HPU/06

HYDROGEN PRODUCTION FROM FOSSIL commodity or use this lower cost secure supply of energy
AND BIOMASS FUELS: CATALYST within the ethanol plant. Because the crude ethanol is
composed of a wide range of oxygenated hydrocarbons, it
DEVELOPMENT, KINETICS, AND means the HTC reformer process can use a wide range of
REACTOR MODELING renewable bio-feedstocks including crude ethanol, glycerol,
cellulose and hemi-cellulose based feedstocks, methanol and
Ahmed Aboudheir and Raphael Idem+ higher order alcohols like propanol, butanol and pentanol, and
HTC Purenergy, Regina, SK, S4P 0S4, Canada; organic acids (i.e. essentially any liquid oxygenated
+
Faculty of Engineering, University of Regina, Regina, SK, hydrocarbon materials containing carbon, hydrogen and
S4S 0A2, Canada oxygen in their molecules).
Catalysts engineering, kinetic studies and reactor
Introduction modeling for these processes as well as their application in a
Fossil and biomass fuels could be made to be a clean multi-feed hydrogen production demonstration will be
and/or alternative energy source if the energy is obtained in an presented.
environmentally sustainable manner. This could be achieved
through the use of hydrogen energy derived from fossil and Catalyst Development
biomass sources with carbon capture. This paper will The developed catalysts for dry reforming and water gas
demonstrate new and innovative ways to produce hydrogen in processes as well as for crude ethanol reforming are very
an environmentally sustainable manner from natural gas and robust and the same catalyst could be used for all the
from crude ethanol as examples of the two fuel sources. oxygenated hydrocarbons reactions, and for the dry reforming
In the first technology, a unique two-stage process of process with little or no modification. The catalysts are based
producing hydrogen by the reforming of natural gas has been on a generic synthetic approach which makes them to be
developed. This process, called “the dry reforming process”, constituent members of the same family of catalysts. The
consists of a two-stage reformer. In the first reformer (dry supports of these catalysts are composed essentially of CeO2
reformer), methane or natural gas reacts with CO2 to produce and/or ZrO2 and/or La2O3 and/or Al2O3 and are prepared via a
H2 and CO while in the second reformer (water gas shift surfactant-assisted method. This approach imparts to the
reformer), CO reacts with steam to produce additional H2 and catalysts nano-crystallinity and high surface area resulting in a
CO2 as the end products; the CO2 is separated and partly high dispersion of the impregnated active non-noble metal (Ni
recycled to the first reformer with the balance being captured. and/or Cu) on the catalyst. The presence of CeO2 in the
The proposed hydrogen reformation technology is highly support leads to a high oxygen storage capacity in the catalyst,
scalable from appliance size to industrial scale plants. As a which together with nano-crystallinity and high metal
result, this technology can be deployed to produce hydrogen dispersion, lead to the avoidance of carbon deposition (or
‘on-site’ for use as a fuel and/or industrial gas for local and/or coking) during reaction. As a result, our catalysts are
widely dispersed consumption. This ‘on-site’ production inherently stable. Consequently, these processes require
capability overcomes one of the main barriers to the minimal replacements of the catalysts unlike what is
widespread adoption of hydrogen as an industrial and commonplace in the literature if nickel is used. For example, a
consumer end-use fuel or industrial gas, i.e. the difficulty in long-term (more than 200 hours) stability tests were
storage and transportation of H2 due to metal embrittlement, conducted to evaluate the durability of the developed catalyst
leakage, and the high cost of compressing H2 for transport. for the reaction with the highest carbon density, the dry
Because of its scalability capability, the technology can be reforming reaction as shown in Figure 1. During catalyst
applied to produce hydrogen from methane obtained from bio- development, activity screening tests are carried out alongside
digesters, and low quality natural gas (both consisting of a an extensive characterization of the catalysts using a wide
large fraction of CO2 in the produced gas). In the latter case, array of characterization tools. The results are used to
the typical practice is to remove the CO2 from the natural gas establish catalyst activity – property relationships, which are
in order to produce a saleable natural gas. The dry reforming then used to fine-tune the catalyst design in order to produce
technology actually takes advantage of the presence of the the improved catalysts. The details of the catalyst
CO2 in the gas as this mixture can provide the direct feed with development and characterization are documented elsewhere
minimal augmented CO2 for the process, and obviates the [1-3].
need and cost associated with CO2 removal.
In the second technology, a new method for hydrogen Kinetics Studies
production from crude ethanol has been developed. The crude For each reaction process, the kinetics experiments were
ethanol term is used in this paper to represent filtered but not carried out at atmospheric pressure in a packed bed tubular
distilled fermentation broth containing water, ethanol and reactor. The experiments to collect the intrinsic kinetic data
other oxygenated hydrocarbons such as glycerol, lactic acid were performed at various reaction temperatures and ratios of
and maltose. By producing hydrogen from crude ethanol, the weight of catalyst to mass flow rate of main reactant (W/FA0).
plant operators have the option to sell the hydrogen as a The reaction temperature for dry reforming, water gas, and

