AIAA JOURNAL

Vol. 44, No. 4, April 2006

Large-Eddy Simulation and Acoustic Analysis
of a Swirled Staged Turbulent Combustor
Charles E. Martin,

Laurent Benoit,

and Yannick Sommerer

Centre Europ een de Recherche et de Formation Avanc ee en Calcul Scientique, 31057 Toulouse, France
Franck Nicoud

Universit e de Montpellier II, 34095 Montpellier Cedex 5, France

and
Thierry Poinsot

Institut de M ecanique de Fluides de Toulouse, 31400 Toulouse, France

The analysis of self-excited combustion instabilies encountered in a laboratory-scale, swirl-stabilized combustion
system is presented. The instability is successfully captured by reactive large-eddy simulation (LES) and analyzed
by using a global acoustic energy equation. This energy equation shows how the source term due to combustion
(equivalent to the Rayleigh criterion) is balanced by the acoustic uxes at the boundaries when reaching the limit
cycle. Additionally, an Helmholtz-equation solver including ameacoustics interaction modeling is used to predict
the stability characteristics of the system. Feeding the ame-transfer function from the LES into this solver allows
to predict an amplication rate for each mode. The unstable mode encountered in the LES compares well with
the mode of the highest amplication factor in the Helmholtz-equation solver, in terms of mode shape as well as in
frequency.
Nomenclature
[A] = square matrix of size N
c = sound velocity, m/s
D
k
= kth species diffusion coefcient, m
2
/s
E = efciency function
E
a
= activation energy, cal/mol
E
1
= instantaneous global acoustic energy term, J
e
1
= acoustic energy, J/m
3
F = ame thickening factor
F
1
= instantaneous global acoustic uxes, W
f = frequency, Hz
i = square root of 1
N = number of nodes of the grid
n = magnitude of the ame transfer function, Pa/m
n = outward normalized normal vector
[

P] = column vector of size N associated to an eigenmode
p = pressure, Pa
R = perfect gas constant, cal/mol K
S
k
c
= kth species Schmidt number
S
1
= instantaneous global Rayleigh term, W
s
0
L
= laminar ame speed, m/s
s
T
= turbulent ame speed, m/s
T = temperature, K
t = time, s
u = velocity vector, m/s
Received 19 November 2004; revision received 21 April 2005; accepted
for publication 31 July 2005. Copyright c 2005 by the American Institute of
Aeronautics and Astronautics, Inc. All rights reserved. Copies of this paper
may be made for personal or internal use, on condition that the copier pay
the $10.00 per-copy fee to the Copyright Clearance Center, Inc., 222 Rose-
wood Drive, Danvers, MA 01923; include the code 0001-1452/06 $10.00 in
correspondence with the CCC.

Ph.D. Student, Computational Fluid Dynamics Team, 42 Avenue G.

Coriolis.

Research Engineer, Computational Fluid Dynamics Team, 42 Avenue G.

Coriolis.

Professor, Math ematique; also at Institut de Mod elisation et de

Math ematiques de Montpellier, Centre National de la Recherche Scien-
tique, Place Bataillon.

Research Director, all ee du Professeur Camille Soula; also at Ecoule-

ments et Combustion, INP de Toulouse, Centre National de la Recherche
Scientique. Associate Fellow AIAA.
Y
k
= kth species mass fraction
Z = local reduced acoustic impedance
= polytropic coefcient

0
L
= ame thermal thickness, m
= cinematic viscosity, Pa/s
= mass density, kg/m
3
= time delay of the n model, s
= phase of the ame transfer function
= pulsation, rad/s

k
= kth species reaction rate, mol/m
3
s

T
= unsteady heat release, J/m
3
Subscripts
L = laminar
ref = related to the reference point of the n model
0 = steady part
1 = uctuating part
Superscripts
th = thickened quantity

= Fourier transformed
Introduction
C
OMBUSTION oscillations are frequently encountered dur-
ing the development of many combustion chambers for gas
turbines.
14
These oscillations cannot be predicted at the design
stage, and correcting actions can be extremely costly at later stages.
Testing burners in simplied combustion chambers is a common
method to verify their stability but is also an ambiguous approach
because most experimentalists knowthat a given burner can produce
unstable combustion in one chamber and not in another. Methods
providingstabilityanalysis before anytests are, therefore, requested.
Large-eddy simulations (LES) are an obvious choice for such
studies: They are powerful tools to study the dynamics of turbu-
lent ames. (See recent works on turbulent combustion.
4,5
) Multiple
recent papers have demonstrated the power of these methods.
612
However, an important limitation of LES is its cost: The intrinsic
nature of LES (full three-dimensional resolution of the unsteady
NavierStokes equations) makes it very expensive, even on todays
computers. Moreover, even when it is conrmed that a combustor is
unstable, LESdoes not indicate whyandhowtocontrol it. Therefore,
741
742 MARTIN ET AL.
tools are needed to analyze LES results but also to provide capac-
ities for optimization and control of thermoacoustic oscillations in
chambers.
A proper framework to analyze combustion stability is the wave
equation in a reacting ow.
4
Such an equation is complex to derive
because most assumptions used in classical acoustics must be revis-
ited in a multispecies, nonisothermal, reacting gas. For low Mach
numbers, an approximate equation controlling the propagation of
pressure perturbations in a reacting gas is

0
c
2
0

_
1

0
p
1
_

2
t
2
p
1
= ( 1)

T1
t

0
c
2
0
u
1
: u
1
(1)
where the subscript 0 refers to mean quantities and the subscript 1 to
small perturbations. Here
T1
is the local unsteady heat release, and
c
0
and
0
are the soundspeedandthe density, respectively, whichcan
change locally because of changes in temperature and composition
due to chemical reactions. These reactions are also the source of the
additional right-hand-side (RHS) source term ( 1)
T1
/t that
is responsible for combustion noise and instabilities. This equation
does not assume a constant polytropic coefcient and, thus, differs
slightly fromthe one derived in a previous work (Eq. 1.1 in Ref. 13).
This approach is better suited to reacting ows where can change
by 30% from fresh to burnt gases. Equation (1) is difcult to use
directly in practice, and multiple methods have been proposed to
solve it.
3,1417
This paper presents a method where Eq. (1) is used
together with LES.
First an acoustic solver based on a Helmholtz equation is devel-
oped to provide all acoustic modes of a combustion chamber. In this
approach, Eq. (1) is solved in the frequency domain by assuming
sinusoidal oscillations. This solver uses information given by the
LES on the mean temperature eld and the ame transfer function.
Second, a new analysis tool to analyze the budget of acoustic
energy in a reacting ow is described. This integral form of Eq. (1)
is a generalization of the well-known Rayleigh criterion
18
(also see
Ref. 4), which allows an evaluation of all terms of the acoustic
energy equation in the LES.
The rst objective of the present work is to couple these three
tools (LES, Helmholtz solver, and acoustic energy budget) and show
how they can be combined to understand combustion instabilities.
This exercice will be performed on a staged swirled combustion
chamber installed at Ecole Centrale de Paris. In this device, the outlet
boundary condition will be changed in the LES from nonreecting
to perfectly reecting (pressure node) to demonstrate the effect of
this condition on the burner unsteady activity and prove that the three
tools used throughout the paper provide reasonable explanations for
this phenomenon.
The presentation starts with a description of the acoustic energy
equation. The LEStool characteristics are recalled before presenting
the Helmholtz tool. The conguration is then described before the
presentation of the results. Stable and unstable regimes evidenced
by LES are discussed. In this last case, a scenario where the com-
bustion instability grows, reaches a limit cycle, and then decays is
studied. This control of the instability is obtained by changing the
outlet boundary condition, and the budget of acoustic energy during
the whole evolution is used to analyze the instability, the mecha-
nisms controlling its limit-cycle amplitude, and its decay. Finally,
the Helmholtz solver results are presented: The ame transfer func-
tion measurement methodology is described and applied to obtain
the frequency as well as the growth rate of the combustor eigen-
modes. It is then veried that the most unstable mode matches the
LES observations.
Acoustic Energy Equation
The total acoustic energy equation is an integral formof the wave
equation (1), which is quite useful to understand basic mechanisms
of combustion instabilites. This equation cannot be used to predict
unstable modes like the Helmholtz solver, but is a powerful method
to analyze the results of an LES as done here. The conservation
equation for the acoustic energy e
1
=
1
2

