ww.sciencedirect.com

i n t e rn a t i o n a l j o u r n a l o f h y d r o g e n en e r g y 3 8 ( 2 0 1 3 ) 1 0 6 5 4e1 0 6 7 0

Available online at w

journal homepage: www.elsevier .com/locate/he

Hetero-/homogeneous combustion of hydrogen/airmixtures over platinum: Fuel-lean versus fuel-richcombustion modes

Marco Schultze, John Mantzaras*

Paul Scherrer Institute, Combustion Research, CH-5232 Villigen PSI, Switzerland

a r t i c l e i n f o

Article history:

Received 30 January 2013

Received in revised form

13 June 2013

Accepted 17 June 2013

Available online 15 July 2013

Keywords:

Fuel-lean and fuel-rich catalytic

combustion of hydrogen over plat-

inum

Hetero-/homogeneous combustion

of hydrogen

Homogeneous ignition

Reactor thermal management

Safe reactor operating regimes

* Corresponding author. Tel.: þ41 56 3104046E-mail address: ioannis.mantzaras@psi.c

0360-3199/$ e see front matter Copyright ªhttp://dx.doi.org/10.1016/j.ijhydene.2013.06.0

a b s t r a c t

The catalytic combustion of hydrogen/air mixtures in platinum-coated channels was

investigated by means of two-dimensional numerical simulations with detailed hetero-

geneous (catalytic) and homogeneous (gas-phase) chemical reaction schemes, detailed

species transport, and heat transfer mechanisms in the solid wall. Three different com-

bustion modes relevant to power generation were analyzed and their ranges of applica-

bility were delineated. The investigated modes included the fuel-lean catalytically

stabilized thermal combustion (CST), the inverse catalytically stabilized thermal combus-

tion (i-CST), and the fuel-rich catalytic/gaseous-lean combustion. In the fuel-lean CST

approach, the less-than-unity Lewis number of hydrogen led to superadiabatic surface

temperatures. Even though the presence of homogeneous combustion moderated the wall

temperature superadiabaticity, reactor and catalyst thermal stability considerations

limited the CST mode to equivalence ratios less than 0.2. In the i-CST mode, the maximum

attained wall temperatures strongly depended on the location of homogeneous ignition, as

it directly affected the magnitude of the wall temperature superadiabaticity induced by

heat recirculation inside the solid walls. The i-CST had advantages over CST in terms of

reactor thermal management and extended the range of applicable equivalence ratios to

0.3. In the fuel-rich catalytic combustion mode, the larger-than-unity Lewis number of the

deficient oxygen reactant and the cooling of the catalytic channel by the resulting bypass

air led to surface temperatures below the adiabatic equilibrium value, even for those cases

where homogeneous combustion was established in the channel. This mode facilitated

safe reactor operation for overall equivalence ratios up to 0.5 and inlet temperatures up to

800 K, an operational range of particular interest for gas turbines.

Copyright ª 2013, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights

reserved.

1. Introduction large-scale power generation, hydrogen-rich syngases can be

Combustion of hydrogen and hydrogen-rich fuels has

attracted increased attention in large power plants and also

in microreactors used for portable power generation [1]. In

; fax: þ41 56 3102199.h (J. Mantzaras).2013, Hydrogen Energy P69

produced via fossil fuel decarbonization, leading to pre-

combustion capture of CO2 [2,3]. Hydrogen-containing fuels

are also intensively investigated in the latest post-

combustion CO2 capture concepts for natural-gas-fired

ublications, LLC. Published by Elsevier Ltd. All rights reserved.

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n en e r g y 3 8 ( 2 0 1 3 ) 1 0 6 5 4e1 0 6 7 0 10655

power plants [4,5]. Therein, the reactants are diluted with

large exhaust gas recycle (EGR) so as to increase the CO2

content in the flue gases and thus to facilitate its subsequent

capture [6,7]. In such approaches, the addition of hydrogen in

the natural gas fuel is highly advantageous as it increases the

combustion stability of the heavily diluted reactive mixture

[8]. Finally, hydrogen and hydrogen-rich fuels are also of in-

terest in microreactors of small (w100 We) power generation

units [9e11]. In such systems, hydrogen can be produced

from methane or higher hydrocarbons in suitable microre-

formers [12e16].

Catalytic combustion methodologies are of interest in

power generation due to their enhanced combustion stability

at very lean equivalence ratios and the ensuing ultra-low NOx

emissions [17e19]. In the conventional catalytically stabilized

thermal combustion (CST) mode [18,20] (see Fig. 1a), frac-

tional fuel conversion is achieved in a heterogeneous (cata-

lytic) reactor, which is typically Pd- or Pt-coated and is

operated at fuel-lean stoichiometries, while the remaining

fuel is combusted in a follow-up homogeneous (gas-phase)

combustion zone, again at fuel-lean stoichiometries. How-

ever, for diffusionally imbalanced limiting reactants with

Lewis numbers (Le) less than unity, such as H2 whereby

LeH2 w 0.3 at fuel-lean stoichiometries in air, CST is com-

pounded by the attainment of superadiabatic surface tem-

peratures [21e23] that endanger the catalyst and reactor

integrity.

Fig. 1 e Hetero-/homogeneous combustion modes: (a) fuel-

lean catalytically stabilized combustion (CST), (b) catalytic-

rich/gaseous-lean combustion, and (c) fuel-lean inverse

catalytically stabilized combustion (i-CST), whereby

gaseous combustion precedes the catalyst.

Alternative to the fuel-lean CST is the more recent

catalytic-rich/gaseous-lean combustion (Fig. 1b), which was

initially proposed for natural-gas-fueled power generation

systems [24,25]. In this mode, part of the air and all of the

natural gas are driven in a catalytic partial oxidation (CPO)

reactor at fuel-rich stoichiometries to produce syngas (CO and

H2). Syngas and unconverted reactants are subsequently

mixed with the bypass air, forming an overall fuel-lean ho-

mogeneous combustion zone. The fuel-rich catalytic com-

bustionmethodology has several advantages compared to the

conventional fuel-lean CST, which include extended catalytic

extinction limits [24e27], control of catalytic combustion by

the availability of oxygen that in turn prevents complete fuel

consumption inside the CPO module even in the event of

accidental gas-phase ignition, and enhanced stability of the

follow-up flame due to the presence of highly-reactive

hydrogen in the syngas [8].

The catalytic-rich/gaseous-lean combustion mode in Fig. 1b

is applicable not only to natural gas but also to pure hydrogen

or syngas fuels. In this case the catalyst does not have a prime

CPO function, but rather acts as a preheater and stabilizer for

the subsequent homogeneous combustion zone. This method

is suitable for a wide range of fuels that include low calorific

value biofuels, whereby flame stability is an issue, and also for

hydrogen-rich syngas fuels for which conventional lean-

premixed gaseous combustion entails the risk of flame flash-

back. The aforementioned advantages have led to the devel-

opment of integrated hetero-/homogeneous combustors based

on the catalytic-rich/gaseous-lean approach for coal-derived

syngas and also for pure hydrogen fuels [24,28]. Finally, in

fuel-rich combustion of hydrogen or syngas, reactor thermal

management is not an overriding issue as the catalyst tem-

perature does not exceed the adiabatic equilibrium tempera-

ture. This is because the limiting reactant is oxygen, which has

a Lewis number larger than unity (LeO2 w 2 at fuel-rich H2/air

stoichiometries).

The fundamentals of hydrogen hetero-/homogeneous

combustion have been studied in the last years. Bui et al. [29]

investigated numerically the combustion of fuel-lean and

fuel-rich H2/air mixtures at atmospheric pressure in a stag-

nation flow over a platinum surface. Appel et al. [22] applied in

situ laser-based measurements over a Pt catalyst surface and,

in conjunctionwith numerical predictions, provided validated

hetero-/homogeneous chemical reaction schemes for non-

preheated fuel-lean H2/air mixtures at atmospheric pressure.

More recently, Mantzaras et al. [30] and Ghermay et al. [31,32]

extended the investigation in Ref. [22] to pressures up to 15 bar

and mixture preheats up to 790 K, establishing the regimes

where the competition between the two reaction pathways

led to suppression of homogeneous combustion. The ignition

propensity of lean hydrogen/air mixtures was investigated

experimentally in stagnation flows over platinum surfaces

[33]. Finally, Maestri et al. [34,35] studied experimentally and

numerically the combustion of fuel-rich H2/air mixtures at

atmospheric pressure in a nearly isothermal Rh-coated cata-

lytic annular reactor.

Of particular interest in the present work is a new fuel-lean

hydrogen catalytic combustion mode proposed in Refs. [31],

which is henceforth termed inverse CST or “i-CST”, wherein

gaseous combustion precedes the catalytic one (see Fig. 1c).

Fig. 2 e Simulated domains for the three modes in Fig. 1. In

all modes, the catalytic channel has a length L [ 75 mm

and a half-height b [ 0.6 mm. In Modes A (a) and C (c) the

outer channel wall surface is insulated, while in Mode B (b)

bypass air flows in adjacent channels with height

H [ 2b [ 1.2 mm.

i n t e rn a t i o n a l j o u r n a l o f h y d r o g e n en e r g y 3 8 ( 2 0 1 3 ) 1 0 6 5 4e1 0 6 7 010656

Homogeneous combustion is initiated inside an inert porous

burner, which is in turn ignited via heat conduction and radia-

tion heat transfer from a downstream catalytic reactor. At

steady operation, the catalytic module is exposed to the hot

combustionproductsof thehomogeneouscombustionzone.As

long as complete fuel conversion is achieved homogeneously,

themaximumtemperatureencounteredbythecatalytic reactor

may be limited by the adiabatic equilibrium temperature of the

reactive mixture. As a matter of fact, superadiabatic surface

temperatures can still be attained in i-CST due to heat recircu-

lation in the solid structure of the gaseous burner, as will be

shown in Section 3.3. Nevertheless, this type of super-

adiabaticity is generally smaller than the catalytic-combustion-

induced one.

The present work undertakes a comparative study of the

three modes in Fig. 1 for hydrogen/air combustion with an

overall (based on the total air and hydrogen) fuel-lean stoi-

chiometry. Detailed simulations are performed in a single

plane channel catalytic reactor using a 2-D numerical model

with elementary heterogeneous and homogeneous chemical

reaction schemes, heat conduction in the solid wall, in-

channel surface radiation heat transfer and detailed species

transport. Main objective is the delineation of the operating

regimes (in terms of fuel-to-air equivalence ratio, mixture

preheat and catalyst maximum tolerable wall temperature)

whereby each of the three examined modes exhibits a clear

advantage. Additional objectives are to address the funda-

mental coupling of hetero-/homogeneous chemistry and

transport in both fuel-lean and fuel-rich hydrogen/air

mixtures.

This article is organized as follows. The numerical model

and kinetic schemes are provided in Section 2, while the un-

derlying physicochemical processes during fuel-lean and fuel-

rich catalytic combustion of hydrogen/air mixtures are

reviewed in Section 3.1. Results for the three different modes

are presented in Sections 3.2e3.4, whereby the interactions of

chemistry, transport and heat transfer mechanisms in the

solid and their impact on the reactor thermal management

are exposed and the safe operating regimes are identified.

Comparisons between the investigated modes and implica-

tions for practical combustion systems are finally outlined in

Section 3.5.

2. Numerical

2.1. Hetero-/homogeneous combustion channel model

For all modes, the catalytic reactor was modeled as a single

planarchannelwitha length L¼ 75mm,height 2b¼ 1.2mmand

awall thickness d¼50mm(seeFig. 2�due tosymmetryonlyhalf

of the domain was considered). This geometry simulated an

individual channel of practical honeycomb reactors [20,26] and

also single-channel catalytic reactors with large cross-flow

aspect ratios [9]. For Modes A (CST) and B (fuel rich combus-

tion), the entire channel length was coated with platinum.

