These

Num ero dordre : 2458
TH
`
ESE
pr esent ee pour obtenir le titre de
DOCTEUR DE
LINSTITUT NATIONAL POLYTECHNIQUE
DE TOULOUSE
Ecole doctorale: TYFEP
Sp ecialit e: Dynamique des Fluides
Directeur de th` ese: Thierry POINSOT
Par M. Alexis GIAUQUE
FONCTIONS DE TRANSFERT DE FLAMME
ET ENERGIES DES PERTURBATIONS
DANS LES ECOULEMENTS REACTIFS
Soutenue le 14 Mars 2007 devant le jury compos e de:
G. SEARBY Directeur de recherche ` a lIRPHE de Marseille Rapporteur
C. BAILLY Professeur ` a lEcole Centrale de Lyon Rapporteur
F. VUILLOT Ing enieur/Chercheur ` a lONERA/DSNA Examinateur
S. DUCRUIX Professeur ` a l
Ecole Centrale de Paris Examinateur

F. NICOUD Professeur ` a lUniversit e de Montpellier II Examinateur
T. POINSOT Directeur de Recherche ` a lIMF de Toulouse Directeur de Th` ese
R ef. CERFACS : TH/CFD/07/15
Figure 1 - Photography of a real wood re
Figure 2 - Picture of a simulated methane ame (detail of gure 4.11)
R esum e
Ces travaux de th` ese pr esentent une etude num erique (utilisant la Simulation aux Grandes Echelles)
des instabilit es de combustion. Lobjectif g en eral est den approfondir la compr ehension en d eveloppant
de nouveaux concepts physiques ainsi que de nouvelles m ethodes danalyses. Deux aspects principaux
des instabilit es de combustion sont trait es :
Les Fonctions de Transfert de Flamme (FTF)
Quatre m ethodes diff erentes pour la d etermination des Fonctions de Transfert de Flamme (FTF)
sont evalu ees ` a laide de la Simulation aux Grandes Echelles (SGE). La m ethode HF-FFT, re-
posant sur un forcage harmonique de la chambre de combustion et sur lanalyse de Fourier pr evoit
des d elais de combustion globaux en accord avec lexp erience. Cette m ethode fournit aussi la FTF
locale (cest ` a dire la r eponse de chaque point de la chambre ` a une excitation en entr ee) et permet
de localiser les zones ayant la plus forte r eponse ` a lexcitation. Elle permet aussi de connatre la
r epartition spatiale des d elais de cette r eponse. Cette m ethode devrait donc se r ev eler utile pour
lanalyse de congurations dans lesquelles la r eponse de la amme nest pas compacte par rapport
` a la longueur donde de lexcitation.
Une nouvelle m ethode WN-WH, reposant sur le forcage par un bruit blanc ltr e de la chambre
de combustion et une analyse utilisant la relation de Wiener-Hopf est compar ee avec succ` es ` a la
m ethode HF-FFT. Bien que WN-WH semble plus difcile ` a mettre en oeuvre que HF-FFT, son
avantage principal r eside en ce quelle donne acc` es au spectre de fr equence de la FTF locale sans
augmenter le co ut de calcul.
Cette etude est en lien etroit avec lanalyse de stabilit e des chambres de combustion. En effet,
la FTF a une inuence sur la fr equence et les taux damplication des modes dans les outils
num eriques utilis ees pour d eterminer la stabilit e des br uleurs. Cette etude montre comment con-
struire cette FTF qui est un el ement essentiel de ce type danalyse.
Les energies des uctuations et les crit` eres de stabilit e dans les ecoulements r eactifs
Poursuivant les travaux de Chu [23] et de Myers [88], cette th` ese d eveloppe une nouvelle equation
de conservation pour une energie nonlin eaire des uctuations en combustion. Un nouvel outil
de post-traitement modulaire est utilis e pour v erier la fermeture des equations des energies des
uctuations sur des ammes laminaires mono et bidimensionnelles .
Cet outil donne par ailleurs acc` es ` a tous les termes physiques et num eriques responsables de
l evolution de ces energies dans l ecoulement. Pour chaque equation, les principaux termes
source sont identi es et cette analyse fournit deux crit` eres de stabilit e pour les ecoulements
r eactifs. Ces crit` eres sont valid es sur le cas dune instabilit e se d eveloppant dans une chambre de
combustion bi-dimensionelle.
Le premier crit` ere etend le crit` ere de Rayleigh lin eaire en y introduisant linuence de la uc-
tuation du ux de chaleur. Ce travail fournit donc un outil d etude lin eaire de la stabilit e des
chambres de combustion.
Il montre egalement que l equation de conservation de l energie entropique des uctuations ne
peut pas etre lin earis ee dans les cas r eactifs en raison de lamplitude des uctuations locales de
temp erature. Le deuxi` eme crit` ere est donc non lin eaire an dinclure linuence de l energie
entropique sur la stabilit e de la amme. Ce crit` ere permet dobtenir des informations sur la
stabilit e de la chambre de combustion lorsquaucune lin earisation de l ecoulement nest possible.
Mots clefs : Simulation aux Grandes Echelles (SGE), Instabilit es thermo-acoustiques, Fonctions
de Transfert de Flamme (FTF), Energies des uctuations dans les ecoulements r eactifs, Crit` eres de
stabilit e en combustion
Abstract
The general objective of this thesis is to extend the understanding of combustion instabilities by testing
models, physical concepts and numerical procedures, and by providing new numerical post-processing
tools to do so. Two main aspects of combustion instabilities are studied and lead to many outputs.
Flame Transfer Function (FTF)
Four different methods for the determination of Flame Transfer Functions (FTFs) in LES have
been tested. HF-FFT method, based on harmonic ame forcing and FFT post-processing gives
results of global combustion delays that compare well with the experiments. This method also
provides the local FTF which gives valuable information about the amplitudes, delays and maxi-
mum locations of the local ame response. This method should reveal useful for congurations
where the response of the ame is not compact compared to the characteristic wavelength of the
excitation, as for distributed reacting cases.
A new WN-WH method based on ltered-white noise forcing and post-processing using the
Wiener-Hopf relation is successfully compared to HF-FFT. Though this method should be han-
dled with care, its main advantage is that it gives access to the frequency spectrum of the local FTF
with no additional computational cost.
An important aspect of this study is its link with stability analysis of combustors. Obviously, FTFs
do have an inuence on the frequency and amplication rates of modes in the numerical methods
used for combustor stability. This study shows how to construct FTFs which are an important
brick of acoustic analysis.
Disturbance Energies and Stability Criteria in reacting Flows.
Following the works of Chu [23] and Myers [88] for non-reacting ows, a new nonlinear conser-
vation equation for a disturbance energy in gaseous reacting ows is derived.
A new modular post-processing tool is used here to check the balance closure of disturbance ener-
gies on laminar 1D and 2D ames. This tool gives access to all the physical and numerical terms
responsible for the evolution of disturbance energies in the ow.
For each equation, major terms are identied and this work proposes two stability criteria for re-
acting ows. These criteria are validated on the case of an instability developing in a 2D reacting
4
conguration.
The rst criterion extends the linear Rayleigh criterion by taking into account the inuence of the
uctuation of the heat ux. This work therefore gives a relevant linear tool for the study of com-
bustion chambers stability.
Besides, it also shows that the entropy disturbance energy cannot be linearized in reacting ows
because of the local amplitude of temperature uctuations. The second criterion is therefore non-
linear to include the inuence of the entropy disturbance energy on the global stability. This
criterion gives relevant information on the stability when no linearization of the ow is possible.
Keywords : Large-Eddy Simulation (LES), Thermo-acoustic instabilities, Flame Transfer Func-
tion (FTF), Disturbance Energies in reacting ows, Stability Criteria in combustion
5
Contents
Remerciements 11
List of symbols 13
Introduction 17
Objectives & Organisation 28
I Description of the numerical tool 33
1 Description of the numerical tool AVBP 35
1.1 Governing Equations and hypothesis for DNS . . . . . . . . . . . . . . . . . . . . . . . 35
1.2 Governing equations for LES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
1.3 The Thickened Flame (TF) model for LES . . . . . . . . . . . . . . . . . . . . . . . . 50
1.4 General aspects of the Boundary Conditions in AVBP . . . . . . . . . . . . . . . . . . . 55
1.5 Articial Viscosity in AVBP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
1.6 Cell-Vertex Discretization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
II Flame Transfer Functions 63
2 Linear models for FTFs (Flame Transfer Functions) 69
2.1 Historical background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
CONTENTS
2.2 Models for FTF . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
3 Measurement methods for FTF in LES 81
3.1 LES of FTF . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
3.2 Postprocessing methods for local FTF . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
3.3 Preliminary comparisons of postprocessing methods. . . . . . . . . . . . . . . . . . . . 85
4 Transfer functions of ames 95
4.1 Conguration A : laminar planar premixed ame (1D) . . . . . . . . . . . . . . . . . . 97
4.2 Conguration B : laminar V ame (2D axi-symmetric) . . . . . . . . . . . . . . . . . . 101
4.3 Conguration C : turbulent burner in cylindrical chamber (3D) . . . . . . . . . . . . . . 104
4.4 Conguration D : turbulent burner in a 15
sector (3D) . . . . . . . . . . . . . . . . . . 127

4.5 Evaluation of FTF measurements methods in LES. . . . . . . . . . . . . . . . . . . . . 138
III Disturbance energies and stability criteria in reacting ows 141
5 Introduction 147
5.1 Previous studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
5.2 An advanced post-processing tool for LES : POSTTIT . . . . . . . . . . . . . . . . . . 161
5.3 Examples of conservation equation balance closure . . . . . . . . . . . . . . . . . . . . 171
6 Disturbance energies in ow 175
6.1 Pressure-Velocity (PV) disturbance energy Eq.(1)[Eq.(6.10)] . . . . . . . . . . . . . . . . . . 175
6.2 Entropy disturbance energy Eq.(2)[Eq.(6.18)] . . . . . . . . . . . . . . . . . . . . . . . . . . 181
6.3 Nonlinear disturbance energy Eq.(3)[Eq.(6.37)] . . . . . . . . . . . . . . . . . . . . . . . . . 184
6.4 The choice of the baseline ow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
6.5 Summary of disturbance energies conservation equations . . . . . . . . . . . . . . . . . 191
7 Results 195
8
CONTENTS
7.1 Conguration A (1-D Reacting Case) . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
7.2 Conguration B (2-D Reacting Case) . . . . . . . . . . . . . . . . . . . . . . . . . . . 210
7.3 Summary of the results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 258
8 Stability criteria in reacting ows 263
8.1 Evolution of disturbance energies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263
8.2 Linear criteria for stability : Rayleigh Criteria, Chu criterion . . . . . . . . . . . . . . . 265
8.3 Deriving stability criteria from Eqs.(1), (2) and (3) . . . . . . . . . . . . . . . . . . . . 268
8.4 Conclusion and prospects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272
General Conclusion 275
Bibliography 277
Appendix 291
A Linear conservation equation for Pressure-Velocity (PV) disturbance energy 291
B Linear conservation equation for Entropy disturbance energy 297
C Exact conservation equation for a nonlinear disturbance energy 301
D Evaluation of numerical corrections for Eqs.(1), (2) and (3) 311
E Short notations for balance closure analysis (Chapter 7). 313
E.1 Equation 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 313
E.2 Equation 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 314
E.3 Equation 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 314
F Publication in Proceedings of the Stanford CTR 2006 Summer Program. 315
9
CONTENTS
G Publication submitted to Combustion and Flame. 329
10
Remerciements
Je remercie tout particuli` erement Anne Clausse, sans qui cette th` ese ne serait pas la m eme, ainsi que
beaucoup dautres choses! Merci...
Je tiens ` a remercier mon directeur de th` ese, le Professeur Thierry Poinsot, pour mavoir permis
dentreprendre cette etude ` a la fois th eorique et num erique concernant les instabilit es de combustion. Je
le remercie tout particuli` erement pour son aide dans l etude des fonctions de transfert de amme.
Je tiens ` a remercier chaleureusement le Professeur Franck Nicoud pour son soutien et son aide, en parti-
culier concernant l etudes des energies des perturbations en combustion. Nous avons eu de nombreuses
discussions fructueuses qui nous ont permis de mener ` a bien une etude int eressante de ces energies.
Je remercie Charles Martin pour son aide concernant l etude de l energie acoustique en combustion. Il a
mis en place une premi` ere approche tr` es utile de l evolution de cette energie. Le lecteur est renvoy e ` a sa
th` ese dans laquelle gurent certaines analyses en lien avec les miennes.
Je remercie egalement Laurent Benoit et Claude Sensiau pour leur aide concernant la compr ehension de
ce quest un solveur fr equentiel des equations de Helmholtz, chose qui n etait pas clair pour moi au
d ebut de cette th` ese....
Je tiens ` a remercier Simon Mendez, El eonore Riber, Val erie Auffray et Gabriel Staffelbach pour leur
soutien, leur bonne humeur, les discussions parfois s erieuses et scientiques et celles qui navaient rien
` a voir avec le sujet de ce m emoire... Merci pour tout ceci et un peu plus!
Finalement, je tiens ` a remercier tous les doctorants, post-doctorants et personnels de recherche du
CERFACS ` a la fois pass e et pr esent pour lambiance chaleureuse, d econtract ee et professionnelle quils
apportent avec eux chaque jour au laboratoire.
REMERCIEMENTS
12
Nomenclature
Roman characters
H
0
j
Enthalpy change of reaction j
S
0
j
Entropy change of reaction j
K
f ,j
Forward reaction rate of reaction j
K
r,j
Reverse reaction rate of reaction j
Q
j
Rate progress of reaction j
J, q
k
Species Diffusive ux
m Momentum
q Heat ux
u Velocity vector
V
k
Species diffusion velocity
W Disturbance energy ux
A(n) LES code variable update
A
+
Downstream propagating acoustic wave
A
Upstream propagating acoustic wave

Al(n) LES code variable update linked with wall
law corrections
B(n) LES code variable update linked with
boundary conditions
c Sound velocity
c
pk
Species heat capacity at constant pressure
c
p
Heat capacity at constant pressure
c
vk
Species heat capacity at constant volume
c
v
Heat capacity at constant volume
D Disturbance energy source term
D
k
Species diffusion coefcient
E Efciency function (Part I)
E, e Total energy
e
k
Species energy
e
1
Disturbance energy in quadratic pressure
and velocity uctuations
E
d
General disturbance energy in ow
e
sk
Species sensible energy
e
s
Disturbance energy in quadratic entropy
uctuations
F Flame Transfer Function
F Thickening factor (Part I)
F
p
Flame Transfer Function relative to pres-
sure uctuations
F
u
Flame Transfer Function relative to veloc-
ity uctuations
F
norm2
Flame Transfer Function (Normalization
2)
F
norm
Flame Transfer Function (Normalization 1)
F
res
Flame Transfer Function for 1D analysis
g Free enthalpy
g
s
Free sensible enthalpy
LIST OF SYMBOLS
g
sk
Species free sensible enthalpy
H, h Enthalpy
h
k
Species enthalpy
h
sk
Species sensible enthalpy
h
m
sk
Molar sensible species mass enthalpy
n Amplitude of the Flame Transfer Function
p Pressure
Q
r
Radiative source term
r Gas constant of the mixture
s Entropy
s
k
Species entropy
s
m
k
Molar species entropy
T Temperature
W Mean molecular weight
W
k
Species molecular weight
X
k
Species molar fraction
Y
k
Species mass fraction
Greek characters
Filter characteristic size
Kronecker operator
0
L
Thickness of the premixed ame
x
Standard mesh size

T
/
T
Heat release
Specic heat ratio
Heat conduction coefcient
Molecular viscosity
t
Turbulent molecular viscosity
T
Volumic integrated heat release
k
Species source term
Dissipation function
Equivalence ratio
Mass density
k
Species mass density
Viscous tensor
Time delay of the Flame Transfer Function
(Part II)
,

Stress tensor
Wrinkling factor
Dimensionless numbers
R Universal gas constant
L
e
Lewis number
M Mach number
P
r
Prandtl number
P
t
r
Turbulent Prandtl number
R
e
Reynolds number
Re
t
Turbulent Reynold number
S Swirl number
S
c,k
Schmidt number
S
t
c,k
Turbulent Schmidt number
Operators
f
t
Turbulent component of the Reynolds l-
tered variable f
Df
Dt
Total derivative of scalar f
f Fourier transform of variable f

f Reynolds ltering operator (Part I)
14
List of symbols
f Time average of variable f
Partial differentiation operator
f Divergence of vector

f
f Gradient of scalar f
f Favre ltering operator

Subscripts
ij
ij component of the stress tensor
f
0
Zeroth order of the linear decomposition of
variable f
f
1
First order of the linear decomposition of
variable f
u
i
Component i of vector u
u
j
Component j of vector u
Superscripts
f Tensor notation of f
f
Subgrid scale component (Part I)

f
Time uctuation of variable f

Short names
AVI Articial VIscosity
BC Boundary Condition
CFD Computational Fluid Dynamics
DNS Direct Numerical Simulation
DTF Dynamically Thickened Flame
HF-FFT Harmonically Forced - Fast Fourier
Transform (post-processed)
LES Large Eddy Simulation
LW Lax-Wendroff
MPI Message Passing Interface
NSCBC Navier-Stokes Characteristic B.C.
PDF Probability Density Function
RANS Reynolds Averaged Navier-Stokes
SGS Sub-Grid Scale
TF Thickened Flame
TFLES Thickened Flame model for LES
TTGC Two-step Taylor-Galerkin Colin
WALE Wall Adapting Linear Eddy (model)
WN-FFT White Noise (forced) - Fast Fourier
Transform (post-processed)
WN-WH White Noise (forced) - Wiener-Hopf
(post-processed)
15
LIST OF SYMBOLS
16
Introduction
Bien quil apparaisse d esormais evident que les r eserves de combustibles fossiles s epuisent, leur
utilisation pour la production d electricit e et pour le transport reste largement majoritaire par rapport
aux formes renouvelables d energie. En effet, limmense majorit e des syst` emes de production des
pays d evelopp es ou en d eveloppement reposent actuellement sur la combustion et la transformation des
d eriv es organiques fossiles. Conscients des dangers de la d ependance energ etique qui en r esulte, de
nombreux chantiers de recherche se sont ouverts pour tenter de sen affranchir.
Concernant la production d electricit e de masse, le nucl eaire, malgr e les risques qui laccompagnent
1
apparat comme le seul candidat capable de soutenir le niveau de consommation actuel. Les r ecentes
d eclarations de lex ecutif am ericain
2
ainsi que celles de la Commission Europ eenne montrent le regain
dint er et pour ce mode de production d energie. Le lancement r ecent du projet ITER (reposant sur
la domestication de la fusion thermonucl eaire) fournit un espoir suppl ementaire de pouvoir disposer
un jour (le projet doit fournir des r esultats industriels ` a lhorizon 2040-2050) dune source d energie
quasi-in epuisable et dont les mati` eres premi` eres seraient mieux distribu ees.
Cependant, du fait de lorganisation economique actuelle, et malgr e laugmentation in evitable de
son co ut, lutilisation d energies fossiles reste pr epond erante dans bien des domaines. Parmi leurs
nombreuses utilisations, ces energies servent deux int er ets majeurs :
Leur combustion entre en jeu dans la production d electricit e par les centrales thermiques.
Elle fournit egalement un travail m ecanique utile servant aux transports, quils soient terrestres ou
a eriens.
R ecemment, la production d electricit e par turbine ` a gaz a connu un regain dint er et en lien avec la
d er egulation du march e de l energie. En effet, gr ace ` a laugmentation de son rendement (cf Fig.3) et ` a
sa grande exibilit e (temps de livraison, temps de maintenance, etc...) ce mode de production d energie
est devenu extr emement concurrentiel pour les petits producteurs d electricit e.
1
les principaux risques sont li es aux accidents nucl eaires en production (Tchernobil (URSS), Three Mile Island (USA),
Saint-Laurent (France), Tokaimura (Japon)....) et aux risques de contamination apr` es utilisation du combustible, ces mat eriaux
n ecessitant un stockage pouvant durer plusieurs dizaines de milliers dann ees.
2
Les Etats-Unis ont pourtant connu un accident nucl eaire majeur en 1979 (Three Mile Island).
INTRODUCTION
Figure 3 - Evolution des co uts de production en fonction de la quantit e d electricit e produite ` a laide des turbines
` a gaz. Chaque point de la courbe repr esente le meilleur compromis ` a une date donn ee. Depuis le d ebut des
ann ees 1990, il est devenu plus rentable de produire de l electricit e ` a laide de turbines de faible puissance.
Source: Charles E.Bayless, Less Is More: Why Gas Turbines Will Transform Electric utilities. Public Utilities
Fortnightly 1994
Bien que des energies alternatives existent concernant le transport routier (syst` emes hybrides essence-
electricit e) et que des syst` emes innovants voient le jour (pile ` a combustible, hydrog` ene, solaire, etc...),
la combustion dessence, de diesel ou de fuel na pas encore de concurrent economiquement s erieux.
Quels sont alors les points ` a am eliorer pour nous permettre de vivre au mieux avec ce mode
de production et de consommation d energie et den limiter les m efaits sur lenvironnement?
Une solution evidente est la limitation de la consommation de carburant de tous les types de chambres
de combustion en am eliorant leur rendement, permettant ainsi dutiliser au mieux les r eserves pr esentes
et de diminuer les emissions de dioxyde de carbone (CO
2
) (produit in evitable de la combustion de
compos es carbon es et gaz ` a effet de serre). Cependant, le CO
2
nest pas le seul polluant produit par
la combustion dhydrocarbures et de nombreuses etudes sattachent aussi ` a limiter la production dune
autre famille de polluants : les oxydes dazote (NO
x
).
Ces d eriv es de la combustion dhydrocarbures dans un milieu riche en azote (comme cest le cas dans
lair) sont ` a lorigine dun cycle de r eactions chimiques qui m` enent ` a la diminution de la concentration
en ozone(O
3
) dans les couches elev ees de latmosph` ere. Or lozone emp eche une partie du rayonnement
ultra-violet de se frayer un chemin jusqu` a la surface de la Terre. Lorsque la concentration en O
3
des hautes couches de latmosph` ere diminue, une plus grande partie ce rayonnement nous parvient et
entrane une augmentation de lincidence des cancers de la peau. Paradoxalement, au cours de ce m eme
cycle chimique, de lozone peut etre cr e e au niveau du sol menant ainsi ` a la formation de brouillards
dozone. Or lozone est toxique par inhalation, et ces brouillards dozone entranent cette fois des
risques respiratoires. La gure 4 r esume ce cycle simpli e de lozone.
Le monoxyde dazote (NO) est lun des repr esentants de la famille des NO
x
. La formation de NO
dans les chambres de combustion a lieu ` a travers deux principaux m ecanismes:
18
Introduction
Figure 4 - Cycle simpli e de lozone incluant linuence de NO originaires de la combustion. source: The
ESPERE Associaton (Environmental Science Published for Everybody Round the Earth)
la formation rapide de NO, due principalement ` a la r eaction entre le radical OH
et la mol ecule
N2 dans les ammes pr e-m elang ees.
la formation thermique du NO, d u ` a la r eaction entre la mol ecule de O
2
et celle de N
2
.
Ces deux m ecanismes de formation encouragent ` a diminuer la temp erature ` a laquelle a lieu la combustion
pour limiter les emissions de NO. Cependant, aux basses temp eratures se forme alors un autre polluant:
le monoxyde de carbone (CO). Comme le montre la gure 5, il existe une temp erature optimale de
combustion ` a laquelle on parvient ` a minimiser ` a la fois la concentration en CO et en NO dans les gaz
br ul es.
Ces temp eratures optimales de fonctionnement etant plus faibles que les temp eratures st-
chiom etriques
3
de combustion, les industriels se sont orient es progressivement vers des technologies de
combustion ` a richesse
4
basse et pr e-m elang ees, ainsi quau d eveloppement de chambres de combustion
etag ees permettant une matrise locale de la richesse du m elange. Cependant, ` a des richesses faibles
(la richesse moyenne de combustion est parfois plus faible que la limite dextinction de la amme), la
combustion est beaucoup plus sensible aux ph enom` enes dinstabilit es. Dans ce cas, des etudes r ecentes
ont montr e [126] que la quantit e de polluants emis pouvait augmenter.
Il convient alors de mieux comprendre les instabilit es de combustion an de permettre lutilisation
industrielle dune combustion plus pauvre et donc plus propre.
3
Temp erature atteinte lorsque la combustion ` a lieu ` a une richesse unitaire
4
La richesse dun m elange caract erise le rapport entre la quantit e de carburant et celle de comburant disponible. Elle
augmente avec la concentration de carburant.
19
INTRODUCTION
1500 1600 1700 1800 1900 2000 2100 K
Temprature de la zone primaire
I
n
d
i
c
e
s

