Proceedings

of the
Combustion
Institute
Proceedings of the Combustion Institute 30 (2005) 2929–2937
www.elsevier.com/locate/proci

Thermophysical characteristics of shear-coaxial

LOX–H2 flames at supercritical pressure
Joseph C. Oefelein*
Combustion Research Facility, Sandia National Laboratories, Livermore, CA 94551-9051, USA

Abstract

The thermophysical characteristics of a liquid-oxygen–hydrogen flame at supercritical pressure are

investigated, with emphasis placed on the near-field region just downstream of a shear-coaxial injector ele-
ment. Results are presented from a series of hierarchical high-fidelity simulations for a condition where
oxygen is injected in a cryogenic state, at a subcritical temperature and supercritical pressure, and hydrogen
is injected in a supercritical state. This condition has significant technical relevance in liquid-rocket engines,
but is not well understood and has not been well quantified to date. For this situation, a diffusion domi-
nated mode of combustion occurs in the presence of exceedingly large thermophysical property gradients.
The flame anchors itself in the interfacial region of high shear that exists between the liquid-oxygen core
and the annular hydrogen jet, and intensifies the respective property gradients such that they approach
the behavior of a contact discontinuity. Significant real-gas effects and transport anomalies coexist locally
in colder regions of the flow, with ideal gas and transport characteristics occurring within the flame zone.
The current analysis provides one of the first quantitative characterizations of the flame structure and asso-
ciated property variations in liquid-oxygen–hydrogen flames at supercritical pressure.

2004 The Combustion Institute. Published by Elsevier Inc. All rights reserved.

Keywords: Turbulent combustion; Cryogenic propellants; Supercritical pressure; Flame structure; Liquid-rocket
propulsion

1. Introduction including the Space Shuttle Main Engine (SSME)

1.1. Objectives
Use of cryogenic propellants in liquid-rocket
engines at chamber pressures that exceed the ther-
modynamic critical point of the propellants has
been well established as a combination that pro-
vides high efficiency and performance for a variety
of launch vehicle applications. In particular, li-
quid-oxygen–hydrogen (LOX–H2) engines have
been developed for a variety of launch systems,
*
Fax: +1 925 294 2595.
E-mail address: oefelei@sandia.gov.
and the Ariane 5 Vulcain engine. The chamber
pressure in these engines exceeds 100 atm, which
is significantly higher than the thermodynamic
critical pressure of both propellants.
Although LOX–H2 rocket engines have been
operated relatively safely for the past several
years, the processes that control combustion are
still not well understood. This paper provides
one of the first quantitative characterizations of
the flame structure and associated property varia-
tions in LOX–H2 flames at supercritical pressure.
Results from a series of hierarchical simulations
are presented for a condition where oxygen is in-
jected in a cryogenic state, at a subcritical temper-
ature and supercritical pressure, and hydrogen is