1 of 4
 HPU/06

crude ethanol were ranged within 700-850 oC, 400-600 oC, bar, respectively. Our laboratory scale membrane reactor has
and 320-520 oC, respectively. Certain criteria for packed-bed been characterized and used for running the water gas shift
reactors were observed, according to Froment and Bischoff, to reaction. The model used for evaluating the membrane reactor
ensure that flow conditions in the reactor were close to plug is presented in Equation 1.
flow in order to obtain isothermal reactor operation, eliminate
back mixing and minimize channeling. These are: (a) ratio of Pm e ( − Ea / RT )
catalyst bed height to catalyst particle size (L/Dp)>50, and (b) N H2 = ( PHhigh − PHlow )F n (1)
ratio of internal diameter of the reactor to the catalyst particle δH 2
2 2

size (D/Dp)>10. These conditions were considered in all where NH2 is the H2 permeation rate through the membrane
kinetic experiments in order to ensure plug flow behavior in [mol/(m2s)], Pm is the pre-exponential factor of the membrane
the reactor. Several reaction mechanisms, new and/or reported [mol m/(s m2 bar0.5)], Ea is the apparent activation energy of
were used to interpret the collected experimental kinetics data. the membrane [J/mol], R is the gas constant [J/(mol K)], T is
For example, Eley Rideal assumptions where the surface the absolute temperature [K], dH2 is the thickness of the
reaction involved an adsorbed species and a free gaseous membrane layer [m], PH2high is the partial pressure of H2 in the
species were used to develop the reaction mechanism and four reaction side [bar], PH2low is the partial pressure of H2 in the
models were proposed based on this mechanism, from which a non-reaction side [bar], ∆P is the total pressure difference
new kinetic model based on the dissociation of adsorbed crude between the reactor side and the permeate side of the
ethanol as the rate-determining step was developed for this membrane (psig), F is the total flow rate [mL/min], and n is
novel catalytic process. The obtained experimental rates for the “reaction” order with respect to the flow rate.
the production of hydrogen by the reforming of crude ethanol
were compared with those predicted by the new developed
mechanistic rate model. The results are shown in the parity 0.025

chart of Figure 2. This figure shows excellent predictive

average absolute deviation = 6.0%
ability of the model with an average absolute deviation of
6.0%. This shows that the assumptions made in terms of rate
determining step for this Eley Rideal mechanism based new
Predicted rate, kmol/kg-cat h

developed model and neglecting the effects of the products for 0.020

deriving the model were justified as documented elsewhere

[2].

100 `
CO2
90 0.015
CH4
80
Conversion, mol%

70
60
50 0.010
40 0.010 0.015 0.020 0.025

30 Measured rate, kmol/kg-cat h

20
Figure 2. Comparison of measured and predicted rates of
10
reaction for crude ethanol reforming within the temperature
0 range 320 - 520 0C and WHSV range 1.9 - 4.6 h-1.
0 50 100 150 200 250
TOS, h Reactor Modeling
A rigorous numerical model was developed to design and
Figure 1. Long-term stability test for dry reforming in terms simulate the production of hydrogen from dry reforming of
of conversion of CH4 and CO2 versus time on stream methane/water gas shift reaction and the reforming of
oxygenated hydrocarbons such as crude ethanol. The model
Membrane Reactor for hydrogen separation was based on the coupling of mass and energy balance
At the bench scale level, a membrane reactor is used for equations as well as the new kinetic models developed for
the water gas shift reaction and the crude ethanol reforming to each process (i.e. dry reforming, water gas shift reaction, and
separate hydrogen from the reformate gas. Commercial reforming of crude ethanol). The general mass, energy, and
membrane tubes were used, which are located inside the momentum balance equations used in our model are presented
membrane reactor. These membrane tubes can handle the in Equations 2 to 4.
temperature and the pressure required for the hydrogen
separation, which are in the range 400 - 600 oC and 20 – 40

