Article

pubs.acs.org/EF

Fuel Processing of Diesel and Kerosene for Auxiliary Power Unit

Applications
Joachim Pasel,*, Remzi Can Samsun, Ralf Peters, and Detlef Stolten,

Electrochemical Process Engineering (IEK-3), Institute of Energy and Climate Research (IEK), Forschungszentrum Julich GmbH,
52425 Julich, Germany

IEK-3 Fuel Cells (FZ Julich), Faculty of Mechanical Engineering, Rheinisch-Westfalische Technische Hochschule (RWTH) Aachen
University, 52072 Aachen, Germany

ABSTRACT: Apart from necessary balance-of-plant components, such as pumps, blowers, sensors, and heat exchangers, the fuel
processing unit of a high-temperature polymer electrolyte fuel cell (HT-PEFC) system based on autothermal reforming contains
three main components: the autothermal reformer, the water-gas shift reactor, and the catalytic burner. In Julich, several
generations of these catalytic reactors have been designed, constructed, and manufactured, with a wide range of thermal powers
between 13 and 140 kW. Characteristic common features of the respective reactor generations are described as well as their
specic structural features. The experimental part of this paper concentrates on investigations with dierent generations of
reactors for autothermal reforming using dierent diesel and kerosene fuels, which were produced either via the gas-to-liquid or
bio-to-liquid process or in a conventional manner from crude oil. They mainly diered from each other with respect to their
boiling ranges and mass fractions of aromatics. It was proven in the scope of the current work that autothermal reforming of
synthetic kerosene and diesel fuel can be performed for 6000 h of time on stream at a fuel conversion of 99.99 and 99.67%,
respectively. The O2/C molar ratio of the educts was found to be an important reaction parameter strongly aecting the
monolith temperatures inside the autothermal reformer and the chemical composition of the product gas. Furthermore, fuel
characteristics had a signicant impact on reactor performance.

1. INTRODUCTION Scheme 3. Reaction Equations for Catalytic Combustion

The strategic approach of work in the fuel processing and
systems group at IEK-3 is governed by the aim of developing a
high-temperature polymer electrolyte fuel cell (HT-PEFC)
system based on the autothermal reforming of middle
distillates, such as diesel and kerosene. The envisaged power
class is 510 kWe. This class is particularly interesting for using
HT-PEFC systems as auxiliary power units (APUs) for on-
board power supply in aircraft and trucks. A process owchart
was developed for this purpose and is described in detail by
Samsun et al.1 The main components of the system besides the
HT-PEFC stack are the autothermal reformer (ATR) (Scheme
1), the water-gas shift (WGS) reactor (Scheme 2), and the
Scheme 1. Reaction Equations for Autothermal Reforming
catalytic burner (CAB) (Scheme 3). These three reactors
constitute the fuel processing unit. The ATR converts liquid
fuels, such as diesel fuel or kerosene, together with air and
steam into a hydrogen-rich gas in a catalytic process (cf. eqs 1
Scheme 2. Reaction Equation for the WGS Reaction
and 2). The exothermic partial oxidation and the endothermic
steam reforming reaction of the hydrocarbons (HCs) contained
in diesel fuel and kerosene proceed consecutively on the same
monolith. In parallel, the WGS reaction and the methanation
reaction also take place in an ATR to the thermodynamic
equilibrium of the reactor temperature. The WGS reactor
reduces the concentration of carbon monoxide in the
hydrogen-rich reformate to a level of 1 vol % (cf. eq 5).
Otherwise, the catalyst of the downstream fuel cell would be
poisoned by adsorbed molecules of carbon monoxide, thus
signicantly reducing the cell voltage. The anode o-gas of the
HT-PEFC stack is converted in the CAB according to the
reactions shown in Scheme 3. The anode o-gas contains
considerable amounts of carbon monoxide, which was not
oxidized in the WGS reactor, as well as unconverted hydrogen
and methane. Methane was formed according to eq 4 during
the reforming process. Such gas concentrations (approximately
Special Issue: Accelerating Fossil Energy Technology Development
through Integrated Computation and Experiment
Received: December 3, 2012
Revised: February 7, 2013
Published: February 11, 2013

