18.4% respectively and similarly the molecular hydrogen in the products over
the hydrogen contained in the reactants ratio (Eq. 3.2) peaks at ϕ=2.4 with a
value of 43.5%, as shown in Fig. 4.6a. The conversion efficiency peaks at ϕ=2.0
with a value of ηLHV =62.7% (Fig. 4.6b) and H2, CO and CO2 experimental dry
molar fractions reach values of 12%, 16.6% and 4.8% respectively at this point.
For richer values of ϕ, the performance drops steeply, as demonstrated by
burner A, and a red flame has been detected on the top of the porous material.
In Fig. 4.7 a quartz tube has been put over the stainless steel tube to visualize
better this effect for Diesel reforming at P=7 kW and ϕ=3.0. In the picture a red
flame coming out from the burner can be noted. It is not due to air entrainment
and it is not a filtration wave reaching the top of the foam as the flame resulted
being stabilized at the layer interface. The preheat temperature T1 has been
raised up to 773 K, by circulating the oxidative air only through the coil wrapped
around the tube, in the attempt to raise the overall reaction temperatures, but
no enhancements have been noticed. Furthermore, this effect was associated with
a steep increase in the soot emission, as demonstrated by Section 4.4. The reason
of this event is not clear, but it seems to be associated with the production of
heavy hydrocarbons in the reforming process, which partially oxidize only at
the exit of the burner. It can also be due to high radiation from the particles
emitted. An investigation with the aid of a spectrometer would confirm one of
these assumptions. Moreover, once the fuel flow has been stopped, the air flowing
96
4.1 Diesel
in the hot porous matrix lead to the oxidation of deposits inside the reforming
layer and a flame was still present for 20-30 seconds more.
A further distinction between n-heptane and diesel reforming in burner A is
the different path taken by the light hydrocarbon formation. n-heptane presents
higher concentrations of acetylene compared to ethylene, whereas for diesel (but
also for Jet A-1 and Bio-diesel, as it will be shown in the following sections)
ethylene is higher than acetylene.
97
4.1 Diesel
equivalence ratio [-]
dr
y
m
o
la
r
fra
ct
io
n
[%
]
1 1.5 2 2.5 30
5
10
15
20
H2 exp
CO exp
CO2 exp
(a)
equivalence ratio [-]
dr
y
m
o
la
r
fra
ct
io
n
[%
]
1 1.5 2 2.5 30
0.5
1
1.5
2
2.5
3
3.5
CH4 exp
C2H2 exp
C2H4 exp
C2H6 exp
(b)
Figure 4.4: Experimental mole fractions (on a dry basis) for a inlet air/fuel
temperature of 573 K at various equivalence ratio for diesel reforming at P=7 kW
for burner A. (a) Hydrogen, carbon monoxide and carbon dioxide. (b) Methane,
acetylene, ethylene and ethane.
98
4.1 Diesel
equivalence ratio [-]
dr
y
m
o
la
r
fra
ct
io
n
[%
]
1 1.5 2 2.5
0
5
10
15
20
H2 exp
CO exp
CO2 exp
(a)
equivalence ratio [-]
dr
y
m
o
la
r
fra
ct
io
n
[%
]
1 1.5 2 2.5
0
0.5
1
1.5
2
CH4 exp
C2H2 exp
C2H4 exp
C2H6 exp
(b)
Figure 4.5: Experimental mole fractions (on a dry basis) for a inlet air/fuel
temperature of 573 K at various equivalence ratio for diesel reforming at P=7 kW
for burner C. (a) Hydrogen, carbon monoxide and carbon dioxide. (b) Methane,
acetylene, ethylene and ethane.
99
4.1 Diesel
(a)
equivalence ratio [-]
co
n
ve
rs
io
n
ef
fic
ie
n
cy
[%
]
1 1.5 2 2.5 3
20
30
40
50
60
70
(b)
Figure 4.6: (a) H2 distribution in hydrogen, water and hydrocarbons present
in diesel reformate as a function of equivalence ratio based on processing of the
measured dry molar fractions. (b) Conversion efficiency defined by Eq. 3.1 against
the equivalence ratio for diesel. Conditions as in Fig. 4.4.
100
4.1 Diesel
Figure 4.7: Flame on the top of the reforming layer of burner A in diesel reforming
at P=7 kW and ϕ=3.0.
101
4.1 Diesel
4.1.3 Porous material effect
The test performed in the previous section has been repeated in burner D (Al2O3
beads). The trend shown in Fig. 4.8 demonstrates a similar behaviour to the one
observed for n-heptane reforming. The syngas quality is enhanced and the curves
describing H2 and CO molar fractions increase monotonically in this configura-
tion, leading to a higher conversion efficiency for ϕ richer than 2.5 compared
to those exhibited by burner A (Fig. 4.9) at these rich values. The broadest
difference between burner A and D has been recorded at ϕ=2.8. At this condi-
tion, burner A presents a dry molar fraction of hydrogen and carbon monoxide
of 9.6% and 12.9% respectively, whereas burner D gives 13.7% and 18.1% respec-
tively. The conversion efficiency passes from 35.6% to 53.8% switching from the
Zirconia foam burner to the Alumina beads reactor at this equivalence ratio.
In the previous section, a test run in burner C has been mentioned. It demon-
strated similar molar fractions to the ones obtained in the Zirconia burner. The
foam burners showed again no significant difference in the syngas composition,
even for different materials.
102
4.1 Diesel
equivalence ratio [-]
H
2
dr
y
m
o
la
r
fra
ct
io
n
[%
]
1 1.5 2 2.5 30
2
4
6
8
10
12
14
16
burner A
burner D
(a)
equivalence ratio [-]
CO
dr
y
m
o
la
r
fra
ct
io
n
[%
]
1 1.5 2 2.5 32
4
6
8
10
12
14
16
18
20
22
burner A
burner D
(b)
103
4.1 Diesel
equivalence ratio [-]
to
ta
lC
1
dr
y
m
o
la
r
fra
ct
io
n
[%
]
1 1.5 2 2.5 30
2
4
6
8
10 burner A
burner D
(c)
Figure 4.8: Experimental mole fractions (on a dry basis) of (a) H2, (b) CO and
(c) C1 (contributions from CH4, C2H2, C2H4 and C2H6) for burner A and D at
conditions as in Fig. 4.4.
equivalence ratio [-]
co
n
ve
rs
io
n
ef
fic
ie
n
cy
[%
]
1 1.5 2 2.5 3
20
30
40
50
60
70
80
burner A
burner D
Figure 4.9: Conversion efficiency, as defined by Eq. 3.1, for burners A and D at
condition as in Fig. 4.4.
104
4.1 Diesel
4.1.4 Heat losses
The heat losses have been assessed using the syngas composition and exit temper-
ature data from Section 4.1.1. The energy lost has been calculated through the
difference between the energy stored in the reactants and the energy stored in the
products, analogously to the procedure explained in Section 3.27. Fig. 4.10 out-
lines the percentage of heat losses versus the thermal power input. At low firing
rate the heat losses are around 41% of the energy of the fuel, whereas at higher
firing rates this ratio drops to 16%. The difference with the same investigation
performed on n-heptane is maybe due to the less accurate carbon balance between
reactants and products, as explained in Section 4.1.1. The missing species may
account for the energy difference in the heat losses analyses between n-heptane
and diesel.
thermal load [kW]
he
at
lo
ss
[%
]
2 4 6 8 10 120
10
20
30
40
50
Figure 4.10: Heat losses for burner A for a thermal load ranging from P=2 to 12
kW and ϕ=2.0 at condition as in Fig. 4.4.
105
4.1 Diesel
4.1.5 Endurance test
An endurance test has been carried out in burner A at P=7 kW and ϕ=2.0. The
burner demonstrated a steady global behaviour over the test, showing no oscilla-
tion in flows and flame position. The test lasted 150 minutes and gas samples were
extracted every 30 minutes, which is the time needed by the GC to process one
sample. The results are shown in Fig. 4.11. The variation of hydrogen, carbon
monoxide and carbon dioxide are limited to 8% from the average value. After
60 and 150 minutes the biggest divergence from the average value is detected.
The hydrocarbons also presents a consistent trend, especially for methane and
ethylene which are the main HC detected in the gas samples. Consequently, the
conversion efficiency results stable around an average value of 63%, as displayed
in Fig. 4.12.
time [min]
dr
y
m
o
la
r
fra
ct
io
n
[%
]
0 50 100 1500
5
10
15
20
H2 exp
CO exp
CO2 exp
Figure 4.11: H2, CO and CO2 experimental dry molar fractions of diesel reform-
ing at P=7 kW and ϕ=2.0 over time at condition as in Fig. 4.4.