However, for Mode C (i-CST), the initial channel length

Li ¼ 15 mm was chemically inert (non-catalytic) while the

remaining 60 mm was coated with platinum (Fig. 2c). This

approach allowed for the establishment of an upstream

homogeneous combustion zone in the inert channel section, as

will be shown in Section 3.3. Thus, the same combustionmode

as the one in Fig. 1c could be obtained, while avoiding the

complex modeling of homogeneous combustion in a porous

medium and its thermal coupling to the follow-up gas-phase

combustion zone.

While the CST Mode A has appeared in many variants [20],

here the conventional approach with adiabatic outer channel

surfaces has been considered. The outer walls in Mode C were

also adiabatic. In Mode B heat transfer occurred with the

bypass air, which flowed in two adjacent channels (one above

and the other below the catalytic channel), each having a

height H ¼ 2b. The gaseous combustion downstream of the

catalyst in Modes A and B (Fig. 1a, b) is not addressed herein.

Of prime interest in this work is the behavior of the catalytic

reactor itself and not the subsequent homogeneous combus-

tion; in any case, the post-catalyst flames involve intricate

aerodynamic stabilization for Mode A and turbulent mixing of

the two streams in Mode B.

To simulate the laminar flow in the catalytic channel, a full

elliptic two-dimensional CFD code was used [31,36] for all

combustion modes. It is stressed that the presence of gaseous

combustionwithin the catalytic channel invalidates the use of

common 1-D models with lumped heat and mass transfer

coefficients. This is because the onset of homogeneous igni-

tion inside the catalytic channels strongly depends on the

boundary layer profiles of temperature and reactive species,

as shown in analytical studies [37,38]. The governing equa-

tions for a steady reacting flow in Cartesian coordinates

become:

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n en e r g y 3 8 ( 2 0 1 3 ) 1 0 6 5 4e1 0 6 7 0 10657

Continuity equation:

vðruÞvx

þ vðrvÞvy

¼ 0: (1)

Momentum equations:

vðruuÞvx

þvðrvuÞvy

þvpvx

� v

vx

�2m

vuvx

�23m

�vuvx

þvvvy

��� v

vy

�m

�vuvy

þvvvx

��¼0;

(2)

vðruvÞvx

þvðrvvÞvy

þvpvy

� v

vx

�m

�vvvx

þvuvy

��� v

vy

�2m

vvvy

�23m

�vuvx

þvvvy

��¼0:

(3)

Energy equation:

vðruhÞvx

þ vðrvhÞvy

þ v

vx

rXKgk¼1

YkhkVk;x � lgvTvx

!

þ v

vy

rXKgk¼1

YkhkVk;y � lgvT

vy

!¼ 0:

(4)

Gas phase species equation:

vðruYkÞvx

þ vðrvYkÞvy

þ v

vxðrYkVk;xÞ þ v

vy

�rYkVk;y

�� _ukWk ¼ 0;

k ¼ 1;.; Kg:

(5)

Species diffusion velocities V!

k were computed using

mixture average diffusion, including thermal diffusion for the

light species H and H2 [39]:

V!

k ¼ �DkmV�ln�YkW=Wk

�þ �DTkWk=

�rYkW

�VðlnTÞ: (6)

Finally, the ideal gas and caloric state laws closed the sys-

tem of equations:

p ¼ rRT

Wand hk ¼ ho

kðToÞ þZTTo

cp;kdT: (7)

At the channel inlet (x ¼ 0), uniform profiles were used for

the temperature, axial velocity and speciesmass fractions. No

slip was applied at the gasewall interface ( y ¼ b) for both

velocity components. At the outlet (x ¼ 75 mm) and the

symmetry plane ( y ¼ 0) the transverse velocity was zero

(v ¼ 0), while zero-Neumann conditions were applied for all

other variables. The interfacial boundary conditions for the

gaseous species at the gasewall interface ( y ¼ b), were:

ðrYkVk;yÞy¼b ¼ Wk _sk; k ¼ 1;.; Kg; (8)

whereby for Mode C _sk ¼ 0 over the non-catalytic length

0 � x � Li. Implicit in Eq. (8) is the assumption that the ratio of

the catalytically active area to the geometrical surface area is

unity. Although in practical systems this ratio depended on

the surface porosity and catalyst dispersion, its specific value

was largely irrelevant for the present study as the catalytic

conversion of the limiting reactant (H2 in fuel-lean and O2 in

the fuel-rich modes) was nearly transport limited. Finally, the

surface species coverage equations over the catalytically-

active sections of the channel were obtained by solving:

sm

_smG

¼ 0; m ¼ 1;.; Ms: (9)

For the solid channel wall, a 2-D approach was also adopted

for the steady heat conduction:

ls

�v2TW

vx2þ v2TW

vy2

�¼ 0; (10)

with ls ¼ 16 Wm�1 K�1 referring to FeCr-alloy, a common

material for catalytic honeycomb reactors in power genera-

tion [20,26].

The interfacial energy boundary condition at y ¼ b was:

_qrad � lg

�vTvy

�y¼b�

þ ls

�vTvy

�y¼bþ

þXKg

k¼1

ð _skhkWkÞ ¼ 0: (11)

The term _qrad accounted for radiation exchange of each

channel surface element with all other channel surface ele-

ments and also with the channel inlet and outlet. The net

radiation method for diffuse-gray areas [10,40] was employed

to compute the radiation exchange between the discretized

channel wall surface elements themselves and between each

surface element and the inlet and outlet channel enclosures.

For the [-th channel surface element, the radiation balance

becomes:

1ε[

qrad;[¼XNþ2

j¼1

F[�jsT4[ �T4

j

�þXNþ2

j¼1

�1�εj

εj

�F[�jqrad;j; l¼1;.;N; (12)

with the index j running over the N discretized wall surface

elements as well as the inlet ( j ¼ N þ 1) and outlet ( j ¼ N þ 2).

The emissivity of each channel surface element was ε ¼ 0.6,

while the inlet and outlet sectionswere treated as black bodies

(εNþ1 ¼ εNþ2 ¼ 1.0). At the inlet and outlet solid wall vertical

faces, radiative boundary conditions were applied:

lsvTwðx ¼ 0; yÞ

vx¼ εs

�T4wðx ¼ 0; yÞ � T4

IN

for b < y ¼ bþ d;

�lsvTwðx¼L;yÞ

vx¼ εs

�T4wðx¼L;yÞ�T4

OUT

forb<y¼bþd: (13)

At the outer surface of the catalytic channel, adiabatic

conditions were imposed for Modes A and C:

vTwðx; y ¼ bþ dÞ=dy ¼ 0; for 0 < x < L: (14)

For Mode B, the outer surface of the bypass channel

( y ¼ b þ d þ H ) was adiabatic (see Fig. 2b). Cooling of the

catalytic channel with the bypass air was modeled with a

lumped heat transfer coefficient. Local Nusselt numbers have

been tabulated in Shah and London [41] for the combined

hydrodynamically and thermally developing flow in a planar

channel with one wall adiabatic and the other having con-

stant heat flux. One-dimensional modeling was a reasonable

simplification since the bypass flow was non-reacting. Using

local Nusselt numbers, Nux,H, from Ref. [41], the boundary

condition at the outer surface of the catalytic channel

( y ¼ b þ d) in Mode B becomes:

_qðxÞ ¼ �Nux;Hlg=dh

�½Twðx; y ¼ bþ dÞ � TmðxÞ�; (15)

with

_mair cpðxÞdTmðxÞ=dx ¼ _qðxÞ: (16)

Table 2 e Homogeneous chemical reaction mechanismfor H2.

a

A b E

H2/O2 reactions

R1 H þ O2 ¼ O þ OH 3.55 � 1015 �0.41 69.45

R2 O þ H2 ¼ H þ OH 5.08 � 104 2.67 26.32

R3 H2 þ OH ¼ H2O þ H 2.16 � 108 1.51 14.35

R4 O þ H2O ¼ OH þ OH 2.97 � 106 2.02 56.07

H2/O2 dissociation-recombination

R5 H2 þ M ¼ H þ H þ M 4.58 � 1019 �1.40 436.73

R6 O þ O þ M ¼ O2 þ M 6.16 � 1015 �0.50 0.00

R7 O þ H þ M ¼ OH þ M 4.71 � 1018 �1.0 0.00

R8 H þ OH þ M ¼ H2O þ M 3.80 � 1022 �2.00 0.00

HO2 formation-consumption

R9 H þ O2 þ M ¼ HO2 þ M 1.48 � 1012 0.60 0.00

H þ O2 þ M ¼ HO2 þ M 6.37 � 1020 �1.72 2.18

R10 HO2 þ H ¼ H2 þ O2 1.66 � 1013 0.00 3.43

R11 HO2 þ H ¼ OH þ OH 7.08 � 1013 0.00 1.26

R12 HO2 þ O ¼ O2 þ OH 3.25 � 1013 0.00 0.00

R13 HO2 þ OH ¼ H2O þ O2 2.89 � 1013 0.00 �2.09

H2O2 formation-consumption

R14 HO2 þ HO2 ¼ H2O2 þ O2 4.22 � 1014 0.00 50.12

R15 HO2 þ HO2 ¼ H2O2 þ O2 1.30 � 1011 0.00 �6.82

R16H2O2 þ M ¼ OH þ OH þ M 2.95 � 1014 0.00 202.63

H2O2 þ M ¼ OH þ OH þ M 1.20 � 1017 0.00 190.37

R17 H2O2 þ H ¼ H2O þ OH 2.41 � 1013 0.00 16.61

R 18H2O2 þ H ¼ H2 þ HO2 4.81 � 1013 0.00 33.26

R19 H2O2 þ O ¼ OH þ HO2 9.55 � 106 2.00 16.61

R20 H2O2 þ OH ¼ H2O þ HO2 1.00 � 1012 0.00 0.00

R21 H2O2 þ OH ¼ H2O þ HO2 5.80 � 1014 0.00 40.00

a From Ref. [47]. Reaction rate k ¼ ATbexp (�E/RT ), A [mole-cm-

Kelvin-s], E [kJ/mol]. Third body efficiencies in reactions R5-R8

and R16 are u(H2O) ¼ 12.0, u(H2) ¼ 2.5, u(CO2) ¼ 3.8, while in R9

u(H2O) ¼ 11.0, u(H2) ¼ 2.0, u(O2) ¼ 0.78, u(CO2) ¼ 3.8. Reaction pairs

(R14, R15) and (R20, R21) are duplicate. Reactions R9 and R16 are

i n t e rn a t i o n a l j o u r n a l o f h y d r o g e n en e r g y 3 8 ( 2 0 1 3 ) 1 0 6 5 4e1 0 6 7 010658

In Eqs. (15) and (16) dh is the hydraulic diameter of the

bypass channel, which is twice the full channel height 2H [41],

Tm(x) is the mean temperature of the bypass air flow, lg the

thermal conductivity of the air evaluated at a temperature

0.2Tw þ 0.8Tm, cp(x) the local mixture heat capacity evaluated

at Tm(x), and _mair the air mass flow rate in one bypass channel

per unit channel width. The above 1-D heat transfer model for

the inert channel gave good agreement with full 2-D simula-

tions, as will be shown in Section 3.4.

An orthogonal staggered grid with 280 � 60 points (in x and

y, respectively), over the channel half-domain produced a

grid-independent solution for the flow. Finer spacing towards

the channel wall and entry section was used. Furthermore, a

280 � 20 grid resolution (in x and y, respectively) discretized

the solid wall. Details of the solution algorithm have been

provided elsewhere [22,23].