2013 American Chemical Society 4386 dx.doi.org/10.1021/ef301976f | Energy Fuels 2013, 27, 43864394
Energy & Fuels
32.0 vol % H2O, 1.4 vol % CO, 7.6 vol % H2, 20.2 vol % CO2,
0.3 vol % CH4, and 38.1 vol % N2) cannot be released into the
environment.
Recently, target values for APUs based on fuel cell systems in
the range of 110 kWe were published by the United States
Department of Energy (U.S. DOE). They are valid for systems
operating on standard ultra-low-sulfur diesel fuel and can be
used for truck applications.2 The value for electrical eciency
was dened as 30% for 2013. The nal target of 40% must be
reached in 2020. A power density of 30 We/L was dened for
2013, while that for 2020 was set to 40 We/L. Peters and
Westenberger evaluated a number of scientic publications in
the area of fuel-cell-based APUs for aircraft.3 The goal for
electrical eciency was dened as 40%. This is identical to the
above-mentioned target for trucks. The power density target of
750 We/L is much higher than that for trucks.
2. REACTOR GENERATIONS FOR FUEL PROCESSING
In Julich, several reactor generations for autothermal reforming
have been designed, constructed, manufactured, and exper-
imentally tested in recent years. These generations cover a wide
range of thermal powers from 13 to 140 kW. The thermal
power is calculated by multiplying the lower heating value of
hydrogen with the number of moles of hydrogen produced via
fuel processing. The reformer eciency refers the thermal
power to the lower heating value of the fuel and its molar ow.
For example, ATR 9.2 from Table 1 delivers a molar stream of
Table 1. Reactor Generations for Autothermal Reforming of
Liquid Fuels (Diesel and Kerosene)
reactor thermal power power density specic power
generation (kW) (kW/L) (kW/kg)
ATR 8 13 2.8 2.2
ATR 9.1 18 3.3 2.3
ATR 9.2 28 3.5 3.3
ATR 10 140 3.3 3.2
ATR 13 13 3.6 2.6
ATR AH1 18 3.3 2.3
ATR AH2 28 3.3 3.3
hydrogen with a thermal power of 28 kW. Assuming a
hydrogen use of 83% in the fuel cell and an eciency of the
HT-PEFC itself of 43.0% at a selected operation point, an
electric power of the system (with ATR 9.2) of 10 kW (gross
power) can be achieved. For each reactor generation in Table 1,
the volume of the catalytically coated monolith was calculated
to obtain 100% reformer load at a gas hourly space velocity
(GHSV) of approximately 30 000 h1, an O2/C molar ratio of
0.47, and a H2O/C molar ratio of 1.9.
Table 1 shows the dierent reactor generations for
autothermal reforming together with their thermal power,
power density, and specic power. The fundamental design of
ATR 8 and its experimental operation during startup and steady
state is explained in detail by Pasel et al.4 It should be noted
here that the general structure of ATR 8 was maintained for all
reactor generations in Table 1. It is a common characteristic of
all reformer generations from Table 1 that they are equipped
with an integrated heat exchanger to produce superheated
steam to be used as the educt for the reforming reaction (cf. eq
1). Further common features of all reformer generations are (1)
use of a RhPt/Al2O3CeO2 catalyst from Umicore AG and Co.
KG, (2) use of a standard pressure-swirl nozzle for fuel
4387
Article
injection and atomization, (3) fuel evaporation using the heat
of the superheated steam, which prevents possible fuel ignition
because of the lack of an oxidant, and (4) addition of air to the
mixture after the fuel evaporation zone.
In the literature,515 Rh, Pt, Ni, and RhPt are the preferred
catalytically active species for autothermal reforming of liquid
HCs. These species were deposited on a number of dierent
washcoats, e.g., CeO2, ZrO2, Al2O3, TiO2, SiO2, La2O3, MgO,
and Y2O3. The various papers describe dierent routes for the
preparation of these catalytic systems as well as their catalytic
activity and characterization by a number of analytical methods,
such as temperature-programmed desorption, oxidation and
reduction, pulse chemisorption, or transmission electron
microscopy. Parmar et al.16 describe in their paper kinetic
studies of the autothermal reforming of the diesel surrogate
tetradecane over a Pt/Al2O3 catalyst. They found that a
LangmuirHinshelwoodHougenWatson model, in which
HC is adsorbed on the catalyst surface as alkyl intermediates by
scission of the CH bond, gave physically meaningful results.
The Korea Advanced Institute of Science and Technology
(KAIST) published a number of papers about autothermal
reforming of liquid HCs, gasoline, and diesel fuel for fuel cells
in recent years.1720 They investigated dierent catalytic
systems, varied the operating conditions (temperature, O2/C
molar ratio, and H2O/C molar ratio), and applied a diesel
ultrasonic injector to obtain a homogeneous mixture of the
reactants.
In Table 1, ATR 8 is the base case with a thermal power of
13 kW, a power density of 2.8 kW/L, and a specic power of
2.2 kW/kg. For reactors ATR 9.1 and ATR 9.2, these values
were increased. ATR 9.2 has a thermal power of 28 kW, a
power density of 3.5 kW/L, and a specic power of 3.3 kW/kg.
The mixing chamber of ATR 9.2 was made of sheet metal
instead of ceramics as in earlier ATR generations. This makes it
possible to manufacture the mixing chamber more easily.
Additionally, the reaction chamber of ATR 9.2, which is
exposed to temperatures between 800 and 1000 C, was also
made of metal sheet instead of turned tubes as used before.
This will make it possible to reduce manufacturing costs if large
numbers of this reformer are produced. ATR 10 in Table 1 is
an exception with respect to its thermal power of 140 kW. ATR
10 was specically designed and manufactured for aircraft
applications, where air conditioning, ice and rain protection,
cabin systems, engine starting, landing gears, ight controls, etc.
consume large amounts of electricity to be potentially supplied
by fuel cell systems. A special characteristic of ATR 10 is that
some of its parts (torospherical head and deecting annulus)
were manufactured by means of precision casting, for which
minimum wall thicknesses of 23 mm are necessary inside the
reformer. Precision casting is much more economical than
metal cutting, which was applied for all other reformer
generations in which the wall thicknesses of the pressure-
loaded parts were in the range of only 1 mm. The maximum
value of 3.6 kW/L for the power density was reached in the
case of ATR 13. The design and construction of this reactor
was particularly optimized with respect to its power density
because ATR 13 will be assembled in an allocated and limited
space at the bottom of a conventional truck. The characteristic
values of reformers ATR AH1 and ATR AH2 in Table 1 are
comparable to those of reformers ATR 9.1 and ATR 9.2.
However, ATR AH1 and ATR AH2 are industrially
manufactured by a medium-sized German company, thus
optimizing production costs and material input, while all other
dx.doi.org/10.1021/ef301976f | Energy Fuels 2013, 27, 43864394
Energy & Fuels Article