0
u
2
1
+
1
2
p
2
1
/(
0
c
2
0
) can be
written
4
e
1
t
= s
1
( p
1
u
1
), s
1
=
( 1)
p
0
p
1

T1
(2)
If integrated over the whole volume V of the combustor bounded
by the surface A, it yields
d
dt
_
V
e
1
dV =
_
V
s
1
dV
_
A
p
1
u
1
ndA (3a)
or
d
dt
E
1
= S
1
F
1
(3b)
where n is the surface normal vector. This surface consists of walls
or of inlet/outlet sections.
In Eq. (3), all terms are time dependent. The RHS source term
S
1
corresponds to the Rayleigh criterion
18
: It measures the corre-
lation between unsteady pressure p
1
and unsteady heat release
T1
averaged over the whole chamber. It can act as a source or a sink
term for the acoustic energy. The other RHS termF
1
is less studied
because it is impossible to measure experimentally. It is an acoustic
ux integrated on all of the boundaries. Walls have zero contribu-
tion in this term because the velocity perturbations u
1
n vanish on
walls. However, F
1
may be large on inlets and outlets where it is
usually a loss term. Equation (2) is, therefore, a generalization of the
Rayleigh criterion: The total acoustic energy in the chamber E
1
will
growif the coustic gain termS
1
is larger than the acoustic losses F
1
.
The magnitudes and relative importance of the two terms S
1
and F
1
are controversial issues in the eld of combustion instabilities. For
example, one important question is to knowwhether acoustic losses
are important in the determination of limit cycles. For these limit
cycles, the acoustic energy E
1
must remain constant over a period
of oscillations, and Eq. (2) shows that such a cycle can be reached
for two situations.
1) The limit cycle may be combustion controlled: If the acoustic
losses are small (F
1
=0), the pressure and heat release signals may
adjust togive S
1
0. The limit cycle is reachedwhenthis phase shift
leads to a zero Rayleigh termS
1
as observed in certain experiments.
Physically, this is often obtained when the heat release oscillations
saturate (because the minimum reaction rate reaches zero at some
instant of the cycle) or when the phase between pressure and heat
release changes so that combustion itself controls the limit cycle
amplitude.
2) The limit cycle may be acoustically controlled: The source
termS
1
may be large (pressure and heat relase oscillating in phase)
but the acoustic losses F
1
are too large and compensate S
1
. In this
case, the nal amplitude of oscillation is controlled by the acoustic
impedances of outlets and inlets.
Clearly, these two solutions lead to very different approaches of
combustion instabilities: If the limit cycle is combustion controlled,
the acoustic behavior of inlets and outlets has a limited effect on the
stability; if it is acousticallycontrolled, acoustic impedances of inlets
and outlets become essential elements of any method (experimental
or numerical). In the present study, the LESresults are postprocessed
to measure all terms of Eq. (2) and determine whether the unstable
mode is combustion or acoustically controlled.
LES for Reacting Flows in Complex Geometries
Numerical Methods for Compressible Reacting LES
Most academic LES are limited to fairly simple geometries for
obvious reasons of cost andcomplexityreduction. Inmanycases, ex-
periments are designed using simple two-dimensional shapes
6,19,20
or axisymmetrical congurations
21,22
and simple regimes (low-
speed ows, fully premixed or fully nonpremixed ames) to allow
research to focus on the physics of the LES (subgrid scale models,
ame/turbulence interaction model) and, more generally, to demon-
strate the validity of the LES concept in academic cases. This ap-
proach is clearly adequate in terms of modeling development, but it
can also be misleading in various aspects when it comes to dealing
MARTIN ET AL. 743
with complex ames in complex geometries, especially in real gas
turbines for which specic problems arise:
1) Real geometries cannot be meshed easily and rapidly with
structured or block-structured meshes: Until now, most LES of re-
acting ows have been performed in combustion chambers where
structured meshes were sufcient to describe the geometry. This is
no longer the case in gas turbines, and this brings additional dif-
culties. Indeed, on structured meshes, building high-order spatial
schemes (typically fourth to sixth order in space) is easy and pro-
vides very precise numerical methods.
2325
For complex geometries
such structured meshes must be replaced by unstructured grids, on
which constructing high-order schemes is a more difcult task.
2) Unstructured meshes also raise a variety of new problems in
terms of subgrid-scale ltering: Dening lter sizes on a highly
anisotropic irregular grid is another open research issue.
2629
Many
LES models, developed and tuned on regular hexahedral grids, may
perform poorly on the low-quality unstructured grids required to
mesh real combustion chambers. For example, the ltered structure
model
24
is difcult to extend to unstructured grids.
3) LES validation is often performed in laboratory low-speed un-
conned ames, in which acoustics do not play a role and the Mach
number remains small so that acoustics and compressibility effects
can be omitted from the equations.
10,21
In most real ames (for ex-
ample in gas turbines), the Mach number can reach high values and
acoustics are important so that taking compressibility effects into
account becomes mandatory. This leads to a signicantly heavier
computational task: Because acoustic waves propagate faster than
the ow, the time step becomes smaller and the boundary conditions
must handle acoustic wave reections.
4
Being able to preserve com-
putational speed on a large number of processors then also becomes
an issue simply to obtain a result in a nite time.
4) At the present time, it is impossible to perform a true LES
everywhere in the ow and it will remain so for a long time. For ex-
ample, the ow between vanes in swirled burners, inside the ducts
feeding dilution jets, or through multiperforated plates would re-
quire too many grid points. Compromises must be sought to offer
(at least) robustness in places where the grid is not sufcient to
resolve the unsteady ow.
In the present work, the full compressible NavierStokes equa-
tions are solved on hybrid (structured and unstructured) grids in a
code called AVBP. Subgrid stresses are described by the wall adapt-
ing local eddy viscosity model.
30
The ame/turbulence interaction
is modeled by the thickened ame (TF) model.
6,31
The numerical
scheme is explicit in time and provides third-order spatial and third-
order time accuracy.
31
TF Model and Chemical Scheme
For this study, the standard TF model
31
is used: In this model, pre-
exponential constants and transport coefcients are both modied to
offer thicker reaction zones that can be resolved on LESmeshes. The
fundamental property justifying this approach has been put forward
by Butler and ORourke
32
by considering the balance equation for
the k-species mass fraction Y
k
in a one-dimensional ame of thermal
thickness
0
L
and speed s
0
L
:
Y
k
t
+
u
i
Y
k
x
i
=

x
i
_
D
k
Y
k
x
i
_
+
k
(Y
j
, T) (4)
Modifying this equation to have
Y
th
k
t
+
u
i
Y
th
k
x
i
=

x
i
_
FD
k
Y
th
k
x
i
_
+
1
F

k
_
Y
th
j
, T
th
_
(5)
leads to a TF equation where F is the thickening factor and su-
perscript th indicates thickened quantities. Introducing the variable
changes X
i
=x
i
/F and =t /F leads to
Y
th
k