1540-7489/$ - see front matter
2004 The Combustion Institute. Published by Elsevier Inc. All rights reserved.
doi:10.1016/j.proci.2004.08.212
2930 J.C. Oefelein / Proceedings of the Combustion Institute 30 (2005) 2929–2937
injected in a supercritical state. Emphasis is placed
on shear-coaxial injection processes in the labora-
tory-scale uni-element rocket configuration devel-
oped at the German Aerospace Center (DLR) in
Lampoldshausen.
1.2. Phenomenological trends
Shear-coaxial injection processes in liquid-
rocket engines exhibit two distinct modes of com-
bustion [1–5]. At subcritical pressures, the liquid
jets atomize. Dynamic forces and surface tension
promote the formation of a heterogeneous spray,
and lifted spray flames form in a manner consis-
tent with the modes of combustion exhibited by
local drop clusters. When chamber pressures ap-
proach or exceed the critical pressure of a partic-
ular propellant, however, injected liquid jets
undergo a transcritical change of state as interfa-
cial fluid temperatures rise above the critical tem-
perature of the local mixture. For this situation,
diminished inter-molecular forces promote diffu-
sion dominated mixing processes prior to atom-
ization. Respective jets vaporize in the presence
of exceedingly large thermophysical gradients,
and well-mixed diffusion flames evolve as a conse-
quence that are anchored by small but intensive
recirculation zones just downstream of the LOX-
post.
Flow visualization studies conducted by Mayer
and Tamura [1] have illustrated the trends de-
scribed above for the case of a LOX–H2 shear-co-
axial injector element. The two extremes are
shown in Fig. 1. When LOX is injected at low-
subcritical pressures (Fig. 1A), atomization occurs
forming a distinct spray. Ligaments are detached
from the jet surface forming drops and drop clus-
Fig. 1. Reacting shear-coaxial LOX–H2 injector oper-
ating at (A) 1.5 MPa (15 atm) and (B) 4.5 MPa (44 atm).
From Mayer and Tamura [1]. Used with permission.
ters that subsequently break up and vaporize. As
the chamber pressure approaches the thermody-
namic critical pressure of the LOX (Fig. 1B), how-
ever, the number of drops present diminishes, and
the injected LOX jet begins to exhibit a pure dif-
fusion mechanism. This mechanism was observed
to dominate at a pressure of 4.5 MPa, which is
slightly below the thermodynamic critical pressure
of oxygen, and significantly above that of hydro-
gen. Studies carried out in the ONERA ‘‘MAS-
COTTE’’ facility [6,7] have produced similar
results. Flame attachment occurs instantaneously
after ignition in the small but intensive recircula-
tion zone that forms just downstream of the annu-
lar LOX-post. Gurliat et al. [8] have recently
verified this trend over a broad range of
conditions.
2. Approach
Results were obtained using the theoretical–
numerical framework developed by Oefelein
[9,10]. This framework has been designed for
application of both the direct numerical simula-
tion (DNS) technique, and the large eddy simula-
tion (LES) technique and solves the fully coupled
compressible conservation equations of mass,
momentum, total-energy, and species in full
geometries. A full multicomponent formulation
has been employed for the current investigation.
The system is assumed to be compressible, chem-
ically reacting, and composed of N species.
The baseline system of equations are cast in
dimensionless form using a reference length-scale
dref, flow speed Uref, and fluid state characterized
by a reference density qref, sound speed cref, con-
stant pressure specific heat C pref , and dynamic vis-
cosity lref. Using these quantities, a reference
Mach and Reynolds number are defined as
M = Uref/cref and Re = qrefUrefdref/lref. With these
definitions, the instantaneous conservation equa-
tions of mass, momentum, total-energy, and
chemical species can be written in conservative
form as follows:
oq
þ r
ðquÞ ¼ 0; ð1Þ
ot


o p
ðquÞ þ r
qu  u þ 2 I ¼ r
s; ð2Þ
ot M
where


l 2  
s¼  ðr
uÞI þ ru þ ruT
Re 3
represents the viscous stress tensor;
o
ðqet Þ þ r
½ðqet þ pÞu
ot
 
¼ r
qe þ r
M 2 ðs
uÞ ; ð3Þ
 J.C. Oefelein / Proceedings of the Combustion Institute 30 (2005) 2929–2937 2931