ATR generations are produced by the workshop at Julich. This 4. RESULTS AND DISCUSSION
transfer of production technology from a research institute to a A statistical experimental design was derived to characterize the
medium-sized company constitutes an important step toward ATR 9.1 described in Table 1. It comprises 11 single
commercialization of the reformer technology developed at experiments, in which the O2/C molar ratio was varied
Julich. The following discussion concentrates on experiments between 0.43 and 0.47, the H2O/C molar ratio was varied
with dierent reactor generations for autothermal reforming, between 1.7 and 1.9, and the GTL kerosene mass ow rate was
applying dierent kinds of diesel fuel and kerosene. varied between 1215 and 2025 g/h. Three experiments at the
center point of the design completed this series of
3. EXPERIMENTAL SECTION investigations to exclude a decline of the catalytic activity and
Seven dierent fuels were used for the autothermal reforming to verify the reproducibility of the results. The ranges for the
experiments. They are described in Table 2 with respect to some variables were selected on the basis of preliminary experi-
ments.23 It was shown in the scope of that work that, when
Table 2. Dierent Kinds of Diesel and Kerosene Applied for applying these molar ratios, the deposition of carbonaceous
Autothermal Reforming substances on the catalyst as well as high temperatures in the
90% average monolith, leading to sintering of the catalytically active species,
recovered aromatics (wt %) molecular can be avoided. Samsun and Peters24 and Box et al.25 give a
fuel supplier (C) /S fraction (ppm) formula very detailed description of the methodology of statistical
GTL Shell MDS, 185 <0.1/<1 C10H22 experimental design in their papers. For the present
kerosene Malaysia
contribution, the inuence of these reaction parameters on
HC BP, 250260 13/<1 C12H24
kerosene Germany characteristic variables was investigated, such as the temper-
EcoPar EcoPar, 338 <0.1/<1 C15H32 atures in the monolith, the concentrations of reaction products
diesel Sweden in the reformate, and the eciency of the reforming process.
NExBTL NESTE Oil, 295 <0.1/<1 C17H36 The temperature of the steam upon entering the reactor was
diesel Finland 460 C. Some of these dependences are presented and
winter- Shell MDS, 307 <0.1/<1 C16H34 described in the following as examples.
grade Malaysia
diesel Figure 1 shows the impact of the O2/C molar ratio and the
summer- Shell MDS, 350 <0.1/<1 C18H38 H2O/C molar ratio on the temperature in the monolith of the
grade Malaysia
diesel
ultimate ARAL, 360 15/<1 C19H38
diesel Germany

special characteristics. They mainly diered in their boiling ranges and

their mass fractions of aromatics. Gas-to-liquid (GTL) kerosene,
EcoPar diesel, NExBTL diesel, winter-grade diesel, and summer-grade
diesel were synthesized from syngas via the GTL or bio-to-liquid
(BTL) process. Therefore, these synthetic fuels had a sulfur mass
fraction of less than 1 ppm and fractions of aromatics far below 1 wt %.
HC kerosene and ultimate diesel, however, were conventionally made
from crude oil in a renery, where they passed a desulfurization step
(hydrocracking), also leading to sulfur mass fractions lower than 1
ppm. Nevertheless, HC kerosene and ultimate diesel contained mass
Figure 1. Inuence of the O2/C molar ratio and the H2O/C molar
fractions of aromatics of approximately 13 and 15 wt %, respectively. It ratio on the temperature of the monolith inside ATR 9.1 at 5 mm in
is well-known from the literature19,21,22 that it is much more dicult to the axial direction at a mass ow of GTL kerosene of 2025 g/h (100%
convert aromatic molecules, such as benzene, toluene, and reformer load).
naphthalene, via autothermal reforming because of their stable
aromatic structure.
In all of the experiments, the product gas was analyzed by a mass ATR 9.1 at 5 mm in the axial direction for a GTL kerosene
spectrometer (Prima 600 S, VG Gas). The concentrations of the mass ow rate of 2025 g/h (100% reformer load). In
following gases were determined: H2, CO, CO2, CH4, N2, and O2.
experiments with earlier generations of reformers,4 it was
Additional analyses of the dry product gas were carried out by a
shown that, because of the very fast kinetics of the exothermic
combined gas chromatograph/mass spectrometer for the detection of
HC molecules with 27 carbon atoms (HP Plot-Q capillary column,
partial oxidation of the HCs (cf. eq 2), high reaction
30 m length and 0.32 mm outer diameter).
temperatures were to be expected just a few millimeters in
A large excess of steam is added to the reformer (H2O/C molar the axial direction of the monolith.
ratio = 1.9) to lower the maximum temperature level in the ATR. Figure 1 illustrates that high temperatures were also observed
Therefore, samples of condensed, non-converted H2O from the ATR during testing of the ATR 9.1, which exceeded 1000 C in the
can be collected directly behind the reactor. These samples were front part of the monolith. The reaction temperatures were
analyzed for the dissolved amounts of total organic carbon (TOC) and higher at higher O2/C molar ratios and lower H2O/C molar
non-purgeable organic carbon (NPOC) by means of infrared ratios. The maximum temperature was 1025 C. Higher oxygen
spectroscopy. The analytical method for the determination of the partial pressures appeared to benet the exothermic partial
NPOC value slightly diers from that for the TOC value. It does not oxidation of the HCs, which led to an increase in the monolith
account for volatile components dissolved in the water. temperature. In the case of lower H2O/C molar ratios, a lower
4388 dx.doi.org/10.1021/ef301976f | Energy Fuels 2013, 27, 43864394
Energy & Fuels Article

total mass ow had to be heated within the process, which also

gave rise to increased temperatures in the monolith.
Figure 2 shows the impact of the O2/C molar ratio and the
H2O/C molar ratio on the temperature in the monolith of the