+
u
i
Y
th
k
X
i
=

X
i
_
D
k
Y
th
k
X
i
_
+
k
_
Y
th
j
, T
th
_
(6)
which has the same solution as Eq. (4) and propagates the ame front
at the same speed s
0
L
. However, Y
th
k
(x, t ) =Y
k
(x/F, t /F) shows that
the ame is thickened by a factor F. The thickened ame thickness
is
th
L
= F
0
L
. Choosing sufciently large values of F allows to obtain
a thickened ame that can be resolved on the LES mesh. Typically,
if n is the number of mesh points within the ame front (n is of the
order of 510) and x the mesh size, the resolved ame thickness

th
L
is nx so that F must be F =nx/s
0
L
. Note that F is not an
additional parameter of the model but is imposed by the preced-
ing relation as soon as the mesh is created. In the framework of
LES, this approach has multiple advantages: When the ame is a
laminar premixed front, the TF model propagates it, in the limit of
an innitely thin front, at the laminar ame speed exactly as in a
G equation approach. However, this ame propagation is due to the
combination of diffusive and reactive terms, which can also act in-
dependently so that quenching (near walls, for example) or ignition
may be simulated. Fully compressible equations may also be used
as required to study combustion instabilities.
The thickeningmodicationof the ame front alsoleads toa mod-
ied interaction between the turbulent ow and the ame: Subgrid-
scale wrinkling must be reintroduced. This effect can be studied and
parameterized using an efciency function E derived fromdirect nu-
merical simulation results.
31,33,34
This efciency function measures
the subgrid-scale wrinkling as a function of the local subgrid turbu-
lent velocity u

e
and the lter width
e
. In practice, the diffusion
coefcient D
k
is replaced by EFD
k
and the preexponential constant
A by AE/F so that the conservation equation for species k is
Y
th
k
t
+
u
i
Y
th
k
x
i
=

x
i
_
EFD
k
Y
th
k
x
i
_
+
E
F

k
_
Y
th
j
, T
th
_
(7)
Such an equation propagates the turbulent ame at a turbulent speed
s
T
= Es
0
L
, while keeping a thickness
th
L
= F
0
L
. In laminar regions,
E goes to unity, and Eq. (7) simply propagates the front at the
laminar ame speed s
0
L
. The subgrid-scale wrinkling function E
was obtained from the initial model of Ref. 31 as a function of the
local lter size
e
, the local subgrid-scale turbulent velocity u

e
, the
laminar ame speed s
0
L
, and the laminar and the ame thicknesses

0
L
and
th
L
.
The TF model uses nite rate chemistry: Here the conguration
corresponds to a lean premixed ame so that a one-step Arrhenius
kinetics is sufcient. This one-step scheme (called 1sCM1) has been
tted with a genetic algorithm-based tool on a laminar ame struc-
ture. The reference mechanism used to t 1sCM1 is the Peters and
Rogg propane scheme.
35
Scheme 1sCM1 takes into account ve
species (C
3
H
8
, O
2
, CO
2
, H
2
O and N
2
):
C
3
H
8
+5O
2
3CO
2
+4H
2
O (8)
The rate of the single step reaction is given by
q = A
_
Y
C
3
H
8
_
W
C
3
H
8
_
n
C
3
H
8 _
Y
O
2
_
W
O
2
_
n
O
2
exp(E
a
/RT) (9)
where the rate parameters are provided in Table 1.
The diffusioncoefcient D
k
of species k is obtainedas D
k
=/S
k
c
,
where is the viscosity and S
k
c
the xed Schmidt number of
Table 1 Rate constants and Schmidt
numbers for 1sCM1 scheme
a
Constants Value
Chemical rate constant
A 3.29E 10
n
C
3
H
8
0.856
n
O
2
0.503
E
a
31526
Schmidt number
C
3
H
8
1.241
O
2
0.728
CO
2
0.941
H
2
O 0.537
N
2
0.690
a
Activation energy is in calories per moles and the
preexponential constants in cgs units.
744 MARTIN ET AL.
species k. The Schmidt number values used in the present simu-
lations are given in Table 1 and correspond to the PREMIX values
measured in the burnt gases. The Prandtl number is set to 0.68.
With this parameter set, the agreement between ame proles ob-
tained using AVBP or PREMIX with the same chemical scheme
is good. The agreement between the Peters and Rogg
35
and the
1sCM1 schemes in terms of laminar ame speed is satisfactory for
the lean to stoechiometric mixtures. Note that other formulations are
available for LES of partially premixed turbulent ows.
5,11
To study
combustion/acoustics coupling, however, the TF model offers the
best compromise. First, the G equation is usually implemented in
low Mach number codes, which do not solve for acoustics, whereas
acoustics are fully represented in the TF model. Second, the TF
approach has been now validated in multiple complex geometry
swirled burners
12,36
making it a proper basis for the present study.
Acoustic Solver for the Helmholtz Equation
The acoustic tool used in this study, called AVSP, solves the eigen-
value problem associated to the wave equation (1). When dealing
withthermoacoustic instabilities, it is usual tomodel the geometryof
the combustor by a network of one-dimensional or two-dimensional
axisymmetric acoustic elements where a simplied form of Eq. (1)
can be solved.
4,15,3739
Jump relations are used to connect all of these
elements, and the amplitude of the forward and backward acoustic
waves are determined so that the boundary conditions are satised.
The main drawback of this approach is that the geometrical details
of a combustor cannot be accounted for and only the rst equiva-
lent longitudinal or orthoradial modes are sought for. In the acoustic
solver, a nite element strategy is used to discretize the exact ge-
ometry of the combustor so that no assumption is made a priori
regarding the shape of the modes. This feature gives the Helmholtz
solver the potential to test the effect of geometrical changes on the
stability of the whole system.
Equation (1) is solved in the frequency domain by assuming har-
monic variations at frequency f =/(2) for pressure, velocity,
and local heat release perturbations:
p
1
= [

P(x) exp(i t )], u
1
= [

U(x) exp(i t )]

T1
= [

T
exp(i t )] (10)
Introducing Eq. (10) into Eq. (1) and neglecting the turbulent noise
p
0
u
1
:u
1
in front of the combustion term( 1)
T1
/t leads
to a modied form of the Helmholtz equation:

0
c
2
0
[(1/
0
)

P] +
2

P = i ( 1)