where employed was developed by Oefelein [10] and is

designed to account for thermodynamic non-ide-
M 2 X
N
p
et ¼ e þ u
u; e¼ hi Y i  ; alities and transport anomalies over a wide range
2 i¼1
q of pressures and temperatures. Here, the extended
corresponding states model [14,15] is used with a
and
Z Z cubic equation of state to evaluate the p–v–T
p T
behavior of the inherent dense multicomponent
hi ¼ h0fi þ C pi ðT ; pÞ dT dp
p0 T0
mixtures. A summary of the cubic equations of
state and recommended constants is given by Reid
are the total internal energy, internal energy, and et al. [16, Chapter 3]. Experience [17,18] has
partial enthalpy of the ith species, and shown that both the Soave–Redlich–Kwong
o (SRK) and Peng–Robinson (PR) equations, when
ðqY i Þ þ r
ðqY i uÞ used in conjunction with the corresponding states
ot
principle, can give accurate results over the range
¼ r
qi þ x_ i ; i ¼ 1; . . . ; N  1; ð4Þ of pressures, temperatures, and mixture states of
where x_ i represents the instantaneous rate of pro- interest in this study. SRK coefficients are fit to
duction of the ith chemical species due to reac- vapor pressure data, and thus more suitable when
tions. Heat release due to chemical reactions in the reduced temperature is less than one. PR coef-
the total-energy equation is accounted for in the ficients, on the other hand, are more suitable when
description of the partial specific enthalpies, hi, the reduced temperature is greater than one. Here,
by including the enthalpy of formation, h0fi , in the PR equation of state was used exclusively
the definition. since the flow involves heat release.
The dimensionless multicomponent energy and Relevant thermodynamic properties are ob-
mass diffusion fluxes are given by the expressions tained in two steps. First, respective component
properties are combined at a fixed temperature
lC p XN
using the corresponding states methodology to
qe ¼  rT þ hi qi obtain the mixture state at a given reference
PrRe i¼1
pressure. A pressure correction is then applied
XN X N
X j DTi using departure functions of the form given by
þ Ru T ðVi  Vj Þ ð5Þ Reid et al. [16, Chapter 5]. These functions are
i¼1 j¼1
W i Dij
derived from MaxwellÕs relations (see VanWylen
and and Sonntag [19, Chapter 10], for example) and
make full use of the real-mixture p–v–T path
qi ¼ qY i Vi dependencies dictated by the equation of state.
XN Molecular transport properties are evaluated in
Y iY j
¼ DTi rðln T Þ þ q Dij dj ; ð6Þ an analogous manner. The viscosity and thermal
j¼1
X iX j conductivity are obtained using the extended
corresponding states methodologies developed
where Ru ð¼ Ru =C p ref Þ is the dimensionless form by Ely and Hanley [20,21]. Mass and thermal
of the universal gas constant, Vi is the dimension- diffusion coefficients are obtained using the
less diffusion velocity of the ith species, and DTi , methodologies outlined by Bird et al. [22],
Dij, and Dij are the dimensionless thermal, mass, Hirschfelder et al. [11], and Takahashi [23].
and binary mass diffusion coefficients, respec- The governing system is discretized on a stag-
tively. These quantities are defined in a manner gered grid in generalized curvilinear coordinates.
consistent with Hirschfelder et al. [11]. Terms Xi This formulation provides non-dissipative spec-
and Yi represent respective mole and mass frac- trally clean damping characteristics and discrete
tions. The term conservation of mass, momentum and total-en-
"   # ergy. Integration is performed using a unique
XN
Xi oli
di ¼ rX j dual-time multistage scheme with a generalized
j¼1
Ru T oX j p;T ;X k ;...;ðk6¼i;jÞ all-Mach-number preconditioning methodology
j6¼i that optimally treats convective, diffusive, geomet-