d
m
i
s
s
i
o
n
NOx CO
Normes
Europennes
Plage de tempratures
faibles missions
Figure 5 - Graphique montrant l evolution des emissions de NO et CO en fonction de la temp erature de
combustion.
20
Introduction
Instabilit es de combustion
L etude de la minimisation des concentrations en polluants dans les gaz br ul es est un aspect ma-
jeur de la recherche actuelle en combustion. Cependant, celle-ci se heurte depuis de nombreuses
ann ees au ph enom` ene des instabilit es de combustion. Comme on la vu pr ec edemment, les br uleurs
actuels fonctionnant ` a des richesses globales faibles sont plus sensibles aux instabilit es de combus-
tion [27, 51, 76, 97, 109]. Dans ce cas, ils sont alors plus susceptibles de g en erer du bruit li e ` a la
uctuation p eriodique du d egagement de chaleur. Par ailleurs, lorsquune chambre de combustion est
soumise ` a une instabilit e, le m elange est moins bon et il se cr ee alors des zones plus riches que la
moyenne et qui vont donc produire plus de NO [7, 36, 63, 126, 148]. La pr evision des instabilit es
de combustion et lam elioration des mod` eles dinteraction entre la amme et les divers ph enom` enes
instationnaires de l ecoulement sont donc actuellement deux points cl es de la recherche en combustion.
Sous le terme dinstabilit es de combustion se cachent de nombreux ph enom` enes physiques quil convient
de d etailler ici. Les instabilit es li ees ` a la combustion peuvent etre s epar ees en deux grandes familles
suivant quelles ne font intervenir que la amme, ou bien quelles n ecessitent un connement de la
zone de combustion pour prendre naissance. Les premi` eres sont nomm ees instabilit es intrins` eques et les
secondes, instabilit es extrins` eques de amme.
Instabilit es intrins` eques
Dans la premi` ere cat egorie sont recens es tous les types dinstabilit es ne faisant intervenir que la amme,
sans consid erer son eventuel connement. Les instabilit es intrins` eques de combustion ont comme orig-
ine,
soit un terme source de type Rayleigh-Taylor, cest ` a dire reposant sur la diff erence de densit e
entre les gaz frais et br ul es,
soit un terme source li e ` a linstabilit e de Darrieus-Landau ou Hydrodynamique reposant sur
la d eviation des lignes de courant en amont et en aval du front de amme et menant ` a la courbure
du front de amme,
soit un terme source menant ` a linstabilit e de type Thermo-diffusive reposant sur la diff erence
entre les coefcients de diffusion desp` eces et de chaleur.
Clavin et Garcia [26] fournissent une analyse compl` ete des instabilit es intrins` eques dans les ammes
planes pr em elang ees. Ces travaux fournissent une relation de dispersion th eorique qui montre que pour
la plupart des vitesses de amme laminaire, le taux de croissance de linstabilit e de Darrieus-Landau est
positif sur une large bande de fr equence (gure 6). Clanet et al. en 1998 [25] ont effectu e la premi` ere
etude exp erimentale de linstabilit e de Darrieus-Landau, et ont montr e la validit e du mod` ele th eorique
de Clavin et Garcia pour des vitesses de amme inf erieures ` a 20cm/s et pour une longueur donde
dinstabilit e x ee. La gure 7 montre un instantan e du front de amme perturb e par une instabilit e
de Darrieus-Landau. Le ph enom` ene est mono-fr equentiel comme en t emoigne la p eriodicit e spatiale des
21
INTRODUCTION
gures li ees ` a linstabilit e. Il est ` a noter que linstabilit e de Darrieus-Landau est actuellement vue comme
lun des m ecanismes possibles de la transition d eagration-d etonation
5
[72] en combustion, ce qui mon-
tre que ce type dinstabilit e peut avoir une importante inuence sur l evolution du front de amme m eme
lorsque celle-ci est largement turbulente.
Figure 6 - Taux de croissance r eduit en fonction du nombre donde r eduit de la perturbation. Source [25]
Figure 7 - Front de amme soumis ` a linstabilit e de Darrieus-Landau. Source [119]
5
La transition d eagration-d etonation se produit lorsquune onde de choc traverse un front de amme se propageant ` a une
vitesse faible. Il se produit alors une transition apr` es laquelle le front de amme suit londe de choc ` a une vitesse supersonique.
22
Introduction
Instabilit es extrins` eques
La deuxi` eme cat egorie concerne les instabilit es pour lesquelles le connement est n ecessaire. La mani` ere
globale dont ces instabilit es se d eveloppent est r esum e sur la gure 8. La amme est l el ement cen-
tral de la boucle dinstabilit e, ampliant et transformant les diff erents types dondes pr esentes dans
l ecoulement. Les instabilit es en milieu conn e peuvent donc etre ordonn ees suivant le type donde
excit ee dans l ecoulement.
A lorigine de linstabilit e, une faible uctuation de l ecoulement amont vient perturber le front de
amme (gure 8-zone2). La amme amplie cette perturbation et cr ee principalement deux types
dondes se propageant dans l ecoulement (gure 8-zone3):
une onde acoustique se propageant ` a la vitesse u + c (u etant la vitesse de l ecoulement et c celle
du son),
une onde entropique se propageant ` a la vitesse u de l ecoulement.
Ces ondes rencontrent les murs et la sortie de la chambre de combustion (gure 8-zone4). Londe
acoustique est en partie r e echie, donnant naissance ` a une onde acoustique remontant l ecoulement ` a
la vitesse u c. Londe entropique cr e ee par la amme peut aussi donner naissance ` a une onde acous-
tique remontant l ecoulement.
6
Londe acoustique se propageant vers lamont peut interagir directement
avec la amme et ainsi exciter une nouvelle onde entropique et/ou acoustique. Elle peut aussi remonter
l ecoulement jusquau niveau de linjection. Au niveau de linjection, linteraction entre lacoustique
et l ecoulement est complexe (gure 8-zone1). De nombreux travaux de recherche traitent ce probl` eme
sp ecique [132] et de mani` ere g en erale, on observe une r eexion de l energie acoustique remontant
l ecoulement sous forme dune onde acoustique, dune onde entropique ou dune onde de vorticit e (cest
` a dire une structure coh erente se propageant ` a sa vitesse de groupe ). Suivant les diff erents types de
uctuations perturbant le front de amme (gure 8-zone2), la boucle dinstabilit e peut etre ferm ee de
diff erentes mani` eres.
Il existe un transfert d energie possible entre les uctuations dentropie et les uctuations acous-
tiques. Cest, par exemple, le cas dune amme subissant une uctuation de richesse p eriodique.
Celle-ci donne lieu ` a une uctuation de d egagement de chaleur cr eant ` a la fois une onde acoustique
et une onde entropique se propageant dans le sens de l ecoulement [140].
La amme peut aussi transformer de l energie de vorticit e en energie acoustique. Ce ph enom` ene a
lieu lorsque par exemple une structure coh erente plisse le front de amme et cr ee une variation lo-
cale du d egagement de chaleur. Comme dans lexemple pr ec edent, cette uctuation du d egagement
de chaleur alimente l energie acoustique derri` ere le front de amme [47].
Il a et e observ e une re-laminarisation dans les gaz br ul es pour certains ecoulements dinjection
faiblement turbulent. Ce comportement vient de lexpansion des gaz br ul es qui a pour effet
d ecarter les lignes de courant et donc de diminuer le niveau de vorticit e. Ceci explique pourquoi
6
De mani` ere g en erale, Kovasnay [64] montre que hors du domaine lin eaire, il existe des echanges entre les energies des
perturbations li ees aux uctuations acoustiques, turbulentes et entropiques
23
INTRODUCTION
la amme tend ` a att enuer la vorticit e qui la traverse. (` eches pointill ees gure 8-zone3).
Figure 8 - Sch ema densemble de la boucle dinstabilit e de combustion dans les milieux conn es.
Les m ecanismes de fermeture de linstabilit e font intervenir diff erents types dondes se propageant ` a
des vitesses diff erentes. Le temps n ecessaire ` a linformation pour parcourir le cycle dinstabilit e nest
donc pas le m eme suivant le type dinstabilit e. Ceci est lune des raisons pour lesquelles diff erentes
fr equences dinstabilit es sont observ ees dans les chambre de combustion. En effet, la fr equence de
linstabilit e ne peut etre inf erieure ` a la fr equence li ee au temps de bouclage.
La gure 9 pr esente une chambre de combustion simpli ee. La amme est situ ee ` a 10 cm de linjection
dans une chambre de 50 cm de long. Le tableau 1 fournit pour cette conguration les vitesses et temps
de propagation des diff erents types dondes pouvant etre impliqu ees dans le ph enom` ene dinstabilit e.
Le tableau 2 r esume (gr ace ` a ces temps de propagation) les fr equences minimales possibles suivant le
type dinteraction. Il est ` a noter que lanalyse ne comprend que les uctuations colin eaires ` a laxe de
la chambre de combustion. On observe que les instabilit es purement acoustiques sont principalement
responsables des uctuations ` a plus haute-fr equence. Tandis que les fr equences interm ediaires (entre 100
et 1000 Hz) sont li ees ` a un m ecanisme faisant intervenir un couplage entre lacoustique et l ecoulement,
via une transmission de la perturbation sous forme de uctuation entropique ou sous forme dune
propagation de vorticit e (par exemple une structure coh erente). Dans son livre Combustion Theory,
Forman A. Williams [154] prend lexemple des instabilit es observ ees dans les chambres de combustion
cryotechniques. Dans ces chambres, deux familles principales dinstabilit es ont et e observ ees :
le screaming est un mode dinstabilit e dont la fr equence est sup erieure ` a 1000 Hz.
le chugging dont la caract eristique est d etre basse fr equence (inf erieure ` a 100 Hz), est princi-
palement d u au couplage entre la chambre de combustion et le syst` eme dinjection.
24
Introduction
Cette diff erentiation en deux familles distinctes (chugging et screaming) correspond bien ` a des
m ecanismes sous-jacents dexcitation diff erents comme on peut le voir dans le tableau 2. Ainsi,
dans lexemple choisi, une instabilit e purement acoustique ne peut se d evelopper pour des fr equences
inf erieures ` a 675Hz. Pour expliquer la pr esence dinstabilit es de plus basse fr equence, on doit consid erer
une interaction entre lacoustique et l ecoulement. On observe (Tableau 2) que lorsque lon invoque une
structure coh erente dans la premi` ere partie de la boucle dinstabilit e, la fr equence minimale pouvant etre
excit ee diminue pour atteindre 191 Hz.
Figure 9 - Chambre de combustion sch ematis ee.
zone type de donde vitesse de propagation (m.s-1) temps de propagation (ms)
1 acoustique (amont) 400 0.25
1 acoustique remontante (amont) 300 0.33
1 entropique (amont) 50 2
1 vorticit e (amont) 25-50 (vitesse de groupe de la structure) 2-4
2 acoustique (aval) 1000 0.4
2 entropique (aval) 100 4
2 acoustique remontante (aval) 800 0.5
Table 1 - Temps caract eristiques n ecessaires au diff erents types dondes pour parcourir les zones d enies sur la
gure 9
Outils danalyse num erique des instabilit es de combustion
Deux principales m ethodes num eriques danalyse peuvent etre appliqu ees pour connatre la sensibilit e
dune chambre de combustion aux instabilit es thermo-acoustiques : les codes dinstabilit e et les crit` eres
25
INTRODUCTION
type dinteraction temps total de la boucle dinstabilit e (ms) fr equence minimale dexcitation (Hz)
1A-2A-RA 1.48 675
1A-2E-RA 5.08 197
1E-2A-RA 3.23 309
1E-2E-RA 6.83 146
1V-2A-RA 3.23-5.08 191-309
1V-2E-RA 6.83-8.83 113-146
Table 2 - Fr equences minimales dexistence des instabilit es suivant le type dinteraction. A:acoustique,
E:entropique, V:vorticit e, R:retour, 1,2:zones d enies sur la gure 9
dinstabilit e.
Codes dinstabilit e
La premi` ere est quantitative et cherche ` a pr evoir lamplitude ainsi que les fr equences des insta-
bilit es pouvant se d evelopper dans la chambre de combustion. Il sagit doutils de r esolution
des equations r egissant le comportement de l ecoulement perturb e. Lacoustique jouant un r ole
pr epond erant, il nest pas surprenant que la plupart de ces outils ressemble ` a des codes dacoustique
dans les domaines ferm es o` u loutil d evelopp e a pour but de donner les fr equences et les taux de
croissance des modes instables. Au sein de ce type danalyse, on peut distinguer trois grandes
m ethodes de r esolution :
Les m ethodes d el ements en r eseaux [8, 43, 44, 78, 108, 112, 145].
Ces m ethodes r esolvent les amplitudes et fr equences des modes acoustiques longitudinaux
dans chacun des el ements. Leur avantage est de permettre une etude param etrique de la sta-
bilit e prenant en compte des imp edances complexes aux fronti` eres du domaine. Cependant,
cet outil requiert de connatre au pr ealable la forme des modes intervenant dans lanalyse et
ne convient pas lorsque la g eom etrie est complexe ou la fr equence des modes elev ee.
Les m ethodes directes spatio-temporelles - Simulation aux Grandes Echelles (SGE) [3, 19,
28, 37, 57, 87, 102, 101, 138]
La SGE fournit l evolution spatio-temporelle des variables de l ecoulement. Elle prend donc
en compte les interactions non lin eaires pouvant se produire au sein du uide. Cette m ethode
demande une importante puissance de calcul et ne donne acc` es quau mode acoustique le
plus ampli e. Par ailleurs, lutilisation de conditions aux limites avec imp edances complexes
peut se r ev eler difcile.
Les m ethodes de r esolution dans lespace des fr equences. [33, 68, 83, 90, 128, 129, 136]
Ces m ethodes fournissent les fr equences ainsi que les amplitudes complexes des modes
acoustiques en lin earisant les equations de Navier-Stokes r eactives. Le principal avantage
de ces m ethodes est de pouvoir prendre en compte des conditions aux limites faisant inter-
venir des imp edances complexes. Mais lanalyse de stabilit e est lin eaire dans ce cas et la
r esolution du probl` eme dans le domaine non lin eaire peut sav erer complexe. Un el ement
26
Introduction
essentiel de ces approches est le besoin de mod` eles d ecrivant la r eponse de la amme (c.` a.d.
le taux de r eaction) ` a des perturbations acoustiques de la chambre. Cet el ement appel e FTF
(Fonction de transfert de amme) est la difcult e principale dans ce type dapproche.
Crit` eres dinstabilit e
Le second type danalyse est qualitatif. Il sagit de d eterminer si un crit` ere simple, bas e sur le
niveau de uctuation des variables primitives (pression, d egagement de chaleur, temp erature, etc...)
peut permettre de pr evoir la pr esence dune instabilit e thermo-acoustique.
Le premier crit` ere de ce type a et e d evelopp e par Rayleigh en 1878 [120] et pr evoit quune insta-
bilit e thermo-acoustique peut se d evelopper si la uctuation de pression est en phase avec celle du
d egagement de chaleur. Depuis, de nombreux travaux se sont attach e ` a etendre ce crit` ere qui ne per-
met pas toujours de d etecter la pr esence de linstabilit e. Deux directions principales ont et e suivies.
La premi` ere consiste ` a etendre la notion d energie acoustique ` a des cas dans lesquels l ecoulement
moyen ne peut etre n eglig e [16, 18, 85]. La seconde cherche ` a d eterminer une equation de conser-
vation non lin eaire pour l energie des uctuations dans les ecoulements [23, 88].
Ces deux aspects, quantitatif et qualitatif, de l etude des instabilit es de combustion sont abord es
dans cette th` ese. Ces travaux sinscrivent donc dans la continuit e d etudes pr ec edentes [8, 82, 135,
139, 143, 149] men ees au CERFACS et utilisant loutil num erique instationnaire (Simulation aux
Grandes Echelles-Simulation Num erique Directe) comme :
un moyen de caract eriser la Fonction de Transfert de Flamme (r eponse de la amme ` a une
perturbation de l ecoulement). Cette information permet dacc eder aux taux de croissance
des modes acoustiques en utilisant les m ethodes de r esolution dans lespace des fr equences
pr ec edemment d ecrites.
un moyen de r ealiser des exp eriences num eriques mettant en evidence les principaux termes
source des instabilit es thermo-acoustiques et ainsi fournir un crit` ere simple permettant de
d etecter ces instabilit es dans les ecoulements r eactifs.
27
OBJECTIVES & ORGANISATION
28
Objectives & Organisation
Objectives
This work presents a numerical study (using both Large-Eddy Simulation (LES) and Direct Numerical
Simulation (DNS)) of two main aspects of combustion instabilities. It is part of the undergoing studies
that tend to address this issue for more than fty years. The nal goal of this long-lasting research effort
is the understanding and the prediction of thermo-acoustic instabilities in combustion. The objectives of
this thesis are:
the extension of existing and the development of new numerical post-processing tools for the study
of Flame Transfer Functions with the help of LES. As far as these transfer functions are concerned,
only global or 1Dresponses are currently used in acoustic codes, assuming a compact reaction zone
compared to the acoustic wavelength. Yet, the ame is not always compact and distributed reaction
zones should also be considered. There is therefore a need for a tool that could determine the local
response of the ame to the incoming perturbations (i.e. the correlation between the inlet velocity
uctuation and the local heat release uctuation). This tool would help to understand the inuence
of the geometry and the injection system on the global response of the ame. This would also
be a step towards the creation of an acoustic tool able to determine the linear stability of reacting
systems.
the derivation of new stability criteria in gaseous reacting ows. Recently, Nicoud and Poinsot [93]
have pointed out the possibility that the classical Rayleigh criterion might not be relevant for all
reacting ows since it does not take into account the disturbance energy created by the uctuation
of entropy. Following Chu [23], they derive a linear criterion based on the correlation between
temperature and heat release uctuations. The rst objective here is therefore to test these linear
criteria on a 2D laminar unstable ame using Direct Numerical Simulations (DNS).
Exact conservation equations for disturbance energies should also be derived and their balance
closed to test the assumptions usually made when deriving stability criteria in reacting ows. Since
the combustion phenomenon creates high temperature uctuations, this work also aims at deriving
a nonlinear stability criterion that could be used when no linearization of the ow is possible.
Organisation
First, Part I provides a description of the numerical tool AVBP used during this work. Then, following
the previously described objectives, the document is split in two parts.
Part II presents a study of Flame Transfer Function (FTF) measurements using LES.
Chapter 2 introduces this issue in its historical background and presents a short review of FTF
linear models in reacting ows.
Chapter 3 focuses on FTF measurement techniques using LES. In particular, one describes a procedure
involving the response of the ame to a ltered-white noise and using the inversion of the Wiener-Hopf
relation to get the FTF.
Chapter 4 presents the results obtained using different methods for FTF measurements in four congu-
rations.
First, the main aspects of FTF are obtained in a 1D laminar ame.
Then, a comparison of three methods of FTF is given in a laminar conical ame.
In a third section, a local comparison of two largely used methods is done in a 3D-turbulent forced
conguration. This section also provides a comparison of global time delays given by these meth-
ods with experimental results.
Finally, a FTF method using ltered-white noise local response is compared to the classical har-
monic method in a 3D-turbulent forced conguration.
Part III presents a theoretical and numerical study of disturbance energies in combustion.
Chapter 5 presents a short sample of the undergoing research effort towards the prediction of thermo-
acoustic instabilities in reacting ows. This chapter also introduces the numerical post-processing tool
developed here to investigate budgets of disturbance energies in ows.
Chapter 6 focuses on the derivation of three disturbance energies.
The rst two conservation equations are linear and respectively refer to Pressure-Velocity (PV)
and Entropy disturbance energies.
The third equation is nonlinear and aims at giving the levels of disturbance energy in nonlinear
cases.
Chapter 7 focuses on the analysis of budgets of these disturbance energies in two congurations.
The rst conguration is a 1D ame and gives access to the main aspects of the unsteady evolution
of disturbance energies.
30
Objectives & Organisation
The second conguration is a 2D laminar ame. This case is harmonically forced and also made
unstable by increasing the outlet reection coefcient. Results obtained in both cases are com-
pared. The inuence of the choice of the reference eld is also investigated. A procedure for
balance analysis in turbulent cases using LES is proposed.
Chapter 8 presents a validation of Rayleigh and Chu linear stability criteria in an unstable 2D laminar
ame. These criteria fail to predict the instability and two new criteria are proposed which succeed.
The rst one extends the Rayleigh linear stability criterion and the second one aims at detecting thermo-
acoustic instabilities when no linearization of the ow is possible.
31
32
Part I
Description of the numerical tool
Chapter 1
Description of the numerical tool AVBP
This chapter focuses on some aspects of the numerical tool used for all simulations during this PhD.
AVBP solver is a modern software tool for Computational Fluid Dynamics (CFD) of high exibility,
efciency and modularity. It is an unstructured solver capable of handling grids of any cells type. In
this work, both structured and unstructured grids have been used. AVBP solves the unsteady laminar
and turbulent compressible Navier-Stokes equations in two and three space dimensions using numerical
method which are second or third order in space and time.
In the following sections, attention will be paid on to two important aspects of the code, the equations
and the main underlying hypothesis used for Direct Numerical Simulations (DNS) and the models used
for Large Eddy Simulations (LES). Some information concerning boundary conditions and the imple-
mentation of articial viscosity in AVBP may also be found.
Finally, some details on the cell-vertex discretization are also provided. This section gives the main dif-
ferences between a central difference and the Lax-Wendroff scheme that will be used in the third Part
of this work.
1.1 Governing Equations and hypothesis for DNS
In this section the compressible Navier-Stokes equations are described (as found in fundamental CFD
text books such as [2, 53] ) . This chapter focuses on unltered equations used in Direct Numerical
Simultation. The LES extensions will be described on page 43.
1.1.1 The governing equations
Throughout this document, the index notation is adopted for the description of the governing equations.
Summation rule is henceforth implied over repeated indices (Einsteins rule of summation). Note how-
ever that index k is reserved to refer to the k
th
species and will not follow the summation rule unles
specically mentioned or implied by the
sign.
DESCRIPTION OF THE NUMERICAL TOOL AVBP
The set of conservation equations describing the evolution of a compressible ow with chemical
reactions of thermodynamically active scalars reads,
u
i
t
+

x
j
( u
i
u
j
) =

x
j
[P
ij
ij
], (1.1)
E
t
+

x
j
( E u
j
) =

x
j
[u
i
(P
ij
ij
) +q
j
] +
T
+Q
r
, (1.2)
k
t
+

x
j
(
k
u
j
) =

x
j
[J
j,k
] +
k
. (1.3)
In Eqs 1.1-1.3 respectively corresponding to the conservation laws for momentum, total energy and
species, the following symbols denote respecticely , u
i
, E,
k
, density, the velocity vector, the total
energy per unit mass and
k
= Y
k
for k = 1 to N ( N is the total number of species). The source term
in the total energy equation, Eq. 1.2s, is decomposed for convenience into a chemical source term and
a radiative source term such that: S =
T
+ Q
r
. Corresponding source terms in the species transport
equations, Eq. 1.3, are noted,
k
.
It is usual to decompose the ux tensor into an inviscid and a viscous component. They are respec-
tively noted for the three conservation equations:
Inviscid terms:
_
_
u
i
u
j
+P
ij
(E +P
ij
) u
j
k
u
j
_
_
(1.4)
where the hydrostatic pressure P is given by the equation of state for a perfect gas (Eq. 1.12).
Viscous terms:
The components of the viscous ux tensor take the form:
_
_
ij
(u
i