Figure 3. Inuence of the O2/C molar ratio and the H2O/C molar
ratio on the concentration of hydrogen in the reformate of ATR 9.1 at
a mass ow of GTL kerosene of 2025 g/h (100% reformer load).

concentration to drop. The impact of the H2O/C molar ratio in

Figure 3 is not quite as pronounced. Higher H2O/C molar
Figure 2. Inuence of the O2/C molar ratio and the H2O/C molar ratios resulted in higher hydrogen concentrations. This can be
ratio on the temperature of the monolith inside ATR 9.1 at 5 mm in explained by the fact that the WGS reaction was promoted by a
the axial direction at a mass ow of GTL kerosene of 1215 g/h (60%
reformer load).
higher water partial pressure. This had a positive impact on the
hydrogen concentration in the reformate. With respect to
Figure 3, it can be additionally pointed out that, at lower O2/C
ATR 9.1 at 5 mm for a GTL kerosene mass ow rate of 1215 g/ molar ratios, the hydrogen yield and, thus, the reformer
h (60% reformer load). Similar observations were made as in eciency increase because the share of endothermic steam
Figure 1 at 100% load. At 60% load, the temperature in the reforming in the reaction process as a whole is enhanced.
monolith was also highest at 5 mm in the axial direction, with Equation 1 makes it clear that the steam reforming reaction
high oxygen and low water partial pressures in the inlet volume delivers more moles of hydrogen than the partial oxidation. The
ow, which is explained as above. A striking dierence between thermodynamic analysis performed by Parmar et al.26 supports
the two load proles is that the temperature level at 60% load this explanation. The authors calculated decreasing hydrogen
was shifted to lower values. The maximum temperature was yields (moles of hydrogen per mole of diesel) with an
only approximately 975 C. A measurement of the temper- increasing O2/C molar ratio.
atures at the outer surface of ATR 9.1 at full and partial loads Figure 4 once again shows the impact of the O2/C molar
(not displayed in Figure 2) showed a temperature dierence of ratio and the H2O/C molar ratio on the concentration of
approximately 30 K. The temperatures were higher at full load. hydrogen in the autothermal reforming product gas, in this
This means that, on the one hand, the heat losses were higher case, for HC kerosene and GTL kerosene at mass ow rates of
at full load. On the other hand, it is well-known from 2025 g/h (100% reformer load). HC kerosene is distinguished
thermodynamics that the absolute heat of formation is lower at
partial load. From a quantitative comparison between the heat
of formation at full and partial loads with the heat losses at full
and partial loads, it can be concluded that the relative heat
loss at partial load is much higher than the relative heat loss
at full load. Here, for the expression relative heat loss, the heat
loss is referred to the heat of formation. As a consequence, the
higher relative heat loss at partial load led to lower
temperatures in the monolith at partial load, as described above.
Figure 3 shows the impact of the O2/C molar ratio and the
H2O/C molar ratio on the hydrogen concentration in the
reformate at a GTL kerosene mass ow rate of 2025 g/h (100%
reformer load). It is clear that the hydrogen concentration in
the reformate of the ATR 9.1 is signicantly inuenced by the
O2/C molar ratio. The average hydrogen concentration was
approximately 37.5 vol % for an O2/C molar ratio of 0.47, and
it increased to an average of 39.0 vol % at an O2/C molar ratio
of 0.43. This trend can be explained by the dilution of the
reformate with nitrogen at a high O2/C molar ratio. Moreover,
higher O2/C molar ratios, as shown in Figure 1, led to higher
temperatures in the monolith, so that the thermodynamic Figure 4. Inuence of the O2/C molar ratio and the H2O/C molar
equilibrium of the WGS reaction, which was also proceeding on ratio on the concentration of hydrogen in the reformate of ATR 9.1 at
the reformer catalyst (cf. eq 3), was shifted to the educt side. In mass ows of GTL kerosene and HC kerosene of 2025 g/h (100%
addition to the dilution eect, this also caused the hydrogen reformer load).

4389 dx.doi.org/10.1021/ef301976f | Energy Fuels 2013, 27, 43864394

Energy & Fuels Article

from GTL kerosene by its signicantly broader boiling range completed in this test. In each cycle, 16 dierent load points
and, above all, by its comparatively high mass fraction of were set between 60 and 100%. This corresponds to kerosene
aromatics of 13 wt % (cf. Table 2). For HC kerosene, the same mass ow rates between 1620 and 2700 g/h. The dwell time at
eect of the O2/C and H2O/C molar ratios was observed as for each of the load points varied between 2 and 10 min. The
GTL kerosene. However, the level of the hydrogen entire experiment took around 70 h.
concentration was signicantly dierent in the case of HC Figure 6 shows the kerosene mass ow rate as a function of
kerosene. While the concentration of hydrogen varied between the duration of the experiment in the time interval between 30
37.5 and 39.0 vol % when GTL kerosene was used, in the case
of HC kerosene, lower values between 35.5 and 37.0 vol %
were observed. This results from the considerably higher
fraction of aromatic compounds in HC kerosene and can be
explained by comparing the stoichiometries, which, for
example, proceed during the partial oxidation of benzene and
n-hexane. Although both substances have a comparable molar
mass of 78 g/mol (benzene) and 86 g/mol (n-hexane), only 3
mol of hydrogen is formed per mole of benzene during the
partial oxidation of benzene, whereas in the case of n-hexane,
this value is 7 mol of hydrogen.
In Figure 5, the impact of the O2/C molar ratio and the
H2O/C molar ratio on the methane concentration in the

