Energy & Fuels 1998, 12, 11-18 11

Plasma Reforming of Methane

L. Bromberg,* D. R. Cohn, and A. Rabinovich
Plasma Science and Fusion Center, Massachusetts Institute of Technology, NW16-108,
77 Massachusetts Ave, Cambridge, Massachusetts 02139

C. O’Brien and S. Hochgreb

Department of Mechanical Engineering, Massachusetts Institute of Technology,
77 Massachusetts Avenue, Cambridge, Massachusetts 02139

Received July 7, 1997. Revised Manuscript Received October 15, 1997X

Thermal plasma technology can be used in the production of hydrogen and hydrogen-rich gases
from a variety of fuels. This paper describes experiments and calculations of high-temperature
conversion of methane using homogeneous and heterogeneous processes. The thermal plasma
is a highly energetic state of matter that is characterized by extremely high temperatures (several
thousand degrees Celsius) and high degree of ionization. The high temperatures accelerate the
reactions involved in the reforming process. Plasma reformers can be operated with a broad
range of fuels, are very compact and are very light (because of high power density), have fast
response time (fraction of a second), can be manufactured with minimal cost (they use simple
metallic or carbon electrodes and simple power supplies), and have high conversion efficiencies.
Hydrogen-rich gas (50-75% H2, with 25-50% CO for steam reforming) can be efficiently made
in compact plasma reformers. Experiments have been carried out in a small device (2-3 kW)
and without the use of efficient heat regeneration. For partial oxidation it was determined that
the specific energy consumption in the plasma reforming processes is 40 MJ/kg H2 (without the
energy consumption reduction that can be obtained from heat regeneration from an efficient
heat exchanger). Larger plasmatrons, better reactor thermal insulation, efficient heat regenera-
tion, and improved plasma catalysis could also play a major role in specific energy consumption
reduction. With an appropriate heat exchanger to provide a high degree of heat regeneration,
the projected specific energy consumption is expected to be ∼15-20 MJ/kg H2. In addition, a
system has been demonstrated for hydrogen production with low CO content (∼2%) with power
densities of ∼10 kW (H2 HHV)/L of reactor, or ∼4 m3/h H2 per liter of reactor. Power density
should increase further with power and improved design.

I. Introduction facturing hydrogen. These shortcomings include cost

and deterioration of catalysts, size and weight require-
Manufacturing of hydrogen from hydrocarbon fuels ments, limitations on rapid response, and limitations
is needed for a variety of applications. These applica- on hydrogen production from heavy hydrocarbons. In
tions include fuel cells used in stationary electric power addition, use of plasma technology could provide for a
production and in vehicular propulsion. Hydrogen can greater variety of operating modes including the pos-
also be used for various combustion engine systems. sibility of virtual elimination of CO2 production by
Hydrogen manufacturing is also needed for industrial pyrolytic operation. This mode of hydrogen production
applications and could be used in refueling stations for may be of increasing importance because of recent
hydrogen-powered vehicles. There is a wide range of additional evidence of global warming.
requirements on the capacity of the hydrogen manu- Disadvantages of plasma reforming are the difficulty
facturing system, the purity of the hydrogen fuel, and of high-pressure operation and the dependence on
capability for rapid response. The overall objectives of electrical energy. High pressure, while achievable,
a hydrogen manufacturing facility are to operate with increases electrode erosion and decreases electrode
high availability at the lowest possible cost and to have lifetime. Dependence on electrical energy results in
minimal adverse environmental impact. energetics that are less favorable than the energetics
of purely thermal processes, especially for endothermic
II. Plasma Reforming reforming reactions.
Plasma devices referred to as plasmatrons can gener-
Plasma technology has potential to significantly al-
ate very high temperatures (>2000 °C) with a high
leviate shortcomings of conventional means of manu-
degree of control, using electricity. The heat generation
is independent of reaction chemistry, and optimum
* To whom correspondence should be addressed. E-mail: BROM@
PFC.MIT.EDU. Telephone: 617-253-6919. Fax: 617-253-0700. operating conditions can be maintained over a wide
X Abstract published in Advance ACS Abstracts, November 15, 1997. range of feed rates and gas composition. Compactness
S0887-0624(97)00109-6 CCC: $15.00 © 1998 American Chemical Society
Published on Web 01/12/1998
12 Energy & Fuels, Vol. 12, No. 1, 1998 Bromberg et al.