2.2. Chemical kinetics

Detailed surface reaction mechanisms [42e44] as well as

reduced ones [45] have been reported for the catalytic oxida-

tion of hydrogen on platinum. The detailed heterogeneous

reaction scheme from Deutschmann et al. [42] (11 irreversible

and three reversible reactions, 5 surface and 6 gaseous spe-

cies, see Table 1) was employed, with a platinum surface site

density G ¼ 2.7 � 10�9 mol/cm2. Surface thermodynamic data

for the three reversible reactions in Table 1 were taken from

Ref. [46].

For homogeneous chemistry, the H2/O2 mechanism of Li

et al. [47] (see Table 2) was employed (21 reversible reactions

and 9 species, including the carrier N2) with its accompanying

gas-phase thermodynamic data. This mechanism has been

Table 1 e Catalytic reaction scheme for H2 oxidation onPt.a

A (g) b E

Adsorption reactions

S1 O2 þ 2Pt(s) / 2O(s) 0.023 0.0 0.0

S2 O2 þ 2Pt(s) / 2O(s) 1.8 � 1021 �0.5 0.0

S3 H2 þ 2Pt(s) / 2H(s) 4.5 � 1010 0.5 0.0

S4 H þ Pt(s) / H(s) 1.0 0.0 0.0

S5 O þ Pt(s) / O(s) 1.0 0.0 0.0

S6 H2O þ Pt(s) / H2O(s) 0.75 0.0 0.0

S7 OH þ Pt(s) / OH(s) 1.0 0.0 0.0

Surface reactions

S8 H(s) þ O(s) ¼ OH(s) þ Pt(s) 3.7 � 1021 0.0 11.5

S9 H(s) þ OH(s) ¼ H2O(s) þ Pt(s) 3.7 � 1021 0.0 17.4

S10 OH(s) þ OH(s) ¼ H2O(s) þ O(s) 3.7 � 1021 0.0 48.2

Desorption reactions

S11 2O(s) / O2 þ 2Pt(s) 3.7 � 1021 0.0 213.2-60qOS12 2H(s) / H2 þ 2Pt(s) 3.7 � 1021 0.0 67.4-6qHS13 H2O(s) / H2O þ Pt(s) 1.0 � 1013 0.0 40.3

S14 OH(s) / OH þ Pt(s) 1.0 � 1013 0.0 192.8

a from Ref. [42]. In all surface and desorption reactions, the reac-

tion rate coefficient is k ¼ ATbexp(-E/RT ), A [mole-cm-Kelvin-s] and

E [kJ/mol]. The surface site density is G ¼ 2.7 � 10�9 mol/cm2. In all

adsorption reactions, except S2 and S3, A denotes a sticking coef-

ficient (g). Reactions S1 and S2 are duplicate. Reaction S3 has an

order of one with respect to platinum. The suffix (s) denotes a

surface species.

Troe reactions centered at 0.8 and 0.5, respectively (second entries

are the low pressure limits).

validated against shock tube, flow-reactor, and laminar flame

speed measurements at pressures up to 87 bar. Moreover, its

aptness for combined hetero-/homogeneous hydrogen com-

bustion over Pt was assessed with in-situ OH-LIF and Raman

measurements for lean stoichiometries in Refs. [31,32] and

more recently for rich stoichiometries in Ref. [48].

Gas-phase and surface reaction rates were evaluated with

CHEMKIN Ref. [49] and Surface-CHEMKIN Refs. [50], respec-

tively, while transport properties were calculated from the

CHEMKIN transport database [39].

3. Results and discussion

The investigated parameter ranges for each mode are sum-

marized in Table 3. The inlet pressure was p ¼ 1 bar, the

preheat temperature ranged from TIN¼ 300 Ke800 K, while the

total equivalence ratio (4TOT, based on the total amount of air)

was fuel-lean and varied from 0.1 to 0.5. For Mode B, the

equivalence ratios inside the catalytic channel ranged from

1.0 to 5.0, while TIN was the same for both reacting and bypass-

air flows. Two total inflow velocities UIN were considered,

10 m/s and 20 m/s. In Modes A and C, the total UIN was the

same with the inflow velocity in the catalytic channel. For

Table 3 e Operating conditions for the three modes inFig. 2.a

4TOT UIN (m/s) TIN (K) 4CAT Bypass air(%)

A (CST) 0.1e0.3 10, 20 300e800 0.1e0.3 0

B (fuel-rich) 0.2e0.5 (2.2e7.8),

(4.4e15.6)

300e800 1.0e5.0 32.7e91.3

C (i-CST) 0.1e0.3 10, 20 300e800 0.1e0.3 0

a Total equivalence ratio (4TOT), inlet velocity in the catalytic

channel (UIN), inlet temperature (TIN), equivalence ratio in the cat-

alytic channel (4CAT, different than 4TOT for Mode B), and bypass air

(% of the total mass flux).

Fig. 3 e Wall temperature profiles in a catalytic channel

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n en e r g y 3 8 ( 2 0 1 3 ) 1 0 6 5 4e1 0 6 7 0 10659

Mode B, however, the inflow velocities in the catalytic channel

and the bypass air were adjusted to yield a total mass flux

(rIN,CATUIN,CAT þ rIN,bypassUIN,bypass) equal to that of the corre-

sponding cases inModes A and C (rINUIN) that shared the same

4TOT, TIN and total UIN. This resulted for Mode C in bypass air

accounting for 32.7%e91.3% of the total incoming air flow (see

Table 3).

with length 75 mm, height 1.2 mm, wall thickness

d [ 0.05 mm and solid wall thermal conductivity

ls [ 16 WmL1 KL1 (geometry in Fig. 2a). Computations at

two H2/air stoichiometries (4 [ 0.3 pertaining to Modes A

and C and 4 [ 6.9 pertaining to Mode B) having the same

adiabatic equilibrium temperature Tad. For 4 [ 0.3 results

are shown for the adiabatic transport-limited Case 1, for

Case 2 with only catalytic reactions (C ), and for Case 3 with

both catalytic and gas-phase reactions (CeG). For 4 [ 6.9

results are shown for the adiabatic transport-limited Case

4 and for Case 5 with catalytic and gas-phase reactions

(CeG).

3.1. Impact of Lewis number of deficient reactant onsurface temperature

The attained catalyst surface temperatures crucially depen-

ded on the Lewis number of the limiting reactant. The nature

of this dependence, which apart from transport also involved

hetero-/homogeneous chemistry coupling and heat transfer

mechanisms in the solid, is firstly clarified in this section.

Computed axial temperature profiles at the gasewall

interface ( y¼ b) are shown in Fig. 3 for caseswith lean (4¼ 0.3)

and rich (4 ¼ 6.9) equivalence ratios; these plots were also

representative of the temperatures inside the solid since,

under the present steady-state conditions, solid temperature

differences in the transverse direction (b � y � b þ d) were less

than 1.5 K. In fuel-lean H2/air combustion, superadiabatic

surface temperatures were attained as illustrated by three

computations in Fig. 3 (Cases 1e3) pertaining to a 4 ¼ 0.3

mixture in the channel geometry of Fig. 2a with UIN ¼ 10 m/s

and TIN ¼ 300 K. Case 1 (curve marked Tw,1) was computed

with an infinitely fast catalytic step H2 þ ½O2 / H2O (i.e.

transport-limited catalytic hydrogen conversion), without

gas-phase chemistry and without the inclusion of heat con-

duction in the solid or surface radiation heat transfer.

Appropriate transport-limited interfacial boundary conditions

for the gas-phase species were imposed (see also [38]) which,

with H2 being the limiting reactant, become:

YH2¼ 0;

and

�rYkVk;y

�y¼b

¼�n0k�n}

k

�Wk

WH2

�rYH2VH2 ;y

�y¼b

; k¼O2;H2O;N2; (17)

with n0k and n}k the stoichiometric coefficients of species k in

the reactants and products, respectively, for the reaction

H2 þ ½O2 / H2O. The corresponding interfacial energy

boundary condition for the simulation of Case 1 was a local

balance between surface heat generation and convection to

the gas:

�lg

�vTvy

�y¼b�

þXKg

k¼1

ð _skhkWkÞ ¼ 0: (18)

The boundary condition Eq. (18) was the limiting case of Eq.

(11) for an infinitely-thin wall (d / 0) without surface radia-

tion. The maximum temperature for Case 1 occurred at the

channel entry (x / 0) and was Tw,1,max ¼ 2143 K, i.e. 954 K

above the adiabatic equilibrium temperature Tad ¼ 1189 K. As

the transport-limited catalytic channel solution at x / 0 was

equivalent to the corresponding transport-limited solution

over a catalytic flat plate [37], the maximum computed tem-

perature Tw,1,max could be compared against theoretical re-

sults. The computed Tw,1,max ¼ 2144 K was in excellent

agreement with the theoretical transport-limited solution,

Tw,max ¼ 2143 K, obtained over a catalytic flat plate by the

formula [22,29]:

Tw;max ¼ TIN þ ðDTÞcLe�2=3H2

; (19)

or equivalently,

Tw;max � Tad ¼ ðDTÞcLe�2=3

H2� 1�; (20)

where (DT )c is the adiabatic combustion temperature rise,

(DT )c ¼ Tad e TIN ¼ 889 K, and LeH2¼ 0.335. The Lewis number

x (mm)0 10 20 30 45 60 75

H2

conv

ersi

on ra

te (m

ol/m

2 s)

0

1

2

3

4

5

6

% H

2 co

nver

sion

-50

-25

0

25

50

75

100

C3

G3

C2

Case 2

Case 3

xig

Fig. 4 e Catalytic and gas-phase hydrogen conversion rates

for Case 3 in Fig. 3 (C3 and G3, respectively) and catalytic

hydrogen conversion rate for Case 2 in Fig. 3 (C2). The %

hydrogen conversions are also shown for Cases 2 and 3.

The arrow marked xig denotes the location of

homogeneous ignition for Case 3.

i n t e rn a t i o n a l j o u r n a l o f h y d r o g e n en e r g y 3 8 ( 2 0 1 3 ) 1 0 6 5 4e1 0 6 7 010660

of the limiting hydrogen reactant, LeH2, was computed for a

4 ¼ 0.3 H2/air mixture using mixture-average diffusion [39]

with transport properties evaluated at (Tw,1,max þ TIN)/2. The

computed wall temperature Tw,1 dropped monotonically with

increasing axial distance to the adiabatic equilibrium tem-

perature Tad ¼ 1189 K (see Fig. 3) upon complete conversion of

hydrogen.

Additional results are shown in Fig. 3 for the 4 ¼ 0.3 H2/air

mixture, using finite-rate chemistry, and the full model of

Section 2.1 with solid heat conduction and surface radiation

heat transfer: Case 2 (Tw,2) referred to predictions using only

the detailed catalytic reaction mechanism (C ) and Case 3

(Tw,3) to simulations with both the catalytic and gaseous (C-G)

reaction mechanisms of Section 2.2. Simulations with only

catalytic chemistry, Tw,2, yielded a maximum wall tempera-

ture of 1756 K, i.e. 567 K above the adiabatic equilibrium

temperature. Although these values were appreciably lower

than those obtained for the transport-limited adiabatic solu-

tion Tw,1, they were still high and posed great concerns for the

reactor thermal management and catalyst stability. The

reduced degree of superadiabaticity in Case 2 was a combined

result of radiation heat losses towards the cold entry, redis-

tribution of energy in the wall via solid heat conduction and

finite-rate surface chemistry. Moreover, the shift of the peak

temperature location from x ¼ 0 (Tw,1) to x ¼ 5.3 mm (Tw,2) was

a result of increased radiation heat losses at the channel entry

due to the corresponding higher view-factors (Fj,Nþ1 in Eq. (12))

and also due to the increased finite-rate surface chemistry

effects at the leading edge (x / 0). Finally, it could be shown

that for Case 2 at axial locations x > 2 mm catalytic reactions

were largely transport-controlled since YH2 ( y ¼ b)/

YH2 ( y ¼ 0) < 0.08.