þ ZðX i  Y i Þrðln pÞ ð7Þ ric, and source term anomalies in a unified man-
ner. The implicit formulation is A-stable, which
is the dimensionless form of the mechanical driv- allows one to set the physical timestep based so-
ing force, where li represents the chemical poten- lely on accuracy considerations. The scheme
tial of the ith species and p is pressure. Radiation accommodates any arbitrary equation of state,
fluxes and body forces are currently neglected. handles thermodynamic non-idealities and trans-
Eqs. (1)–(4) were solved with full finite-rate port anomalies over a wide range of conditions,
hydrogen–oxygen kinetics using the mechanism and provides full thermophysical coupling (com-
developed by Westbrook and Dryer [12] and pressible and incompressible) between processes.
Yetter et al. [13]. The property evaluation scheme The algorithm has been optimized to provide
2932 J.C. Oefelein / Proceedings of the Combustion Institute 30 (2005) 2929–2937
excellent parallel scalability attributes using a dis-
tributed multiblock domain decomposition with
generalized connectivity.
3. Results and discussion
The theoretical–numerical framework has been
used to perform a series of hierarchical high-fidel-
ity simulations of LOX–H2 flames at supercritical
pressure. Both LES and DNS techniques have
been applied to gain a further understanding of
the flow dynamics, flame structure, and associated
property variations inherent to shear-coaxial
injection processes. In this paper, primary empha-
sis has been placed on a series of fundamental
DNS calculations performed on a sector of the
German Aerospace Center (DLR) windowed
combustor operated in the high-pressure test facil-
ity (P8) in Lampoldshausen.
A schematic of the computational domain is
given in Fig. 2. The geometry has been non-
dimensionalized using a reference length-scale of
d = 4 mm, which is the physical diameter of the
LOX jet. The combustion chamber is cylindrical,
with a uni-element LOX–H2 shear-coaxial injector
mounted on the centerline of the face plate and a
converging nozzle at the exit. The chamber and
nozzle assembly is 110 dimensionless units long,
the chamber diameter is 12.5 units, and the nozzle
exit diameter is 4 units. The dimensionless inner
and outer diameters of the annular H2 jet are
1.150 and 1.625, respectively. The LOX-post is
0.075 units thick and is not tapered or recessed.
The injector flow conditions and fluid proper-
ties are listed in Table 1. Calculations were per-
formed using a reference chamber pressure of
10.1 MPa (100 atm). The oxygen and hydrogen
streams are injected into the system at 100 and
150 K, respectively. For this set of conditions,
the hydrogen stream enters in a gaseous supercrit-
ical state. The LOX stream, however, enters in a
transcritical state that is subcritical with respect
to temperature but supercritical with respect to
pressure. The corresponding densities are 1119
and 15.4 kg/m3, respectively. The mean injection
Fig. 2. Schematic of the computational domain.
Table 1
Injector flow, fluid, and critical properties
O2 H2
Pressure (MPa) 10.1
Temperature (K) 100 150
Density (kg/m3) 1119 15.40
Viscosity (m2/s) 1.431 · 107 3.590 · 107
Velocity (m/s) 30.0 125
Flowrate (kg/s) 421.9 · 103 27.09 · 103
Mach Number 0.0394 0.120
Reynolds Number 8.386 · 105 5.571 · 105
Critical properties of O2: pc = 5.04 MPa (49.7 atm) and
Tc = 155 K.
Critical properties of H2: pc = 1.30 MPa (12.8 atm) and
Tc = 33.2 K.
velocities are 30 and 125 m/s, respectively. The
corresponding mass flowrates are 421.9 · 103
and 27.09 · 103 kg/s, and the corresponding
Mach numbers are 0.0394 and 0.120.
The flow Reynolds numbers in the oxygen and
hydrogen ducts are 8.386 · 105 and 5.571 · 105.
These quantities are an order of magnitude larger