Figure 6. Mass ow of GTL kerosene in a load change experiment

with ATR 9.2 at an O2/C molar ratio of 0.47 and a H2O/C molar ratio
of 1.9.

and 40 h for clarity of the diagram. The 16 dierent recurrent

load points are easily recognizable. The mass ow rates of the
other reactants (air and water) were adjusted to the kerosene
mass ow. They are not shown here.
Figure 7 shows the temperatures in the mixing chamber of
the ATR 9.2 during the experiment as well as those in the

Figure 5. Inuence of the O2/C molar ratio and the H2O/C molar
ratio on the concentration of methane in the reformate of ATR 9.1 at a
mass ow of GTL kerosene of 2025 g/h (100% reformer load).

reformate of the ATR 9.1 is described and discussed for a GTL

kerosene mass ow rate of 2025 g/h (100% load). A strong
inuence of the oxygen partial pressure on the methane
concentration in the reformate is visible. At an O2/C molar
ratio of 0.47, the methane concentration ranged between 0.2
and 0.4 vol %. This increased to approximately 0.81.0 vol % at
an O2/C molar ratio of 0.43.
High methane concentrations in the reformate are undesired
because methane has to be combusted in the CAB of the fuel
processing system and must not be released into the
atmosphere. Additionally, 3 mol of hydrogen is consumed for Figure 7. Temperatures in the monolith and in the mixing chamber in
a load change experiment with ATR 9.2 at an O2/C molar ratio of 0.47
each mole of methane (cf. eq 4). This, of course, has a negative and a H2O/C molar ratio of 1.9 using GTL kerosene.
impact on the reformer eciency. However, methane is a very
stable molecule. The molar ow to be converted is decisive for
the design of the monolith in the CAB and should not exceed monolith at 5 and 140 mm in the axial direction. For the sake of
values of more than 0.5 vol %. Low-oxygen partial pressures in clarity, only the time interval between 30 and 40 h is shown in
the educt ow of the ATR are therefore unsuitable for a fuel this case as well. Dependent upon the individual load points,
processing system. The inuence of the water partial pressure the temperature in the lower mixing chamber varied between
on the methane concentration was very low, as shown in Figure 370 and 410 C, while the temperature in the monolith at 5
5. mm uctuated between 790 and 880 C. This indicated that a
Experimental testing of the ATR 9.2 from Table 1 involved a slight homogeneous pre-reaction occurred in the mixing
load change test at an O2/C molar ratio of 0.47 and a H2O/C chamber of the ATR 9.2 during the entire load change
molar ratio of 1.9. Approximately 40 consecutive cycles were experiment. The temperature in the monolith at 140 mm in the
4390 dx.doi.org/10.1021/ef301976f | Energy Fuels 2013, 27, 43864394
Energy & Fuels
axial direction remained constant despite the load changes and
was between 680 and 685 C. The endothermic steam
reforming of HCs in the lower area of the monolith appears
to somewhat constrain the temperature uctuations in the
upper area of the monolith.
Figure 8 shows the concentrations of carbon monoxide and
hydrogen in the same time interval of the load change
Figure 8. Concentrations of hydrogen and carbon monoxide in a load
change experiment with ATR 9.2 at an O2/C molar ratio of 0.47 and a
H2O/C molar ratio of 1.9 using GTL kerosene.
experiment. The two concentrations were hardly aected by the
load changes. They were 38.2 vol % for hydrogen and 9.1 vol %
for carbon monoxide. No deactivation of the catalyst was
observed as a result of the recurrent load changes throughout
the full 70 h of the experiment (not shown here).
One decisive criterion for the maturity of a new technology is
its durability for the respective application. To evaluate this
issue, a long-term experiment with ATR 9.2 was conducted.
Figure 9 shows the concentrations of H2, CO, CO2, and CH4 as
a function of time on stream using GTL kerosene and BTL
diesel as fuel (cf. Table 2). GTL kerosene was used for the rst
5000 h of time on stream. After that, the experiment was
continued by applying BTL diesel fuel. According to Scheme 1,
H2 and CO are the desired products of autothermal reforming,
while CH4 and CO2 are formed during methanation and the
Figure 9. Concentrations of hydrogen, carbon monoxide, carbon
dioxide, and methane in a long-term experiment with ATR 9.2 at an
O2/C molar ratio of 0.47 and a H2O/C molar ratio of 1.9 using GTL
kerosene and BTL diesel fuel.
4391
Article
WGS reaction. In Figure 9, the concentrations of H2, CO, CO2,
and CH4 remained constant during the rst 4500 h of time on
stream. The H2 concentration was in the range of 38.038.6
vol %, and the concentration of CO amounted to
approximately 9.210.0 vol %, while the concentration of
CO2 varied between 12.5 and 11.5 vol %. CH4 displayed a
concentration of approximately 0.3 vol %. In the last 500 h of
time on stream, a slight deactivation of the catalyst occurred
with respect to the WGS reaction. The concentrations of H2
and CO2 decreased slightly, while the concentration of CO
increased simultaneously. After switching to BTL diesel fuel,
the values for the gas concentrations changed slightly but
remained almost constant for another 1000 h of time on
stream. The concentration of H2 decreased to the range of
34.736.0 vol %, whereas the concentration of CO was in the
range of 11.012.4 vol %. The content of CO2 in the reformate
varied between 9.5 and 10.5 vol %, while the content of
methane remained almost constant in comparison to the
experiment with GTL kerosene and displayed values of around
0.3 vol %. For stoichiometric reasons (cf. Scheme 1), the
concentration of H2 in the reformate must be lower when using
BTL diesel fuel with a molecular formula of C17H36 in
comparison to GTL kerosene with a molecular formula of
C10H22 (cf. Table 2).
The conversion of HCs in GTL kerosene and BTL diesel fuel
is dened as being complete if no molecules apart from H2,
CO, CO2, or CH4 are detected in the gas phase or in the
samples of condensed, non-converted H2O. The analysis of the
gas phase with respect to C2C7 HCs, such as ethene, propene,
butene, or even benzene, was negative in the case of the
experimental phase with GTL kerosene and is therefore not