T
(11)
where the unknown quantities are the complex amplitude

P of the
pressure oscillation at frequency f and pulsation . Note that

T
,
the amplitude of the heat release perturbation, is also unknown and
must be modeled. This is obviously the difcult part of the mod-
eling, and it remains an open research issue today. In the present
work, the simplest linear approach, initally proposed by Crocco,
14
was chosen as a rst step. A direct extension of the standard n x
model
4,14,40
was used to write
T1
nu
1
(x
ref
, t ), where u
1
is
the axial velocity uctuations. In one-dimensional approaches, the
interaction index n and time delay are two parameters describing
the acoustic behavior of a compact ame located at the axial position
x
ref
. In the Helmholtz solver, where the geometry of the combustor
is fully described, the ame is distributed and the interaction index
and time delay depend on space. These data can be extracted from
LES results by postprocessing either a self-excited or a forced oscil-
lating regime. Once measured in LES, the elds n(x, ) and (x, )
are used to model the unsteady heat release in Eq. (11) as

T
= n(x, ) exp[i (x, )]

U(x
ref
) n
ref
(12)
The linearized momentumEuler equation

U =

P/i can be used
to relate

T
to

P and close Eq. (11).
Three types of boundary conditions can be prescribed for Eq. (11),
where nis the outward unit normal vector to the boundary: 1) Dirich-
let condition, which imposes

P =0, on fully reecting outlets;
2) Neumann condition, which imposes

P n=0, on fully rigid
walls or reecting inlets; and 3) Robin condition, which imposes
cZ

P n=i

P, on general boundaries, where Z is the local re-
duced complex impedance Z =

P/
0
c
0

U n. In this study, the re-

duced boundary impedance Z has been obtained using Eq. (16) as
will be described later.
Knowing the boundary impedance Z, the sound speed c
0
and the
density
0
distribution, the ame response [n(x, ), (x, )], and
assuming that Z does not depend on , a Galerkin nite element
method is used to transform Eq. (11) into a nonlinear eigenvalue
problem of size N (the number of nodes in the nite element grid
used to discretize the geometry) of the form
[A][

P] +[B][

P] +
2
[C][

P] = [D()][

P] (13)
where [

P] is the column vector containing the eigenmode at pul-
sation and [A], [B], and [C] are square matrices depending only
on the discretized geometry of the combustor. (Note that the same
result holds if 1/Z =1/Z
0
+ Z
1
+ Z
2
/, where Z
0
, Z
1
, and Z
2
are complex-valued constants.) If the impedances Z change with ,
Eq. (13) can be solved iteratively and independently for each eigen-
mode, by using a Z value adapted to each eigenfrequency. [D()] is
the unsteady contribution of the ame and depends on the pulsation
through the combustion term n(x, ) exp[i (x, )]. No efcient
numerical method exists to solve this nonlinear eigenvalue problem.
However, inthe case where the unsteadyame response is neglected,
namely, [D()] =0, Eq. (13) simplies into a quadratic eigenvalue
problemdepending only on and
2
. Avariable transformation can
then be used to obtain an equivalent linear eigenvalue problem of
size 2 N (Ref. 41). Several numerical methods can then be used
to assess the eigenmodes. Direct methods like the quadratic regu-
lator approach are exact and have the advantage to provide all of
the eigenmodes. However, they can be expensive to solve for large
problems (N >10
3
). Because only the rst fewfrequencies are usu-
ally of interest from a physical point of view, it is more appropriate
to use an iterative method that can be applied for large problems
(N >10
5
) without difculty. In the Helmholtz solver, we are us-
ing a parallel implementation of the Arnoldi method (see Ref. 42),
whichenables to solve complex problems of size N 20000 ina few
minutes.
Setting [D()] =0 is equivalent to nding the eigenmodes of the
burner, taking into account the presence of the ame through the
mean temperature eld but neglecting the ame effect as an acous-
tically active element. The boundary conditions are also acounted
for, and this approximation can provide relevant information on the
shape and real frequency of the rst few modes of the combustor.
However, because there is no coupling between the acoustics and
the ame, there is no hope to discriminate between stable and unsta-
ble modes, which is the ultimate objective of this study. When it is
assumed that the unsteady ame response creates a small perturba-
tion of the modes, a linear expansion technique can be developed to
assess the imaginary part of , hence, the stability of the perturbed
modes.
43,44
Another path has been followed in this study to handle
cases where the unsteady response of the ame changes the modes
signicantly and when the linear expansion is not justied. The non-
linear eigenvalue problem Eq. (13) is then solved iteratively, the kth
iteration consisting in solving the quadratic eigenvalue problem in