Simulations with combined catalytic and gas-phase

chemistry (CeG) at 4 ¼ 0.3 (Case 3 in Fig. 3) yielded a lower

superadiabaticity compared to the transport-limited Case 1

and the only-catalytic-chemistry (C ) Case 2, with a maximum

wall temperature of 1698 K. Reason for this quite interesting

behavior was the presence of gaseous combustion, which

partially shielded the catalyst from the hydrogen-rich channel

core and hence reduced the heterogeneously-driven super-

adiabaticity. For Case 3, axial profiles are given in Fig. 4 of the

catalytic (C3) and gas-phase (G3) conversion rates of hydrogen

as well as the % total conversion of hydrogen; superposed are

also the catalytic conversion rate (C2) and the % hydrogen

conversion for Case 2. The C rates in Fig. 4 accounted for the

catalytic conversion on one catalytic surface ( y ¼ b), while the

G3 rate was calculated by integrating the volumetric gaseous

H2 conversion over the 0.6mm channel half-height. The C and

G rates are herein considered positive when referring to con-

sumption of hydrogen. Homogeneous ignition for Case 3 was

achieved close to the channel entry, at xigz 2mm; the ignition

distance xig was defined as the far-upstream position whereby

OH reached 5% of its maximum value in the channel and this

location closely corresponded to the sharp rise in the G3

conversion of hydrogen in Fig. 4. It is stressed that although all

modes in Fig. 1 presumed only heterogeneous combustion

inside the catalytic module, in reality gaseous combustion

could not be neglected under realistic operating conditions

either for hydrogen and syngas [23,32,51] or for lower hydro-

carbon fuels [10,52,53].

For Case 3 in Fig. 3, over the extent of appreciable gaseous

hydrogen conversion (2 mm � x � 22 mm), catalytic conver-

sion (C3) occurred in parallel with G3, albeit with decreasing

magnitude at increasing axial distances (Fig. 4). Hydrogen

could still leak through the gaseous combustion zone to be

subsequently converted catalytically at the channel wall,

although at diminished magnitudes when compared to the

only catalytic combustion Case 2. Therefore, the presence of

gaseous combustion limited the amount catalytic conversion

which was responsible for the superadiabatic wall tempera-

tures, as further seen by comparing the C3 and C2 curves in

Fig. 4. As in matter of fact, the proposed i-CST Mode C in Figs.

1c and 2c could be envisioned as a limiting process for Case 3

in Fig. 4 whereby G3 completely dominated over C3, hence

lowering the wall temperature down to Tad. Apart from the

moderation of the wall temperature, the presence of gaseous

combustion had the additional advantage of allowing com-

plete conversion of hydrogen at shorter reactor lengths; this

was because the gas-phase reaction pathway did not exhibit

the strong transport limitations of the heterogeneous

pathway. Characteristically, 99.9% of hydrogen conversion

was attained at x ¼ 40.2 mm for Case 3 and at x ¼ 65.5 mm for

Case 2 (see Fig. 4).

Additional simulations are provided in Fig. 3 for the fuel-

rich stoichiometry 4 ¼ 6.9 (Cases 4e5), which had the same

adiabatic equilibrium temperature as 4 ¼ 0.3 (Tad ¼ 1189 K). To

make a direct comparison with the forgoing 4 ¼ 0.3 results,

simulations were performed in the channel geometry of

Fig. 2a. Predictions in Fig. 3 referred to adiabatic, transport-

limited conditions (Case 4, Tw,4) and to finite-rate chemistry

with detailed hetero-/homogeneous reaction mechanisms

and the full model of Section 2.1 (Case 5, Tw,5). With increasing

axial distance, the transport-limited solution Tw,4 approached

the adiabatic equilibrium temperature Tad ¼ 1189 K from

underadiabatic values. The Lewis number of the deficient

reactant, calculated using a mixture-average transport model

as in the previous lean case, was LeO2 ¼ 2.268. Substituting LeO2

to Eq. (19), the resulting theoretical minimum temperature

x (mm)0 15 30 45 60 75O

2 co

nver

sion

rate

(mol

/m2 s

)

0.0

0.5

1.0

1.5

2.0

% O

2 co

nver

sion

0

25

50

75

100

C

G

Fig. 5 e Catalytic (C ) and gas-phase (G) oxygen conversion

rates and % oxygen conversion for the 4 [ 6.9 Case 5 in

Fig. 3.

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n en e r g y 3 8 ( 2 0 1 3 ) 1 0 6 5 4e1 0 6 7 0 10661

(flat plate solution) was Tw,min ¼ 815 K, again in excellent

agreement with the computed Tw,4,min¼ 814 K at x/ 0 (Fig. 3).

Computations using finite-rate chemistry with surface radia-

tion and heat conduction (Tw,5 in Fig. 3) yielded temperatures

20e88 K lower than Tw,4; these modest differences (compared

to the larger differences in the lean Cases 1e3) were attributed

to the less significant radiation heat losses, a result of the

substantially lower wall temperatures at the reactor entry.

Catalytic and gas-phase conversion rates as well as the

percentage conversion of the limiting O2 reactant are provided

in Fig. 5 for Case 5. Gas-phase (G) conversion was practically

zero, as the wall temperature Tw,5 was too low to ignite ho-

mogeneous combustion. Nonetheless, it will be shown in the

coming Section 3.4 that gaseous combustion could be initiated

at rich stoichiometries for higher inlet temperatures. Fig. 5

further indicated a 97.3% O2 conversion at the channel exit

for Case 5. The incomplete oxygen conversion was mainly an

outcome of the lower molecular diffusivity of O2 rather than a

result of finite-rate surface chemistry: it could be shown than

even for the transport-limited Case 4 the corresponding O2

conversion was 98.6%. Finite-rate surface chemistry effects

x (mm)0 15 30 45 60 75

Tem

pera

ture

(K)

350

650

950

1250

1550

1850

H2

conv

ersi

on (m

ol/m

2 s)

0.0

1.2

2.4

3.6

4.8

6.0

Tw

Tm,gasG

C

ad=1233T KTw,max=1717K

0 15 30 45 60 75 x (mm)

(a)

(b)

214010700

-0.6

0.60.0

y(m

m)

CST

xig

00 15 30 45 60 75x (mm)x

ig

OH

Fig. 6 e Predictions at 4 [ 0.3, UIN [ 10 m/s, TIN [ 350 K for the

concentration (in ppm mass), (b, d) axial profiles of wall temper

gas-phase (G) hydrogen conversion. The horizontal lines marke

incoming mixture.

were present, especially close to the entry. These effects were

stronger than those of the lean Case 2 (Fig. 4); for example, at

x ¼ 0.5 mm the ratio YH2 ( y ¼ b)/YH2 ( y ¼ 0) was 0.098 whereas

YO2 ( y ¼ b)/YO2 ( y ¼ 0) was 0.185.

The foregoing discussion has exposed fundamental as-

pects of the three combustion modes. In Mode A, super-

adiabatic surface temperatures were attained, in Mode B the

surface temperatures were mainly underadiabatic, while in

Mode C the surface temperatures were largely adiabatic.

Detailed simulations for the specific geometries of each mode

(Fig. 2), with emphasis on the fundamental combustion pro-

cesses and reactor thermal management issues will be pre-

sented next.

3.2. Catalytically stabilized combustion (CST)

Fundamental processes involving the coupling of hetero-/

homogeneous kinetics, transport and heat loss mechanisms

are outlined, and the parameter space of suitability for the

CST mode is delineated. Fig. 6 provides the OH distributions,

the catalytic and gas-phase hydrogen conversions, and also

the wall and mean gas temperatures for CST (and i-CST as

well, for later comparison) at 4 ¼ 0.3, UIN ¼ 10 m/s and

TIN ¼ 350 K. Concentrating on the CST results (Fig. 6a, b), the 2-

D OH map in Fig. 6a illustrates the creation of two separate

flame branches near the catalytic channel walls. This was a

direct consequence of the LeH2 < 1: fuel was transported to-

wards the hot wall more efficiently that heat was transported

away from it, leading to the confinement of the ensuing

flames at the wall vicinity. It has been shown [22] that an

artificial increase of the hydrogen Lewis number to unity led

to formation of a closed V-shaped flame extending over the

entire channel height.

The confinement of the gaseous flame near the wall had

profound implications for the underlying combustion pro-

cesses and the reactor thermal management. The maximum

wall temperature in Fig. 6b was Tw,max ¼ 1717 K, i.e. 484 K

x (mm)0 15 30 45 60 75

Tem

pera

ture

(K)

350

650

950

1250

1550

1850

H2

conv

ersi

on (m

ol/m

2 s)

0.0

1.2

2.4

3.6

4.8

6.0

Tw

Tm,gasG

C

ad=1233KTw,max=1648K

0 15 30 45 60 75 x (mm)

(c)

(d)

328016400

-0.6

0.60.0

y(m

m)

i-CST

OH

xig

xig

T

CST (a, b) and the i-CST (c, d) modes; (a, c) 2D maps of OH

ature (Tw), mean gas temperature (Tm,gas), catalytic (C ) and

d Tad provide the adiabatic equilibrium temperature of the

Hydrogen mass fraction YH2

0.000 0.002 0.004 0.006 0.008 0.010

y (m

m)

0.0

0.1

0.2

0.3

0.4

0.5

0.6

Temperature (K)300 600 900 1200 1500 1800

x = 1.6 mm

x= 3.3 mm

x=8.

2 mm

x = 1.6 mmx= 3.3 mm

x=8.2

mm

Fig. 7 e Transverse profiles of hydrogen mass fraction and

temperature at three axial locations within the extent of

appreciable gas-phase G hydrogen conversion for the CST

case in Fig. 9(a, b). The gasewall interface is located at

y [ 0.6 mm.

i n t e rn a t i o n a l j o u r n a l o f h y d r o g e n en e r g y 3 8 ( 2 0 1 3 ) 1 0 6 5 4e1 0 6 7 010662

above the adiabatic equilibrium temperature Tad ¼ 1233 K.

Simulations without the inclusion of gaseous chemistry (not

shown in Fig. 6b) yielded Tw,max ¼ 1784 K, signifying that the

flame provided a modest thermal shielding of 67 K. However,

despite the lower wall temperature in the presence of gaseous

chemistry, the surface temperature moderation in Fig. 6b was

constrained by two facts; the presence of a short gaseous in-

duction zone (0 � x � xig, where xig w 0.7 mm, see Fig. 6a, b)

with pure catalytic conversion, and the inability of the flame

to fully suppress catalytic conversion downstream of the ho-

mogeneous ignition position, x > xig. Characteristically, over

the extent xig � x � xTw,max, where xTw,max ¼ 3.3 mm denotes

the location of maximum wall temperature, the ratio of the

CST

x (mm)0 15 30 45 60 75

Tem

pera

ture

(K)

350

590

830

1070

1310

1550

H2

conv

ersi

on (m

ol/m

2 s)

0.0

0.3

0.6

0.9

1.2

1.5

Tw

Tm,gas

G

C

ad=971T K

Tw,max=1431K

0 15 30 45 60 75 x (mm)

(a)

(b)

31 620

-0.6

0.60.0

y(m

m)

0 15 30 45 60 75x (mm)

OH

Fig. 8 e Predictions at 4 [ 0.2, UIN [ 10 m/s, TIN [ 350 K for the

concentration (in ppm mass), (b, d) axial profiles of wall temper

gas-phase (G) hydrogen conversion. The horizontal lines marke

incoming mixture.

catalytic to gaseous hydrogen conversion C/G ranged from 54

to 0.24 (Fig. 6b). The transverse plots of H2 mass fraction and

temperature in Fig. 7 (pertaining to the CST case in Fig. 6)

further attested the leakage of hydrogen through the gaseous

flame, as the gradients (dYH2=dy)y¼b were clearly non-zero over

the length xig � x � xTw,max. Moreover, as seen in Fig. 7, the

maximum temperatures in the reactor occurred at the wall

and not in the gas. This was also the case for all examined CST

cases.