ij
) +q
j
J
j,k
_
_
(1.5)
J
j,k
is the diffusive ux of species k and is presented in subsection on page 38 (Eq. 1.23). The stress
tensor
ij
is explicited in subsection on page 39 (Eq. 1.24). Finally, subsection on page 39 is devoted to
the heat ux vector q
j
(Eq. 1.27).
1.1.2 Thermodynamical variables
The standard reference state used is P
0
= 1 bar and T
0
= 0 K. The sensible mass enthalpies (h
s,k
) and
entropies (s
k
) for each species are tabulated for 51 values of the temperature (T
i
with i = 1...51) ranging
from 0K to 5000K with a step of 100K. Therefore these variables can be evaluated by:
36
Governing Equations and hypothesis for DNS
h
s,k
(T
i
) =
_
T
i
T
0
=0K
C
p,k
dT =
h
m
s,k
(T
i
) h
m
s,k
(T
0
)
W
k
, and (1.6)
s
k
(T
i
) =
s
m
k
(T
i
) s
m
k
(T
0
)
W
k
, with i = 1, 51 (1.7)
The superscript m corresponds to molar values. The tabulated values for h
s,k
(T
i
) and s
k
(T
i
) can
be found in the JANAF tables [146]. With this assumption, the sensible energy for each species can be
reconstructed using the following expression :
e
s,k
(T
i
) =
_
T
i
T
0
=0K
C
v,k
dT = h
s,k
(T
i
) r
k
T
i
i = 1, 51 (1.8)
Note that the mass heat capacities at constant pressure c
p,k
and volume c
v,k
are supposed constant
between T
i
and T
i+1
= T
i
+ 100. They are dened as the slope of the sensible enthalpy (C
p,k
=
h
s,k
T
) and sensible energy (C
v,k
=
e
s,k
T
). The sensible energy henceforth varies continuously with the
temperature and is obtained by using a linear interpolation:
e
s,k
(T) = e
s,k
(T
i
) + (T T
i
)
e
s,k
(T
i+1
) e
s,k
(T
i
)
T
i+1
T
i
(1.9)
The sensible energy and enthalpy of the mixture may then be expressed as:
e
s
=
N
k=1
k
e
s,k
=
N
k=1
Y
k
e
s,k
(1.10)
h
s
=
N
k=1
k
h
s,k
=
N
k=1
Y
k
h
s,k
(1.11)
1.1.3 The equation of state
The equation of state for an ideal gas mixture writes:
P = r T (1.12)
where r is the gas constant of the mixture dependant on time and space: r =
R
W
where W is the mean
molecular weight of the mixture:
1
W
=
N
k=1
Y
k
W
k
(1.13)
37
The gas constant r and the heat capacities of the gas mixture depend on the local gas composition as:
r =
R
W
=
N
k=1
Y
k
W
k
1 =
N
k=1
Y
k
r
k
(1.14)
C
p
=
N
k=1
Y
k
C
p,k
(1.15)
C
v
=
N
k=1
Y
k
C
v,k
(1.16)
where 1= 8.3143 J/mol.K is the universal gas constant. The adiabatic exponent for the mixture is given
by = C
p
/C
v
. Thus, the gas constant, the heat capacities and the adiabatic exponent are no longer
constant. Indeed, they depend on the local gas composition as expressed by the local mass fractions
Y
k
(x, t):
r = r(x, t), C
p
= C
p
(x, t), C
v
= C
v
(x, t), and = (x, t) (1.17)
The temperature is deduced from the the sensible energy, using Eqs. 1.9 and 1.10. Finally boundary
conditions make use of the speed of sound of the mixture c which is given by:
c
2
= r T (1.18)
1.1.4 Conservation of Mass: Species diffusion ux
In multi-species ows the total mass conservation implies that:
N
k=1
Y
k
V
k
i
= 0 (1.19)
where V
k
i
are the components in directions (i=1,2,3) of the diffusion velocity of species k. They are
often expressed as a function of the species gradients using the Hirschfelder Curtis approximation:
X
k
V
k
i
= D
k
X
k
x
i
, (1.20)
where X
k
is the molar fraction of species k : X
k
= Y
k
W/W
k
. In terms of mass fraction, the approxi-
mation 1.20 may be expressed as:
Y
k
V
k
i
= D
k
W
k
W
X
k
x
i
, (1.21)
38
Summing Eq. 1.21 over all ks shows that the approximation 1.21 does not necessarily comply with
equation 1.19 that expresses mass conservation. In order to achieve this, a correction diffusion velocity
V
c
is added to the convection velocity to ensure global mass conservation (see [107]) as:
V
c
i
=
N
k=1
D
k
W
k
W
X
k
x
i
, (1.22)
and computing the diffusive species ux for each species k as:
J
i,k
=
_
D
k
W
k
W
X
k
x
i
Y
k
V
c
i
_
, (1.23)
Here, D
k
are the diffusion coefcients for each species k in the mixture (see subsection on page 40);
J
i,k
. Using Eq. 1.23 to determine the diffusive species ux implicitly veries Eq. 1.19.
1.1.5 Viscous stress tensor
The stress tensor
ij
is given by the following relations:
ij
= 2(S
ij
1
3
ij
S
ll
), (1.24)
and
S
ij
=
1
2
(
u
j
x
i
+
u
i
x
j
), (1.25)
Eq. 1.24 may also be written:
xx
=
2
3
(2
u
x

v
y

w
z
),
xy
= (
u
y
+
v
x
)
yy
=
2
3
(2
v
y

u
x

w
z
),
xz
= (
u
z
+
w
x
)
zz
=
2
3
(2
z
w

u
x

v
y
),
yz
= (
v
z
+
w
y
)
(1.26)
where is the shear viscosity (see subsection on page 40).
1.1.6 Heat ux vector
For multi-species ows, an additional heat ux term appears in the diffusive heat ux. This term is due
to heat transport by species diffusion. The total heat ux vector then writes:
q
i
=
T
x
i
. .
Heat conduction
k=1
_
D
k
W
k
W
X
k
x
i
Y
k
V
c
i
_
h
s,k
. .
Heat ux through species diffusion
=
T
x
i
+
N
k=1
J
i,k
h
s,k
, (1.27)
where is the heat conduction coefcient of the mixture (see subsection on page 40).
39
1.1.7 Transport coefcients
In CFD codes for multi-species ows the molecular viscosity is often assumed to be independent of
the gas composition and close to that of air
1
so that the classical Sutherland law can be used. In a rst
step, the same assumption for the multi-gas AVBP solver is made, yielding:
= c
1
T
3/2
T +c
2
T
ref
+c
2
T
3/2
ref
(1.28)
where c
1
and c
2
must be determined so as to t the real viscosity of the mixture. For air at T
ref
= 273 K,
c
1
= 1.71e-5 kg/m.s and c
2
= 110.4 K (see [153]). These values are given by the user. A second law is
available, called Power law:
= c
1
(
T
T
ref
)
b
(1.29)
with b typically ranging between 0.5 and 1.0. For example b = 0.76 for air.
The heat conduction coefcient of the gas mixture can then be computed by introducing the molecular
Prandtl number of the mixture as:
=
C
p
P
r
(1.30)
with P
r
supposed as constant in time and space and is given by the user. The computation of the species
diffusion coefcients D
k
is a specic issue. These coefcients should be expressed as a function of the
binary coefcients D
ij
obtained from kinetic theory (Hirschfelder et al. [54]). The mixture diffusion
coefcient for species k, D
k
, is computed as (Bird et al. [11]):
D
k
=
1 Y
k
N
j=k
X
j
/D
jk
(1.31)
The D
ij
are complex functions of collision integrals and thermodynamic variables. For a DNS code
using complex chemistry, using Eq. 1.31 makes sense. However in most cases, DNS uses a simplied
chemical scheme and modeling diffusivity in a precise way is not needed so that this approach is much
less attractive. Therefore a simplied approximation is used in AVBP for D
k
. The Schmidt numbers
S
c,k
of the species are supposed to be constant so that the binary diffusion coefcient for each species is
computed as:
D
k
=

S
c,k
(1.32)
Note that the Schmidt number for each species k is assumed to be constant in time and space and is
given by the user. P
r
and S
c,k
model the laminar (thermal and molecular) diffusion. Usual values of
Schmidt and Prandtl numbers for premixed ames are those given by PREMIX in the burnt gas.
1
This introduces errors that are less important than those related to the thermodynamic properties.
40
1.1.8 Kinetics
The source term on the right hand side of Eqs. 1.2 & 1.3 respectively writes:
_

T

k
_
where
T
is the rate of heat release and
k
the reaction rate of species k. Source terms in the momentum
equation, Eq. 1.1, may also appear and are in this case used to impose momentum components, for
example an imposed pressure gradients in periodic ows. In most cases however, they are null.
The combustion model of AVBP is an Arrhenius law written for N reactants /
k
and for M reactions
as:
N
k=1
kj
/
kj
k=1
kj
/
kj
, j = 1, M (1.33)
The reaction rate of species k (
k
) is the sum of rates
kj
produced by all M reactions:

k
=
M
j=1

kj
= W
k
M
j=1
kj
Q
j
(1.34)
where
kj
=
kj
kj
and Q
j
is the rate progress of reaction j and is written:
Q
j
= K
f,j
N
k=1
(
Y
k
W
k
)
kj
K
r,j
N
k=1
(
Y
k
W
k
)
kj
(1.35)
K
f,j
and K
r,j
are the forward and reverse rates of reaction j:
K
f,j
= A
f,j
exp(
E
a,j
1T
) (1.36)
where A
f,j
and E
a,j
are the pre-exponential factor and the activation energy given in the input le. K
r,j
is deduced from the equilibrium assumption:
K
r,j
=
K
f,j
K
eq
(1.37)
where K
eq
is the equilibrium constant dened by Kuo [70]:
K
eq
=
_
p
0
1T
_
P
N
k=1

kj
exp
_
S
0
j
1

H
0
j
1T
_
(1.38)
where p
0
= 1 bar. H
0
j
and S
0
j
are respectively the enthalpy (sensible + chemical) and the entropy
changes for reaction j:
H
0
j
= h
j
(T) h
j
(0) =
N
k=1
kj
W
k
(h
s,k
(T) + h
0
f,k
) (1.39)
41
S
0
j
=
N
k=1
kj
W
k
s
k
(T) (1.40)
where h
0
f,k
is the mass enthalpy of formation of species k at temperature T
0
= 0 K and is given in the
input le (molar value).
The heat release is calculated as:

T
=
N
k=1

k
h
0
f,k
(1.41)
42
1.2 Governing equations for LES
1.2 Governing equations for LES
1.2.1 The LES Concept
Large Eddy Simulation (LES) [122, 116] is nowadays recognized as an intermediate approach in com-
parisons to the more classical Reynolds Averaged Navier-Stokes (RANS) methodologies. Although
conceptually very different these two approaches aim at providing new systems of governing equations
to mimic the characteristics of turbulent ows.
The derivation of the new governing equations is obtained by introducing operators to be applied to
the set of compressible Navier-Stokes equations. Unclosed terms arise from these manipulations and
models need to be supplied for the problem to be solved. The major differences between RANS and
LES come from the operator employed in the derivation. In RANS the operation consists of a temporal
or ensemble average over a set of realizations of the studied ow [116, 22]. The unclosed terms are
representative of the physics taking place over the entire range of frequencies present in the ensemble
of realizations under consideration. In LES, the operator is a spatially localized time independent lter
of given size, , to be applied to a single realization of the studied ow. Resulting from this spatial
average is a separation between the large (greater than the lter size) and small (smaller than the lter
size) scales. The unclosed terms are in LES representative of the physics associated with the small
structures (with high frequencies) present in the ow. Figure 1.1 illustrates the conceptual differences
between (a) RANS and (b) LES when applied to a homogeneous isotropic turbulent eld.
(a) (b)
Figure 1.1 - Conceptual representation of (a) RANS and (b) LES applied to a homogeneous isotropic turbulent
eld.
Due to the ltering approach, LES allows a dynamic representation of the large scale motions whose
contributions are critical in complex geometries. The LES predictions of complex turbulent ows are
henceforth closer to the physics since large scale phenomena such as large vortex shedding and acoustic
waves are embedded in the set of governing equations [107].
43
For the reasons presented above, LES has a clear potential in predicting turbulent ows encountered
in industrial applications. Such possibilities are however restricted by the hypothesis introduced while
constructing LES models.
This chapter describes the equation solved for LES of reacting ows in AVBP. First, the ltered equa-
tions solved by AVBP for a turbulent non-reacting ow are described (subsection on page 44). Subsection
on page 47 presents the models used for turbulent viscosity. Subsection on page 50 describes specically
the models for ame/turbulence interactions (the TF model) and shows how this model is coupled to the
ltered equations.
1.2.2 The Governing Equations for Non-Reacting Flows
The ltered quantity f is resolved in the numerical simulation whereas f
= ff is the subgrid scale part

due to the unresolved ow motion. For variable density , a mass-weighted Favre ltering is introduced
such as:
f = f (1.42)
The balance equations for large eddy simulations are obtained by ltering the instantaneous balance
equations 1.1, 1.2 and 1.3:
u
i
t
+

x
j
( u
i
u
j
) =

x
j
[P
ij
ij
ij
t
], (1.43)

E
t
+

x
j
(

E u
j
) =

x
j
[u
i
(P
ij
ij
) +q
j
+q
j
t
] +
T
+Q
r
, (1.44)
Y
k
t
+

x
j
(
Y
k
u
j
) =

x
j
[J
j,k
+J
j,k
t
] +
k
, (1.45)
where a repeated index implies summation over this index (Einsteins rule of summation). Note also
that throughout the document, the index k is reserved to refer to the k
th
species and does not follow the
summation rule (unless specically mentioned).
For generality, the ltered chemical source terms are mentioned here if a user wants to implement a
new combustion model. For the standard AVBP models for ame / turbulence interactions (TF and DTF),
a specic implementation is done (see Section on page 50). The cut-off scale corresponds to the mesh
size (implicit ltering). As usually done, we assume that the lter operator and the partial derivative
commute.
In Eqs. 1.43, 1.44 and 1.45, the ux tensor can be divided in three parts: the inviscid part, the viscous
part and the subgrid scale turbulent part.
Inviscid terms:
The three spatial components of the inviscid ux tensor are the same as in DNS but based on the
ltered quantities:
44
Governing equations for LES
_
_
u
i
u
j
+P
ij
E u
j
+P u
j
ij
k
u
j
_
_
(1.46)
Viscous terms:
The components of the viscous ux tensor take the form:
_
_
ij
(u
i

ij
) +q
j
J
j,k
_
_
(1.47)
Filtering the balance equations leads to unclosed quantities, which need to be modeled.
Subgrid scale turbulent terms:
The components of the turbulent subgrid scale ux take the form:
_
_
ij
t
q
j
t
J
j,k
t
_
_
(1.48)
The ltered viscous terms in non reactive ows
The ltered diffusion terms are (see T. Poinsot and D. Veynante, Chapter 4 [107]) :
the laminar ltered stress tensor
ij
is given by the following relations:
ij
= 2(S
ij
1
3
ij
S
ll
),
2(
S
ij
1
3
ij
S
ll
),
(1.49)
and
S
ij
=
1
2
(
u
j
x
i
+
u
i
x
j
), (1.50)
Eq. 1.49 may also be written:
xx

2
3
(2
e u
x

e v
y

e w
z
),
xy
(
e u
y
+
e v
x
)
yy

2
3
(2
e v
y

e u
x

e w
z
),
xz
(
e u
z
+
e w
x
)
zz

2
3
(2
e w
z

e u
x

e v
y
),
yz
(
e v
z
+
e w
y
)
(1.51)
the ltered diffusive species ux vector is:
J
i,k
=
_
D
k
W
k
W
X
k
x
i
Y
k
V
i
c
_

_
D
k
W
k
W
e
X
k
x
i
Y
k
V
i
c
_
,
(1.52)
45
where higher order correlations between the different variables of the expression are assumed
negligible.
the ltered heat ux is :
q
i
=
T
x
i
+
N
k=1
J
i,k
h
s,k

e
T
x
i
+
N
k=1
J
i,k
h
s,k
(1.53)
These forms assume that the spatial variations of molecular diffusion uxes are negligible and can be
modelled through simple gradient assumptions.
Subgrid scale turbulent terms for non-reacting LES
As highlighted above, ltering the transport equations yields a closure problemevidenced by the so called
Sub-Grid Scale (SGS) turbulent uxes (see Eq. on page 44). For the system to be solved numerically,
closures need to be supplied. Details on the forms and models available in AVBP are given in this
subsection.
the Reynolds tensor is :
ij
t
= ( u
i
u
j
u
i
u
j
), (1.54)
ij
t
is modeled as:
ij
t
= 2
t
(
S
ij
1
3
ij
S
ll
), (1.55)
The modelisation of
t
is explained in subsection on page 47.
the subgrid scale diffusive species ux vector:
J
i,k
t
= (
u
i
Y
k
u
i
Y
k
), (1.56)
J
i,k
t
is modeled as:
J
i,k
t
=
_
D
k
t
W
k
W

X
k
x
i
Y
k
V
i
c,t
_
, (1.57)
with
D
t
k
=

t
S
t
c,k
(1.58)
The turbulent Schimdt number S
t
c,k
= 1 is the same for all species and is xed in the source code
(like P
t
r
). Note also that having one turbulent Schmidt number for all the species does not imply,
V
c,t
= 0 because of the W
k
/W term in Eq. 1.57.
46
the subgrid scale heat ux vector:
q
i
t
= (
u
i
E u
i
E), (1.59)
where e is the sensible energy. In the source code, the modelisation for q
t
is written :
q
i
t
=
t
T
x
i
+
N
k=1
J
i,k
t
h
s,k
, (1.60)
with
t
=

t
C
p
P
t
r
. (1.61)
The turbulent Prandtl number is xed in AVBP. Usually, P
t
r
= 0.9.
The correction diffusion velocities are henceforth obtained in AVBP from:
V
c
i
+

V
c,t
i
=
N
k=1
_

S
c,k
+

t
S
t
c,k
_
W
k
W

X
k
x
i
, (1.62)
and where Eqs. 1.32 and 1.58 are used.
1.2.3 Models for the subgrid stress tensor
ij
t
:
In AVBP, three models are available:
the Smagorinsky model,
the Wall Adapting Linear Eddy (WALE) model,
the Filtered Smagorinsky model.
A description of the characteristics of each model is given in the next subsection.
These LES models are derived on the theoretical ground that the LES lter is spatially and temporally
invariant. Variations in the lter size due to non-uniform meshes or moving meshes are not directly
accounted for in the LES models. Change of cell topology is only accounted for through the use of the
local cell volume, that is = V
1/3
cell
.
Use of articial viscosity is in theory prohibited in LES. It should therefore be used with a lot of
caution and in no way overshadow the LES model contribution.
The ltered compressible Navier-Stokes equations exhibit sub-grid scale (SGS) tensors and vectors
describing the interaction between the non-resolved and resolved motions. The inuence of the SGS
47
on the resolved motion is taken into account in AVBP by a SGS model based on the introduction of a
turbulent viscosity,
t
. Such an approach assumes the effect of the SGS eld on the resolved eld to be
purely dissipative.
The previous hypothesis is essentially valid within the cascade theory of turbulence. The notion of
turbulent viscosity can therefore be introduced and yields a general model for the SGS which reads
ij
t
= ( u
i
u
j
u
i
u
j
)
= 2
t

S
ij
1
3

ll
t
ij
, (1.63)
with
S
ij
=
1
2
_
u
i
x
j
+
u
j
x
i
_
1
3
u
k
x
k
ij
. (1.64)
In Eqn. (1.63)
ij
t
is the SGS tensor to be modeled,
t
is the SGS turbulent viscosity, u
i
is the Favre
ltered velocity vector (compressible ows) and

S
ij
is the resolved strain rate tensor. The three models
in AVBP only differ through the estimation of
t
whose expressions are given below.
Smagorinsky model
t
= (C
S
)
2
_
2

S
ij

S
ij
, (1.65)
where denotes the lter characteristic length (cube-root of the cell volume), C
S
is the model constant
set to 0.18 but can vary between 0.1 and 0.18 depending on the ow conguration. The Smagorinsky
model [141] was developed in the sixties and heavily tested for multiple ow congurations. This closure
has the particularity of supplying the right amount of dissipation of kinetic energy in homogeneous
isotropic turbulent ows. Locality is however lost and only global quantities are maintained. It is known
as being too dissipative and transitioning ows are not suited for its use [122].
Note: in AVBP, the expression for
t
is not exactly as given in (1.65). Rather a simplied version of
S
ij
(as shown in (1.64) ) is used; compressibility effects are neglected and u
k
/x
k
0.
WALE model
s
d
ij
=
1
2
( g
2
ij
+ g
2
ji
)
1
3
g
2
kk

ij
, (1.66)
t
= (C
w
)
2
(s
d
ij
s
d
ij
)
3/2
(
S
ij
S
ij
)
5/2
+(s
d
ij
s
d
ij
)
5/4
, (1.67)
where denotes the lter characteristic length (cube-root of the cell volume), C
w
= 0.4929 is the model
constant and g
ij
denotes the resolved velocity gradient. The WALE model [40] was developed for wall
bounded ows in an attempt to recover the scaling laws of the wall. Similarly to the Smagorinsky model
locality is lost and only global quantities are to be trusted.
48
Filtered Smagorinsky model
t
= (C
S
F
)
2
_
2 HP(
S
ij
) HP(
S
ij
), (1.68)
where denotes the lter chracteristic length (cube-root of the cell volume), C
S
F
= 0.37 is the model
constant and HP(
S
ij
) denotes the resolved strain rate tensor obtained from a high-pass ltered velocity
eld. This model was developed in order to allow a better representation of local phenomena typical of
complex turbulent ows [40]. With the Filtered Smagorinsky model transition is better predicted and
locality is in general better preserved.
49
1.3 The Thickened Flame (TF) model for LES
A difcult problem is encountered for Large Eddy Simulation of premixed ames: the thickness
0
L
of a
premixed ame is generally smaller than the standard mesh size
x
used for LES. For this reason, the
Thickened Flame (TF) model has been developed so as to resolve the ame fronts on a LES mesh. In
turbulent ows, the interaction between turbulence and chemistry is altered: eddies smaller than T
0
L
do
not interact with the ame any longer. As a result, the thickening of the ame reduces the ability of the
vortices to wrinkle the ame front. As the ame surface is reduced, the reaction rate is underestimated.
In order to correct this effect, an efciency function c has been developed [28] from DNS results and
implemented into AVBP (see Fig. 1.2). It is described in the next subsection.
Figure 1.2 - Direct Numerical Simulation of ame/turbulence interactions by Veynante ([4], [104]). Left: non
thickened ame, right: thickened ame (T = 5).
1.3.1 The combustion subgrid scale model: c
A complete description of the efciency function is given in ref [28]. The underlying model philosophy
can be summarized through 3 main steps:
The wrinkling factor of the ame surface is estimated fromthe ame surface density , assuming
an equilibrium between the turbulence and the subscale ame surface:
1 +
e
S
0
L
a
T
)
s
(1.69)
where a
T
)
s
is the subgrid scale strain rate,
e
is the lter size and is a model constant.
a
T
)
s
is estimated from the lter size
e
and the subgrid scale turbulent velocity u
e
: a
T
)
s
=
u
e
/
e
. The function corresponds to the integration of the effective strain rate induced by
50
The Thickened Flame (TF) model for LES
all scales affected by the articial thickening, i.e. between the Kolmogorov
K
and the lter
e
scales (see also [84]). is written as:
1
L
,
u
e
S
0
L
_
= 0.75 exp
_
1.2
_
u
e
/S
0
L
_
0.3
_
_
1
L
_2
3
(1.70)
Finally, the efciency function is dened as the wrinkling ratio between the non-thickened refer-
ence ame and the thickened ame:
c =
(
0
L
)
(
1
L
)
=
1 +
_
0
L
,
u
e
S
0
L
_
u
e
S
0
L
1 +
_
1
L
,
u
e
S
0
L
_
u
e
S
0
L
(1.71)
S
0
L
and
0
L
are the laminar ame speed and the laminar ame thickness, respectively, when T = 1
and
1
L
= T
0
L
.
c varies between 1 (weak turbulence) to c
max
T
2/3
(large wrinkling at the subgrid scale). In
turbulent premixed zones, the efciency function is determined to ensure that the turbulent ame
speed will be c S
0
L
= S
T
. The efciency function is required when the vortex size r is dened
by
0
L
> r >
c
L
for a real ame and by
1
L
= T
0
L
> r >
c
L
for a thickened ame.
c
L
is
a cut-off length scale: for vortices lower than
c
L
, the ame remains unaffected.
c
L
is dened in
[28], Eq. 31.
The lter size
e
corresponds to the greatest scale affected by the ame thickening, that is to
say
1
L
. In practice,
e
= 10
x
with
x
=
3
voln. The subgrid scale turbulent velocity u
e
is estimated using the operateur OP
2
, based on the rotational of the velocity eld to remove the
dilatational part of the velocity which must not be counted as turbulence. A Laplacian operator
is directly applied to the velocity :
u
e
= c
2
3
x
[