106
4.1 Diesel
time [min]
co
n
ve
rs
io
n
ef
fic
ie
n
cy
[%
]
0 50 100 15020
30
40
50
60
70
80
90
100
Figure 4.12: Conversion efficiency of diesel reforming at P = 7 kW and ϕ=2.0
over time at condition as in Fig. 4.4.
107
4.2 Biodiesel
4.2 Biodiesel
RME biodiesel has also been tested. The equivalence ratio effect has been in-
vestigated up to ϕ=3.0 for a thermal load of P=7 kW in burner A. The results
are shown in Fig. 4.13. Biodiesel data show a similar behaviour compared to
diesel reforming. H2 and CO dry molar fractions keep increasing up to ϕ=2.3
and then the performance decreases considerably. Consequently, the conversion
efficiency resulted maximum at ϕ=2.2 with ηLHV =64.9% with H2, CO and CO2
experimental dry molar fractions reaching values of 14.0%, 19.1% and 4.4% re-
spectively. The same test has been run in burner C. Fig. 4.14 shows the results,
which are comparable with burner A. Burner A provided higher syngas quality
between 2.2<ϕ<2.7 with conversion efficiencies higher than about 10% in this
range (Fig. 4.15). The HCs presence in the reformate is different compared
to diesel and n-heptane. The difference is the much higher molar fraction of
hydrocarbons at very rich equivalence ratio but also the predominant presence
of ethylene at ϕ higher than 2.7. This trend, together with the lower hydrogen
and carbon monoxide yield compared to other fuels, shows the complexity in the
reforming mechanism of heavy fuels.
The biodiesel test showed a shifted stability range towards richer equivalence
ratios compared to the other fuels investigated. Specifically, the flame managed
to penetrate the flame holder layer up to ϕ=1.7 in burner C and ϕ=1.8 in burner
A. The mechanism leading to this differing behaviour is not clear and it may due
to different kinetics paths taken by an oxygenated fuel like biodiesel compared to
diesel or kerosene during the pre-ignition reactions.
108
4.2 Biodiesel
equivalence ratio [-]
dr
y
m
o
la
r
fra
ct
io
n
[%
]
1 1.5 2 2.5 30
5
10
15
20
H2 exp
CO exp
CO2 exp
(a)
equivalence ratio [-]
dr
y
m
o
la
r
fra
ct
io
n
[%
]
1 1.5 2 2.5 30
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
CH4 exp
C2H2 exp
C2H4 exp
C2H6 exp
(b)
Figure 4.13: Experimental mole fractions (on a dry basis) for a inlet air/fuel
temperature of 523 K at various equivalence ratio for RME biodiesel reforming
at P=7 kW for burner A. (a) Hydrogen, carbon monoxide and carbon dioxide.
(b) Methane, acetylene, ethylene and ethane.
109
4.2 Biodiesel
equivalence ratio [-]
dr
y
m
o
la
r
fra
ct
io
n
[%
]
1 1.5 2 2.5 30
5
10
15
20
H2 exp
CO exp
CO2 exp
(a)
equivalence ratio [-]
dr
y
m
o
la
r
fra
ct
io
n
[%
]
1 1.5 2 2.5 30
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
CH4 exp
C2H2 exp
C2H4 exp
C2H6 exp
(b)
Figure 4.14: Experimental mole fractions (on a dry basis) for a inlet air/fuel
temperature of 523 K at various equivalence ratio for RME biodiesel reforming
at P=7 kW for burner C. (a) Hydrogen, carbon monoxide and carbon dioxide.
(b) Methane, acetylene, ethylene and ethane.
110
4.2 Biodiesel
equivalence ratio [-]
co
n
ve
rs
io
n
ef
fic
ie
n
cy
[%
]
1 1.5 2 2.5 3
20
30
40
50
60
70
burner A
burner C
Figure 4.15: Conversion efficiency, as defined by Eq. 3.1, for biodiesel reforming
in burners A and C at condition as in Fig. 4.13.
111
4.3 Kerosene
4.3 Kerosene
Finally kerosene has been reformed, at P=7 kW, up to an equivalence ratio of
3.0. For this fuel, only burner A has been tested. The conversion efficiency,
shown in Fig. 4.16, peaks at ϕ=2.1 with the value of ηLHV =69.6% and H2,
CO and CO2 experimental dry molar fractions reach values of 13.8%, 18.9%
and 3.2% respectively. The reformate composition follows closely the calculated
equilibrium data up to ϕ=2.0 and afterwards the presence of hydrocarbons in the
experimental results lead to a divergence between the curves. It is noted that the
experimental CO molar fraction follows equilibrium closely up to ϕ=2.3. After a
peak of H2 and CO at ϕ=2.5, the molar fractions start to decrease, as described
also from the equilibrium calculations. The equilibrium calculations could not go
beyond ϕ=2.9, because the adiabatic flame temperature of C12H24 with enthalpy
of formation at 298.15 K of -383.894 kJ/mol [122] could not be calculated by the
method used for this project [116].
This fuel presented a very low concentration ofHCs in the reformate and most
of the molecular hydrogen missing from the equilibrium calculation data is stored
in CH4, with very low concentration of C2 species. Specifically, at ϕ=3.0 the
maximum content of light hydrocarbons has been detected, with 2.9% methane,
0.12% of acetylene and 0.65% and 50 ppm of ethylene and ethane respectively.
Also for this fuel ethylene is predominant over the acetylene molar fraction.
112
4.3 Kerosene
equivalence ratio [-]
dr
y
m
o
la
r
fra
ct
io
n
[%
]
1 1.5 2 2.5 30
5
10
15
20
25
30
H2 exp
CO exp
CO2 exp
H2 num,TAD
CO num,TAD
CO2 num,TAD
(a)
equivalence ratio [-]
dr
y
m
o
la
r
fra
ct
io
n
[%
]
1 1.5 2 2.5 30
0.5
1
1.5
2
2.5
3
3.5
CH4 exp
C2H2 exp
C2H4 exp
C2H6 exp
(b)
Figure 4.16: Calculated equilibrium at the adiabatic flame temperature and the
experimental mole fractions (on a dry basis) for a inlet air/fuel temperature of
523 K at various equivalence ratio for Jet A-1 kerosene reforming at P=7 kW
for burner A. (a) Hydrogen, carbon monoxide and carbon dioxide. (b) Methane,
acetylene, ethylene and ethane.
113
4.4 Soot emission
4.4 Soot emission
Fig. 4.17 gives an indication of the soot content in the sample for the three
different fuels described in this chapter, at P=7 kW for burner A at different
equivalence ratios. The graph represents a qualitative comparison between the
fuels and the soot emission trend, rather than a quantitative evaluation of the
particulate properties (soot number and size) at the different conditions consid-
ered. The main considerations drawn from the graph are, first, that soot starts
to appear in the sample above ϕ=2.0-2.1 for all fuels tested. Secondly, diesel
presents the highest FSN value and hence the largest particulate emission, as
expected. Finally, note the low soot generation observed in biodiesel reforming.
This trend is maybe due to the presence of oxygen atoms in the fuel molecule,
which lowers the rate of particulate formation.
equivalence ratio [-]
FS
N
1.5 2 2.5 30
0.5
1
1.5
2
2.5
n-heptane
Jet A-1 kerosene
Diesel
RME Biodiesel
Figure 4.17: Soot emissions measured by a filter-method machine (FSN defined
by Eq. 2.1) for n-heptane, Diesel oil, Jet A-1 Kerosene and Biodiesel over a range
of equivalence ratio for burner A at P=7 kW.
114
4.5 Discussion
4.5 Discussion
Comparing the results from the three fuels tested in this Chapter, the following
comments can be drawn. The total C1 molar fraction, as contribution from CH4,
C2H2, C2H4 and C2H6 of diesel, kerosene and biodiesel is shown in Fig 4.18.
The trend is similar for the three fuels up to ϕ=2.3. From this value onwards,
higher presence of HC is observed in biodiesel reformate. The complexity of the
fuel molecule seems to play the major role in the reforming mechanism, probably
because of the different components present in the fuel molecule as previously
shown in the GC-MS outputs (Figs. 2.8, 2.9 and 2.10), leading to a high sensitivity
of the reformate composition to the process temperature. Consequently, the
conversion efficiency is slightly different for the three fuels. ηLHV of the three
fuels is compared in Fig. 4.19. The graph shows similar trends and values, with
kerosene demonstrating a higher performance compared to diesel and biodiesel.
As demonstrated previously, the higher hydrogen and carbon monoxide yield is
in kerosene reforming, together with a low HCs emission, lead to the highest
performance among the fuels over all the equivalence ratio investigated.