The behavior in Figs. 6b and 7 was consistent with

asymptotic studies of gaseous combustion over stagnation

flow catalytic surfaces having deficient reactants with Lewis

numbers less than unity [54]. It was therein shown that a

continuous increase of the strain rate pushed the flame to-

wards the wall, leading to incomplete gaseous conversion of

the fuel (deficient reactant) and consequently to catalytic

conversion of the fuel that leaked through the flame. A further

increase of the strain rate pushed the flame against the stag-

nation surface and eventually extinguished it [54]. In the

present channel-flow geometry, the aerodynamic strain

(characterized by the transverse mass flux towards the

channel walls) was imposed by the incoming mass flux

(wrINUIN) whereas the gaseous reactivity was largely

controlled by the equivalence ratio and the inlet temperature.

Reducing the equivalence ratio to 4 ¼ 0.2, while keeping all

other parameters the same as in Fig. 6, led to a drastic

reduction of the G hydrogen conversion as shown in Fig. 8a, b.

The OH levels in Fig. 8a and the G magnitude in Fig. 8b man-

ifested a hydrogen G conversion that was insignificant when

compared to the corresponding C conversion. The maximum

wall temperature in Fig. 8b was Tw,max ¼ 1431 K, i.e. 460 K

above the adiabatic equilibrium temperature Tad ¼ 971 K.

Although the maximum superadiabaticity, Tw,max e Tad, for

the 4¼ 0.2 case in Fig. 8b was comparable to that of the 4¼ 0.3

case in Fig. 6b (Tw,max e Tad ¼ 484 K), the suppression of

superadiabaticity was much more pronounced in the latter

case due to the appreciable gaseous combustion. This was

x (mm)0 15 30 45 60 75

Tem

pera

ture

(K)

350

590

830

1070

1310

1550

H2

conv

ersi

on (m

ol/m

2 s)

0.0

0.3

0.6

0.9

1.2

1.5

Tw

Tm,gas

G

C

ad=971K

Tw,max=1444K

0 15 30 45 60 75 x (mm)

(c)

(d)

120600

-0.6

0.60.0

y(m

m)

i-CST

OH

0 15 30 45 60 75x (mm)

OH

T

CST (a, b) and the i-CST (c, d) modes; (a, c) 2-D maps of OH

ature (Tw), mean gas temperature (Tm,gas), catalytic (C ) and

d Tad provide the adiabatic equilibrium temperature of the

1738K

0 15 30 45 60 75x (mm)

(a)

(b)

290015000

0 15 30 45 60 75x (mm)

-0.6

0.60.0

y(m

m)

CST

x (mm)0 5 10 15 30 45 60 75

Tem

pera

ture

(K)

650

900

1150

1400

1650

1900

H2

conv

ersi

on (m

ol/m

2 s)

0.0

3.2

6.4

9.6

12.8

16.0Tw

Tm,gas

G

C

ad=1497T T K

Tw,max=1492K

0 15 30 45 60 75 x (mm)

(c)

(d)

360018000

-0.6

0.60.0

y(m

m)

0 15 30 45 60 75x (mm)

i-CST

OH OH

Fig. 9 e Predictions at 4 [ 0.3, UIN [ 10 m/s, TIN [ 650 K for the CST (a, b) and the i-CST (c, d) modes; (a, c) 2-D maps of OH

concentration (in ppm mass), (b, d) axial profiles of wall temperature (Tw), mean gas temperature (Tm,gas), catalytic (C ) and

gas-phase (G) hydrogen conversion. The horizontal lines marked Tad provide the adiabatic equilibrium temperature of the

incoming mixture. For clarity, in (b, d) the first 15 mm are shown in an expanded scale.

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n en e r g y 3 8 ( 2 0 1 3 ) 1 0 6 5 4e1 0 6 7 0 10663

evident from Eq. (20) since LeH2 changed only modestly (0.321

at 4 ¼ 0.2 and 0.336 at 4 ¼ 0.3) while (DT )c was 621 K at 4 ¼ 0.2

and 883 K at 4 ¼ 0.3.

It is thus seen that a promotion of gaseous combustion

could appreciably reduce the attained surface super-

adiabaticity. This is further demonstrated in Fig. 9a, b, per-

taining to a case with 4 ¼ 0.3 and a higher preheat TIN ¼ 650 K.

In this case the gaseousG conversion was substantially higher

than the corresponding one at TIN ¼ 350 K (Fig. 6b). Moreover,

the onset of homogeneous ignition occurred practically at

xig z 0, thus diminishing the extent of pure catalytic conver-

sion. The combined effects of stronger gaseous combustion

and diminishing homogeneous ignition distance led to pro-

portionally lower superadiabaticity. The maximum wall

temperature in Fig. 9b was Tw,max ¼ 1738 K, i.e. only 241 K

above the adiabatic equilibrium temperature Tad ¼ 1497 K.

Characteristically, computations without the inclusion of

gaseous chemistry, yielded a much higher Tw,max ¼ 1878 K.

In summary, two important conclusions could be reached

from Figs. 6e9. The first one is that, contrary to the common

premise that gas-phase combustion could pose a concern for

the catalytic reactor thermal management, gas-phase chem-

istry was actually desirable in lean catalytic combustion of

hydrogen. The second outcome was the enhanced suppres-

sion of the degree of superadiabaticity with increasing

equivalence ratio and mixture preheat. This was highly

desirable, since such operating conditions would have other-

wise led to much higher superadiabaticities that would have

in turn endangered the catalyst and reactor thermal stability.

Finally, as seen in Figs. 6, 8 and 9, the attained temperatures

(wall and mean gas temperatures as well) at the channel exit

were modestly lower than the corresponding Tad (by 25 K in

Fig. 9b and by 6 K in Fig. 8b) due to radiation heat losses to-

wards the colder entry.

The attained reactor temperatures for all examined CST

cases have been summarized in Fig. 10. For the three

investigated equivalence ratios, plots are provided for the

computed maximum wall temperatures Tw,max (symbols) as a

function of the inlet temperature, TIN, at the two investigated

velocities UIN ¼ 10 and 20 m/s. For three selected TIN at 4 ¼ 0.2

and 0.3 and forUIN¼ 10m/s, the attainedwall temperatures in

the presence of only catalytic chemistry are further shown

(filled triangles). Also plotted in Fig. 10 are the corresponding

adiabatic equilibrium temperatures, Tad, the theoretically

maximum wall temperatures, Tw,max-theory, from Eq. (19) and

an upper limit for the tolerable wall temperature

Tw,limit ¼ 1500 K that warranted thermal stability of the FeCr-

alloy structure. This limit temperature obviously depended on

the specific wall material and catalyst formulation. For the

leaner cases at 4 ¼ 0.1 it could be shown that gas-phase

combustion was absent, apart for the highest preheat

TIN ¼ 800 K, whereby the integrated (over-x) G rate reached

25% of the corresponding C rate. The attained maximum

surface temperatures Tw,max for 4 ¼ 0.1 never exceeded the

limiting temperature Tw,limit ¼ 1500 K (Fig. 10a). With

increasing 4 and TIN, the differences between Tw,max-theory and

Tw,max increased (see Fig. 10a, b, c) due to the corresponding

enhancement in gaseous combustion. Moreover, the attained

Tw,max at 4¼ 0.3 were always higher than Tw,limit (Fig. 10c), and

this was of course the case for all 4 > 0.3. At 4 ¼ 0.2 the Tw,limit

was not exceeded only for TIN <w500 K (Fig. 10b). Therefore, the

standard CST mode in Fig. 1a with hydrogen fuel was pri-

marily limited to the sufficiently lean stoichiometries 4 � 0.2.

The results in Fig. 10 showed that the UIN ¼ 20 m/s cases

always yielded higher Tw,max than the corresponding

UIN ¼ 10 m/s cases. An increase of the inlet velocity

augmented the transverse reactant fluxes towards the cata-

lytic wall and this could have potentially increased the fuel

leakage through the gaseous flame (and thus the catalytic

hydrogen conversion) according to the foregoing discussion in

the context of [54]. However, while a modest change of

parameters directly affecting gaseous chemistry (4, TIN)

(a) ϕ = 0.1

600

800

1000

1200

1400

1600

Tw,max-theory

UIN=20m/s

800

1200

1600

2000

Inlet temperature TIN (K)300 400 500 600 700 800

Max

imum

wal

l and

adi

abat

ic e

quilib

rium

tem

pera

ture

s (K

)

1000

1400

1800

2200

2600

Tad

Tw,limit

UIN=10m/s

(b) ϕ = 0.2

Tad

Tw,max-theory Tw,limit

Tw,limit

(c) ϕ = 0.3

Tw,max-theory

Tad

Fig. 11 e Computed maximum wall temperatures at

UIN [ 10 m/s (filled circles) and UIN [ 20 m/s (open circles)

versus inlet temperature (TIN) for the i-CST mode at three

different equivalence ratios: a) 4 [ 0.1, b) 4 [ 0.2 and c)

4 [ 0.3. The adiabatic equilibrium temperatures (Tad), the

maximum theoretical wall temperature (Tw,max-theory) from

Eq. (19) and a limiting wall temperature for reactor thermal

stability Tw,limit [ 1500 K are also plotted.

(a) ϕ = 0.1

600

800

1000

1200

1400

1600

Tw,max-theory

UIN=20m/s

800

1200

1600

2000

Inlet temperature TIN (K)300 400 500 600 700 800

Max

imum

wal

l and

adi

abat

ic e

quilib

rium

tem

pera

ture

s (K

)

1000

1400

1800

2200

2600

Tad

Tw,limit

UIN=10m/s

(b) ϕ = 0.2

Tad

Tw,max-theory

Tw,limit

Tw,limit

(c) ϕ = 0.3

Tw,max-theory

Tad

Fig. 10 e Computed maximum wall temperatures at

UIN [ 10 m/s (filled circles) and UIN [ 20 m/s (open circles)

versus inlet temperature (TIN) for the CST mode at three

different equivalence ratios: a) 4 [ 0.1, b) 4 [ 0.2 and c)

4 [ 0.3. The solid triangles provide the computed

maximum wall temperatures in the presence of only

catalytic reactions at selected cases for 4[ 0.2 and 0.3. The

adiabatic equilibrium temperatures (Tad), the maximum

theoretical wall temperature (Tw,max-theory) from Eq. (19)

and a limiting wall temperature for reactor thermal

stability Tw,limit [ 1500 K are also plotted.

i n t e rn a t i o n a l j o u r n a l o f h y d r o g e n en e r g y 3 8 ( 2 0 1 3 ) 1 0 6 5 4e1 0 6 7 010664

produced appreciablemodifications in the gas-phase/catalytic

hydrogen chemistry coupling, the aerodynamic parameterUIN

required a substantial change to affect this coupling. It could

be shown that the enhancement of Tw,max in Fig. 10 with the

increase of UIN mostly reflected the reduced importance of

heat losses compared to the increased heat generation and

also the increased homogeneous ignition distance xig that in

turn allowed for detrimental pure catalytic conversion

without shielding from gaseous combustion.

The CST mode yielded nearly complete hydrogen conver-

sion at the channel outlet (>99.9% for UIN¼ 10m/s and>99.0%

for UIN ¼ 20 m/s except for the 4 ¼ 0.2, TIN ¼ 300 K case with

95.1% conversion) due to the high molecular diffusivity of

hydrogen. A fractional hydrogen conversion (so as to accom-

modate a post-catalyst flame zone in Fig. 1a) would thus

require shorter channel lengths. Even then, superadiabatic

surface temperatures cannot be mitigated since the peak wall

temperature is always close to the channel entry.