2
x
j
x
j
_
lmn
u
n
x
m
_
[ (1.72)
with c
2
2 and where
lmn
stands for the permutation tensor.
Estimation of the model constant
The model constant is estimated to match the asymptotic behavior of the wrinkling factor versus
RMS velocity u for thin ames when
e
goes to the integral length scale l
t
, the ame wrinkling goes
to
max
dened by:
max
= 1 +u
/S
0
L
(1.73)
with u the velocity at length scale l
t
. is then deduced from Eq. 1.73:
=
2 ln(2)
3c
ms
[Re
1/2
t
1]
(1.74)
where Re
t
=
u
l
t
is the turbulent Reynolds number and c

ms
= 0.28. The reader is referred to [28] for
more details.
51
The TFLES1 model: In this method, the user has to provide an evaluation for . For that, Re
t
=
u
l
t
has to be evaluated. Typically, l

t
is of the order of the largest vortex in the computation domain. However,
strong variations of Re
t
may be observed whether the vortex is near the injectors or in the burner, in the
fresh gas or in the burnt gas. In usual combustion chamber, Re
t
varies between 100 and 100,000, then
varies between 0.165 and 0.0052 (assuming is of the order of unity).
The TFLES2 model: The objective of this model is to compute Re
t
as a function of space and
time and to give a new formulation for . The different steps to obtain the efciency function are the
following:
First, the integral length scale is estimated. Near the wall, the log-law in turbulent channel ow
is used to estimate l
t
: l
t
0.4d
w
, where d
w
is the distance to the wall. Numerically, d
w
is the
minimum distance from walls to the considered node. Far from walls, the user must provide an
estimation of the integral length scale l
0
t
in the whole chamber. Usually l
0
t
is of the order of 1/10
to 1/4 of the transversal size of the burner. Finally, the local integral length scale is:
l
t
min(0.4d
w
, l
0
t
) (1.75)
Assuming a Kolmogorov cascade with constant transfer rate , the uctuation velocity u is de-
duced by:
=
u
3
l
t
=
u
3
e
(1.76)
The subgrid scale velocity u
e
is overestimated when using Eq. 1.72 close to the wall, that is to
say in boundary layers. For this reason a damping function depending on the distance to the wall
is used and u
e
is then determined as:
u
e
=
_
1 exp
_
d
w
0.2
1
L
__
c
2
3
x
[

2
x
j
x
j
_
lmn
u
n
x
m
_
[ (1.77)
Near the wall, l
t
goes to zero while the lter scale
e
is nearly constant. When
e
< l
t
, the
Kolmogorov assumption is not valid any more, that is why
e
and u
e
are replaced by l
t
and u.
Such an approximation is acceptable, because most of the combustion occurs in the center of the
chamber.
A correction for the turbulent Reynolds number is done wherever the cut-off length scale
c
L
is
smaller or greater than the Kolmogorov scale
k
. If
c
L
<
k
,
k
is the smallest efcient structure
size and
c
L
is irrelevant. The Reynolds number is then Re
t
=
u
l
t
. Recalling that the ame

Reynolds number is dened by: Re
f
=
S
0
L
0
L
1, and that the Kolmogorov scale Reynolds

number is written as: Re(
k
) =
u
(
k
)
k
(
k
)
k
S
0
L
0
L
= 1, the turbulent Reynolds number may be
written: Re
t
=
_
l
t
k
_
4/3
. If
c
L
>
k
, the smallest efcient structure size is
c
L
and the Reynolds
number estimation should be corrected with a function , such as:
2
Re
t
=
_
l
t
c
L
_
4/3
. The nal
model for is then written:
=
2ln(2)
3c
ms
[(Re
t
)
1/2
1]
(1.78)
52
The Thickened Flame (TF) model for LES
= min(
_
c
L
_
2/3
, 1) (1.79)
_
c
L
_
2/3
= (Re
t
)
1/2
_
l
t
0
L
_
2/3
ln(2)
2c
ms
exp
_
_
_
1.2
_
u
S
0
L
_
0.3
_
_
_
(1.80)
The model constant has to be xed by the user in the input le and is of the order of 0.3.
Other forms of efciency function have been derived by Charlette and Meneveau [20, 21] but are not
implemented in AVBP yet.
1.3.2 Implementation of the standard Thickened Flame (TF) model
The ltered equations for total energy and for species (Eqs. 1.44 & 1.45) must be modied in reactive
ows when the TF or DTF model is used. In this case, only the ltered equations for velocities (Eq. 1.43)
are unchanged. For the species and energy, the ltered equations are replaced by the thickened equations
as follows:
Viscous terms
the ltered diffusive species ux vector is given by:
J
i,k
= c T

S
c,k
W
k
W

X
k
x
i
+
k
V
c
i
, (1.81)
with
V
c
i
= c T
N
k=1
S
c,k
W
k
W

X
k
x
i
, (1.82)
the ltered heat ux is:
q
i
= c T
C
p
P
r
T
x
i
+
N
k=1
J
i,k
h
s,k
, (1.83)
The source term
The ltered source term vector of Eq. 1.42 is written:
_
E
T
(
e
Y
k
,
e
T)
F
E
k
(
e
Y
k
,
e
T)
F
_
, (1.84)
where
T
(
Y
k
,

T) and
k
(
Y
k
,

T) are reaction rates computed with the Arrhenius expression and the
ltered values of Y
k
and T. Note that this model should be used only for perfectly premixed cases since
53
mixing in the fresh gases, for example, is modied by thickening and not correctly handled with the
ltered terms of Eqs. 1.44 & 1.45. The actual transport equations for the TF model are summarized
below.
Use of the TF model implies the following relation for the correction diffusion velocities:
V
c
i
+

V
c,t
i
=
N
k=1
c T

S
c,k
W
k
W

X
k
x
i
, (1.85)
1.3.3 Final equations solved for the TF model
The nal set of LES equations solved for when performing LES of reacting ows with theTF model
nally reads:
u
i
t
+

x
j
( u
i
u
j
) =

x
j
_
P
ij
2 ( +
t
) (
S
ij
1/3

S
ll

ij
)
_
,

E
t
+

x
j
(

E u
j
) =

x
j
_
u
i
P
ij
2 u
i
(
S
ij
1/3

S
ll

ij
)
_
+

x
j
_
C
p
c T

P
r
T
x
j
_
+

x
j
_
N
k=1
_
c T

S
c,k
W
k
W

X
k
x
j

Y
k
(
V
c
j
+

V
c,t
j
)
_

h
s,k
_
+
c
T
T
,
Y
k
t
+

x
j
(
Y
k
u
j
) =

x
j
_
c T

S
c,k
W
k
W

X
k
x
j

Y
k
(
V
c
j
+

V
c,t
j
)
_
+
c
k
T
,
54
1.4 General aspects of the Boundary Conditions in AVBP
1.4 General aspects of the Boundary Conditions in AVBP
Boundary Conditions are an essential part in any CFD code, and especially in AVBP because of acoustics
present in the governing equations [127],[104]. The time integration in AVBP is performed with a multi-
stage Runge-Kutta (RK) method. For simplicity, we only consider here a single-stage RK, which is
actually a simple Euler integration, eventually combined with a Lax-Wendroff method for stability.
Knowing the solution w
n
at time t, the solution w
n+1
at time t + t is computed for each node i as:
w
n+1
i
= w
n
i

t
V
i
.dw
n
i
(1.86)
where w
n
i
= w(t, x
i
) and w
n+1
i
= w(t + t, x
i
); x
i
is the coordinate vector, V
i
is the nodal volume
around node i and dw
n
i
is the nodal residual at node i, as computed by the numerical scheme.
The n superscript is here to remind that AVBP is an explicit code, hence this nodal residual only depends
on quantities known at the timestep n.
This formula is applied to each node of the computational domain (). If no physical BC was imposed,
the computed solution at each timestep would only depend on the initial solution and on the numerical
scheme.
In order to impose a BC on the border of the domain (), we write:
_
_
_
w
n+1
i
= w
n
i

t
V
i
(dw
n
i
)
scheme
x
i
/
w
n+1
i
= w
n
i

t
V
i
(dw
n
i
)
BC
x
i

(1.87)
For each node lying on the boundary, we have replaced the scheme-predicted residual by a BC-corrected
residual. This operation is known as imposing the BC in a hard way. A weak method is additionally
used in conjunction with this hard method for certain BC. In this case, some gradients and uxes are
modied before applying the numerical scheme. It is mainly used for Von Neuman like conditions, as
adiabaticity or impermeability.
The AVBP BC can be classied through two categories:
Non-characteristic BC, that work directly on conserved variables,
Characteristic BC, that use a wave decomposition to modify residuals.
55
1.5 Articial Viscosity in AVBP
The numerical discretization methods in AVBP are spatially centered. These types of schemes are known
to be naturally subject to small-scale oscillations in the vicinity of steep solution variations. This is
why it is common practice to add a so-called articial viscosity (AV) term to the discrete equations, to
avoid these spurious modes (also known as wiggles) and in order to smooth very strong gradients.
We describe here the different AV methods used in AVBP. These AV models are characterized by the
linear preserving property which leaves unmodied a linear solution on any type of element. The
models are based on a combination of a shock capturing term (called 2
nd
order AV) and a background
dissipation term (called 4
th
order AV). In AVBP, adding AV is done in two steps:
rst a sensor detects if AV is necessary, as a function of the ow characteristics,
then a certain amount of 2
nd
and 4
th
AV is applied, depending on the sensor value and on user-
dened parameters.
1.5.1 The sensors
A sensor
j
is a scaled parameter which is dened for every cell
j
of the domain that takes values from
zero to one.
j
= 0 means that the solution is well resolved and that no AV should be applied while
j
= 1 signies that the solution has strong local variations and that AV must be applied. This sensor
is obtained by comparing different evaluations (on different stencils) of the gradient of a given scalar
(pressure, total energy, mass fractions, . . . ). If these gradients are identical, then the solution is locally
linear and the sensor is zero. On the contrary, if these two estimations are different, local non-linearities
are present, and the sensor is activated. The key point is to nd a suitable sensor-function that is non-zero
only at places where stability problems occur.
The Colin-sensor (
C
j
) [28] used in this thesis is described here.
The Colin sensor
For most unsteady turbulent computations it is necessary to have a sharp sensor, which is very small
when the ow is sufciently resolved, and which is nearly maximum when a certain level of non-
linearities occurs.
S is the scalar quantity the sensor is based on (usually S is the pressure)
k
1
and
k
2
functions are
dened as:
k
1
= S
j
S
k

k
2
= (
S)
k
.(x
j
x
k
) (1.88)
where a k subscript denotes cell-vertex values while
j
is the subscript for cell-averaged values.
(
S)
k
is the gradient of S at node k as computed in AVBP.
k
1
measures the variation of S inside the cell
j
(using only quantities dened on this cell).
k
2
is an
56
Articial Viscosity in AVBP
estimation of the same variation but on a wider stencil (using all the neighbouring cell of the node k).
the aim of the so-called Colin-sensor is to preserve sharp gradients in the computation. its properties
can be summarized as follows:

C
j
is very small when both
k
1
and
k
2
are small compared to S
j
. This corresponds to low
amplitude numerical errors (when
k
1
and
k
2
have opposite signs) or smooth gradients that are
well resolved by the scheme (when
k
1
and
k
2
have the same sign).

C
j
is small when
k
1
and
k
2
have the same sign and the same order of magnitude, even if they
are quite large. This corresponds to stiff gradients well resolved by the scheme.

C
j
is big when
k
1
and
k
2
have opposite signs and one of the two term is large compared to the
other. This corresponds to a high-amplitude numerical oscillation.

C
j
is big when either
k
1
or
k
2
is of the same order of magnitude as S
j
. This corresponds to a
non-physical situation that originates from a numerical problem.
The exact denition of the Colin-sensor is:
j
=
1
2
_
1 + tanh
_
__
1
2
_
1 + tanh
_
__
(1.89)
with:
= max
k
j
_
0,

k
[
k
[ +
1
S
k
J
k
_
(1.90)
k
= [
k
1
k
2
[
k
max
_
[
k
1
[, [
k
2
[
_
(1.91)
k
=
2
_
1
3
max
_
[
k
1
[, [
k
2
[
_
[
k
1
[ +[
k
2
[ +S
k
_
(1.92)
The numerical values used in AVBP are:
0
= 2.10
2
= 1.10
2
1
= 1.10
2
2
= 0.95
3
= 0.5 (1.93)
WARNING:
Note, that these denitions of and
k
apply only for the Navier-Stokes variables. For species, the
reference value is not S
k
but 1, which is the maximum value of a species mass fraction:
= max
k
j
_
0,

k
[
k
[ +
1
J
k
_
and
k
=
2
_
1
3
max
_
[
k
1
[, [
k
2
[
_
[
k
1
[ +[
k
2
[ + 1
_
(1.94)
57
1.5.2 The operators
There are two AV operators in AVBP: a 2
nd
order operator and a 4
th
order operator. All AV models in
AVBP are a blend of these two operators. These operators have the following properties:
2
nd
order operator: it acts just like a classical viscosity. It smoothes gradients, and introduces
articial dissipation. It is thus associated to a sensor which determines where it must be applied.
Doing this, the numerical scheme keeps its order of convergence in the zones where the sensor is
inactive, while ensuring stability and robustness in the critical regions. Historically, it was used to
control shocks, but it can actually smooth any physical gradient.
4
th
order operator: it is a less common operator. It acts as a bi-Laplacian and is mainly used to
control spurious high-frequency wiggles.
The way they are combined is determined both by the sensor and by user-dened parameters (smu2 and
smu4).
Both operator contributions are rst computed on each cell vertex, and are then scattered back to nodes
(there is no divergence here, as it is done directly during the scattering operation).
The 2
nd
order operator
A cell contribution of the 2
nd
order AV is rst computed on each vertex of the cell
j
:
R
k
j
=
1
N
v
V
j
t
j
smu2
j
(w
j
w
k
) (1.95)
The nodal residual is then found by adding the surrounding cells contributions:
dw
k
=
j
R
k
j
(1.96)
For example, on a 1D uniform mesh, of mesh size x, and for
j
= = cste :
dw
k
=
smu2
2
x
t
(w
k1
2w
k
+w
k+1
) (1.97)
which can be interpreted as:
dw
k
=
AV
_
(
FD
k,x
w) dx (1.98)
with:
AV
=
smu2 x
2
2t
=
smu2 x[u +c[
2 CFL
and
FD
k,x
w =
w
k1
2w
k
+w
k+1
x
2
(1.99)
where
FD
k,x
is exactly the classical FD Laplacian operator evaluated at k and of size x.
This shows that
AV
can be seen as an articial viscosity (it has the same units as a physical viscosity),
which is controlled by the user-dened parameter smu2. The smu2 parameter is therefore dimension-
less.
58
Articial Viscosity in AVBP
The 4
th
order operator
The technique used for the 4
th
order operator is identical to the technique of the 2
nd
order operator. A
cell contribution is rst computed on each vertex:
R
k
j
=
1
N
v
V
j
t
j
smu4
_
(
w)
j
(x
j
x
k
) (w
j
w
k
)
_
(1.100)
The nodal value is then found by adding every surrounding cells contributions:
dw
k
=
j
R
k
j
(1.101)
For example, on a 1D uniform mesh, of mesh size x, this yields:
R
k
left
=
smu4
2
x
t
__
1
2
(
w
k
w
k2
2x
+
w
k+1
w
k1
2x
)
_
_
x
2
_
_
w
k1
+w
k
2
w
k
__
(1.102)
R
k
right
=
smu4
2
x
t
__
1
2
(
w
k+1
w
k1
2x
+
w
k+2
w
k
2x
)
_
_
x
2
_
_
w
k
+w
k+1
2
w
k
__
(1.103)
Adding these 2 contributions gives:
dw
k
= smu4
x
16t
(w
k2
4w
k1
+ 6w
k
4w
k+1
+w
k+2
) (1.104)
which can be interpreted:
dw
k
=
AV
_
(
FD
k,x
w) dx (1.105)
with:
AV
=
smu4.x
4
16t
=
smu4.x
3
[u +c[
16 CFL
and
FD
k,x
w =
w
k2
4w
k1
+ 6w
k
4w
k+1
+w
k+2
x
4
(1.106)
where
FD
k,x
is exactly the classical FD bi-Laplacian operator evaluated at k and of size x.
This shows that
AV
can be seen as an articial 4
th
order hyper-viscosity, which is controlled by the
user-dened parameter smu4. Just like smu2, the smu4 parameter is dimensionless.
59
DESCRIPTION OF THE NUMERICAL TOOL
1.6 Cell-Vertex Discretization
The ow solver used for the discretization of the governing equations is based on the nite volume
(FV) method. There are two common techniques for implementing FV methods: the so called cell-
vertex and the cell-centered formulation. In the latter, not used in AVBP, discrete solution values are
stored at the center of the control volumes (or grid cells), and neighbouring values are averaged across
cell boundaries in order to calculate uxes. The alternative cell-vertex technique, used as underlying
numerical discretization method of AVBP, the discrete values of the conserved variables are stored at the
cell vertices (or grid nodes). The mean values of the uxes are then obtained by averaging along the cell
edges.
1.6.1 Weighted Cell Residual Approach
For the description of the weighted cell-residual approach the laminar Navier-Stokes equations are con-
sidered in their conservative formulation:
w
t
+

T = 0, (1.107)
where w is the vector of conserved variables and

T is the corresponding ux tensor. For convenience,
the latter is divided into an inviscid and a viscous part,

T =

T
I
(w) +

T
V
(w,

w). The spatial terms
of the equations are then approximated in each control volume
j
to give the residual
R
j
=
1
V
j
_
T n do, (1.108)
where
j
denotes the boundary of
j
with normal n.
This cell-vertex approximation is readily applicable to arbitrary cell types and is hence straightforward
to apply for hybrid grids. The residual (1.108) is rst computed for each element by making use of a
simple integration rule applied to the faces. For triangular faces, a straightforward mid-point rule is
used, which is equivalent to the assumption that the individual components of the ux vary linearly on
these faces. For quadrilateral faces, where the nodes may not be co-planar, in order to ensure that the
integration is exact for arbitrary elements if the ux functions do indeed vary linearly, each face is divided
into triangles and then integrated over the individual triangles. The ux value is then obtained from the
average of four triangles (two divisions along the two diagonals). This so-called linear preservation
property plays an important part in the algorithm for ensuring that accuracy is not lost on irregular
meshes. Computationally, it is useful to write the discrete integration of (1.108) over an arbitrary cell as
R
j
=
1
N
d
V
i
j
T
i

dS
i
, (1.109)
60
Cell-Vertex Discretization
where

T
i
is an approximation of

T at the nodes, N
d
represents the number of space dimensions and
i
j
are the vertices of the cell. In this formulation the geometrical information has been factored
into terms

dS
i
that are associated with individual nodes of the cell but not faces;

dS
i
is merely the
average of the area-weighted normals for triangulated faces with a common node i, i
j
. Note, that
for consistency one has
i
j
dS
i
=

0. A linear preserving approximation of the divergence operator
is obtained if the volume V
j
is dened consistently as
V
j
=
1
N
2
d
i
j
x
i

dS
i
, (1.110)
since x = N
d
.
Once the cell residuals are calculated, one may then dene the semi-discrete scheme
dw
k
dt
=
1
V
k
j|k
j
D
k
j
V
j
R
j
, (1.111)
where D
k
j
is a distribution matrix that weights the cell residual from cell center
j
to node k (scatter
operation), and V
k
is a control volume associated with each node. Conservation is guaranteed if
k
j
D
k
j
= I. In the present context, (1.111) is solved to obtain the steady-state solution using
explicit Euler or Runge-Kutta timestepping.
The family of schemes of interest makes use of the following denition of the distribution matrix:
D
k
j
=
1
n
n
(I +C
t
j
V
j

dS
k
), (1.112)
where n
n
is the number of nodes of
j
and

/ is the Jacobian of the ux tensor. The simplest central
difference scheme is obtained by choosing C = 0 and is neutrally stable when combined with Runge-
Kutta time-stepping. A Lax-Wendroff type scheme may also be formulated in which case C is chosen
to be a constant that depends on the number of space dimensions and the type of cells used it
may be shown that this takes the simple form C = n
2
v
/2N
d
.
61
DESCRIPTION OF THE NUMERICAL TOOL
62
Part II
Flame Transfer Functions
Table of Contents
2 Linear models for FTFs (Flame Transfer Functions) 69
2.1 Historical background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
2.2 Models for FTF . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
2.2.1 Laminar FTF . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
2.2.2 Turbulent FTF . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
Convective hypothesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
Global formulation for FTF model . . . . . . . . . . . . . . . . . . . . . . . . . 75
Local formulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
Normalizations of FTF amplitudes . . . . . . . . . . . . . . . . . . . . . . . . . 77
Pressure/velocity FTF model . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
General remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
3 Measurement methods for FTF in LES 81
3.1 LES of FTF . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
3.1.1 LES as the proper tool for FTF . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
3.1.2 Introducing acoustic waves through the burner inlet . . . . . . . . . . . . . . . . 81
3.2 Postprocessing methods for local FTF . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
3.2.1 Time delay or Flight time method (TD method). . . . . . . . . . . . . . . . 83
3.2.2 Harmonic forcing; FFT postprocessed method (HF-FFT method). . . . . . . . . 83
3.2.3 White noise forcing; FFT postprocessed method (WN-FFT method). . . . . . . . 84
3.2.4 White noise forcing; Wiener-Hopf postprocessed method (WN-WH method). . . 84
3.3 Preliminary comparisons of postprocessing methods. . . . . . . . . . . . . . . . . . . . 85
3.3.1 Validation of signal post-processing methods: FFT versus Wiener-Hopf inversion. 86
First test : inuence of a noisy response signal . . . . . . . . . . . . . . . . . . 86
Second test : inuence of signal sampling . . . . . . . . . . . . . . . . . . . . . 87
Third test : inuence of a shift of the response frequency. . . . . . . . . . . . . . 87
TABLE OF CONTENTS
3.3.2 Admittance of a diffuser : HF-FFT versus WN-WH methods. . . . . . . . . . . 89
Mean ow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
Admittance calculation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
4 Transfer functions of ames 95
4.1 Conguration A : laminar planar premixed ame (1D) . . . . . . . . . . . . . . . . . . 97
4.1.1 Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
4.1.2 Amplitude . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
4.1.3 Time delay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
4.2 Conguration B : laminar V ame (2D axi-symmetric) . . . . . . . . . . . . . . . . . . 101
4.2.1 Comparison of global FTF (F
norm
) . . . . . . . . . . . . . . . . . . . . . . . . 101
Amplitude . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
Phase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
4.3 Conguration C : turbulent burner in cylindrical chamber (3D) . . . . . . . . . . . . . . 104
4.3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
4.3.2 A double-swirler partially premixed burner . . . . . . . . . . . . . . . . . . . . 107
4.3.3 Large Eddy Simulations for gas turbines . . . . . . . . . . . . . . . . . . . . . . 107
4.3.4 Introducing acoustic waves through the burner inlet . . . . . . . . . . . . . . . . 109
4.3.5 Reacting steady ow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
4.3.6 Forced reacting ow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
4.3.7 Coherent structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
4.3.8 Comparison of LES, RANS and experiment : global transfer function . . . . . . 118
4.3.9 Comparison of LES, RANS and experiment : local transfer function . . . . . . 118
4.3.10 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
sector (3D) . . . . . . . . . . . . . . . . . . 127

4.4.1 General computational remarks. . . . . . . . . . . . . . . . . . . . . . . . . . . 128
4.4.2 Reference signals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
Comparison of global FTF. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
Comparison of local FTF. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
norm
) . . . . . . . . . . . . . . . . . . . . . . . . 130
Amplitude . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
Phase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
4.4.4 Local comparison of FTF (F
dim
) . . . . . . . . . . . . . . . . . . . . . . . . . . 132
Amplitude . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
66
TABLE OF CONTENTS
Time delay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
Weighted Probability density functions of time delays. . . . . . . . . . . . . . . 135
4.4.5 Concluding remarks for conguration D . . . . . . . . . . . . . . . . . . . . . . 137
4.5 Evaluation of FTF measurements methods in LES. . . . . . . . . . . . . . . . . . . . . 138
67
TABLE OF CONTENTS
68
Chapter 2
Linear models for FTFs (Flame Transfer
Functions)
2.1 Historical background
Since the beginning of the 50s, the development of higher power propulsion systems has been disrupted
by the issue of thermoacoustic instabilities. Thanks to the use of mathematical tools from the eld of
electronics, like the notion of transfer functions, the understanding of these instabilities has also grown
at the same time. There are two main methods to study the stability of combustors:
the Purely Acoustic (PA) method. In this approach, the burner is considered as a black box and a
two-port formulation is used to reconstruct a transfer matrix linking acoustic uctuations on both
sides of the burner [1, 95, 98, 99, 111]. Figure 2.1 illustrates the idea that the whole combustion
chamber can be treated as a black box and its acoustic response obtained from the four signals of
inlet and outlet velocity and pressure uctuation signals. U
inlet
and U
outlet
are respectively the
Figure 2.1 - The combustion chamber seen with a two-port formulation.
velocity uctuations at the inlet and at the outlet of the conguration. P
inlet
and P
outlet
are the
pressure uctuations at the inlet and at the outlet of the system. These quantities are linked by the
LINEAR MODELS FOR FTFS (FLAME TRANSFER FUNCTIONS)
acoustic matrix M of the conguration by the following relation :
_
U
outlet
P
outlet
_
= M
_
U
inlet
P
inlet
_
(2.1)
Note that the coefcients of the M matrix are complex. The element containing the ame is treated
like all others so that the ame effect is included in the matrix of this element. The major drawback
of this method is that in experiments, the measurement of unsteady pressure and velocities in burnt
gases can be a difcult task.
the Flame Transfer Function methods. With this method, the velocity and pressure signals after the
burner are not used anymore. The transfer acoustic matrix is recovered using the inlet velocity (and
pressure) signal(s) as well as the instantaneous heat release signal. When using these methods, the
dimensional ame transfer function is called F
dim
() =

and a priori depends on the frequency.