of the plasma reformer is ensured by high-energy

density associated with the plasma itself and by the
reduced reaction times, resulting in a short residence
time.
Hydrogen-rich gas (50-75% H2, with 25-50% CO for
steam reforming) can be efficiently produced in plasma
reformers with a variety of hydrocarbon fuels (gasoline,
diesel, oil, biomass, natural gas, jet fuel, etc.) with
conversion efficiencies to hydrogen-rich gas close to
100%.1,2 Plasma reformers are relatively inexpensive
(they use relatively simple metallic or carbon elec-
trodes).
The plasma conditions (high temperatures and a high
degree of ionization) can be used to accelerate thermo-
dynamically favorable chemical reactions without a
catalyst or to provide the energy required for endother-
mic reforming processes. Plasma reformers can provide
a number of advantages: (1) compactness and low
weight (due to high power density); (2) high conversion Figure 1. Schematic diagram of research plasmatron used
efficiencies; (3) minimal cost (simple metallic or carbon in the experiments.
electrodes and simple power supplies); (4) fast response
time (fraction of a second); (5) operation with a broad cooled cathode with zirconium tip and water-cooled copper
tubular anode. The plasma-forming gas is air or nitrogen. The
range of fuels, including heavy hydrocarbons; (6) no
length of the arc is stabilized by the ledge in the internal
need for catalysts and therefore no problems of catalyst diameter of the anode. The hydrocarbons (methane) are
sensitivity and deterioration; (7) operation with a broad injected radially downstream from the anode root of the arc.
range of fuels, including heavy hydrocarbons. The This type of plasmatron has a long lifetime (1000-2000 h) at
technology could be used to manufacture hydrogen for currents less than 100 A (corresponding to ∼10 kW). The
a variety of stationary applications, e.g., distributed, zirconium tip provides thermionic emission in air or nitrogen
low-pollution electricity generation from fuel cells.3 It atmosphere without excessive cathode erosion.
could also be used for mobile applications (e.g., on-board The reactor consists of steel tube with an outer diameter of
generation of hydrogen for fuel cell-powered vehicles) 0.09 m (3.5 in.) and a length of 0.2 m (8 in.). The inside of the
and for refueling applications (stationary sources of steel tube was covered with a 0.012 m (0.5 in.) ceramic lining
hydrogen for vehicles). (Sauereisen No. 8), leaving a 0.06 m (2.25 in.) inner diameter
channel. The plasmatron was connected to the top of reactor.
Plasma reformers also offer a unique potential for
An inefficient heat exchanger was used in some of the
pyrolytic generation of hydrogen without production of experiments, with the air and methane being preheated in
CO2, thus providing an option for reduction of gases that tubing wrapped around the reactor.
could contribute to global warming. The size of the plasmatron is 0.075 m (3 in.) long and 0.04
This paper describes the investigation of plasma m (2 in.) outer diameter. The weight is 0.7 kg (1.5 lbs).
reforming with use of different oxidizing feedstocks: (1) The gas chromatograph (GC) used (MTI Model M200)
pyrolytic decomposition (no oxidizer); (2) partial oxida- consists of a sample loop, vacuum pump, and two columns,
tion (air); (3) modified partial oxidation (air/water); (4) each with an injector valve and thermal conductivity detector
heterogeneous partial oxidation. (TCD). Gas samples are provided directly from the reactor
The overall objective of this work is to investigate the by means of a water-cooled copper tube. The analyzed
dependence of the hydrogen yield and the specific energy components are H2, O2, N2, CO, CH4, CO2, C2H2, C2H4, and
C2H6. Bottled gas was used for calibration. Water content in
input of thermal, noncatalytic processes and to prelimi-
the reformate is not monitored.
narily explore plasma catalysis.

