Modélisation de la combustion turbulente : application des méthodes de tabulation de la chimie détaillée l'allumage forcé, Numerical simulation of forced ignition using LES coupled with a tabulated detailed chemistry approach

Subramanian V. Subramanian

Description

Sous la direction de Pascale Domingo, Luc Vervisch
Thèse soutenue le 12 janvier 2010: INSA de Rouen
L'optimisation des systèmes d'allumage est un paramètre critique pour la définition des foyers de combustion industriels. Des simulations aux grandes échelles (ou LES pour Large-Eddy Simulation) d'un brûleur de type bluff-body non pré-mélangé ont été menées afin de comprendre l'influence de la position de la bougie sur la probabilité d'allumage. La prise en compte de la combustion est basée sur une méthode de tabulation de la chimie détaillée (PCM-FPI pour Presumed Conditional Moments - Flame Prolongation of ILDM). Les résultats de ces simulations ont été confrontés des résultats expérimentaux disponibles dans la littérature. Dans un premier temps, les mesures de vitesse et du champ de richesse à froid sont comparées aux résultats de la simulation pour évaluer les capacités de prédiction en terme de structure de l'écoulement et de mélange turbulent. Un suivi temporel des vitesses et de la fraction de mélange est réalisé à différents points pour déterminer les fonctions de densité de probabilité (ou PDF)des variables caractéristiques de l'écoulement, à partir des champs résolus en LES. Les PDFs ainsi obtenues servent l'analyse des phénomènes d'allumages réussis ou déficients rencontrés expérimentalement. Des simulations d'allumage forcé ont été effectuées pour analyser les différents scénarios de développement de la flamme. Les corrélations entre les valeurs locales (fraction de mélange, vitesse) autour de la position d'allumage et les chances de succès de développement du noyau de gaz brûlés sont alors discutées. Enfin, une extension de la méthode PCM-FPI avec prise en compte des effets d'étirement est développée à l'aide d'une analyse asymptotique, puis confrontée aux résultats de mesures expérimentales.
-Simulation aux grandes échelles
-Allumage forcé
-Flammelettes laminaires
-Flammes turbulentes
The optimization of the ignition process is a crucial issue in the design of many combustion systems. Large eddy simulation (LES) of a conical shaped bluff-body turbulent non-premixed burner has been performed to study the impact of spark location on ignition success. The chemistry part of the simulation is done using tabulated detailed chemistry approach. This burner was experimentally investigated by Ahmed et al at Cambridge (UK). The present work focuses on the case without swirl for which detailed measurements are available. First, cold fkow measurements of velocities and mixture fraction are compared with their LES counterparts, to assess the prediction capabilities of simulations in terms of flow and turbulent mixing. Time history of velocities and mixture fraction are recorded at selected spots, to probe the resolved probability density function (pdf) of flow variables, in an attempt to reproduce, from the knowledge of LES resolved instantaneous flow conditions, the experimentally observed reasons of success or failure of spark ignition. A flammability map is also constructed from the resolved mixture fraction pdf and compared with its experimental counterpart. LES of forced ignition is then performed using flamelet fully detailed tabulated chemistry combined with presumed pdfs (PCM-FPI). Various scenarios of flame kernel development are analyzed and correlated with typical flow conditions observed in this burner. The correlations between velocities and mixture fraction values at the sparking time and the success or failure of ignition are then further discussed and analysed. The rate of flame development during successful or unsuccessful ignition events are analysed and compared against experimental observations. Finally, from asymptotic flame analysis, a novel approach has been proposed to include flame straining effects in the PCM-FPI method developped at CORIA-CNRS. The new model overcomes the problem associated with classical PCM-FPI closure to model kernel quenching due to intense local turbulence. Computations are done including the flame straining effects and the effect brought by the new model on kernel development is analysed in detail.
-Large eddy simulation
-Spark ignition
-Laminar flamelets
-Turbulent flames
Source: http://www.theses.fr/2010ISAM0001/document

Informations

• Publié par: Thesee
• Nombre de lectures: 68
• Langue: English
• Poids de l'ouvrage: 10 Mo