3.3. Inverse catalytically stabilized combustion (i-CST)

i-CST as well as CST results are shown in Figs. 6, 8 and 9, so as

to facilitate comparison of the two fuel-lean modes. For low

inlet temperatures and/or for high inlet velocities, gaseous

combustion in i-CST was stabilized close to the end of the

inert section (x ¼ 15 mm) as shown in Fig. 6c, d. This always

resulted in superadiabatic wall temperatures irrespective of

the relative magnitudes of the C and G hydrogen conversion

rates. As seen in Fig. 6d, the maximum wall temperature was

Tw,max ¼ 1648 K, i.e. 415 K above the adiabatic equilibrium

temperature Tad ¼ 1233 K; this happened despite the strong G

conversion (homogeneous ignition was achieved at

xig ¼ 13.5 mm) that nearly diminished the C conversion. The

attained superadiabaticity was thus not catalytically-induced

(as in the CST case of Fig. 6b) but was a result of the upstream

heat conduction inside the solid wall (also known as heat

recirculation in the solid). While down to the homogeneous

ignition position xig ¼ 13.5 mm no hydrogen was depleted

(owing to the catalytically inert section), the gas-temperature

was significantly higher than TIN ¼ 350 K. At xig ¼ 13.5 mm the

temperature at the gasewall interface was Tw ¼ 1010 K while

the mean gas temperature was Tm,gas ¼ 430 K. The ignition-

x (mm)0 15 30 45 60 75

Wal

l tem

pera

ture

(K)

900

1000

1100

1200

1300

1400

1500

H2

conv

ersi

on ra

te (g

/m2 s

)

0

2

4

6

8

10Deutschmann et al. [42]Park et al. [43]Fridell et al. [44]

Fig. 12 e Predictions for the case in Fig. 8a, using three

different surface reaction mechanisms [42e44]. The wall

temperatures and the H2 catalytic conversion rates are

compared. The wall temperatures predicted with [42] and

[43] practically overlap, while the H2 catalytic conversion

rates of the three surface schemes practically overlap.

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n en e r g y 3 8 ( 2 0 1 3 ) 1 0 6 5 4e1 0 6 7 0 10665

relevant preheat temperature for the gas was in turn closer to

Tw ¼ 1010 K rather than Tm,gas ¼ 430 K since homogeneous

ignition was always achieved at the wall proximity, as evi-

denced by the flame shape in Fig. 6c. Under the operating

conditions of Fig. 6, the advantage of i-CST over CST in terms

of wall superadiabaticity reduction was hence marginal.

For cases where the catalytically-induced super-

adiabaticity was smaller (e.g. lower 4 and hence lower (DT )c in

Eq. (19)) and also where the G conversion of hydrogen was

much weaker than the C conversion, the i-CST mode could

even result in higher superadiabatic temperatures that the

CST mode, as illustrated in Fig. 8. The effect of heat recircu-

lation in i-CST was stronger than the catalytically-induced

effect in CST, leading to a higher wall temperature for the

former (Tw,max ¼ 1444 K, Fig. 8d) compared to the latter

(Tw,max ¼ 1431 K, Fig. 8b). The main advantage of i-CST could

be seen for cases with sufficiently high inlet temperatures TIN

(Fig. 9d). In such cases, the flame practically anchored at xw 0

(see Fig. 9c, d), thus diminishing heat recirculation effects in

the solid and at the same time consuming all hydrogen ho-

mogeneously and thus suppressing the catalytically-induced

superadiabaticity. Fig. 9d indicated a slightly underadiabatic

maximumwall temperature Tw,max¼ 1492 K for the i-CST case

(due to radiation heat losses), which was in turn substantially

lower than the superadiabatic Tw,max¼ 1738 K shown in Fig. 9b

for the corresponding CST case.

Computed temperatures have been summarized in Fig. 11

for all examined i-CST cases. For a given equivalence ratio,

there existed a value of TIN above which the maximum wall

temperature approached the adiabatic equilibrium tempera-

ture Tad, since in these cases flames anchored close to xw0.

This clearly demonstrated that for practical preheats of large

power generation systems (600 K < TIN < 800 K), the i-CST

mode was limited by the Tad value and did not yield super-

adiabatic surface temperatures. The dependence of the wall

temperature on the inlet velocity was stronger in the i-CST

mode when compared to CST (see Figs. 10 and 11), the reason

being that the magnitude of UIN determined how close the

flame anchored to x w0 and hence the degree of suppression

of the heat-recirculation-induced superadiabaticity.

When compared to the CST results in Fig. 10, the i-CST

mode had a wider safe operational envelope (Tw < Tw,limit),

which for UIN ¼ 10 m/s encompassed the entire examined TIN

range at 4 ¼ 0.2 and considerable part of the TIN range at

4 ¼ 0.3. Certainly the design of more complex upstream

gaseous combustion zones (such as porous burners with

increased residence times) could further improve the higher

UIN characteristics in Fig. 11 by reducing heat recirculation in

the solid matrix. Similarly to the CST mode of the foregoing

section, the i-CST mode led to nearly complete hydrogen

conversion at the channel exit (>99.9% for UIN ¼ 10 m/s, and

>99.0% for UIN ¼ 20 m/s except for the 4 ¼ 0.1, TIN ¼ 300 K case

with 98.5% conversion).

Finally, the effect of different reaction mechanisms is dis-

cussed. In an earlier study [22] we evaluated the impact of

different catalytic and gaseous reaction mechanisms on the

hetero-/homogeneous combustion of H2/air over platinum.

For the case in Figs. 8a and 12 provides comparisons of the

wall temperature and the H2 catalytic conversion rate axial

profiles, which were obtained using three different catalytic

reaction mechanisms, the mechanism of Deutschmann [42]

used throughout this work, and the mechanisms from Park

et al. [43] and Fridell et al. [44]; the gas-phase scheme was the

same [47] in all simulations of Fig. 8a. The fist two catalytic

schemes [42,43] gave almost the same result, while the last

one [44] resulted in wall temperatures at most 15 K lower. The

overall good agreement between the three surface mecha-

nisms was largely attributed to the nearly mass-transport-

limited H2 catalytic conversion, which in turn yielded nearly

the same H2 catalytic conversion rates in Fig. 12.

3.4. Fuel rich catalytic combustion

Stable combustion (pure heterogeneous or combined hetero-/

homogeneous) could be achieved for all conditions of Mode B

in Table 3 (i.e. even with the largest examined stoichiometry

in the catalytic channel, 4CAT¼ 5.0).While in the fuel-lean CST

and i-CST modes nearly complete hydrogen conversion was

attained, for the fuel-rich mode there was always uncon-

verted hydrogen (for 4CAT > 1.0) at the channel outlet, which

increased with increasing 4CAT,and required a post-catalyst

flame to complete fuel conversion.

Predictions for a total equivalence ratio 4TOT ¼ 0.2 and

TIN ¼ 300 K are illustrated in Fig. 13 for two fuel-rich stoichi-

ometries in the catalytic channel, 4CAT ¼ 1.0 and 5.0. The inlet

velocities in the catalytic channel and the bypass air (provided

in the caption of Fig. 13) resulted in the same mass flux as the

CST and i-CST modes at 4 ¼ 0.2, TIN ¼ 300 K and UIN ¼ 10 m/s.

The 4CAT ¼ 1.0 case was an extreme limit for the fuel-rich

combustion mode and was only included for comparison

purposes. Fig. 13(a2, b) gives the catalyst wall temperature, Tw,

the mean gas temperature in the catalytic channel, Tm,gas, the

mean temperature in the bypass air, Tm,bypass, the adiabatic

equilibrium temperature based on the total stoichiometry

4TOT ¼ 0.2, Tad, and the catalytic (C ) and gas-phase (G)

hydrogen conversions. Homogeneous ignition was achieved

only for the 4CAT ¼ 1.0 case, as shown from the 2-D OH dis-

tribution in Fig. 13(a1) and the G conversion in Fig. 13(a2),

while for 4CAT � 2.0 only catalytic conversion was present. For

x (mm)0 15 30 45 60 75

Tem

pera

ture

(K)

500

770

1040

1310

1580

1850

H2

conv

ersi

on (m

ol/m

2 s)

0

5

10

15

20

25

Tw

Tm,gas

G

C

ad=1812T KTw,max=1414 KTcenter

Tm,bypassTm,bypass2

0 15 30 45 60 75 x (mm)

(a)

(b) 5802900

-0.6

0.60.0

y(m

m)

ϕCAT

= 3.0

0 15 30 45 60 75x (mm)

(b)

OH

Fig. 14 e Predictions at 4TOT [ 0.5, TIN [ 500 K for the fuel-

rich mode (Fig. 2b), 4CAT [ 3.0, UIN,CAT [ 5.28 m/s,

UIN,bypass [ 5.84 m/s. In (a) 2-D maps of OH concentration

(in ppm mass) are provided. In (b) axial profiles of wall

temperature (Tw), mean gas temperature (Tm,gas),

temperature at the channel center (Tcenter), bypass air mean

temperature (Tm,bypass), catalytic (C ) and gas-phase (G)

hydrogen conversion are plotted. Tm,bypass2 provides the

mean bypass air temperature computed with a 2-D model.

The horizontal lines marked Tad provide the adiabatic

equilibrium temperature of the total mixture.

x (mm)0 15 30 45 60 75

Tem

pera

ture

(K)

300

475

650

825

1000

H2

conv

ersi

on (m

ol/m

2 s)

0.0

0.2

0.4

0.6

0.8

TwTm,gas

G

C

ad =926T KTw,max=764 K

Tm,bypass

Tem

pera

ture

(K)

300

750

1200

1650

2100

H2

conv

ersi

on (m

ol/m

2 s)

0

10

20

30

40

Tw

Tm,gas

G

C

ad =926 K

Tw,max=1645 K

Tcenter

Tm,bypass

(b) ϕCAT

=5.0

xig

0 15 30 45 60 75 x (mm)

880044000

-0.6

0.60.0

y(m

m)

OH

(a2)

(a1) ϕCAT

=1.0

T

Fig. 13 e Predictions at 4TOT [ 0.2, TIN [ 300 K for the fuel-

rich mode (Fig. 2b); (a1, a2) 4CAT [ 1.0, UIN,CAT [ 4.35 m/s,

and UIN,bypass [ 6.13 m/s, (b) 4CAT [ 5.0, UIN,CAT [ 2.19 m/s

and UIN,bypass [ 8.47 m/s. In (a1) 2-D maps of OH

concentration (in ppm mass) are provided. In (a2, b) axial

profiles of wall temperature (Tw), mean gas temperature

(Tm,gas), catalytic (C ) and gas-phase (G) hydrogen

conversion are plotted. In (a2) Tcenter is the temperature

along the channel center ( y [ 0). The horizontal lines

marked Tad provide the adiabatic equilibrium temperature

of the total mixture.

i n t e rn a t i o n a l j o u r n a l o f h y d r o g e n en e r g y 3 8 ( 2 0 1 3 ) 1 0 6 5 4e1 0 6 7 010666

the case in Fig. 13(a2) with combined heterogeneous and ho-

mogeneous combustion, the gas-temperature at the channel

center ( y ¼ 0), Tcenter, is also shown. Similar graphs are pro-

vided in Fig. 14 for the highest 4TOT ¼ 0.5, TIN ¼ 500 K, and for

4CAT ¼ 3.0 (homogeneous combustion was achieved only for

4CAT � 3.0). Moreover, mean bypass air temperatures in Fig. 14

are given for two different simulations; the 1-D lumped heat

transfer model (Tm,bypass) discussed in Section 2.1 and a full 2-

D model for the bypass air channel (Tm,bypass2). The latter was

obtained by simulating a bypass channel subject to the

following wall boundary conditions: the lower wall tempera-

ture Tw was the solution of the standard model in Section 2.1

while the upper wall was adiabatic. The agreement between

the two simulations was quite good, justifying the simpler 1-D

model for the non-reacting bypass channel.