[F
dim
()[ is amplitude of the response and Arg(F
dim
())/(2) is the time delay of the response.
Depending on the model used, both can vary with frequency or not.
As shown by Trufn et Al.[150], both methods (PA and FTF) are mathematically equivalent and a two-
port matrix can be obtained by the knowledge of the transfer function F
dim
().
In the following, focus is mainly put on Flame Transfer Function models and measurement methods.
Figure 2.2 illustrates the idea that in a combustion chamber, the whole reaction zone can be treated as
a black box and that the global heat release response of the ame can be correlated to the incoming
uctuating velocity and/or pressure signal(s). The rst introduction of ame transfer function is due to
Figure 2.2 - FTF representation of the reaction zone.
the work on linear stability of combustor of Gunder et al. [49] for whom the ame is seen as a reacting
device with a xed time delay to mass ow uctuations coming from the inlet. Following this work
and assuming that heat release uctuations are mostly due to the inlet velocity uctuations
1
; Crocco et
al. [31] introduced and developed in 1951 the idea that the heat release uctuation is linked to the inlet
velocity uctuation by a simple relation known as the n model. The n model is the simplest FTF
approach. It assumes the following relation between the heat release uctuation in the whole chamber
T
and the inlet velocity uctuation:
T
=
_
V

T
(t)dv nu
(t ) (2.2)
1
This assumption comes from experiments where the ame surface varies linearly with the velocity in front of the ame.
70
2.1 Historical background
where
T
is the local heat release uctuation and u
the reference (inlet) velocity uctuation. n and

are constant over time and do not depend on the frequency of the uctuation. This model implies that
any inlet velocity uctuation acts on the ame after a time delay under the shape of a heat release
uctuation whose amplitude is proportional to n times the initial velocity perturbation. The previous
relation assumes that the incoming perturbation is harmonic. It is intrinsically linked to the linear stability
studies of combustion systems.
In these rst studies, n and are constant. This model therefore has the drawback of considering all
frequencies equivalent concerning the ame transfer function. But it is known for a long time (De Soete
in 1964 [142] for laminar ames and Becker et al. in 1971 for turbulent ames [6]) that FTF are not
constant and can be approximated in amplitude by a frequency low-pass lter. Figure 2.3 represents
Figure 2.3 - Typical shape of the ame transfer function. The ame behaves as a low pass lter with a given
cut-off frequency
a typical amplitude of the FTF as observed experimentally. More recently, Le Helley [52] and Noiray
et al. [94] also conrmed this fact experimentally showing that the amplitude of the ame response
decreases with the frequency.
The n model has been extended and rened in many aspects and present FTF models use varying
modulus and phase with frequency. Variables n and are linked to the FTF by the following relations :
n() = [F
dim
()[ (2.3)
() =
Arg(F
dim
())
2
(2.4)
Yet, the dependance of the time delay (()) to the frequency (

2
) is not clear. Other authors [14, 67, 66]
have measured almost linearly increasing phase of the ame response with frequency or, in other words,
almost constant time delays of the response of the ame. This observation leads to an hybrid model
where the modulus of F varies with the frequency but the time delay () is the mean convective time
needed by the perturbation to go from the inlet of the combustion chamber to the ame. As shown in
Section 2.2, this leads to a particular method to measure the phase response of the ame : the Time
delay or Flight time method (TD).
71
2.2 Models for FTF
The response of turbulent ames to acoustic forcing is studied in the present work. This section focuses
on hypothesis and phenomenological models that have been developed for such ows. Great progress
has recently been made in the comprehension of the linear and non-linear response of laminar ames to
acoustic disturbances [42, 74, 134]. The derived models will be mentioned since they illustrate the work
in progress in the derivation of models for FTF of turbulent ames. First, the model of Ducruix et al. [42]
is discussed as a particular example of a theoretical laminar FTF model and its main trends are analyzed
(Section 2.2.1). Then phenomenological models for turbulent FTF are discussed along with their main
hypothesis (Section 2.2.2).
2.2.1 Laminar FTF
The laminar ame transfer function model proposed by EM2C in a series of publications [42, 94, 134]
is dedicated to laminar ames anchored as V ames. The ame movement is studied analytically or
using the so-called G-equation. The uctuation of the ame surface is correlated to the uctuation of the
incoming velocity and the uctuation of heat release is directly linked with the uctuation of the ame
surface. Ducruix et al. [42] and Schuller et al. [134] develop the following theoretical model for the
response of laminar conical-ame :
T
=
a
u
2
[(1 cos
)cos(t) + (
sin
)sin(t)] (2.5)
T
=
u
u
F
norm
() (2.6)
where
L
f
0
/S
l
is a reduced pulsation. L
f
, S
l
and
0
are respectively the ame length, the
laminar ame velocity and the half angle of the cone dened by the ame front (see g 2.4).
T
and

T
are respectively the uctuation and the mean value of the volumic integrated heat release. u
= acos(t)
and u are the uctuation and the mean value of the velocity eld. Equation 2.6 leads to the following
values for [F
norm
()[ and Arg(F
norm
()):
[F
norm
()[ = 2/
2
[(1 cos
)
2
+ (
sin
)
2
]
1
2
Arg(F
norm
()) = tan
1
[(
sin
)/1 cos
)]
Figure 2.5 shows a sketch of the results obtained both theoretically and experimentally. It shows a good
agreement between experimental results and model both for the amplitude and the phase of the FTF for
low reduced pulsations
. These results point out some of the main features of the FTF to be met with
turbulent FTF.
The FTF can be well reproduced using a rst-order low-pass lter at least for low frequencies. At
higher frequencies, the heat release uctuation and the velocity uctuation at the burner mouth can
be out of phase whereas the maximum phase is /2 for a rst order low-pass lter. Obviously, the
FTF phase in turbulent cases will take any value between 0 and 2 making it hard to represent it
72
2.2 Models for FTF
Figure 2.4 - Theroretical experiment used for the derivation of the laminar FTF model. a) steady state. b) forced
case ([42])
Figure 2.5 - Sketch of the amplitude and phase of the laminar FTF. Experiment First-order model
([42])
73
with low-order low-pass lter models which are limited to less than 2 for the maximum phase of
the FTF.
Even with no reections, the FTF amplitude is not continuously decreasing with the frequency and
shows peaks. One might infer that if any reection on walls should occur, the ame might support
instabilities at the proper frequencies.
Yet, this laminar FTF theoretical model cannot be directly applied to turbulent combustion because of
multiple hypothesis that cannot be used with turbulent combustion.
Velocity uctuations are directly correlated to pressure oscillations. In other words, the baseline
ow is laminar and no turbulent uncorrelated velocity uctuations are present.
The speed of the ame front is assumed to be independent of the local curvature and is equal to
the laminar ame velocity.
The ame front is continuous. No local extinction occur, which will be the case for turbulent
ames.
The ame is anchored to the burner lip. No movement of the ame base is considered.
The incoming acoustic perturbation is one-dimensional which is not the case for turbulent com-
bustion where the ow eld is greatly perturbed by the ame.
If one of the previous hypothesis is relaxed, the equations that lead to the laminar FTF become too
complex to be solved by hand and need to be solved numerically. Other models of FTF for laminar
ames have been described by Hathout et al [51] or Lieuwen [73] but they have similar limitations.
2.2.2 Turbulent FTF
Methods used for laminar FTF cannot be directly applied to FTF of turbulent ames. One therefore needs
to consider more phenomenological hypothesis along with numerical tools to obtain the turbulent FTF.
This section focuses on models for turbulent Flame Transfer Function. These models are all extentions of
the classical n model [31] and therefore suppose that the response to a harmonic forcing of the inlet
velocity is a harmonic heat release uctuation at the same frequency. Still, it is of interest to mention
the work of Armitage et al. [5] who recently investigated the non-linear response of turbulent premixed
ames.
Convective hypothesis
As presented in the introduction, most qualitative analysis of the instability phenomenon often mention
that it is enough to know the time delay since this parameter will impose the frequency of the instability.
Therefore most efforts have been made to determine the phase of the ame transfer function and not the
74
2.2 Models for FTF
gain. Modeling the time delay of the ame response as being constant is something widely done when
constructors want to have a rough idea of the stability of their combustor [77].
The assumptions used in this case are:
that the upstream generated perturbation has a convective nature and therefore that the ame re-
sponds to a propagation of an entropy, a vorticity or a mass ow rate uctuation which is convected
by the mean ow,
depending on the authors, different distributions of time delays between heat release uctuations
and forcing velocity uctuations are chosen, from continuous [39] to discrete [66] distributions. At
least a distribution for should be chosen for partially premixed ames as argued by Dowling [39]
who gives a simple explanation in the case of lean combustors. At lean conditions, the equivalence
ratio uctuation in front of the ame is the main reason for heat release uctuation. A simple ame
transfer function is in this case :
T
= k
(2.7)
If there is a uniform convection time from the fuel injection to the ame, then
(t)
=
u
(t )
u
(2.8)
where u is the fuel injection velocity. But substituting relation 2.8 into relation 2.7 leads to a FTF
amplitude which is independent of the frequency; which cannot be true, as argued in the previous
section. Choosing an homogeneous repartition of the convective time between and +
leads to the following relation between heat release uctuations and inlet velocity uctuations.
T
()
T
= k
u()
u
1
2
_
+
e
it
= k
sin()
u()
u
e
i
(2.9)
The shape of the FTF obtained from this model better ts the experimental and numerical mea-
surements of Flame Transfer Function amplitudes which state that the FTF amplitude decreases
with the frequency of forcing.
Global formulation for FTF model
A normalized version of this model is often used to compare the global heat release in the burner and the
inlet velocity. It enables to compare results from one conguration to an other. In this case, the transfer
function F is dened by:
1
p
0
S
inj
_
V

T
(t)dv = F()u
(t) (2.10)
where S
inj
is the forced reference inlet surface,
T
and u
are respectively the local heat release and

the inlet velocity uctuations taken at the reference point. This normalization is introduced by analytical
calculation of the asymptotic behavior for innitely low frequency ame response as derived in [61] and
presented below : for premixed ames, this asymptotic value links the ratio of burnt gas temperature (T
b
)
75
and fresh gas temperature (T
f
) with the amplitude of the ame response to innitively low frequency
forcing.
Replacing integral heat release by fuel consumption in Eq.(2.10) and balancing fuel consumption with
fuel supply leads to :
QS
inj
u
inj
0
Y
F
=
_
V

T
dv (2.11)
where Q is the heat released by the combustion of 1kg of fuel. Then in the low frequency limit ( 0),
an estimate can be derived for the limit value of the amplitude of the FTF ([F()[) :
[F()[( 0) =
1
p
0
Q
0
Y
0
F
min(1, 1/) (2.12)
Assuming constant heat capacity (c
p
(T
b
T
f
) = QY
0
F
min(1, 1/)), the previous equation can be writ-
ten:
lim
0
[F()[ =
T
b
T
f
1 (2.13)
which can be a useful upper limit for F(). For typical methane ames, it lies between 1 and 10,
depending on the inlet temperature.
Local formulation
Conventional FTF model focus on the correlation between the volumic integral of heat release uctua-
tions with a reference point velocity uctuation. It implies that the ame is compact compared to the
characteristic wavelength of the perturbation. But although the ame is acoustically compact, it is often
not compact compared to the velocity uctuation characteristic lengths. For such cases, an extension of
these global models can be obtained by linking the correlation of the local heat release uctuation with
the reference velocity uctuation. The objective of this extension of the FTF model is therefore to give
the spatial distribution of the Flame Transfer Function. Experiments can also be used to provide the
same information by measuring unsteady heat release in 1-D sections (using a moveable slit (Varoquie
et al [152])) or in 3-D using local diagnostics (Poinsot et al [103]).This extension can give either 1-D or
3-D representations of the FTF :
For the 3D extension, the relation is :
1
pS
inj
T
(x, y, z) = F(, x, y, z)
(2.14)
where

T
(x, y, z) is the local fourier transform of the heat release uctuation. This model is
directly linked with the usual FTF model by :
[F()[ =
_
v
F(, x, y, z)dv
(2.15)
Arg(F()) = Arg
__
v
F(, x, y, z)dv
_
(2.16)
76
2.2 Models for FTF
For the 1D extension, the relation is :
1
pS
inj
T
(x) = F(, x)
(2.17)
where

T
(x) is now the Fourier transform of the integrated (over y,z) heat release uctuation. This
model is directly linked with the usual FTF model by :
[F()[ =
_
x
F(, x)dx
(2.18)
Arg(F()) = Arg
__
x
F(, x)dx
_
(2.19)
This model extension gives 1D curves of the ame transfer function which can be used in 1D
acoustic codes to gives eigen modes of reacting congurations where the ame cannot be con-
sidered as compact anymore (for example in the ICLEAC experiment of Varoqui e [151] or Truf-
n [149].)
Normalizations of FTF amplitudes
Two other normalizations both for the global and local formulations of the FTF are used in this work.
The dimensional FTF amplitude is also used.
The FTF can be normalized by the ratio between the mean velocity at the reference point and
the integral of the mean heat release over the whole volume leading to a non-dimensional FTF
amplitude. In other words :
F
norm
() =
u
(2.20)
This normalization is used for the analysis of global FTF in sections 4.2 and 4.4.
For the local formulation, the FTF amplitude can also be normalized by its maximum value in the
combustor. This normalization is introduced in section 4.1 for the analysis of the main features of
the response of 1D ames to inlet velocity forcing. It is described by the following relation:
[F
norm2
()(x, y, z)[ =
max
x,y,z
_
(2.21)
Finally, the dimensional amplitude FTF is also used in section 4.3 and 4.4 for the local analysis of
the ame response. The following relation links the amplitude of the FTF in this case with heat
release and velocity uctuations:
[F
dim
(x, y, z)[ =
(2.22)
77
Note that all the formulations for the amplitude of the FTF do not inuence the FTF phase which remains
equal to Arg
_

T
u
_
for the global analysis and to Arg
_
T
u
_
for the local one.
Pressure/velocity FTF model
All the previously exposed extensions for the FTF model heavily rely on the choice of the reference point.
Trufn and Poinsot [150] have shown that the distance between the reference point and the ame zone has
to be compact compared to the acoustic wave length to give relevant amplitude and phase values when
only the velocity is measured at the reference point. These values can then be used to reconstruct the
acoustic matrix of the combustor. This can be an issue for experimentalists since measuring a velocity
eld near the ame front requires expensive heat resistant probes. Trufn and Poinsot show that this
issue can be avoided by measuring also pressure uctuation at the reference point. In this case, the
global model has four parameters :
_
V
T
dv = F
u
()
+F
p
()
(2.23)
where F
u
, F
p
respectively represent the transfer function of the response due to the velocity and pressure
uctuations. When
_
V
T
is expressed as a function of

u
and

p
at the reference point, this reference

point can be anywhere and the difculty of the classical model (based on
_
V
T
versus

u
) disappears.
However there is a drawback to this approach : to determine F
u
and F
p
, two independent calculations
must be performed or two different forced experimental cases must be set : one forced at the inlet and
the other one forced at the outlet. Trufn and Poinsot show that in this case, the ame transfer function
can be used to reconstruct the global acoustic matrix of the combustion chamber linking pressure and
velocity uctuations upstream and downstream of the ame. To assess this achievement, in [150] the
acoustic matrix is reconstructed with two methods.
The rst method uses the pressure and velocities upstream and downstream of the ame front. The
acoustic matrix is recovered using usual acoustic relations.
The second method uses only pressure and velocity measurements upstream from the ame. Fi-
nally the FTF obtained from their model is used to reconstruct the acoustic matrix of the combus-
tion chamber.
The two methods give similar results and these results do not depend on the distance between the refer-
ence point and the ame front.
General remarks
All previously presented formulations (except the pressure/velocity FTF model) rely on the compacity
of the ame front compared to the acoustic wave length. If the ame is not compact, the obtained FTF is
not compatible with the PA approach. This means that there is a lower limit to the acoustic wavelength
78
2.2 Models for FTF
that can be handled properly by these models. For actual combustion chambers used in gas turbines, the
characteristic length of the ame is a few decimeters. This means that those models can be applied for
acoustic waves which have at least a wavelength (L) of a few meters. In hot gases (c = 900m.s
1
), the
frequency cut-off (f
off
) is therefore of a few hundred Hz only (f
off
=
c
L
). Using other models than the
pressure/velocity FTF model, the system identication of combustion chambers is therefore limited to
frequencies lower than a few hundred Hz.
79
80
Chapter 3
Measurement methods for FTF in LES
3.1 LES of FTF
This section describes how simulation can be used to measure FTF in ames (laminar or turbulent).
3.1.1 LES as the proper tool for FTF
Considering the measurement of turbulent FTF, the advantage of LES is clear. Turbulent FTF are mainly
needed for acoustic study purposes in complex geometries. Typical acoustic frequencies are of the order
of 500 Hz in gas turbines corresponding to a period of 2 ms which correspond to time scales that are
now affordable for LES even in complex geometries thanks to the use of unstructured meshes.
To determine FTF, one needs either the instantaneous volumic integral of the heat release uctua-
tion (to determine the global response of the ame) or the instantaneous eld of heat release uctua-
tions (to determine the local response of the ame). In this work, LES (AVBP code from CERFACS
(www.cerfacs.fr/cfd)) is chosen to get these elds.
3.1.2 Introducing acoustic waves through the burner inlet
To determine the transfer function of a burner, the usual procedure is to introduce an acoustic wave into
the burner (usually through the inlet) and measure the perturbation of heat release. The phase between
the incoming unsteady ow rate and the unsteady heat release is an essential ingredient of acoustic
approaches for combustor stability [32, 55, 69, 96, 107, 113]. However, the exact numerical procedure
to introduce acoustic waves in a computation is a difcult topic and may lead to numerical artefacts
[41, 105]. In subsonic compressible LES, each boundary must be specied in terms of mean conditions
(velocity or pressure for example) but also in terms of acoustic impedance. Many methods for such
MEASUREMENT METHODS FOR FTF IN LES
problems are based on characteristic methods [48, 107] but must be used with care: the well - known
non reecting conditions actually impose an impedance (which is often unknown) and can have a
strong inuence on the results [138]. For a simple laminar ame, Kaufmann et al [61] show that a proper
method to excite a combustion chamber is to pulsate the incoming acoustic wave and not the incoming
local velocity. This procedure is used here and avoids false numerical resonances when the chamber is
forced. The acoustic wave is therefore introduced into the combustion chamber through the inlet. Note
that when this forcing technique is used, the velocity evolution at the inlet is not xed anymore and
depends on the combination of the imposed wave (A
+
) entering the domain and the one coming back
(A
) from the burner. The inlet velocity and pressure are linked to the acoustic waves at the inlet though
the following relation:
u
=
1
c
(A
+
A
) (3.1)
p
= (A
+
+A
) (3.2)
That means that the inlet velocity uctuation can only be known before the run if the acoustic wave
coming from the combustor is negligible.
Another issue when forcing such a conguration comes from the amplitude of the waves reected at the
chamber outlet which must remain low in order to minimise self-excited oscillations. This is done using
the NSCBC boundary method [86, 105, 107] which allows the control of the acoustic impedances on
the boundaries.
3.2 Postprocessing methods for local FTF
This section presents four methods related to the identication of turbulent combustors. These four
methods are expected to give the local FTF. Table 3.1 summarizes their characteristics. Sections 3.2.1 to
3.2.4 present their principles.
Method Inlet forcing Inlet data Outlet data
Post processing
method
Advantage
TD None
Position of
reference
point
Mean ame
position
Measure ight time
No forced
computation
needed
HF-FFT
Single
frequency
u
(t)
T
(x, y, z, t)
Fast Fourier Trans-
form
Precise
WN-FFT
Filtered
white noise
u
(t)
T
(x, y, z, t)
Fast Fourier Trans-
form
Fast
WN-WH
Filtered
white noise
u
(t)
T
(x, y, z, t)
Wiener-Hopf inver-
sion
Fast
Table 3.1 - Characteristics of methods for system identication.
82
3.2 Postprocessing methods for local FTF
3.2.1 Time delay or Flight time method (TD method).
This method only focuses on the determination of the time delay between heat release and inlet velocity
forcing uctuations using a steady state CFD result [14]. As explained in the introduction, the delay
between these two uctuations is considered as the main information for the stability of the combustor.
This method consists of three main steps:
First a steady state simulation of the conguration is done (using RANS) or the mean eld of an
unsteady computation is post-processed (using LES or URANS).
The mean position of the ame is extracted from the eld of heat release.
Using the stream lines of the mean ow, the convection time between the inlet and the ame is
identied as the time at the intersection of these lines with the mean ame position. This provides a
probability density function of the delay between heat release and inlet velocity uctuations. Note
that the probability density function is therefore independent of the frequency: the same delay is
used for all frequencies.
This method is not as straigthforward as it seems: innite time delays can be obtained in recirculating
zones. Moreover, determining the age of ow particules as they move in the combustor often requires
to add Lagrangian formulations to the code or an additional conservation equation.
3.2.2 Harmonic forcing; FFT postprocessed method (HF-FFT method).
This method consists in introducing a downstream acoustic wave at a xed pulsation () as described in
section 3.1.2. It assumes that the response to a harmonic perturbation of the owis a harmonic uctuation
of heat release. The local ame transfer function is then obtained through the following relation:
F
dim
() =
(3.3)
where F
dim
compares the local uctuation of heat release to the uctuation of reference velocity,