III. Experimental Facility IV. Methane Pyrolysis

The experimental facility includes research plasmatrons, In pyrolysis reforming, the goal is to manufacture
reactors, and GC diagnostic. hydrogen gas and solid carbon (soot) in an oxygen-free
A plasma system has been developed based on Air Plasma environment. Equilibrium calculations indicate that at
Cutting System PAK MASTER 100. The power supply
low temperatures, methane decomposes into hydrogen
provides 75 A of maximum current at arc voltages of ∼120-
160 V. The standard system has been customized in order to and soot. In practice, the process is dominated by
control and measure the arc current and voltage. The system kinetics and nucleation and particle growth, and it is
was equipped with gas and water rotameters, pressure trans- very difficult to model. High temperatures, such as
ducers, and a digital thermocouple meter. those achievable with plasmas, are required to drive the
A schematic of the research plasmatron used in these process.
studies is shown in Figure 1. The plasmatron has a water- Research in the area of methane plasma reforming
has in the past been driven by industrial production of
(1) Rudiak, E. M.; Rabinovich, A.; Tul, N. A. USSR Patent 700935,
August 1979.
acetylene. Acetylene is produced at high temperatures,
(2) Kaske, G.; Kerke, L.; Muller, R. Hydrogen Production in the Huls and in order to freeze its concentration and optimize
Plasma-Reforming Process. Hydrogen Energy Prog. 1986, VI (1). its production, fast quenching of the reformate is
(3) Bromberg, L.; Cohn, D. R.; Rabinovich, A. Plasma Reformer/ Fuel
Cell Systems for Decentralized Power Applications. Int. J. Hydrogen needed, as in the Huls and DuPont processes.4,5 More
Energy 1997, 22, 83-94. recent experiments for hydrogen/carbon conversion in
Plasma Reforming of Methane Energy & Fuels, Vol. 12, No. 1, 1998 13