115
4.5 Discussion
equivalence ratio [-]
C 1
dr
y
m
o
la
r
fra
ct
io
n
[%
]
1 1.5 2 2.5 30
4
8
12
16
Diesel
Jet A-1 kerosene
RME biodiesel
Figure 4.18: Experimental mole fractions (on a dry basis) of C1 (contributions
from CH4, C2H2, C2H4 and C2H6) for burner A at P=7 kW for diesel, biodiesel
and kerosene.
equivalence ratio [-]
co
n
ve
rs
io
n
ef
fic
ie
n
cy
[%
]
1 1.5 2 2.5 3
20
40
60
80 Diesel
RME bio-diesel
Jet A-1 Kerosene
Figure 4.19: Conversion efficiencies of kerosene, diesel and RME biodiesel reform-
ing, as defined by Eq. 3.1, at P=7 kW for burner A.
116
4.5 Discussion
Diesel thermal partial oxidation has been studied experimentally in Ref. [123].
That experimental setup differs in the vaporization stage, reforming stage and in
the ignition strategy. The vaporizer in [123] is a separated section from the rest of
the reactor, the porous matrix is a single layer and the igniter is on the top of the
burner. A ZrO2-10 ppi foam has been implemented in [123], whereas different
dimensions and a porosity of 30 ppi was used in this paper. The operational
differences occur in the preheat temperature and the thermal load (873.15 K
and P= 10.5 kW in [123]). Despite these differences, comparable results were
achieved: the maximum H2 and CO concentration was obtained at ϕ=2.2 with a
molar fraction of 16.0% and 18.5% respectively. As previously shown in Section
3.2, we obtained maximum hydrogen and carbon monoxide molar fraction at
ϕ=2.5 with values of 14.5% and 18.4% respectively. The small difference from
the results of [123] is possibly due to the lower thermal load and the lower preheat
temperature used in this investigation, which lead to slower reaction rates of the
reforming process.
A catalytic partial oxidation reformer has been developed by [124] for an APU
system embedded with a SOFC. The authors obtained a conversion efficiency of
78% for a thermal load of 6 kW and ϕ=2.5, compared to 60% efficiency obtained
in this project for reforming in burner D at P=7 kW and at the same equivalence
ratio. A staged injection (a first lean oxidation followed by a second fuel stream
injected in the hot off-gas without any additional air) is implemented to provide
the system with the heat necessary to the fuel evaporation and catalytic bed
heating. Another experimental demonstration of CPOX of ethanol, isooctane,
hexadecane, synthetic JP8, desulfurized kerosene and desulfurized diesel for solid
oxide fuel cell applications has been performed by [59]. The results obtained by
desulfurized diesel reforming at ϕ=2.2 are H2=25%, CO= 18% and CO2=3%.
It can be noted that, apart from hydrogen, the results are similar to the results
obtained by burner D. The catalytic activity may enhance the production of hy-
drogen, which in this research resulted being “trapped” in the light hydrocarbons
and water.
Many other published works on diesel reforming focus on ATR of diesel oil and
its surrogates, highlighting the sensitivity of catalysts to the temperature profile
of the reactor and to the poisoning species (sulfur) and the degradation of the
117
4.5 Discussion
catalytic bed with time. In [125], a noble metal catalyst has been used to reform
hexadecane and commercial diesel in an ATR reformer. The experimental results
show molar fractions of H2, CO, CO2 and CH4 of 53%, 25%, 18% and 3.2%
respectively, for commercial diesel reforming in a furnace kept at 850 ◦C. Further
energy was needed to produce the large amount of steam used to run the steam
reforming stage of the autothermal process and the WGS reaction. Synthetic
diesel (sulfur-free) and JP-8 (1000 ppmw S) have been reformed in an ATR
reactor in [126]. The effect of H2S, which is produced from the sulfur content
of the fuel at low temperatures (300 ◦C-400 ◦C) and of SO2, formed at higher
temperatures (>600 ◦C) is assessed separately and results to affect dramatically
and irreversibly the hydrogen yield, which dropped from 75% to 40% and 55%
respectively. For this reason, in a similar work [127] a ZnO desulfurizer bed has
been implemented in a 5 kW reactor before the reforming stage. The degradation
of noble catalyst activity has been demonstrated in [128], in which the yield of
H2 dropped from 39% to 30% and CO increased from 6% to 10% after 20 hours
of operation in the oxidative steam reformer of low sulfur (15 ppm) diesel. A
good example of integration of an ATR reformer, a WGSR stage and a PROX
carbon monoxide remover is [72], in which high importance has been attributed
to the homogeneity of the diesel/air/steam mixture. Before the clean-up stages,
the syngas composition obtained by the authors was 38% H2, 10% CO and 14%
CO2.
Steam reforming of desulfurized diesel has been carried out by [129]. Even if
short-run experiments showed good results with over 70% hydrogen concentration
in volume, shortly degradation of the catalyst and heavy deposits on the cata-
lyst surface have been observed. Lower temperature steam reforming operations
were also demonstrated by [130], by using supercritical water which enhance the
solubility of heavy liquid fuels.
Kerosene and bio-diesel non-catalytic reforming results are not available in
the open literature, to the author’s knowledge. The majority of experimental
investigations in this field focused on the autothermal [131; 132; 133] and steam
reforming approach, which resulted in higher performance than the technique
presented here. For example, in [131], dry molar fractions of 38% H2 and of
17% of CO have been obtained by autothermal reforming of desulphurised Jet
118
4.6 Chapter summary
A-1. In [134], steam reforming of kerosene led to hydrogen molar fraction over
70% and carbon monoxide molar fraction at around 10%. Similarly, autothermal
reforming of biodiesel has been achieved by [74] with a dry gas composition after
the ATR reactor of 41.65% hydrogen and 8.62% carbon monoxide.
The slightly poorer performance of the present non-catalytic reformer com-
pared to the catalytic reformer is balanced, in practical applications, by the lower
cost expected by the absence of a catalyst and the more robust behaviour by the
capability to use a great variety of fuels in the same burner.
4.6 Chapter summary
The 70 mm ID burner has been tested for commercial fuel reforming in ceramic
foams and beads (diesel) porous layers. Diesel reforming has been tested for the
thermal load, equivalence ratio and porous material effect with similar general
results as for n-heptane reforming. The diesel reformate proved to be sensitive to
the process temperature and to the equivalence ratio, which cannot be decoupled
in the thermal POX technique used in this project. High thermal loads led
to high reforming temperatures, leading to conversion efficiency up to 71.1%
at the maximum thermal load tested and at ϕ=2.0, conditions for which, the
reformer also showed the lowest heat losses. At this point the hydrogen balance
between reactants and products is matching better than at lower thermal loads.
Further tests demonstrated that the maximum conversion efficiency is achieved at
ϕ=2.0 and the use of alumina beads for the reforming layer (burner D) improved
the hydrogen and carbon monoxide fraction for ϕ higher than 2.5. Finally, an
endurance test has demonstrated a syngas composition fluctuating around the
8% of the mean molar fraction values.
Biodiesel has been successfully reformed to syngas in burner A and C, match-
ing the behaviour showed by the diesel tests at different equivalence ratios. Hy-
drogen and carbon monoxide reach a peak and then start to decrease steeply,
along with a substantial formation of light hydrocarbons, in the highest percent-
age among the four fuel tested in this research. The top conversion efficiency
obtained was at ϕ=2.2 with a value of 64.9%.
119
4.6 Chapter summary
Kerosene reforming has showed the highest conversion efficiency over the fu-
els tested. This is due mainly to the low presence of hydrocarbons in the refor-
mate, with significant presence only of methane. The conversion efficiency was
maximum at ϕ=2.1 with a value of 69.6%. Finally, the soot emission analysis
highlighted the low particulate formation of biodiesel reforming compared to the
other fuel outcome.
Of particular interest are (a) the different pathways taken by the hydrocarbon
formations (ethylene compared to acetylene in n-heptane and diesel, biodiesel,
kerosene reforming), (b) the presence of deposits and sooty flames at ϕ>2.5 in
diesel reforming and (c) the ease of flash-back demonstrated by rich biodiesel
flames compared to the other fuels. These issues require further studies by means
of experimental and numerical investigations to achieve a fuller understanding of
the chemistry involved.
An attempt to reform crude rapeseed oil has been made, but tests failed for
coking of the burner, high deposits and ignition failure. A potential alternative
may be the reforming of a blend of crude oil and kerosene or biodiesel, examining
the stability and the functionality of the reformer. Nevertheless, the reforming
of a carbon neutral feedstock such as RME Biodiesel in a robust and reliable
device has been demonstrated in this Chapter, showing only minor drawbacks in
comparison to the major benefits obtained.
120
Chapter 5
Syngas clean-up
In this chapter, some experiments concerning syngas clean-up have been under-
taken. The target was to test the possibility of reducing the carbon monoxide
content and transforming the remaining hydrocarbons to extra hydrogen in order
to approach the requirements of a high-temperature PEM a fuel cell.