shown in Figure 10. However, from the very beginning of the
Figure 10. Concentrations of byproducts in the gas phase in a long-
term experiment with ATR 9.2 at an O2/C molar ratio of 0.47 and a
H2O/C molar ratio of 1.9 using BTL diesel.
experiment with BTL diesel fuel, undesired byproducts, such as
ethene, ethane, propene, propane, butenes, pentenes, and
hexene, were detected in the gas phase of ATR 9.2. The
concentration of ethene was highest with 100 ppmv after 1000
h of time on stream with BTL diesel fuel (6000 h total time on
stream). Their total concentration values, i.e., the content of
non-methane hydrocarbons NMHCs, increased to 282 ppmv
after 1000 h of time on stream in the experiment with BTL
diesel fuel. A maximum NMHC value of 460 ppmv was found
after 900 h of time on stream after an unintentional shutdown
of the test rig. This caused additional thermal stress on the
dx.doi.org/10.1021/ef301976f | Energy Fuels 2013, 27, 43864394
Energy & Fuels
catalyst. Fortunately, the catalytic activity could be at least
partially restored in the following 100 h of time on stream.
Ethene is known from the literature to be a precursor of
carbon formation on the catalyst, thus reducing the number of
active sites. Yoon et al.20 describe in their paper dierent
methods to suppress the ethene-induced carbon deposition in
diesel autothermal reforming. They successfully varied the O2/
C molar ratio and used a diesel ultrasonic injector to reduce the
formation of ethene. Karatzas et al.1215 published a large
number of papers about autothermal reforming of diesel fuel.
They investigated a number of RhPt catalysts on dierent
washcoats (CeO2, ZrO2, Al2O3, TiO2, SiO2, La2O3, MgO, and
Y2O3) on a 400 cells per square inch (cpsi) cordierite monolith
for the autothermal reforming of diesel fuel with low sulfur
content.14 During their experiments, they found a minimum
ethene concentration of approximately 1000 ppmv at 98%
conversion of diesel fuel with their most promising catalyst
system RhPt/CeO2ZrO2.
Additionally, during the whole experiment, traces of
byproducts of the reforming process were found in the
condensed, non-converted H2 O. Figure 11 shows the
Figure 11. Dissolved amounts of TOC and NPOC in the condensed
non-converted water from a long-term experiment with ATR 9.2 at an
O2/C molar ratio of 0.47 and a H2O/C molar ratio of 1.9 using GTL
kerosene and BTL diesel.
concentrations of non-purgeable organic carbon (NPOC)
for both experiments with GTL kerosene and BTL diesel fuel
and total organic carbon (TOC) for the investigation with
BTL diesel fuel in the condensed, non-converted H2O as a
function of time on stream. The NPOC value for the rst 5000
h of time on stream in Figure 11 varied between 10 and 25
ppmw and increased somewhat with time on stream, indicating
a slight deactivation of the catalyst. After switching to BTL
diesel, the NPOC value increased to 53 ppmw, while the TOC
value was 90 ppmw after an additional 1000 h of time on
stream. These ndings are in agreement with the results in
Figure 10 and the data in Table 2 with respect to the chemical
and physical characteristics of the two fuels applied. The results
indicate that it is more dicult to run the autothermal
reforming reaction with BTL diesel fuel than with GTL
kerosene because the amount of byproducts is signicantly
higher in the case of BTL diesel fuel. BTL diesel fuel displays
higher boiling temperatures and a larger molar mass. Therefore,
its components require a higher enthalpy ow in the
evaporation chamber of the reformer. There is a greater risk
of liquid fuel droplets hitting the catalyst, undergoing pyrolysis
4392
Article
on the catalyst, and thus damaging active sites of the catalyst in
the case of BTL diesel fuel. Additionally, zones with a non-
homogeneous mixture of educts (oxygen, steam, and
evaporated fuel) in the mixing chamber of the reformer
might also occur more frequently. Nevertheless, conversion of
HCs in GTL kerosene was still very high at approximately
99.99% after 5000 h of time on stream. After switching to BTL
diesel, HC conversion was slightly reduced and amounted to
99.67% after an additional 1000 h. These values are still very
promising with respect to the application of ATR 9.2 in an APU
onboard aircraft or trucks. The way in which these conversions
were calculated is explained in detail in ref 24. It considers both
the undesired byproducts in the gas and liquid phases from
Figures 10 and 11.
The nal part of this paper investigates the inuences of the
chemical and physical properties of the dierent diesel fuels
from Table 2 on the performance and the operational behavior
of ATR 8. For these experiments, the ATR was operated at full
load (100%). Saturated steam was fed into the internal heat
exchanger of the reactor, where it was superheated to more
than 440 C before entering the mixing chamber of the ATR.
Figure 12 shows the contents of hydrogen, carbon monoxide,
Figure 12. Concentrations of hydrogen, carbon monoxide, carbon
dioxide, and methane during operation of ATR 8 with ve dierent
diesel fuels at an O2/C molar ratio of 0.47 and a H2O/C molar ratio of
1.9, with 100% reformer load and saturated steam.
carbon dioxide, and methane at an O2/C molar ratio of 0.47
and a H2O/C molar ratio of 1.9. The gure shows that there
were only small dierences in the gas concentrations in the
reformate. The hydrogen concentrations were in the range of
37.0 and 38.0 vol %. The concentrations of carbon monoxide
varied between 10.1 and 11.4 vol %, and the concentrations of
carbon dioxide varied between 11.2 and 11.5 vol %, while the
methane contents were rather stable, displaying values of 0.1
0.2 vol %. The experiment with ultimate diesel was an
exception. Strong decreases in the hydrogen content to 32.1 vol
% and in the carbon dioxide content to 9.0 vol % were
observed, in parallel to the highest value for the carbon
monoxide concentration of 13.6 vol %. On the one hand, this
result is due to the high mass fraction of aromatics contained in
ultimate diesel. For stoichiometric reasons, the chemical
reaction of aromatic molecules during autothermal reforming
delivers fewer moles of hydrogen than that of aliphatic HCs.
Additionally, the observed changes in the concentrations result