Figure 3. Specific energy consumption for hydrogen for the

Figure 2. Hydrogen yield in plasma pyrolysis of methane as same conditions as in Figure 2.
a function of the power consumption per unit mass of methane
for several reactor setups.
refers to a simple heat regeneration system consisting
of a few turns around the outside of the steel reactor; it
a plasma (without quench) with large plasma devices did not operate effectively as a heat exchanger. Straight
have been performed by Kvaerner in Norway and refers to the conventional setup, with vertical plasma
Sweden.6 Thermal cracking of methane has in the past facing up. Big anode refers to a larger anode and a
been used for carbon black manufacturing and has been longer residence time. For all of the experimental
proposed for hydrogen manufacturing.7 In France, setups, the average specific power consumption was in
Fulcheri8 performed calculations on methane pyrolysis the range 200-250 MJ/kg H2. The hydrogen yield,
in a plasma and conducted experiments whose main defined as the ratio of measured hydrogen concentration
objective was that of carbon black manufacturing.9 Both to that of complete conversion, increases from ∼30% to
the Kvaerner and French work use graphite electrodes, 70% when the specific power input increases from 250
operating at near-atmospheric pressure. These experi- to 650 kJ/mol CH4. The product gas composition in the
ments indicate high-energy consumption requirements. case of the highest hydrogen yield was 33.3 vol % H2,
The goal of the present experiments is to investigate 54.4 vol % N2, 6.8 vol % CH4, 3.2 vol % C2H2, and 1.2
the means of decreasing specific energy consumption, vol % C2H4. No other peaks were observed in the GC.
both through plasmatron improvements and reactor Future research will concentrate on development of
modifications. different types of plasmatron (with uncooled graphite
The methane pyrolysis experiments were carried out electrodes) and reactor chambers. In a plasmatron with
with a plasmatron operating with nitrogen gas. The these characteristics, methane can be used as the
methane gas was injected downstream from the anode plasma gas, eliminating the diluent (nitrogen in the
and allowed to cool slowly (no quenching, to minimize present experiments) and decreasing the specific energy
the acetylene production). The ratio of flow rates of CH4 consumption. Specific energy consumption could also
to N2 has been varied from 1:1 to 1:3. be reduced by heat regeneration and by operation with
Several arrangements of the plasmatron and the conditions that result in a longer residence time, such
reactor were tested. These included inverted vertical as increased reactor volume.
injection into reactor chamber (plasma jet facing down),
side injection into the reactor chamber with extracting V. Partial Oxidation/Steam Reforming of
spout (cyclone configuration), and fluidized bed. The Methane
cyclone configuration was motivated by the hope of The process explored in the experiments involved
removing the condensed carbon (soot) from the flow. The using a combination of air and steam as the oxidizer.
fluidized bed was done with activated charcoal particles The plasmatron was operated with air, and methane
to investigate if providing high surface area for the and water were injected downstream from the anode.
deposition of the carbon atoms resulted in a decrease Partial oxidation is an exothermic reaction, while
in the specific energy consumption. steam reforming is strongly endothermic. The purpose
The results of experiments are shown in Figures 2 of the reagent mixture was to explore the effectiveness
and 3. The cyclone, downstream injection, and fluidized of near-energy-neutral processes. In addition, it may
bed setups were described above. The CH4 preheat be possible to achieve one-step reforming/water shift
reactions, with a compact design.
(4) Muller, R. The Use of Hydrogen Plasma Processes in the The results of plasma reforming of methane by air-
Petrochemical and Iron-Smelting Industries. Hydrogen Energy Prog.
1988, YII (2), 885-900. water mixture at constant power are shown in Figures
(5) Holmes, J. M.Evaluation of DUPONT Arc Process for Acetylene 4 and 5. Kair is the amount of air normalized to the air
and Vinyl Chloride Monomer Production. U.S. Department of Com- required for stoichiometric partial oxidation of the
merce/National Bureau of Standards, 1969 (30 pages).
(6) Gaudernack, B.; Lynum, S. Hydrogen from Natural Gas without methane. Similarly, Kwater is the amount of water
Release of CO2 to the Atmosphere. Hydrogen Energy Prog., Proc. World normalized to the water required for stoichiometric
Hydrogen Energy Conf., 11th 1996, 511-523.
(7) Steinber, M. Production of Hydrogen and Methanol from Natural
steam reforming of the methane. As could be seen from
Gas with Reduced CO2 Emissions. Hydrogen Energy Prog., Proc. World Figure 4 at 0.4 < Kair < 0.7, the maximum H2 yield
Hydrogen Energy Conf., 11th 1996, 499-510, (∼60%) is achieved at water excess Kwater close to 0 and
(8) Fulcheri, L.; Schwob, Y. From Methane to Hydrogen, Carbon
Black and Water. Int. J. Hydrogen Energy 1995, 20, 197-202. decreased to 40% as the water excess increases to 0.7.
(9) Lynum, S. Private communication. At this air-to-methane ratio (close to pyrolysis) the
14 Energy & Fuels, Vol. 12, No. 1, 1998 Bromberg et al.

Figure 4. Contours of constant hydrogen yield as a function Figure 6. Contours of constant hydrogen yield as a function
of Kair and Kwater for partial oxidation/steam reforming of of Kair and Kwater for plasma catalysis of methane.
methane.