5.1 n-heptane
Single steps were undertaken to assess the improvement obtained by implement-
ing each stage and they are described by Fig. 5.1. The columns represent the
experimental mole fractions of hydrogen, carbon monoxide, carbon dioxide and
the total C1 (contributed from methane, acetylene, ethylene and ethane), while
the symbols represent the corresponding calculated equilibrium results at the
adiabatic flame temperature.
The first set of data depicts the results for the previously described n-heptane
for burner E (a). The second set is the syngas composition with water injection
(m˙H2O/m˙C7H16=1) without any catalyst (b). The experimental data show a slight
increase in H2 dry molar fraction associated with an increase in CO2 and decrease
of CO molar fraction for this configuration in agreement with a “thermal” water-
gas shift reaction of the reformate.
Afterwards, the SR catalyst has been inserted in the reactor and, firstly, the
reaction has been driven by the water produced by the partial oxidation process
only (no extra water injected - configuration (c) in Fig. 5.1). In this case, the
121
5.1 n-heptane
molar fraction of HCs is significantly reduced (from 4.5% to 1% C1) through the
reaction (5.1):
CmHn +mH2O = mCO + (n/2 +m)H2 (5.1)
while H2 molar fraction increased from 12.9% to 17.9%. A similar improvement
has been observed by injecting extra water (d), but in this case the G-90 catalyst
acted also as a WGSR: H2 reached 20.8% and CO2 achieved 12% whereas CO
dropped to 7.3% while the HC presence is only moderately reduced.
Eventually, the high temperature water-gas shift reaction stage has also been
implemented injecting a H2O/C7H16 mass flow ratio of 1 (configuration (e)) and
subsequently raised to 2 (f). The final composition of the syngas is 22.3% H2,
5.4% CO and 15.6% CO2 with only 0.39% of CH4, 70 ppm of C2H2 and 10-20
ppm of C2H4 and C2H6 left un-reformed.
The same operation has been carried out for n-heptane reforming in the same
burner at the same thermal load but at ϕ=2.5. The results displayed in Fig.
5.2 illustrate similar trends as those at ϕ=2.0, with the sole difference of higher
levels of the molar fractions of the syngas components over all the different steps.
In more detail, the final syngas composition for configuration (f) is: 27.1% H2,
8.7% CO, 11.8% CO2, 0.98% CH4, 0.02% C2H2, 40 ppm of C2H4 and 16 ppm of
C2H6.
From the comparison with the equilibrium calculations it can be noted that the
experimental and the calculated molar fractions of hydrogen differ by less than 2%
for conditions (c), (d), (e) and (f). In particular, at condition (f) the experimental
H2 dry molar fraction is 22.3% and the equilibrium value is 22.6%. The biggest
gap is observed at condition (b), in which the experimental H2 dry molar fraction
is 12.9% and the equilibrium value is 19.1%. This is possibly due to the lack
of energy (resulting in lower temperature) to run the steam reforming process
(endothermic reaction). When the catalyst is added, the steam reforming process
is enhanced since the catalytic effect consists in the acceleration of the rates of
the chemical reactions at low temperatures. Experimental CO and CO2 dry
molar fractions in the configurations involving a catalytic activity (configurations
(c), (d), (e) and (f)) are respectively lower and higher compared to the calculated
equilibrium. Moreover, the equilibrium calculation carried out at T3 instead of the
122
5.1 n-heptane
adiabatic temperature shows an opposite behaviour, with experimental hydrogen
molar fraction far from the calculated value and CO and CO2 experimental results
comparable with the calculated data. This disparity can be observed also in Fig.
5.2, for configurations (c), (d) and (e).
At ϕ=2.5, the experimental hydrogen molar fraction is again very close to
the calculated equilibrium for configurations (c), (d), (e) and (f), in which the
catalytic bed is implemented. The best match is achieved for configuration (e)
with an experimental H2 dry molar fraction of 27.7% and a calculated equilibrium
value of 27.4%.
Photographs of the syngas flame ignited on top of the burner are displayed
in Fig. 5.3, 5.4 and 5.5. The colour of the flame gives a qualitative indication of
the species present in the syngas produced by the reformer. Passing from Fig.
5.3 to Fig. 5.4, the flame in configurations (c) and (d) is less sooty and it starts
to be slightly blue because of the substantial reduction of hydrocarbons present
in the off-gas. Eventually, with the implementation of the WGSR (Fig. 5.5)
in configurations (e) and (f), the major component is hydrogen and hence the
flame assumes a pale blue colour, difficult to visually detect. Furthermore, in
configurations (e) and (f), the almost sole presence of hydrogen as species makes
the flame much shorter than the flame of the other configurations. This occurs
because H2 diffuses more rapidly than most other hydrocarbon fuels do and in a
laminar non-premixed flame (as the one in Fig. 5.5) molecular diffusivity plays
the major role [135], together with the higher flame speed of hydrogen/air flames
[136].
123
5.1 n-heptane
Figure 5.1: Calculated equilibrium at the adiabatic flame temperature and the
experimental mole fractions (on a dry basis) for steam reforming and water-
gas shift reaction experimental results of n-heptane reformate obtained at P=7
kW and ϕ=2.0 in burner E. (a) ceramic inert foam only; (b) water injection
(m˙H2O/m˙C7H16=1); (c) steam reforming with no extra water injected; (d) steam
reforming with water injection m˙H2O/m˙C7H16=1); (e) steam reforming and WGSR
with water injection (m˙H2O/m˙C7H16=1); (f) steam reforming and WGSR with
water injection (m˙H2O/m˙C7H16=2).
124
5.1 n-heptane
Figure 5.2: Calculated equilibrium at the adiabatic flame temperature and the
experimental mole fractions (on a dry basis) for steam reforming and water-
gas shift reaction experimental results of n-heptane reformate obtained at P=7
kW and ϕ=2.5 in burner E. (a) ceramic inert foam only; (b) water injection
(m˙H2O/m˙C7H16=1); (c) steam reforming with no extra water injected; (d) steam
reforming with water injection m˙H2O/m˙C7H16=1); (e) steam reforming and WGSR
with water injection (m˙H2O/m˙C7H16=1); (f) steam reforming and WGSR with
water injection (m˙H2O/m˙C7H16=2).
125
5.1 n-heptane
Figure 5.3: Off-gas flame at the burner’s outlet for n-heptane at P=7 kW and
ϕ=2.0 in burner E at configuration (a) and (b).
126
5.1 n-heptane
Figure 5.4: Off-gas flame at the burner’s outlet for n-heptane at P=7 kW and
ϕ=2.0 in burner E at configuration (c) and (d).
127
5.1 n-heptane
Figure 5.5: Off-gas flame at the burner’s outlet for n-heptane at P=7 kW and
ϕ=2.0 in burner E at configuration (e) and (f).
128
5.1 n-heptane
5.1.1 Endurance test
The functionality of the SR catalyst (Su¨d-chemie G 90-EW) has been also assessed
over time for configuration (c). The flow conditions have been set to thermal
load P=7 kW and equivalence ratio ϕ=2.5. Gas samples were analyzed every 15
minutes and the data are displayed in Fig. 5.6. The result is that the hydrogen
molar fraction reduced from 20.8% to 19.4%, carbon monoxide passes from 17.2%
to 16.9% and carbon dioxide stayes constant around 4.5% after 75 minutes of
running showing a slight degradation of the catalyst with time.
time [min]
dr
y
m
o
la
r
fra
ct
io
n
[%
]
0 15 30 45 60 750
5
10
15
20
25
30
H2 exp
CO exp
CO2 exp
Figure 5.6: Experimental mole fractions (on a dry basis) over time for steam
reforming (configuration (c) in Fig. 5.1) of n-heptane at P=7 kW and ϕ=2.5 in
burner E.
129
5.2 Biodiesel
5.2 Biodiesel
An attempt of syngas clean-up has been also made on the RME biodiesel refor-
mate. The main issue concerns in the water injection. It has been noted that
with the current configuration, the injection of water together with the fuel at
the nozzle level, as explained in Section 2.1.4, caused the reduction of the tem-
perature T1 below the necessary evaporation temperature for biodiesel, resulting
in high oil condensation on the burner’s walls. However, transitory results have
been achieved for configuration (f). For 1-2 minutes the gas samples demon-
strated the functionality of the catalyst stages, before the first stage temperature
dropped. Specifically, in Fig. 5.7 three different steps in the clean-up process are
displayed. For the configurations (a) and (c), the flame is stable since no water
was injected. The first one is the syngas composition obtained in burner E at
P=7 kW and ϕ=2.5. Hydrogen molar fraction is around 12.6%, carbon monoxide
and carbon dioxide 15.4% and 5.4% respectively and the total contribution from
the hydrocarbons reached 6.03%. By adding the steam reforming catalyst only
without any extra water injection, as in configuration (c), the hydrocarbon pres-
ence significantly decreased to a 0.53%. Also, hydrogen molar fraction reached
almost 15% and carbon monoxide and carbon dioxide shifted to 13.3% and 6.02%
respectively, demonstrating a WGSR operation of that catalytic bed. Eventually,
the WGSR catalyst has been added as in configuration (f). The final composition
obtained is 18.9% H2, 4.5% CO, 16.4% CO2, 0.34% CH4, 0.002% C2H2, 40 ppm
of C2H4 and 20 ppm of C2H6.