k
dened as
([A] [D(
k 1
)])[

P] +
k
[B][

P] +
2
k
[C][

P] = 0 (14)
A natural initialization is to set [D(
0
)] =0 so that the computation
of the modes without combustion is in fact the rst step of the iter-
ation loop. Usually, less than ve iterations are enough to converge
toward the complex pulsation and associated mode. This linearized
approach to describe the stability of the burner in terms of modes has
drawbacks but remains one of the basic tools to study instabilities:
1) The linearization is valid only for small-amplitude perturba-
tions, a condition which is obviously not true when limit cycles typ-
ical of combustion instabilities are observed in gas turbines. How-
ever, this assumption is valid when the instability grows
45
and helps
MARTIN ET AL. 745
to determine the unstable modes: Such modes have to appear and
grow before they reach a limit cycle, and any analysis adapted to
this early phase is of interest.
2) Most acoustic tools work on linear regimes for which each os-
cillatory mode is independent of other modes. Many combustion in-
stabilities exhibit nonlinear coupling, where high-frequency modes
couple with low-frequency oscillations.
46
These were also observed
in the experiment in Ref. 1, in which a 530-Hz mode (often called
rumble) was systematically accompanied by a high-frequency mode
(called screech) at 3750 Hz. The fact that combustion instabilities
involve more than one mode of oscillation is one of the basis of
the approach of Yang and Culick.
47
The tool presented earlier treats
each mode individually and cannot simulate such phenomena.
3) The description of the coupling between acoustics and com-
bustion in such models is extremely crude. The response of the ame
excited by an acoustic wave depends on several physical phenom-
ena such as chemical reactions, species diffusion, vortex shedding,
vortexame interaction, etc. All of these phenomena are not ne-
glected in the present study, but their cumulative effect is modeled
through the global timescale and index n.
Despite these limitations, such tools are useful because they pro-
vide relevant information about the modes triggered by the acoustic
ame coupling while running fast: For the current conguration,
only 8000 grid points were necessary to describe the geometry and
obtain the rst four modes. For comparison, half a million nodes
were used to perfomthe LES. Atypical run for solving the quadratic
eigenvalue problemof type Eq. (14) onthis gridlasts 10minbyusing
15 processors (R14000 500-MHz IP35) on an SGI O3800 parallel
machine. Such a tool can, thus, be used in the design process of
new gas turbines to characterize their thermoacoustic modes. By
describing the whole geometry between the compressor and the tur-
bine, including all of the injectors dispatched around the combustion
chamber, such simulations would give unique information about the
swirling modes that sometimes show up in large gas turbines. The
difcult and computationally expensive task would be to compute
the ame transfer function by performing a LES of the turbulent
ame. Such a simulation would be performed by considering an an-
gular sector corresponding to only one injector, saving grid points
and CPU resources.
Conguration
Geometry: Swirled Premixed Combustor
The methodologies described in the preceding sections were
tested for a swirled combustor shown in Fig. 1. The conguration is
typical of swirled combustion: Premixed gases are introduced tan-
gentially into a long cylindrical duct feeding the combustion cham-
ber. The tangential injection creates the swirl required for stabiliza-
tion. The fuel is propane. The two independant swirler elements
allow fuel staging. The staging parameter is dened as the ratio
of fuel ow rate of the rst to the second swirler.
The regime studied here corresponds to the parameters given in
Table 2. The staging of the burners corresponds to =0.3.
Fig. 1 Staged swirled combustor conguration.
Table 2 Flow parameters for
combustion cases
Parameter Value
Flow rate
Total 22.10
3
Axial 4.10
3
Equivalence ratio
a
0.8
Reynolds number 4.67 10
4
a
Burner mouth.
Table 3 Acoustic inlet and outlet boundaries characteristics
for REF and LEAK
Case
Boundary characteristics LEAK REF
Inlet 1,000 1,000
Outlet 1,000 10,000
Characteristic Nonreecting Reecting
Outlet impedance at 380 Hz
Measured in LES 0.040.21i
Calculated with Eq. (16) 0.850.36i 0.050.23i
Fig. 2 Mean axial velocity eld: white line, iso-u
x
=0 and black line:
iso-T =1500 K for stable combustion.
Boundary Conditions
Specifying boundary conditions is a critical issue for compress-
ible ows. Here, the NavierStokes characteristic boundary condi-
tion technique
4,48
was used at the outlet. The level of reection of
this boundary can be controlled by changing the relaxation coef-
cient of the wave correction,
49
which determines the amplitude
of the incoming wave L
1
entering the computational domain:
L
1
= ( p p
t
) (15)
where p
t
is the prescribed pressure value at innity. Equation (15)
acts on the ow like a spring mechanism with a stiffness . The
impedance of the boundary is a function of and , which can be
obtained analytically
49
for simple cases,
2
Z = (i /)/(1 i /) (16)
(This impedance can be taken into account by the Helmholtz solver
under the following form: 1/Z =1 +i /.)
For small values of , Eq. (15) keeps the pressure p close to
its target value p
t
while letting acoustic waves go out at the same
time
4
: The outlet is nonreecting. When large values of are used,
the outlet pressure remains strictlyequal to p
t
andthe outlet becomes
totally reecting. Two sets of computation will be shown (Table 3).
The rst one, LEAK, corresponds to a case where the spring stiffness
is small so that the outlet is nonreecting and the acoustic waves
are evacuated with very small reection levels. For the second set,
REF, is large and the outlet is reecting. (The pressure oscillation
is almost zero.)
LES Results
Stable Flow
The rst computation corresponds to the case where the out-
let section is nonreecting (case LEAK in Table 3): The acoustic
feedback is minimized, and the ame does not exhibit any strong
unstable movement. The mean velocity and fuel mass fraction elds
are shown in Figs. 2 and 3. As expected, the downstream part of the
746 MARTIN ET AL.
Fig. 3 Mean fuel mass fraction eld: black lines, isoreaction rate for
stable combustion.
Fig. 4 Meannormalizedvalues: , pressure; , heat release; and
, phase angle between pressure and heat release.
Fig. 5 Mean values: , Rayleigh criterion S
1
and . . . . , acoustic
uxes F
1
.
central recirculation zone is lled by burnt gases and stabilize the
turbulent ame.
Instability Sequence
LES also reveals that the combustor can exhibit a strong unstable
mode when the outlet is acoustically closed (case REF). In this
case, soon after ignition, the pressure and the global heat release
start oscillating (Fig. 4) at 380 Hz. To analyze the behavior of this
instability, the following sequence is set up:
1) Starting from a stable ame (LEAK), the outlet impedance is
changed to become reecting (case REF) at time t =0.127 s (Fig. 4).
The oscillation grows and reaches a limit cycle at a frequency of
380-Hz mode.
2) At time t =0.173 s, the outlet impedance is switched again to a
nonreecting condition (case LEAK) and the instability disappears.
This scenarioprovides four phases whichare studiedsequentially:
1) a linear growth between times 0.127 and 0.145 s, 2) an overshoot
phase between 0.145 and 0.160 s, 3) a limit cycle between times
0.160 and 0.173 s, and 4) a decay phase starting at t =0.173 s. For
each phase, the instability is analyzed in terms of ame shape, ame
oscillation, and phase between heat release and pressure. Moreover,
the acoustic energy equation budget is closed, and all terms are
analyzed.
Growth Phase
Once the outlet boundary is acoustically closed (t =0.127 s), the
thermoacoustic instability starts. Figure 5 shows the time variations
of the combustion source term S
1
and the acoustic losses F
1
. The
total acoustic energy evolution of the chamber E
1
is shown in Fig. 6.
Figure 7 shows that the budget of Eq. (3) is quite well closed by
the LES data: The difference S
1
F
1
matches the time derivative
of E
1
. This validates both the LES results and the acoustic energy
equation (3). It is also the rst example of such a treatment for
a resonating combustor. Because the budget is closed, individual
terms can then be analyzed.
First, the phase angle between pressure and heat release is shown
in Fig. 4. During the growth phase, it is close to zero and slowly
Fig. 6 Evolution of burner acoustic energy E
1
.
Fig. 7 Comparison: . . . . , time derivative of the acoustic energy
E
1
/t and , S
1
F
1
.
shifting toward /4, leading to a strong coupling between pressure
and heat release, that is, a positive S
1
term. During the growth phase,
the source term S
1
is large and always positive (Fig. 5) because the
phase angle stays in the [/2; /2] range. Figure 5 shows that the
acoustic losses balance the reacting term S
1
in the acoustic budget
equation. The limit cycle is controlled by acoustic losses and not by
combustion.
Overshoot Phase and Limit Cycle
At t =0.160 s, the instability reaches a limit cycle at 380 Hz. Be-
fore reaching this limit cycle, a large overshoot of acoustic energy is
observed: This is typical of combustion instabilities, and it has been
observed experimentally in other systems.
45,50
Figure 4 shows that,
reaching the nonlinear zone, the phase difference between pressure
and heat release increases from zero to /4 in the limit-cycle zone.
The drift of this phase difference together with increasing acoustic
losses lead to the saturation of the instability.
The coupling loop between p
1
and
T1
can be identied from
LES as follows. The longitudinal mode induces the formation of a
vortex ring at the dump plane. This vortex ring strongly interacts in
phase with the ame. Figure 8 shows the interaction between the
acoustically induced vortex ring and the ame brush. Figure 9 allows
to locate LES snapshots in the acoustic period and displays the mean
pressure uctuation p
1
in the ame zone, the heat release uctuation

T1
, andthe uctuationof the meanvelocityinthe dumpplane u
dump
1
.
At instant 1, a vortex ring appears at the dump plane when du
dump
1
/dt
is maximum. The ring structure detaches and is convected through
the ame by the mean ow (instants 2, 3, and 4). Between instants 1
and 3, the ow stretches the ame, increasing its area, whereas the
ame wrinkling by the vortex ring remains weak. Consequently