2 of 4
 HPU/06

∂Ci ∂Ci ⎛ ∂2Ci 1 ∂Ci ∂2Ci ⎞ using the model to simulate a well-documented reaction
+ vZ = D⎜⎜ 2 + + ⎟ + ρBrj (2) process in the literature [4] as shown in Figure 3. This
∂t ∂z ⎝ ∂r r ∂r ∂z 2 ⎟⎠ confirmed the validity of the numerical model to design pilot
and commercial plants as reported elsewhere [5].
⎛ ∂T ∂T ⎞ ⎛ ∂2T 1 ∂T ∂2T ⎞
ρg CP ⎜ + vZ ⎟ = λ⎜⎜ 2 + + ⎟ + ρB ∑−∆H j rj
∂z ⎠ ⎝ ∂r r ∂r ∂z 2 ⎟⎠
[ ] (3)
⎝ ∂t Multi-feedstock Hydrogen Production demonstration
plant
Based on our successful lab scale results demonstrated in
⎛ ∂vz ∂v ⎞ ∂P ⎡∂2v 1 ∂vz ∂2vz ⎤ the previous sections, a multi-feedstock demonstration plant
ρg ⎜ + vz z ⎟ = − + µ⎢ 2z + + ⎥ + ρg g z (4)
⎝ ∂t ∂z ⎠ ∂z ⎣ ∂r r ∂r ∂z 2 ⎦ was designed to produce 1.0 kg/h of hydrogen. A schematic
diagram of the hydrogen production demonstration plant is
shown in Figure 4. The pilot plant will be used for dry
where D denotes effective diffusivity, v denotes velocity, C i
reforming/water gas shift process (operation mode # 1) or
denotes concentration of each chemical species, ρ B denotes steam reforming of oxygenated hydrocarbons (operation mode
the catalyst bulk density, rj denotes reaction rate (if the # 2).
diffusant is being consumed by the reaction, rj will be
negative in these equations), t denotes time, ρ g denotes gas Table 1. Outlet Concentration Profiles Of The Reactor At
density, Cp denotes heat capacity, λ denotes effective thermal Feed Temperature Of 320 0C And W/FA0 Of 2016 Kg-Cat
conductivity, T denotes temperature, ∆H denotes heat of
S/Kg-Crude
Fluid Measured Predicted AD%
reaction, gz denotes acceleration gravity, and P denotes mol% mol%
pressure. The pressure drop in packed bed can be calculated Crude ethanol 2.9 2.8 3.45
from Ergun’s formula of Equation 5 as recommended by Water 80.5 82.2 2.11
Froment and Bischoff, 1990. CO2, CO, CH4, etc. 4.3 3.9 9.30
Hydrogen 12.3 11.1 9.76
dP ρ g v Z2 1 − ε ⎡ 150 (1 − ε ) ⎤ (5) Average absolute deviation, AAD% 6.15
− = 1 . 75 +
dz d p ε 3 ⎢⎣ Re ⎥⎦
where

⎡ (d d − 2 )
2
⎤
ε = 0 . 38 + 0 . 073 ⎢1 + ⎥
t p 1.0 675
(d )
2
o-xylene
⎢⎣ t d p ⎥⎦ 0.9 670
fluid temperature
0.8 665
d ρ gvZ
Re = , dP denotes particle diameter, dt
p 0.7 660
phthalic anhydride
µ

Temperature, K
Mole fraction

0.6 655
denotes internal tube diameter, ε denotes void fraction of 0.5 650
packing, Re denotes Reynolds number, and µ denotes dynamic 0.4 645
viscosity. Diffusion is the only transport mechanism in the 0.3 640
membrane phase (a vertical cylindrical membrane). Thus, the
0.2 635
unsteady state mass balance in cylindrical coordinate for the CO and CO2
0.1 630
permeable gases through the inert membrane can be
0.0 625
represented as in Equation 6. 0.0 0.5 1.0 1.5 2.0 2.5 3.0

∂ C ⎛ ∂ 2C 1 ∂ C ⎞ (6)
Distance from the top of the reactor, m
i
= D ⎜⎜ i
+ i
⎟⎟
∂ t ⎝ ∂ r
2
r ∂ r
m
⎠ Figure 3. Center line axial concentration and temperature
where Dm denotes effective diffusivity in the membrane.
profiles in multi-tubular o-xylene oxidation reactor, industrial
scale.
These equations for the model were solved subject to the
initial and boundary conditions for each process in order to In operation Mode # 1, Dry reforming process and water
design or simulate the behavior of the reactors. For example, gas shift reaction, the natural gas and carbon dioxide is the
the simulation results for crude ethanol conversion were found feed stream comprising of 50% methane and 50% CO2. This
to be in accordance with our experimental data as shown in stream is delivered at about 2 bar to the desulphurization unit
Table 1. A further validation of our model was obtained by and then to the multi-tubular dry reforming reactor. The