from a reduced activity of the catalyst for the WGS reaction.
In addition to the results shown in Figure 12 with respect to
the main products, Figure 13 displays the respective
dx.doi.org/10.1021/ef301976f | Energy Fuels 2013, 27, 43864394
Energy & Fuels
Figure 13. Concentrations of byproducts in the gas and liquid phases
during operation of ATR 8 with ve dierent diesel fuels at an O2/C
molar ratio of 0.47 and a H2O/C molar ratio of 1.9, with 100%
reformer load and saturated steam.
concentrations of byproducts in the gas and liquid phases at an
O2/C molar ratio of 0.47 and a H2O/C molar ratio of 1.9.
Among the byproducts in the gas phase, propene showed the
highest values and is therefore shown in Figure 13 as an
example. Because a large excess of water was added to the
reformer (H2O/C molar ratio = 1.9), samples of condensed,
non-converted H2O from the ATR can be collected directly
behind the reactor. These samples were analyzed for the
dissolved amounts of TOC and NPOC. These values are also
shown in Figure 13. Acetic acid displayed the highest contents
among the substances in the condensed water and is
additionally shown in the diagram. Other dissolved byproducts
were acetaldehyde, propanal, acetone, butanal, 2-butanone,
benzene, 2-pentanone, toluene, and benzaldehyde.
Figure 13 shows very low values for the concentrations of the
byproducts when NExBTL diesel is used. The NMHC value
was only 13 ppmv. According to Table 2, this diesel fuel is free
of aromatics, with the lowest temperature for 90% recovery
during fractionated distillation. When using Ecopar diesel, the
concentrations for the byproducts were slightly higher. The
NMHC value reached 16 ppmv. The concentrations of all
byproducts increased when Shell summer-grade diesel fuel was
used. The NMHC value rose to 261 ppmv, and the
concentration of propene rose to 73 ppmv. The NPOC and
TOC concentrations amounted to 16 ppmw. Shell summer-
grade diesel fuel is also free of aromatics but has a remarkably
high temperature for 90% recovery of 350 C. Some of the
components in Shell summer-grade diesel fuel demand a higher
enthalpy ow for their evaporation before entering the ATR
monolith. Therefore, the danger of liquid fuel droplets
impinging on the monolith, undergoing pyrolysis on the
catalyst, and thus damaging active sites is higher when Shell
summer-grade diesel fuel is used in comparison to NExBTL
diesel. Obviously, the catalyst inside ATR 8 was slightly
deactivated by the experiments with Shell summer-grade diesel
fuel because the concentration values for the byproducts in
Figure 13 when using Shell winter-grade diesel fuel were higher
(NMHC = 310 ppmv) than those observed with NExBTL
diesel and Ecopar diesel, although their physical and chemical
characteristics in Table 2 were comparable. Finally, Figure 13
shows the results obtained when applying ultimate diesel, which
has by far the highest mass fraction of aromatics among the
fuels investigated (approximately 15 wt %) and a comparatively
high temperature for 90% recovery during fractionated
4393
Article
distillation of approximately 360 C. Accordingly, the
concentrations of the byproducts were found to be highest.
The NMHC value reached 915 ppmv, and the concentration of
propene reached 157 ppmv. The NPOC and TOC
concentrations were 138 and 169 ppmw, respectively.
5. CONCLUSION
Fuel processing of liquid fuels, such as kerosene and diesel, is an
ecient and environmentally friendly process of producing
hydrogen for fuel cell systems. The reactors for autothermal
reforming can be designed and constructed at dierent thermal
power levels (13140 kW) relevant for various applications,
such as aircraft and trucks. The power densities and specic
powers of these reactors are very promising, showing values
from 2.8 to 3.6 kW/L and from 2.2 to 3.3 kW/kg, respectively.
Samsun et al.1 report on Julichs packaged fuel processor, which
contains ATR 9.2, a WGS reactor, a CAB (all in the thermal
power class of 28 kW), a heat exchanger for cooling the
reformate before it enters the fuel cell, two heating cartridges
for startup, and the necessary pipework. This packaged fuel
processor has a thermal power density of 330 W/L. This value
is very suitable for the spatial requirements of aircraft and truck
applications.
Autothermal reforming of GTL kerosene with ATR 9.2 at a
thermal power level of 28 kW was proven to run for 5000 h at a
HC conversion of 99.99%. After switching to BTL diesel fuel,
the amounts of undesired byproducts in the gas and liquid
phases increased signicantly because of the broader boiling
range of BTL diesel fuel. However, after an additional 1000 h of
time on stream, the HC conversion still amounted to 99.67%.
A severe impact of the chemical and physical properties of
dierent diesel fuels was found during autothermal reforming.
For stoichiometric reasons, high mass fractions of aromatic
molecules in ultimate diesel caused a decrease in the
concentration of hydrogen in the reformate in parallel to an
increase in the content of carbon monoxide. Using diesel fuels
with broader boiling ranges and higher mass proportions of
aromatics resulted in the formation of larger amounts of
undesired reaction byproducts, thus decreasing HC conversion.
AUTHOR INFORMATION
Corresponding Author
*Telephone: +49-2461-61-5140. Fax: +49-2461-61-6695. E-
mail: j.pasel@fz-juelich.de.
Notes
The authors declare no competing nancial interest.
ACKNOWLEDGMENTS
Parts of this work were funded by the German Federal Ministry
of Economics and Technology within the EFFESYS project, by
the Ministry for Climate Protection, Environment, Agriculture,
Nature Conservation, and Consumer Protection of the German
State of North Rhine-Westphalia within the ADELHEID
project, and nally, by the FCGEN project, which is conducted
within the Fuel Cells and Hydrogen Joint Undertaking
programme of the European Commission. The authors are
particularly grateful to the fuel processing and HT-PEFC teams
at Julich and all project and cooperation partners.
REFERENCES
(1) Samsun, R. C.; Wiethege, C.; Pasel, J.; Janen, H.; Lehnert, W.;
Peters, R. Energy Procedia 2012, 29, 541551.
dx.doi.org/10.1021/ef301976f | Energy Fuels 2013, 27, 43864394
Energy & Fuels Article