T1
increases with a medium slope (Fig. 9). Between points 3 and
5, the vortex ring is stretching the ame and
T1
increases faster.
Moreover, the vortex ring is gradually destroyed, and its global
coherence disappears between instants 4 and 5, at a moment when
du
dump
1
/dt is minimum. At instant 5, some coherent structures are
still interacting with the ame, producing (noisy) ame pockets and
cusps. After instant 5, the ame burns out the fresh gases present in
the chamber and propagates back to the injection pipe decreasing
the overall ame surface and
T1
.
Decay Phase
The decay phase is triggered by the sudden change in the acous-
tic outlet boundary condition switching to nonreecting (LEAK) at
t =0.173 s. The phase angle
p
between pressure and heat release
increases by a large amount:
p
>/2 at t =0.174 s, that is, 1.2 ms
after relaxing outlet pressure (Fig. 4). At this time, S
1
becomes glob-
ally negative for the rst time and the instability is rapidly damped.
The acoustic losses actually become positive (gain term) during this
last oscillation but the instability engine is broken and the Rayleigh
MARTIN ET AL. 747
Fig. 8 Vortex ring shedding at six instants (Fig. 9) during the limit
cycle: isosurface, Q vortex criterion and black lines, isoreaction rate in
the burner central plane.
Fig. 9 Time signals during limit cycle and snapshots corresponding to
Fig. 8: , pressure; . . . . , inlet velocity; and , total heat release.
termS
1
becomes negative during a half-cycle (0.173 <t <0.175 ms
in Fig. 5) leading to the immediate decay of all unstable activity.
Acoustic Analysis Results
This section describes the results obtained with the Helmholtz
solver for the conguration of Fig. 1. The impedances are calcu-
lated from Eq. (16) using the value of given in Table 3. Equa-
tion (16) shows that, for a xed coefcient, the impedance is a
function of the pulsation . To verify that Eq. (16) is sufciently
accurate, Table 3 gives the value of the impedance at 380 Hz (which
is the frequency of the rst mode observed in the LES) predicted
by Eq. (16) and measured in the LES. For these computations, the
ame transfer function is needed. In the present work, n and are
obtained by postprocessing LES results as described in the follow-
ing section. These results are then used to predict the frequency and
growth rates of all modes using the Helmholtz equation (11) and to
compare them to the LES results of the preceding section.
Measurement of Flame Transfer Function
The key mechanism to predict combustion instabilities in
Helmholtz codes is the ame transfer function (FTF). In the model
chosen in this study [Eq. (12)], the heat release uctuations are re-
lated to the velocity uctuations at a reference point through the
Fig. 10 Flame transfer function magnitude (n parameter) eld in
central plane evaluated at f =432 Hz: black line, iso-n=2.5
10
6
Pa m
1
and +, reference point location.
magnitude n and the phase of the FTF dened as
n(x, ) =

T
(x)

U(x
ref
) n
ref

(17)
(x, ) = (x, ) = arg
_

T
(x)

U(x
ref
) n
ref
_
(18)
Because the ame is more prone to interact with longitudinal
oscillations, the reference normal vector n
ref
is set collinear to the
axial direction so that the scalar product

U(x
ref
) n
ref
corresponds
to the axial velocity uctuation at the reference point. This point
is located following the Crocco approach,
14
that is, in the fresh gas
near the inlet combustor. As shown in Fig. 2, a recirculation zone
located at the entrance of the dump combustor is induced by the
swirling ow. Consequently, the reference point is located upstream
of this zone
3
at 113 mm from the inlet of the device (Fig. 10). Note
that other positions in the vicinity of this location have been tested
leading to very similar results with the Helmholtz solver.
Flame transfer functions are usually measured for the whole
combustor
39,51,52
or by zones.
53
In the present approach, n and
must be obtained locally at each point of the combustor, and this
can be done by analyzing snapshot series from LES. Moreover, the
frequency dependence of these parameters should be taken into ac-
count. To do so, it is assumed that FTF depends mostly on the real
part of the frequency and not of the growth rate of the mode. Thus,
at each subset k of the algorithm [Eq. (14)], the FTF parameters
are evaluated at [
k
/(2)], and the rst guess corresponds to the
eigenmodes determined without active ame (where n is set to zero
everywhere). (Further studies are needed to assess this assumption
used for practical reasons.) From a practical point of view, n and
can be extracted from LES results by Fourier processing the local
heat release and the unsteady velocity at the reference point and
using Eqs. (17) and (18). Two methodologies can be considered for
such an analysis: 1) use a stable regime and force the ow (usually
by introducing acoustic waves through the inlet
39,40,54
) or 2) use a
self-excited regime.
The choice of the methodology raises other fundamental issues,
which are still open today:
1) When it is assumed that Eq. (12) is valid for both forced and
self-excited cases, both methodologies are expected to provide the
same n and elds. This is a theoretical argument that has not been
checked yet and is left for further work.
2) The possible dependence of the transfer function on the acous-
tics wave amplitude raises an additional difculty. If n and depend
on the wave amplitude, then both methodologies have drawbacks:
The forcing method will provide different n and elds when the
forcing amplitude changes, whereas the self-excited method uses a
nonlinear limit cycle to evaluate n and .
These questions are beyond the scope of the present study. Here
the elds of n and were measured by postprocessing the self-
excited regime between t =0.145 and t =0.177 s shown in Fig. 4.
Results given in the next section prove the validity of this method,
but further studies are obviously needed. The resulting ame trans-
fer function parameters, shown only within the ame zone (when
n 2.5 10
6
Pa m
1
), are shown in Figs. 10 and 11.
748 MARTIN ET AL.
Table 4 Helmholtz solver results
No active ame Active ame
Mode F, Hz , rad s
1
F, Hz , rad s
1
1 312 48 334 588
2 431 32 432 600
3 841 4.5 825 6
4 1140 63 1116 363
Fig. 11 Flame transfer function phase ( parameter) eld in the
central plane evaluated at f =432 Hz: black line, iso-n=2.5
10
6
Pa m
1
and +, reference point location.
Fig. 12 Acoustic pressure modulus for the rst four modes calculated
by the Helmholtz solver with active acoustic ame in REF case, ||

P||
isolines.
Helmholtz Solver Results
Using the mean elds given by LES, the Helmholtz solver is
applied to obtain the thermoacoustic eigenmodes of the burner. This
tool can be run using either a nonactive ame, that is, setting n to
zero everywhere, or an active ame, using the postprocessed n and
elds displayed. Moreover, the impedance values are determined
from the LES boundary conditions settings (Table 3).
First, when the outlet impedance corresponds to the LEAK case,
the Helmholtz solver predicts that all modes are damped. This con-
rms the LES result of Fig. 2 for which no acoustic mode was found
for this regime.
Second, when the outlet impedance corresponds to the REF case,
Table 4 presents the modes that are obtained for both the nonactive
and the active ames. These four modes in the REF case are all
longitudinal except in the vicinity of the swirler: The structure of
the modes in the central plane of the burner is shown in Fig. 12,
whereas a one-dimensional cut of the rms pressure eld is given in
Fig. 13. The real frequencies of the active and nonactive cases are
very similar: Typically, taking into account the active effect of the
ame shifts the eigenmode frequency by a fewpercent. However the
effect on the growth rate is more dramatic: All modes computed with
the nonactive ame are damped, whereas the modes computed with
an active ame are excited for modes 2, 3, and 4. The fastest growing
Fig. 13 Longitudinal structure of rst four modes obtained from
Helmholtz solver: , normalized p
rms
1
evolution along burner axis
with acoustics/ame coupling and , without.
Fig. 14 RMS pressure uctuations p
rms
1
along burner axis; , LES
and , solvers.
mode is mode 2 at 432 Hz, which is close to the mode observed in the
LES (380 Hz). To verify that the mode 2 obtained by the Helmholtz
solver is indeed the mode appearing in the LES, Fig. 14 shows a
comparison of the |