130
5.2 Biodiesel
Figure 5.7: Experimental mole fractions (on a dry basis) for steam reforming and
water-gas shift reaction experimental results of biodiesel reformate obtained at
P=7 kW and ϕ=2.5 in burner E. Configurations (a), (c) and (f), as defined in
Fig. 5.1.
131
5.3 Chapter summary
5.3 Chapter summary
The implementation of SR and WGSR stages along with water injection has been
experimentally demonstrated for n-heptane reformate clean-up. Moreover, the
potential of biodiesel syngas clean-up in the reformer has been introduced, even
if more work is needed to obtain stable operation and better thermal integration
to avoid oil condensation. The experimental results obtained by processing n-
heptane showed consistency with the calculated equilibrium data, especially for
hydrogen molar fraction. Eventually, a slight degradation of the G-90 EW steam
reforming catalyst activity has been observed over 75 minutes of operation.
132
Chapter 6
Conclusions
6.1 Summary
The objectives of this thesis were to design a reactor based on the two-layer
porous inert medium partial oxidation, to study the behaviour of the reformer
varying the characteristic parameters of the system (equivalence ratio, thermal
load, materials) and to demonstrate the flexibility of the reactor with different
heavy liquid commercial fuels. In order to achieve these targets, a compact re-
actor has been designed and investigated by means of temperature detectors and
of a gas analyser. A vapouriser based on a commercial air-assisted nozzle fed
with hot air has been designed and embedded in the reactor and an automotive
spark-plug system has been implemented for the ignition of the air/fuel mixture.
Temperatures were measured at three different stages with K-type and R-type
thermocouple to monitor the temperature of the pre-vapouriser, of the porous
medium interface and of the gas samples and a Gas Chromatograph calibrated
for H2, O2, N2, CO, CO2, CH4, C2H2, C2H4 and C2H6 measured the species
present in the reformate. Different fuels have been successfully reformed.
First of all, n-heptane has been processed in several burner setups with dif-
ferent material, diameter and length. The most successful configuration was the
alumina bead packed bed burner, which displayed the highest conversion efficien-
cies over the equivalence ratio values investigated. Specifically, the conversion
efficiency peaked at 75% at ϕ=2.5 for P=7 kW with H2, CO and CO2 exper-
imental dry molar fractions of 19.3%, 19.9% and 2.3%. Preliminary tests have
133
6.1 Summary
been carried out for this fuel to assess the homogeneity of the air/fuel mixture
and demonstrated a good uniformity of fuel vapour, especially for the 70 mm
ID burner. Moreover, the reformate over the 100 mm ID burner surface was
quite uniform in composition. The operating range of burner C was explored
and the equivalence ratio of 2 displayed the broadest range of flame speed that
could be stabilised in the two-layer porous medium reactor. At this equivalence
ratio, the reformate composition approached very closely the equilibrium data,
calculated by means of the NASA CEA database. Composition closer to the
equilibrium were achieved with high thermal loads, which also demonstrated low
heat losses - approximately 10% - calculated from the energy balance between
the products and the reactants. A particulate emission investigation has been
performed, pointing to ϕ=2.0 as the soot limit; above this equivalence ratio soot
was generated.
The reforming investigation has been extended to commercial heavy liquid fu-
els and diesel, kerosene and biodiesel have been successfully converted to hydrogen-
enriched gas. These fuels demonstrated a similar beahaviour to n-heptane, but
diesel and biodiesel exhibited lower conversion efficiency and hydrogen content
due to their low H2 : C ratio and the complexity of the fuel molecule, which
led to slow kinetics, sensitive to the process temperature and leading to signif-
icant light hydrocarbon formation at rich equivalence ratios. Diesel conversion
efficiency peaked at ϕ=2.0 for P=7 kW in burner A with a value of ηLHV =62.7%
and H2, CO and CO2 experimental dry molar fractions reach values of 12%,
16.6% and 4.8% respectively. Biodiesel conversion efficiency resulted to a max-
imum at ϕ=2.2 for P=7 kW in burner A with ηLHV =64.9% with H2, CO and
CO2 experimental dry molar fractions reaching values of 14.0%, 19.1% and 4.4%
respectively. Kerosene was the only fuel which could be compared to equilibrium
calculations and demonstrated close results to the calculated fractions up to an
equivalence ratio of 2. Its efficiency peaked at ϕ=2.1 for P=7 kW in burner A
with the value of ηLHV =69.6% and H2, CO and CO2 experimental dry molar
fractions reach values of 13.8%, 18.9% and 3.2% respectively. The soot investiga-
tion showed that biodiesel has the lowest particulate emissions out of the three
fuels, most likely due to the oxygen content in the molecule.
134
6.2 Suggestion for further research
Finally, steam reforming and water-gas shift reactions have been applied to
the reformate obtained from n-heptane and biodiesel in order to remove light
hydrocarbons and carbon monoxide from the syngas obtained by the partial ox-
idation process. Commercial catalysts have been inserted in the reactor and
different configurations involving water injection have been tested. The exper-
imental results gave confidence in the feasibility of the process by approaching
equilibrium calculations and for n-heptane a H2 mole fraction of 27.1% has been
achieved with 8.7% CO and around 1% C1 at ϕ=2.5. Better thermal integration
in order to avoid sharp temperature drops due to the water injections must be
adopted in potential further research, but the “pale blue” off-gas flame and the
gas analysis demonstrated the functionality of the concept with hydrogen as the
main component of the syngas obtained.
6.2 Suggestion for further research
Further research may focus on experimental and modelling work. On the experi-
mental side, research on better use of the heat produced by the partial oxidation
process would help to improve the global efficiency of the system. Moreover,
more resistant material to thermal cycles should be found and tested and maybe
a combination of foams and beads may lead to use the benefits given by the two
solutions (quick response for the foam and high efficiency for the bead layer).
A very important step forward would be a more detailed analysis of the syngas
composition, which would allow to detect all the species which in this research
were missing from the carbon balance, especially high hydrocarbons. Eventually,
a clean-up stage based on the results here obtained may be implemented and a
integration with a fuel cell system should be attempted.
On the modelling side, a study of the structure of the flame for n-heptane
and possibly for a real fuel (kerosene) would be beneficial in the understand-
ing of the reforming mechanisms and in the main step controlling the reactions.
The experimental data described in this thesis may help validating a potential
computational model.
135
References
[1] U.S. DOE. Annual energy review 2009. Energy Information Administration,
2010. xi, 2
[2] C. Song. Fuel processing for low-temperature and high-temperature fuel
cells: Challenges, and opportunities for sustainable development in the 21st
century. Catalysis Today, 77:17–49, 2002. xi, 5, 6, 11, 14, 15
[3] A.J. Barra, G. Diepvens, J.L. Ellzey, and M.R. Henneke. Numerical study of
the effects of material properties on flame stabilization in a porous burner.
Combustion and Flame, 134:369–379, 2003. xi, 16
[4] M.J. Dixon, I. Schoegl, C.B. Hull, and J.L. Ellzey. Experimental and numer-
ical conversion of liquid heptane to syngas through combustion in porous
media. Combustion and Flame, 154:217–231, 2008. xi, 17, 19, 20, 89
[5] G. Vourliotakis, G. Skevis, M.A. Founti, Z. Al-Hamamre, and D. Trimis.
Detailed kinetic modelling of the T-POX reforming process using a reactor
network approach. International Journal of Hydrogen Energy, 33:2816–
2825, 2008. xi, 20, 21
[6] Paul Smith Customer relations. Private communication. xii, 46
[7] Thomas Starke Sales manager. Private communication. xii, 47
[8] Conoco Phillips. Product information.
http://www.conocophillips.co.uk/EN/fuels/productinfo/pages/index.aspx.
xii, 48
136
REFERENCES
[9] Rotexo. COSILAB. http://www.rotexo.com/cms/index.php. xiii, 60
[10] M. Chaos, A. Kazakov, Z.W. Zhao, and F.L. Dryer. A high-temperature
chemical kinetic model for primary reference fuels. International Journal
of Chemical Kinetics, 39:339–414, 2007. xiii, 19, 39, 60
[11] EIA. Greenhouse gases, climate change, and energy.
http://www.eia.doe.gov/oiaf/1605/ggccebro/chapter1.html. 1
[12] P. Moriarty and D. Honnery. Hydrogen’s role in an uncertain energy future.