than that exhibited in typical laboratory-scale de-
vices at atmospheric pressures. To account for the
inherently smaller and wider range of scales effi-
ciently (and feasibly), a quasi-2d DNS was per-
formed on a 30 cylindrical sector with periodic
conditions specified in the azimuthal direction. A
grid with 12-million hexahedral cells was con-
structed in generalized coordinates. The bulk of
the cells were clustered in the near-field region just
downstream of the injector such that the esti-
mated local Kolmogorov time and length-scales
were resolved with an average of 20 timesteps
and 16 cells per respective interval. The system
was integrated in time using a dimensionless time-
step of 0.001 based on the characteristic time of
the incoming LOX jet, dref/Uref = 133 ls.
All wall surfaces were assumed to be hydrauli-
cally smooth. Fully developed turbulent velocity
profiles were generated inside the injector ducts
by recycling the unsteady velocity field from radial
planes at an axial distance of 2 dimensionless units
upstream of the face-plate. The stripped fields
were corrected to maintain the mass flow rates
 J.C. Oefelein / Proceedings of the Combustion Institute 30 (2005) 2929–2937
specified in Table 1 and then imposed at the injec-
tor inlet 8 units upstream using a non-reflecting
pressure condition. The walls inside the injector
ducts were assumed to be adiabatic. The walls in
the combustion chamber were assumed to be
300 K. A zero-gradient pressure condition was
applied at the nozzle exit. The current investiga-
tion is focused exclusively on the near-field region
just downstream of the injector. This region is de-
picted in Fig. 2 by the 15 · 4.875 rectangle adja-
cent to the injector face-plate. Figure 3 shows
contours of the normalized turbulent kinetic en-
ergy field in this region to highlight the complex
flow structure. This result was taken from a com-
panion LES calculation using a full three-dimen-
sional grid with 8.8-million cells.
Analysis of the subsequent DNS results is fo-
cused on the region of high shear near the injector
Fig. 3. Normalized turbulent kinetic energy field.
Fig. 4. Instantaneous (A) vorticity, (B) density, and (C)
temperature fields.
2933
exit. Results have been non-dimensionalized using
a reference state corresponding to the LOX jet in-
flow condition. Figure 4 shows the instantaneous
character of the dimensionless vorticity, density,
and temperature fields just downstream of the
injector tip. The plot of vorticity includes OH mass
fraction contours to mark the flame location. Com-
bustion occurs at near stoichiometric conditions
and produces a wake that effectively separates the
hydrogen and oxygen streams as the flow evolves
downstream. Because of the liquid-like characteris-
tics of the LOX, an extremely large density gradient
exists within the shearlayer. The close proximity of
the dense LOX core on one side of the flame and ex-
tremely low density fluid on the other side promotes
the formation of small-scale vortical structures
within the shearlayer that evolve dynamically in a
manner analogous to that produced by a backward
facing step. These structures emanate from the
upper corner of the LOX-post, are initially much
smaller than the LOX-post thickness, and coalesce
within the resultant downstream wake.
The degree to which the properties vary across
the flame and shearlayer in the near-field region
are shown in Figs. 5–9. Symbols mark the location
of respective grid nodes. Figure 5 shows the radial
Fig. 5. Radial variation in the Kolmogorov time and
length scales at an axial location of 0.03 units.
Fig. 6. Radial variation of temperature and density at
an axial location of 0.03 units.
2934 J.C. Oefelein / Proceedings of the Combustion Institute 30 (2005) 2929–2937