3 of 4
 HPU/06

required reaction temperature is in the range 700-850 oC with Conclusions and Recommendations
the endothermic heat obtained via a natural gas/molten salt New stable catalysts and new kinetics models were
furnace. The reformate gas product stream is cooled to about developed for dry reforming of natural gas and water gas shift
60 oC and then compressed to about 20-40 bar. The reformate reaction as well as for crude ethanol reforming. These
gas and steam are then heated and delivered to the multi- catalysts are very robust and the same catalyst could be used
tubular membrane reactor at about 400-600oC. The water gas for any oxygenated hydrocarbons reactions, and for the dry
shift reaction and the separation of hydrogen will be carried reforming process with little or no modification.
out simultaneously in this membrane reactor. The residue In addition, a comprehensive model for the design and
(raffinate) from the membrane reactor returns to the furnace simulation of these hydrogen production processes was
and combusted for additional heating to the reformer. Another developed. On both laboratory and industrial scales, the
option to explore is to add a membrane filter at the end of the developed mathematical model is capable of accurately
raffinate stream to further remove additional hydrogen from predicting the concentration profiles of all the chemical
the raffinate according to Equation 1. The resulting raffinate species and the temperature profiles of the fluid in both axial
could be recycled to the dry reformer. In this case, there will and radial directions. The simulation results on the laboratory
be a purge stream downstream of the membrane filter in order scale was able to demonstrate that plug flow and isothermal
to keep the CO2 levels in the various streams in the process behaviors could not always be guaranteed even by following
fixed. The pure hydrogen stream will be cooled, compressed well-known operating criteria. As a result, it is recommended
and stored after cooling at about 400 bar for utilization. to use a comprehensive model to verify whether or not plug
flow behavior is attained for each kinetic experimental
In operation Mode # 2 for oxygenated hydrocarbon condition. Utilizing the experimental and modeling findings, a
reforming, only the multi-tubular membrane reactor and the demonstration plant is designed to produce 1 kg/h of hydrogen
second will be utilized. The biomass feed (i.e. oxygenated by dry reforming natural gas and oxygenated hydrocarbons. It
hydrocarbons such as crude ethanol) will be heated and is recommended to implement this demonstration plant in
delivered to the membrane reactor at 400-550 oC and 20-40 order to optimize the design configuration and all the
bar using a liquid pump prior to the heating furnace that uses operating parameters, as well as to validate the integration
natural gas/raffinate as a fuel to maintain the temperature of between the unit operations before moving to a commercial
the reactor that is immersed in molten salt. The residue scale.
(raffinate) from the membrane reactor is returned to the
furnace and combusted for additional heating required by the References
reactor. Also, another option to explore is to add a membrane (1) Idem, R.O., Kumar, P.; Sun. Y.; Catalysts for Hydrogen
filter at the end of the raffinate stream to further remove Production, US Patent, Application Number: 2006/0216227,
additional hydrogen from the raffinate. The pure hydrogen Publication date September 28, 2006.
stream will be cooled, compressed and stored after cooling at (2) Akande A.J.; Idem, R.O.; Dalai, A.K.; Applied Catalysis A:
General, 2005, 287, 159–175.
about 400 bar for utilization.
(3) Kumar, P.; Sun, Y.; Ibrahim, H.H.; Akpan, E.R.; Aboudheir, A.;
Idem, R. O.; Appl. Catal., submitted, 2006.
Mode #2 Feed:
Oxygenated Product: H2
(4) Froment G.F.; Bischoff K.B.; Chemical reactor analysis and
Hydrocarbons design. 2nd ed., New York: Wiley; 1990.
CO & H2
(5) Aboudheir, A.; Akande, A.J.; Idem, R.O.; Dalai, A.K.;
CO2 for International Journal of Hydrogen Energy, 2006, 31, 752–761.
Utilization

CO2 for
Mode #1 Feed: Recycle
CH4 H2O
CO2 (start up)

Dry Reforming Reaction WGS or Biomass Reactions

Figure 4. A schematic diagram of Hydrogen Production

Demonstration Plant: Mode # 1, Dry Reforming of Methane
followed by the Water Gas Shift Reaction for hydrogen
production or Mode # 2, Liquid Oxygenated Hydrocarbons
(Biomass) Reforming

4 of 4