P| proles on the axis and conrms the good
agreement of LESand Helmholtz solver. The difference between the
frequency observed in the LES(380 Hz) and the frequency predicted
by the Helmholtz solver (432 Hz) is probably due to the zero Mach
number approximation for the Helmholtz solver.
Conclusions
Three tools have been used to analyse ameacoustics coupling
mechanisms in a staged swirled combustor: Full compressible LES,
Helmholtz analysis, and budget of acoustic energy. The two latter
methods are based on the wave equation in reacting ows. They
use LES results but provide essential new elements: The Helmholtz
results allow to predict the stability of the combustor and the exact
identication of modes appearing during the instability, whereas the
budget of acoustic energy demonstrates that the Rayleigh criterion is
not the only or even the largest termin the acoustic energy equation.
Acoustic losses at the outlet of the combustor contribute signicantly
to the budget of acoustic energy and determine the levels of oscil-
lation amplitudes as well as their appearance. More generally, this
study conrms the need of coupling classical computational uid
dynamics (here LES) and acoustic analysis to understand combus-
tion instabilities. It also demonstrates the crucial effect of acoustic
boundary conditions on the stability of combustors. This has a prac-
tical implication. The stability of a given combustion chamber is
controlled by acoustic impedances upstreamand downstreamof the
combustor. Removing a combustor section froma full gas turbine to
install it in a laboratory setup obviously becomes very dangerous to
MARTIN ET AL. 749
study the stability of the burner. Indeed, the combustion may prove
to be stable in one case and unstable in the other one. This result
calls for the development of more coupled acoustics analysis of the
whole turbine.
Acknowledgments
Certain numerical simulations have been conducted on the com-
puters of the Centre Informatique National de lEnseignement
Sup erieur and Institut du D eveloppement et des Ressources en In-
formatique Scientique French national computing centers. Sim-
ulations have been supported partly by Alstom Power and by the
European Community Program (WP2) FUELCHIEF.
References
1
Poinsot, T., Trouve, A., Veynante, D., Candel, S., and Esposito, E., Vor-
tex Driven Acoustically Coupled Combustion Instabilities, Journal of Fluid
Mechanics, Vol. 177, 1987, pp. 265292.
2
Candel, S., Combustion Instabilities Coupled by Pressure Waves and
Their Active Control, 24th Symposium (International) on Combustion,
Combustion Inst., Pittsburgh, PA, 1992, pp. 12771296.
3
Crighton, D. G., Dowling, A., Ffowcs Williams, J. E., Heckl, M., and
Leppington, F., Modern Methods in Analytical Acoustics, Springer-Verlag,
Berlin, 1992, Chap. 13.
4
Poinsot, T., and Veynante, D., Theoretical and Numerical Combustion,
R. T. Edwards, Philadelphia, 2005, Chap. 8.
5
Peters, N., Turbulent Combustion, Cambridge Univ. Press, Cambridge,
England, U.K., 2000.
6
Angelberger, C., Egolfopoulos, F., and Veynante, D., Large Eddy Simu-
lations of Chemical and Acoustic Effects on Combustion Instabilities, Flow
Turbulence and Combustion, Vol. 65, No. 2, 2000, pp. 205220.
7
Caraeni, D., Bergstr om, C., and Fuchs, L., Modeling of Liquid Fuel In-
jection, Evaporation and Mixing in a Gas Turbine Burner Using Large Eddy
Simulation, FlowTurbulence and Combustion, Vol. 65, 2000, pp. 223244.
8
Colin, O., and Rudgyard, M., Development of High-Order Taylor-
Galerkin Schemes for Unsteady Calculations, Journal of Computational
Physics, Vol. 162, No. 2, 2000, pp. 338371.
9
DesJardin, P. E., and Frankel, S. H., Two Dimensional Large Eddy
Simulation of Soot Formation in the Near Field of a Strongly Radiating
Nonpremixed AcetyleneAir Jet Flame, Combustion and Flame, Vol. 119,
No. 1/2, 1999, pp. 121133.
10
Pierce, C. D., and Moin, P., Progress-Variable Approach for Large
Eddy Simulation of Nonpremixed Turbulent Combustion, Journal of Fluid
Mechanics, Vol. 504, 2004, pp. 7397.
11
Pitsch, H., and Duchamp de la Geneste, L., Large Eddy Simulation of
Premixed Turbulent Combustion Using a Level-Set Approach, Proceedings
of the Combustion Institute, Vol. 29, 2002, pp. 20012008.
12
Selle, L., Lartigue, G., Poinsot, T., Koch, R., Schildmacher, K. U.,
Krebs, W., Prade, B., Kaufmann, P., and Veynante, D., Compressible
Large-Eddy Simulation of Turbulent Combustion in Complex Geometry
on Unstructured Meshes, Combustion and Flame, Vol. 137, No. 4, 2004,
pp. 489505.
13
Martin, C., Benoit, L., Nicoud, F., and Poinsot, T., Analysis of Acoustic
Energy and Modes in a Turbulent Swirled Combustor, Proceedings of the
Summer Program, Center for Turbulence Research, NASA Ames Research
Center/Stanford Univ., Stanford, CA, 2004, pp. 377394.
14
Crocco, L., Research on Combustion Instability in Liquid Propellant
Rockets, 12th Symposium(International) on Combustion, Combustion Inst.
Pittsburgh, PA, 1969, pp. 8599.
15
Stow, S. R., and Dowling, A. P., Thermoacoustic Oscillations in an
Annular Combustor, American Society of Mechanical Engineers, ASME
Paper 2001-GT-0037, July 2001.
16
Culick, F. E. C., Combustion Instabilities in Liquid-Fueled Propulsion
SystemsAn Overview, AGARD, Vol. 72B, 1988, pp. 173.
17
Dowling, A. P., The Calculation of Termoacoustic Oscillations, Jour-
nal of Sound and Vibration, Vol. 180, No. 4, 1995, pp. 557581.
18
Rayleigh, L., The Explanation of Certain Acoustic Phenomena, Na-
ture, Vol. 18, July 1878, pp. 319321.
19
L egier, J.-P., Poinsot, T., and Veynante, D., Dynamically Thickened
Flame Large Eddy Simulation Model for Premixed and Non-Premixed Tur-
bulent Combustion, Summer Program 2000, Center for Turbulence Re-
search, Stanford Univ., Stanford, CA, 2000.
20
Duchamp de Lageneste, L., and Pitsch, H., Progress in Large Eddy
Simulation of Premixed and Partially Premixed Turbulent Combustion,
Annual Research Briefs, Center for Turbulence Research, NASA Ames Re-
search Center/Stanford Univ., Stanford, CA, 2001, pp. 97107.
21
Kempf, A., Forkel, H., Chen, J.-Y., Sadiki, A., and Janicka, J., Large-
Eddy Simulation of a Counterow Conguration with and Without Com-
bustion, Proceedings of the Combustion Institute, Vol. 28, 2000, pp. 3540.