Fig. 7. Radial variation of H2, O2, OH, and H2O mass

fractions at an axial location of 0.03 units.

Fig. 8. Radial variation of Z and derivative relations Zp

and ZT at an axial location of 0.03 units.

variation in the dimensionless Kolmogorov time

and length scales at an axial location of 0.03 units.
The Kolmogorov time-scales approach values
as small as 102 in the flame and 103 on the hydro-
gen side of the shearlayer. The Kolmogorov length-
scales approach values as small as 104 and 105,
respectively. This figure also shows how the grid
spacing distribution compares to the local Kol-
mogorov length-scales across the shearlayer to ver-
ify the degree of resolution achieved in the
calculation. The grid spacing in the axial direction Fig. 9. Radial variation of kinematic viscosity, Prandtl,
provides a similar degree of resolution. Schmidt, and Lewis numbers at an axial location of 0.03
Figures 6 and 7 show the corresponding radial units.
variations in temperature, density, and the H2, O2,
OH, and H2O mass fractions. The change in den- files of the compressibility factor Z and the
sity across the flame is of the order of 1000 to 1 derivative relations

  
and occurs over an interval that is only a fraction p oZ
of the LOX-post thickness. An equally distinct Zp ¼ 1  ð8Þ
Z op T
change in temperature due to combustion and sca-
lar-mixing is also evident. Though continuous, and
"  #
gradients in the flame zone approach the behavior T oZ
of a contact discontinuity. Depletion of all of the ZT ¼ 1 þ ð9Þ
Z oT p
oxygen in the flame, the flame structure, and the
fuel-rich characteristics on the hydrogen side of are shown. The compressibility factor indicates
the flame are also apparent. the degree to which the local mixture state ap-
Figure 8 provides a further characterization of proaches that of an ideal gas. Eqs. (8) and (9)
the thermodynamic properties across the shear- are directly related to the isothermal compressibil-
layer at the same axial location. Here, radial pro- ity and the coefficient of thermal expansion and
 J.C. Oefelein / Proceedings of the Combustion Institute 30 (2005) 2929–2937
indicate the degree to which these fundamental
forms of compressibility deviate from that of an
ideal gas. The isothermal compressibility charac-
terizes the change in volume that results from a
change in pressure while the temperature remains
constant. The coefficient of thermal expansion
characterizes the change in volume that results
from a change in temperature while the pressure
remains constant. Figure 8 shows that real-gas
and liquid effects exist locally in colder regions of
the flow on either side of the shearlayer, and ideal
gas conditions exist within the hot fuel-rich region
and flame zone. This latter observation can be di-
rectly attributed to the inherent diminished effect
of pressure with increasing temperature.
Figure 9 shows the radial variation in the
dimensionless kinematic viscosity and the Fickian
(mixture-averaged) values of the Prandtl, Schmidt,
and Lewis numbers, respectively. Relative to the
value in the LOX core, the kinematic viscosity in-
creases by a factor of approximately 100 in the
flame zone and varies by a factor of 10–100 in the
shearlayer. The Prandtl, Schmidt, and Lewis num-
bers all decrease from values greater than 1 on the
LOX side of the flame to values less than 1 on the
hydrogen side. Relative magnitudes on the LOX
side suggest that the diffusion of momentum domi-
nates over energy and mass, and that the diffusion
of energy dominates over mass. Mass diffusion
rates into the flame on the LOX side are thus rate-
limiting. Values associated with the fuel-rich prod-
ucts, on the other hand, suggest that the opposite
trend will be present. On this side of the flame, the
relative magnitudes of the Prandtl, Schmidt, and
Lewis numbers suggest that the diffusion of energy
will dominate over momentum, and that mass diffu-
sion dominates over both momentum and energy.
The results presented above suggest that the
smoothness of the LOX jet can be attributed to
high axial inertia coupled with rapid radial heat-
ing and vaporization within the shearlayer. The
initial vortical expansion of the low-density
hydrogen induces strong recirculating backflow
in the vicinity of the LOX-post. The net effect of
the local thermophysical interactions is to produce
a strong counter-recirculating zone of hot fuel-