Figure 7. Contours of specific energy consumption for the

Figure 5. Contours of specific energy consumption for the same conditions as in Figure 6.
same conditions as in Figure 4.
The reactor was filled with NiO catalyst on an Al2O3
reaction is strongly endothermic and addition of water support. The air/methane ratio in all experiments was
decreases gas temperature and consequently the rate Kair ≈ 1 (partial oxidation), and water excess has been
of reaction. changed to be in the range 0 < Kwater < 4.
The same behavior is observed at Kair ≈ 0.9, which is In the first set of experiments the catalyst volume was
close to stoichiometric partial oxidation. Although the 2.5 × 10-4 m3. With this volume, the H2 yield was 65-
reaction is slightly exothermic, the H2 yield (80%) is as 70% and the specific power consumption was ∼72-75
high as can be achieved with these operational condi- MJ/kg H2. The results of doubling the volume of
tions. As in the case with lower Kair, water addition only catalyst to 5 × 10-4 m3 are shown in Figures 6 and 7 as
decreases the temperature and H2 yield. functions of Kair and Kwater. The larger catalyst volume
At Kair ≈ 1.1-1.4 and water excess close to 0, the H2 increases the hydrogen yield to 90-100% at specific
yield (60%) is less than at stoichiometric air/methane power consumptions of -45 to 55 MJ/kg H2.
ratio. Since the reaction is exothermic, water addition As shown in Figures 6 and 7, the optimum operating
increases H2 yield to 80%. conditions are Kair = 1 and Kwater ≈ 4. In this regime
Figure 5 shows the specific energy consumption (for the hydrogen yield reaches 100% at specific power
hydrogen) as a function of Kair and Kwater. The minimum consumptions of ∼45 MJ/kg H2. Increasing water
power consumption without the use of heat regeneration excess beyond Kwater ≈ 4 significantly decreases the gas
(∼100 MJ/kg H2) is achieved at ratio air-to-methane temperature and reaction rate, with a corresponding
close to 1 (stoichiometric partial oxidation) where H2 decrease of the hydrogen yield to 70% at a specific power
yield is maximum. Decreasing Kair toward the pyrolysis consumption of ∼70 MJ/kg H2.
mode or increasing to complete combustion leads to
increased specific power consumption. This value is VII. Discussion
reduced from previously obtained values of the partial
oxidation of methane5 with a reactor with reduced Comparison of homogeneous plasma and plasma
volume (and enhanced increased space velocity) and catalytic modes of operation shows the significant
without ceramic lining. advantage of the later regime. Figures 8 and 9 show
the hydrogen yield and the specific energy consumption
VI. Plasma Catalytic Reforming as a function of the energy input per unit mass of fuel
To decrease the specific power consumption further, for partial oxidation (previous results), partial oxidation/
experiments on plasma catalytic reforming of methane steam reforming (present results), and plasma catalysis
by the air-water mixture have been carried out. (present results). The results of partial oxidation had
Plasma Reforming of Methane Energy & Fuels, Vol. 12, No. 1, 1998 15