22
Pitsch, H., and Steiner, H., Large Eddy Simulation of a Turbulent
Piloted Methane/Air Diffusion Flame (Sandia Flame D), Physics of Fluids,
Vol. 12, No. 10, 2000, pp. 25412554.
23
Lele, S., Compact Finite Difference Schemes with Spectral Like Reso-
lution, Journal of Computational Physics, Vol. 103, No. 1, 1992, pp. 1642.
24
Ducros, F., Comte, P., and Lesieur, M., Large-Eddy Simulation of
Transition to Turbulence in a Boundary Layer Developing Spatially over a
Flat Plate, Journal of Fluid Mechanics, Vol. 326, 1996, pp. 136.
25
Gamet, L., Ducros, F., Nicoud, F., and Poinsot, T., Compact Finite Dif-
ference Schemes on Non-UniformMeshes. Application to Direct Numerical
Simulations of Compressible Flows, International Journal for Numerical
Methods in Fluids, Vol. 29, No. 2, 1999, pp. 159191.
26
Sagaut, P., Large Eddy Simulation for Incompressible Flows, Springer-
Verlag, 2000, Chap. 5.
27
Scotti, A., Meneveau, C., and Lilly, D. K., Generalized Smagorin-
ski Model for Anisotropic Grids, Physics of Fluids, Vol. 5, No. 9, 1993,
pp. 23062308.
28
Scotti, A., Meneveau, C., and Fatica, M., Generalized Smagorinski
Model for Anisotropic Grids, Physics of Fluids, Vol. 9, No. 6, 1997,
pp. 18561858.
29
Vasilyev, O. V., Lund, T. S., and Moin, P., A General Class of Commu-
tative Filters for LES in Complex Geometries, Journal of Computational
Physics, Vol. 146, 1998, pp. 82104.
30
Nicoud, F., and Ducros, F., Subgrid-Scale Stress Modelling Based on
the Square of the Velocity Gradient, Flow Turbulence and Combustion,
Vol. 62, No. 3, 1999, pp. 183200.
31
Colin, O., Ducros, F., Veynante, D., andPoinsot, T., AThickenedFlame
Model for Large Eddy Simulations of Turbulent Premixed Combustion,
Physics of Fluids, Vol. 12, No. 7, 2000, pp. 18431863.
32
Butler, T. D., and ORourke, P. J., A Numerical Method for Two-
Dimensional Unsteady Reacting Flows, 16th Symposium (International)
on Combustion, Combustion Inst., Pittsburgh, PA, 1977, pp. 15031517.
33
Angelberger, C., Veynante, D., Egolfopoulos, F., and Poinsot, T., Large
Eddy Simulations of Combustion Instabilities in Premixed Flames, Pro-
ceedings of the Summer Program, Center for Turbulence Research, NASA
Ames Research Center/Stanford Univ., Stanford, CA, 1998, pp. 6183.
34
Charlette, F., Veynante, D., andMeneveau, C., APower-LawWrinkling
Model for LES of Premixed Turbulent Combustion: Part INon-Dynamic
Formulation and Initial Tests, Combustion and Flame, Vol. 131, No. 12,
2002, pp. 159180.
35
Peters, N., and Rogg, B., Reduced Kinetic Mechanisms for Applica-
tions in Combustion Systems, Springer-Verlag, Heidelberg, Germany, 1993,
Chap. 8.
36
Roux, S., Lartigue, G., Poinsot, T., Meier, U., and Berat, C., Studies
of Mean and Unsteady Flow in a Swirled Combustor Using Experiments,
Acoustic Analysis, and Large Eddy Simulations, Combustion and Flame,
Vol. 141, No. 1, 2005, pp. 4054.
37
Paschereit, C. O., Flohr, P., and Schuermans, B., Prediction of Com-
bustion Oscillations in Gas Turbine Combustors, AIAA Paper 2001-484,
Jan. 2001.
38
Krueger, U., Hueren, J., Hoffmann, S., Krebs, W., Flohr, P., and Bohn,
D., Prediction and Measurements of Thermoacoustic Improvements in Gas
Turbines with Annular Combustion Systems, American Society of Mechan-
ical Engineers, ASME Paper 2000-GT-567, Munich, May 2000.
39
Polifke, W., Poncet, A., Paschereit, C. O., and Doebbeling, K., Re-
construction of Acoustic Transfer Matrices by Instationary Computational
Fluid Dynamics, Journal of Sound and Vibration, Vol. 245, No. 3, 2001,
pp. 483510.
40
Kaufmann, A., Nicoud, F., and Poinsot, T., Flow Forcing Techniques
for Numerical Simulation of Combustion Instabilities, Combustion and
Flame, Vol. 131, No. 4, 2002, pp. 371385.
41
Chatelin, F., Eigenvalues of Matrices, Wiley, 1993, pp. 121, 122.
42
Lehoucq, R., Maschoff, K., Sorensen, D., and Yang, C., ARPACK
Homepage, http://www.caam.rice.edu/software/ARPACK, 1996.
43
Nicoud, F., and Benoit, L., Global Tools for Thermo-Acoustic Insta-
bilities in Gas Turbines, Bulletin of the American Physical Society, Vol. 48,
No. 10, 2003, pp. 240, 241.
44
Benoit, L., and Nicoud, F., Numerical Assessment of Thermo-Acoustic
Instabilities in Gas Turbines, ICFD Conference of Numerical Methods for
Fluid Dynamics, Oxford Univ. Press, 2004, pp. 13631370.
45
Poinsot, T., Veynante, D., Bourienne, F., Candel, S., Esposito, E., and
Surjet, J., Initiation and Suppression of Combustion Instabilities by Ac-
tive Control, 22nd Symposium (International) on Combustion, Combustion
Inst., Pittsburgh, PA, 1988, pp. 13631370.
46
Rogers, D. E., and Marble, F. E., AMechanismfor High Frequency Os-
cillations in Ramjet Combustors and Afterburners, Jet Propulsion, Vol. 26,
No. 1, 1956, pp. 456462.
47
Yang, V., and Culick, F. E. C., Analysis of Low-Frequency Combus-
tion Instabilities in a Laboratory Ramjet Combustor, Combustion Science
Technology, Vol. 45, No. 12, 1986, pp. 125.
750 MARTIN ET AL.
48
Poinsot, T., and Lele, S., Boundary Conditions for Direct Simula-
tions of Compressible Viscous Flows, Journal of Computational Physics,
Vol. 101, No. 1, 1992, pp. 104129.
49
Selle, L., Nicoud, F., and Poinsot, T., The Actual Impedance of Non-
Reecting Boundary Conditions: Implications for the Computation of Res-
onators, AIAA Journal, Vol. 42, No. 5, 2004, pp. 958964.
50
Lang, W., Poinsot, T., and Candel, S., Active Control of Com-
bustion Instability, Combustion and Flame, Vol. 70, No. 3, 1987,
pp. 281289.
51
Flohr, P., Paschereit, C. O., and Belluci, V., Steady CFD Analysis
for Gas Turbine Burner Transfer Functions, AIAA Paper 2003-1346, Jan.
2003.
52
Pankiewitz, C., Fischer, A., Hirsch, C., and Sattelmayer, T., Computa-
tion of Transfer Matrices for Gas Turbine Combustors Including Acoustics/
Flame Interaction, AIAA Paper 2003-3295, May 2003.
53
Varoquie, B., Legier, J., Lacas, F., Veynante, D., and Poinsot, T., Ex-
perimental Analysis and Large Eddy Simulation to Determine the Response
of Non-Premixed Flame Submitted to Acoustic Forcing, Proceedings of the
Combustion Institute, Vol. 29, 2002, pp. 19651970.
54
Ducruix, S., and Candel, S., External Flow Modulation in Computa-
tional Fluid Dynamics, AIAA Journal, Vol. 42, No. 8, 2004, pp. 15501558.
J. C. Oefelein
Guest Editor