International Journal of Hydrogen Energy, 34:31–39, 2009. 1
[13] N.Z. Muradova and T.N. Veziroglu. “Green” path from fossil-based to hy-
drogen economy: An overview of carbon-neutral technologies. International
Journal of Hydrogen Energy, 33:6804–6839, 2008. 2
[14] J. Wee. Contribution of fuel cell systems to CO2 emission reduction in their
application fields. Renewable and Sustainable Energy Reviews, 14:735–744,
2010. 2
[15] A. Brooks. Why hydrogen? CARB Hydrogen Highway Network Workshop,
2005. 3
[16] M. Balat. Potential importance of hydrogen as a future solution to envi-
ronmental and transportation problems. International Journal of Hydrogen
Energy, 33:4013–4029, 2008. 3
[17] IEA. Hydrogen Production and Distribution. IEA Energy Technology Es-
sentials, 2007. 3, 7
[18] T.E. Lipman. What will power the hydrogen economy? Analysis and report
prepared for the natural resources defense council., 2004. 3
[19] J. Brouwer. On the role of fuel cells and hydrogen in a more sustainable
and renewable energy future. Current Applied Physics, 10:9–17, 2010. 4
137
REFERENCES
[20] L. Barreto, A. Makihira, and K. Riahi. The hydrogen economy in the
21st century: a sustainable development scenario. International Journal of
Hydrogen Energy, 28:267–284, 2003. 4
[21] P. Agnolucci. Economics and market prospects of portable fuel cells. In-
ternational Journal of Hydrogen Energy, 32:4319–4328, 2007. 4
[22] S. Carnot. Reflexions sur la Puissance Motrice du Feu. Paris, 1824. 4
[23] IEA. Fossil fuel-fired power generation., 2007. 4
[24] M. Jones. Hybrid vehicles the best of both worlds. Chemistry and Industry,
15:589–592, 1995. 4
[25] F. Barbir and T. Gomez. Efficiency and economics of proton exchange
membrane (PEM) fuel cells. International Journal of Hydrogen Energy,
22:1027–1037, 1997. 4
[26] R. Farrauto, S. Hwang, L. Shore, W. Ruettinger, J. Lampert, T. Giroux,
Y. Liu, and O. Ilinich. New material needs for hydrocarbon fuel processing:
Generating hydrogen for the PEM fuel cell. Annual Review of Materials
Research, 33:1–27, 2003. 5, 7
[27] A. Naidja, C.R. Krishna, T. Butcher, and D. Mahajan. Cool flame partial
oxidation and its role in combustion and reforming of fuels for fuel cell
systems. Progress in Energy and Combustion Science, 29:155–191, 2003. 5,
11
[28] A.L. Dicks. Hydrogen generation from natural gas for the fuel cell systems
of tomorrow. Journal of Power Sources, 61:113–124, 1996. 5
[29] L.J. Pettersson and R. Westerholm. State of the art of multi-fuel reformers
for fuel cell vehicles: problem identification and research needs. Interna-
tional Journal of Hydrogen Energy, 26:243–264, 2001. 6
[30] C.N. Satterfield. Heterogeneous Catalysis in Industrial Practice. Krieger
Publishing Company, 1996. 6
138
REFERENCES
[31] J.C. Amphlett, K.A.M. Creber, J.M. Davis, R.F. Mann, B.A. Peppley, and
D.M. Stokes. Hydrogen production by steam reforming of methanol for
polymer electrolyte fuel cells. International Journal of Hydrogen Energy,
19:131–137, 1994. 6, 8
[32] S. Ahmed and M. Krumpelt. Hydrogen from hydrocarbon fuels for fuel
cells. International Journal of Hydrogen Energy, 26:291–301, 2001. 7, 13,
14
[33] Q. Ming, T. Healey, L. Allen, and P. Irving. Steam reforming of hydrocar-
bon fuels. Catalysis Today, 77:51–64, 2002. 7, 14
[34] D.L. Trimm and Z.I. Onsan. Onboard fuel conversion for hydrogen-fuel-
cell-driven vehicles. Catalysis Reviews, 43:31–84, 2001. 7, 10
[35] D. Dalle Nogare, P. Baggio, C. Tomasi, L. Mutri, and P. Canu. A thermo-
dynamic analysis of natural gas reforming processes for fuel cell application.
Chemical Engineering Science, 62:5418–5424, 2007. 7
[36] A. Ristori, P. Dagaut, and M. Cathonnet. Production of hydrogen by steam
reforming of methanol on CeO2 promoted Cu/Al2O3 catalysts. Journal of
Molecular Catalysis A: Chemical, 194:99–105, 2003. 8
[37] J. Sun, X. Qiu, F. Wu, W. Zhu, W. Wang, and S. Hao. Hydrogen from
steam reforming of ethanol in low and middle temperature range for fuel
cell application. International Journal of Hydrogen Energy, 29:1075–1081,
2004. 8
[38] P. Yaseneva, S. Pavlova, V. Sadykov, E. Moroz, E. Burgina, L. Dovlitova,
V. Rogov, S. Badmaev, S. Belochapkin, and J. Ross. Hydrogen production
by steam reforming of methanol over CuCeZrYOx-based catalysts. Catalysis
Today, 138:175–182, 2008. 8
[39] S. Patel and K.K. Pant. Experimental study and mechanistic kinetic mod-
eling for selective production of hydrogen via catalytic steam reforming of
methanol. Chemical Engineering Science, 62:5425–5435, 2007. 8
139
REFERENCES
[40] A. Iulianelli, T. Longo, and A. Basile. Methanol steam reforming reaction in
a Pd-Ag membrane reactor for CO-free hydrogen production. International
Journal of Hydrogen Energy, 33:5583–5588, 2008. 8
[41] T. Huang and H.M. Chen. Hydrogen production via steam reforming of
methanol over Cu/(Ce,Gd)O2−x catalysts. International Journal of Hydro-
gen Energy, 35:6218–6226, 2010. 8
[42] T.H. Yen, W.T. Hong, W.P. Huang, Y.C. Tsai, H.Y. Wang, C.N. Huang,
and C.H. Lee. Experimental investigation of 1 kW solid oxide fuel cell
system with a natural gas reformer and an exhaust gas burner. Journal of
Power Sources, 195:1454–1462, 2010. 8
[43] J. Mathiak, A. Heinzel, J. Roes, T. Kalk, H. Kraus, and H. Brandt. Cou-
pling of a 2.5 kW steam reformer with a 1 kWel PEM fuel cell. Journal of
Power Sources, 131:112–119, 2004. 9
[44] D.G. Lffler, K. Taylor, and D. Mason. A light hydrocarbon fuel proces-
sor producing high-purity hydrogen. Journal of Power Sources, 117:84–91,
2003. 9
[45] P.C. Hulteberg, H. Burford, K. Duraiswamy, B. Porter, and R. Woods.
A cost effective steam reformer for a distributed hydrogen infrastructure.
International Journal of Hydrogen Energy, 33:1266–1274, 2008. 9
[46] J.M. Sohn, Y.C. Byun, J.Y. Cho J. Choe, and K.H. Song. Development
of the integrated methanol fuel processor using micro-channel patterned
devices and its performance for steam reforming of methanol. International
Journal of Hydrogen Energy, 32:5103–5108, 2007. 9
[47] J.G.M. Furtado, G.C. Gatti, E.T. Serra, and S.C.A. de Almeida. Perfor-
mance analysis of a 5 kW PEMFC with a natural gas reformer. Interna-
tional Journal of Hydrogen Energy, In Press, 2010. 9
[48] M. Prettre, C. Eichner, and M. Perrin. The catalytic oxidation of methane
to carbon monoxide and hydrogen. Transactions of the Faraday Society,
43:335–340, 1946. 10
140
REFERENCES
[49] D.A. Hickman and L.D. Schmidt. Synthesis gas formation by direct oxi-
dation of methane over Pt monoliths. Journal of Catalysis, 138:267–282,
1992. 10
[50] P.D.F. Vernon, M.L.H. Green, A.K. Cheetham, and A.T. Ashcroft. Partial
oxidation of methane to synthesis gas, and carbon dioxide as an oxidising
agent for methane conversion. Catalysis Today, 13:417–426, 1992. 10
[51] S.S. Bharadwaj and L.D. Schmidt. Catalytic partial oxidation of natural
gas to syngas. Fuel Process Technology, 42:109–127, 1995. 10, 14
[52] S.A. Seyed-Reihani and G.S. Jackson. Catalytic partial oxidation of n-
butane over Rh catalysts for solid oxide fuel cell applications. Catalysis
Today, In Press, 2009. 11
[53] I. Aartun, B. Silberova, H. Venvik, P. Pfeifer, O. Grke, K. Schubert, and
A. Holmen. Hydrogen production from propane in Rh-impregnated metallic
microchannel reactors and alumina foams. Catalysis Today, 105:469–478,
2005. 11
[54] P. Corbo and F. Migliardini. Hydrogen production by catalytic partial
oxidation of methane and propane on Ni and Pt catalysts. International
Journal of Hydrogen Energy, 32:55–56, 2007. 11
[55] S.N.K.K. Aki and M.A. Abraham. Catalytic partial oxidation of methane
in supercritical water. The Journal of Supercritical Fluids, 7:259–263, 1994.