amounts of light hydrocarbons were observed (<0.1 vol

%). Water content is not monitored.
These results, although preliminary, indicate that by
the control of operating parameters (water excess,
power, water injection point, etc.), it is possible to
eliminate the water shift reaction stage and produce
relatively pure hydrogen in one stage.
Previous calculations5 quantify the benefits of efficient
preheating. With a good heat regenerator, capable of
preheating the incoming gases to temperatures of 700-
800 °C, the required specific energy requirement is
expected to drop by about 1/2. It may be possible
therefore to decrease the energy consumption to about
20 MJ/kg H2.
Figure 8. Hydrogen yields as a function of the energy The results were obtained with a setup that has not
consumption per unit mass for partial oxidation (previous yet been optimized. Among the items that could be
results), partial oxidation/steam reforming, and for plasma
catalysis of methane.
improved are the use of a heat exchanger to preheat
the reagents prior to introduction into the plasmatron,
the use of larger power/size plasmatron (10-20 kW),
since larger plasmas are more efficient than smaller
ones (larger plasmas have lower surface-to-volume
ratios), better performing catalysts, including using
separate high-temperature and low-temperature cata-
lysts, and finally, better thermal insulation by the use
of ceramic linings with lower thermal conductivity at
high temperatures.
The cost of the hydrogen produced by the plasma
reformer system is still under investigation. The cost
of the produced hydrogen is a sensitive function of the
cost of electricity and the cost of the feedstock. On the
other hand, the capital expense for the plasmatron, the
Figure 9. Specific energy consumption for the same condi- power conditioning equipment, the monitoring and the
tions as in Figure 8. controls plus the operating cost of the plasmatron
electrodes (which are consumable) are relatively small.
been obtained previously,10 while the results labeled Under these conditions, assuming a cost of gas of $0.1
partial oxidation/steam reforming were discussed above. m-3 ($2.7/MMBTU) and a cost of electricity of 50 mils
The H2 yield for partial oxidation has been increased kW-1 h-1 , then the cost of the hydrogen produced is
by ∼20-50% from previous results, owing to improved about $6/MMBTU. The cost distribution is approxi-
plasmatron and reactor designs. The specific energy mately 40% electricity and 55% cost of fuel (methane),
consumption has also been decreased by about 20-30%, and the rest is the depreciation cost (assuming a 3 year
mostly owing to increased yield. linear depreciation). This is a first attempt to determine
the cost of the produced hydrogen, and further refine-
The plasma catalytic regime results are substantially
ment will be necessary.
better than previous results and most recent partial
oxidation/steam reforming results. The hydrogen yields
are 2-3 times better at significantly lower (1/3) specific VIII. Theoretical Calculations
power consumption. The chemical processes discussed in the sections
In addition to an increase of the yields and a decrease above can be modeled both from thermodynamic and
of the energy consumption, there is an added advantage chemical kinetic viewpoints in order to better under-
of plasma catalytic reforming vs homogeneous plasma stand the chemical and physical parameters that influ-
reforming. With water addition, it is possible to ac- ence the reforming process (e.g., reaction heat release,
complish in one stage both reforming and water shift thermal energy input from the plasmatron, reactor heat
processes. It is possible to produce hydrogen with losses, residence times, and mixing characteristics).
sufficiently low CO concentrations for use in high- Thermodynamic (i.e., chemical equilibrium) modeling
performance PEM fuel cells. Conventional process of allows prediction of the operating characteristics of
hydrogen production includes stages of partial oxidation idealized reformers and thus identification of the most
(steam reforming), water shift reaction, and preferential productive regions of the reactor design parameter space
oxidation. In plasma catalytic reforming, the composi- to investigate, whereas chemical kinetic models allow
tion of the reformate is 40 vol % H2, 38 vol % N2, 3.4 more detailed simulations of particular reactor designs
vol % CH4, 3.4 vol % CO, and 13.5 vol % CO2. Trace and aid in interpretation of experimental results. The
following section will include some results from ther-
(10) Bromberg, L.; Cohn, D. R.; Rabinovich, A.; O’Brien, C.; Hoch- modynamic models of the various reaction categories
greb, S. In Proceedings of the 1996 U.S. DOE Hydrogen Program discussed above (partial oxidation, partial oxidation plus
Review. National Renewable Energy Laboratory Report NREL/CP-430-
21968; National Renewable Energy Laboratory: Golden, CO, 1996; p water shift, and pyrolysis), as well as a kinetic model
553. of the thermal (noncatalytic) plasma reactor.
16 Energy & Fuels, Vol. 12, No. 1, 1998 Bromberg et al.

Figure 10. Availability diagram of the states of the reforming

process. State i is the initial fuel-air-water mixture, state r
is the reformed product, state m is a hypothetical intermediate
state after energy addition. Figure 12. Output species concentrations and temperature
for an equilibrium model of a partial oxidation methane
reformer (relative air-to-fuel ratio λ ) 0.25).