rich products in the vicinity of the LOX-post
tip. Inside this recirculation zone is an unsteady
Fig. 10. Flame-holding mechanism.
2935
stagnation region that provides the primary
flame-holding mechanism. The fluid dynamic as-
pects of this mechanism are shown in Fig. 10.
Flame development in the stagnation region is
dominated by diffusion and enhanced by convec-
tive energy transport from the backflow of fuel-
rich combustion products. The backflow of hot
products sustains ignition and promotes vaporiza-
tion and mixing of the dense LOX core.
4. Conclusions
The thermophysical characteristics of a liquid-
oxygen–hydrogen flame at supercritical pressure
have been investigated using the direct numerical
simulation technique. Calculations were performed
on a sector of a computational domain designed to
match the German Aerospace Center (DLR) win-
dowed combustor operated in the high-pressure
test facility (P8) in Lampoldshausen. The combus-
tion chamber is cylindrical, with a uni-element
shear-coaxial injector mounted on the centerline
of the face-plate and a converging nozzle at the exit.
Oxygen was injected in a cryogenic state at a sub-
critical temperature and supercritical pressure.
The hydrogen was injected in a pure supercritical
state. Emphasis was placed on the near-field region
just downstream of the injector element. In this re-
gion, a diffusion dominated mode of combustion
exists in the presence of exceedingly large thermo-
physical property gradients. The flame anchors it-
self within the interfacial region of high shear that
exists between the liquid-oxygen core and the annu-
lar hydrogen jet. Resultant variations in the ther-
mophysical properties across the flame intensify
respective gradients such that they approach the
behavior of a contact discontinuity.
Results provide detailed quantitative character-
izations of the flame structure, associated property
variations, and the dominating effects of the dense
liquid-oxygen core on the near-field shearlayer
dynamics. Diminished mass diffusion rates inside
the dense core create a fuel-rich condition within
the near-field shearlayer region. Combustion oc-
curs at near stoichiometric conditions and produces
a wake that effectively separates the hydrogen and
oxygen streams as the flow evolves downstream.
The extremely low values of density in the hot reac-
tion zone cause a considerable reduction of aerody-
namic interactions between the liquid-oxygen jet
and the hydrogen stream. Significant real-gas and
liquid behavior is observed to coexist locally in
colder regions of the flow, with ideal gas and trans-
port characteristics existing within the reaction and
product zones. The latter can be directly attributed
to the inherent diminished effect of pressure with
increasing temperature.
The coupled effect of local thermophysical
interactions is to produce a strong counter-
recirculating zone of hot fuel-rich products in
2936 J.C. Oefelein / Proceedings of the Combustion Institute 30 (2005) 2929–2937
the vicinity of the injector tip. This recirculation
zone provides the primary flame-holding mecha-
nism. The flame-holding mechanism is dominated
by an unsteady stagnation region. Flame develop-
ment within the stagnation region is dominated by
diffusion and enhanced by convective energy
transport from the backflow of fuel-rich combus-
tion products. The backflow of hot products both
sustains the ignition process and promotes vapor-
ization of the dense liquid-oxygen core.
Acknowledgment
This work was funded by the NASA Marshall
Space Flight Center under Contract No. H-
33477D. The support provided is gratefully
acknowledged.
References
[1] W. Mayer, H. Tamura, J. Propul. Power 12 (6)
(1996) 1137–1147.
[2] W. Mayer, A. Schik, B. Vieille, C. Chaveau, I.
Gökalp, D. Talley, R. Woodward, J. Propul. Power
14 (5) (1998) 835–842.
[3] W. Mayer, A. Schik, M. Schäffler, H. Tamura, J.
Propul. Power 16 (5) (2000) 823–828.
[4] G. Herding, R. Snyder, C. Rolon, S. Candel, J.
Propul. Power 13 (2) (1998) 146–151.
[5] S. Candel, G. Herding, R. Snyder, P. Scouflaire, C.
Rolon, L. Vingert, J. Propul. Power 14 (5) (1998)
826–834.
[6] D. Kendrick, G. Herding, P. Scouflaire, J.C. Rolon,
S. Candel, Combust. Flame 118 (1999) 327–339.
[7] M. Juniper, A. Tripathi, P. Scouflaire, J.C. Rolon,
S. Candel, Proc. Combust. Inst. 28 (1) (2000)
1103–1109.
Comment
Sebastien Candel, EM2C Lab, CNRS, Ecole Centale