Extrait

Ecole Doctorale { S.P.M.I.I.
THESE
Presentee par
V. Subramanian
Pour l’obtention du grade de
Docteur de l’Institut National des Sciences Appliquees de Rouen
Discipline : Energetique.
Specialite : Mecanique des Fluides.
Formation Doctorale : Sciences Physiques, Mathematiques
et de l’Information pour l’Ingenieur.
Laboratoire d’Accueil : UMR-CNRS-6614-CORIA.
Numerical Simulation of Forced Ignition Using LES
Coupled with a Tabulated Detailed Chemistry Approach
Le 12 janvier 2010
Membres du Jury
Rapporteurs :
Epaminondas Mastorakos Professeur, Cambridge University, Royaume-Uni
Olivier Colin Ingenieur Chercheur, Institut Fran cais du Petrole (IFP)
Examinateurs :
Denis Veynante Directeur de Recherche, Ecole Centrale Paris (EM2C)
Ananias Tomboulides Professeur, University of Western Macedonia
Directeurs de these :
Pascale Domingo Chargee de Recherche CNRS, INSA de Rouen
Luc Vervisch Professeur, INSA de RouenAcknowledgments
Words fall short as I express my heartfelt acknowledgement to all those people who make me
feel fortunate for where I stand today. With this ecstatic feeling of completing my doctoral
thesis, when I rewind and play back the wonderful memories of my stay from day one, I see
plenty of faces who had supported me through this journey in one way or the other. It is my
greatest pleasure to use few front pages of my thesis to express my heartfelt gratitude to them.
My earnest and foremost acknowledgement goes to my thesis advisors Dr. Pascale Domingo
and Prof. Luc Vervisch, who o ered me a great opportunity to pursue my Ph.D in France, un-
der their esteemed guidance. I enjoyed an overwhelming technical support from Dr. Domingo
throughout the entire course of this thesis work. Every-time, when I faced challenges with
my research work, Dr. Domingo spared her valuable time and o ered inspirational advices
and solutions through series of interactive dialogues. She had the patience to explain me the
functioning of CFD routines, an important process for a person like me who never had any
exposure to CFD coding at the start of this work. Beside being a guide, Dr. Domingo o ered
an excellent moral support and showed great caring, which, I otherwise would have missed
being so far from my family.
Prof. Vervisch inspired me with his ever energetic and enthusiastic approach in dealing
with research problems. During the three years of my Ph.D, I have seen in him an excel-
lent advisor who can bring the best out from his students, an outstanding researcher who
can constructively criticize research and a nice human being who is honest, fair and helpful
to others. I cannot forget his invaluable inputs and hints during the developmental stage of
mono-dimension amelet solver, which nally helped us to have a beautiful publication in
Combustion and Flame journal. Prof. Vervisch also provided me plenty of opportunities to
attend project meetings and seminars, which not only helped me to improve my presentation
skills but also gave me a chance to travel around Europe.
I am deeply indebted to Prof. Mastorakos of Cambridge University (UK), for his support
and suggestions during this work. He also was kind enough to provide the experimental data
for validation and also o ered advices and clari cations at critical junctures. Dr. Ahmed of
Cambridge University deserves my sincerest gratitude for providing me with the experimental
data. I wish to thank all the members of my thesis jury, for their presence on the nal day of
presentation and their constructive criticism as well as kind encouragement.
A special thanks to Dr. Ganesan, who had introduced me to CORIA lab and helped me
set-up my stay in Rouen. I am grateful to Mrs. Isabelle Lebon, who had kindly tolerated my
ibroken French and helped me through many administrative formalities and booked the travel
tickets for many of my trips. Thanks to my dear colleagues Guillaume Godel (who o ered me a
good company in lab for three years and also helped me in preparing French abstract), Nicolas
Enjalbert, Cindy Merlin and Gregory Bonomeau. I really enjoyed our mini basketball sessions
we used to have in our bureau during the break. I would like to appreciate the help o ered
by my seniors Guido Lodata and Alexandre Naudin during the group meeting and several
other occasions. Hearty thanks to each and everyone in the research team of Laboratoire de
Mecanique des Fluides Numerique.
CORIA lab is an unforgettable place to do research work and I would like to thank all
the members of the CORIA family. I would like to specially thank my dearest friends Arnab,
Ouissem, Dahn, Fon and Mechline. The lunch and co ee break chit-chats were really refreshing
and helped me overcome the work stress.