Figure 11. Output species concentrations and temperature

for an equilibrium model of a pyrolytic methane reformer as
functions of the energy input (expressed as a fraction of the Figure 13. Output species concentrations and temperature
fuel lower heating value). for an equilibrium model of a partial oxidation methane
reformer with water shift (relative air-to-fuel ratio λ ) 0.25;
Thermodynamic equilibrium calculations can greatly steam-to-carbon ratio ) 10).
simplify the comparison of different reforming reactions
or technologies, since they allow determination of limit-
ing system behaviors with minimal computation. If all
chemical reactions are assumed to reach equilibrium
within the reactor, the system can be specified com-
pletely by the energy inputs and losses along with input
mixture composition (and hence chemical energy). In
the present analysis, the chemical energy of a mixture
is defined as its availability with respect to products of
complete combustion. Thus, given a reformer with a
specified input composition and known energy interac-
tions with the environment, the final equilibrium com-
position and temperature are determined. Figure 10
gives a diagrammatic representation of the states and
available interactions of the reforming process.
Figure 14. Specific energy consumption (MJ/kg H2) as a
Figures 11-13 indicate the variations in equilibrium function of energy input for pyrolysis, partial oxidation, and
output composition from reformers operating under partial oxidation with water shift reformers.
conditions of pyrolysis, partial oxidation, and partial
oxidation with water shift, respectively, as functions of the hydrogen output is maximized. Another important
the net energy input to the reformer (i.e., thermal energy conclusion from these results is that significant hydro-
input from the plasmatron less heat losses in the gen yields from pyrolysis can be achieved only at quite
reactor), expressed as a fraction of the lower heating high energy inputs, limiting the utility of this process
value of the input fuel (50 MJ/kg for methane). Several to applications where efficiency is not a primary concern.
points can be noted from these figures. First, it can be This effect can be seen more quantitatively in Figure
seen that even for the exothermic reaction of partial 14, which shows the specific energy consumption (in MJ/
oxidation, additional energy input from the plasmatron kg H2) for the three processes at equilibrium. The
can result in larger hydrogen yields; for the cases of dashed horizontal line corresponds to the lower heating
partial oxidation and partial oxidation with water shift, value of hydrogen (120 MJ/kg). Hence, there is only a
there is a particular range of energy input values where limited range of input energies for which pyrolysis of
Plasma Reforming of Methane Energy & Fuels, Vol. 12, No. 1, 1998 17