Thewall temperatures could bemaintained below the limit

wall temperature Tw,limit¼ 1500 K, at least for sufficiently large

4CAT as shown in Figs. 13 and 14. This was due to the cooling of

the catalytic channel by the bypass air and also due to the

catalytic process itself that led to underadiabatic surface

temperatures for the LeO2 > 1 deficient fuel, as discussed in

Section 3.1 and the accompanying Fig. 3. Despite these factors

that resulted in reduced wall temperatures, flames were

established, at least for high TIN; computations illustrated that

for 4TOT � 0.3 and TIN � 700 K homogeneous combustion was

always present for all examined 4CAT. It is noted that our

recent fundamental kinetic studies [48] have shown that,

contrary to pure gaseous combustion whereby the reactivity

of lean hydrogen/air mixtures was higher than the corre-

sponding reactivity of rich mixtures, in catalytic combustion

homogeneous ignition could be achieved easier for rich

hydrogenmixtures. Thiswas an outcome of the lowmolecular

diffusivity of oxygen that led to reduced catalytic depletion of

the deficient O2 reactant over the gaseous induction zone; on

the other hand, in lean catalytic combustion the large

hydrogenmolecular diffusivity led to a fast catalytic depletion

of the deficient H2, thus depriving it from the gaseous reaction

pathway. Finally, in Figs. 13(a2) and 14(b) it was evident that

the catalytic conversion rate C became slightly negative

(indicating catalytic production of hydrogen) at the flame

location. Reaction flux analysis indicated that this was a result

of the large concentration of H radicals produced homoge-

neously inside the flame, which in turn adsorbed on the

catalyst (reaction S4 in Table 1) and then shifted the balance of

hydrogen adsorption/desorption reactions (S3 and S12,

respectively) towards desorption.

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n en e r g y 3 8 ( 2 0 1 3 ) 1 0 6 5 4e1 0 6 7 0 10667

In rich catalytic combustion closed flames were formed

(see Fig. 14), in contrast to the fuel-lean CST mode whereby

the low Lewis number of the deficient hydrogen reactant led

to open flames with two separate branches (Figs. 6 and 8).

Moreover, whenever a flame was established in the reactor,

the peak temperatures were located in the gas-phase, and

specifically at the plane of symmetry ( y¼ 0, see Tcenter profiles

in Fig. 13(a2) and Fig. 14(b)) rather than at the channel walls,

thus greatly facilitating the reactor thermalmanagement. The

reactor wall temperatures for all examined cases in Mode B

have been summarized in Fig. 15 for UIN ¼ 10 m/s; results for a

total UIN ¼ 20 m/s are also provided in Fig. 15 at 4CAT ¼ 3.0.

Comparing Fig. 15 to Figs. 10 and 11, it was evident that the

fuel-rich mode had a much wider range of applicability

(a) ϕTOT

= 0.2

600

1000

1400

1800

600

1000

1400

1800

2200

Inlet temperature TIN (K)300 400 500 600 700 800

Max

imum

wal

l and

adi

abat

ic e

quilib

rium

tem

pera

ture

s (K

)

600

1000

1400

1800

2200

600

1000

1400

1800

2200

Tw,limit

Tw,limit

(b) ϕTOT

= 0.3

(d) ϕTOT

= 0.5

(c) ϕTOT

= 0.4

Tad

Tad

Tad

ϕCAT =5.0

ϕCAT=1.0

ϕCAT=5.0

ϕCAT=1.0

ϕCAT =5.0

ϕCAT =5.0

ϕCAT=1.0

Tw,limit

ϕCAT=1.0Tad

Fig. 15 e Computed maximum wall temperatures at

UIN [ 10 m/s (open symbols) versus inlet temperature (TIN)

for the fuel-rich mode at four different total equivalence

ratios: a) 4TOT [ 0.2, b) 4TOT [ 0.3, c) 4TOT [ 0.4 and c)

4TOT [ 0.5. The equivalence ratios in the catalytic channel

are 4CAT [ 1.0 (open circles), 4CAT [ 2.0 (lower open

triangles), 4CAT [ 3.0 (open squares), 4CAT [ 4.0 (open

diamonds), and 4CAT [ 5.0 (upper open triangles). The

filled squares provide the maximum wall temperatures for

4CAT [ 3.0 and UIN [ 20 m/s. The adiabatic equilibrium

temperatures (Tad based on 4TOT) and a limiting wall

temperature for reactor thermal stability Tw,limit [ 1500 K

are also plotted.

(Tw,max < Tw,limit). Specifically, the fuel-rich mode could

accommodate gas-turbine relevant operating conditions (total

equivalence ratios 4TOT ¼ 0.5 and inlet temperatures

TIN > 650 K), at least for 4CAT > 3.0.

3.5. Mode comparison

Operating regimes for the three investigated modes are

summarized in Fig. 16, which was constructed using results

from Figs. 10, 11 and 15. The lines in Fig. 16 demarcated the

operating envelopes in the total equivalence ratio versus inlet

temperature parametric space: areas below the corresponding

lines signified safe reactor operation (Tw,max < 1500 K). For

UIN ¼ 10 m/s (Fig. 16a) CST exhibited the narrower operating

envelope; nonetheless, it offered the simplest reactor design

solution for applications with 4TOT < 0.2. The operating en-

velope in i-CST was broader than that of CST; moreover, the

highest acceptable 4TOT in i-CST exceeded 0.3 in the inter-

mediate temperature range 450 K < TIN < 650 K.

For the fuel-richmode, results are shown in Fig. 16a for two

catalytic equivalence ratios 4CAT ¼ 2 and 3. For 4CAT ¼ 3,

operation with 4TOT > 0.4 for TIN < 750 K became possible. In

addition, all of the 4TOT - TIN parametric space in Fig. 16 was

accessible for 4CAT ¼ 4 and 5. Thus, for hydrogen fuel, this

modewas suitable for gas turbines (4TOT up to 0.5 and TIN up to

800 K) and justified its application despite the more complex

design requirements. It is finally noted that although opera-

tion at 4CAT ¼ 5 was feasible (experiments in a subscale gas-

turbine reactor with hydrogen fuel have even shown safe

and stable combustion up to 4CAT ¼ 6 [55]), it was desirable to

limit 4CAT in the fuel richmode below about 4. The reasonwas

0.1

0.2

0.3

0.4

0.5

Inlet temperature TIN (K)300 400 500 600 700 800

Tota

l equ

ival

ance

ratio

ϕTO

T

0.1

0.2

0.3

0.4

0.5

UIN=10m/s

CST

i-CST

fuel-rich (ϕCAT =3)

fuel-rich(ϕ

CAT =2)

UIN =20m/s

(a)

(b)

CST

fuel-rich

( ϕCAT =3)

i-CST

Fig. 16 e Operating regimes for the three combustion

modes in Fig. 2, in terms of total equivalence ratio (4TOT)

and mixture preheat (TIN). The areas below the lines define

safe operating envelopes for each mode: a) UIN [ 10 m/s

and b) UIN [ 20 m/s.

i n t e rn a t i o n a l j o u r n a l o f h y d r o g e n en e r g y 3 8 ( 2 0 1 3 ) 1 0 6 5 4e1 0 6 7 010668

that the resulting NOx emissions at the end of the post-

catalyst gaseous combustion zone (Fig. 2b) dropped appre-

ciablywith rising hydrogen conversion inside the catalyst [55].

Finally, the UIN ¼ 20m/s results (Fig. 16b) have qualitatively

the same trends as those at UIN ¼ 10 m/s. For UIN ¼ 20 m/s,

however, the i-CST mode exhibited narrower envelope

compared to that for UIN ¼ 10 m/s, the reason being the

sensitivity of the homogeneous ignition distance on UIN as

discussed in Section 3.3.

4. Conclusions

The catalytic combustion of fuel-lean hydrogen/air mixtures

(overall equivalence ratios, 4TOT, ranging from 0.1 to 0.5) was

investigated numerically in Pt-coated plane channels for three

different combustion modes: the conventional catalytically

stabilized thermal combustion (CST), the inverse catalytically

stabilized thermal combustion (i-CST), and the fuel-rich cat-

alytic/gaseous-lean combustion. In the lean CST mode, the

less-than-unity Lewis number of hydrogen led to super-

adiabatic surface temperatures that endangered the reactor

integrity and the catalyst thermal stability. Open flames could

be established in CST for 4TOT � 0.2, which extended along the

catalytic walls and moderated the surface temperature

superadiabaticity by shielding the catalyst from the hydrogen

fuel. Even though the gaseous-combustion-induced suppres-

sion of the above-mentioned superadiabaticity was propor-

tionally stronger at higher 4TOT and mixture preheats (TIN),

reactor and catalyst thermal stability considerations limited

the CST mode to 4TOT � 0.2.

The maximum attained wall temperatures in the i-CST

mode strongly depended on the location of homogeneous

ignition. Long ignition distances facilitated further preheating

of the incomingmixture via heat recirculation inside the solid

walls and led to superadiabatic surface temperatures, even in

cases where homogeneous combustion dominated over cat-

alytic combustion. The i-CST mode had advantages over CST

in terms of reactor thermal management particularly at

higher TIN and extended the range of applicable 4TOT up to 0.3.

In the fuel-rich catalytic combustion mode, the combina-

tion of bypass cooling air and the high Lewis number of the

deficient oxygen reactant led to surface temperatures below

the adiabatic value and allowed safe operation for 4TOT up to

0.5 and TIN up to 800 K, at least for sufficiently high fuel-rich

catalytic stoichiometries in the catalytic channel (4CAT > 3).

Homogeneous ignition could be achieved in the catalytic

channel at low 4CAT and high TIN resulting in the formation of

closed flames, which however did not pose thermal manage-

ment concerns. The operational range of the fuel-rich mode

made it suitable for gas-turbines fueled with either hydrogen

or syngases.

Acknowledgments

Support was provided by the European Union project Hydro-

geneIGCC and the Swiss Center of Energy andMobility (CCEM)

project Carbon Management.

Nomenclature

b channel half-height, m

cp,k specific heat at constant pressure of k-th gaseous

species, J/(kg K)

Dkm mixture-average species diffusion coefficientm,m2/s

DTk species thermal diffusion coefficient, m2/s

h, hok total enthalpy, chemical enthalpy of k-th gaseous

species, J/kg

Kg total number of gaseous species

L catalytic channel length, m

Li catalytically inert channel length, m

Le Lewis number (thermal over mass diffusivity)_m mass flow rate per unit channel width, kg/(s$m)

Ms total number of surface species

p pressure, Pa

R universal gas constant, J/(mol$K)_sk catalytic production rate of k-th species, mol/(m2$s)

T, To temperature and reference temperature, K

u, UIN streamwise velocity and inlet streamwise velocity,

m/s

v transverse (-y) velocity component, m/s

V.

k species diffusion velocity vector, m/s

Wk, W species molecular weight and average molecular

weight, kg/mol

Yk mass fraction of k-th gaseous species x, y streamwise

and transverse coordinates

Greek symbols

G surface site density, mol/m2

lg, ls thermal conductivity of gas and solid, J/(s$m$K)

m dynamic viscosity, kg/(m$s)

r density, kg/m3

sm surface species site occupancy

4 hydrogen-to-oxygen equivalence ratio

_uk homogeneous molar production rate of k-th species,

mol/(m3s)

Subscripts

IN inlet

ig homogeneous ignition

g gas

s surface

k, m indices for gas-phase and surface species

W wall

r e f e r e n c e s

[1] Dunn S. Hydrogen futures: toward a sustainable energysystem. Int J Hydrogen Energy 2002;27:235e64.