Paris, France. Your calculations are quite close to some
of the data gathered in cryogenic flame experiments car-
ried out in recent years [1] but there is also one differ-
ence. Your flame is attached at the point where the
oxygen leaves the injector. In our experiments, the flame
has an edge, and it is anchored at a small but finite dis-
tance from the injection lip.
You also mentioned that some extinction takes place
at a distance from the injection plane. We found that
oxygen/hydrogen flames require excessively large strain
rates to be extinguished. Is there a possibility that this
extinction could be related to an insufficient resolution
of the flame reaction layer?
Reference
[1] M. Juniper, A. Tripathi, P. Scouflaire, J.-C. Rolon,
S. Candel, Proc. Comb. Inst. 28 (2000) 1103–1109.
[8] O. Gurliat, V. Schmidt, O.J. Haidn, M. Oschwald,
Aerospace Sci. Technol. 7 (2003) 517–531.
[9] J.C. Oefelein, General numerical framework
for reacting multiphase flow with complex
thermochemistry, thermodynamics and trans-
port, copyright 1992–2004 All Rights Reserved
(2004).
[10] J.C. Oefelein, General package for evaluation of
multicomponent real-gas and liquid mixture states
at all pressures, copyright 1992–2004 All Rights
Reserved (2004).
[11] J.O. Hirschfelder, C.F. Curtiss, R.B. Bird, Molec-
ular Theory of Gases and Liquids, second ed., Wiley,
New York, 1964.
[12] C.K. Westbrook, F.L. Dryer, Prog. Energy Com-
bust. Sci. 10 (1) (1984) 1–57.
[13] R.A. Yetter, F.L. Dryer, H. Rabitz, Combust. Sci.
Technol. 79 (1991) 97–128.
[14] T.W. Leland, P.S. Chappelear, Ind. Eng. Chem.
Fundam. 60 (7) (1968) 15–43.
[15] J.S. Rowlinson, I.D. Watson, Chem. Eng. Sci. 24 (8)
(1969) 1565–1574.
[16] R.C. Reid, J.M. Prausnitz, B.E. Polling, The Prop-
erties of Liquids and Gases, fourth ed., McGraw-
Hill, New York, 1987.
[17] J.C. Oefelein, Simulation and Analysis of Turbulent
Multiphase Combustion Processes at High Pressures,
Ph.D. thesis, The Pennsylvania State University,
University Park, Pennsylvania, 1997.
[18] J.C. Oefelein, V. Yang, J. Propul. Power 14 (5)
(1998) 843–857.
[19] G.J. VanWylen, R.E. Sonntag, Fundamentals of
Classical Thermodynamics, third ed., Wiley, New
York, 1986.
[20] J.F. Ely, H.J.M. Hanley, Ind. Eng. Chem. Fundam.
20 (4) (1981) 323–332.
[21] J.F. Ely, H.J.M. Hanley, Ind. Eng. Chem. Fundam.
22 (1) (1981) 90–97.
[22] R.B. Bird, W.E. Stewart, E.N. Lightfoot, Transport
Phenomena, Wiley, New York, 1960.
[23] S. Takahashi, J. Chem. Eng. Jpn. 7 (6) (1974)
417–420.
Reply. Differences between the current simulation
(see Fig. 4) and data gathered from recent cryogenic
flame experiments can be directly attributed to the man-
ner in which the wall boundary conditions are specified.
This is an extremely important and valid observation.
Results from the simulation are based on the assumption
that the injector and faceplate walls are adiabatic. This
assumption produces a much higher wall temperature
in the vicinity of the flame than would be observed in
the actual experiment and does not accommodate
quenching in the vicinity of the wall. In reality, the flux
distribution across the walls is quite complex, particu-
larly in the vicinity of the LOX-post tip. Enhanced
heat-transfer within the walls will tend to both smooth
and reduce the wall temperature distribution to values
much closer to the propellant injection temperatures of
100 and 150 K (see Table 1). Detailed treatment of wall
effects is clearly an important factor that must eventually
be incorporated into these types of simulations.
 J.C. Oefelein / Proceedings of the Combustion Institute 30 (2005) 2929–2937 2937

Subsequent analysis of the coupled flow-flame interac-
tions has suggested that the vortical structures emanating
from the hydrogen side of the LOX-post can produce a rel-
atively strong dilatation field. This field coalesces progres-
sively within the LOX-post wake and subsequently
interacts with the flame a short distance downstream. In
regions where the dilatation field compresses the flame
the reaction-rate associated with the production of OH
radicals was observed to drop significantly by a factor of
1–2 orders of magnitude. Within this region, the OH pro-
duction rate profiles are resolved with approximately 12
cells in each spatial direction and 8–16 time-steps locally
as the drop in production rate occurs. This is a distinctly
interesting observation that requires further investigation.