11
[56] M. Dajiang, C. Yaoqiang, Z. Junbo, W. Zhenlinga, M. Di, and G. Maochu.
Catalytic partial oxidation of methane over Ni/CeO2 -ZrO2 -Al2O3. Journal
of Rare Earths, 25:311–315, 2007. 11
[57] R. Subramanian, G.J. Panuccio, J.J. Krummenacher, I.C. Lee, and L.D.
Schmidt. Catalytic partial oxidation of higher hydrocarbons: reactivities
and selectivities of mixtures. Chemical Engineering Science, 59:5501–5507,
2004. 11
141
REFERENCES
[58] D.J. Haynes, A. Campos, D.A. Berry, D. Shekhawat, A. Roy, and J.J.
Spivey. Catalytic partial oxidation of a diesel surrogate fuel using an Ru-
substituted pyrochlore. Catalysis Today, In Press, 2009. 11
[59] P.K. Cheekatamarla and C.M. Finnerty. Synthesis gas production via cat-
alytic partial oxidation reforming of liquid fuels. International Journal of
Hydrogen Energy, 33:5012–5019, 2008. 11, 117
[60] H. Pedersen-Mjaanes, L. Chan, and E. Mastorakos. Hydrogen production
from rich combustion in porous media. International Journal of Hydrogen
Energy, 30:579–592, 2005. 11, 18
[61] A. Pastore and E. Mastorakos. Hydrogen production from rich combustion
in porous media. Experimental Thermal and Fluid Science, 34:359–365,
2010. 11, 18, 89
[62] M.K. Drayton, A.V. Saveliev, L.A. Kennedy, A.A. Fridman, and Y. Li.
Syngas production using superadiabatic combustion of ultra-rich methane-
air mixtures. Proceeding of the Combustion Institute, 27:1361–1367, 1998.
11, 17, 59, 89
[63] I. Schoegl, S.R. Newcomb, and J.L. Ellzey. Ultra-rich combustion in paral-
lel channels to produce hydrogen-rich syngas from propane. International
Journal of Hydrogen Energy, 34:5152–63, 2009. 11, 59
[64] Z. Al-Hamamre, S. Voß , and D. Trimis. Hydrogen production by ther-
mal partial oxidation of hydrocarbon fuels in porous media based reformer.
International Journal of Hydrogen Energy, 34:827–832, 2009. 11, 20
[65] F.J. Weinberg, T.G. Bartleet, F.B. Carleton, P. Rimbotti, J.H. Brophy,
and R.P. Manning. Partial oxidation of fuel-rich mixtures in a spouted bed
combustor. Combustion and Flame, 72:235–239, 1988. 11
[66] D. Trimis, K. Wawrzinek, O. Harzfeld, K. Lucka, A. Rutsche, F. Haase,
K. Kru¨ger, and C. Ku¨chen. High modulation burner for liquid fuels based
142
REFERENCES
on porous media combustion and cool flame vaporization. 6th Interna-
tional Conference on Technologies and Combustion for a Clean Environ-
ment (Clean Air VI), 2001. 11
[67] S.C. Reyes, J.H. Sinfelt, and J.S. Feeley. Evolution of processes for synthesis
gas production: Recent developments in an old technology. Industrial &
Engineering Chemistry Research, 42:1588–1597, 2003. 12, 14
[68] M. Krumpelt, T.R. Krause, J.D. Carter, J.P. Kopasz, and S. Ahmed. Fuel
processing for fuel cell systems in transportation and portable power appli-
cations. Catalysis Today, 77:3–16, 2002. 12
[69] A. Docter and A. Lamm. Gasoline fuel cell systems. Journal of Power
Sources, 84:194–200, 1999. 12, 13
[70] B.J. Bowers, J.L. Zhao, M. Ruffo, R. Khan, D. Dattatraya, N. Dushman,
J.C. Beziat, and F. Boudjemaa. Onboard fuel processor for PEM fuel cell
vehicles. International Journal of Hydrogen Energy, 32:1437–1442, 2007.
13
[71] C. Severin, S. Pischinger, and J. Ogrzewalla. Compact gasoline fuel pro-
cessor for passenger vehicle APU. Journal of Power Sources, 145:675–682,
2005. 13
[72] B. Lindstro¨m, J.A.J. Karlsson, P. Ekdunge, L. De Verdier, B. Ha¨ggendal,
J. Dawody, M. Nilsson, and L.J. Pettersson. Diesel fuel reformer for au-
tomotive fuel cell applications. International Journal of Hydrogen Energy,
34:3367–3381, 2009. 13, 118
[73] S. Yoon and J. Bae. A diesel fuel processor for stable operation of solid
oxide fuel cells system: I. Introduction to post-reforming for the diesel fuel
processor. Catalysis Today, In Press, 2010. 13
[74] G.J. Kraaij, S. Specchia, G. Bollito, L. Mutri, and D. Wails. Biodiesel fuel
processor for APU applications. International Journal of Hydrogen Energy,
34:4495–4499, 2009. 13, 119
143
REFERENCES
[75] Y.S. Seo, A. Shirley, and S.T. Kolaczkowski. Evaluation of thermodynam-
ically favourable operating conditions for production of hydrogen in three
different reforming technologies. Journal of Power Sources, 108:213–225,
2002. 14
[76] J.M. Ogden. Review of small stationary reformers for hydrogen production,
2001. IEA report-UC Davies. 14
[77] E. Almqvistt. History of Industrial Gases. Plenum Publishers, 2003. 14
[78] J.D. Holladay, J. Hu, D.L. King, and Y. Wang. An overview of hydrogen
production technologies. Catalysis Today, 139:244–260, 2009. 15
[79] C.D. Dudfield, R.Chen, and P.L. Adcock. A compact CO selective oxidation
reactor for solid polymer fuel cell powered vehicle application. Journal of
Power Sources, 86:214–222, 2000. 15
[80] A. Rehmat and S.S. Randhava. Selective methanation of carbon monox-
ide. Industrial & Engineering chemistry product research and development,
9:512–515, 1970. 15
[81] S. Gottesfeld and J. Pafford. A new approach to the problem of carbon
monoxide poisoning in fuel cells operating at low temperatures. Journal of
the Electrochemical Society, 135:2651–2652, 1988. 15
[82] Johnson Matthey. Palladium membrane hydrogen purifiers: outside-in or
inside-out flow - which design is best for compound semiconductors? Tech
bulletin. 15
[83] D.R. Hardesty and F.J. Weinberg. Burners producing large excess en-
thalpies. Combustion Science and Technology, 8:201–214, 1974. 16
[84] Y. Kotani and T. Takeno. An experimental study on stability and combus-
tion characteristics of an excess enthalpy flame. Proceeding of the Combus-
tion Institute, 19:1503–1509, 1982. 16
[85] F.J. Weinberg. Combustion temperatures: the future?, 1971. Nature. 16
144
REFERENCES
[86] W.A. Bone. Surface combustion. J. Franklin Inst. State Penn., 173:101–
131, 1912. 16
[87] Y.M. Laevskii and V.S. Babkin. Stabilized gas combustion wave in an inert
porous medium. Combustion, Explosion, and Shock Waves, 44:502–508,
2008. 16
[88] V.S. Babkin, V.I. Drobyshevich, Y.M. Laevskii, and S.I. Potytnyakov. Fil-
tration combustion of gases. Fizika Goreniya i Vzryva, 19:17–26, 1983.
16
[89] K. Hanamura, R. Echigo, and S.A. Zhdanok. Superadiabatic combustion
in a porous medium. International Journal of Heat and Mass Transfer.,
36:3201–3209, 1993. 16
[90] D.J. Diamantis, E. Mastorakos, and D.A. Goussis. Simulations of premixed
combustion in porous media. Combustion Theory and Modeling, 6:383–411,
2002. 16, 59
[91] J.P. Bingue, A.V. Saveliev, A.A. Fridman, and L.A. Kennedy. Hydrogen
production in ultra-rich filtration combustion of methane and hydrogen
sulfide. Journal of Hydrogen Research, 27:643–649, 2002. 17
[92] V.S. Babkin, A.A. Korzhavin, and V.A. Bunev. Propagation of premixed
gaseous explosion flames in porous media. Combustion and Flame, 87:182–
190, 1991. 17, 18, 59
[93] Z. Al-Hamamre and A. Al-Zoubi. The use of inert porous media based re-
actors for hydrogen production. International Journal of Hydrogen Energy,
35:1971–1986, 2010. 17
[94] D. Trimis and F. Durst. Combustion in a porous medium-advances and
applications. Combustion Science and Technology, 121:153–168, 1996. 18
[95] W.M. Mathis and J.L. Ellzey. Flame stabilization, operating range, and
emissions for a methane/air porous burner. Combustion Science and Tech-
nology, 175:825–839, 2003. 18
145
REFERENCES
[96] S. Wood and A.T. Harris. Porous burners for lean-burn applications.