Figure 15. Predicted hydrogen yield from a kinetic model of Figure 16. Predicted specific energy consumption (in MJ/kg
the thermal plasma partial oxidation-water shift methane H2) from a kinetic model of the thermal plasma partial
reformer (steam-to-carbon ratio ) 0.1) as a function of relative oxidation-water shift methane reformer (steam-to-carbon
air-to-fuel ratio (Kair ) 1 is stoichiometric for partial oxidation) ratio ) 0.1) as a function of relative air-to-fuel ratio (Kair ) 1
and energy input (relative to methane LHV of 50 MJ/kg CH4). is stoichiometric for partial oxidation) and energy input
Circles are experimentally observed points; stars are the same (relative to methane LHV of 50 MJ/kg CH4).
points corrected for heat loss (i.e., lower net energy input).
) 0.1). Since the heat losses in the experimental system
methane leads to a net production of hydrogen energy. were not completely known, the calculations were
In all other cases, more energy must be spent to produce performed for a range of input energies, shown on the
the hydrogen than can be extracted from it. vertical axes of the plots. Some relevant experimental
It should be noted that several important issues points, with observed hydrogen yields, are plotted in
cannot be addressed by these thermodynamic calcula- Figure 15 as circles, using the arc power to determine
tions. These include the required size and weight of a the energy input. It can be seen that the experimental
fuel processor for a given desired conversion, response hydrogen yields are lower than the calculated values.
times, and costs per kW of hydrogen production capac- This effect can be attributed to the heat losses in the
ity, all of which are strongly affected by the kinetics of reactor; the stars plotted in the figure correspond to the
the reaction processes. same observed points corrected for 50% heat loss, at
As an example of chemical kinetic analysis of specific which point the model matches the observed yields well,
reformer designs and as an aid to interpretation of the indicating that this portion of the input energy is
experimental results presented above, the thermal probably lost from the reactor instead of remaining in
plasma partial oxidation-water shift reactor has been the gas mixture.
modeled with full kinetics. The reactor (described The major advantage of the use of chemical kinetic
previously) is approximated as a plug-flow reactor; the models is the ability to examine different reactor
heat input from the plasma is modeled as an extremely configurations without the expense and complications
rapid enthalpy addition to the input reactant flow (i.e., of experiments. For example, Figures 15 and 16 indi-
fast compared to the important chemical reaction time cate that the optimal operating conditions for the
scales). Given the initial conditions specified by the thermal plasma partial oxidation-water shift reactor
input composition and flow rate and the arc power, the with K ) 0.1 lie in the range 0.7 < Kair < 1 with arc
set of chemical rate equations given by the reaction energies between 30 and 40% of the fuel energy, with
mechanism (in this case, a modified version of GRI- better yields at lower Kair and higher energy input and
Mech 2.11) and the energy equation are integrated over lower specific energy consumption at higher Kair and
time, assuming adiabatic conditions. The residence lower energy input.
time of the mixture in the reformer is calculated from
the resulting temperature and pressure versus time IX. Conclusions
profiles and the reactor volume, assuming ideal gas Plasma reforming of methane, especially plasma
behavior. The predicted output gas composition of the catalysis, has been demonstrated with sufficiently low-
reformer can then be found from the plug-flow reactor energy consumption and with high hydrogen yields to
composition at the calculated residence time. be economically interesting for energy applications.
Figures 15 and 16 show the results of a series of such The pyrolysis results require substantially higher
calculations for hydrogen yield and specific energy energy consumption than could be obtained from the
consumption, respectively, spanning a parameter space hydrogen produced. Substantial reduction of specific
that corresponds to the experiments shown in Figures energy consumption will be required before this becomes
4 and 5. In both cases, methane fuel is used, and the feasible for energy applications. The results should
air-to-fuel ratio (the horizontal axes of the figures) is improve with increasing reactor size and with use of
varied around the stoichiometric value for partial oxida- uncooled graphite electrode plasmatrons.
tion (0.6 < Kair < 1.2). The amount of water injected is For homogeneous reforming with partial oxidation
fixed at 0.1 mole of water per mole of methane (Kwater with air, improvements in plasmatron and reactor
18 Energy & Fuels, Vol. 12, No. 1, 1998 Bromberg et al.

design resulted in a decrease of specific energy con- Plasmatron generation of hydrogen could be used for
sumption but with values that are still relatively high. decentralized energy production with fuel cells. Another
Further improvements in plasmatron and reactor will possible application of this technology is for onboard
be required. reforming for the manufacturing of hydrogen-rich gases.
Initial results with plasma catalysis indicate hydro- The compactness of the reformer, with high power
gen yields of about 100% and moderate energy con- densities, makes the technology attractive for this
sumption without the use of heat regeneration and application.
sufficient residence time. There were no problems with Plasma reforming could also find uses for decentral-
soot buildup on the catalyst or on the surfaces of reactor. ized hydrogen production, such as in transportation
In addition, with water injection, it was possible to refueling systems, either on-demand generation of
combine reforming and water shift reactions in a single hydrogen (requiring 200-500 kW plasmatrons) or small
stage. The device is very compact, with power densities 20 kW CW plasmatrons (with stationary storage) for
of ∼10 kW (H2 HHV)/L of reactor, or ∼4 m3/h H2 per 30-50 vehicle fleets. The technology could also be used
liter of reactor. for industrial applications. Finally, plasma reformers
There are several areas where improvements on could be used with hard-to-use fuels, such as raw
plasmatron and reactor are possible. They include biofuels or heavy oils.
system optimization (such as heat regeneration), im-
proved design (increased thermal insulation in the
Acknowledgment. This work was performed under
reactor chamber), and increased power operation (larger
the auspices of DOE under Contract FG04-95AL88002.
plasmas with larger volume-to-surface ratios). These
improvements are actively being pursued. EF9701091