(2) United States Department of Energy (U.S. DOE). Revised APU

Targets, DOE Program Record; U.S. DOE: Washington, D.C., 2010;
Record 11001.
(3) Peters, R.; Westenberger, A. In Innovations in Fuel Cell
Technologies; Steinberger-Wilckens, R., Lehnert, W., Eds.; Royal
Society of Chemistry (RSC): London, U.K., 2010; RSC Energy and
Environment Series 2.
(4) Pasel, J.; Meiner, J.; Pors, Z.; Samsun, R. C.; Tschauder, A.;
Peters, R. Int. J. Hydrogen Energy 2007, 32, 48474858.
(5) Ferrandon, M.; Krause, T. Appl. Catal., A 2006, 311, 135145.
(6) Ferrandon, M.; Mawdsley, J.; Krause, T. Appl. Catal., A 2008,
342, 6977.
(7) Mayne, J. M.; Dahlberg, K. A.; Westrich, T. A.; Tadd, A. R.;
Schwank, J. W. Appl. Catal., A 2011, 400, 203214.
(8) Mayne, J. M.; Tadd, A. R.; Dahlberg, K. A.; Schwank, J. W. J.
Catal. 2010, 271, 140152.
(9) Shamsi, A.; Baltrus, J. P.; Spivey, J. J. Appl. Catal., A 2005, 293,
145152.
(10) Harada, M.; Takanabe, K.; Kubota, J.; Domen, K.; Goto, T.;
Akiyama, K.; Inoue, Y. Appl. Catal., A 2009, 371, 173178.
(11) Kaila, R. K.; Gutierrez, A.; Krause, A. O. I. Appl. Catal., B 2008,
84, 324331.
(12) Karatzas, X.; Dawody, J.; Grant, A.; Svensson, E. E.; Pettersson,
L. J. Appl. Catal., B 2011, 101, 226238.
(13) Karatzas, X.; Jansson, K.; Dawody, J.; Lanza, R.; Pettersson, L. J.
Catal. Today 2011, 175, 515523.
(14) Karatzas, X.; Jansson, K.; Gonzalez, A.; Dawody, J.; Pettersson,
L. J. Appl. Catal., B 2011, 106, 476487.
(15) Karatzas, X.; Creaser, D.; Grant, A.; Dawody, J.; Pettersson, L. J.
Catal. Today 2011, 164, 190197.
(16) Parmar, R. D.; Kundu, A.; Thurgood, C.; Peppley, A. B.; Karan,
K. Fuel 2010, 89, 1212.
(17) Kang, I.; Bae. J. Power Sources 2006, 159 (2), 1283.
(18) Yoon, S.; Kang, I.; Bae, J. Int. J. Hydrogen Energy 2008, 33 (18),
4780.
(19) Kang, I.; Bae, J.; Bae, G. J. Power Sources 2006, 163 (1), 538.
(20) Yoon, S.; Kang, I.; Bae, J. Int. J. Hydrogen Energy 2009, 34 (4),
1844.
(21) Borup, R. L.; Inbody, M. A.; Semelsberger, T. A.; Tafoya, J. I.;
Guidry, D. R. Catal. Today 2005, 99, 263270.
(22) Kaila, R. K.; Krause, A. O. I. Int. J. Hydrogen Energy 2006, 31,
1934.
(23) Palm, C.; Cremer, P.; Peters, R.; Stolten, D. J. Power Sources
2002, 106, 231.
(24) Samsun, R. C.; Peters, R. Methodologies for fuel cell process
engineering. In Fuel Cell Science and Engineering; Stolten, D., Emonts,
B., Eds.; Wiley-VCH Verlag and Co. KGaA: Weinheim, Germany,
2012; Vol. 1, pp 597642.
(25) Box, G. E. P.; Hunter, W. G.; Hunter, J. S. Statistics for
Experimenters; John Wiley and Sons: Hoboken, NJ, 1978.
(26) Parmar, R. D.; Kundu, A.; Karan, K. J. Power Sources 2009, 194,
1007.

4394 dx.doi.org/10.1021/ef301976f | Energy Fuels 2013, 27, 43864394