T
is
the Fourier transform of the local heat release uctuation and

u
is the Fourier transform of the forced

inlet velocity.
The method to retrieve the FTF can be described as follows.
Harmonically forced Large Eddy Simulations of the conguration are done for a discrete set of
excitation frequencies.
3Delds of heat release uctuation are extracted as well as the reference (usually the inlet) velocity
uctuation.
3D elds of F
dim
() are obtained using FFTs of
T
(t) and u
(t) and constructing F

dim
()
through equation 3.3. F
dim
() is a local information about the location and the phase of the
ame response.
83
This method requires FFTs of
T
at all points of the combustor during its forced operation. In the
present work, this is obtained by using typically 100 snapshots with 10 snapshots per period which gives
a frequency resolution of 1/10th of the forcing frequency. Section 4.3 will present a comparison between
the TD, HF-FFT methods and experimental results. This method is expensive because it must be repeated
for a representative set of discrete frequencies (typically 10 to 50), leading to many forced LES.
3.2.3 White noise forcing; FFT postprocessed method (WN-FFT method).
This method consists in introducing a downstream ltered white noise at the inlet and to obtain the FTF
using one LES only. The local ame transfer function is then obtained through relation 3.3. The method
to retrieve the FTF for the concerned conguration can be described as follows.
A ltered white-noise forced Large Eddy Simulation of the conguration is done.
3Delds of heat release uctuation are extracted as well as the reference (usually the inlet) velocity
uctuation.
3D elds of F
dim
() are obtained using equation 3.3 and FFTs of
T
(t) and u
(t)
This method requires FFTs of
T
at all points of the combustor but is much faster than HF-FFT because
only one forced LES is required.
3.2.4 White noise forcing; Wiener-Hopf postprocessed method (WN-WH method).
This method using white-noise forcing and the Wiener-Hopf relation inversion was introduced in FTF
measurements by Schuermans et al. [131] and used for real devices by many authors [113, 110, 156,
130]. To use this method, the inlet has to be forced with a ltered white-noise including all frequencies
between 0 and a cut-off frequency. This method takes advantage of the Wiener-Hopf equation linking
the autocorrelation matrix of the inlet signal (inlet velocity), and the cross-correlation vector c :
h = c (3.4)
where h is the impulse response lter (i.e the lter corresponding to the unit response of the system).
However, in the present work, h corresponds to the white-noise response lter.
The method to retrieve the FTF(F
dim
) for the concerned conguration can be described as follows:
A LES forced with ltered white noise is performed.
The response signal (local heat release) and the reference signal (inlet velocity) are stored.
Knowing the white noise inlet velocity signal (s), the autocorrelation matrix can be determined:
i,j
=
1
M
N
l=L
s
li
s
lj
for i, j = 0, ..., L.
(M=NL+1)
(3.5)
84
3.3 Preliminary comparisons of postprocessing methods.
Using the inlet (velocity)(s) and the response (heat release)(r) signals, the cross-correlation vector
can be obtained :
c
i
=
1
M
N
l=L
s
li
r
l
for i = 0, ..., L. (3.6)
so that the equation 3.4 can be inverted to obtain h.
Using relation 3.7, the FTF is obtained from the lter h.
L
k=0
h
k
e
itk
= 1 +F
dim
() (3.7)
where F
dim
is the dimensional FTF given by the following relation :
F
dim
() =
(3.8)
The parameter L is the length of the lter and is critical to the validity of this analysis. tL represents
the time memory of the lter. It means that the biggest time delay
max
available is determined by the
value of L (
max
= tL). Attention has also to be paid not to choose L too big to avoid numerical
problems during the inversion of the matrix since its size varies like L
2
.
The greatest advantage of this method is that it gives full three dimensional elds of the amplitude and
the phase response of the ame. It is also able to give the response parameters at all frequencies between
0Hz and the cut-off frequency. This work will focus on the validation of this method against the single-
frequency forcing method (comparison WN-WH/HF-FFT (Section 4.4)).
Another interest of the WN-WH method compared to the HF-FFT method is that it can give the FTF at
all frequencies below the cut-off frequency with only one LES.
Preliminary comparisons are made to assess the validity of the post-processing methods for system iden-
tication described in Section 3.2. Two main tests are computed :
The rst type of identication procedures is based on FFT of response and reference signals. The
second type is based on the inversion of the Wiener-Hopf relation. Therefore, in a rst test, the
inuence of the signal properties on the response of the signal post-processing method (FFT/WH
inversion) is studied.
Then HF-FFT and WN-WH post-processing methods are compared on the identication of the
admittance of a diffuser.
None of these examples include combustion or FTF but they allow the verication of the post-processing
methods.
85
3.3.1 Validation of signal post-processing methods: FFT versus Wiener-Hopf inversion.
To test the validity of each signal post-processing method used for the identication, different simple
tests cases are done. For all tests, the reference signal is s(t) = cos(t) with = 2f and f = 100Hz.
The total signal is composed of ten periods with twenty points per period except for the second test where
the number of samples per period is changed. The response signal (mimicking heat release) is always
composed of a signal equal to r(t) = 2.3cos((t 0.004)) with = 2f and f = 100Hz, except for
the third test where the frequency of the response signal is shifted. The goal of this section is therefore
to verify in which cases these methods are able to get the known amplitude of the response (i.e. [F
dim
[ =
2.3) and its phase (i.e. Arg(F
dim
) = 2 = 2.5133 rad).
First test : inuence of a noisy response signal
Other frequencies are progressively added in the response signal to test the capacity of each method
to retrieve the information from a noisier signal. Table 3.2 and 3.3 sum up the differents frequency
components of the response signals. Note that signal 5 is very different from the reference signal and
that its component at 100 Hz is not the most important.
Table 3.4 sums up the amplitudes and phases obtained with both methods from the different response
signals. Both methods give good results with an error lower than 5% even for the signal 5. FFT results
are better, retrieving the exact amplitude and phase with an error lower than one percent.
Freq.(Hz) Reference signal : n, Signal 1: n, Signal 2: n,
100 2.3, 4.10
3
2.3, 4.10
3
2.3, 4.10
3
50 0, 0 0, 0 0, 0
180 0, 0 0.03, 2.10
3
0.4, 2.10
3
300 0, 0 0.02, 1.10
3
0.05, 1.10
3
500 0, 0 0.01, 5.10
4
0.2, 5.10
4
Table 3.2 - Composition of the different response signals : Part 1
Freq.(Hz) Signal 3: n, Signal 4: n, Signal 5: n,
100 2.3, 4.10
3
2.3, 4.10
3
2.3, 4.10
3
50 0, 0 1.5, 2.10
3
1.5, 2.10
3
180 1.5, 2.10
3
1.5, 2.10
3
1.5, 2.10
3
300 0.5, 1.10
3
0.5, 1.10
3
0.5, 1.10
3
500 1.0, 5.10
4
1.0, 5.10
4
3.0, 5.10
4
Table 3.3 - Composition of the different response signals : Part 2
86
Signal W-H inversion method: [F
dim
[, Arg(F
dim
) (rad) FFT method : [F
dim
[, Arg(F
dim
) (rad)
Reference Signal 2.229, 2.508 2.304, 2.508
Signal 1 2.3, 2.512 2.304, 2.508
Signal 2 2.31, 2.515 2.3, 2.507
Signal 3 2.34, 2.521 2.285, 2.504
Signal 4 2.229, 2.548 2.295, 2.507
Signal 5 2.23, 2.549 2.286, 2.505
Table 3.4 - Results : amplitude and phase obtained with both methods. Exact results : [F
dim
[ = 2.3,
Arg(F
dim
) = 2.5133rad
Second test : inuence of signal sampling
For this test, the discretisation of both source and response signals is lowered progressively from 20
points per period to the Shannon criterion limit of two time samples per period. The amplitude of the
response is still 2.3 and the time delay 4 ms, corresponding to a phase of 2.5133 rad. Table 3.5 sums up
the number of samples per period for all the signals used in this test. This table also show the results
given by both methods for the amplitude and the phase of the response. Table 3.5, Fig.(3.1) a) and b)
Reference signal Signal 1 Signal 2 Signal 3
Sampling (Points/Period) 20 10 4 5
FFT : [F
dim
[, Arg(F
dim
) (rad) 2.304, 2.508 2.3, 2.503 2.29, 2.507 2.302, 2.514
WH : [F
dim
[, Arg(F
dim
) (rad) 2.229, 2.508 2.3, 2.513 2.288, 2.498 2.289, 2.519
Signal 4 Signal 5 Signal 6 Signal 7
Sampling (Points/Period) 3 2.5 2.3 2
FFT : [F
dim
[, Arg(F
dim
) (rad) 2.354, 2.514 2.397, 2.496 2.404, 2.481 2.373, 2.454
WH : [F
dim
[, Arg(F
dim
) (rad) 2.3, 2.513 2.311, 2.522 2.3, 2.513 2.3, 2.513
Table 3.5 - Decreasing sampling for source and response signals : results in amplitude and phase. Exact results :
[F
dim
[ = 2.3, Arg(F
dim
) = 2.5133rad
show that both methods give good results even near or at the Shannon limit. They are robust to the
decrease of the time sampling. Note that for the Wiener-Hopf inversion method, the size of the lter (L
parameter) has been decreased with the time sampling to keep the same time memory of the system (see
Section 3.2). This test shows that these methods can be used even with poorly discretized signals.
Third test : inuence of a shift of the response frequency.
For this test, the frequency of the response signal is shifted progressively from 100 to 120Hz. The
total time of the response signal is kept constant and is equal to 0.1s which corresponds to a frequency
87
a) b)
Figure 3.1 - a) Amplitude of the FTF versus time sampling of signals (number of points per period)
b) Phase (rad) of the FTF versus time sampling of signals (number of points per period):
Theory FFT WH.
resolution of 10Hz. Since the reference signal frequency is 100Hz, the FFT is therefore supposed to
consider all frequencies lower than 110Hz as being 100 Hz for the resulting amplitude and phase. For
frequencies very different from 100 Hz , the amplitude and phase should decrease to 0 for both methods.
Table 3.6 gives the frequencies of all the response signals considered. It compares the results obtained
with both methods in amplitude and phase. Fig.(3.2) shows that both methods show a selectivity of about
Reference signal Signal 1 Signal 2 Signal 3
Frequency (Hz) 100 101 103 107
FFT 2.304, 2.508 2.27, 2.224 1.968, 1.654 0.87, 0.463
WH 2.229, 2.508 2.26, 2.195 2.02, 1.545 1.105, 0.252
Signal 4 Signal 5
Frequency (Hz) 110 120
FFT 0.022, 6.118 0.023, 6.126
WH 0.25, 5.627 0.233, 5.587
Table 3.6 - Increasing frequency of the response signal : results in amplitude and phase. Exact results :
[F
dim
[ = 2.3, Arg(F
dim
) = 2.5133rad
3Hz with an error lower than 10%. The residual amplitude at 120Hz is 1% of the signal amplitude for
the Fourier coefcient method and 10% for the wiener-Hopf inversion method. This test shows that the
Fourier method has a better selectivity than the Wiener-Hopf method. This means that for real signals,
close frequencies might pollute each other for short signals when the post-processing uses the Wiener-
Hopf inversion relation.
88
Figure 3.2 - Amplitude of the FTF versus frequency of the response signal : Exact FFT WH.
3.3.2 Admittance of a diffuser : HF-FFT versus WN-WH methods.
This section presents a preliminary validation of the WN-WH and HF-FFT methods against the theory
for the calculation of the admittance at the outlet of a subsonic diffuser. The admittance of a diffuser is
computed using AVBP and forcing the outlet section (Fig.3.3). Results are post-processed using either
HF-FFT or WN-WH. Even though this example is not a ame yet, it corresponds to an important issue
in combustion instability studies : the measurement of impedance of diffusers and distributors. And for
this example like for FTF computation, post-processing methods are crucial so that this case constitutes
a good accuracy check of the post-processing HF-FFT and WN-WH techniques. An other advantage of
this case is that a theoretical solution is available : here, the theoretical admittance can be evaluated with
NOZZLE, a code which solves the analytical relations of Marble and Candel [81]. Computations have
been done by M. Myrczik [89] during an internship at CERFACS. NOZZLE is a software written by
Lamarque [71]. It provides the impedance of the diffuser. The mesh used for this comparison is shown
in gure 3.3. Table 3.7 presents the geometry, mesh and boundary conditions of the diffuser.
Figure 3.3 - Mesh of the diffuser
Mean ow
The validation starts by making sure that NOZZLE provides the same values for the steady ow as
the LES code AVBP. In order to be able to compare the results obtained by AVBP and NOZZLE, it
is important to remind that NOZZLE uses a quasi-1D model in contrast to AVBP (2D). Therefore, the
values provided by the stabilized AVBP calculation are averaged for each section (here: averaging along
89
Geometry Value Boundary conditions Values
L
x
(m) 0,2 u
in
(
m
s
) 30
L
y
(m) 0,015 p
in
(bar) 0,9292059
n
x
101
in
(
kg
m
3
) 1,07478
n
y
11 T
in
(K) 300,0
1,399 r (J.kg/K) 288,19
In AVBP the walls are assumed to be adiabatic with zero normal velocity
Table 3.7 - Characteristics of the diffuser and boundary conditions
y) to nally compare them to the values obtained with NOZZLE. Note that the boundary conditions in
table 3.7 are imposed on the averaged values, which are then fed to the NOZZLE code.
Figure 3.4 presents the compared mean ow. The values obtained with NOZZLE coincide very well with
AVBP.
Figure 3.4 - Comparison between AVBP/NOZZLE for the mean ow in diffuser.
90
Admittance calculation
As shown in the previous section, the mean ows in AVBP and NOZZLE are the same. This section
presents now the calculation of admittance with three methods:
NOZZLE which is based on an analytical model.
LES computations with AVBP using the HF-FFT method.
LES computations with AVBP using the WN-WH method.
Figure 3.5 shows how the comparisons proceed. A major question for this calculation is which boundary
Figure 3.5 - Procedure for the comparison of post-processing techniques.
condition to select at the entrance of the diffuser. If the diffuser was choked, the admittance would not
depend on the compressor elements located upstream of the choked section. Unfortunately, most dif-
fusers are not choked and computing the diffuser outlet admittance requires to know its inlet admittance.
Here, in a rst step, one assumes a zero acoustic velocity at the entrance.
91
Thus, several calculations have been performed with the HF-FFT method. Every calculation starts
from the initial solution described in section 3.3.2.
An upstream acoustic wave A
e
it
= ip
0
e
it
is injected at t = 0 in the exit plane. Once the ow is
established (after a few acoustic times), the admittance calculation can be done using u
and p
in the
exit plane at the forcing frequency. 18 frequencies from 0 to 3000 Hz are used for the HF-FFT method.
For the WN-WH method a low-pass-ltered white-noise (0 < f < 4000Hz) is injected at the outlet
of the domain. One computation of 50 ms is done in order to match a frequency resolution of 20Hz.
In this case, the WN-WH method is approximately 10 times faster than the HF-FFT method because it
requires only one unsteady computation instead of 18.
Fig.(3.6) shows the comparison between the three methods (WN-WH/ HF-FFT/ NOZZLE). Results
obtained with the HF-FFT method match very well the results obtained with NOZZLE. With the WN-
WH method, one nds the peaks around the same frequencies, with a phase shift for higher ones. In
addition, between 1000 and 1200 Hz, the results of WN-WH deviate signicantly from the analytical
solution.
Figure 3.6 - Comparison WN-WH/HF-FFT/NOZZLE for the diffuser (real and imaginary part of admittance). The
solution obtained by NOZZLE is the reference solution.
As mentioned before in section 3.2.4, the quality of results obtained with the WN-WH method,
depends on the lter size L used for the Wiener-Hopf signal identication method. The inuence of the
different L (190, 170, 130) is examined. Fig.(3.7) shows that the values obtained with L = 130 differ
the most from the NOZZLE solutions. The lter appears to be too small and hence triggers only small
delay times.
L = 170 and L = 190 seem to be appropriate quantities. As shown on Fig. 3.7, it is not obvious which
is the best one; because for one range of frequency L = 170 gives better results, but for another range
L = 190 ts better.
92
The overall conclusion is that WN-WH methods are faster than HF-FFT methods but they will also
require more care.
93
Figure 3.7 - Comparison of three WN-WH results for L= 190, 170 and 130 for the diffuser (Top: Real part of
admittance; Bottom: Imaginary part of admittance)
94
Chapter 4
Transfer functions of ames
Results presented in this chapter have been obtained in four congurations (see Fig.(4.1)):
Conguration A : a laminar planar premixed ame (1D)
Conguration B : a laminar conical ame (2D axi-symmetric)
Conguration C : a turbulent gas turbine burner installed in a round combustion chamber at EBI
Karsruhe [14] (3D)
Conguration D : a turbulent gas turbine burner installed in a sector of an annular gas turbine,
representative of a real machine (3D)
The 4 methods presented in chapter 3.2 will be applied to some or all of these set-ups:
Time Delay or Flight Time (TD) method
Harmonically forced ; FFT post-processed (HF-FFT) method
White Noise forced ; Wiener-Hopf post-processed (WN-WH) method
White Noise forced ; FFT Post-processed (WN-FFT) method
Table 4.1 summarizes the different tests which were performed. In most cases, two methods were used
to allow comparisons. The presentation is organized by experimental set-ups. For each case, the cong-
uration is described rst and results on FTF follow.
TRANSFER FUNCTIONS OF FLAMES
Figure 4.1 - Congurations used for FTF tests.
Method TD HF-FFT WN-WH WN-FFT
Conguration A X
Conguration B X X X
Conguration C X X
Conguration D X X X
Table 4.1 - Summary of methods and congurations.
96
4.1 Conguration A : laminar planar premixed ame (1D)
4.1.1 Description
The goal of this section is to determine the main features of the 1D ame response to an incoming
acoustic uctuation in terms of amplitude and time delays. A 1D ame is forced with an incoming
acoustic wave at 200Hz with an amplitude of 0.5m.s
1
. The reference point for velocity is the domain
inlet so that the distance between this point and the mean ame front is 1mm. The mean inlet velocity is
equal to the laminar ame speed so that the ame is stable when there is no forcing. The structured mesh
is composed of 1000 quadrilateral elements and 2002 nodes. Its dimensions are 2.10
2
m 2.10
9
m.
Figure 4.2 sums up the numerical experiment and shows the location of three sample points A, B and
C. The ame width is 0.5mm and the pulsated ame extension (the zone over which the ame moves
during forcing) is 1mm.
Figure 4.2 - numeric experimental setup.
Signals of heat release are collected along a line crossing the ame. Figure 4.3 gives the heat release
signals at points A, B and C (normalized by the maximum value) as well as the reference inlet velocity.
It is interesting to note that signals of heat release are not harmonic because the pulsated ame extension
is greater than the ame width. This particular feature will be encountered in 3D turbulent congurations
since the pulsated velocity amplitude at the ame front will be even higher than the present value.
Only the HF-FFT post-processing technique is applied to the instantaneous signals of velocity and heat
release collected along the line crossing the ame to determine the general behavior of the local response
of the ame.
4.1.2 Amplitude
Figure 4.3 shows that the maximum instantaneous heat release is the same for all three points. But the
amplitude of the response ([F
norm2
[ dened in Eq.(2.21)) differs between A and B and C, as presented on
97
Figure 4.3 - a) Signals of heat release at points A, B and C. b) Reference velocity signal
table 4.2. The response at point C which is in the mean ame is almost negligible because the frequency
of the heat release uctuating signal at that point is not 200Hz. Actually, the ame passes exactly twice at
that location during a period, so that the response at the mean ame position is mainly at 400Hz. The 200
Hz component increases as one approaches from the maximum extension position of the forced ame
front as shown on gure 4.4.
Point A Point B Point C
Normalized response amplitude 0.462 0.555 0.068
Table 4.2 - Normalized amplitude of the response at points A, B and C (HF-FFT)
4.1.3 Time delay
At points A and B, the ame has almost the same amplitude of response, but very different time delays.
Point B is closer to the forced inlet. One therefore expects that the time delay at that point will be smaller.
It is not the case and table 4.3 sums up the time delays for points A and B . Point A has the lowest time
delay, and since this point is farer than B from the inlet, it is interesting to explain this result.
In the present case, the pulsated ame zone is compact when compared to the acoustic wavelength which
98
Figure 4.4 - Normalized amplitude of the response ([F
norm2
[) along the axial line.
is 1.7m long. Therefore, the velocity uctuation at point C (which is at the mean ame position) is
almost in phase with the inlet velocity. When the velocity at the inlet increases, it pushes the ame from
C to A. The ame rst reaches that point before the inlet velocity decreases and comes back to C. After
more than half a period, the ame reaches B. In this case the ame mean position therefore separates
two differents zones of time delays with a discontinuity of half the time period. Figure 4.5 represents the
values of time delays along the axial line crossing the mean ame position. This type of behavior will
also be observed in 3D cases where time delays exhibit abrupt changes at the mean ame position.
Point A Point B
Time delay (ms) 1.228 3.761
Table 4.3 - Time delays at points A, B
99
Figure 4.5 - Time delay (ms) of the ame response along the axial line.
100
4.2 Conguration B : laminar V ame (2D axi-symmetric)
This conguration has been extensively studied experimentally by Le Helley [52] and numerically by
Kaufmann [61] and Trufn [149]. This burner corresponds to a laminar ame. The structured mesh
used for this study is axisymmetric and takes into account for a large part of the feeding line for acoustic
reasons (see Figure 4.6). It consists of 19189 nodes and 9318 quadrilateral elements.
Figure 4.6 - Laminar conguration B. Le Helley [52]
This conguration is used here to compare FTF obtained using either HF-FFT, WN-FFT and WN-WH
methods (see section 4.2.1). The conguration is forced at ve frequencies between 100 and 500 Hz.
norm
)
Computations have been made at CERFACS by Patrick Schmitt and post-processing has been done
by the thermodynamic group at Munich University. Only a global comparison will be made for this
conguration.
Amplitude
Figures 4.7 presents the amplitudes of the FTF ([F
norm
[) obtained with HF-FFT, WN-FFT and WN-WH
methods. The three methods reasonably agree for all frequencies. Yet, HF-FFT and WN-WH method
match very well and the WN-FFT method gives a noisy response which shows undesired peaks. Note
that all methods give values of [F
norm
[ above unity for all frequencies. This means that the relative
uctuation of the integrated heat release is bigger than the relative uctuation of inlet velocity. For
example, if the amplitude of the uctuation of the inlet velocity represents 10% of the average inlet
velocity, the uctuation of the total heat release in the system represents 11% of the mean power released
by the ame at 100Hz and 16% at 400Hz.
Using Eqs.(2.10) and (2.20), the amplitude [F[ of the normalized FTF at 100Hz can also be estimated.
It is equal to 8.02. This value can be compared to the asymptotic value given by Eq.(2.13). With a mean
temperature of 2387K in burnt gases and a temperature of 300K at the inlet, the asymptotic value of the
interaction index for low frequencies is equal to 6.96. The value for [F[ at 100Hz therefore overestimates
101
by 15% this theoretical amplitude.
Note that in this case, the amplitude response is not dependant of the lter size for the Wiener-Hopf
inversion and the black curve represents the mean response for L parameter varying from 20 to 40.
Figure 4.7 - Amplitude of FTF : HF-FFT WN-FFT WN-WH.
Phase
Figure 4.8 compares the times delays (Arg(F
norm
)/(2)) obtained with the three different methods.
It shows that HF-FFT and WN-WH methods give almost the same time delay for all frequencies. The
WN-FFT method gives strongly varying results at least for frequencies lower than 200 Hz.
It is interesting to note that for all methods the global time delay of the FTF is not constant over the
frequency range for this conguration. This implies that the heat release uctuation can not be due only
to a pure convective phenomenon. With values for the time delay varying from 0.05 ms to 0.3 ms, the
heat release signal goes from an almost in phase uctuation to a phase of more than /4.
102
Figure 4.8 - Time delay of FTF : HF-FFT WN-FFT WN-WH.
103
4.3 Conguration C : turbulent burner in cylindrical chamber (3D)
The burner used for this study is a modied version of a Siemens hybrid burner used here at atmo-
spheric pressure (maximum power = 400 kW). The burner (Fig. 4.9) is mounted on a cylindrical com-
bustion chamber (Fig. 4.10). The non-reacting and reacting ows (Fig. 4.11) for this burner installed in
a slightly different combustion chamber have been computed in previous studies and successfully com-
pared to experiments in terms of axial and tangential velocities (mean and RMS) and mean temperature
elds [137, 138]. The mesh used in this study contains 311047 nodes and 1777730 elements.
This conguration is used in the present work to compare time delays obtained with the TD and HF-FFT
methods.
Figure 4.9 - Burner geometry.
This section presents a comparison of time delay of turbulent FTFs obtained with TD and HF-FFT
methods. Both results are successfully compared with experimental results in terms of time delay. This
section mainly shows that the TD method fails to predict the global delay between heat release and
forcing velocity uctuations. Besides, it also shows that the response of the turbulent ame is highly non
homogeneous and the HF-FFT method gives access to the spatial variations of the turbulent FTF both
for the amplitude and phase. It therefore indicates locations where the ame responds the most to the
incoming acoustic wave depending on its frequency.
This section follows almost exactly the paper published in Journal of Turbulence, 6(21):1-20, 2005.
104
Figure 4.10 - Burner mounted on a circular combustion chamber
Figure 4.11 - Iso-surface of heat release colored by axial velocity.
105
4.3.1 Introduction
Being able to predict the forced response of a combustion device is a mandatory step to design stable
burners, which exhibit as little sensitivity to acoustic waves as possible. Indeed, ame acoustic coupling
is a phenomenon which is found in multiple combustors[155, 114, 138, 133]: it creates a considerable
industrial risk if it cannot be avoided at the design stage. However, predicting ame / acoustics coupling
is still a challenge [65, 93, 150]. Only sophisticated numerical methods can be used to characterise
burner responses in real geometries while at the design stage. LES (Large Eddy Simulations) is now
the most precise numerical tool for turbulent ames as shown by numerous recent examples [138, 34,
57]. However, the validation of LES as a tool to study acoustics / ame coupling is far from complete.
Few comparisons including experiments and simulations can be found in the literature on LES of ame
transfer functions in complex congurations. The main reason for this, is that ame / acoustics studies
require the computation of turbulent combustion (something that LES can do reasonably well) but also
of acoustics. Even in codes dedicated to acoustics in ames where turbulent combustion modelling is
extremely simplied, the prediction of ame responses to acoustic perturbations is a difcult topic [114,
65, 30, 15, 17, 107]. Therefore coupling such acoustic analysis with state-of-the-art LES is still an open
eld of investigation and one objective of this study is to investigate this issue.
This study shows howexperiments and Large Eddy Simulations (LES) can be used in realistic geometries
to provide one parameter controlling ame stability: the ame response to acoustic waves entering the
burner. This response is an essential ingredient of most acoustic tools for ame / acoustics codes. It is
characterised through the transfer function between unsteady heat release in the chamber and oscillating
inlet velocity. Using simultaneously LES and experiments also allows an investigation of the various
coherent structures found in swirling combustors [56]:
It is well known [50, 10] that non-reacting swirled ows exhibit strong hydrodynamic modes called
PVC (Precessing Vortex Core). These spiral modes induce a rotation of the swirl motion axis and
have often been suspected of generating instabilities when combustion is activated even though
certain recent LES [138] actually suggest that the PVCs are sometimes damped when combustion
starts.
When a combustor (swirled or not) enters a self-excited mode or when it is forced at large am-
plitudes, another vortical structure appears: it has a mushroom shape in two-dimensional burn-
ers [106] and a ring shape in swirled axisymmetric burners [15]. This structure is typical of impul-
sively starting jets and appears only at sufciently large amplitudes of the velocity uctuations.
Determining when these structures are found in real combustors and what their impact can be, is still an
open eld of investigation which can be addressed with LES and advanced experimental diagnostics as
proposed here. For the present study, the target conguration is a large-scale industrial turbine burner
(400 kW) to be as representative as possible of real combustion devices. This partially premixed system
was studied both experimentally (at Univ. Karlsruhe) and numerically (at CERFACS).The experimental
conguration is described in Section 4.3.2. The LES tools and the numerical methodology used to
introduce acoustic waves through boundaries are then presented (Sections 4.3.3, 4.3.4) before describing
results in terms of mean ow (Section 4.3.5, 4.3.6). The coherent structures appearing during forcing
are discussed in Section 4.3.7 while a comparison of LES and experimental data is performed in Section
106
4.3.8 in terms of phase between inlet velocity and global heat release. Finally, a local analysis of ame
response is described in Section 4.3.9 to compare LES results with other estimates of combustion phases
used in Reynolds Averaged (RANS) codes [65] which are based on a measurement of ight times along
uid trajectories between burner inlet and combustion zone.
4.3.2 A double-swirler partially premixed burner
The burner used for this study is a modied version of a single Siemens hybrid burner used here at
atmospheric pressure (maximum power = 400 kW). This burner (Fig. 4.9) is mounted on a cylindrical
combustion chamber (Fig. 4.10). The conguration used for experiments is described on Fig. 4.12. Fully
premixed methane / air is injected through two coaxial swirlers (diagonal and axial) at an equivalence
ratio of 0.8 for the axial and 0.5 for the diagonal swirler. The ow rates injected in the combustion
chamber are respectively 0.018kg.s
1
for the axial and 0.162kg.s
1
for the diagonal swirler. The pul-
sating unit is a rotating valve modulating the diagonal mass ow rate up to amplitudes of 30 percent. The
unsteady velocity induced by forcing is measured using a hot wire probe located in the diagonal swirler
(Fig. 4.12) while the global unsteady heat release is evaluated through a measurement of the OH* radi-
cal. Since 90% of the injected mixture is at a constant equivalence ratio, the OH* signal can be reliably
linked to the unsteady heat release [79]. Figure 4.13 displays phase-locked OH* images obtained using
an ICCD camera and an appropriate OH* lter ( = 307.8nm). Typical ame shapes are observed
without forcing or with low-frequency forcing (left), and with high-frequency forcing (right). The ame
is quasi-steady as long as the frequency of the excitation remains below 50Hz. Beyond this limit, ring
vortices interact with the ame imposing their axis-symmetrical shape to the ame.
4.3.3 Large Eddy Simulations for gas turbines
Recent studies [138, 3, 19, 29, 102, 124, 117] have shown that Large Eddy Simulations (LES) are pow-
erful tools to study the dynamics of turbulent ames and especially their response to acoustic forcing.
However, the extension of LES methods to complex geometries creates new problems to adapt high-
precision numerical techniques (required for LES) to arbitrarily complex meshes. Here a parallel LES
solver called AVBP (see www.cerfacs.fr/cfd/) is used to solve the full compressible Navier Stockes equa-
tions on hybrid (structured and unstructured) grids with second or third-order spatial and temporal ac-
curacy [29]. Subgrid stresses are described by the WALE model [92]. The subgrid ame / turbulence
interaction is modelled by the Thickened Flame (TF) model [3, 28]. The validity of the TF model used
here has been checked in many recent papers [138, 121] and will not be discussed here. For the present
application, methane / air combustion is modelled using six species (CH
4
,O
2
, CO
2
, CO, H
2
O and N
2
)
107
Figure 4.12 - Experimental setup.
quasi-steady ame ring vortices
Figure 4.13 - Instantaneous Phase-locked OH* images of the ame structure.
108
and two reactions [137]. These reactions are :
CH
4
+ 3/2O
2
=> CO + 2H
2
O (4.1)
CO + 1/2O
2
<=> CO
2
(4.2)
The forward rate of the rst irreversible reaction is:
q
1
= AY
0.9
CH
4
Y
1.1
O
2
exp(T
1
a
/T)
and the rate of the second reaction is:
q
2
= B[Y
CO
Y
0.5
O
2
(1/K
y
) Y
CO
2
] exp(T
2
a
/T)
where A = 6.1.10
8
mol.m
3
.s
1
, B = 6.1.10
5
mol.m
3
.s
1
, T
1
a
= 17613 K, T
2
a
= 6038 K and
K
y
(T) is the equilibrium constant of reaction (2).
The mesh used for the burner of this study contains 1.7 million elements. The characteristic size of the
mesh in the combustion chamber is 3mm which has to compared to the laminar ame thickness (1.5mm).
The simulations were achieved in Centre Informatique National de lEnseignement Superieur (CINES)
and the average computing time needed is around 10 hours on 64 processors on SGI 03800 for one period
at the forcing frequency of 120Hz.
4.3.4 Introducing acoustic waves through the burner inlet
To determine the transfer function of a burner, the usual procedure is to introduce an acoustic wave
into the burner (usually through the inlet) and measure the perturbation of heat release. The phase
between the incoming unsteady ow rate and the unsteady heat release is an essential ingredient of
acoustic approaches for combustor stability [107, 32, 55, 69, 96, 113]. Experimentally, this is usually
done with loudspeakers forcing the ow upstream or downstream of the combustion chamber. At high
chamber powers, however, rotating valves are required as done here. Because of the cost of such devices
and the danger of pulsating large ow rates of premixed gases, performing the same task with LES is
an obvious alternative path. However, the exact numerical procedure to introduce acoustic waves in a
computation is a difcult topic and may lead to numerical artefacts [105, 41]. In subsonic compressible
LES, each boundary must be specied in terms of mean conditions (velocity or pressure for example)
but also in terms of acoustic impedance. Many methods for such problems are based on characteristic
methods [107, 48] but must be used with care: the well - known non reecting conditions actually
impose an impedance (which is often unknown) and can have a strong inuence on the results [138]. For
a simple laminar ame, Kaufmann et al [62] showthat a proper method to excite a combustion chamber is
to pulsate the incoming acoustic wave and not the local velocity. This procedure is used here and avoids
false numerical resonances when the chamber is forced. Another issue when forcing such a conguration
comes from the amplitude of the waves reected at the chamber outlet which must remain low in order
to minimise self-excited oscillations. This is done using the NSCBC boundary method [107, 105, 86]
which allows the control of the acoustic impedances on the boundaries.
109
4.3.5 Reacting steady ow
Since the focus of this section is the forced response of the burner, no experimental results will be pro-
vided for the unforced case. The non-reacting and reacting ow for this burner installed in a slightly dif-
ferent combustion chamber have been computed in previous studies and successfully compared to experi-
ments in terms of axial and tangential velocities (mean and RMS) and mean temperature elds [138, 137]
this validation is not repeated here.
The reacting case corresponds to a global equivalence ratio of 0.51, air owrate of 180 g/s, a Reynolds
number of 120000 (based on bulk velocity and burner diameter) and power of 277 kW. A 2D snapshot
of the axial velocity eld (Fig. 4.14) shows a strongly turbulent ow. The swirled motion creates a large
recirculation zone in the centre of the combustion chamber and a circular outer recirculation zone (which
appears as two separates zones on the 2D axial cut). An iso-line of temperature (1100K) reveals the ame
position close to the burner mouth, stabilised both on the inner recirculation zone created by swirl and on
the outer recirculation zone. A temperature isosurface at 1000 K (Fig. 4.15) in an unsteady non-pulsated
case shows the typical ame shape: the edges of the ame are disturbed but the central zone, near the
axial inlet, is very stable due to the injection of richer premixed gases (equivalence ratio of 0.8) through
the axial swirler.
Figure 4.14 - Example of instantaneous eld of axial velocity in the axial plane, iso-line of temperature
(1100K). Unforced ow.
110
Figure 4.15 - Snapshot of temperature isosurface (1000 K) coloured by axial velocity. Unforced ow.
111
4.3.6 Forced reacting ow
In the experiment, the combustor was forced at frequencies changing from 10 to 120 Hz while the LES
were performed at 80, 120 and 250 Hz.
1
The inlet forcing immediately drives a heat release oscillation. Figure 4.16 compares evolutions of
heat release, diagonal inlet velocity and total mass of fuel in the chamber for the 120Hz case. The
inlet velocity (circles) and the enclosed fuel mass (deltas) oscillate sinusoidally, perfectly correlated
to the inlet acoustic pulsation. Heat release exhibits more nonlinearities, especially between instants
= 0 and = /2. The fast decreasing of the heat release occurs when the fresh gases previously
injected burn out. This main event during the cycle of excitation has been observed experimentally
for pulsated and self-excited combustors at the origins of the study of combustion instability. mettre
des refs. The ame shape (visualised by an isosurface of temperature) is displayed for four phases of
the cycle (Fig. 4.17) ( = 0, /2, , 3/2). The phase = 0 corresponds to the minimum value
of the axial velocity whereas = corresponds to the maximum value of the axial velocity. The
ame surface varies through the cycle and exhibits various complex structures especially at = 3/2
(Fig. 4.17d) where the ring vortex dislocates into smaller structures. The total heat release oscillates
from 0.7 to 1.3 times its steady value (Fig. 4.16) at of the inlet velocity. The ame goes through phases
of rapid expansion (between instants = and = 3/2 on Fig 4.17 for example) during which
ring vortices created by the forcing distort the ame front and expand it. Later in the cycle, the ame
shrinks very fast (between instants = 0 and = /2 on g 4.17) as already pointed out on g 4.16.
4.3.7 Coherent structures
LES reveals that two coherent structures determine the ame shape and the structure of the ow during
forcing. Figure 4.18 displays the classical ring vortex structure observed in this conguration. These
structures are displayed for four phases of the cycle (Fig. 4.18) ( = 0, /2, , 3/2). The phase = 0
corresponds to the minimum value of the axial velocity whereas = corresponds to the maximum
value of the axial velocity. The excitation creates a large axisymetrical ring which is convected by the
mean ow and interacts with the ame, increasing the ame surface as it passes through. This ring is
visualised here using an iso-surface of the detection criterion of Hussain and Jeong which is based on the
second invariant of the velocity tensor [59]. When the values of this criterion is positive, the rotation rate
is greater than the deformation rate, which implies the presence of a coherent structure. LES shows that
this ring is coherent and axisymmetric. At times = and = 3/2, the structure is ring-shaped
and well identied in Figure 4.18. This ring structure directly shapes the ame surface leading to the
mushroom-shaped ame of Figure 4.17 = and = 3/2. At instants = 0i and = /2
however, the ring becomes unstable and loses coherence. At instant = /2 , it almost disappears.
A second structure generated by the axial forcing is the precessing vortex core. This instability is typical
of non-reacting swirling ows [80] but was not observed for the unforced reacting case. Visualising an
1
At low forcing frequencies, the ame response is quasi-steady. Moreover, LES of low frequency cases become expensive
as characteristic physical times increase. This explains why LES were not performed at frequencies smaller than 80 Hz.
112
Figure 4.16 - Heat release (squares), inlet normal velocity (circles) and enclosed fuel mass (deltas) uctuations in
percents of mean values versus phase angle for the 120Hz LES.
113
= 0 = /2
= 3/2 =
Figure 4.17 - Snapshots of temperature isosurface (1000 K) coloured by the axial velocity for four different
phases of the cycle : f = 120Hz.
114
iso-surface of the vortex criterion (Fig. 4.19) in an inner cone excluding the large vortex ring of Fig. 4.18
reveals a coherent structure rotating around the chamber axis. The frequency of this precessing structure
is independent of the pulsation frequency and remains around 75 Hz.
115
(a) = 0 (b) = /2
(d) = 3/2 (c) =
Figure 4.18 - Snapshots of vortex criterion coloured by the temperature (Hussain [59]) for four different phases of
the cycle : f = 120Hz.
116
(a) = 0 (b) = /2
(d) = 3/2 (c) =
Figure 4.19 - Snapshots of vortex criterion coloured by the axial velocity visualising the precessing vortex core
for four different phases of the cycle : f = 120Hz.
117
4.3.8 Comparison of LES, RANS and experiment : global transfer function
Figure 4.20 compares numerical predictions and experimental measurements of the ame transfer func-
tion using the n model [107, 32]. This model, commonly used when dealing with reacting system
identication, is based on a rst harmonic analysis. Therefore it does not account for non-linear phe-
nomena which could be relevant in case of self-excited ames congurations [133, 75]. This model links
the Fourier transform of the volume-integrated heat release uctuation