Progress in Energy and Combustion Science, 34:667–684, 2008. 18
[97] V. Khanna, R. Goel, and J.L. Ellzey. Measurement of emissions and radia-
tion for methane combustion within a porous medium burner. Combustion
Science and Technology, 133:133–142, 1994. 18
[98] R. Mital, J.P. Gore, and Viskanta. A study of the structure of submerged
reaction zone in porous ceramic radiant burners. Combustion and Flame,
111:175–184, 1997. 18
[99] F. Von Issendorff, V. Stamatov, A. Mach, and D. Trimis. Application
of porous burners in industrial furnaces for glass production. Proceedings
of the Eighth International Conference on Technologies Combustion for a
Clean Environment (Clean Air VIII), 2005. 19
[100] S. Mo¨ßbauer, O. Pickena¨cker, and D. Trimis. Application of the porous
burner technology in energy- and heat-engineering. Proceedings of the Fifth
International Conference on Technologies Combustion for a Clean Environ-
ment (Clean Air V), 1999. 19
[101] A. Ristori, P. Dagaut, and M. Cathonnet. The oxidation of n-hexadecane:
experimental and detailed kinetic modeling. Combustion and Flame,
125:1128–1137, 2001. 20
[102] Sigma-Aldrich Co Ltd. Alumina Oxide for drying properties.
http://www.sigmaaldrich.com. 28
[103] Lanik. Vukopor HT and Vukopor S properties. http://www.lanik.cz/.
29, 30
[104] H. Freund, T. Zeiser, F. Huber, E. Klemm, G. Brenner, F. Durst, and
G. Emig. Numerical simulations of single phase reacting flows in randomly
packed fixed-bed reactors and experimental validation. Chemical Engineer-
ing Science, 58:903–910, 2003. 30
146
REFERENCES
[105] Almatis. T-162 Tabular Alumina balls Data sheet.
http://www.almatis.com. 30
[106] A. Petzold and J. Ulbricht. Feuerbeton und betonartige feuerfeste Massen
und Materialien. Verlag Deutscher Verlag fu¨r Grundstoffindustrie, 1994. 30
[107] TC Ltd. Guide to Thermocouple and Resistance Thermometry.
http://www.tc.co.uk/. 36, 37
[108] D.J. Cookson, P. Iliopoulos, and B.E. Smith. Composition-property rela-
tions for jet and diesel fuels of variable boiling range. Fuel, 74:70–78, 1995.
39, 40
[109] A. Sarioglan, H. Olgun, M. Baranak, A. Ersoz, H. Atakul, and S. Ozdogan.
Diesel evaporation as the first step of hydrogen production. International
Journal of Hydrogen Energy, 32:2895–2901, 2007. 39
[110] S. Jugjai and N. Polmart. Enhancement of evaporation and combustion
of liquid fuels through porous media. Experimental Thermal and Fluid
Science, 27:901–909, 2003. 40
[111] T. Hoshino, Y. Iwata, and H. Koseki. Oxidation stability and risk evaluation
of biodiesel. Thermal Science, 11:87–100, 2007. 40
[112] K. Arisoy. Oxidative and thermal instability of biodiesel. Energy Sources
A, 30:1516–1522, 2008. 40
[113] T.A. Foglia, K.C. Jones, A. Nunez, J.G. Phillips, and M. Mittelbach. Com-
parison of chromatographic methods for the determination of bound glyc-
erol in biodiesel. Chromatographia, 60:305–311, 2004. 40
[114] A.H. Lefebvre. Gas Turbine Combustion. McGraw-Hill, 1999. 50
[115] NASA. Computer Program CEA (Chemical Equilibrium with Applica-
tions). http://www.grc.nasa.gov/WWW/CEAWeb/ceaHome.htm. 51
[116] Turns. HPFLAME - module for adiabatic flame temperature calculation.
51, 112
147
REFERENCES
[117] P.F. Hsu, W.D. Evans, and J.R. Howell. Experimental and numerical study
of premixed combustion within nonhomogeneous porous ceramics. Combus-
tion Science and Technology, 90:149–172, 1993. 59
[118] C.Y. Nah and S. Palanki. Analysis of heptane autothermal reformer to gen-
erate hydrogen for fuel cell applications. International Journal of Hydrogen
Energy, 34:8566–8573, 2009. 67
[119] B.S. Haynes and H.G. Wagner. Soot formation. Progress in Energy and
Combustion Science, 7:229–273, 1981. 88
[120] A.B. Mhadeshwar and D.G. Vlachos. Hierarchical multiscale mechanism
development for methane partial oxidation and reforming and for thermal
decomposition of oxygenates on Rh. Journal of Physical Chemistry B,
109:16819–16835, 2005. 89
[121] K.V. Dobrego, N.N. Gnesdilov, S.H. Lee, and H.K. Choi. Overall chemi-
cal kinetics model for partial oxidation of methane in inert porous media.
Chemical Engineering Journal, 144:79–87, 2008. 89
[122] V.A. Levin, G.D. Smekhov, and A.N. Khmelevskii. Modeling of equilibrium
combustion of kerosene by a mixture of gaseous hydrocarbons. Combustion,
Explosion, and Shock Waves, 43:373–377, 2007. 112
[123] Z. Al-Hamamre, S. Deizinger, A. Mach, F. Von Issendorff, and D. Trimis.
Thermal partial oxidation of diesel in porous reactors for synthesis gas
production. Clean Air, 7:391–407, 2006. 117
[124] A. Lindermeir, S. Kah, S. Kavurucu, and M. Muhlner. On-board diesel fuel
processing for an SOFC-APU-technical challenges for catalysis and reactor
design. Applied Catalysis B: Environmental, 70:488–497, 2007. 117
[125] I. Kang and J. Bae. Autothermal reforming study of diesel for fuel cell
application. Journal of Power Sources, 159:1283–1290, 2006. 118
[126] P.K. Cheekatamarla and A.M. Lane. Catalytic autothermal reforming of
diesel fuel for hydrogen generation in fuel cells I. Activity tests and sulfur
poisoning. Journal of Power Sources, 152:256–263, 2005. 118
148
REFERENCES
[127] S. Roychoudhury, M. Lyubovsky, D. Walsh, D. Chub, and E. Kallio. Design
and development of a diesel and JP-8 logistic fuel processor. Journal of
Power Sources, 160:510–513, 2006. 118
[128] F. Rosa, E. Lopez, Y. Briceno, D. Sopena, R.M. Navarro, M.C. Alvarez-
Galvan, J.L.G. Fierro, and C. Bordons. Design of a diesel reformer coupled
to a PEMFC. Catalysis Today, 116:324–333, 2006. 118
[129] C. Mengel, M. Konrad, R. Wruck, K. Lucka, and H. Kohne. Diesel Steam
Reforming for PEM Fuel Cells. Journal of Fuel Cell Science and Technology,
5, 2008. 118
[130] K. Pinkwart, T. Bayha, W. Lutter, and M. Krausa. Gasification of diesel oil
in supercritical water for fuel cells. Journal of Power Sources, 136:211–214,
2004. 118
[131] B. Lenz and T. Aicher. Catalytic autothermal reforming of jet fuel. Journal
of Power Sources, 149:44–52, 2005. 118
[132] J. Pasel, J. Meissner, Z. Pors, R.C. Samsun, A. Tschauder, and R. Peters.
Autothermal reforming of commercial Jet A-1 on a 5 kWe scale. Interna-
tional Journal of Hydrogen Energy, 32:4847–4858, 2007. 118
[133] B.J. Dreyer, I.C. Lee, J.J. Krummenacher, and L.D. Schmidt. Autothermal
steam reforming of higher hydrocarbons: n-decane, n-hexadecane, and JP-
8. Applied Catalysis A: General, 307:184–194, 2006. 118
[134] X. Yu, S. Zhang, L. Wang, Q. Jiang, S. Li, and Z. Tao. Hydrogen produc-
tion from steam reforming of kerosene over Ni-La and Ni-La-K/cordierite
catalysts. Fuel, 85:1708–1713, 2006. 119
[135] A.R. Choudhuria and S.R. Gollahalli. Characteristics of hydrogen-
hydrocarbon composite fuel turbulent jet flames. International Journal
of Hydrogen Energy, 28:445–454, 2003. 123
[136] F. Halter, C. Chauveau, and I. Go¨kalp. Characterization of the effects of
hydrogen addition in premixed methane/air flames. International Journal
of Hydrogen Energy, 32:2585–2592, 2007. 123
149