Most important of all, I would like to thank my beloved mother, my sister, my brother
in-law and all my relatives for being a source of constant moral courage and inspiration.
Lastly, I would like acknowledge the nancial support given by the European project
TIMECOP-AE (Towards innovative methods for combustion prediction in aero- engines) and
CNRS (Centre national de la recherche scienti que), who paid my salary during the complete
duration of the thesis.
iiAbstract
Numerical Simulation of Forced Ignition Using LES coupled with a
Tabulated Detailed Chemistry Approach
The optimization of the ignition process is a crucial issue in the design of many combustion
systems. Large eddy simulation (LES) of a conical shaped blu body turbulent non-
premixed burner has been performed to study the impact of spark location on ignition
success. The chemistry part of the simulation is done using tabulated detailed chemistry
approach. This burner was experimentally investigated by Ahmed et al at Cambridge
(UK).
The present work focuses on the case without swirl for which detailed measurements are
available. First, cold ow measurements of velocities and mixture fraction are compared
with their LES counterparts, to assess the prediction capabilities of simulations in terms
of ow and turbulent mixing.
Time history of velocities and mixture fraction are recorded at selected spots, to probe
the resolved probability density function (pdf) of ow variables, in an attempt to reproduce,
from the knowledge of LES resolved instantaneous ow conditions, the experimentally ob-
served reasons of success or failure of spark ignition. A ammability map is also constructed
from the resolved mixture fraction pdf and compared with its experimental counterpart.
LES of forced ignition is then performed using amelet fully detailed tabulated chemistry
combined with presumed pdfs (PCM-FPI).
Various scenarios of ame kernel development are analyzed and correlated with typical
ow conditions observed in this burner. The correlations between velocities and mixture
fraction values at the sparking time and the success or failure of ignition are then further
discussed and analysed. The rate of ame development during successful or unsuccessful
ignition events are analysed and compared against experimental observations.
Finally, from asymptotic ame analysis, a novel approach has been proposed to in-
clude ame straining e ects in the PCM-FPI method developped at CORIA-CNRS. The
new model overcomes the problem associated with classical PCM-FPI closure to model
kernel quenching due to intense local turbulence. Computations are done including the
ame straining e ects and the e ect brought by the new model on kernel development is
analysed in detail.
Keywords: Large eddy simulation, Spark ignition, Laminar amelets, Turbulent ames
iiiResume
Modelisation de la combustion turbulente. Application des methodes de
tabulation de la chimie detaillee l’allumage force
L’optimisation des systemes d’allumage est un parametre critique pour la de nition des
foyers de combustion industriels. Des simulations aux grandes echelles (ou LES pour Large-
Eddy Simulation) d’un bruleur^ de type blu -body non premelange ont ete menees a n de
comprendre l’in uence de la position de la bougie sur la probabilite d’allumage. La prise
en compte de la combustion est basee sur une methode de tabulation de la chimie detaillee
(PCM-FPI pour Presumed Conditional Moments - Flame Prolongation of ILDM). Les
resultats de ces simulations ont ete confrontes des resultats experimentaux disponibles
dans la litterature. Dans un premier temps, les mesures de vitesse et du champ de richesse
froid sont comparees aux resultats de la simulation pour evaluer les capacites de prediction
en terme de structure de l’ecoulement et de melange turbulent. Un suivi temporel des vi-
tesses et de la fraction de melange est realise di erents points pour determiner les fonctions
de densite de probabilite(ou PDF)des variables caracteristiques de l’ecoulement, partir
des champs resolus en LES. Les PDFs ainsi obtenues servent l’analyse des phenomenes
d’allumages reussis ou de cients rencontres experimentalement. Des simulations d’allumage
force ont ete e ectues pour analyser les di erents scenarios de developpement de la amme.
Les correlations entre les valeurs locales (fraction de melange, vitesse) autour de la posi-
tion d’allumage et les chances de succes de developpement du noyau de gaz brles sont
alors discutees. En n, une extension de la methode PCM-FPI avec prise en compte des
e ets d’etirement est developpee l’aide d’une analyse asymptotique, puis confrontee aux
resultats de mesures experim