T
to the Fourier transform of
the velocity uctuation at the inlet u
by :
ne
iw
=
1
S
D
p
0
(4.3)
where w is the forcing pulsation (w = 2 f), p
0
is the mean pressure in the combustion chamber and
S
D
the diagonal swirler inlet surface of fuel injection. Note that the phase is directly calculated from
, by multiplying by w. As no time dependent variables are available in RANS results, is extracted
using the ight time method described in subsection 4.3.9. Experimental results are given by the phase
comparison between the signal given by a photomultiplier (with a OH* frequency lter) which is directly
linked to the heat release, and a hot wire probe signal at the inlet of the combustion chamber. Values of
the LES phase are in good agreement with experimental values (Fig. 4.20). The experimental phase
between heat release and velocity uctuations is of the order of 3.5 rad at 120 Hz.
No comparison of the amplitudes of the Flame Transfer Function given by the HF-FFT method and the
experiment is provided. The reason is that the amplitude of the FTF given by the HF-FFT method greatly
changes with the acoustic boundary conditions. Since the experimental acoustic boundary conditions are
unknown, no trustful amplitude can be obtained with the HF-FFT method.
Experiment LES RANS
(rad) -3.5 -3.2 -1.1
Table 4.4 - Comparison of phases (rad) for the 120Hz case.
4.3.9 Comparison of LES, RANS and experiment : local transfer function
The previous subsection has shown differences in terms of global phase between LES, RANS and ex-
periments. To analyse these differences, local transfer function must be studied. The usual n
w
-
w
model
links inlet velocity perturbations to global heat release providing one delay
w
(expressed here as a phase
w
=
w
w) and one interaction index n
w
. In simulations, it is possible, however, to construct a map
of local n
w
-
w
values linking the inlet velocity perturbation to the local heat release variation to have a
local phase
w
(x, y, z) and a local index n
w
(x, y, z). These quantities are dened by:
n
w
(x, y, z) = MOD(

) (4.4)
118
Figure 4.20 - Comparison of LES, RANS and experiment phases.
w
(x, y, z) = ARG(

T
u
) (4.5)
where

T
and

u
are the Fourier transforms of the local heat release and of the inlet velocity uctuations
respectively. MOD and ARG are respectively the real and imaginary part of the complex Fourier
coefcient of the transfer function at pulsation w.
Two methods can be used to evaluate n
w
and
w
:
direct method (using time-resolved elds of the forced ow): using the denitions of equations 4.4
and 4.5,

T
(t, x, y, z) and u
(t)
inlet
can be used to evaluate
w
(x, y, z) and n
w
(x, y, z) for each
forcing frequency. This method is possible when LES forced elds are available as done here.
ight time method (using averaged elds of the unforced ow): quite often,
w
is extracted from
averaged elds, simply by measuring the time needed for one particle to travel from the diagonal
swirler inlet to a given location in the combustor. In this case, convection delays to the ame front
are evaluated for example by picking ight time values to the isosurface Y
CH4
= 0.004 (where the
maximum mean heat release occurs.). This method is common when RANS codes are used and
no forced response has been computed. It is based on the calculation of the dispersion of a passive
scalar in a stationary eld. More detailed information about the approach is given by Krebs et al.
in [65]. It can also be tested here using the averaged elds given by LES.
To avoid ambiguities due to the mean ow used for the ight time method, the averaged LES eld
was used and no direct RANS code was run. This ensures that the ight time and the direct methods
119
are applied on exactly the same ows.
Figure 4.21 shows a longitudinal cut coloured by the amplitude response at 120Hz using the direct
method. Since the area including the mean position of the ame mainly responds at twice the frequency
of excitation, the two ame-shaped like structures show the maximum extension of the ame through a
cycle of excitation (where the ame only goes once during a period of excitation). The ame in front of
the axial swirler has the strongest response as shown by the high n
w
values in this zone (Fig 4.21). This
may be due to the equivalence ratio difference between the two ames, the richer one producing more
heat release and therefore more heat release uctuation when excited by the acoustic wave.
Figure 4.22 displays the same cut coloured by the phase (
w
(x, y, z) given by equation 4.5) between
heat release and inlet velocity uctuations and obtained by the direct method. It appears that the ame
does not move in a compact way but rather gets stretched by the incoming acoustic excitation.
Figure 4.23 displays again the same axial 2D cut coloured by the phase
w
obtained with the ight
time method. It is assumed in this analysis that only convective time lower than two periods of
excitation are worth being considered (i.e. 16.6ms) [65]. It can be observed that certain regions of the
mean ame are not reached by the ow within two periods after the start of the excitation. This implies
that the heat release uctuation is not linked to the convection of an initial perturbation but is the result
of a complex interaction between acoustics and the ow in the chamber. The resulting structures may
appear in the recirculation zone (located in the centre of the combustion chamber (Fig 4.14)) before any
purely convective phenomenon reaches it.
Figure 4.24 shows the isosurface of control (Y
CH4
= 0.004) coloured by
w
(x, y, z) obtained with the
direct method. It can be seen that the axi-symmetry of the ow is conserved in terms of combustion
phase.
Figure 4.25 shows the isosurface of control coloured by
w
(x, y, z) obtained with the ight time
method. The phases obtained with this method still have an axisymmetrical repartition. As already done
for the axial 2D cut (Figure 4.23), non-represented phases correspond to convective times greater than
two periods of excitation.
Figure 4.26 compares the PDF of phases and its cumulant measured either by the direct method
necessarily issued from LES (thin line) or by the ight time method (thick line) on the isosurface
of control Y
CH4
= 0.004. The PDF of phases obtained by LES exhibit a few peaks around -3.5 and
0 rad while the ight time method gives a most probable value of
w
around -1.1 rad. This means
that according to this last method the heat release uctuation should be in phase with the inlet velocity
uctuation signal which is incompatible with the results of Figure 4.16 or with the experimental data .
Figure 4.26 is obtained from unweighed phases but it is reasonable to expect that points where
large unsteady heat release takes place, will have the largest effect on the instability and that they should
account for a larger part in the average phase. This means that another method to post-process phases is
to weigh them using the local n
w
index value.
120
Figure 4.27 compares the two weighted PDF
2
of phases using the following weighted phase:
P
w
(
0

w

0
+d
0
) =
n
d
0
1
norm
i
N
1
norm
i
in the region dened by:
n
w
(x, y, z) >
n
w
(x, y, z)
max
100
n
d
0
is the number of mesh points for which
0

w

0
+d
0
, N is the total number of mesh points
in the volume and
norm
i
= Int(
100 n(x
i
, y
i
, z
i
)
n
max
)
One can observe that the two weighted PDF (Figure 4.27) do not match. Although the ight time
method shows a wide maximum between -3.5 and 0 rad, the phases obtained with equation 4.5 have
an evident maximum at -3.2. This last most represented value of
w
is in agreement with a global heat
release uctuation (i.e.. integrated over the whole volume) out of phase with the inlet velocity uctuation
(Figure 4.16 ). Clearly in this example, the ight time method does not give accurate results.
2
Note that this normalisation is usually not available in ight time method, because the local n
w
values of the ame
transfer function are needed.
121
Figure 4.21 - Longitudinal cut coloured by response amplitude n
w
(x, y, z) at 120Hz. (Direct method)
122
Figure 4.22 - Longitudinal cut coloured by phase
w
(x, y, z) (rad) between heat release and inlet velocity
uctuations (wrapped between -2 and 0). iso-line Y
CH4
= 0.004. (Direct method)
Figure 4.23 - Longitudinal cut coloured by convection phase for diagonal injection (wrapped between -2 and
0). iso-line Y
CH4
= 0.004. (Flight time method)
123
Figure 4.24 - Isosurface of Y
CH4
= 0.004 coloured by phase (rad) between heat release and inlet velocity
uctuations. (Direct method)
Figure 4.25 - Isosurface of Y
CH4
= 0.004 coloured by convection phase (rad) for diagonal injection. (Flight
time method)
124
Figure 4.26 - Comparison of the phasesPDF and its cumulant on the isosurface Y
CH4
= 0.004.
PDF : ight time method. direct method.
Cumulant : ight time method . direct method.
Figure 4.27 - Comparison of the phasesPDF and its cumulant obtained inside maximum extension ame zone.
PDF : ight time method. direct method.
Cumulant : ight time method . direct method.
125
4.3.10 Conclusions
Large Eddy Simulations are used to compute the forced response of an acoustically excited swirled
burner. The conguration is a 1:1 representation of a prototype gas turbine provided by Siemens and in
which premixed gases are injected through two complex-geometry swirlers at two different equivalence
ratios (0.5 and 0.8). The LES solver uses hybrid meshes and high-order schemes. Combustion/turbulence
interaction is modelled using the thickened ame model. A two-step scheme for methane/air combustion
is used to represent chemistry. A comparison of combustion phases given by LES and measurements
performed in Karlsruhe yields good agreement. Exciting the main inlet at frequencies varying from 80
to 250 Hz, two main ow structures are evidenced by LES:
A toroidal structure which modulates the global heat release by interacting with the ame surface.
A precessing vortex core attached to the centre of the axial swirler and whose frequency is inde-
pendent of the inlet excitation frequency.
Finally, the maps of combustion phase and response amplitude obtained by LES show that the phase
between heat release and inlet velocity uctuations cannot be simply related to a convective phase from
the burner inlet to the ame front as currently done when no LES results are available.
More generally, this study conrms the potential of LES to study unsteady turbulent combustion in
complex geometries.
126
sector (3D)
sector (3D)
This conguration uses almost the same burner as conguration C but mounted on a sector of an annular
chamber (see Figure 4.28). The simulation includes a 15
sector of the whole combustion chamber.