[2] Gnanapragasam NV, Reddy BV, Rosen MA. Hydrogenproduction from coal gasification for effective downstreamCO2 capture. Int J Hydrogen Energy 2010;35:4933e43.

[3] Castaldi MJ, Dooher JP. Investigation into a catalyticallycontrolled reaction gasifier (CCRG) for coal to hydrogen. Int JHydrogen Energy 2007;32:4170e9.

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n en e r g y 3 8 ( 2 0 1 3 ) 1 0 6 5 4e1 0 6 7 0 10669

[4] Tock L, Marechal F. H2 processes with CO2 mitigation:thermo-economic modeling and process integration. Int JHydrogen Energy 2012;37:11785e95.

[5] Cormos CC. Evaluation of power generation schemes basedon hydrogen-fuelled combined cycle with carbon captureand storage (CCS). Int J Hydrogen Energy 2011;36:3726e38.

[6] Winkler D, Mueller P, Reimer S, Griffin T, Burdet A,Mantzaras J, et al. Improvement of gas turbine combustionreactivity under flue gas recirculation condition with in situhydrogen addition. ASME 2009. GT2009e59182.

[7] Schneider A, Mantzaras J, Bombach R, Schenker S, Tylli N,Jansohn P. Laser induced fluorescence of formaldehyde andRamanmeasurements ofmajor species during partial catalyticoxidation of methane with large H2O and CO2 dilution atpressures up to 10 bar. Proc Combust Inst 2007;31:1973e81.

[8] Griffin T, Winkler D, Wolf M, Appel C, Mantzaras J. Stagedcatalytic combustionmethod for the advanced zero emissionsgas turbine power plant. ASME 2004. GT2004e54101.

[9] Norton DG, Wetzel ED, Vlachos DG. Fabrication of single-channel catalytic microburners: effect of confinement on theoxidation of hydrogen/air mixtures. Ind Eng Chem Res2004;43:4833e40.

[10] Karagiannidis S, Mantzaras J. Numerical investigation on thestart-up of methane-fueled catalytic microreactors. CombustFlame 2010;157:1400e13.

[11] Karagiannidis S, Mantzaras J. Numerical investigation on thehydrogen-assisted start-up of methane-fueled, catalyticmicroreactors. Flow Turbul Combust 2012;89:215e30.

[12] Nabavi M, Yan Y, Santis AJ, Poulikakos D. A spatially-resolved temperature-dependent model for butanereforming over rhodium catalyst. Int J Hydrogen Energ2012;37:9067e75.

[13] Lopez E, Irigoyen A, Trifonov T, Rodriguez A, Llorca J. Amillion-channel reformer on a fingertip: moving down thescale in hydrogen production. Int J Hydrogen Energy2010;35:3472e9.

[14] Newson E, Truong TB. Low-temperature catalytic partialoxidation of hydrocarbons (C-1-C-10) for hydrogenproduction. Int J Hydrogen Energy 2003;28:1379e86.

[15] Izquierdo U, Barrio VL, Cambra JF, Requies J, Guemez MB,Arias PL, et al. Hydrogen production from methane andnatural gas steam reforming in conventional andmicroreactor reaction systems. Int J Hydrogen Energy2012;37:7026e33.

[16] Thormann J, Maier L, Pfeifer P, Kunz U, Deutschmann O,Schubert K. Steam reforming of hexadecane over a Rh/CeO2

catalyst in microchannels: experimental and numericalinvestigation. Int J Hydrogen Energy 2009;34:5108e20.

[17] Smith LL, Karim H, Castaldi MJ, Etemad S, Pfefferle WC,Khanna VK, et al. Rich-catalytic lean-burn combustion forlow-single digit NOx gas turbines. J Eng Gas Turbines Power2005;127:27e35.

[18] Beebe KW, Cairns KD, Pareek VK, Nickolas SG, Schlatter JC,Tsuchiya T. Development of catalytic combustiontechnology for single-digit emissions from industrial gasturbines. Catal Today 2000;59:95e115.

[19] Schlegel A, Benz P, Griffin T, WeisensteinW, Bockhorn H.Catalytic stabilization of lean premixed combustion: methodfor improvingNOxemissions.CombustFlame1996;105:332e40.

[20] Carroni R, Griffin T, Mantzaras J, Reinke M. High-pressureexperiments and modeling of methane/air catalyticcombustion for power generation applications. Catal Today2003;83:157e70.

[21] Pfefferle WC, Pfefferle LD. Catalytically stabilizedcombustion. Prog Energy Combust Sci 1986;12:25e41.

[22] Appel C, Mantzaras J, Schaeren R, Bombach R, Inauen A,Kaeppeli B, et al. An experimental and numericalinvestigation of homogeneous ignition in catalytically

stabilized combustion of hydrogen/air mixtures overplatinum. Combust Flame 2002;128:340e68.

[23] Mantzaras J. Catalytic combustion of syngas. Combust SciTechnol 2008;180:1137e68.

[24] Smith LL, Karim H, Castaldi MJ, Etemad S, Pfefferle WC. Rich-catalytic lean-burn combustion for fuel-flexible operationwith ultra-low emissions. Catal Today 2006;117:438e46.

[25] Schneider A, Mantzaras J, Jansohn P. Experimental andnumerical investigation of the catalytic partial oxidation ofCH4/O2 mixtures diluted with H2O and CO2 in a short contacttime reactor. Chem Eng Sci 2006;61:4634e46.

[26] Schneider A, Mantzaras J, Eriksson S. Ignition and extinctionin catalytic partial oxidation of methane-oxygen mixtureswith large H2O and CO2 dilution. Combust Sci Technol2008;180:89e126.

[27] Zhu H, Kee RJ, Engel JR, Wickham DT. Catalytic partialoxidation of methane using RhSr- and Ni-substitutedhexaaluminates. Proc Combust Inst 2007;31:1965e72.

[28] AlavandiSK,EtemadS,BairdBD.LowsingledigitNOxemissionscatalytic combustor for advanced hydrogen turbines for cleancoal power systems. ASME 2012. GT2012e68128.

[29] Bui PA, Vlachos DG, Westmoreland PR. Homogeneousignition of hydrogen/air mixtures over platinum. ProcCombust Inst 1996;26:1763e70.

[30] Mantzaras J, Bombach R, Schaeren R. Hetero-/homogeneouscombustion of hydrogen/air mixtures over platinum atpressures up to 10 bar. Proc Combust Inst 2009;32:1937e45.

[31] Ghermay Y, Mantzaras J, Bombach R. Effects of hydrogenpreconversion on the homogeneous ignition of fuel-lean H2/O2/N2/CO2 mixtures over platinum at moderate pressures.Combust Flame 2010;157:1942e58.

[32] Ghermay Y, Mantzaras J, Bombach R, Boulouchos K.Homogeneous combustion of fuel lean H2/O2/N2 mixturesover platinum at elevated pressures and preheats. CombustFlame 2011;158:1491e506.

[33] Brady K, Sung CJ, T’ien J. Ignition propensity of hydrogen/airmixtures impinging on a platinum stagnation surface. Int JHydrogen Energy 2010;35:11412e23.

[34] Maestri M, Beretta A, Faravelli T, Groppi G, Tronconi E. Roleof gas-phase chemistry in the rich combustion of H2 and COover a Rh/Al2O3 catalyst in annular reactor. Chem Eng Sci2007;62:4992e7.

[35] Maestri M, Beretta A, Faravelli T, Groppi G, Tronconi E,Vlachos DG. Two-dimensional detailed modeling of fuel-richH2 combustion over Rh/Al2O3 catalyst. Chem Eng Sci2008;63:2657e69.

[36] Karagiannidis S, Mantzaras J, Jackson G, Boulouchos K.Hetero-/homogeneous combustion and stability maps inmethane-fueled microreactors. Proc Combust Inst2007;31:3309e17.

[37] Mantzaras J, Benz P. An asymptotic and numericalinvestigation of homogeneous ignition in catalyticallystabilized channel flow combustion. Combust Flame1999;119:455e72.

[38] Mantzaras J, Appel C. Effects of finite rate heterogeneouskinetics on homogeneous ignition in catalytically stabilizedchannel-flow combustion. Combust Flame 2002;130:336e51.

[39] Kee RJ, Dixon-Lewis G, Warnatz J, Coltrin ME, Miller JA. AFortran computer code package for the evaluation of gas-phase multicomponent transport properties. Report No.SAND86e8246 1996.

[40] Siegel R, Howell JR. Thermal radiation heat transfer. NewYork: Hemisphere; 1981. p. 271.

[41] Shah RK, London AL. Laminar flow forced convection inDucts. New York: Academic Press; 1978.

[42] Deutschmann O, Maier LI, Riedel U, Stroemman AH,Dibble RW. Hydrogen assisted catalytic combustion ofmethane on platinum. Catal Today 2000;59:141e50.

i n t e rn a t i o n a l j o u r n a l o f h y d r o g e n en e r g y 3 8 ( 2 0 1 3 ) 1 0 6 5 4e1 0 6 7 010670

[43] Park YK, Aghalayam P, Vlachos DG. A generalized approachfor predicting coverage-dependent reaction parameters ofcomplex surface reactions: application to H2 oxidation overplatinum. J Phys Chem A 1999;103:8101e7.

[44] Fridell E, Rosen A, Kasemo B. A laser-induced fluorescencestudy of OH desorption from Pt in H2O/O2 and H2O/H2

mixtures. Lagmuir 1994;10:699e708.[45] Trevino C. Catalytic ignition of very lean mixtures of

hydrogen. Int J Hydrogen Energy 2011;36:8610e8.[46] Warnatz J, Allendorf MD, Kee RJ, Coltrin ME. A model of

elementary chemistry and fluid mechanics in thecombustion of hydrogen on platinum surfaces. CombustFlame 1994;96:393e406.

[47] Li J, Zhao Z, Kazakov A, Dryer FL. An updated comprehensivekinetic model of hydrogen combustion. Int J Chem Kinet2004;36:566e75.

[48] Schultze M, Mantzaras J, Bombach R, Boulouchos K. Anexperimental and numerical investigation of the hetero-/homogeneous combustion of fuel-rich hydrogen/airmixtures over platinum. Proc Combust Inst 2013;34:2269e77.

[49] Kee RJ, Rupley FM, Miller JA. Chemkin II: a Fortran chemicalkinetics package for the analysis of gas-phase chemicalkinetics. Report No. SAND89e8009B 1996.

[50] Coltrin ME, Kee RJ, Rupley FM. Surface Chemkin: a Fortranpackage for analyzing heterogeneous chemical kinetics atthe solid surface-gas phase interface. Report No.SAND90e8003C 1996.

[51] Ghermay Y, Mantzaras J, Bombach R. Experimental andnumerical investigation of hetero-/homogeneouscombustion of CO/H2/O2/N2 mixtures over platinum atpressures up to 5 bar. Proc Combust Inst 2011;33:1827e35.

[52] Reinke M, Mantzaras J, Bombach R, Schenker S, Inauen A.Gas phase chemistry in catalytic combustion of methane/airmixtures over platinum at pressures of 1 bar to 16 bar.Combust Flame 2005;141:448e68.

[53] KaragiannidisS,Mantzaras J,BoulouchosK.Stabilityofhetero-/homogeneous combustion in propane andmethane fueledcatalytic microreactors: channel confinement andmoleculartransport effects. Proc Combust Inst 2011;33:3241e9.

[54] Law CK, Sivashinsky GI. Catalytic extension of extinctionlimits of stretched premixed flames. Combust Sci Technol1982;29:277e86.

[55] Bolanos F, Winkler D, Piringer F, Griffin T, Bombach R,Mantzaras J. Study of a rich/lean staged combustion conceptfor hydrogen at gas turbine relevant conditions. ASME TurboExpo 2013. GT2013e94420, June 3e7, San Antonio Texas, USA.