Side faces are enclosed with slip adiabatic rigid walls. The mesh contains 339417 nodes and 1859248
tetrahedral elements. Validation concerning mean and RMS results will not be presented here. This same
Figure 4.28 - Burner mounted on an annular combustion chamber.
mesh has been used to explore the inuence of piloting on combustion stability [140]. By adding two
side burners to this conguration, it is also used by Staffelbach et al. [144] to evaluate the inuence of
the extension of the mesh on the resulting unstable modes of the conguration.
Conguration D is used here to compare results of FTF (F
dim
and F
norm
) using HF-FFT, WN-FFT and
WN-WH methods.
127
4.4.1 General computational remarks.
Two main methods using LES to get the FTF are compared. The rst method (HF method) requires the
injection of a harmonic signal at the main inlet of the conguration. The frequency of forcing is changed
to get the dependance of the FTF to frequency. The conguration is forced at six frequencies between
90 and 540 Hz.
With an explicit resolution of the Navier-Stokes equations and a time step of the order of 4.0e
7
s,
25000 iterations are needed to compute one period if the conguration is forced with a 100 Hz acoustic
wave. Since the HF-FFT method uses the Fourier transform of signals, the spectral resolution is equal
to the inverse of the total computed time. Therefore, to get a resolution of 20Hz, 0.05s have to be
computed, which corresponds to 125000 explicit iterations. On 64 processors on a CRAY XD1, it
represents 50h of computation. Therefore, no calculations for lower frequencies than 90Hz have been
made to save computational time.
Since WN-WH method is based on the correlation of signals and not on their Fourier transform,
theoretically only one period has to be resolved to state for the time delay, even if the third simple test
shows a selectivity issue concerning the Wiener-Hopf inversion method (see Section 3.3). Therefore, to
enable a simpler comparison between the three methods and to avoid this issue, they are applied on the
same elds of heat release and velocity uctuations corresponding to a minimum of ve periods for all
concerned frequencies.
Table 4.5 summarizes the physical simulated times and CPU costs for the turbulent case and for
the three methods. All computations have been done on a CRAY XD1 (120 processors).
Forced cases Turbulent combustor
HF-FFT 90 Hz Physical time: 55ms Cpu Cost : 54h
HF-FFT (global) Total Cpu Cost : 192h
WN-WH or WN-FFT (F
cut
= 600Hz) Physical time: 55ms Cpu Cost : 54h
Table 4.5 - Computational costs and simulated times for the different runs of the turbulent conguration.
Note that the physical simulated time can be decreased as the frequency increases because the period
of the forcing also decreases. Also note that the frequency resolution is not kept constant. No less than
ve periods have been computed for the harmonically forced cases. The last two lines compare the global
computational cost of the two types of methods. Since different computations have to be done for each
128
sector (3D)
frequency when using HF-FFT method, the cost of this method is four times bigger. The goal of the
following section is therefore to validate WN-FFT and WN-WH methods in an industrial conguration.
4.4.2 Reference signals
For both methods, investigations of global and local FTF are done. Depending on global or local FTF
analysis, different response signals have to be compared to the reference signal.
Comparison of global FTF.
In this case, the reference inlet velocity signal is the harmonic or white-noise uctuating velocity, and
the response signal is the integrated heat release uctuation over the whole volume. Figure 4.29 presents
these signals for the turbulent conguration and for the white noise forced case. Inlet velocity and heat
release signals are normalized by their mean values. There is a clear correlation between these two
signals. One can observe that a maximum of inlet velocity induces a maximum of heat release after
approximately 2-3 ms .
Figure 4.29 - Normalized signals of reference for the white-noise forced case: inlet velocity heat
release
Comparison of local FTF.
In this case, 3D elds of heat release uctuation are used. The transfer function of each grid point
is calculated using HF-FFT, WN-FFT and WN-WH methods. Figure 4.30 shows a 2D longitudinal
cut colored with the instantaneous local heat release. The LES are done using the optimal local ame
thickening model [125] with 7 points in the ame front. It leads to a thickening factor ranging from
129
approximately 5 to 50 depending on the local rening of the mesh. A point shows the location of the
reference inlet point. This local treatment enables amplitude and time delay from FTF 3D elds to be
compared using different system identication techniques.
Figure 4.30 - 2D longitudinal cut colored by instantaneous heat release. Black point marks the reference inlet
point.
norm
)
Amplitude
Figures 4.31 present the amplitudes of the FTF obtained with HF-FFT, WN-FFT and WN-WH methods
from signals of Fig.(4.29). It appears that HF-FFT and WN-WH methods give similar results for the
amplitude of the turbulent FTF, with a minimum taken from the Wiener-Hopf curve at approximately
300Hz. In this case, the WN-FFT results are not in agreement with the other results. Figure 4.31 also
shows that the result obtained with WN-WH method is sensitive to the size of the lter used to compute
auto-correlation matrix and cross-correlation vectors. Depending on the value chosen for the lter length
L, the response experiences a variation of approximately 10 percent. When approaching the cut-off
frequency (600 Hz) of the low pass lter used to obtain the white noise signal, the discrepancy between
results for different L values increases. In contrary to what is observed for the laminar FTF amplitude
in chapter 4.2, the amplitude of the FTF is not greater than one for the frequencies of interest. It means
that the relative uctuation of heat release is lower than the relative uctuation of the inlet velocity for
all frequencies. Figure 4.31 shows that a maximum of the FTF amplitude is reached around 500Hz. If
130
sector (3D)
the inlet velocity is forced at 10% of its mean value at 500Hz, the heat release will uctuate with an
amplitude of 8% of its mean value.
Using Eqs.(2.10) and (2.20), the amplitude [F[ at 100Hz can be estimated. It is equal to 1.77. This value
can be compared to the asymptotic value given by Eq.(2.13). With a mean temperature of 1835K in
burnt gases and a temperature of 685K at the inlet, the asymptotic value of the interaction index for low
frequencies is equal to 1.68. The value for [F[ at 100Hz therefore compares well with this theoretical
amplitude.
Figure 4.31 - Amplitude of FTF : HF-FFT method. + WN-FFT method. .... WN-WH method for various values
of the lter size L.
Phase
Figure 4.32 compares the times delays for the conguration D obtained with HF-FFT, WN-FFT and WN-
WH methods. As for conguration B, it shows that HF-FFT and WN-WH time delays almost match for
all frequencies. WN-FFT method gives almost incoherent time delays that can hardly be compared to the
other results. Results show a discontinuity around 300Hz. This artefact is only due to the phase crossing
the 2 limit and going back to zero. Without reference time for the delay, values above 300 Hz are either
or T + without changing the phase between the heat release uctuation and the reference velocity
uctuation. Though, uctuations (depending on L) of the results of Wiener-Hopf relations inversion are
important near 300 Hz and 600 Hz. The last frequency corresponds to the cut-off frequency and has
already been pointed out concerning the amplitude analysis. The source of uctuations around 300 Hz is
not clear but corresponds to the fact that the phase crosses the 2 limit for slightly different frequencies
depending on the value of the L parameter.
131
Figure 4.32 - Time delay of FTF : HF-FFT method. + WN-FFT method. .... WN-WH method (for various values
of the lter size L).
4.4.4 Local comparison of FTF (F
dim
)
It has been shown in the previous section that the global response of the system is almost identical when
forced with a harmonic signal or with a ltered white noise. It is therefore interesting to focus on the
local response of the ame. The analysis is done by comparing results obtained with the two methods
HF-FFT and WN-WH at 90 Hz. the WN-WH method gives the FTF for all the frequency range included
in the spectrum of the imposed ltered white noise. Results at 90 Hz are therefore extracted from the
total resolved spectrum for the WN-WH method. The frequency of 90Hz was chosen because it is the
value observed experimentally in the full combustor (it is an azimutal mode). Longitudinal cuts colored
by the local amplitude and time delay are used to support this analysis. Finally the weighted probability
density function of time delays for both cases is commented.
Amplitude
Figures 4.33 and 4.34 show 2D longitudinal plane cuts colored by the amplitude of the FTF at 90Hz,
respectively obtained using the HF-FFT and WN-WH methods. Note that the mean ame position is
slightly different whether the ame is forced using HF-FFT method or WN-WH method. This is due
to the fact that the simulation time is not long enough to ensure a perfect convergence of the mean heat
release eld.
Figure 4.33 shows that the response amplitude to the 90Hz acoustic excitation is almost axi-symmetric
although the geometry of the combustion chamber is not. It means that the shape of the forced ame
is mostly imposed by the injector which is axi-symmetric. The 2D plane cut colored by the response
amplitude is separated in two zones which react differently to the excitation.
In Zone 1, the response amplitude is intense. It corresponds to high values of local heat release
132
sector (3D)
uctuations. This is due to the equivalence ratio of the axial swirler ( = 0.8) which is higher
than the equivalence ratio of the diagonal swirler ( = 0.5). The mean heat release in this zone is
therefore higher and leads to more heat release uctuations for a given incoming velocity uctua-
tion.
In Zone 2 the level of mean heat release is almost constant along the ame front but the side
anchored at the burner mouth and the side stabilized by the precessing vortex core [47, 147] show
differences in their response amplitude to the uctuation. The part of the ame front attached to
the outer part of the burner reacts more weakly to the perturbation showing that this part of the
ame is well anchored and less prone to instability.
The separation of the response of the white noise forced ame in two zones is not so clear on gure 4.34
: some parts of the zone 2 show a response which has the same amplitude as in zone 1. The response
amplitude is not perfectly axi-symetric. This can be due to the effects of the geometry in the case of the
white-noise forced ame. Also note that as for the harmonically forced case, the outer ame front reacts
less intensively to the incoming perturbation than the inner side.
133
Figure 4.33 - 2D longitudinal cut colored with the amplitude of FTF: HF-FFT method. Black line marks the
mean ame position.
Figure 4.34 - 2D longitudinal cut colored with the amplitude of FTF: WN-WH method. Black line marks the
mean ame position.
134
sector (3D)
Time delay
Figures 4.35 and 4.36 show 2D longitudinal plane cuts colored by the time delay of the FTF at 90Hz
respectively obtained using the HF-FFT and WN-WH methods. The delays change between 0 and 11 ms
(the forcing period at 90Hz).
The zones described for the amplitude analysis are also present on those gures. They also correspond
to different types of responses:
In zone 2, the behaviour of the time delay for the present ame is well described by the 1Danalysis,
with lower time delays after the mean ame front and higher time delays in the inner uctuating
area.
In zone 1, the distribution of time delays is reversed: shorter time delays are observed before the
mean ame front. It is due to the fact that the ame rst contracts in the cycle, burning nearer to
the injection, and then expands to reach the black zone which corresponds to higher time delays.
The dimensions of the two zones are different whether the ame is forced with a harmonic signal or with
ltered white noise. It shows that even if the global time delay is almost the same for the two forcing
methods, the inuence of the other frequencies present with WN-WH method changes the local response
of the ame (i.e. the localisation of zones 1 and 2) at 90Hz.
The maps of amplitudes (Fig.(4.33) and (4.34)) and of local time delays (Fig.(4.35) and (4.36)) exhibit
obvious structures. Their interpretation however is difcult. Zone 1 seems to be the place where the most
intense interaction takes place, while zone 2 yields more diffuse ame responses but with time delays
which exhibit the pattern of oscillating ames seen in Fig.(4.5) for 1D ames. This suggests that the
inner ame, stabilized on the axial swirler hub, is the most unstable one and the rst source of the ame
response. Being more precise is difcult and using these results to study stability requires an acoustic
code and is done in other PhDs [8].
Weighted Probability density functions of time delays.
The 3D elds of amplitude and time delay obtained with both methods are used to reconstruct the
Weighted-PDF of time delays. The way these PDF of time delays are obtained is described in sec-
tion 4.3. Figure 4.37 compares the two weighted PDF of time delays. Both show a maximum around
2 ms which is the global value of the time delay at 90Hz. They also both show a peak around 6ms. It
means that even if the global value of the time delay is 2ms a non negligible part of the uctuations of
heat release that occur out of phase with the inlet velocity. Figure 4.35 shows that this value corresponds
to heat release uctuations occurring inside the mean ame position in zone 1 and 2. The main difference
between the two pdf of time delays is the probability of nding a time delay around 1 ms which is much
smaller when the ame is forced with ltered white-noise. This may be due to the inuence of higher
frequencies on the 90Hz turbulent FTF using WN-WH method.
135
Figure 4.35 - 2D longitudinal cut colored with the time delay of FTF: HF-FFT method. Black line marks the
mean ame position.
Figure 4.36 - 2D longitudinal cut colored with the time delay of FTF: WN-WH method. Black line marks the
mean ame position.
136
sector (3D)
Figure 4.37 - PDF of weighted time delays : WN-WH method. HF-FFT method.
4.4.5 Concluding remarks for conguration D
The inuence of frequency on the FTF of a turbulent combustor has been studied using three different
methods of forcing and post-processing. The rst output of this study concerns the global FTF com-
parison. The WN-WH method gives results that are similar to the results obtained using the harmonic
method though its computational cost is much lower (4 times in this case). Yet, it has to be noticed that
the white-noise forcing method requires the use of the Wiener-Hopf relation inversion and therefore has
specic drawbacks that have to be carefully addressed.
First the WH method is very sensitive to the size of the lter used for the inversion. The number
of elements of the lter multiplied by the time sampling of the signal represents the time memory
of the system. When inverted, this time represents the minimum frequency this method will be
able to treat. But due to the matrix inversion, the size of the lter also has to remain as low as
possible to give precise results. This leads to an optimal value of the size of the lter which may
be sometimes hard to nd.
When forcing with white noise, extreme care has to be given to the lter cut-off frequency and to
the type of lter used. In this study, the cut-off frequency is set to 600Hz which has been shown as
too close to the frequencies of interest. A fourth-order low-pass lter has also been used to avoid
any unwanted high frequency inuences on the measured FTF.
The local response of the turbulent ame is studied at 90Hz. Both methods (HF-FFT and WN-WH) give
similar FTF and probability density functions of time delays. They both show that as already observed in
chapter 4.3 the local response of the turbulent ame can be very different from its global response, with
locations where the uctuations of HR can be out of phase with the uctuations of velocity even though
the global response is almost in phase.
137
4.5 Evaluation of FTF measurements methods in LES.
Four different methods for the measurement of FTF in LES have been tested. Table 4.6 summarizes the
main advantages and drawbacks of these different methods that derive from the previous results. + and
- signs under the name of each method is the global author evaluation for their use in the measument of
FTF in LES. Signs ! emphasize the major advantages and drawbacks of the diffrent methods.
An important aspect of this study is their link with stability analysis of combustors. Obviously,
FTF do have an inuence on the frequency and amplication rates of modes in the numerical methods
used for combustor stability [9, 91]. The present results show that methods like WN-WH and HF-FFT
provide slightly different local FTF maps. Whether these differences will affect or not signicantly the
frequencies and amplication rates of modes remains to be studied. What this study has shown is how
to construct FTFs which is the important brick of acoustic analysis : further studies are needed to
determine which method will be the most precise. Note also that experimental results on global FTF are
available but that no-one has studied local FTF yet. This is obviously a required step for the future.
138
4.5 Evaluation of FTF measurements methods in LES.
M
e
t
h
o
d
s
A
d
v
a
n
t
a
g
e
s
D
r
a
w
b
a
c
k
s
T
D
m
e
t
h
o
d
-
-
C
o
m
p
u
t
a
t
i
o
n
a
l
c
o
s
t
:
O
n
l
y
r
e
q
u
i
r
e
s
t
h
e
s
i
m
-
u
l
a
t
i
o
n
o
f
t
h
e
s
t
e
a
d
y
s
t
a
t
e
s
o
l
u
t
i
o
n
.
C
o
m
p
l
e
x
i
t
y
:
I
n
j
e
c
t
i
o
n
o
f
a
p
a
s
s
i
v
e
s
c
a
l
a
r
t
o
g
e
t
t
h
e
o
w
t
i
m
e
h
i
s
t
o
r
y
!
R
e
s
u
l
t
s
:
O
n
l
y
r
e
c
o
v
e
r
s
t
i
m
e
d
e
l
a
y
s
!
P
r
e
c
i
s
i
o
n
:
L
a
c
k
o
f
p
r
e
c
i
s
i
o
n
w
h
e
n
t
i
m
e
d
e
l
a
y
i
s
n
o
t
d
i
r
e
c
t
l
y
l
i
n
k
e
d
w
i
t
h
c
o
n
v
e
c
t
i
v
e
p
h
e
n
o
m
e
n
a
H
F
-
F
F
T
m
e
t
h
o
d
+
+
!
P
r
e
c
i
s
i
o
n
:
c
o
m
p
a
r
e
s
w
e
l
l
w
i
t
h
e
x
p
e
r
i
m
e
n
t
s
i
n
t
e
r
m
s
o
f
t
i
m
e
d
e
l
a
y
s
!
P
r
e
c
i
s
i
o
n
:
c
o
m
p
a
r
e
s
w
e
l
l
w
i
t
h
a
n
a
l
y
t
i
c
a
l
m
o
d
e
l
s
R
e
s
u
l
t
s
:
g
i
v
e
s
a
c
c
e
s
s
t
o
t
h
e
l
o
c
a
l
r
e
s
p
o
n
s
e
o
f
t
h
e
a
m
e
!
C
o
m
p
u
t
a
t
i
o
n
a
l
c
o
s
t
:
e
a
c
h
f
r
e
q
u
e
n
c
y
r
e
q
u
i
r
e
s
a
d
i
f
f
e
r
e
n
t
c
o
m
p
u
t
a
t
i
o
n
W
N
-
F
F
T
m
e
t
h
o
d
-
-
C
o
m
p
u
t
a
t
i
o
n
a
l
c
o
s
t
:
o
n
l
y
o
n
e
c
o
m
p
u
t
a
t
i
o
n
o
f
t
h
e
w
h
i
t
e
-
n
o
i
s
e
f
o
r
c
e
d
c
o
n
g
u
r
a
t
i
o
n
i
s
n
e
e
d
e
d
P
r
e
c
i
s
i
o
n
:
l
a
c
k
o
f
p
r
e
c
i
s
i
o
n
d
u
e
t
o
F
F
T
p
o
s
t
-
p
r
o
c
e
s
s
i
n
g
o
f
n
o
n
-
h
a
r
m
o
n
i
c
s
i
g
n
a
l
o
n
c
o
n
g
u
r
a
t
i
o
n
B
!
R
e
s
u
l
t
s
:
T
o
t
a
l
l
y
i
n
c
o
h
e
r
e
n
t
r
e
s
u
l
t
s
f
o
r
c
o
n
g
u
r
a
-
t
i
o
n
D
W
N
-
W
H
m
e
t
h
o
d
+
+
(
+
)
!
C
o
m
p
u
t
a
t
i
o
n
a
l
c
o
s
t
:
o
n
l
y
o
n
e
c
o
m
p
u
t
a
-
t
i
o
n
o
f
t
h
e
w
h
i
t
e
-
n
o
i
s
e
f
o
r
c
e
d
c
o
n
g
u
r
a
t
i
o
n
i
s
n
e
e
d
e
d
!
P
r
e
c
i
s
i
o
n
:
c
o
m
p
a
r
e
s
w
e
l
l
w
i
t
h
H
F
-
F
F
T
r
e
-
s
u
l
t
s
R
e
s
u
l
t
s
:
g
i
v
e
s
a
c
c
e
s
s
t
o
t
h
e
l
o
c
a
l
r
e
s
p
o
n
s
e
o
f
t
h
e
a
m
e
P
r
e
c
i
s
i
o
n
:
r
e
s
u
l
t
s
a
r
e
d
e
p
e
n
d
e
n
t
t
o
t
h
e
L
p
a
r
a
m
e
t
e
r
f
o
r
t
h
e
W
i
e
n
e
r
-
H
o
p
f
i
n
v
e
r
s
i
o
n
!
C
o
m
p
l
e
x
i
t
y
:
d
i
f
c
u
l
t
i
e
s
i
n
n
d
i
n
g
t
h
e
o
p
t
i
m
a
l
v
a
l
u
e
o
f
t
h
e
L
p
a
r
a
m
e
t
e
r
.
T
a
b
l
e
4
.
6
-
C
o
m
p
a
r
i
s
o
n
o
f
t
h
e
a
d
v
a
n
t
a
g
e
s
a
n
d
d
r
a
w
b
a
c
k
s
o
f
T
D
,
H
F
-
F
F
T
,
W
N
-
F
F
T
a
n
d
W
N
-
W
H
m
e
t
h
o
d
s
.
139
140
Lire
la seconde partie
de la thse

These

Transféré par

Informations du document

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

These

Transféré par

Droits d'auteur :

Formats disponibles

Num ero dordre : 2458

Ecole Centrale de Paris Examinateur

sector (3D) . . . . . . . . . . . . . . . . . . 127

Upstream propagating acoustic wave

f Fourier transform of variable f

f Favre ltering operator

Subgrid scale component (Part I)

Time uctuation of variable f

= ff is the subgrid scale part

voln. The subgrid scale turbulent velocity u

is the turbulent Reynolds number and c

has to be evaluated. Typically, l

. Recalling that the ame

1, and that the Kolmogorov scale Reynolds

sector (3D) . . . . . . . . . . . . . . . . . . 127

and a priori depends on the frequency.

the reference (inlet) velocity uctuation. n and

are respectively the local heat release and

at the reference point, this reference

is the Fourier transform of the forced

(t) and constructing F

sector of the whole combustion chamber.

Vous aimerez peut-être aussi